Done1-4 - NASA · through some of the science team contracts. The Project also purchased 12 MicroTops hand-held sun photometers, 2 PREDE (Japanese built) sun photometers, and a micro-pulse

SIMBIOS Project Annual Report

NASA/TM--1999-208645

SIMBIOS Project 1998 Annual Report

Charles R. McClain, Goddard Space Flight Center, Greenbelt, MarylandGiulietta S. Fargion, SAIC General Sciences Corporation, Beltsville, Maryland


PREFACE

The purpose of this series of technical reports is to provide current documentation of the SIMBIOS Project activities,NASA Research Announcement (NRA) research status, satellite data processing, data product validation and field calibration. This documentation is necessary to ensure that critical information is related to the scientific community andNASA management. This critical information includes the technical difficulties and challenges of combining ocean color data from an array of independent satellite systems to form consistent and accurate global bio-optical time series products. This technical report is not meant to substitute for scientific literature. Instead, it will provide aready and responsive vehicle for the multitude of technical reports issued by an operational project.


Table of Contents

1. SIMBIOS BACKGROUND................................................................................................................................................. 1

2. SIMBIOS PROGRAM AND PROJECT GOALS................................................................................................................ 4

3. SIMBIOS: NRA CONTRACTS............................................................................................................................................ 7

4. SIMBIOS PROJECT ACTIVITIES (1997-1998) .............................................................................................................. 11

5. VALIDATION OF SURFACE BIO-OPTICAL PROPERTIES IN THE GULF OF MAINE AS A MEANS

FOR IMPROVING SATELLITE PRIMARY PRODUCTION ESTIMATES ................................................................... 32

6. OCTS AND SEAWIFS BIO-OPTICAL ALGORITHM AND PRODUCT VALIDATION AND

INTERCOMPARISON IN U.S. COASTAL WATERS ..................................................................................................... 35

7. VALIDATION OF OCEAN COLOR SATELLITE DATA PRODUCTS IN UNDER-SAMPLED MARINE AREAS .. 39

8. STRAY LIGHT AND ATMOSPHERIC ADJACENCY EFFECTS FOR LARGE-FOV, OCEAN VIEWING

SPACE SENSORS. ............................................................................................................................................................ 42

9. BIO-OPTICAL MEASUREMENTS AT OCEAN BOUNDARIES IN SUPPORT OF SIMBIOS. ................................... 44

10. REMOTE SENSING OF OCEAN COLOR IN THE ARCTIC: ALGORITHM DEVELOPMENT AND

COMPARATIVE VALIDATION. ..................................................................................................................................... 47

11. HIGH FREQUENCY, LONG TIME SERIES MEASUREMENTS FROM THE BERMUDA TESTBED

MOORING IN SUPPORT OF SIMBIOS. .......................................................................................................................... 50

12. HIGH-LATITUDE INTERCOMPARISON AND VALIDATION EXPERIMENT (HIVE)............................................. 52

13. SATELLITE OCEAN COLOR VALIDATION USING MERCHANT SHIPS................................................................. 55

14. HIGH ATTITUDE MEASUREMENTS OF RADIANCE AT HIGH SPECTRAL AND SPATIAL

RESOLUTION FOR SIMBIOS SENSOR CALIBRATION,VALIDATION, AND INTERCOMPARISONS................ 58

15. MERGING OCEAN COLOR DATA FROM MULTIPLE MISSIONS............................................................................. 61

16. VALIDATION OF BIO-OPTICAL PROPERTIES IN COASTAL WATERS: A JOINT NASA NAVY PROJECT. ...... 63

17. BIO-OPTICAL MEASUREMENT AND MODELING OF THE CALIFORNIA CURRENT AND

POLAR OCEANS…………………………………………………………………………………………………………66

18. SIMBIOS NORMALIZED WATER LEAVING RADIANCE CALIBRATION AND VALIDATION:

SENSOR RESPONSE, ATMOSPHERIC CORRECTIONS, STRAY LIGHT AND SUN GLINT................................... 69


Table of Contents (cont.)

19. VALIDATION OF CARBON FLUX AND RELATED PRODUCTS FOR SIMBIOS AT THE CARIACO

CONTINENTAL MARGIN TIME SERIES....................................................................................................................... 71

20. THE BERMUDA BIO-OPTICS PROGRAM (BBOP)....................................................................................................... 74

21. SIMBIOS DATA PRODUCT AND ALGORITHM VALIDATION WITH EMPHASIS ON THE

BIOGEOCHEMICAL AND INHERENT OPTICAL PROPERTIES. ............................................................................... 77

22. VALIDATION OF THE SEAWIFS ATMOSPHERIC CORRECTION SCHEME USING MEASUREMENTS

OF AEROSOL OPTICAL PROPERTIES. ......................................................................................................................... 80

23. MEASUREMENTS OF AEROSOL, OCEAN AND SKY PROPERTIES AT THE HOTS SITE IN THE

CENTRAL PACIFIC. ......................................................................................................................................................... 83

24. ASSESSMENT OF THE CONTRIBUTION OF THE ATMOSPHERE TO UNCERTAINTIES IN NORMALIZEDWATER-LEAVING RADIANCE: A COMBINED MODELING AND DATA ANALYSIS APPROACH..................... 86

25. VALIDATION OF THE WATER-LEAVING RADIANCE DATA PRODUCT. ............................................................. 88

26. OCI DATA VALIDATION IN THE WATER ADJACENT TO TAIWAN. ..................................................................... 96

GLOSSARY................................................................................................................................................................................. 97

REFERENCES............................................................................................................................................................................. 99



1

Chapter 1

SIMBIOS Background

Charles R. McClainNASA Goddard Space Flight Center

Greenbelt, Maryland

The Sensor Intercomparison and Merger forBiological and Interdisciplinary Ocean Studies(SIMBIOS) program was conceived in 1994 as aresult of a National Aeronautics and SpaceAdministration (NASA) management review of theagency's strategy for monitoring the bio-opticalproperties of the global ocean through ocean colorremote sensing from space. At that time, the NASAocean color flight manifest included two data buymissions, the Sea-viewing Wide Field-of-view Sensor(SeaWiFS) and Earth Observing System (EOS) Color,and two sensors, Moderate Resolution ImagingSpectroradiometer (MODIS) and Multi-angle ImagingSpectro-Radiometer (MISR), scheduled for flight onthe EOS-AM (1998) and -PM (2001) satellites.

Considerable effort was spent by Dr. McClain(Project Scientist) and Dr. Kirk (Study Manager) onexamining mission scenarios for EOS Color becauseof the slips being encountered with SeaWiFS.However, with the delay of SeaWiFS and an uncertainlaunch schedule, it was not clear that EOS Color wasneeded to fill a potential gap between SeaWiFS andMODIS, especially when five additional ocean colorsystems with similar global capabilities [Ocean Colorand Temperature Sensor (OCTS), Japan; GlobalImager (GLI), Japan; Polarization DetectingEnvironmental Radiometer-1 (POLDER-1) and -2,France; and Medium Resolution ImagingSpectrometer (MERIS), European Space Agency] andseveral other non-global missions by Argentina,Germany, Taiwan, India, Korea, the U.S. Navy, andthe People’s Republic of China, were planned forlaunch during the late 1990s.

The review led to a decision that the internationalassemblage of ocean color satellite systems providedample redundancy to assure continuous globalcoverage, with no need for the EOS Color mission. Atthe same time, it was noted that non-trivial technicaldifficulties attended the challenge (and opportunity) ofcombining ocean color data from this array ofindependent satellite systems to form consistent andaccurate global bio-optical time series products(Figure 1). Thus, it was announced at the October,1994 EOS Interdisciplinary Working Group meetingthat some of the resources budgeted for EOS Colorshould be redirected into an intercalibration andvalidation program. NASA Goddard Space Flight

Center (GSFC) was directed to develop a plan forsubmission to NASA Headquarters (HQ) by May1995. As a result of the directive from NASA/HQ, theocean color group lead by Dr. McClain atNASA/GSFC organized an internationalorganizational meeting in February 1995 at theUniversity of Miami Rosenstiel School for Marine andAtmospheric Sciences. The objective was develop aconceptual plan for a comparison program. The plan(Ocean Color Multisensor Data Evaluation andUtilization Plan, under SIMBIOS Documents,http://simbios.gsfc.nasa.gov/) which outlined NASA’scontribution to the international effort was completed,submitted, externally reviewed, and revised in 1995.

Based on the final approved plan, a NASAResearch Announcement (NRA) was released in July1996, and the SIMBIOS Project Office wasestablished at the NASA (GSFC) in January 1997 (co-located with the SeaWiFS Project). The initialSIMBIOS Program was scoped for five years (1997-2001) and included separate support for a scienceteam (NRA selections) and the Project Office.

Dr. Mueller (San Diego State University) actedas an interim project manager at NASA/GSFC under aone-year assignment to assist in getting the projectoffice organized and the science team contractsexecuted. His assistance in this capacity was essentialas the SIMBIOS Project was beginning just as theSeaWiFS Project was preparing for launch.

In parallel with the NASA SIMBIOS Programplanning, the international effort was being organized.The initial meeting was held in Vittoria, BritishColumbia during September, 1995. As a result of therecommendations from that meeting, the InternationalOcean Colour-Coordinating Group (IOCCG) wasformed (http://www.ioccg.org/). The IOCCGpresently operates under the auspices of the ScientificCommittee on Oceanic Research (SCOR) andchairmanship of Dr. Platt (Bedford Institute ofOceanography). The IOCCG meets one or two timesper year and is generating a series of special reports ontopics essential to the coordination of the internationalocean color community (e.g., IOCCG, 1998).

During calendar year 1997, Dr. Mueller’s tenureas SIMBIOS Project Manager, the primaryaccomplishments included the following:


2

• The science team contracts or interagencyagreements were negotiated (3-year contracts)and first year funds were distributed. Severaladditional investigations, primarily foratmospheric correction studies, were solicited andfunded.

• The initial Project Office staffing actions werecompleted.

• The SIMBIOS processing systems werepurchased and installed.

• An instrument pool was established. Submersibleinstruments for field studies were purchasedthrough some of the science team contracts. TheProject also purchased 12 MicroTops hand-heldsun photometers, 2 PREDE (Japanese built) sunphotometers, and a micro-pulse lidar as part of theinstrument pool. Twelve CIMEL (French built)sun photometers were also purchased to augmentthe Aerosol Robotic Network (AERONET) fordeployment at coastal sites.

• A satellite overflight prediction service wasestablished to assist in coordinating field studiesin synchronizing data collection with satellitepasses (http://simbios.gsfc.nasa.gov/, and http: //simbios.gsfc.nasa.gov/~schedule/Predictions/Current_Cruises.html).

• The first SIMBIOS science team meeting washeld in August at Solomons Island, Maryland.

During the second year of the SIMBIOS Project,Dr. McClain assumed project management of both theSeaWiFS and SIMBIOS as both Dr. Cleave and Dr.Mueller stepped down in their roles as projectmanagers of these two projects, respectively. In thesecond year, many of the original objectives ofSIMBIOS began to come to fruition, some of whichare discussed in more detail in subsequent chapters.Project accomplishments included:

• The first SIMBIOS intercalibration round-robinexperiment (Riley and Bailey, 1998) wascompleted.

• The CIMEL sun photometers were re-engineeredby Meridian Engineering of Hanover, Marylandto improve their survivability in the field.Confirmed (delivered or negotiated) CIMEL sitesinclude Lanai (Hawaii) with a backup inHonolulu, Ascension Island, Bahraine, Turkey(Black Sea), Tahiti, Wallops Island (Virginia),and South Korea. Additional sites beingconsidered are Perth, Australia and Iceland.Delivery of equipment for calibrating polarizedbands was received from the University of Lilleas part of the collaboration with the AERONETgroup at NASA/GSFC [Brent Holben, principalinvestigator (P.I.)]. An automated procedure forextracting data from the AERONET archive forsatellite match-up comparisons was completed.

• The micropulse lidar environmental container wasre-engineered by Meridian Engineering toimprove temperature stability.

• A second copy of the SeaWiFS TransferRadiometer (SRX), the SXR-2, was built byReyer Corporation of New Market, Maryland foruse in the calibration round-robin.

• A Marine Optical Spectroradiometer (MOS)ground station subsystem was purchased fromAntrix Corporation Limited (Bangalore, India) forinstallation at Wallops Flight Facility.

• Data processing (Level-0 to Level-3) software forOCTS and MOS were completed. This activityincluded evaluations of the navigation, bandcoregistration, calibration, destripping,atmospheric correction, and bio-opticalalgorithms. Match-up comparisons with field dataand OCTS were also completed using match-upsubscenes provided by National SpaceDevelopment Agency of Japan (NASDA) EarthObservation Research Center (EORC).

• SeaWiFS Bio-Optical Archive and StorageSystem (SeaBASS) holdings were greatlyexpanded as the science team field activities gotunderway.

• The SIMBIOS website was substantiallyenhanced including monthly reports posted by thescience team investigators.

• Science team contract performance was reviewedand the contracts were revised as necessary andrenewed.

• The second science team meeting was held atScripps Institution of Oceanography (SIO) inSeptember 1998.

In 1999, the SIMBIOS Project will finalize theprocessing of it holdings of OCTS data, i.e., all theOCTS data collected at Wallops Flight Facility (eastcoast of North America, Gulf of Mexico, and theCaribbean Sea), which should be completed by earlyspring. The OCTS processing software will beincorporated into SeaWiFS Data Analysis System(SeaDAS) Version 3.3. The next ocean colorinstrument data set to be evaluated will be ADEOS-1/POLDER in cooperation with Centre Nationald’Edudes Spatialle (CNES) and the POLDER scienceteam. The SIMBIOS Project and science team willassist the ROCSAT Ocean Color Imager (OCI)(launched in January) and MODIS (launch scheduledfor July) instrument teams in their evaluations,primarily by providing match-up data for calibrationand validation analyses. The SIMBIOS Project willalso assist NASA/HQ in drafting the secondSIMBIOS NRA which is scheduled for releasesometime this summer. Finally, the third science teammeeting will be held in September in the Washington,D.C. area.

Ins

tru

me

nt

(Mis

sio

n:

Co

un

try

)7

87

98

08

18

28

38

48

58

68

78

88

99

09

19

29

39

49

59

69

79

89

90

00

10

2

7879

8081

8283

8485

8687

8889

9091

9293

9495

9697

9899

0001

02

Pro

of-o

f-C

once

ptC

ZC

S (N

imbu

s-7:

U.S

.)

Glo

bal

OC

TS

(AD

EO

S-I:

Jap

an)

POL

DE

R (A

DE

OS-

I/II

: Fra

nce

& J

apan

)

SeaW

iFS

(Orb

Vie

w-2

: U.S

.)

MO

DIS

(E

OS-

AM

: U.S

.)

MIS

R (E

OS-

AM

: U.S

.)

ME

RIS

(EN

VIS

AT

: ESA

)

GL

I (A

DE

OS-

II: J

apan

)

MO

DIS

(E

OS-

PM: U

.S.)

Lim

ited

Cov

erag

e

MO

S (P

riro

da: G

erm

any

& R

ussi

a)

MO

S (I

RS:

Ger

man

y &

Indi

a)

OC

I (R

OC

SAT

: Tai

wan

)

OSM

I (K

OM

PSA

T: S

outh

Kor

ea)


4

Chapter 2

SIMBIOS Program and Project Goals

Giulietta S. FargionSAIC General Sciences Corporation

Beltsville, Maryland

2.1 PROGRAM GOALS

The purpose of the SIMBIOS project is todevelop a methodology and operational capability tocombine data products from various ocean colormissions in a manner that ensures the best possibleglobal coverage and best exploits the complementarymissions of the sensors. NRA 96-MTPE-04specifically defines the scientific goals in support ofthis undertaking. “Specifically, the objectives are (1)to quantify the relative accuracies of the products frominternational ocean color missions, (2) to improve thelevel of confidence and compatibility among theproducts, and (3) to generate merged, improved level-3 products” (NASA, 1996). A major goal of theProgram is to foster collaborations with the variousspace agencies and science working groups to assisteach other in achieving these objectives. Presently,SIMBIOS is collaborating with Japan on OCTS andGLI, with India and Germany on MOS and just started a relationship with France on POLDER.

2.2 PROJECT OBJECTIVES

The SIMBIOS Program consists of the SIMBIOSScience Team and the SIMBIOS Project. TheSIMBIOS Science Team, as initially defined by theNRA selections, was expanded to include additionalinvestigations for atmospheric correction algorithmvalidation. The SIMBIOS Project has been establishedto provide support and coordination for the SIMBIOSProgram such as administration, projectdocumentation, complete data processing ofcomplementary ocean color missions, support for eachof the individual elements, and oversight of dataprocessing systems supported by the Project.“Specifically, the SIMBIOS Project is responsible forthe following:• Assistance in the collection, processing, archiving

and documentation of the calibration (includinginstrument round-robins, pre- and post-launchsensor calibration) and bio-optical data sets;

• Maintenance of a data archive and storage systemfor tracking instrument calibrations and validatingbio-optical algorithms or merger methodologieswith in situ data;

• Planning and execution of instrument andanalysis round-robins which will includecoincident intercomparisons of hardware andsoftware performance;

• Oversight of the regular revision of the oceanoptics protocols which will establish themethodologies and standards for calibratinginstruments, collecting data, and producing finalresults from approved analysis procedures.

• Assistance in data acquisition during field studies;• Management of a centralized facility for

providing portable field sources which will beused for tracking the calibration of instrumentswhile they are used in the field;

• Assistance in product validation and qualitycontrol assessment;

• Technical monitoring of the prototyping,deployment, or production of new technologydevelopments;

• Assistance in the development, evaluation, andimplementation of product merger schemes andalgorithms; and

• Administer and oversee the regular meeting ofworking subgroups concerned with specific issuesimportant to the success of the SIMBIOS Project”(NASA, 1996).

The SIMBIOS Program is augmented by theparticipation of the NASA-supported MODIS OceansTeam and SeaWiFS Calibration and ValidationProgram. SIMBIOS calibration activities areaccomplished using the calibration activities of thevarious independent participating missions,coordinating activities of the SIMBIOS Project, andcomplementary activities of individual investigatorsselected by the NRAs. The SIMBIOS Project goal isto help science users with a long time series ofcalibrated radiances extending across the boundariesof individual missions. Close cooperation with theSeaWiFS Project is essential to the success of theSIMBIOS Project. The SIMBIOS Project is co-locatedwith the SeaWiFS Project at GSFC, and many ofSIMBIOS staff are shared with SeaWiFS to guaranteea close cooperation to quantify the relative accuracy ofthe data products from different ocean color sensorsand the improve level of confidence and compatibilityamong the products.


5

Table 1. SIMBIOS ProjectSIMBIOS Project Address

NASA/Goddard Space Flight CenterSIMBIOS ProjectCode 970.2, Building 28Greenbelt, MD 20771Telephone: (301) 286-9676FAX: (301) 286-0268E-mail: [email protected]

Personnel of SIMBIOS Project

Sean BaileyRobert Barnes *Patty Clow *Giulietta FargionGene Feldman *Ian Durham *Bryan Franz *Lynne Hoppel *Alice IsaacmanCharles McClain *Fred Patt *Christophe PietrasTom RileyIndia Robinson *Kellie Ruebens *Brian Schieber *Linwood Smith *Paul Smith *Sung Lee *Tamara Tucker *Grey Valenti *Menghua WangBill Woodford *

FutureTech CorporationSAIC General Sciences CorporationNASASAIC General Sciences CorporationNASASAIC General Sciences CorporationSAIC General Sciences CorporationNASASAIC General Sciences CorporationNASASAIC General Sciences CorporationSAIC General Sciences CorporationNASANASASAIC General Sciences CorporationSAIC General Sciences CorporationFutureTech CorporationSAIC General Sciences CorporationSAIC General Sciences CorporationSAIC General Sciences CorporationSAIC General Sciences CorporationUniv. of Maryland BaltimoreFutureTech Corporation

Note: total civil servant staff ~ 2.5 man year

*shared with SeaWiFS Project

2.3 PROJECT ORGANIZATION

The SIMBIOS Project organization structure isscoped to support the primary activities: 1) sensorengineering and calibration, 2) satellite dataprocessing; 3) data product validation, 4) SIMBIOSScience Team, and 5) SIMBIOS Project Office. TheSIMBIOS Project organization chart is shown inFigure 1 and illustrates the individual subtasks undereach primary activity with staff associations. TheSIMBIOS Project key personnel and their areas ofresponsibility at GSFC are also given in Table 1. Thesatellite data processing elements requires closecoordination between the data product validation andthe sensor engineering and calibration. A completeoverview of the primary activities 1 to 4 will bedescribed in the following chapter. The Project Officeadministrative structure is as follows: system support,administrative support, technical support, scienceteam support, procurement support, and resources

support. At the present time, only SeaWiFS andSIMBIOS participants and collaborators associatedwith other relevant programs, and science teams whoactively contribute data or algorithms, are givenaccess to SIMBIOS Project resources, such as theSeaWiFS Bio-optical Archive and Storage System(SeaBASS) and Field Instrument Pool. This limitationis to protect the interests of the researchers currentlysubmitting to the SeaBASS archive.

The SIMBIOS Project policy is to have alldocuments easily available to the SIMBIOS PIs and tothe larger ocean color scientific community. Access tothe information is gained through the SIMBIOSwebsite. The SIMBIOS website was organized toserve as the main information resource to access theProject activities, Project Office and Science Team.The website is organized under five main topics:News and Information, Support Services andSchedule, Project Status, Instrument Pool andContacts. Briefly, under Project Status are foundmonthly progress updates from the Project and theScience Team. The Project update covers processingand calibration accomplishments for the variousmissions, the instrument pool and real-time cruisesupport, and other key events or Project issues. TheScience Team oversight is divided into two categories:funded PIs and international collaboration. UnderStatus of Contracted PIs, the abstracts of the fundedresearch, monthly progress reports, documents inprocess, and other key events can be found.

Under News and Information, is an archive ofmost of the produced SIMBIOS documentation andcalendar of events. In this section, a user can easilyfind out which conferences or symposia the Projectstaff is attending, documents on SIMBIOS workshopsand team meetings, and special activities, such as theSIMBIOS participation in Indian Ocean Experiment(INDOEX). An online bibliography includesSIMBIOS technical memoranda or papers presentedand published, with online abstracts. SeaWiFSscheduling and the SeaBASS database are accessibleonline. For further information see http://simbios.gsfc.nasa.gov. All sections are updated as needed.

In addition, the SIMBIOS Project has an internalweb interface to closely follow the contractual statusof the funded PIs. Contractual issues such as datadelivery, contract evaluation performance, contractmodifications, requested extensions, procurementsupport performance, and other items are tracked here.In addition, the SIMBIOS Project holds bimonthlystaff meetings, weekly review on technical papers orproject issues (such as calibration, in situ matchups,etc.), and key staff attends the SeaWiFS meetings.Monthly SIMBIOS Project progress reports aregenerated and posted to the SIMBIOS website. TheProject also gives a formal oral presentation to theEarth Science Directorate (Code 900) management.

* T

he S

IMB

IOS

Pro

ject

is c

o-lo

cate

d w

ith

the

Sea

WiF

S P

roje

ct a

nd m

any

of th

e S

IMB

IOS

Sta

ff a

re s

hare

d.

G. F

argi

onA

ssis

tant

Pro

ject

Man

ager

F. P

att*

Con

trac

t Sta

ff S

uper

viso

r

M

. He

O

. Kop

elev

ich

G

. Kor

otae

v

H. L

i

S. S

aito

h

F. C

have

zB

. Che

nG

. Cot

aT.

Dic

key

P. F

late

au

R. F

roui

nR

. Gre

enW

. Gre

ggM

. Mil

ler

R. M

ille

r

US

SIM

BIO

S P

I's

A. S

ubra

man

iam

K. S

tam

nes

R. Z

anev

eld

Inte

rnat

iona

l SIM

BIO

S P

IsW

. Bal

chJ.

Bro

ckC

. Bro

wn

D. C

apon

eK

. Car

der

G. M

itch

ell

J. M

uell

erF.

Mul

ler-

Kar

ger

J. P

orte

rD

. Sie

gel

SIM

BIO

S S

cien

ce T

eam

N

RA

-96

G. F

eldm

anS

atel

lite

Dat

a P

roce

ssin

g

Sate

llite

Dat

a P

roce

ssin

gSa

telli

te D

ata

Ana

lysi

sD

ata

Set

Man

agem

ent

B. F

ranz

*J.

Gal

es

Pro

cess

ing

Syst

em S

uppo

rtS

. Lee

*P.

Sm

ith

*B

. Woo

dfor

d *

C. M

cCla

in*

Dat

a P

rodu

ct V

alid

atio

n

Mer

ged

Dat

a Se

t Ass

embl

yA

lgor

ithm

Dev

elop

men

tA

lgor

ithm

Val

idat

ion

A. I

saac

man

M. W

ang

SeaB

ASS

Int

erfa

ceJ.

Wer

dell

*

Cru

ise

Sche

dule

Sate

llite

Ove

rflig

ht S

uppo

rtS

. Bai

ley

*

Mat

ch-u

p D

ata

Ana

lysi

sS

. Bai

ley

*J.

Wer

dell

G. F

argi

onS

enso

r E

ngin

eeri

ng

& C

alib

rati

on

SIR

RE

X P

rogr

amA

ER

ON

ET

Int

erfa

ceC

. Pie

tras

G. M

eist

er

Sate

llite

Sen

sor

Cha

ract

eriz

atio

n an

dD

ata

Cal

ibra

tion

R. B

arne

s*

Fie

ld I

nstr

umen

t P

ool

S. B

aile

y*

Pro

cure

men

t Su

ppor

tP.

Clo

w *

L. H

oppe

l*C

. Tig

er*

Tec

hnic

al S

uppo

rtI.

Ant

onen

ko*

Syst

em S

uppo

rtA

dmin

istr

atio

n Su

ppor

tR

esou

rces

Sup

port

I. R

obin

son

*J.

Stu

bble

fiel

d*T.

Tuc

ker*

SIM

BIO

S Sc

ienc

eT

eam

Sup

port

G. F

argi

on

SIM

BIO

S P

roje

ct O

ffic

e

C. M

cCla

in*

Pro

ject

Sci

enti

st &

Man

ager

..

SIM

BIO

S P

roje

ct O

rgan

izat

ion

Cha

rt


7

Chapter 3

SIMBIOS: NRA Contracts

Giulietta S. FargionSAIC General Sciences Corporation


3.1 PROJECT RESPONSIBILITY

Contract administration involves ensuringcompliance with contract terms and conditions duringthe performance of the contract. The performance ofSIMBIOS Project contracts are monitored in severalways by several individuals in the Project office.These individuals include a Contracting Officerspecialist, a Resource/Finance person, the AssistantProject Manager with technical background andexperience with projects and their costs, and theProgram Manager. The SIMBIOS Project contractpersonnel and their areas of responsibility at GSFC aregiven in Table 1. The most important step of contractadministration is to review the requirements andspecific obligations set forth in the contract. This roleis performed primarily by the Assistant ProjectManager. In matters related to contract laws andinterpretation, assistance is given by the ContractingOfficer. In matters related to the status of payments orprovisions relating to payments, assistance is given bythe Resource/Finance person. While SIMBIOScontractors have historically addressed questionsabout contract administration to the Assistant ProjectManager, they are entirely free to address questionsdirectly to the appropriate personnel.

A wide selection of contract types is available foruse between the Project office and contractors. Thisselection provides a wide selection of services and alarge degree of flexibility of contract conditions (FAR,1998). GSFC Procurement Office selected a Researchand Development Service, firm-fixed price (FFP)contract-type for the NASA Research Announcement(NRA) selections. The FFP contract provides for aprice that is not subject to any adjustment by reason ofthe cost experience of the contractor in theperformance of the contract. Use of the FFP contractimposes a minimum administrative burden on thecontracting parties. Contracts are very different fromthe usual scientific research grants in three mainaspects: 1) a work statement concerning the endobjectives of the research (i.e., the proposal), 2) thetechnical data to be delivered under the contract withtime delivery specifications, and 3) scientific andtechnical reports, consistent with the objectives of theeffort involved, as a permanent record of the workaccomplished under the contract.

3.2 CONTRACTS OVERVIEW

The initial SIMBIOS program was scoped for fiveyears (1997-2001) and includes support for the NRAselections (i.e., science team) and the SIMBIOSProject. The Interim SIMBIOS Office (ISPO)successfully negotiated specific work statements forthe 21 NRA selections and issued contracts startingfrom late summer 1997. SIMBIOS funded contracts,PI names and delivery requirements are summarizedin Table 2.

NASA Procurement requires formal evaluationsfor all contracts at the end of each contract year. Theseevaluations are to go into a database and will beshared with the PIs’ institutions. The performanceevaluation has four categories: quality, time, other andprice (price is not relevant here because contracts arefixed cost). SIMBIOS Project Office procedure is tocoordinate an inside panel to perform an across-the-board evaluation of all funded contracts. Under qualitywill be considered 1) data quality and completeness,2) ancillary information provided on the data(metadata), 3) the data's usefulness in relation toSIMBIOS goals, i.e., calibration, validation, andalgorithm development, and 4) quality of technicalreports. Time is a mixed bag, but will be viewed inrespect to data and documentation (monthly, year-endreports, and special topic publications) and deliverytimes. Under “other” will be considered 1) scientificpublications and scientific achievements, 2) scienceteam collaboration and involvement, and 3) othersignificant events occurring during the contract periodevaluate.

Table 1. Contract key personnel

SIMBIOS NRA Organization

Contracting Officer: Lynne HoppelResource/Financial Officer: Patty ClowAssistant Project Manager: Giulietta FargionProgram Manager: Charles McClain


8

3.3 SCIENCE TEAM

NASA SIMBIOS Science Team PrincipalInvestigators are composed of those selected under theNRAs, some members of the MODIS Ocean Teams,and certain members the SeaWiFS Project. TheScience Team can be grouped under three workingareas: 1) Ocean Bio-optical and SensorCharacterization Studies, 2) Data Merger Studies, and3) Atmospheric Correction Studies. There are manymore U.S. and international co-investigators and

collaborators actively participating in the NASAcomponent of the international SIMBIOS program.

The funded Principal Investigators andcollaborators are shown in Table 3. SIMBIOS ScienceTeam meeting is held each year, information anddocumentation’s produced in the two previousmeetings (1997 and 1998) are posted on the SIMBIOSwebsite.

Figure 1 shows the global distribution of theNRA-selected SIMBIOS field studies, and followingchapters from 5 to 27 describe the funded researchtopics, field studies and the first year research results.

Table 2. SIMBIOS contracts and government interagency agreements *.

Principal Investigator

Con

trac

tN

umbe

r

Begin Date

Mon

thly

Rep

ort

End

-yea

r T

M

Cru

ise

Rep

ort

Met

adat

a

Spec

ial T

opic

TM

Fie

ld D

ata

Wit

hin

3 or

6 M

onth

sA

lgor

ithm

Dev

elop

men

tSc

ienc

e T

eam

Mee

ting

William Balch 97268 9/23/97 ÷ ÷ ÷ ÷ ÷

J. Brock / C. Brown * 7/21/97 ÷ ÷ ÷ ÷ ÷

D. Capone / E. Carpenter 97131 9/11/97 ÷ ÷ ÷ ÷ ÷

Ken Carder 97137 9/23/97 ÷ ÷ ÷ ÷ ÷

Francisco Chavez 97134 9/26/97 ÷ ÷ ÷ ÷ ÷

Glenn Cota 97132 9/26/97 ÷ ÷ ÷ ÷ ÷

Tom Dickey 97127 7/15/97 ÷ ÷ ÷ ÷ ÷

Dave Eslinger 97133 9/17/97 ÷ ÷ ÷ ÷ ÷

Robert Frouin 97135 9/22/97 ÷ ÷ ÷ ÷ ÷

Robert Green * 8/6/97 ÷ ÷ ÷ ÷ ÷

Watson Gregg * 2/5/97 ÷ ÷ ÷ ÷ ÷

Rick Miller * 12/16/97 ÷ ÷ ÷ ÷ ÷

Greg Mitchell 97130 8/18/97 ÷ ÷ ÷ ÷ ÷ ÷

Jim Mueller 97126 8/18/97 ÷ ÷ ÷ ÷ ÷

Frank Müller-Karger 97128 8/18/97 ÷ ÷ ÷ ÷ ÷

Dave Siegel 97125 7/11/97 ÷ ÷ ÷ ÷ ÷ ÷

Ron Zaneveld 97129 8/18/97 ÷ ÷ ÷ ÷ ÷ ÷

Piotr Flatau 97139 9/22/97 ÷ ÷ ÷ ÷ ÷ ÷

Mark Miller * 9/23/97 ÷ ÷ ÷ ÷ ÷ ÷

John Porter 97136 9/29/97 ÷ ÷ ÷ ÷ ÷

Knut Stamnes 97138 9/15/97 ÷ ÷ ÷ ÷ ÷ ÷ ÷

* Interagency agreement


9

Table 3. Funded Principal Investigators (PI) and collaborators *. Collaborators are funded via EOS(MODIS Ocean Team) and/or SeaWiFS Project.

I. Ocean Bio-optical and Sensor Characterization Studies

K. Arrigo * (NASA)W. Balch (Bigelow Lab.)J. Brock (NOAA)C. Brown (NOAA)D. Capone (U. of Maryland)E. Carpenter (State U. of New York)K. Carder (U. South Florida)F. Chavez (Monterey Bay Aquarium Res. Inst.)D. Clark * (NOAA)G. Cota (Old Dominion U.)T. Dickey (U. of California/Santa Barbara)M. He (Peoples Republic of China, Ocean U. of Qingdao)W. Esaias * (NASA)D. Eslinger (U. of Alaska)R. Evans *(U. of Miami)R. Frouin (Scripps Inst. Oceanography)R. Green (Jet Propulsion Lab.)F. Hoge *(NASA)S. Hooker *(NASA)O. Kopelevich (Soviet Union, P.P. Shirshov Inst. of Oceanology)G. Korotaev (Ukraine, Marine Hydrophysical Inst.)H. Li (Taiwan, Nat. Taiwan Ocean U.)R. Miller (NASA)G. Mitchell (Scripps Inst. of Oceanography)J. Mueller (San Diego State U.)H. Li (Taiwan, Nat. Taiwan Ocean U.)R. Miller (NASA)G. Mitchell (Scripps Inst. of Oceanography)J. Mueller (San Diego State U.)

F. Müller-Karger (U. South Florida) S. Saitoh (Japan, Hokkaido U.) D. Siegel (U. of California/Santa Barbara) R. Zaneveld (Oregon State U.)

II. Data Merger Studies

W. Gregg (NASA) D. Siegel (U. of California/Santa Barbara)

III Atmospheric Correction Studies

P. Flateau (Scripps Inst. of Oceanography)H. Gordon * (U. of Miami)M. Wang (UMD/Baltimore)M. Miller (Brookhaven National Lab.)J. Porter (U. of Hawaii)K. Stamnes (U. of Alaska)


10

1. G

loba

l dis

trib

utio

n of

the

NR

A-s

elec

ted

SIM

BIO

S fi

eld

stud

ies.

Uni

ted

Stat

es: (

1) B

alch

; (2

) B

row

n/B

rock

Car

der;

(5)

Cha

vez;

(6)

Cot

a; (

7) D

icke

y; (

8) E

slin

ger;

(9)

Fro

uin;

(4/

14)

Gre

en;

Hog

e; (

11)

Mill

er; (

12)

Mitc

hell;

(13

) M

ülle

r-K

arge

r; (

14)

Port

er; (

15)

Sieg

el; (

16)

Zan

evel

d/M

uelle

r;K

opel

evic

h; (

19)

Kor

otae

v; (

20)

Li;

and

(21

) Sa

toh


11

Chapter 4

SIMBIOS Project Activities (1997-1998)

Menghua WangUniversity of Maryland Baltimore County

at Goddard Space Flight Center, Greenbelt, Maryland

Brian Franz, Alice Isaacman, Christopher Pietras, Brian Schieber and Paul SmithSAIC General Sciences Corporation


Tom Riley and Gene FeldmanNASA Goddard Space Flight Center

Greenbelt, Maryland

Sean BaileyFutureTech Corporation

Greenbelt, Maryland

The project accomplishments during this firstperiod (1997-1998) under (a) satellite data processing,(b) data product validation, and (c) sensor engineeringand calibration activities are described below.

4.1 SATELLITE DATAPROCESSING

One of the primary goals of the SIMBIOS projectis to develop methods for meaningful comparison andpossible merging of data products from multiple oceancolor missions. Direct comparison of such products iscomplicated by differences in sensor characteristicsand processing algorithms, as well as spatial andtemporal coverage.

4.1.1 MOS

The German Modular Optoelectronic Scanner(MOS) (Zimmermann and Neumann, 1997) is animaging pushbroom Charge-Coupled Device (CCD)spectrometer that was launched in a sun-synchronouspolar orbit in the spring of 1996 on the Indian RemoteSensing (IRS-P3) satellite.

MOS is a technology demonstrator instrumentwith limited geographic coverage capabilities. Withthe successful launch of NASA’s SeaWiFS on August1 of 1997, there are now two ocean color missions inconcurrent operation. Table 1 provides characteristicsof MOS compared with SeaWiFS. Therefore, we havean unprecedented opportunity to compare ocean color

data from two sensors in simultaneous operation ontwo different satellite platforms. We have extensivelystudied the ocean color measurements comparedbetween MOS and SeaWiFS and developed avicarious calibration approach in which the MOSspectral bands can be recalibrated from the SeaWiFSmeasurements, which are considered as “truth,”thereby allowing remotely retrieved ocean colorresults from the two sensors to be meaningfullycompared (Wang and Franz, 1998 and 1999). Thevicarious calibration method has also beensuccessfully applied to Japan’s OCTS data, and willalso be tested with the French POLDER data. In here,we give some summaries about these studies.

The MOS calibrated and geo-located data (L1Bdata) are provided by the MOS project at Institute ofSpace Sensor Technology, Deutshe ForshungsanstaltLuft-und Raumfarhrt (DLR), Berlin, Germany. TheMOS L1B data file is prepared as size of 384 x 384.In the following sub-sections, we give brief discuss ofour efforts in comparing the MOS results withSeaWiFS.

Atmospheric Correction and Destriping

It is well known that the atmospheric correction,which removes more than 90% of sensor-measuredsignals contributed from atmosphere in the visible, isthe key procedure in ocean color imagery dataprocessing. We have implemented the SeaWiFSatmospheric correction algorithm (Gordon and Wang,1994b) into the MOS imagery data processing using


12

the method outlined by Wang (1999). Therefore, aconsistent atmospheric correction method can beapplied to both SeaWiFS and MOS.

The MOS radiance image has along-track stripesdue to variations in the relative response of the in-dividual detectors on the MOS CCD array (total of384 CCD detectors). Therefore, we have developed asimple destriping algorithm and applied it to the MOSradiance imageries. The MOS destriping procedurecan be briefly outlined as follows. First, for each scan(along the detector array) and a given spectral band, fitthe radiance to a least-square cubic polynomial alongthe scan and compute relative gain at each detector.Next, for each detector select the median gain over allscans in the scene to derive the nominal gain factor forthat detector. Finally, the MOS radiance image can berecomputed with the destriping correction using themedian gain factor.

A complicated stray light and striping correctionalgorithm is now available from the MOS project.Some study and tests are needed to see the efficacy ofthis algorithm. We applied the atmospheric correctionto both MOS and SeaWiFS for co-located images andcompared the retrieved ocean and atmospheric opticalproperties. Two MOS-SeaWiFS co-located imagesacquired on January 29, 1998 in the Atlantic ocean(latitude 27° and longitude -32°) and February 28,1998 in the Mediterranean Sea (latitude 38° andlongitude 3°) were first tested. These two scenes,acquired one month apart, differ significantly in theirocean and atmospheric optical properties. In compar-ing the MOS retrieved ocean and atmospheric opticalresults with that of SeaWiFS, however, we found that:

• the MOS retrieved aerosol optical thickness at theNIR band was usually a factor of 2-3 times higherthan that of SeaWiFS;

• the MOS retrieved ε (7,8) which characterizes thespectral variation of aerosol optical properties isunreasonably low; and

• the MOS retrieved normalized water-leavingreflectances [ ρw λ( )]N in the visible are

significantly different from those of SeaWiFS.

Since we are applying an identical atmosphericcorrection process to the two sets of measurements,the large discrepancy in the retrieved results betweenthe two sensors can probably be interpreted as adifference in sensor calibrations. It is thereforenecessary to recalibrate one sensor to the other, toallow for meaningful comparisons of the retrievedocean optical properties.

A Vicarious Intercalibration for MOS

The vicarious intercalibration procedure betweenMOS and SeaWiFS can be described as follows. Weassume that the gain of the MOS 868 nm band is

unchanged because of differences in the orbits ofMOS and SeaWiFS, thereby using the aerosolconcentration from the MOS measurements, and onlyaccept that the aerosol model determined by SeaWiFSis still valid. Next, by using the SeaWiFS retrievedaerosol models we can theoretically predict theatmospheric effects in the MOS imagery. Thewhitecap radiance contribution can be estimated in thesame way as SeaWiFS (Gordon and Wang, 1994a).Finally, using the SeaWiFS retrieved normalize water-leaving reflectance, [ ρw λ( )]N , the water-leaving and

total radiance at the TOA in the MOS imagery can becomputed, and the gain coefficients for the MOSbands can be derived. To reduce the variation of thederived gain coefficients with various scans, multiplescans within the MOS scene can be used to obtaincoefficient data and derive a best fit for the MOS 384detectors. Table 2 provides the derived MOS re-calibration gain coefficients fitted with the least-square cubic polynomial. These coefficients werederived from two MOS data set acquired on January29 and February 28, 1998. We found that the derivedgain coefficients for the MOS bands 1-6 have verysimilar values in the two different cases, indicatingthat they are nearly independent of temporal andspatial variations. The derived gain coefficients forband 7, however, are different in the two cases. Itappears that the MOS 750 nm band performance isrelated to the atmospheric optical conditions and itsgain adjustment is in opposite to other bands (gaincoefficient > 1).

Since the derived MOS 750 nm band re-calibration gain coefficients depend on theatmospheric optical properties, we have modified theatmospheric correction algorithm such that thecorrection can also be operated using the MOS 685and 868 nm bands. Therefore, a consistent re-calibration gain coefficients for the MOS bands 1-6and band 8 can be applied.

Results and Discussions

We applied the derived MOS gain coefficients asin Table 2 to the MOS measured-radiance at the TOA,and retrieved ocean and atmospheric optical propertiesfor comparison with results from the SeaWiFSmeasurements. Figure 1(a) and 1(b) provide examplesof the histogram (%) for the retrieved normalizewater-leaving reflectances (%) for bands 2 from MOSin comparison with SeaWiFS for the case of January29 and February 28, 1998. There are four cases ineach figure: (i) results from the SeaWiFSmeasurements with the bands 7 and 8 used in theatmospheric corrections, (ii) results from the MOS re-calibrated radiances with the bands 7 and 8 used in thecorrections, (iii) same as in (ii) except that the MOSbands 6 and 8 were used in the corrections, and (iv)results from the MOS original radiance data with the


13

bands 7 and 8 used in the corrections. Note that forcases in which the MOS bands 7 and 8 were used forthe atmospheric corrections, two different calibrationgain coefficients were applied for the MOS 750 nmband for cases of January 29 and February 28, 1998.However, when the MOS bands 6 and 8 were used inthe corrections, the MOS band 7 reflectance data weresimply not used, thereby allowing a consistent set ofrecalibration gain coefficients for the MOS bands 1-6and 8 to be applied for both cases. Figures 1(a) and1(b) show that the vicarious calibration improves theagreement significantly.

To further test the efficacy of the vicarious re-calibration approach, we have applied the MOS re-calibration gain coefficients, which were derived from

January 29 and February 28, 1998 data, to a MOSimage acquired on September 24, 1997 (4-5 monthsbefore) at a location of about latitude 45° andlongitude 13° in the Adriatic Sea, and compared theresults to those obtained from a co-located SeaWiFSimage. Again, MOS results were improvedsignificantly with vicarious re-calibration (Wang andFranz, 1999).

We conclude from these studies that, with thevicarious calibration approach, the retrieved resultsfrom different sensors can be meaningfully comparedand possibly merged. More importantly, with thesame procedure, one can re-calibrate satellite sensorsusing in situ ocean and atmospheric optical propertymeasurements.

Table 1. Characteristics of MOS with SeaWiFS.

Instrument MOS SeaWiFSPlatform I RS-P3 OrbView-2Launch date March 21, 96 August 1, 97Altitude (km) 817 705Equatorial crossing time 10:30 AM 12:00 NoonResolution at nadir (km) 0.5 1.1Scan swath (km) 200 2800Time for one orbit (minutes) 101 99Spectral range (nm) 408-1010 412-865Instrument calibration Lamp Solar & Lunar

Table 2. The derived MOS gain coefficients as G λ, i( ) = cn

n= 0

3

∑ λ( ) in , for i = 1-384

MOS

λ (nm)

c0(λλ) c1(λλ) c2(λλ) c3(λλ)

408 0.9029 3.5×10-4 -9.0×10-7 1.4×10-9

443 0.8453 3.8×10-4 -7.0×10-7 6.5×10-10

485 0.8097 3.8×10-4 -5.3×10-7 2.1×10-10

520 0.8693 1.7×10-4 -4.9×10-8 2.8×10-10

570 0.8701 1.8×10-4 2.2×10-7 -4.6×10-10

685 0.9287 7.6×10-4 -2.6×10-6 3.5×10-9

750† 1.3208 -3.5×10-4 -2.9×10-6 7.4×10-9

750‡ 1.2287 5.1×10--4 -5.3×10-6 -3.2×10-10

868 1.0000 0.0 0.0 0.0

† From case of Jan. 29, 1998 ‡ From case of Feb. 28, 1998


14

Figure 1. The histogram (%) of the MOS retrieved normalized water-leaving reflectances (%) with andwithout recalibrations in comparison with the SeaWiFS measurements for 443 nm band for case of (a) January29, 1998 (Atlantic Ocean) and (b) February 28, 1998 in the Mediterranean Sea.

0

5

10

15

20

0 1 2 3 4 5 6

MOS and SeaWiFS (Band 2)

SeaWiFS MOS (bands 7 & 8) MOS (bands 6 & 8) MOS (original)

% o

f Pix

els

[ ρw(2) ]

N (%)

Atlantic Ocean (Jan. 29, 1998)(a)

0

5

10

15

0 1 2 3 4 5

MOS and SeaWiFS (Band 2)

SeaWiFS MOS (bands 7 & 8) MOS (bands 6 & 8) MOS (original)

% o

f Pix

els

[ ρw(2) ]

N (%)

Mediterranean Sea (Feb. 28, 1998)(b)


15

4.1.2 OCTS

The OCTS is an optical radiometer which flew onthe Japanese Advanced Earth Observing Satellite(ADEOS) from August 1996 to June 1997, collecting10-months of global ocean color data (Kawamura,1998). During the ADEOS mission lifetime,approximately 450 GB of real-time, 700m-resolutionOCTS data was collected by the SeaWiFS projectthrough NOAA ground stations at Wallops, Virginiaand Fairbanks, Alaska. The archive consists of 337scenes of the U. S. East Coast and 1311 scenes overAlaska.

Since no standard software was publiclyavailable, the coastal U.S. data set was originallyprocessed and distributed in near-real time usingNASA software developed under rapid prototypingconditions. Over the past year, the SIMBIOS projecthas been engaged in a comprehensive program toenhance the algorithms and software used to processthe OCTS data, with specific attention given toimproving the geolocation and image registration,atmospheric correction algorithms, and vicariouscalibration. In addition, through its close workingrelationship with the OCTS team at the JapaneseSpace Agency (NASDA) Earth Observation ResearchCenter (EORC), the project has collected an archive ofOCTS Level-1B and Level-2 NASDA productscorresponding to times and locations where in situdata is available. This match-up data set has beenused to validate the Level-2, Version-3 products fromNASDA (Shimada, 1998b), as well as the Level-2products generated using SIMBIOS project softwareand algorithms.

Processing to Level-0

The U.S. Coastal data set must first be convertedfrom a raw 10-bit format to a 16-bit Level-0 format.The original processing made use of a NASA-definedLevel-0 format, which was inconsistent with theNASDA Level-0 format (NASDA, 1994). Toeliminate this inconsistency, the SIMBIOS softwarewas enhanced to generate NASDA format, includingheaders containing the definitive spacecraftephemeredes and spacecraft clock timing data. Theentire raw-data archive was then reprocessed toNASDA Level-0 format, and the data set was againmade available to the public. It can be accessed via the"File Request and Staging" function of the SeaWiFSdata processing system at http://falefa.gsfc.nasa.gov/~seawifsd/sdpsdoc/html/main.html.

Processing to Level-1B

In general, the processing of Level-0 data toLevel-1B includes conversion of raw sensor counts tophysical units and the assignment of geolocation

information to each observation (pixel, detector, andband). Several factors in the design of the OCTSinstrument complicate this process. Each OCTS bandis divided into 10 individual detectors, where eachdetector is associated with a separate scan line. Eachdetector has a slightly different and apparently non-linear responsively, which gives rise to horizontalstriping in the OCTS imagery. The basis for theSIMBIOS calibration is the NASDA Version 3preflight calibration and relative detector calibration(Shimada, 1998a). The relative detector calibrationsubstantially reduces the striping effects, but low-levelartifacts are still visible and they can becomesignificant following removal of the large,comparatively smooth atmospheric signal.

The design of the OCTS scan mirror and focalplane results in severe misalignment of the line-of-sitefor different spectral bands viewing at essentially thesame time. To perform atmospheric correction orderive chlorophyll concentration at a given locationrequires that the bands be co-registered to thatlocation.

The SIMBIOS project developed a registrationtechnique in which an idealized focal-plane geometryis defined, and all bands are re-sampled using anearest neighbor approach to match the idealizedground track. This method is similar in concept to theNASDA approach (NASDA, 1994), but it will resultin slightly different re-sampling. Furthermore, whilethe registration process can correct for differences ingeolocation between bands, the re-sampledobservations come from different lines-of-sight withunique sensor-to-ground and sun-to-groundgeometries.

To ensure accurate atmospheric correction, thetrue solar and sensor view angles per band are savedin the Level-1B file for use in Level-2 processing.Before the registration process can begin, however,the observations from each band must be accuratelynavigated. The geolocation algorithm used forSIMBIOS OCTS processing is a modified version ofthe method used for SeaWiFS navigation (Patt andGregg, 1994). This technique has the potential to yieldan exact solution to the geolocation of each scan line,when the spacecraft position and sensor geometry areaccurately known.

Some sources of error include: inexact knowledgeof the instrument design and installed, post-launchsensor and mirror alignments (NASDA, 1994),uncertainties in orbit location or spacecraft attitude,and variations in the scan-rate or sensor tilt.

To improve geolocation accuracy, the SIMBIOSproject derived time-dependent attitude and tiltadjustments using an automated navigationassessment technique based on island targets (Patt etal., 1997). The final navigation is still being assessed,but preliminary indications show that it is accurate toapproximately one kilometer.


16

Processing to Level-2

Within the SIMBIOS project, Level-1B to Level-2 processing includes application of the vicariouscalibration, removal of the atmospheric signal toretrieve the water-leaving radiances, and generation ofderived products such as chlorophyll concentration. Itis most appropriate to apply the vicarious calibrationat this stage, as it must account for deficiencies in theatmospheric correction algorithm, as well as forchanges in the preflight calibration. Since this sameapproach was used by NASDA, the Level-1B productsfrom the Japanese Space Agency and the SIMBIOSproject are calibration equivalent, though theatmospheric correction algorithms and file formatsdiffer.

The SIMBIOS OCTS atmospheric correctionalgorithm is based on the SeaWiFS algorithm ofGordon and Wang (1994a). The SIMBIOS approachhas been to identify those few parts of the algorithmwhich are sensor or band-pass specific, and develop asoftware package which can process data frommultiple ocean color remote sensors. The sensitivity ofthe algorithm to differences in spectral bands wascarefully assessed by Wang (Wang, 1999), wherein heshowed by simulation that these differences can beaccurately accounted for through exact calculation ofthe Rayleigh reflectances and minor modifications tothe diffuse transmittance calculation. The multi-sensorLevel-1B to Level-2 code (MSl12) has been usedsuccessfully to process and intercalibrate MOS andSeaWiFS data (Wang and Franz, 1999), and it is nowbeing employed to process the entire SIMBIOS OCTSarchive.

When processing OCTS data, the MSl12 softwareis able to make use of the band-dependent solar andview angles. If these angles are provided in the Level-1B input file, they will be used in the atmosphericcorrection process when computing the Rayleighradiances, the air-mass calculations that affectestimates of white-cap radiance and diffusetransmittance, and the normalization of the water-leaving radiances.

The atmospheric correction process includesmasking for land, clouds, and saturation and maskingfor stray-light in the vicinity of those bright sources.To mitigate the effects of residual striping or otherforms of systematic or random noise, the MSl12 codeincludes options for various types of statistical, spatialfiltering. For OCTS processing, a 5x5-pixel medianfilter is applied to the Rayleigh-subtracted radiances inthe near-IR bands, based on the assumption that thespatial scale of aerosol variability is expected to begreater than a few kilometers. This smoothingeffectively reduces the tendency for the atmosphericcorrection algorithm to select different aerosol typesfrom pixel-to-pixel. Additional smoothing with a 3x3-

pixel median on the Rayleigh-subtracted radiances isapplied to the visible bands, reducing the effect ofresidual striping at the cost of degraded resolution ofthe in-water spatial structure.

For purposes of vicarious calibration, theSIMBIOS project is fortunate to have time and space-coincident in situ water-leaving radiance spectra fromthe Marine Optical Buoy (MOBY) buoy (Clark et al.,1997) to match a series of OCTS over-flights. Thissame buoy is the basis for the SeaWiFS calibration,providing a convenient bridge to connect the twomission life-spans. Unfortunately, there were someproblems with the early deployment of MOBY,limiting the number of usable calibration samples toseven measurements spanning the period November1996 to February 1997. In addition, MOBYmeasurements are not sufficient to calibrate the near-IR bands at 765 nm and 865 nm, where water-leavingradiances are very close to zero. It was thereforeassumed that the 865 nm vicarious calibration wasequal to the value suggested in the NASDA Version-4calibration (Fukushima, pers. comm.), and the 765 nmband was adjusted to force the atmospheric correctionalgorithm to select, on average, a maritime aerosoltype over the MOBY site. This assumption isconsistent with the typical aerosol-type retrievalsderived from SeaWiFS data in the vicinity of MOBY.The entire vicarious calibration process can beperformed in a single processing step in which theOCTS match-up data set is corrected to retrieve thenormalized water-leaving radiances, the retrievednormalized water-leaving radiances are replaced withthe in situ values, and the atmospheric correctionalgorithm is reversed to derive the theoretical top-of-atmosphere radiances. These theoretical radiances aredivided by the observed radiances to derive correctionfactors for each in situ match-up point, and thoseratios are averaged within each band to derive thevicarious calibration coefficients.

The final calibration coefficients are listed in theTable 3, with the equivalent NASDA Version-4calibration shown for comparison. Considering thatthe two projects use different atmospheric correctionalgorithms ( Fukushima et al., 1998) and different insitu measurements for calibration, the results are ingood agreement.

Table 3.Band NASDA SIMBIOS

1 1.14 1.132 1.03 1.013 0.9394 0.94

4 1.00 1.005 1.04 1.036 1.00 0.997 1.02 0.918 0.89 0.89


17

The largest difference is in the 765 nm band,which NASDA does not use for atmosphericcorrection (Shimada, 1998). Since SIMBIOS doesmake use of the 765 nm band for aerosol modelselection, a correction for oxygen absorption wasnecessarily performed, which accounts for thediscrepancy.

Validation Analyses

An extensive set of in situ data taken during thetime of the OCTS mission was made available to theSIMBIOS project for match-up analysis from theSeaBASS database. Over 600 data points taken on134 separate days exist in this set.

OCTS Satellite data used in the match-up analysiswas obtained from two different sources and twodifferent processing methods were used. Files in theU.S. Coastal archive which matched in situ data pointswere processed from Level 0 to Level 2 using theSIMBIOS algorithms and calibration described above.Level 1B data files supplied to the SIMBIOS Projectby NASDA/EORC were processed to Level 2 usingthe SIMBIOS atmospheric correction algorithms andvicarious calibration. Level 2 data supplied byNASDA/EORC were also matched to in situ data, butthese data were processed using Version 3 of theNASDA calibration and so results obtained are ofhistoric interest only.

Normalized water-leaving radiance (nLw) valueswere extracted from the Level 2 records andacceptable pixels within a 1.05-km radius wereweighted by the inverse of their distance to the match-up site and then averaged to derive nLw values forcomparison to in situ data. This technique usuallyresulted in obtaining data from 9 pixels in a squarecentered about the in situ data point, although theactual number can vary due to the complex scangeometry of the OCTS instrument. Match-up valueswere also calculated for an area within a 2-km radiusif there were not enough acceptable pixels within the1.05-km limit.

The in situ water-leaving radiances werenormalized using theoretical estimates of surfaceirradiance (Es). The Es values were calculated fromthe Solar radiance model of Neckel and Labs (Neckeland Labs, 1984), weighted by the OCTS band passes,and transmitted to the surface using measurements ofthe local ozone concentration, knowledge of theRayleigh optical thickness, and estimates of theaerosol transmittance. The resulting nLw values werethen compared to the weighted average nLw valuesfrom OCTS.

Where possible, the match-ups of satellite to insitu data were judged acceptable by the same criteriaused to judge SeaWiFS match-ups to in situ data.Specifically, match-ups were rejected if:

• The time difference between satellite and in situmeasurements was more than 4 hours (6 hours forchlorophyll).

• Difference in solar zenith angle was more than15º

• In situ data were from a duplicate cast and had alower K490 value than other casts.

• Wavelengths of in situ data differed from OCTSwavelengths by more than 5 nm.

• Satellite data coefficient of determination (theratio of standard deviation to average) was greaterthan 0.5

• The usable number of satellite pixels was lessthan 5, after tests for clouds and sunglint.

For the data processed from NASDA Level 1B toLevel 2 using SIMBIOS methods, 13 match-up pointswere deemed acceptable. An additional 10 match-uppoints were obtained from the Wallops Level 0dataset. Data collected over the MOBY site wereexcluded from this analysis, since they were used toderive the vicarious calibration.

Table 4.

BandNumber ofMatch-ups

nLw ratioOCTS retrieval / in situ

1 18 1.202 18 1.063 19 1.014 13 0.9985 9 1.04

The Table 4 summarizes the match-up resultsbetween in situ water-leaving radiance data (excludingMOBY) and OCTS retrievals processed using theSIMBIOS atmospheric correction algorithms.

Sargasso Sea Validation Study

Due to the small number of usable match-up insitu data in existence for the OCTS mission, additionalvalidation of the SIMBIOS OCTS calibration wassought. A time series of averaged normalized water-leaving radiances derived from 27 scenes centered inthe Sargasso Sea was generated and compared tonominal water-leaving radiances derived from MOBYin situ data and to a time series from one year later inthe Sargasso Sea area created from SeaWiFS data.

The Sargasso Sea area was chosen for the OCTSdata because it yielded the largest number of clear-water scenes within the OCTS data set acquired bythe Wallops receiving station. In addition, six clear-water scenes centered over Bermuda were generatedfrom the same archive and compared to in situ datafrom that area.

The SeaWiFS Sargasso time-series data wastaken from the standard 8-day time-binned dataset, fora 3° square box centered at Latitude 25.5° and


18

Longitude -71.5° from 57 8-day periods betweenNovember 1997 and December 1998. The SeaWiFStime binning procedure averages all acceptable valuesover each time period on a pixel-by-pixel basis. Thenominal clear-water radiances used for comparisonwere derived from MOBY in situ data tuned to theOCTS and SeaWiFS wavelengths respectively. In situchlorophyll and water leaving radiances for theBermuda site were taken from the Bermuda AtlanticTime Series (BATS) conducted by the Center forRemote Sensing and Environmental Optics at theUniversity of California, Santa Barbara (Garver andSiegel, 1997). For each scene in the Sargasso Sea andBermuda area, normalized water-leaving radianceswere averaged, excluding values where chlorophyllwas not within acceptable limits, i.e. data not between.0005 µg/L and .30 µg/L, as well as data ofunacceptable quality, e.g., clouds, etc. The averagedSargasso Sea nLw values were compared to nominalMOBY-derived OCTS nLw values and to SeaWiFSvalues from 1997-1998 while the Bermuda area nLwvalues were compared to nominal MOBY-derivednLw values and to contemporary in situ data from theBATS experiment. For both Sargasso Sea andBermuda area data, chlorophyll values were computedfrom the average nLw using both the OC2 algorithmof O'Reilly et al. (1998) adjusted to the OCTS spectralbands and the standard NASDA OCTS-C algorithm(Kishino et al., 1998). In addition, pigmentconcentrations were computed using the NormalizedDifference Pigment Index (Frouin, personal comm.).

The agreement in the general behavior of theOCTS data with the SeaWiFS data from one year latersuggests that the variability seen in the Sargasso SeanLw values is due to seasonal variation rather thaninstrument instability. The mean nLw values fromOCTS agree reasonably well with MOBY-derivednominal nLw values, suggesting that the SIMBIOScalibration is working well. The three chlorophyllalgorithms used on OCTS data agree in the averagebut the distribution of the chlorophyll derived by theOC2 algorithm is somewhat broader than thatobtained using the NASDA chlorophyll or NDPIpigment algorithm. The results of this study werepresented at the Alps ’99 Conference in Meribel,France in January, 1999

4.2 PRODUCT VALIDATION

The Project is comparing and validatingSeaWiFS satellite measurements with in situatmospheric and bio-optical measurements.

4.2.2 SeaWiFS AOT Comparisons

The theoretical basis of the SeaWiFS aerosolretrieval algorithm are described in both Gordon andWang, (1994b) and Wang et al., (1999). In thissection, we briefly outline our efforts in comparingand validating the SeaWiFS aerosol optical productswith the in situ measurements mainly from the data ofthe Aerosol Robotic Network (AERONET) (Holben etal., 1998). The majority of the instruments within theAERONET network are in continental locations. TheSIMBIOS Project augumented this network withcoastal and island stations. Some other in situmeasurements from calibration and validationcampaigns within the SIMBIOS project are alsoanalyzed. The primary objectives for thesecomparisons are to:

• validate the SeaWiFS aerosol optical thickness(AOT) products, and

• determine the validity of the suite of aerosolmodels currently used by SeaWiFS foratmospheric correction.

In Situ Data Acquisition

The ground-based measurements utilized for theaerosol optical thickness matchup analyses come fromtwo primary sources: automated CIMEL sun/skyscanning radiometer managed as part of theAERONET network, and hand-held MicroTops IIsunphotometers. A select group of the ground stationsfrom the AERONET was chosen. These instrumentswere located at either coastal or island stations andwere operational for a reasonable length of time afterSeaWiFS went into operation. Table 1 provides theAERONET station name, location (latitude andlongitude), and corresponding responsible AERONETprinciple investigator (PI). The hand-held MicroTopsII sunphotometer data, on the other hand, werecollected by various investigators in field campaignsassociated with the SIMBIOS Project. Data from theMicroTops instruments are reprocessed from rawvoltages using code adapted from the AERONETstandard CIMEL processing code. This ensures thatthe τa

data derived from the MicroTops

measurements will be comparable to the data providedby the AERONET.

SeaWiFS Data Acquisition

The SeaWiFS aerosol optical thickness data wereobtained by spatially co-locating a 5x5 pixel grid boxaround the pixel containing the ground-basemeasurement station, thereby providing a maximum25 SeaWiFS retrievals in each matchup. TheSeaWiFS operational code has been modified to


19

output, at a pixel by pixel level, the aerosol opticalthickness at wavelength 865 nm, values of retrievedtwo aerosol models, as well as the model partitionratio ra value. Therefore, aerosol optical thicknesses

at all the SeaWiFS wavelengths can be calculated(Wang et al., 1999).

Data Analyses

SeaWiFS data were obtained by averaging over a5x5 pixel grid box spatially, whereas the τa data from

the CIMEL measurements were derived by a time-weighted averaging during the SeaWiFS overpass (± 1hour). Usually, the CIMEL instruments routinely takeone measurement every 15 minutes. Therefore, for agiven SeaWiFS file there may be as many as eightAERONET measurements that qualify as a match forthe 2-hour time window. The number of hand-heldMicroTops II measurements that match a givenSeaWiFS file, however, varies greatly since themeasurement protocol for these instruments is not yetwell-defined. In general, there should be a minimumof three MicroTops measurements per matchedSeaWiFS file. The ground-based measurements areaveraged after weighting by the time differencebetween the in situ measurement and the SeaWiFSoverpass. Once averaged, the ground basedmeasurements are compared with the SeaWiFSderived values on a band by band basis for eachground station. Due to a limited number ofMicroTops data, these data are grouped together asone “station.”

Preliminary Results

We compared the SeaWiFS derived aerosoloptical thicknesses with those from the ground in situmeasurements. Figure 2 provides an overallcomparison results of τa λ( ) between SeaWiFS and

CIMEL measurements at wavelength 865 nm (870 nmfor CIMEL). The CIMEL measurements were fromthe AERONET stations listed in Table 5. The dottedlines in Figure 2 is the 1:1 line. Though thecomparison results varying both in time and location,it appears that SeaWiFS has tendency ofoverestimating τa λ( ) with respect to the in situ

measurements. Note that, however, since theSeaWiFS band 8 has not been absolutely calibrated onorbit, any error in calibration may contribute to theerror in the τa 865( ) evaluations. On the other hand,

some in situ data are suspected to be erroneous due toinstrument calibration. Obviously, more studies areneeded to understand all of these. We want toemphasize that all results are preliminary.

Similarly, the in situ MicroTops II data, whichwere from the various SIMBIOS calibration andvalidation campaigns, have been compared with theSeaWiFS measurements. Although this work is stillin the initial phase, some results are promising. Table6 shows three sample comparison results from threefield experiments. In these three examples, theSeaWiFS results were almost all underestimated ascompared with the MicroTops II measurements (Ratio< 1), though the three results usually agreedreasonable well.

Figure 2. The retrival SeaWiFS aerosol optical thickness τa λ( ) compared with the ground in situ

measurements from AERONET at 865 nm (870 nm for CIMEL Data).

0

0.1

0.2

0.3

0.4

0.5

0 0.1 0.2 0.3 0.4 0.5

Cim

el τ

a (87

0)

SeaWiFS τa (865)

SeaWiFS (5x5)


20

Table 5. AERONET sites utilized for the aerosol matchup analyses.

AERONET Station Latitude Longitude AERONET PI

Bahrain 26.32 50.50 Charles McClain*Bermuda 32.37 -64.70 Brent HolbenDry Tortugas 24.60 -82.80 Ken Voss/Howard GordonKaashidhoo 4.97 73.47 Brent HolbenLanai 20.83 -156.99 Charles McClain*San Nicolas Island 33.26 -119.49 Robert Frouin

*SIMBIOS Project Office

Table 6. Three samples of MicroTops II data compared with SeaWiFS

Aerosol Optical Thickness τ a λ( )

Jason ProjectG. Feldman

GOCAL97-98J. Mueller

JULNAN98B. Schieber andA. Subramaniam

λ (nm)

SeaWiFS Ratio SeaWiFS Ratio SeaWiFS Ratio440500670865

0.0555 0.76030.0509 0.92040.0395 0.88570.0304 0.9102

0.1635 0.86780.1469 0.97670.1076 1.18110.0774 0.8113

0.1775 0.91680.1538 0.90310.1013 0.92170.0641 0.8456

4.2.3 SeaBASS INTERFACE

The SeaWiFS Bio-optical Archive and StorageSystem (SeaBASS) was created by the SeaWiFSProject to provide a data archive for in situ productsused in scientific analysis. This system has beenexpanded to additionally contain data sets of interestto the SIMBIOS Project.

The in situ data in SeaBASS includemeasurements of water-leaving radiance and otherrelated optical and pigment measurements, from ships,moorings and drifters. Various methods are deployedin the collection of SeaBASS data, including the useof standard profiles, floating radiometers and above-water measurement devices. In addition to the opticaland pigment products listed, measurements of totalsuspended matter (TSM), chromophoric dissolvedorganic matter (CDOM), and other typically non-plankton derived optical components are recorded inSeaBASS.

The products are used by the SIMBIOS Project invalidation of SeaWiFS and other (OCTS, POLDER,etc.) postlaunch imagery and development of newoperational chlorophyll algorithms. Additional usesunder SIMBIOS will likely include field datacalibration comparisons, algorithm workshops, and

time series studies. A detailed description of theoriginal SeaBASS design and scope is provided in theSeaWiFS Technical Report Series, specificallyvolume 20. A current description of the SeaBASSsystem is available via the World Wide Web athttp://seabass.gsfc.nasa.gov.

Data Design

The SeaBASS system contains data from wellover 150 separate experiments with thousands of filesrequiring over 22 megabytes of storage space. Inaddition, the historical optics database holds over300,000 records related to phytoplankton pigmentsand is expected to grow as new pigment data iscollected. Since the SeaBASS system is continuouslygrowing with new data, a method for efficientlyingesting and storing field data was needed. It wasdeemed important that the data ingest be asstraightforward and effortless as possible on the partof the contributing investigators, while still offering auseful and correct data archive for analysis efforts.The following aspects were considered to be the mostimportant in the design of the system:

• Simple data format• Global portability


21

• Data update simplicity• Web accessible data holdings• Support for existing Web standards

To accomplish the above goals, the SeaBASSsystem was designed to support standard ASCII fileswhich can be managed from any computer platform.The files use a simple descriptive header format whichcontains descriptor=value pairs to explain the specifics(“metadata”) of a data file. For example:

/begin_header/affiliations = University_of_Maryland/investigators = John_Creamer/[email protected]/parameters=Ed,Lu,Kd,CHL/north_latitude=59.10[DEG]/south_latitude=58.65[DEG]/east_longitude=-13.11[DEG]/west_longitude=-13.50[DEG]! Comments: data file to be updated…/end_header@

The header or metadata section is intended tofully describe the data in a file. Each header containsinformation on time and location of the in situmeasurements, the investigators involved, valuesmeasured, comments, and other related descriptiveinformation. Frequently, SeaBASS files will containreferences to other files with additional descriptivedetails (e.g., README files).

Following the header information, the actualmeasured optics or pigment data is listed in columnarformat. Typical data files include optical profiles,summary pigment files, or along track measurements.Specific information and examples of the formats canbe referenced at the SeaBASS web site.

Following an series of redesign modifications inthe first year of the SIMBIOS effort, the SeaBASS fileformats have not changed significantly. However,additional changes may need to be made in the nearfuture as new instrumentation and collection methodscontinue to populate the archive, some of which maywarrant variations in the current data archive approach(i.e., hyper-spectral data).

Format Checking

Early in the development of the SeaBASS website it was realized that each data provider can havevery different views on data format or presentation. Inorder to establish a standard format for SeaBASSincoming data files which met the dual goals ofsimplicity and portability, feedback software wasdeveloped.

The primary component of this software is theFCHECK script. FCHECK is written in the PERL

scripting language with connections to look-up tablesand UNIX mail handling utilities. Data providersfrom any platform can test their data for compatibilitywith the SeaBASS format by simply e-mailing theirfile to [email protected]. The FCHECKprogram will receive the incoming mail file from thesystem mail handler, parse the data, and thoroughlycompare it to the established SeaBASS format.FCHECK will then report back to the data provider onthe file’s compliance. This file testing functionhappens automatically with no intervention required,but the SeaBASS administrator is sent a copy of allincoming data files, for review. The FCHECKprogram has been used by many SeaBASS dataproviders as a quick format checker and has savedmany hours of processing time for both the SeaBASSadministrator and contributing investigators.

Once the data provider is sure that their files willwork with the SeaBASS archive, they can send thedata, and related documents, to the SeaBASS archivevia the file transfer protocol (FTP). From there, theSeaBASS administrator will move the files into theirappropriate position in the archive and will initiate adatabase update to register the new data to theSeaBASS online system.

Data Security Issues

A concern of some contributors to SeaBASSinvolves outside user accessibility. Some investigatorshave been concerned that when data is quicklysubmitted to the archive, others may use the datawithout proper notification or acknowledgement. Toaddress these concerns, and to afford continued rapidsubmission of data sets, the SeaBASS web server hasbeen configured as a password protected system.Typically, only SeaWiFS or SIMBIOS authorizedinvestigators who also submit data to the archive areallowed access. Others may be allowed access on acase-by-case basis. Additionally, the web server andSeaBASS software log all user activity. Thisinformation is available to contributing investigators.It is expected that all users who use data from theSeaBASS archive will offer collaboration orauthorship to the data provider.

Specific Matchup Format

To expedite comparisons between in situ data andSeaWiFS (or other satellite) products, a specific fileformat has been developed. This file format, known asthe “matchup” format, follows a similar logic to thatof the other SeaBASS in situ files, that being asimple, easy to implement format in standard ASCII.

However, unlike the general format for SeaBASSdata, the matchup format expects a strict adherence inthe positions and units of data fields. This requirementis to allow for automated comparisons between in situ


22

and satellite data. The format is also used ininteractions with the SeaWiFS/SIMBIOS satellite dataarchive. The matchup format is structured in standardASCII columnar format with the following fields ineach data record:

Year, Month, Day, Hour, Minute, Second, Latitude,Longitude, Lw412, Lw 443, Lw 490, Lw 510, Lw 555,Lw 670, Es412, Es 443, Es 490, Es 510, Es 555, Es 670,Kd490, Chl a

Where the time and location values identify whenand where the measurement was taken. Water-leavingradiance (Lw) and downwelling irradiance (Es) valuesmay be extrapolated from below the surface ormeasured above the surface, depending upon theinstrumentation used. The vertical attenuation ofirradiance at 490 nm (Kd 490) as well as in situmeasured values of chlorophyll a (Chl a) are alsorecorded if available. Missing data values arerepresented by a placeholder. As in the case of otherSeaBASS submitted data, the matchup file alsocontains a metadata section to describe the enclosed insitu data.

Data Contributions

The SIMBIOS project funds numerousinvestigators in order to obtain in situ optical andpigment data for Case I (optically characterized solely

by phytoplankton) water characterization. Additionalinvestigators are also supported to develop newalgorithms or scientific approaches in accordance withthe goals of the SIMBIOS Project. As describedabove, the data and information provided are used forboth internal analysis and satellite validation and topopulate the SeaBASS archive for future activities andcommunity use. Table 7 summarizes the data setswhich have been received from investigators duringthe first year of funding. As shown in the table, themajority of funded principal investigators providefield data for the archive. However, some additionalinvestigators were funded solely to develop algorithmsor analysis techniques.

Future Plans

The SIMBIOS project has created an environmentby which in situ data may be archived, accessed andanalyzed to improve satellite retrievals throughscientific data validation. In its first year of operation,the Project has implemented software and formatstandards which should greatly aid in future satelliteefforts. Additionally, the Project’s focus on thoroughdata documentation should aid in both current andfuture research activities. Subsequent years will focuson improving the usefulness of in situ data bycontinuing to improve the data ingest and dataanalysis approach.

Table 7. SIMBIOS supported data submitted to SeaBASS

PrincipleInvestigator

Universityor Laboratory

OpticsData

PigmentData

DocumentationProvided

AlgorithmDevelopment

William Balch Bigelow Labs ÷ ÷ ÷

J. Brock / C. Brown USGS / NOAA ÷ ÷ ÷

Douglas Capone U. MD ÷ ÷ ÷

Ken Carder U. South Florida ÷ ÷ ÷

Francisco Chavez Mont. Bay ARI ÷ ÷

Glenn Cota ODU ÷ ÷ ÷

Tom Dickey UC SantaBarbara ÷ ÷

Dave Eslinger U. AK, NOAA/CSC ÷ ÷ ÷

Piotr Flatau Scripps, UCSD ÷ ÷

Robert Frouin UCSD / Scripps ÷ ÷

Robert Green JPL ÷ ÷

Watson Gregg GSFC ÷ ÷

Mark Miller BNL ÷ ÷

Rick Miller Stennis Space Ctr. ÷

Greg Mitchell Scripps, UCSD ÷ ÷ ÷ ÷

Jim Mueller San Diego St. U. ÷ ÷ ÷

Frank Müller-Karger U. South Florida ÷ ÷ ÷

John Porter U. Hawaii ÷ ÷

Dave Siegel UC Santa Barbara ÷ ÷ ÷ ÷

Knut Stamnes U. Alaska ÷ ÷ ÷

Ron Zaneveld OSU ÷ ÷ ÷


23

4.2.4 Support Services

In an effort to improve the quality and quantity ofcalibration and validation data sets, the SIMBIOSProject offers several support services to fieldinvestigators. These services include; scheduling ofon-board Local Area Coverage (LAC) recording forSeaWiFS; overflight predictions for operationalsensors (currently SeaWiFS, MOS-B and OCI); nearreal time SeaWiFS imagery for cruise locations; andoptical instrumentation from a pool of investigator-and project-owned instruments. These services maybe requested via the World Wide Web athttp://simbios.gsfc.nasa.gov.

In return for these services, the SIMBIOS Projectrequests that the field investigators provide in situvalidation data to the Project's bio-optical archive,SeaBASS. Since October of 1997, when theseservices were initially offered, the SIMBIOS Projecthas supported 81 cruises (Table 8).

SeaWiFS Scheduling

Since much of the worlds oceans are not coveredby a SeaWiFS High Resolution Picture Transmission(HRPT) station, high-resolution data may be recordedonboard the SeaWiFS sensor. As a service to thescience community, the SIMBIOS project inconjunction with the SeaWiFS project can scheduleSeaWiFS onboard LAC for cruises that occur outsideHRPT coverage. SeaWiFS has the ability to record amaximum of 10 minutes of high-resolution data perdownlink. Typically, a 30-second interval is allottedfor LAC target, which corresponds to 180 scan linesor approximately 200 km along track at nadir.Detailed information on LAC scheduling is availableon the SIMBIOS web site.

Overflight Predictions

For calibration and validation purposes, in situmeasurements should be made as close to the sensoroverflight time as is possible. To aid investigators indetermining when sampling should occur, theSIMBIOS project offers overflight predictions for alloperational ocean color remote sensors. Currently,the sensors supported are SeaWiFS and MOS-B.With the successful launch of ROCSAT-1 on 26January 1999, OCI is being added to this list. Detailedinformation on overflight predictions is available onthe SIMBIOS web site.

Near-Real Time SeaWiFS Imagery

In addition to providing predictions for satelliteoverflight times, the SIMBIOS project offers near realtime imagery of the operations SeaWiFS products inJPEG format to cruises at sea. These images providefield investigators with additional information withwhich they may maximize in situ sampling oftransient oceanographic features. The defaultspecifications for the images provided include:

• Available LAC, HRPT, and Global AreaCoverage (GAC);

• Chlorophyll-a and pseudo-true color images;• 2-degree box about a designated location or the

entire designated region;• Image Width: 600 pixels; and• Minimum percent valid chlorophyll pixels: 5% .

Images may be customized to best accommodateindividual investigators needs. Detailed informationon near real time imagery is available on theSIMBIOS web site.

Table 8. Table of SIMBIOS supported cruises with services provided.

Cruise Location Begin Date End Date

On

-bo

ard

LA

C

Ove

r-fl

igh

tP

red

icti

on

Nea

r R

eal T

ime

Imag

ery

Inst

rum

ent

Po

ol

Principal InvestigatorChesapeake Bay 09/28/1997 11/23/1997 ÷ Curt Davis

Black Sea 10/03/1997 10/13/1997 ÷ Oleg Kopelevich

Alaska 10/10/1997 10/17/1997 ÷ Glenn Cota

Gulf Of Maine 11/10/1997 11/21/1997 ÷ ÷ William Balch

Taiwan 11/23/1997 11/25/1997 ÷ ÷ Hsien-wen Li

Equatorial Pacific 09/27/1997 12/16/1997 ÷ ÷ Francisco Chavez

Ross Sea 10/01/1997 12/14/1997 ÷ ÷ ÷ Greg Mitchell

Equatorial Atlantic 11/14/1997 12/12/1997 ÷ ÷ ÷ C. Menkes

HOTS 12/01/1997 12/21/1997 ÷ ÷ John Porter


24


On

-bo

ard

LA

C

Ove

r-fl

igh

tP

red

icti

on

Nea

r R

eal T

ime

Imag

ery

Inst

rum

ent

Po

ol

Principal InvestigatorJamstec / Kaiyo 12/07/1997 12/26/1997 ÷ ÷ ÷ Marlon Lewis

Southern Polar Front 10/19/1997 02/15/1998 ÷ ÷ Greg Mitchell

HOTS Cruise 01/09/1998 01/12/1998 ÷ ÷ John Porter

Jamstec / Mirai 01/31/1998 02/11/1998 ÷ ÷ ÷ Marlon Lewis

SeaWiFS Initialization Cruise 01/26/1998 02/12/1998 ÷ ÷ ÷ ÷ Dennis Clark

Florida Bay 01/25/1998 02/07/1998 ÷ ÷ Eurico D'Sa

Antarctic 02/10/1998 03/18/1998 ÷ ÷ ÷ Greg Mitchell

Azores / Canary Islands 03/02/1998 03/27/1998 ÷ ÷ Marcel Wernand

Gulf of California 03/05/1998 03/16/1998 ÷ ÷ ÷ ÷ Jim Mueller

Gulf Of Maine 03/14/1998 03/27/1998 ÷ ÷ ÷ Bruce Monger

Indian Ocean 02/28/1998 03/15/1998 ÷ ÷ John Morrison

East China Sea 03/16/1998 03/27/1998 ÷ ÷ ÷ Gwo-Ching Gong

Bahamas 04/1/1998 04/07/1998 ÷ ÷ ÷ Robert Steward

HOTS 04/14/1998 04/16/1998 ÷ ÷ Craig Motell

Sicily Channel 03/27/1998 04/20/1998 ÷ ÷ ÷ M. Ribera d'Alcala'

California Coast 04/02/1998 04/24/1998 ÷ ÷ B.G. Mitchell

Yellow Sea 04/03/1998 04/23/1998 ÷ ÷ ÷ Sonia Gallegos

Sea Of Cortez 04/17/1998 04/26/1998 ÷ ÷ ÷ Gene Feldman

SW Pacific 03/24/1998 04/30/1998 ÷ ÷ ÷ ÷ Ajit Subramaniam

North Sea 04/6/1998 05/01/1998 ÷ ÷ ÷ Marcel Wernand

Atlantic Meridional Transect 04/6/1998 05/01/1998 ÷ ÷ ÷ Stan Hooker

Equatorial Pacific 04/26/1998 05/2/1998 ÷ ÷ ÷ Peter Strutton

Subtropical North Pacific 04/21/1998 05/06/1998 ÷ ÷ ÷ Jeffrey Polovina

HOTS 05/11/1998 05/15/1998 ÷ ÷ John Porter

Atlantic Ocean - Baltic Sea 05/01/1998 05/18/1998 ÷ ÷ Roland Doerffer

Gulf Of Cadiz 05/09/1998 05/19/1998 ÷ ÷ Andreas Neumann

Baltic Sea 05/26/1998 06/03/1998 ÷ Peter Land


Gulf Of Maine 06/01/1998 06/15/1998 ÷ ÷ ÷ Bruce Bowler

Equatorial Pacific 06/6/1998 06/16/1998 ÷ ÷ ÷ ÷ Peter Strutton

Massachusetts Bay 07/05/1998 07/10/1998 ÷ ÷ ÷ Ajit Subramaniam

Labrador Sea 06/22/1998 07/10/1998 ÷ ÷ ÷ Glenn Cota


South China Sea 06/28/1998 07/03/1998 ÷ ÷ Hsien-Wen Li

Gulf Of Mexico 06/27/1998 07/03/1998 ÷ Andrew Thomas

Arctic Ocean 06/18/1998 07/20/1998 ÷ ÷ Dariusz Stramski

Resolute Bay 07/28/1998 08/25/1998 ÷ ÷ ÷ Glenn Cota

Equatorial Atlantic 07/15/1998 08/15/1998 ÷ ÷ ÷ C. Menkes

Long Island Sound 07/20/1998 07/30/1998 ÷ S. Papanikolaou

Puget Sound 06/04/1998 06/25/1998 ÷ ÷ ÷ Ron Zaneveld

Puget Sound 07/29/1998 08/22/1998 ÷ ÷ ÷ Ron Zaneveld

Arctic Ocean 08/05/1998 08/07/1998 ÷ ÷ Sergey Shirshov

Puget Sound 07/24/1998 08/22/1998 ÷ ÷ ÷ M.J. Perry

Taiwan Straits 08/10/1998 08/20/1998 ÷ Ming-Xia HE


25


On

-bo

ard

LA

C

Ove

r-fl

igh

tP

red

icti

on

Nea

r R

eal T

ime

Imag

ery

Inst

rum

ent

Po

ol

Principal InvestigatorHOTS 08/08/1998 08/12/1998 ÷ John Porter

Gulf of Maine 08/14/1998 08/18/1998 ÷ ÷ ÷ Bruce Bowler

SE Bering Sea 08/15/1998 09/09/1998 ÷ ÷ ÷ Stephan Zeeman

South West U.S. 08/17/1998 09/25/1998 ÷ ÷ ÷ Edward Zalewski

South China Sea 09/11/1998 09/27/1998 ÷ ÷ ÷ Rick Miller

California Coast 09/13/1998 10/02/1998 ÷ ÷ Gregg Mitchell

North Pacific 08/30/1998 09/21/1998 ÷ ÷ ÷ ÷ Ron Zaneveld

HOTS 09/26/1998 09/30/1998 ÷ ÷ John Porter

North Pacific 10/01/1998 10/12/1998 ÷ ÷ ÷ Carrie Leonard


Barents Sea 9/15/1998 10/10/1998 ÷ ÷ Oleg Kopelevich

U.S. East Coast 10/5/1998 10/13/1998 ÷ ÷ Stephan Howden

Gulf Of Maine 9/10/1998 10/31/1998 ÷ ÷ ÷ ÷ Bruce Bowler

HOTS 10/18/1998 10/20/1998 ÷ John Porter


Equatorial Pacific 10/23/1998 11/09/1998 ÷ ÷ ÷ Peter Strutton

HOTS 11/10/1998 11/12/1998 ÷ ÷ John Porter

South Atlantic Bight 10/27/1998 11/24/1998 ÷ ÷ ÷ Ajit Subramaniam

Gulf of California 11/25/1998 12/09/1998 ÷ ÷ ÷ ÷ Andrew Barnard

Gulf of Maine 12/02/1998 12/18/1998 ÷ ÷ Heidi Sosik

HOTS 12/13/1998 12/16/1998 ÷ ÷ John Porter

Indian Ocean 01/04/1999 02/23/1999 ÷ ÷ ÷ J. Le Fevre

Indian Ocean - INDOEX 01/14/1999 04/01/1999 ÷ ÷ ÷ Piotr Flatau

Southern Ocean 01/01/1999 04/01/1999 ÷ ÷ Steve Groom

California Coast 01/09/1999 01/31/1999 ÷ ÷ Greg Mitchell

Field Instrument Pool

The SIMBIOS project provided funding toseveral of the science team members for the purchaseof in situ ocean optical instrumentation. The fundingwas provided with the stipulation that theseinstruments would be made available for three years toan instrument pool to be maintained by the ProjectOffice. The Project augmented this instrument poolwith atmospheric instrumentation. Table 10summarizes the instruments available.

4.3 SENSOR ENGINEERING ANDCALIBRATION

AERONET is a network of ground-basedautomated sun photometers owned by nationalagencies and universities. AERONET data providesglobally distributed, near-real time observations of

aerosol spectral optical depths, aerosol sizedistributions, and precipitable water. These data allowfor algorithm validation of satellite aerosol retrievalsas well as characterization of aerosol properties. Suchvalidation and characterization are critical foratmospheric correction of ocean color sensors. Themajority of the instruments within the AERONETnetwork are in continental locations. The CIMEL SunPhotometer manufactured in France is the instrumentused for AERONET.

The SIMBIOS Project augmented his instrumentpool with atmospheric instruments by purchasing 12MicroTops hand-held sun photometers, 2 PREDE(Japanese built) sun photometers, one micro-pulselidar and the 12 CE318 CIMEL sun photometers.

Specifically, the 12 CIMEL instruments augmentthe AERONET network with coastal and islandstations. These CIMEL instruments better withstandthe corrosive marine environment after undergoing arobust re-engineering. These “hardened” CIMEL


26

instruments are placed, with other instruments, at sitesnear marine environments. Confirmed (delivered ornegotiated) CIMEL sites include Lanai-Hawaii (with abackup in Honolulu), Ascension Island, Bahraine,Tahiti, Wallops Island (Virginia), South Korea, andTurkey (Black Sea).

CIMEL Re-engineering

In its normal configuration, the CIMEL SunPhotometer is not suitable for use in a corrosivemarine environment. Meridian Engineering ofHanover (Maryland) under Tom Riley’s guidancehardened these instruments for placement in locationswhere they may be subject to salt spray. Themodifications include:

• Replacement of the base housing — The newbase housing is welded and has more weatherresistant connectors.

• Replace motor housing — The new housing iswelded.

• Replace instrument housing — Again the newhousing is welded.

• Cable and connectors — Connectors werereplaced with military style connectors. Cablesreplaced with ones rated for this environment.The cables were then sealed with both 3M 2242Linerless Electrical Rubber Splicing Tape andthen 3M Scotch Super 88 Vinyl Electrical Tape.

• Hardened moisture sensor — The moisture sensorcable and connector were replaced.

The following devices were added to allowremote operation:

• Electronic boxes — The two electrical boxes(Control Box and Power Box) are now sealedplastic NEMA boxes and are painted white.

• Photovoltaic panel mount — The twophotovoltaic panels and the moisture sensor aremounted on a fiberglass panel that is held to thepower box with stainless steel hardware.

• Sun Shield — The Control Box also has a fiberglass sun shield held one inch above the box andhas stainless steel hardware.

• Mounting Plate — A heavy galvanized steelmounting plate is now available.

Sun Photometer Calibration:Non-Polarized Channels

The calibration consists of using a sun photometercalibrated in high altitude conditions (Mauna Loa,Hawaii) and transferring the calibration to the othersun photometers by taking simultaneous sun directmeasurements from ground during a clear day. The

top of atmosphere (TOA) signals (V0) of the non-calibrated sun photometer are deduced from those ofthe CIMEL sun photometer. This method is currentlyapplied to all the CIMEL sun photometers involved inthe AERONET network (Holben et al., 1998).

The processing steps for analyzing the calibrationmeasurements performed by a sun photometer aredescribed below and summarized in Diagram 1:

• Download the raw signal of the uncalibrated sunphotometer [Vu (i)] and of the calibrated sunphotometer [Vr (i)].

• Select the simultaneous measurements (timedifference lower than 36 seconds).

• Compute for each channel, the TOA signals forthe un-calibrated sun photometer using the TOAsignals of the calibrated sun photometer (obtainedat Mauna Loa).

• The TOA signals, computed for eachsimultaneous event, are averaged

The comparison with the aerosol optical thickness(AOT) retrieved from the calibrated sun photometerallows for validation of the protocol. AOT areretrieved using the same steps as in Diagram 2 exceptfor the Ozone retrieval. Total Ozone MappingSpectrometer (TOMS) measurements are used for thecomputation of ozone contribution.

Currently, eight MicroTops, one SIMBAD andtwo PREDE sun photometers have been successfullytransfer calibrated according to the inter-calibrationmethod. The TOA signals of the MicroTops #3773,SIMBAD #972396 and PREDE #PS090063 aresummarized during several transfers of calibrationbetween August and December 1998 in Table 1.Means and standard deviation are presented for eachchannel of each instrument. The standard deviation ofthe TOA signal is overall lower than 2%. The resultsare especially good for the MicroTops and SIMBADsun photometers, considering the long period used, thedesign (i.e., hand held) and affordable cost. Theaccuracy of the CIMEL sun photometer TOA signals(calibrated at Mauna Loa and used for our calibrationtransfer) is 0.5 % for the channels between 440 and1020nm (Holben et al., 1998). The accuracy of thecalibration transfer for the PREDE PS090063 is onlypresented on Table 11. Another PREDE (#PS090064)instrument, used between October and December1998, is still under study.

Sun Photometer Calibration:Polarized Channels

The polarized version of the CIMEL sunphotometers is composed of three channels at 870nmmounted on the filter wheel. Three polarizing sheetsare located in front of the filters and are respectivelyadjusted according to 0, -60 and +60 degrees.


27

The polarization rate of the light is deduced from threemeasurements in these three channels.

The calibration of the polarized version of theCIMEL sun photometer is performed using acalibration device designed by the Laboratoired’Optique Atmosphérique. The calibration device islocated ahead of a non-polarized source and allows thepolarization of the light in the output of the device.The reading of the angular position of the twoSF11windows mounted inside the device allows forthe computation of the polarization rate of the light.The calibration consists of plotting the computedpolarization rate versus the measured one anddetermining the slope and the intercept of the trendline.

The calibration have been performed on sixpolarized instruments belonged to the AERONETgroup. The results are summarized in Table 12. Thesame calibration is scheduled for the SIMBIOSpolarized CIMEL sun photometers. The calibration isperformed for a polarization rate ranging between 0 to65%, because of the dimensions of the device. Thelinear squared fit determined for the calibration of thesix CIMEL sun photometers ranges between 0.9993 to1.

Sun Photometer Data Processing

The data processing is applied to all SIMBIOSsun photometers (i.e., CIMEL, PREDE, MicroTops IIand SIMBAD). The data processing principle isdescribed below and summarized in Diagram 2:

1. Download the raw signal Vi corresponding to themeasurements performed in the channel i.

2. Compute the ratio between the raw signal and thetop of Atmosphere (TOA) signal Vi0 provided bythe calibration of the sun photometer, Vi/Vi0.

3. Compute the ratio between the averaged Sun-Earth distance and the Sun-Earth distance of theday, (d0/d).

4. Compute the air mass m, according to the day, thedate, the time and the location when/where themeasurements were taken.

5. Compute the total optical thickness in the channeli, τtot(i).

6. According to the location, compute and removefrom the total optical thickness the Rayleigh andthe ozone optical depth for each channel, τR(i)and τO(i). The same equation and Dobson table isused for the data processing of all the sunphotometers.

7. Compute the aerosol optical thickness (AOT),τA(i).

Several AOT measurements have been performedon the roof of a building at GSFC (from August toDecember 1998), from cruises during several

campaigns and from ground, allowed performancecomparisons with SeaWiFS sensor (Wang et al.,1999).

Calibration Round Robin

NASA personnel carried out the first SeaWiFS-SIMBIOS Intercalibration Round-Robin Experiment(SIRREX-6) from August 1997 to February 1998.SIRREX-6 was performed in a completely differentmanner from SIRREX-1 through SIRREX-5. In thosetests, laboratory references were brought to a centrallocation and tested against one another. In SIRREX-6,the same four common field instruments (e.g.,Satlantic in-water radiometers) were taken to nineseparate laboratories and tested using the laboratories’standards and procedures. Two of the sensors wereseven-channel radiance heads and two were seven-channel irradiance heads. The calibration and datareductions procedures used at each site followed thatlaboratories’ normal procedures. The reference lampsnormally used for the calibration of these types ofinstruments by the various laboratories were also usedfor this experiment. NASA personnel processed thedata to produce calibration parameters from thevarious laboratories for comparison. These testsshowed an overall agreement at better than the +/-2%level. Test equipment, calibration procedures, datareduction, and specific handling procedures have beenpublished in the Riley and Bailey (1998).

SXR-II

The first SeaWiFS Transfer Radiometer (SXR)was built for the SeaWiFS Project to verify andcompare measurements of spectral radiance at sixdiscrete wave-lengths in the visible and near infraredJohnson et al. (1998). In addition SXR is used tocompare these sources to standards of spectralradiance maintained at the National Institute ofStandards (NIST). SIMBIOS project had a secondcopy of the SeaWiFS Transfer Radiometer (SXR-II)built for use in the calibration round robin. This unitwill supplement the first unit and is designed foreasier travel. The SXR-II is currently beingcharacterized and calibrated. It will be used in futureSIRREX round robin experiments.

Additional SQMs

NASA developed a portable optical calibrationsource called the SeaWiFS Quality Monitor (SQM)with NIST. Last year, NASA provided limitedfinancial support for the development of the SQM as acommercial product. Two companies elected to takeadvantage of the opportunity, Satlantic Instrumentsand Yankee Environmental Systems, Inc. TheSatlantic version was evaluated by Dr. Stan Hooker.


28

The Yankee Environmental Systems, OpticalCalibration Source prototype was evaluated at GSFCby Tom Riley. Their computer interface is asignificant improvement over the earlier models. Anumber of small problems were uncovered andreported back to the manufacter who has takencorrective action.

4.4 SIMBIOS COMPUTINGRESOURCES

The SIMBIOS computing facility contains manydifferent types of equipment (Diagram 3). The maingoal of the facility is to provide support for thefollowing areas: 1) routine bulk data processing, 2)large scale analysis and match-up between sensors, 3)focused analysis on data subsets, and 4) storage oflarge sensor data sets.

To support these efforts, the facility has grown toinclude two large server systems, a 6 terabyte tapelibrary, four desktop workstations, and numerouspersonal computers and printers. Co-location of thefacility with the SeaWiFS Project has also madeavailable other resources such as high-quality colorprinters, network equipment, and other supportingequipment. Large scale tasks such as bulk dataprocessing and large scale analysis tasks areperformed by two Silicon Graphics (SGI) Origin 2000servers. To prepare for increased workload, the

systems were upgraded to 6 processors each, andapproximately 3 gigabytes (GB) of RAM. The influxof large data sets for additional sensors requiresincreasing amounts of space, especially whenperforming comparative analysis. The Data Processingserver has been increased to 98 GB of disk, and theData Analysis server has over 372 GB of total diskspace. Long term storage of data sets is handled bythe 6 terabyte (TB) tape library attached to the DataAnalysis server.

Silicon Graphics O2 workstations are used forsoftware development and more focused analysistasks. The sizing of the workstations varies with thetask requirements. One workstation has 320megabytes (MB) of RAM and 28 GB of disk space, asecond workstation has 640 MB of RAM combinedwith 72 GB of disk space. The third workstation has640 MB of RAM and 50 GB of disk space.Acquisition of a fourth workstation is planned for thisfiscal year.

SIMBIOS is connected to the Internet via a 100megabit link, making network transfer of large datasets more feasible. Internal connections are also 100megabit for rapid transfer of data between the serversand workstations. Data ingest for SIMBIOS is alsodone via removable media. The Project can read awide variety of removable media including CD-ROM,8 millimeter tape, 4 millimeter (DAT) tape, and DLTmedia up through DLT 7000.

Table 10. SIMBIOS Pool Instruments

Instrument Quantity Description Manufacturer Custodian(s)

MicroTops IISunphotometerw/GPS

12 5 channel handheld sun photometerw/Garmin GPS-38

Solar LightCompany

SIMBIOS Project

HISTAR Package 12 hyper-spectral absorption/beam attenuationmeters mounted on a cage with a SeaBirdCTD

WETLabs, Inc Zaneveld/Pegau

Hydroscat 6 3 Backscattering meter HOBILabsSiegelCarderMitchell

Pure Water System 3Water purification system for calibration ofWETLabs AC-9 and HISTAR absorption andattenuation meters

Barnstead Zaneveld/Pegau

AC-9 3 Absorption/beam attenuation meter WETLabs, Inc.CotaCaponeMuller-Karger

SIMBAD 2 5 channel radiometer and sun photometer Frouin

SeaWiFSMultichannelProfiling Radiometer

2

Free-fall profiling radiometer measuring Lu(z),Ed(z), Es, Lu(1m) with cables, deck unit andPC with data acquisition and processingsoftware

Satlantic, Inc. CaponeChavez

Micropulse LIDAR 1 Continuous operation LIDAR system SESI SIMBIOS ProjectPrede SunPhotometer 2 Automated ship-board sun photometer Prede SIMBIOS Project

CE 318CIMEL SunPhotometer

12 Marine-hardened, automated sun photometer Cimel SIMBIOS Project


29

Table 11. Top of Atmosphere signals (V0) of MicroTops, Simbad and Prede sun photometers determined bytransfer calibration from a calibrated CIMEL at GSFC between August and December 1998.

MicroTops #3773 440nm 500nm 675nm 870nm 940nm08/20/98 – Cimel # 37 1236.95±0.78 980.63±1.37 1214.27±0.98 821.80±1.12 1419.70±1.2208/21/98 – Cimel # 37 1243.64±1.92 989.01±1.48 1220.24±0.86 825.59±0.76 1420.55±1.8010/16/98 – Cimel # 27 1216.46±0.91 970.49±0.89 1187.49±0.77 822.38±0.58 1400.10±0.5511/24/98 – Cimel # 27 1219.75±0.45 976.43±0.46 1191.42±0.34 821.64±0.61 1401.36±0.75MEAN 1229.20 979.14 1203.35 822.85 1410.43STD % 1.07 0.79 1.35 0.22 0.79Simbad #97230608/20/98 – Cimel # 37 393564.40±1.89 482820.40±1.07 403245.16±0.35 420247.88±0.84 307582.92±0.5908/21/98 – Cimel # 37 393525.05±3.13 481855.71±1.02 407705.35±0.41 423453.93±1.03 305498.45±0.5610/16/98 – Cimel # 27 381894.63±1.25 469676.76±0.99 393682.49±1.82 409544.15±0.56 306692.23±0.9811/24/98 – Cimel # 27 392974.50±0.89 477633.99±0.97 409380.37±1.16 412874.93±0.41 307275.49±0.1712/14/98 – Cimel # 101 386234.51±0.95 471653.55±0.37 393092.41±0.21 409462.25±0.51 310487.83±0.22MEAN 389638.62 476728.08 401421.16 415116.63 307507.39STD % 1.36 1.24 1.91 1.54 0.60Prede #PS09006308/20/98 – Cimel # 37 1.52E-04±3.38 2.86E-04±1.43 3.59E-04±2.22 2.81E-04±0.36 2.84E-04±2.6308/21/98 – Cimel # 37 1.48E-04±1.13 2.84E-04±1.45 3.57E-04±1.75 2.81E-04±0.45 2.92E-04±2.30MEAN 1.50E-04 2.82E-04 3.58E-04 2.81E-04 2.88E-04STD % 1.48 0.51 0.27 0.11 1.87

Table 12: Slope (δp), intercept (p0) and linear squared fit applied to the polarization rate given by the deviceand plotted versus the polarization rate measured by six polarized CIMEL sun photometers.

Instrument # δp p0 R2

92 1.1088 -0.1051 0.999625 1.2256 -0.0523 0.9977111 1.0791 0.0476 145 1.0608x 0.2354 0.999843 1.1935x - 0.7816 114 1.1665 -11.179 0.9993


30

Diagram 1. Sun photometer transfer calibration

Calibration InputTOA signal of the calibrated sun photometer

Vr0(i)

TOA computation for the uncalibrated signalVu0(i)=Vr0(i)*(Vu(i)/Vr(i))

for the selected simultaneous measurementsTime difference<36 seconds

Sun photometer data processingapplied to all measurements

(see Diagram 2except ozone computation)

Ozone content retrievedusing TOMS sensor

according to day and location

Aerosol optical thickness computationττA(i)

Raw signals (in V) or (in A) in the channel iVu(i) for the uncalibrated sun photometer

Vr(i) for the calibrated sun photometerDate, Time, Longitude, Latitude

Diagram 2. Sun photometer data processing

Calibration InputTop of Atmosphere signal Vi0

Earth-Sun distance computationaccording to the averaged distance

d0/d(Iqbal M., 1983)

airmass computationm

(Kasten et al, 1989)

Total optical thickness computationττtot(i)=Ln[(Vi/Vi0)/(d0/d)^2]/m

Rayleigh optical thickness computationττR(i)

(Penndorf R., 1957)

Ozone optical thickness computationττO(i)

Lookup table by 5degrees of latitude(London et al., 1976)

Rayleigh and Ozone correctionτΑ(ι)=ττΑ(ι)=τtot(i)-ττR(i)-ττO(i)

Aerosol optical thickness computationττA(i)

Raw signals (in V) or (in A) in the channel i: ViDate, Time, Longitude, Latitude


31

Diagram 3. SIMBIOS Analysis/Data Processing System


32

Chapter 5

Validation of Surface Bio-optical Propertiesin the Gulf of Maine as a Means for Improving

Satellite Primary Production Estimates

William M. BalchBigelow Laboratory for Ocean Sciences

West Boothbay Harbor, Maine

5.1 INTRODUCTION

There is a strong need to have sea-truth opticaldata in order to evaluate remotely-sensed ocean color.Our SIMBIOS contract is to use the M/S ScotianPrince ferry as a ship of opportunity, running betweenPortland, Maine and Yarmouth (NV). While the Gulfof Maine is cloudy and foggy more than it is clear; theclimatology shows on average, about 1 in 4 days iscompletely clear and clear days are slightly more fre-quent in the late summer and early fall months. Aship of opportunity program (where one has choice ofthe days of sampling) provides much better flexibilityto sample during clear periods with good satellite cov-erage. Measurements include continuous, surface,along-track fluorescence, light scattering, absorption,beam attenuation, above-water remote sensing reflec-tance, calcite-dependent light scattering, temperature,and salinity. Expendable Bathythermograph (XBT)drops allow acquisition of vertical temperature infor-mation, useful for defining isopycnal slope, whichaffects primary production. These data are compara-ble to a previous program from early 1982, where aShip Of Opportunity Program (SOOP) was run on thetruck ferry, M/V Marine Evangeline, which ran alongthe same transect (Boyd, 1985). These surface datawere combined with satellite-derived sea surface tem-perature fields to examine the Maine coastal current(Bisagni et al., 1995). Unfortunately, this programstopped in 1982. The ongoing SIMBIOS results willdovetail nicely with the previous work (which alsohad CZCS coverage) for looking at any long-termchanges in the Gulf of Maine hydrography or bio-optics.

The Gulf of Maine is an ideal site for gatheringsea-truth optical data. It is a semi-enclosed basin withan average depth of about 150m. Isopycnal slope (andproductivity) appears to be strongly topographicallydriven, with the water column becoming isothermal atthe 60m isobath around Georges Bank. Integratedproductivity in the region varies seasonally by a factor

of 45x (0.1 to 4.4 gC m-2 d-1; O'Reilly, 1987), greaterthan in the California Current (15x variability; 0.2-3.0

gC m-2 d-1). The Marine resources Monitoring As-sessment and Prediction (MARMAP) database alongthe eastern seaboard of the U.S. has been the equiva-lent to the California Cooperative Oceanic FisheriesInvestigation (CalCOFI) database (West Coast) interms of its sheer size and length of sampling. How-ever, MARMAP does not maintain pigment or inher-ent optical property data. This SIMBIOS project at-tempts to provide this unique data for the region.

5.2 RESEARCH ACTIVITIES

We participated in several field cruises to fieldtest equipment, and collect data. We have completedthe following field activities:

• June '98-Second set of sea trials of sampling andunderway system in the Gulf of Maine and Geor-ges Bank (11d cruise).

• September-October '98-Eleven cruises aboardM/S Scotia Prince completed.

Real time enumeration of coccolithophores is es-sential to this work (as well as cell counts done afterthe cruise). We have an Olympus BH2 microscopewith epi-fluorescence and polarization optics whichallows us to quantify birefringent CaCO3 coccoliths,

empty coccospheres, and fluorescing, plated coccoli-thophore cells on board the ship. Due to limited timeat each station, only about two samples typically canbe enumerated. However, counts are recorded onvideo, which provides an important record for post-cruise comparison to the preserved counts, as well asfor any retrospective analyses. We take preserved cellcount samples at all stations (preserving with 4%buffered formalin) and, after the cruise, measure theconcentration of detached coccoliths and plated coc-colithophores (Utermöhl, 1931 and 1958).

The technique of Fernandez et al. (1993) is usedto measure CaCO3 concentrations. Briefly, samples

are filtered onto 0.4µm polycarbonate filters, andrinsing first with filtered sea water, then borate buffer


33

(pH=8) to remove seawater calcium chloride. Filtersare placed in trace metal free centrifuge tubes with 5ml 0.5% Optima grade Nitric acid. Next, the Ca con-centration is measured using graphite furnace atomicabsorption spectrometry. The sensitivity of the tech-nique, after correction to the volume of seawater fil-

tered, is about 2ng Ca l-1.Particulate organic carbon and nitrogen are meas-

ured according to Sharp (1974; 1991). One liter sam-ple bottles will be emptied via tygon tubing, throughin-line pre-combusted glass fiber filters. Low vacuumwill be maintained during filtration. These filters willbe placed in plastic petri dishes using clean forceps,and stored at -20°C. Prior to running the samples, theywill be fumed over HCl to remove any carbonates,then dried in a dessicator, and shipped on dry ice tothe Bermuda Biological Laboratory for analysis.Chlorophyll is measured fluorometrically according toStrickland and Parsons (1972) and modified accordingto Joint Global Ocean Flux Study (JGOFS) (1996)which involves filtering 200 ml seawater with Gelman

GFF filters, extracting overnight at 4oC in 10 ml of90% acetone, and measuring fluorescence with aTurner 111 fluorometer.

We locate high densities of coccolithophores inthe field with a flow-through system which estimatessuspended CaCO3 using an optical technique. AWyatt Technologies laser-light scattering photometer(equipped with a flow-through cell) is used to measureoptical volume scattering at 18 angles. Integration ofthis signal in the backward direction allows calcula-tion of backscattering in real time. A pump injectsweak acid every 4 minutes to dissolve the CaCO3, andthen backscattering is re-measured. The differencebetween the total bb and the acidified bb is called the“acid-labile bb,” and is calibrated to suspended

CaCO3 concentrations. The acidified bb represents

backscattering due to organic matter (which can becalibrated to CHN measurements). This value can beconverted to POC concentration using our POC-specific backscattering coefficients (Balch et al.,1999). The system monitors chlorophyll fluorescence,pH, temperature, salinity and 18-channel volumescattering (with and without CaCO3). An AC-9 pro-vides estimates of absorption and attenuation, whichby difference, also provides scattering. A Global Po-sitioning System is interfaced, as well. The flow-through system has been field-tested over carbonatebanks off of Florida, which provided data over 3 or-ders of magnitude in acid-labile bb. Such high con-

centrations of suspended CaCO3 are found in the mostdense coccolithophore blooms that we have ever vis-ited. During the Arabian Sea expedition, as well as 5recent Gulf of Maine cruises, the instrument ran virtu-ally flawlessly over thousands of kilometers, provid-ing a wealth of data on the distribution of particulate

inorganic and organic carbon (Balch et al., 1999). Thefinal cruise of this year was completed in October; theraw data have been worked up and distributed to theNASA SIMBIOS Project. Discrete samples for sus-pended calcite, POC, PON, and coccolith counts werealso taken, and will be processed over the comingmonths.

5.3 RESEARCH RESULTS

We completed 33 days at sea in 1998, with 11trips aboard the M/S Scotia Prince ferry (Table 1).Two of 11 trips were cloudy (82% were sunny). Inthe Gulf of Maine, where 28% of the time it is sunny,we have demonstrated that we take full advantage offerry as a ship of opportunity. For the year, we haveprocessed about 147 stations in the Gulf of Maine andGeorges Bank, and provided continuous underwaydata along 5000 miles of ship track for SeaWiFS cali-bration. Below are summarized the main samplingactivities:

• R/V Delaware 10-21 November 1997 R/V Dela-ware, Gulf of Maine and Georges Bank. Under-way data analysis completed and submitted to Se-aBASS database. Data collected: underway fluo-rescence, backscattering and calcite-dependentbackscattering (510 nm; Wyatt light scatteringphotometer, absolutely calibrated with a glassstandard), absorption, scattering (AC-9; SeaWiFSbands; recently calibrated), temperature, salinity.Discrete measurements at 27 stations of sus-pended calcite, POC/PON, chlorophyll, coccolithand coccolithophore concentration. Discrete cellcounts and analytical analyses are still beingprocessed.

• R/V Albatross 3-12 June 1998, Gulf of Maine andGeorges Bank. Data collected: underway fluores-cence, backscattering and calcite-dependent back-scattering (510 nm), absorption, scattering (AC-9:SeaWiFS bands), temperature, salinity, Lw and Edat SeaWiFS wavelengths measured with SatlanticSAS system. Discrete measurements at 28 sta-tions of suspended calcite, POC/PON, chloro-phyll, coccolith and coccolithophore concentra-tion.

• M/S Scotia Prince on a total of 1l cruises (Table1). Data collected: underway fluorescence, back-scattering and calcite-dependent backscattering(510 nm; Wyatt light scattering photometer, ab-solutely calibrated with a glass standard), absorp-tion, scattering (AC-9; SeaWiFS bands; recentlycalibrated), temperature, salinity. Lw and Ed atSeaWiFS wavelengths measured with SatlanticSAS system. Microtops sun photometer measur-ments performed for overpasses. Discrete meas-urements at stations of suspended calcite,POC/PON, chlorophyll, coccolith and coccolitho-


34

phore concentration. Discrete cell counts andanalytical analyses being processed.

5.4 WORK PLAN

Our contract calls for 20 ferry trips next year. Weanticipate no problem completing these trips since our

laboratory container and sample arm are already com-pleted and ready.

Moreover, we anticipate no problems in timelydata work-up and submission as our processing soft-ware is now streamlined for post-cruise calibrations.

Table 1. Summary of cruises and in situ data collected.

Cruise LocationCruiseName Date

Flu

ore

scen

ce

Bac

ksca

tter

ing

Cal

cite

-dep

end

ent

bac

ksca

tter

ing

Ab

sorp

tio

n

Sca

tter

ing

Tem

per

atu

rean

d s

alin

ity

Mic

roT

op

s su

np

ho

tom

eter

Sta

-ti

on

sS

tati

on

s

Su

spen

ded

calc

ite

Ch

loro

ph

yll

PO

C/P

ON

Co

cco

lith

an

dC

occ

olit

ho

ph

. Co

n.

Dis

cret

e ce

ll co

un

t

Lw a

nd

Ed

(Sat

lan

tic

SA

S)

Delaware, Gulf of Maine andGeorge Bank

10d 10-21 Nov 1997

÷ ÷ ÷ ÷ ÷ ÷ 27 ÷ ÷ ÷ ÷ ÷

Gulf of Maine and George Bank 11d 3-12June 1998

÷ ÷ ÷ ÷ ÷ ÷ 28 ÷ ÷ ÷ ÷ ÷

Portland, Maine andYarmouth (NV)

Ferry 1 10-11Sept. 1998

÷ ÷ ÷ ÷ 2 ÷ ÷ ÷ ÷ ÷ ÷


Ferry 2 16-17Sept. 1998

÷ ÷ ÷ ÷ ÷ ÷ ÷ 10 ÷ ÷ ÷ ÷ ÷ ÷


Ferry 3 17-18Sept. 1998

÷ ÷ ÷ ÷ ÷ ÷ ÷ 9 ÷ ÷ ÷ ÷ ÷ ÷


Ferry 4 28-29Sept. 1998

÷ ÷ ÷ ÷ ÷ ÷ ÷ 9 ÷ ÷ ÷ ÷ ÷


Ferry 5 3-4Oct. 1998

÷ ÷ ÷ ÷ ÷ ÷ ÷ 10 ÷ ÷ ÷ ÷ ÷

Portland, Maine and Yarmouth (NV)

Ferry 6 4-5Oct. 1998

÷ ÷ ÷ ÷ ÷ ÷ ÷ 10 ÷ ÷ ÷ ÷ ÷


Ferry 7 11-12Oct. 1998

÷ ÷ ÷ ÷ ÷ ÷ ÷ 10 ÷ ÷ ÷ ÷ ÷


Ferry 8 12-13Oct. 1998

÷ ÷ ÷ ÷ ÷ ÷ ÷ 8 ÷ ÷ ÷ ÷ ÷


Ferry 9 16-17Oct. 1998

÷ ÷ ÷ ÷ ÷ ÷ ÷ 9 ÷ ÷ ÷ ÷ ÷ ÷


Ferry 10 17-18Oct. 1998

÷ ÷ ÷ ÷ ÷ ÷ ÷ 9 ÷ ÷ ÷ ÷ ÷


Ferry 11 18-19Oct. 1998

÷ ÷ ÷ ÷ ÷ ÷ ÷ 8 ÷ ÷ ÷ ÷ ÷


35

Chapter 6

OCTS and SeaWiFS Bio-Optical Algorithm andProduct Validation and Intercomparison in

U.S. Coastal Waters

Christopher W. BrownNOAA/NESDIS, Camp Springs, Maryland

John C. BrockUSGS Center for Coastal Geology and Regional Marine Studies

St. Petersburg, Florida

6.1 INTRODUCTION

The launch of the National Space DevelopmentAgency of Japan (NASDA) Ocean Color and Tem-perature Sensor (OCTS) in August 1996, and thelaunch of Orbital Science Corporation’s (OSC)SeaWiFS in August 1997 signaled the beginning of anew era for ocean color research and application. Theocean color data provided by these satellites promisesthe ability to remotely evaluate and monitor 1) waterquality, 2) transport of sediments and adhered pollut-ants, 3) primary production, upon which commercialshellfish and finfish populations depend for food, and4) harmful algal blooms which pose a threat to publichealth and economies of affected areas. Accordingly,several US government agencies have recently ex-pressed interest in using optical remote sensing in UScoastal waters for these purposes.

The goal of this project is to evaluate the per-formance of standard bio-optical algorithms and vali-date satellite ocean color products derived from OCTSand SeaWiFS data in Case I and Case II coastal USwaters. To accomplish this goal, in situ bio-opticalobservations were collected and analyzed. Thisdocument briefly describes the methods used in sam-ple collection and analysis, and presents preliminaryresults obtained during the first year of the project. Asa result of the failure of the Japanese satellite on June30, 1997, and the consequent termination of the OCTSdata stream, we have been unable to evaluate the per-formance of OCTS algorithms and data products.


In-situ bio-optical data were collected duringeight cruises in diverse US Case I and Case II watersalong the Atlantic coast and in the Great Lakes sinceOCTS became operational in November 1996 (Table1). Optical instruments were deployed on cruises to

measure surface spectral downwelling irradiance, in-water spectral downwelling irradiance and upwellingradiance. Deployment of other instruments and thecollection of bottle samples enabled additional meas-urements that included temperature, chlorophyll fluo-rescence, light scattering, quantum scalar irradiance,total suspended solids concentration, chlorophyll andpigment concentrations from fluorometric and HighPressure Liquid Chromatography (HPLC) techniques,and absorption coefficients of colored dissolved or-ganic matter and particles. Although samplingstrategies and instrument packages varied betweencruises, a Biospherical Instruments Profiling Reflec-tance Radiometer (PRR) cage was typically deployedoff the stern of the boat in conjunction with a refer-ence surface unit with matching channels. The PRRcage contains a split PRR600s that measures sevenchannels of downwelling irradiance, seven channels ofupwelling radiance, depth, tilt, roll, and temperature.A reference surface unit PRR610 that measures sevenmatched channels of surface downwelling irradiancewas also used. PRR600s channels 1 to 6 are narrowband (10-nanometer [nm], full width half maximum[FWHM]) centered at 380 nm, 412 nm, 443 nm, 490nm, 510 nm, and 555 nm, while channel 7 on thedownwelling sensor and PRR610 measures broadband Photosynthetically Available Radiation (PAR)(400 to 700 nm). In addition, the PRR cage contains a10-centimeter pathlength, 660 nm Light Emitting Di-ode (LED) -based SeaTech transmissometer, a Bio-spherical Instruments Quantum Scalar Profiling sensor(QSP200), a SeaTech light scattering sensor (LSS),and a WETLabs Wetstar chlorophyll fluorometer.Water is drawn through the fluorometer using a Sea-Bird pump running at 2,000 revolutions per minute.The data from these instruments are multiplexedthrough the PRR600s such that each record contains adepth and the parameters from every instrument.

In some cases, a Hydro-Optics, Biology, and In-strumentation Laboratories, Inc. (HOBI Labs) Hydro-


36

Scat-6 spectral backscattering sensor was lowered byhand following the deployment of the PRR cage.Typically, an along-track system was used to measurethe position (latitude, longitude), time, course andspeed of the vessel, temperature, and salinity. In situtemperature, salinity, and density were also measuredat some stations with a Conductivity-Temperature-Depth (CTD) instrument. Water samples for chloro-phyll biomass, particulate, and dissolved absorptionwere obtained from a separate cast using a Niskin ar-ray. Discrete water samples were collected followingthe PRR cast, from just below the sea surface using aNiskin bottle.

Bio-Optical Data Processing

The PRR optical data was processed using theBermuda Bio-Optics Project (BBOP) processing soft-ware (Siegel et al, 1995c). The data were separatedinto upcast and downcast profiles and then binned to0.5 meter bins. Spectral attenuation coefficients werecalculated for the optical channels over a five pointmoving window. Subsurface downwelling irradianceand upwelling radiance were extrapolated to just be-low the surface using data from the top 3 meters. TheHydroScat-6 data were processed using software pro-vided by HOBI Labs, Inc. The data were despiked, intwo passes using a difference threshold, and a movingaverage was calculated for these channels. These datawere also separated into upcast and downcast profilesand then binned to 0.5-m bins.

For fluorometric chlorophyll a and HPLC pig-ment determinations, water samples are filteredthrough glass fiber. The chlorophyll a samples arecold extracted in 10 ml of 90% acetone (10% water)for 24 hours in the dark, and the biomass is deter-mined fluorometrically with a Turner Designs fluo-rometer using the method of Yentsch and Menzel(1963). Filtered HPLC pigment samples are cut intosmall pieces, ground, and extracted in a 1.5 ml micro-centrifuge tube with 1.5 ml 90% acetone and areplaced in a freezer overnight. They are centrifuged at0°C for 15 minutes, 0.5 ml is filtered through a Nal-gene nylon syringe filter, the samples are diluted to60% acetone with the addition of 0.25 ml water, and0.5 ml is injected into the HPLC. A Hewlett-Packard1050 Series HPLC with a Phenomonex SphericloneODS (2) reverse-phase column (250 mm x 4.5 mmwith 5 µm particles) is used with a ternary gradient toseparate and identify the photosynthetic pigments(Wright et al, 1991). On cruises that included along-track measurements, water from about one meter be-low the sea surface was pumped through a hose into abucket in which a Hydrolab Datasonde 3 Multiprobelogger was immersed. The Datasonde was used tomeasure temperature, specific conductivity, and salin-ity. Data from many of these cruises are made avail-able via cruise reports (Culver et al. 1998, Subrama-

niam et al. 1997a, Subramaniam et al. 1997b, Subra-maniam et al. 1997c, Subramaniam et al. 1998), andalso through a website at the NOAA Coastal ServicesCenter (http://www.csc.noaa.gov/crs/cruises/).


Our group has conducted eight cruises since thebeginning of October 1996, including six cruises inthe South Atlantic Bight (SAB), one in the Gulf ofMaine, and one in the vicinity of Nantucket Shoals.Coverage of the SAB has been extensive due to pro-ductive collaboration with the NOAA Southeast Fish-eries Science Center.

The SAB consists of a variety of environmentsincluding near-coastal and continental shelf regimes,the Gulf Stream, and the Sargasso Sea. Smooth tran-sitions from low temperature waters with high pig-ment concentrations and attenuation coefficients, thatare typical of nearshore environments, to the highertemperature waters with low pigment concentrationsand attenuation coefficients, that are typical of off-shore environments, are often disrupted by GulfStream eddies that appear on the continental shelf.The variability in the biological and optical character-istics of these regimes complicates temporal or spatialanalyses of changes in, for example, phytoplanktonspecies composition, primary production rates, col-ored dissolved organic matter (CDOM) concentration,suspended sediments, or temperature.

Coastal waters of the SAB can be classified asoptical Case II waters. Data from the MAY97OBcruise (Subramaniam et al., 1998), on May 5, 1997and May 8, 1997, indicate that the waters of OnslowBay and especially Pamlico Sound were sediment-dominated. Data from the May cruise suggest that theOC2 algorithm (O’Reilly et al., 1998) over-estimatedchlorophyll and pigment concentrations in these wa-ters by a factor of at least 1.5 and up to 15. Chloro-phyll a and total suspended solids concentrations inwaters of the SAB during the APR98SAB cruise inApril 1998 were highly variable. Chlorophyll a con-centrations tended to be higher nearshore, and totalsuspended solids concentration showed no discerniblegeographic pattern, with high and low concentrationsbeing measured at both nearshore and offshore sta-tions. High attenuation coefficients for the bluewavelengths indicated the presence of high concentra-tion of colored dissolved organic material near thecoasts of South Carolina, Georgia, and Florida how-ever, these high coefficients were not observed nearNorth Carolina. Overall, the OC2 version 2 algorithmsignificantly over-estimated the chlorophyll a con-centration by a factor of 4 and as much as a factor of50, however, CZCS pigment concentration was esti-mated more successfully, with ratios of measured toestimated pigment ranging from 1 to 5.


37

Table 1. Summary of cruises.

Cruise Location Date Aircraft Ship Cruise NameSouthern California Bight 1/29/96 to 2/18/96 None Jordan JAN96CALMid-Atlantic Bight 3/1/96 to 3/10/96 NASA P-3 Endeavour MAR96OMPSouth Atlantic Bight 4/22/96 to 4/26/96 None Ferrel APR96FERNew York Bight 5/13/96 to 5/16/96 None Osprey MAY96NYGulf of Maine 5/18/96 to 5/24/96 None Gulf Challenger MAY96MELake Erie 8/26/96 to 8/30/96 None R/V Bio Lab AUG96GLBeaufort, North Carolina 3/13/97 to 3/13/97 None R/V Onslow Bay MAR97OCCCoastal North Carolina 5/5/97 to 5/8/97 None R/V Onslow Bay MAY97OBGulf of Maine 6/3/97 to 6/10/97 None Gulf Challenger JUN97MESouth Atlantic Bight 9/6/97 to 9/24/97 None Cape Hatteras SEP97SABSouth Atlantic Bight 11/4/97 to 11/6/97 None R/V Palmetto NOV97SARSouth Atlantic Bight 4/5/98 to 4/27/98 NOAA Twin Otter Cape Hatteras APR98SABNantucket Shoals 7/6/98 to 7/10/98 None Gulf Challenger JUL98NANSouth Atlantic Bight 10/28/98 to 11/25/98 None Cape Hatteras NOV98SAB

Figure 1. OC2 version 2 vs. fluorometric chlorophyll.

0.10

1.00

10.00

0.10 1.00 10.00

Measured ChlF (mg/m^3)

OC

2 ve

r 2

(mg/

m^

3)

r2=0.83Slope=0.8


38

The results of cruise NOV97SAR (Culver et al,1998), conducted from November 3 to 5, 1997, showthat the OC2 algorithm performs reasonably well atstations well off-shore, although, in general the algo-rithm tends to over-predict chlorophyll concentrationsin coastal waters. Also, we found good agreementbetween our in situ optical measurements of normal-ized water-leaving radiance and that measured by theSeaWiFS sensor. We interpret this to mean that our insitu measurements are of good quality, and that ourgoal of validating satellite chlorophyll algorithms us-ing these measurements is reasonable and tractable.

The consistent overestimation of chlorophyll aconcentration in the SAB by the standard algorithms islikely due to the influence of sediments and CDOMon ocean color. Preliminary analysis suggests thatsediment may be the dominant factor near NorthCarolina, and CDOM may be the dominant factor forthe southern waters.

When these components do not covary with chlo-rophyll a, the algorithms will overestimate chlorophylla concentration. The agreement between our in situoptical measurements of normalized water-leavingradiance and that measured by the SeaWiFS sensorindicate that a chlorophyll algorithm for these watersis tractable, however, the coefficients will require ad-justment to compensate for the effect of sediment andCDOM on the radiance measurements. Overall, ourresults show that it is necessary to establish empiricalbio-optical relationships for separate coastal regions,such as the SAB.

There is a reasonable relationship between theOC2 algorithm-derived chlorophyll concentration andthe fluorometrically measured chlorophyll concentra-tion in the SAB (Figure 1). Small changes in thepower equation coefficients should yield a better re-gional algorithm for the SAB. The relationship be-

tween HPLC-derived chlorophyll a and the OC2 algo-rithm is not as robust and requires further study.

6.4 WORK PLAN

The project plans to complete three additionalcruises in the SAB during October 1998 through Sep-tember 1999 in order to collect a more complete dataset using our established sampling protocol. Giventhat analyses thus far have demonstrated that the OC2Version 2 algorithm estimates of chlorophyll a con-centration for SAB waters exceed acceptable errorlimits, our project will strive to develop a region-specific algorithm for SeaWiFS for these waters.During the recent NOV98SAB cruise, in addition tothe deployment of the NOAA CSC PRR600, a Satlan-tic free falling profiler (SPMR), the SIMBAD, and anabove water hyperspectral Spectrex instrument wereused to obtain water leaving radiance data. This coin-cident multiple instrument data set will be used tocompare the efficacy of each instrument in coastalCase II waters. In response to the observation thaterroneous negative radiances occur in SeaWiFS dataat 412 nm band, especially in late spring/early sum-mer, sunphotometer data will be used to determine thesuitability of the standard atmospheric correction algo-rithms for the SAB.

ACKNOWLEDGMENTS

We sincerely thank Drs. A. Subramaniam and M.Culver for their invaluable assistance in this project.In addition to collection and analysis of the data, theyreviewed and suggested improvements in a previousversion of this Technical Memorandum.


39

Chapter 7

Validation of Ocean Color Satellite Data Productsin Under-Sampled Marine Areas

Douglas G. Capone and Ajit SubramaniamChesapeake Biological Laboratory

Solomons, Maryland

Edward J. CarpenterSUNY at Stony Brook, New York

7.1 INTRODUCTION

The marine diazotroph, Trichodesmium is aplanktonic cyanobacterium which occurs throughoutthe tropical and sub-tropical oligotrophic seas. It canform large blooms and is thought to be responsible forthe major fraction of N2 fixation in the pelagic zone ofworld's oceans. The total rate of N2 fixation by thisorganism has been thought to be about 10 Tg annu-ally. However, estimates of Trichodesmium's contri-bution to total marine N2 fixation are relatively crudeand other independent estimates suggest a higher rateof marine N2 fixation. The annual rate was derived byscaling an average trichome N2 fixation rate with bio-mass abundance and distribution data largely takenfrom historical plankton surveys. However, there areunique difficulties in quantifying the biomass of abuoyant, colonial alga such as Trichodesmium and thequantitative accuracy of many of these surveys aresuspect. One of the most appealing prospects of usingsatellites in biological oceanography has been the po-tential to derive flux estimates from inventories ofstanding stock calculated from changes in opticalproperties of the ocean. Using algorithms to estimateglobal distributions of Trichodesmium biomass, wecan calculate nitrogen fixed by this organism, andhence model its contribution to new production in theworld oceans. Towards this goal, we proposed to usea previously developed hyperspectral optical modelfor the Trichodesmium to construct sensor specificbiomass algorithms.

The goal of our proposed work was to developsensor specific algorithms to quantify the biomass ofTrichodesmium in the world oceans. The specific ob-jectives of our proposal were to:

• Use a hyperspectral optical model for Trichodes-mium to derive sensor specific Trichodesmiumchlorophyll algorithms.

• Constrain and validate these algorithms usingfield measurements. Also use the field measure-ments to validate standard chlorophyll algorithms.

• Estimate the error in using standard algorithms toestimate chlorophyll in a Trichodesmium bloom.Develop flags for switching algorithms if neces-sary.

We were required by our contract to acquire andprocess data from at least one cruise per year. Thedata included measurements of vertical profiles ofspectral downwelling irradiance, upwelling radiance,spectral attenuation and remote-sensing reflectancecalculated from the profile data, photosyntheticallyactive radiation, primary productivity, and concentra-tions of chlorophyll a and other pigments. We wererequired to purchase and make available to the instru-ment pool, an AC-9 instrument and a turnkey free-fallprofiling radiometer system. Here we outline our pro-gress during the first year of our SIMBIOS project todevelop an algorithm for detection and quantificationof Trichodesmium biomass for the SeaWiFS sensor.We also describe our field program for validating theOC2 algorithm and obtaining in situ bio-optical datain under-sampled marine areas.


The work detailed below is of an effort in prog-ress and intended to provide an interim update to theSIMBIOS project. Our initial approach was to derivea Trichodesmium chlorophyll specific algorithm froman optical model, and then constrain this algorithmbased on field data. Our optical model detailed inSubramaniam et al. (1999) showed that at low con-centrations, Trichodesmium does not have singularenough optical properties to be uniquely identified(Figure 1). But as the concentration of Trichodesmiumin the water column increases, its optical propertiesdominate and it is possible to identify and quantify


40

this organism (Figure 1). Therefore, we initially de-cided to develop a branching algorithm, where at lowTrichodesmium concentrations the algorithm producedtotal chlorophyll (i.e. not Trichodesmium specific), butat higher Trichodesmium concentrations the algorithmresulted in Trichodesmium specific chlorophyll. Weformulated the algorithm as (R555-R490/R510) whereR555, R490, R510 are the SeaWiFS remote sensing re-flectance corresponding to bands 3, 4 and 5 respec-tively. We parameterized this formulation by fitting apower equation to the results of the previously devel-oped remote-sensing reflectance based hyperspectraloptical model for Trichodesmium. However, we werenot satisfied by the results of this formulation andhave been working on other approaches and bandcombinations.

Although the branching algorithm approach isvery attractive for producing global maps of Tricho-desmium abundance based on a sound theoreticalmodel, it is essential to characterize the chlorophyllconcentration threshold where the algorithm results inTrichodesmium specific chlorophyll concentration.The optical model suggested that this threshold wasabout 0.5 mg/m3 Trichodesmium Chl a. While we didnot encounter such high concentrations of Trichodes-mium during the Trichonesia cruise, preliminaryanalysis of field data from a cruise in the South Atlan-tic Bight shows that this threshold is probably muchhigher – around 3 mg/m3. We need more field datawith Trichodesmium cell counts to characterize thisthreshold. In the meantime we have developed a flagto study global occurrence of Trichodesmium blooms.This flag is based on the combination of high reflec-tance at 555 nm, and absorption at 490 nm. We arecomparing this to climatological data on Trichodes-mium occurrence - in regions and seasons whereTrichodesmium is known to occur historically. Wethink we are very close to completing an initialTrichodesmium chlorophyll algorithm and will com-municate it to the SIMBIOS office as soon as we aresatisfied with it. Our 1998 field campaign was a 35-day researchcruise in the South Western Pacific ocean. We leftLyttleton, New Zealand on the 24 March 1998 andarrived in Suva, Fiji on the 30 April 1998. We hadrequested cruise time in January-February when theorganism is most abundant in this region. However,due to ship availability, our cruise was pushed toApril. This, in combination with the effects of the El-Nino, resulted in our encountering lower Trichodes-mium populations than we had anticipated. After ourport call in New Caledonia on 30 March, we had tohead north to avoid Cyclone Zumann. Thus we wereunder overcast skies during the portion of the cruisewith some of the highest chlorophyll concentrations.In total, we occupied 51 stations and made 33 opticsdeployments including 2 diel deployments for Inher-

ent Optical Property measurements. The in situ opticaldata, chlorophyll concentrations, and carbon andnitrogen fixation rates have been submitted toSeaBASS.

We acquired a Wetlabs AC-9 in January 1998.This instrument was field tested during a SIMBIOScruise (CALCOFI 9802) and recalibrated after thatcruise. This instrument was used on the Trichonesiacruise and is now available at CBL.

A Satlantic SeaWiFS Profiling Multispectral Ra-diometer (SPMR) with a matched SeaWiFS Mul-tispectral Surface Reference (SMSR) was procured inMarch 1998. The SPMR is a 13 channel free fallingprofiler that measures in situ downwelling irradiance(Ed-) and upwelling radiance (Lu-) at 340, 380, 412,443, 490, 510, 520, 555, 565, 619, 665, 670, and 683nm. The SMSR is a floating reference that measuresdownwelling irradiance (Ed+) just above water andupwelling radiance (Lu-) 70 cm below the surface.The SPMR includes a 300-m power telemetry cable, 2axis tilt sensors, and external temperature and con-ductivity sensors. The SMSR includes a 100-m powertelemetry cable and 2 axis tilt sensors. The packageincludes a deckbox with computer interface and powersupply, data logging and display software, and a Hi-tachi pentium laptop computer with an extra PCMCIAserial card and cables.

The SPMR was used on the Trichonesia cruise inMarch-May, on NOAA cruises in July and November,and a CALCOFI cruise in September. Initially therewere problems with gain switching and some datafrom the Trichonesia cruise and JUL98NAN cruisewere compromised because of this. This has beensubsequently fixed by Satlantic. The instrument hasbeen recalibrated twice at Satlantic and twice at UCSanta Barbara since March 1998. It has been shippedto the Indian Ocean for use in the Indian Ocean Ex-periment (INDOEX).

7.3 WORK PLAN

We will continue to work on refining the Tricho-desmium algorithm. Specifically we will analyze thefield data collected during the Trichonesia (1998) andINDOEX (1999) cruises to define the Trichodesmiumspecific chlorophyll concentration that can be detectedby the algorithm and test the sensitivity of the algo-rithm. We will work on algorithms specific to theOCTS, the MOS, and the MODIS sensors. We willparticipate in the INDOEX cruise in March 1999 anda cruise off northern Australia in October 1999.


41

Figure 1. Optical model of Trichodesium remote sensing reflectance for a) 0.1 mg Chl/m3, b) 1.0 mgChl/ m3, and c) 10.0 mg Chl/ m3.

Wavelength (nm)

400 450 500 550 600 650 700 750

% R

emot

e se

nsin

g re

flec

tanc

e

0.000

0.005

0.010

0.015

0.020

0.025SeawaterTrichodesmium coloniesMixture of colonies and trichomesAverage phytoplanktonSynechococcusTrichomes onlyColonies - modeled backscatter

11

2

2

3

3

4

4

5

5

6

6 7

7

Wavelength

400 450 500 550 600 650 700 7500.000

0.005

0.010

0.015

0.020

0.025

SeawaterTrichodesmium coloniesMixture of colonies and trichomesAverage phytoplanktonSynechococcusTrichomes onlyColonies - modeled backscatter

Low Gilvin 1

2

1

2

3

3

44

556

6

77

400 450 500 550 600 650 700 7500.000

0.005

0.010

0.015

0.020

0.025

0.030SeawaterTrichodesmium coloniesAverage phytoplanktonSynechococcusColonies - modeled backscatter

% R

emot

e se

nsin

g R

efle

ctan

ce


42

Chapter 8

Stray Light and Atmospheric Adjacency Effects forLarge-FOV, Ocean-Viewing Space Sensors

Kendall L. CarderUniversity of South Florida,

St. Petersburg, Florida

8.1 INTRODUCTION

The radiance of dark targets (e.g. oceans andlakes) when measured by large-field imagers such asMODIS and SeaWiFS can be enhanced by stray lightwithin the sensor and by radiance atmosphericallyscattered from bright, adjacent targets. One of thegoals of this project is evaluation of this stray lightand estimation of its effect on derived products ofremote color imagery.


The first year of this project focused on collectingfield data to evaluate the performance of SeaWiFS.To assure field data quality, a significant effort wasexpended in characterizing and calibrating field in-struments as well as standardizing field methodology.Field sites were selected where adjacency artifactswould be likely. One site is a deep ocean canyon sur-rounded by bright carbonate banks, the Tongue of theOcean (TOTO); another is Lake Okeechobee, Floridawhere dark water is surrounded by bright vegetation.With funding from other sources, additional field ex-periments were applied to this study and help withevaluation of SeaWiFS performance. These includedsampling on Tampa Bay, the West Florida Shelf,Exuma Sound, the San Juan Islands in Puget Sound,and Lake Tahoe.


The degradation of SeaWiFS infrared bands,which are used to evaluate the type and amount ofaerosol path radiance, caused a perturbation in thetype of aerosol correction to be made by the SeaWiFSatmospheric correction scheme. This resulted in overestimation of the blue-richness of the aerosol scatter-ing. Furthermore, high cirrus clouds have a tendencyto exaggerate this effect, due to enhanced aerosol ab-sorption in the oxygen band near 762 nm. To obviatethe worst effects of both problems, we developed an

aerosol-typing algorithm using bands 6 and 8 overclear waters. This approach forced the calibration ofband 6 relative to band 8 to provide normalized water-leaving radiance values at band 6 consistent with theGordon and Clark (1981) for scenes with known aero-sol type. For scenes of unknown aerosols, the new 6-8band combination delivered appropriate aerosol typeswith SEADAS. This prevented blue exaggeration ofthe aerosol type and prevented excessive removal ofblue radiance from the scene, which otherwise wouldhave resulted in blue-poor water-leaving radiance val-ues. This approach provided atmospherically correctedscenes that did not became negative at blue wave-lengths for gelbstoff-rich (blue-absorbing) areas. Thisapproach also provided clear water, normalized water-leaving radiance values at 555 nm consistent withGordon and Clark (1981) using the calibration factorsdistributed by the SeaWiFS Project on 6 January1998. Furthermore, it provided a data distribution forclear waters using ratios of bands 1 and 2 versus bands2 and 5 that are consistent with those found by Carderet al. (1999) using the SeaBASS in situ data set.

Using the above calibration scheme, SeaWiFSimages of the TOTO, the West Florida Shelf, and re-gions between were compared to field data using thedefault NASA chlorophyll algorithm and the Carder etal. (1999), MODIS-like, semianalytic chlorophyll al-gorithm. For these spring scenes, the pigments weresomewhat packaged with little in the way of photo-protective pigments. Both algorithms performedsimilarly (within 5%) for Case 1 waters. Whengelbstoff absorption overwhelmed the pigment ab-sorption, however, the semianalytic algorithm deliv-ered chlorophyll values within 25% of measured val-ues; while the default NASA algorithm overestimatedchlorophyll by as much as a factor of 4. These sceneswill be re-evaluated with the new calibration-degradation curves of SEADAS 3.2 during the nextcontract year, but the new results are not expected todeviate significantly from these results using the 6-8band combination except for turbid waters.

In terms of evaluating atmospheric adjacency andstray light effects due to infrared radiance emanatingfrom the bright foliage surrounding Lake Okeechobee


43

or blue-green light reflected from the bright shallowbanks surrounding TOTO, our preliminary findingssuggest the following:

• Lake Okeechobee has so much horizontal vari-ability in the water-leaving radiance field at infra-red wavelengths that it is at times an impracticalsite at which to evaluate stray infrared light fromits surrounds. It does offer, however, so muchgelbstoff absorption at 412 nm that we are con-sidering performing atmospheric correctionsbased upon the blue end of the spectrum.

• Data we collected at the Friday Harbor Labora-tory in the San Juan Islands suggest that this siteprovides a viable alternative to Lake Okeechobeefor evaluation of stray infrared light and/or at-mospheric adjacency effects. HyperspectralImager for Low Light Spectroscopy (PHILLS)aircraft data Curtis Davis (NRL) suggests that onecan find a red-edge effect (exaggerated infraredlight) in the water near sunlit tree lines in datacollected at 10,000 feet and with visibility in ex-cess of 50 km. These data will be evaluated nextyear along with SeaWiFS data collected at highresolution over Griffin Bay (San Juan Island)which is surrounded by forest land.

• Chlorophyll values farther than 2 pixels offshoreand east from bright, shallow banks of the TOTOwere not significantly different than those foundfarther offshore. Extreme caution must be taken,however, to avoid misinterpreting conditionswhere gelbstoff- and particle-laden waters fromthe shallows are transported offshore. Next yearan evaluation of the electronic over-shoot, and/orhysteresis on the west sides of bright targets willalso be evaluated on the eastern edge of theTOTO. More accurate calibration of theSeaWiFS sensor will help in this effort.

• A recently published paper by Reinersman et al.(1998) on calibration of aircraft and small foot-print, spacecraft sensors using cloud shadows wasdependent upon the development of a MonteCarlo model of radiance from single sphericalclouds. This model is being modified to permitevaluation of atmospheric adjacency effects ofclouds on radiance values received by a spacesensor when viewing the ocean through an open-ing (elliptical cylinder) in a cloud bank. Algo-rithms that are more immune to the adjacency ef-fects of clouds will be sought. Much of the lightscattered into the sensor from clouds will be frommolecules and stratospheric aerosols and thin cir-rus clouds. A strategy using the effect of the oxy-gen absorption line on band 7 versus results frombands 6 and 8 may provide an indication of theeffective height of any aerosol scattering forSeaWiFS, whereas MODIS has a cirrus discrimi-nator built into its data processing scheme

8.4 WORK PLAN

The main goal of the project will remain thesame, to evaluate stray light effects on derived prod-ucts from ocean color imagery. Since MODIS andOCTS will not be available, the effort will be concen-trated on SeaWiFS, Advanced Visible and InfraredImaging Spectrometer (AVIRIS), and PHILLS.

The TOTO site will be visited in mid-April. It isplanned to use PHILLS overflights of Griffin Bay,Tampa Bay, and the West Florida Shelf. The aircraftdata will be used to refine techniques to detect atmos-pheric adjacency effects and bright infrared reflec-tance due to vegetation.


44

Chapter 9

Bio-Optical Measurements at Ocean Boundariesin Support of SIMBIOS

Francisco P. ChavezMonterey Bay Aquarium Research Institute

Moss Landing, California

9.1 INTRODUCTION

This project consists of a mooring program andsupporting cruise-based measurements aimed at quan-tifying the spectrum of biological and chemical vari-ability in the equatorial Pacific. The mooring programwas designed to obtain continuous time series of bio-logical and chemical properties on a time scale that isequivalent to measurements of currents, local windsand temperature structure.The project has the follow-ing general objectives:

• to understand the relationships between physicalforcing, primary production, nutrient supply andthe exchange of carbon dioxide between oceanand atmosphere in the equatorial Pacific;

• to describe the biological and chemical responsesto climate and ocean variability;

• to describe the spatial, seasonal and interannualvariability in near surface plant pigments, primaryproduction, carbon dioxide and nutrient distribu-tions, and

• to obtain near real-time bio-optical measurementsto ground-truth satellite measurements of oceancolor.


Moorings

Bio-optical and chemical sensors were deployedon two moorings of the TAO array in the equatorialPacific at 0°, 155°W and 2°S, 170°W. The followingsensors formed the core set:

• A Biospherical PRR-620 located 3m above thewater surface, to measure downwelling irradianceat 412, 443, 490, 510, 555, 656nm and PAR.

• A ∆pCO2 system to measure the difference inpartial pressure of CO2 between ocean and at-mosphere (supported by NOAA).

• Two Satlantic OCR-100s located approximately1.5m below the surface, to measure upwelling ra-

diance at 412, 443, 490, 510, 555, 670 and683nm.

• Two Wetlabs miniature fluorometers at approxi-mately 1.5m and 20m below the surface to meas-ure stimulated in vivo fluorescence.

• Two Biospherical MCP-200s located at 10m and30m below the surface to measure downwellingirradiance at 490nm.

• A Biospherical PRR-600 located 20m below thesurface to measure downwelling irradiance at412, 443, 490, 510, 555, 670nm and PAR, plusupwelling radiance at 412, 443, 490, 510, 555,670 and 683nm.

Daily noon-time bio-optical data were deliveredvia service ARGOS in near real time to MBARI, andthen via automated FTP to the SeaBASS database,after some processing and quality control. Higher fre-quency, publication-quality data (10 or 15-minuteintervals) were recovered at approximately six monthintervals, and sent to the SeaBASS database afterprocessing and quality control. Some derived prod-ucts, such as OC2V2 chlorophyll from surface up-welling radiances, and mean chlorophyll over the up-per 20m of the water column (Morel, 1988) were in-cluded in these data files. Using simple algorithmsderived from optical profile data the upwelling radi-ances measured by the moored radiometers have beenconverted to normalized water-leaving radiances(Lwn), for the SeaWiFS calibration/validation effort.

In addition to the two mooring installations de-scribed above, MBARI has also maintained smallerbio-optical packages at four sites in the equatorial Pa-cific: 2°N, 180°; 2°S, 140°W; 2°N, 140°W and 2°N,110°W. These bio-optical packages produced meas-urements of downwelling incident irradiance at490nm, upwelling radiance from ~1.5m depth at 412,443, 490, 510, 555, 670 and 683nm, plus stimulated invivo fluorescence. Data, including derived Lwn (seeabove) were delivered in near real time to MBARI andthe SeaBASS database. A bug in the original softwareprecluded collection of significant data prior to Marchof 1998.


45

In situ measurements

All cruises were undertaken aboard the NOAAship Ka’imimoana, with the exception of GP6-98-RBaboard the NOAA ship Ronald H. Brown (Table 1).Essentially, the cruise-based measurements consistedof chlorophyll (using the fluorometric method de-scribed by Chavez et al., 1995) and nutrient profiles (8depths, 0-200m) obtained at CTD stations between8°N and 8°S across the Pacific from 95°W to 165°E.On selected cruises (primarily the 155°W and 170°Wmeridional transects), primary productivity measure-ments and optical profiles, using the Satlantic Profil-ing Multispectral Radiometer, were also performed.These data were archived at MBARI and the chloro-phyll and optical profile measurements provided toNASA’s SeaBASS database.


Chavez et al. (1998) describe the biological-physical coupling observed in the central equatorialPacific during the onset of the 1997-98 El Niño. Fig-ure 1 shows the time series of chlorophyll, SST and∆pCO2 from the mooring located at 0°, 155°W for thetime period covering the peak of the 1997-98 El Niñoand the subsequent recovery of the phytoplanktoncommunity. Our data from the four smaller bio-opticalpackages describe the physical and biological effectsof the passage of Tropical Instability Waves. Whilethe corresponding SeaWiFS data in large part agreeclosely with the mooring-derived chlorophyll concen-trations, there are time periods where this is not the

case. This may in part be due to the spatio-temporalaveraging that is inherent in the SeaWiFS 9km/8-Daydata. We have performed SeaWiFS chlorophyll com-parisons with extracted chlorophyll concentrations forall of the 1998 equatorial Pacific cruises, with goodagreement between the two data types.

9.4 WORK PLAN

During 1999 data collection will continue as for1998. The two major mooring installations at 0°,155°W and 2°S, 170°W are now operating well aftersome initial problems with damage and loss of instru-ments. The smaller bio-optical packages at four addi-tional locations across the Pacific are also performingwell and data collection at those sites, as describedabove, will also continue. The program of cruise-based measurements of chlorophyll, nutrients, primaryproductivity and bio-optical profiles will continue onup to eight equatorial Pacific cruises during 1999,with scheduled SeaWiFS LAC where possible, to en-hance the probability of obtaining valid matchups.

The major data processing objective for 1999 is toautomate the near real-time data provision to the Sea-BASS database as much as possible to aid in the cali-bration/validation effort. In addition we aim to im-prove our data processing routines for the high resolu-tion data that are downloaded from the 0°, 155°W and2°S, 170°W moorings at approximately 6 month in-tervals. This should decrease the lag time betweendata download and provision of products to the Sea-BASS database.


46

Table 1. Summary of cruises during which in situ data have been obtained by MBARI in support of SIM-BIOS. All cruises were undertaken aboard the NOAA ship Ka’imimoana, with the exception of GP6-98-RBaboard the NOAA ship Ronald H. Brown. Meridional transects indicate the lines occupied by the ship. Alongeach line, CTD stations were performed approximately every degree of latitude from 8°N to 8°S. Measurementsconsisted of extracted chlorophyll (Chl) plus nitrate, phosphate and silicate (Nutrients) at 8 depths between 0 and200m. On selected cruises, primary productivity (PP) measurements were also made using 14C incubation tech-niques, and daily optical profiles with the Satlaintic Profiling Multispectral Radiometer (SPMR) were obtained.During GP7-98-KA, a SIMBAD radiometer was also used, on loan from the SIMBIOS instrument pool.

Cruise ID Dates Meridional transects MeasurementsGP6-97-KA 27-Sep-97 to 30-Oct-97 125°W and 140°W Chl, PP, Nutrients, SPMRGP7-97-KA 06-Nov-97 to 17-Dec-97 155°W, 170°W and 180° Chl, PP, Nutrients, SPMRGP1-98-KA 05-Feb-98 to 13-Mar-98 95°W and 110°W Chl, NutrientsGP2-98-KA 18-Apr-98 to 20-May-98 125°W and 140°W Chl, NutrientsGP3-98-KA 02-Jun-98 to 03-Jul-98 155°W and 170°W Chl, PP, Nutrients, SPMRGP4-98-KA 07-Jul-98 to 03-Aug-98 165°E and 180° Chl, NutrientsGP5-98-KA 05-Sep-98 to 09-Oct-98 125°W and 140°W Chl, NutrientsGP6-98-RB 11-Oct-98 to 13-Nov-98 95°W and 110°W Chl, NutrientsGP7-98-KA 19-Oct-98 to 13-Nov-98 155°W and 170°W Chl, PP, Nutrients, SPMRGP8-98-KA 18-Nov-98 to 12-Dec-98 180° and 170°W Chl, NutrientsGP1-99-KA 22-Jan-99 to 24-Feb-99 125°W and 140°W Chl, Nutrients

Figure 1. Time series of Sea Surface Temperature (SST – heavy shaded line), chlorophyll a averaged overthe upper 20m of the water column from moored spectroradiometer data (heavy solid line) SeaWiFS OC2V2chlorophyll (closed triangles) and ∆pCO2 (thin solid line).

-0.25

0

0.25

0.5

0.75

1

1.25

1.5

Dec-96 Mar-97 Jun-97 Sep-97 Dec-97 Mar-98 Jun-98 Sep-98

Date

Ch

l a [

ug

l-1

] / ∆∆

pC

O2

[uat

m/1

00]

22

24

26

28

30

32

Tem

per

atu

re [

°C]

in-situ chlSeaWiFSpCO2/100SST


47

Chapter 10

Remote Sensing of Ocean Color in the Arctic:Algorithm Development and Comparative Validation

Glenn F. CotaOld Dominion University

Norfolk, Virginia

10.1 INTRODUCTION

The Optical Research Consortium of the Arctic(ORCA) is a collaborative project between G.F. Cotaof Old Dominion University (ODU), T. Platt andW.G. Harrison of the Bedford Institute of Oceanogra-phy (BIO), S. Sathyendranath of Dalhousie Universityand S. Saitoh of Hokkaido University. ORCA initi-ated bio-optical data collections in August, 1994 andhas conducted 10 high latitude cruises in the BeringSea (7/95 = Ber95, 7/96 = Ber96), the Labrador Sea(10-11/96 = Lab96, 5-6/97 = Lab97, 6-7/98 = Lab 98),the Gulf of Alaska (10/97 = Goa97), and in theNorthwest Passage near Resolute Bay, NWT, Canada(8/94 = Res94, 8/95 = Res95, 8/96 = Res96, 8/98 =Res98). ORCA has now collected almost 500 in-water optical profiles and most of them have beensubmitted to NASA’s SeaBASS database.

The ORCA program depends upon coordinationand cooperation between international co-investigatorswith support from NASA, NASDA and Canadiansources. ODU has had primary responsibility for bio-optical and biogeochemical data collection with as-sistance from all collaborators. ODU has been re-sponsible for data processing, distribution and submis-sion to SeaBASS. Data analyses are being conductedby all co-investigators. Modeling is being done pri-marily by BIO and Dalhousie. Our algorithm devel-opment efforts encompass most of the potential dataproducts listed for SeaWiFS, MODIS and GLI, andaddresses product validation for a specific area ofconcern, i.e. high latitudes and coastal waters.

Observational suites depend primarily on fieldprogram duration, the sampling time, berths and per-sonnel available. Longer, dedicated field programs(i.e. Resolute) include more comprehensive observa-tions. Core observations always include in-waterspectroradiometer profiles for apparent optical prop-erties (AOPs) and discrete profiles of fluorometricchlorophyll. Absorption spectra for particulate, ex-tracted particulates, and soluble materials are alsoavailable from most cruises. The observational suitesfor Resolute and the Labrador Sea are typically morecomprehensive with in situ inherent optical property

(IOPs) profiles (AC-9, Hydroscat-6: Resolute only),above-water spectral remote sensing reflectanceRrs(λ), spectrophotometric and/or HPLC pigments,primary production and/or photosynthesis versus irra-diance, inorganic nutrients, 15N-nutrient uptake, par-ticulate organic carbon and nitrogen.


From May, 1997 through August, 1998 fourSIMBIOS validation cruises for OCTS or SeaWiFSwere conducted with one on an early-start supple-mental contract (Labrador Sea 5-6/97) and three underthe main contract (Gulf of Alaska 10/97, Labrador Sea6-7/98, and Resolute Bay 8/98). Our scheduledAugust, 1997 Resolute field program was cancelledbecause OCTS had failed and SeaWiFS was not yetoperational.

Bio-optical, biogeochemical and experimental ob-servations were made as frequently as logistical andweather considerations permitted. Data collectionsand processing for biogeochemical and optical obser-vations conform closely to Joint Global Ocean FluxStudies (JGOFS, 1991) and SeaWiFS protocols(Mueller and Austin 1992, 1995).

Water column profiling consisted of a bottle orpump cast for discrete samples preceded or followedby optical casts. Discrete water samples were col-lected from standard depths and depths correspondingto features of particular interest such as fluorescence,beam attenuation and/or density maxima. Biogeo-chemical observations included fluorometric chloro-phyll and phaeopigments, spectrophotometric (Reso-lute) or HPLC (Labrador Sea) pigments, absorptionspectra of particulates, methanol-extracted particu-lates, and dissolved colored organic material(CDOM), inorganic nutrients, particulate organic car-bon and nitrogen, and selected other observations suchas total suspended matter and cell count samples.Experimental observations at Resolute and the Labra-dor Sea included 14C and 13C primary production and15N-nutrient uptake of nitrate and ammonium.

Optical stations include hand-held sun photome-try and photographs of sea and sky conditions. In-


48

water optical observations were made with a SatlanticSeaWiFS Profiling Multichannel Radiometer (SPMR)and SeaWiFS Multichannel Surface Reference(SMSR). Downwelling irradiance Ed(0

-,λ) and up-welling radiance Lu(0

-,λ) were measured just belowthe sea surface at 13 wavebands with the SMSR sen-sors mounted on a floating, tethered buoy with an um-bilical cable for power and data transmission. TheSPMR also has 13 channels for vertical profiles ofdownwelling irradiance Ed(z, λ) and upwelling radi-ance Lu(z, λ), and was deployed in a free-fall modewith a kevlar cable for power and data transmission.The SPMR includes sensors for tilt and pitch, pres-sure, conductivity and temperature (Ocean Sensors),and fluorescence (WETLabs WetStar). The SMSRand SPMR measurements were always made concur-rently and the sensors were deployed 20-100 m fromthe vessel to avoid ship-shadow effects. Above-waterreflectance measurements of the sea surface weremade with an Analytical Spectral Devices (ASD) Per-sonal Spectrometer II (350-1050 nm, 1.4 nm band-pass,). Incident spectral irradiance Ed(0

+,λ) was esti-mated with a gray Spectralon card (10% reflectance)and water-leaving radiance Lu(0

+,λ) was corrected forreflected sky light (K. Carder and C. Davis, pers.Comm.). Spectrometers have been calibrated 1-2times per year depending upon the number of fieldcampaigns. Profiles of IOPs were measured with 1-2WET Labs AC-9 absorption and beam attenuationmeters with 9 channels and 25 cm path lengths fortotal and/or soluble constituents <0.2um. A HOBILabs Hydroscat-6 was used to measure backscatter in6 spectral bands. The latter three types of measure-ments were only made at Resolute.

AOPs, fluorometric chlorophyll and absorptionspectra data from all cruises have been submitted toSeaBASS, except the Lab98 data set. The downwel-ling irradiance Ed sensor malfunctioned during Lab98,and the data processing will require modifications ofSatlantic software or estimation of normalized water-leaving radiance Lwn from models. Additional ancil-lary data are also being submitted as they becomeavailable from our lab and collaborators.


High northern latitude waters have unique bio-optical properties compared with temperate data andmodels (Sathyendranath et al. 1999b). These waters

have a large dynamic range of biomass, which must beconsidered in algorithm development.

Results to date reveal that diatoms usually domi-nate phytoplankton assemblages, but prymenesiophyteblooms may alter bio-optical signatures (Sathyen-dranath et al. 1999a). The relatively large diatom cellsare highly packed with high chlorophyll-specific at-tenuation in the blue. Regional differences have beenfound in soluble attenuation with the highest values inthe Canadian Arctic Archipelago. Blue-green bandratio algorithms provide reasonable chlorophyll andnet daily primary production retrievals when properlytuned for high latitudes. However, the currentSeaWiFS OC2 chlorophyll algorithm underestimatesbiomass at moderate to high concentrations (Figure 1).Solar-induced fluorescence should provide reasonableinstantaneous estimates of primary production withrobust backscatter corrections. The universality ofany algorithms requires extensive evaluation and vali-dation.

10.4 WORK PLANNED

Cruise plans for 1999 are not yet finalized, butshould include a spring (May-June) Labrador Seacruise and one more to be determined. WithNOAA/NDBC, a time-series of bio-optical observa-tions from the Chesapeake Light Tower is being initi-ated. As proposed ORCA is conducting 1-3 cruisesper year, and contributing critical core observationaldata as well as a variety of ancillary bio-optical andbiogeochemical data.

Core observations have been delivered within 90days and include profiles of AOPs made with ourSatlantic profiling spectroradiometer, fluorometricchlorophyll, and normally particulate and dissolvedabsorption spectra. Additional observations for whichODU has primary responsibility for are being deliv-ered within 12 months, while contributions from in-ternational collaborators are submitted as they becomeavailable.

Our goal is to make raw data and derived prod-ucts generally available via the web but our site is stillunder construction. Efforts in 1999 will concentrate ondata analyses, modeling, and publication of results.Two manuscripts have been submitted and severalmore are in preparation.


49

Figure 1. The current best-fit line of our high latitude data (minus Lab98) versus the current SeaWiFS OC2chlorophyll algorithm.


50

Chapter 11

High Frequency, Long-Time Series Measurements fromthe Bermuda Testbed Mooring in Support of SIMBIOS

Tommy Dickey, S. Zedler and D. SigurdsonUniversity of California at Santa Barbara,

Santa Barbara, California

11.1 INTRODUCTION

The primary goal of the present activity is to pro-vide high frequency, long-term time series bio-opticaldata from the Bermuda Testbed Mooring (BTM) insupport of SIMBIOS (Dickey et al., 1998). The site islocated about 80km southeast of Bermuda at ~31° 43'N, 64° 10' W in waters of ~4530m depth.

The BTM study, which is funded by NASA, NSF,ONR, and NOPP, supports a variety of scientific re-search including SIMBIOS calibration and validationactivities. Our study contributes to the developmentof methodologies and capabilities for synthesizingdata derived from several satellite ocean color mis-sions (Esaias et al., 1995), particularly SeaWiFS. KeyNASA supported BTM optical measurements (every15min during daylight) presently include: surfacedownwelling spectral irradiance (7 λ's, SeaWiFSmatched) and subsurface (nominal depths of 14 and32m) downwelling spectral irradiance and upwellingspectral radiance (7 λ's for each, SeaWiFS matched).

Products include: radiance and irradiance ratios,spectral attenuation coefficients, and upwelling spec-tral radiances just beneath the surface (0-). These dataprovide necessary links to and interpolation of radio-metric data for remotely sensed observations of oceancolor (suite of satellite color imagers) which can beused to estimate biomass and primary productivityglobally.

Our newly developed telemetry system is nowproviding near real-time radiometric data to our lab.We are in the process of developing system softwarewhich will make key data available to the SIMBIOSproject on a daily basis.

Highly complementary, value-added optical, me-teorological, and physical data products are providedthrough our complementary NSF BTM study (e.g.,Dickey et al., 1998). Our comprehensive measure-ments can be used to relate bio-optical signals to in-herent optical properties and to biogeochemical meas-urements and to test interdisciplinary models of phy-toplankton biomass and productivity and carbon flux.


A major objective of the SIMBIOS program is toestimate normalized spectral water-leaving radiance,Lwn, by using in situ measurements. Our approach isto utilize moored instruments to obtain high temporalresolution measurements of fundamental radiometricvariables to estimate upwelled spectral radiance justbeneath the ocean surface, Lu(0-), which can be usedto determine Lwn.

The mooring methodology, which allows for col-lection of very high temporal resolution, long-termdata, has several advantages including the ability toobtain high numbers of "match-up" data for compari-sons with SeaWiFS (e.g., Mueller and Austin, 1992).However, the use of moorings for obtaining opticaldata is a relatively new approach and has been used byvery few groups (e.g., Dickey. 1991; Smith et al.,1991; Clark et al., 1997; Dickey et al., 1998).

Moored optical observations are more challengingthan their ship-based counterparts for several reasons(e.g., less frequent calibration, biofouling, etc.). Thus,it is desirable to intercompare fundamental radiomet-ric data and derived products based on nearly concur-rent, co-located data from moored and ship-based in-strumentation. We have been collaborating with Dr.David Siegel (UCSB), who has been collecting com-parable vertical profile data during approximatelymonthly (every other week in springtime) shipboardcruises near the BTM site.

Our BTM data products are being intercomparedwith ship-based data collected as part of Dr. Siegel'sBermuda Bio-optical Program (BBOP) program (Sie-gel et al., 1995a,b; Siegel and Michaels, 1996). Itshould be noted that the BTM and BBOP measure-ments are displaced in space and to a lesser extent intime. In addition, there are several other significantdifferences. Thus, correlations between the BTM andBBOP data sets should not necessarily be expected tobe great.

The BTM Deployment 7, 8, and 9 (May 3, 1997 -March 31, 1998) data sets encompass periods when 1)only the Japanese ocean color imager (OCTS) wascollecting data, 2) neither OCTS nor SeaWiFS was


51

collecting data, and 3) when only SeaWiFS was col-lecting data. The time series of meteorological, physi-cal, and optical data are described in Dickey et al.(1998). Time series measurements of surface and nearsurface optical properties were made using mooredoptical radiometer systems. Measurements includesurface spectral downwelling irradiance (Ed (0+, λ) atλ = 412, 443, 490, 510, 555, 665, and 683nm (10nmbandwidth at half power) at 6Hz for 20sec every15min during daylight and at midnight. Subsurface(nominal depths of 14 and 32m) downwelling irradi-ance (Ed (z, λ) and upwelling radiance, Lu (z, λ) dataat λ = 412, 443, 490, 510, 555, 665, and 683nm arecollected following the same regimen.

Our work is demonstrating that moored radiomet-ric instrumentation can be used to provide a largenumber of in situ match-up data for comparisons withSeaWiFS observations in near real-time. The presentapproach uses a multi-purpose mooring, the BTM, andis relatively inexpensive. The use of such mooringsfor optical measurements is relatively new, and thusseveral complicating issues need to be considered andevaluated. We are working with data from the BTMand are doing statistical comparisons of data obtainedfrom the BTM and those obtained nearly concurrentlyby the BBOP program. Despite several complicatingfactors, generally favorable correlations have beenobtained for fundamental radiometric variables. Re-

sults including estimates of Lu(0-) have been trans-ferred to the SeaBASS system. More detailed analy-ses and comparisons will continue to be done in thefuture. Particular areas of investigation will includecalibration, sampling issues (e.g., BBOP spatial vs.BTM temporal), and biofouling. In addition, it isnoteworthy that our near real-time data telemetry sys-tem has been functioning during the present Deploy-ment 10. Lu(0-) estimates for this deployment havealso been transferred to SeaBASS.

11.3 WORK PLAN

It is our intention to place an additional radiome-ter at 5-7m depth for a future deployment to determineif our estimates of Lu (0-) can be improved signifi-cantly. Deployment 10 full bandwidth data will berecovered in March 1999 and Deployment 11 will beinitiated.

ACKNOWLEDGMENTS

We wish to thank David Siegel and MargaretO'Brien for sharing the BBOP data used for intercom-parisons and for their assistance.


52

Chapter 12

The High-latitude Intercomparison andValidation Experiment (HIVE)

Dave L. EslingerInstitute of Marine Science, University of Alaska Fairbanks

(Present affiliation: NOAA, Charleston, South Carolina)

12.1 INTRODUCTION

The HIVE project is a three-year series of cruisesin both the Bering Sea and Prince William Sound/Gulfof Alaska. Both locations are at similar latitudes(~60°N) and have similar planktonic communities;however, the physical environments are dramaticallydifferent. These locations provide excellent test sitesto determine the performance of bio-optical algo-rithms at high latitudes. We are making measurementsof pigments (fluorometric and HPLC), radiometricprofiles, total suspended matter, and ocean aerosols.The HIVE project builds on additional funding fromthe Japanese ADEOS program and the EXXON Val-dez Oil Spill Trustee Council, which are supportingresearch projects in the Bering Sea and Prince WilliamSound, Alaska, respectively.

We are contracted to perform two or more cruisesper year in the Gulf of Alaska, Prince William Sound,and/or Bering Sea depending on availability of sam-pling opportunities on US and Japanese vessels. In1997, we received additional funding from the NASASIMBIOS program to participate in a collaborativeJapanese cruise in the Bering Sea prior to the start ofthe actual SIMBIOS project. In the first year of theSIMBIOS contract, we participated in three cruises,two in Prince William Sound and one in the BeringSea. This technical memorandum reports on all fourHIVE cruises to date.

The goal of the HIVE project is to validate high-latitude, satellite data products derived from imageryfrom different ocean color sensors. This goal is to beaccomplished through meeting the following objec-tives:

• Make accurate measurements of in situ data fieldsof potential ocean color data products in northern-hemisphere, high-latitude waters.

• Compare measured data fields with data fieldsdetermined from ocean color satellites.

• Intercompare data fields retrieved from differentocean color satellites.

• Intercompare data fields retrieved from the samesatellite at different times of day.

The failure of the ADEOS spacecraft precludedus from getting any match-up OCTS data during our1997 cruise, which was prior to the SeaWiFS launch.In 1998, clouds, fog and poor weather have preventedany match-ups with the SeaWiFS data sets. Thereforepoor weather and the ADEOS failure have preventedus from making any progress towards objectives 2 and4. However, we have been successful in making fieldmeasurements, even under less-than-ideal conditions.


We are making field measurements of phyto-plankton pigments using both fluorometric and HPLCtechniques, profiles of upwelling radiance and down-welling irradiance, total suspended matter, and oceanaerosols. These activities are described in detail be-low. We describe our activities in four cruises. Thefirst occurred in July, 1997, in the Bering Sea, aboardthe Japanese training ship Oshoro Maru and has acruise identifier of OS97. The second was in June,1998 was in Prince William Sound, AK, aboard a 58-foot charter vessel, the M/V Auklet and has a cruiseidentifier of AUK9806. The third was in July and wasback to the Bering Sea aboard the T/S Oshoro Maruand has a cruise identifier of OS98. The fourth HIVEcruise was in August, again in Prince William Soundaboard the M/V Auklet, and has identifier AUK9808.

On all cruises, water samples were collected from6 depths in the upper water column using Niskin bot-tles. During AUK9806, samples were taken from thesurface and 5, 10, 15, 25 and 50 meter depths. Duringthe OS97, OS98, and AUK9808 cruises, samplingdepths were selected based on standard light levelsand were generally taken at the depths correspondingto1, 10, 20, 36, 50, and 95% of surface light. Depthswere altered when the water column was too shallowto permit sampling to the one-percent light level. Sur-face samples generally came from one-meter depth,but were occasionally deeper under rough sea condi-tions, when necessary to avoid firing bottles above thesurface.


53

Phytoplankton Pigments

Water samples were filtered under low vacuumonto GF/F filters, followed by freezing in liquid nitro-gen for HPLC, and by grinding and extraction in 90%acetone for immediate fluorometric analyses. Chloro-phyll and phaeophytin concentrations were measuredin the acetone extracts aboard the ship immediatelyafter centrifugation using the technique of Stricklandand Parsons (1972). HPLC samples were shipped inliquid nitrogen back to Fairbanks, where they werestored. Our group does not have the necessary facili-ties to perform HPLC analyses, therefore, after stor-age, samples from the first three cruises were shippedvia FedEx to the Grice Marine Lab, Charleston, SC,for analysis. A combination of inadequate packingand shipment delays in both pick-up and transit led tothe samples arriving in Charleston at room tempera-ture. HPLC analysis of a subset of these filters indi-cated significant degradation of pigments had oc-curred and the samples were discarded. The filtersfrom the fourth cruise remain in Fairbanks, in liquidnitrogen, while we determine exactly where the delaysin handling and timing occurred and how to preventthem in the future.

Total Suspended Mater Concentrations

Water samples were taken for determination oftotal suspended matter (TSM) according to the JGOFSprocedures (JGOFS, 1991) as outlined in the SeaWiFSprotocols (Mueller and Austin, 1995). A modificationof this was to use pre-combusted glass fiber filters.Filtered samples were dried and returned to Fairbanksfor weighing on a precision analytical balance.Weighed dry filters were then recombusted andreweighed to give both total suspended matter andorganic and inorganic fractions.

Radiance and Irradiance Profiles

For OS97, we were able to borrow a BiosphericalInstruments MER-2048 system from the NOAACoastal Services Center. Using this MER, we mademeasurements of Ed at 340, 380, 395, 412, 443, 455,490, 510, 532, 555, 565, and 671 nm. Eu and Lu pro-files measurements were made at 380, 412, 443, 490,510, 555, and 683 nm. For surface reference, Es wasmeasured at 340, 380, 395, 412, 443, 490, 510, 555,565, 671, 780, and 875 nm. Temperature was alsomeasured during these profiles.

For AUK9806 cruise Satlantic loaned us anSPMR system (serial # 022). The profiler measuredED at 405.4, 411.9, 435.6, 442.8, 455.4, 489.6, 510.4,531.9, 554.8, 590.5, 665.6, 669.6, 700.4 nm and Lu at405.1, 411.9, 435.1, 443.5, 456.3, 489.4, 510.3, 531.5,554.3, 590.5, 665.3, 670.2, and 700.3 nm. The surface

reference measured Ls at 406.5, 412.2, 435.3, 443.4,455.9, 489.9, 510.4, 531.6, 554.6, 590.3, 665.1, 670.0,and 700.6 nm and Es 405.1, 412.4, 435.6, 442.9,456.1, 489.3, 510.4, 531.5, 554.5, 590.4, 664.8, 670.0,and 700.6 nm. This system was only used for theAUK9806 cruise. Due to failure of a PCMCIA serialport card, both the profiler and surface reference couldnot be used simultaneously. Instead, we made ameasurement of Es and Ls, did three profiles, and thenrepeated the surface reference measurements. Giventhe heavy overcast conditions during this cruise, wedo not expect the surface light values to have variedsignificantly during the time we were profiling.

For the OS98 and AUK9808 cruises, we used ournewly purchased SPMR system, which simultaneouslymeasured Ed, Lu, Es temperature, conductivity, and invivo fluorescence profiles. Ed was measured at 412.3,443.1, 490.7, 510.0, 555.1, 665.6 and 682.4 nm. Lu

was measured 412.3, 442.1, 490.9, 510.1, 554.9,665.4, and 682.9 nm. Es was measured at 412.3,442.6, 490.3, 510.2, 554.4, 665.0, and 683.8 nm.

Instrument Calibration

All instruments have been calibrated as requiredby the SeaWiFS protocol and SIMBIOS contract.Calibration files have been submitted with the bio-optical data to SeaBASS.


Table 1 summarizes the cruise time, location, sta-tion, and measurement data. All cruises had completeovercast at almost every station. Although these highlatitudes are generally cloudy, this was an exception-ally bad year for our cruises. In 1994 through 1996cruises, we had much better weather at these times andexpected similar intermittent clouds these last twoyears.

12.4 WORK PLAN

This last year we participated in three cruises in-stead of only the two in which we were contracted toparticipate. Next year, we plan to consolidate our two5-day Prince William Sound cruises into one 10-daycruise. By increasing our at-sea time to ten days, wehope to get the satellite match-ups that we havemissed up to the present time. At the present, we planto schedule the cruise for middle to late April, to cap-ture the high, spring-bloom phytoplankton concentra-tions. However, we will review the previous yearsAVHRR data to try and determine if expecting cloud-free days during this time period is realistic. We willstill plan to participate in the T/S Oshoro Maru cruisein July.


54

Table 1. Summary of cruises and in situ data collected.

Cruise ID Location Cruise Dates Numberof Stations

Non-opticalmeasurements*

OS97 Bering Sea 18 July –2 Aug.1997 20 Pigments (fluor),TSM, T

AUK9806 Prince WilliamSound

5 –9 June 1998 21 Pigments (fluor),TSM, T

OS98 Bering Sea 18 July –1 Aug. 1998 17 Pigments (fluor),TSM, T, S, Fluor

AUK9808 Prince WilliamSound

8 Aug. – 2 Sept., 1998 15 Pigments (fluor &HPLC), TSM, T, S,Fluor

* TSM = Total Suspended Matter, T = temperature profile, S = Conductivity profile, Fluor = in vivo fluo-rescence profile


55

Chapter 13

Satellite Ocean Color Validation Using Merchant Ships

Robert Frouin and David L. CutchinScripps Institution of Oceanography

University of California San Diego, La Jolla, California

Pierre-Yves DeschampsLaboratoire d'Optique Atmospherique

Universite des Sciences et Technologies de Lille, Villeneuve d'Ascq, France

13.1 INTRODUCTION

The usual approach to satellite ocean color vali-dation is to organize dedicated experiments at sea, e.g.in regions that are likely to cause atmospheric correc-tion errors, and to measure concomitantly water-leaving radiance and aerosol optical properties at thetime of satellite overpass (Clark et al., 1997). The dataare necessary not only to compare satellite estimateswith in situ measurements, but also to interpret thedifferences and, eventually, adjust atmospheric cor-rection algorithms. These experiments, unfortunately,are expensive; they cannot be carried out over the fullrange of expected atmospheric and oceanic conditions.Furthermore, because of cloud cover only a fewmatch-ups are generally obtained during a campaign,making the approach even less cost-effective.

A complementary approach --and the focus of thepresent project-- is to use ships of opportunity, in par-ticular, merchant ships traveling routes in the world'soceans. For the approach to be cost-effective, the shipsmust participate on a volunteer basis or at a very smallcost, and the measurement program must not interferewith the ship’s activities. Importantly, the ship shouldnot be required to stop, i.e. one should be able to col-lect the data en route, while the ship is steaming. Be-cause of these limitations, one cannot expect to meas-ure the full set of ocean-color variables using ships ofopportunity. Still, to be useful the measurement pro-gram should meet the basic requirements for satelliteocean color validation.

It is in this context, and also according to a studyof the measurements required for satellite ocean-colorvalidation (Schwindling et al., 1998), that the SIM-BAD concept was developed. SIMBAD is a portableradiometer designed for use onboard ships of opportu-nity. It measures normalized water-leaving radiance(or diffuse marine reflectance) and aerosol opticalthickness in typical spectral bands of satellite oceancolor sensors. Knowledge of both variables is neededand generally sufficient to evaluate atmospheric cor-

rection algorithms (Schwindling et al., 1998). Theinstrument is easy to operate, so that any ordinarycrew can learn quickly how to make measurements.To achieve adequate sampling, i.e. acquire data in awide range of oceanic and atmospheric conditions, aseries of ten radiometers was built for use in two com-plementary networks of ships/routes, one operated bythe Scripps Institution of Oceanography, the other bythe University of Lille.


The SIMBAD radiometer was designed and built bythe Laboratoire d’Optique Atmospherique (LOA) ofthe University of Lille. It measures water-leaving ra-diance from above the surface as opposed to below thesurface the usual way. This eliminates constraints as-sociated with deploying underwater instruments atsea, e.g. stopping the ship. Viewing of the ocean sur-face is made from the side of the ship lit by the sun,outside the glitter region, at a nadir angle of 45 de-grees. A vertical polarizer reduces the skylight re-flected by the surface in the instrument’s field-of-view. The best location to operate the instrument isgenerally the bow, making it easy to avoid ship-generated foam. The radiometer also measures aerosoloptical thickness, by viewing the sun like a standardsun photometer. The same optics (2.5 degree field-of-view) and detectors are used in both ocean- and sun-viewing modes, but at different electronic gain.

The radiometric measurements are made simulta-neously in five spectral bands centered at 443, 490,560, 670, and 865 nm. These bands are not exactlythose of the ocean color sensors that are flying orscheduled to fly, but result from a compromise. Theyconstitute, however, a basic set of bands for atmos-pheric correction and bio-optical algorithms. Viewingof the ocean and sun is made sequentially, but thenon-simultaneity of the measurements is not limiting.SIMBAD does not measure incoming solar irradiance,a variable that must be known to normalize water-


56

leaving radiance. It is calculated accurately in clearsky conditions or when the sky is partly cloudy(<30%) with the sun not obscured by clouds. Thisdefines the sky conditions under which SIMBADshould be operated. In addition to radiometry, SIM-BAD automatically acquires viewing angles (using amagnetometer and two inclinometers), time and geo-graphic location (using a GPS), pressure, and tem-perature. It requires about 15 minutes to collect acomplete data set (ocean, sun, optical zero), includingdeploying the instrument and logging ancillary data(wind speed, sea state, cloud cover, etc.). Internalmemory and batteries allow for 3 months of opera-tions in normal mode (typically 5 sets of measure-ments per day). At the end of the campaign or voyage,after the ship returns to port, the data are downloadedonto computer disk and diskette and the radiometer iscalibrated. In the case of long campaigns, the data aremailed from selected ports along the ship’s route, al-lowing for a more rapid data turnover.

Justification and Verification

The principle of the SIMBAD radiometer, in par-ticular its ability to reduce reflected skylight with thehelp of a vertical polarizer, has been justified theoreti-cally and verified experimentally (Fougnie et al.,1999b). Theoretical calculations show that, for a nadirangle of 45 degrees and a relative azimuth angle be-tween solar and viewing directions of 135 degrees, therecommended geometry, reflected skylight is reducedto typically 10-3 in reflectance units at 443 nm. Thisrepresents 2 to 10% of the diffuse marine reflectance,the signal of interest. Furthermore, the effects of sur-face roughness on skylight reflection and, hence, un-certainties in sea state (wind speed) are minimized.Taking into account typical uncertainties in windspeed and geometry, the residual reflected skylight arecorrectable to a few 10-4 in reflectance units. For mostoceanic waters, the resulting error on the diffuse ma-rine reflectance in the blue and green is less than 1%.The theoretical results have been verified experimen-tally at the Scripps Institution of Oceanography pier,by viewing the ocean surface with a scanning polari-zation radiometer (Fougnie et al., 1999b). The variousangular and spectral effects predicted by theory havebeen evidenced in the measurements.

Since the water body polarizes incident sunlight(e.g., Ivanoff, 1975, Frouin et al., 1994), the measuredpolarized radiance differs from the total radiance. Forthe recommended geometry, the underwater scatteringangle varies only between 148 and 158 degrees, andthe maximum polarization factor varies between 0.83and 0.92 (Fougnie et al., 1999b). One should be ableto correct the polarization effects to within 5% relativeaccuracy, which compares with other errors, such asthose due to bi-directional effects and radiometriccalibration.

The SIMBAD radiometer was evaluated at sea,during California Cooperative Fisheries Investigation(CalCOFI) cruises in the southern California Bight.The diffuse marine reflectance measured by the radi-ometer was compared with that measured by an un-derwater MER optical system. The agreement be-tween the two types of measurements was good, towithin 1.6 10-3 in reflectance units. Aerosol opticalthickness measured by the SIMBAD radiometer alsocompared well with that measured by sun photometersduring the October-November 1996 CalCOFI cruise(Nakajima et al., 1999) and during the Second AerosolCharacterization Experiment (ACE-II).

The SIMBAD radiometer proved useful to checkvicariously the radiometric sensitivity of the Polariza-tion and Directionality of the Earth’s Reflectance(POLDER) instrument onboard ADEOS (Fougnie etal., 1999a). The satellite radiance was compared withthat computed for the same geometry using a radia-tion-transfer model. SIMBAD measurements of aero-sol optical thickness and marine reflectance at the timeof satellite overpass were used as input to the model.The accuracy of the vicarious calibration coefficientswas estimated to be better than 3%. A large decreaseof the POLDER instrument response was found in theblue, confirming the results previously obtained usingalternative calibration techniques.

Data Format

The SIMBAD data sets are fairly completefor satellite ocean color validation. They include near-concurrent values of spectral aerosol optical thicknessand diffuse marine reflectance, necessary variables forchecking radiometric calibration and evaluating at-mospheric correction algorithms. They also includefractional cloud coverage and wind speed (useful toestimate whitecap reflectance). The data sets can beaccessed at the following web addresses:http://genius.ucsd.edu/~frouin/ and http://www-loa-univ-lille1.fr/~poteau/.


Since the launch of the SeaWiFS in September1997 and until December 1998, SIMBAD measure-ments have been made during 12 research cruises and9 merchant ship voyages (Table 1). A total of 156SIMBAD data sets have been collected under clear ormostly clear sky (fractional cloud coverage less than0.2) within a few hours of satellite overpass. A varietyof oceanic and atmospheric conditions have beensampled, but no data have been acquired in the SouthAtlantic and Indian Oceans. The research cruises andmerchant ship voyages are listed below in chronologi-cal order with location or route, date, name of theSIMBAD operator, and network (SIO or LOA).


57

Table 1. SIMBAD cruise measurements from 9/1997 to 12/1999

CalCOFI 9709, Southern California Bight, 09/20/98-10/08/98, Wieland, SIO.M/V Asian Challenger, Valparaiso-Auckland, 10/12/98-10/18/98, Cutchin, SIO.R/V Melville, San Diego-Acapulco-Callao-Valparaiso-Easter-Papeete-Honolulu, 11/02/97-06/07/98,Comer, SIO.M/V Nacre, Honolulu-Coronel, 12/20/98-01/02/98, Cutchin, SIO.CalCOFI 9802, Southern California Bight, 01/23/98-02/10/98, Subranamian, SIO.CalCOFI 9804, Southern California Bight, 04/02/98-04/24/98, Mitchell, SIO.R/V Revelle, Eastern Tropical Pacific Ocean, 04/09/98-04/27/98, Dupouy, LOA.M/V Toucan, Le Havre-Cayenne-le Havre, 05/04/98-05/23/98, Poteau, LOA.M/V Chevron Mississippi, Honolulu-Valdez, 05/18/98-05/23/98, Cutchin, SIO.M/V Toucan, Le Havre-Cayenne-le Havre, 06/08/98-06/20/98, Poteau, LOA.CalCOFI 9807, Southern California Bight, 07/09/98-07/27/98, Wieland, SIO.ATMARA 9807, English Channel, 07/21/98-08/06/98, Wartel, LOA.NAN 9807, Cape Cod, 07/06/98-07/10/98, Shieber, SIO.M/V Mokihana, Oakland-Guam-Taiwan-Japan-Oakland, 07/20/98-08/15/98,Bousseloub, SIO.CalCOFI 9809, Southern California Bight, 09/13/98-10/01/98, Wieland, SIO.BREST 9809, Northeastern Atlantic Ocean, 09/25/98-10/01/98, Gohin, LOA.M/V Mokihana, Oakland-Guam-Taiwan-Japan-Oakland, 08/25/98-09/08/98, Bousseloub, SIO.M/V Micronesian Navigator, California-Micronesia-Philippines-California,08/25/98-10/28/98, Sumogat, SIO.R/V Cape Hatteras, Beaufort-Jacksonville, 10/21/98-12/01/98, Subranamian, SIO.M/V Micronesian Navigator, California-Micronesia-Philippines-California, 10/30/98-12/31/98, Sumogat, SIO.R/V Ka’imimoana, Honolulu-Fiji-Honolulu, 10/19/98-12/11/98, Strutton, SIO.

13.4 WORK PLAN

A collaborative LOA-SIO measurement programfor satellite ocean color validation based on ships ofopportunity has been developed. This has been ac-complished in a series of successive, logical steps.First, a study of validation requirements was per-formed, indicating that measurements of spectral wa-ter-leaving radiance and aerosol optical thickness atthe time of satellite overpass are generally sufficient toevaluate satellite-derived ocean color (Schwindling etal., 1999). Second, in view of the study and limitationsof dedicated validation experiments at sea, the SIM-BAD radiometer was designed, for use of ships onopportunity such as merchant ships traveling regularroutes in the world’s oceans. Third, the instrument’sinnovative principle (use of a vertical polarizer to re-duce skylight reflection) was verified experimentallyat the Scripps Institution of Oceanography pier (Foug-nie et al., 1999b). Fourth, the instrument was tested atsea, on station and en route, during various oceano-graphic campaigns (Fougnie, 1998, Nakajima et al.,1999). Fifth, suitable merchant ships and routes in thePacific and Atlantic Oceans were identified and se-lected officers were trained. Some of the ships, forexample M/V Micronesian Navigator (SIO network)and M/V Toucan (LOA network), are now providingSIMBAD data routinely. Sixth, SIMBAD data wereused to check vicariously the radiometric calibrationof the POLDER instrument onboard ADEOS (Fougnieet al., 1999a).

Regarding measurement opportunities, the pro-gram’s focus has been on research campaigns during

which water-leaving radiance and/or aerosol opticalthickness were measured by other instrumentation(e.g., CalCOFI campaigns), so that comparisons couldbe made. Although more evaluation of the SIMBADdata is necessary, and should be done on a continuousbasis, in order to increase the number of sur-face/satellite match-ups the focus should be shifted toships that do not carry out the same type of measure-ments as the SIMBAD radiometer (e.g., more mer-chant ships). The program also needs to be extendedto ships traveling in other oceanic regions, in particu-lar the southern Oceans and the Indian Ocean, to com-plement the present coverage. There are many possi-bilities, and the demand for SIMBAD radiometers hasbeen increasing. The realization of a new series ofradiometers is envisioned, with the instruments avail-able for the next generation of satellite ocean colorsensors.

ACKNOWLEDGMENTS

The authors wish to thank all the officers and sci-entists that have collected SIMBAD data, P. Lecomteand M. Verwaerde of the University of Lille for real-izing the SIMBAD radiometer, B. Fougnie, J.McPherson, and A. Poteau for processing the data.This work has been supported by the National Aero-nautics and Space Administration under ContractNAS-97135 (to R. Frouin), the Centre Nationald’Etudes Spatiales, the Centre National de la Recher-che Scientifique, and the Region Nord-Pas de Calais.


58

Chapter 14

High Attitude Measurements of Radiance at High Spectraland Spatial Resolution for SIMBIOS Sensor

Calibration, Validation and Intercomparisons

Robert Green and Thomas G. ChrienNASA, Jet Propulsion Laboratory

Pasadena, California

14.1 INTRODUCTION

The successful combination of data from differentocean color sensors depends on the correct interpreta-tion of signal from each of these sensors. Ideally, thesensor measured signals are calibrated to geophysicalunits of spectral radiance, and sensor artifacts are re-moved and corrected. The calibration process re-samples the signal into a common radiometric dataspace so that subsequent ocean color algorithms thatare applied to the data are based in physical processesand inherently sensor independent. The objective ofthis SIMBIOS effort: “High Altitude Measurements ofRadiance at High Spectral and Spatial Resolution ForSIMBIOS Sensor Calibration, Validation, and Inter-comparisons is to conduct a series of direct calibrationcomparisons using the NASA Airborne Visi-ble/Infrared Imaging Spectrometer (AVIRIS),” is toassess the calibration of the satellite sensors whichparticipate in SIMBIOS.

The AVIRIS sensor flies at 20 km altitude on aNASA ER-2 near the top of the absorbing and scat-tering portions of the atmosphere. AVIRIS measuresthe solar reflected spectrum from 370 nm to 2500 nmthrough 224 contiguous 10 nm wide spectral channels.Data are acquired as images of 11 by up to 800 kmextent with 20 m spatial resolution. The stability andrepeatability of AVIRIS calibration has been and con-tinues to be validated through a series of inflight cali-bration experiments. With pre and post flight calibra-tions of AVIRIS, coupled with the on-board calibra-tor, calibration accuracy of better than 2% spectral,3% radiometric and 3% spatial have been achieved.

The activities and results presented here are pri-marily for SeaWiFS. However, related work has beencarried out for ADEOS OCTS. In the future, it isplanned that AVIRIS will underfly EOS.


On the 2nd of October 1997 AVIRIS under flewthe SeaWiFS sensor of the coast of Southern Califor-nia near latitude 35° and longitude 127°. The

SeaWiFS sensor was tilted to the Northeast to avoidsun glint. In order for the AVIRIS spectral images tooverlap the area, the azimuth, and the zenith angles ofthe SeaWiFS measurements, AVIRIS was flown in acircle beneath SeaWiFS. The AVIRIS scan angle (±15 degrees) plus the aircraft roll angle (20 degrees)assured overlap of both area and observation geome-try. The AVIRIS data from the 2nd of October 1997flight were spectrally, radiometrically and spatiallycalibrated through the AVIRIS data system algo-rithms. The SeaWiFS data and the SEADAS proc-essing software were requested and received. Boththe AVIRIS and the SEAWIFS data sets include theposition and pointing information at the time of acqui-sition. The simultaneous acquisition of the AVIRISand SeaWiFS with overlapping observation geometryprovides the essential data set to complete the calibra-tion objective.


Determination of the areas of the SeaWiFS andAVIRIS images with the same observation geometriesrequires projection of the data based on the positionand pointing knowledge of the sensor. A set of algo-rithms were developed to accept the SeaWiFS positionand point information and calculate the azimuth andzenith angles for the 2nd of October SeaWiFS data.All angles were calculated from the perspective of thesurface. The observation geometry at the surface isthe correct geometry for calibration comparison.AVIRIS receives and records both Global PositioningSystem (GPS) and Inertial Navigation System (INS)information from the ER-2 aircraft. The GPS pro-vides latitude, longitude and altitude information,while the INS provides roll, pitch, and yaw informa-tion. These position and pointing data were used withan analogous set of algorithms to calculate theAVIRIS spectral image azimuth and zenith angleswere calculated. With both the AVIRIS and SeaWiFSimage azimuth and zenith angles determined, the tar-get area of overlapping observation geometry wasidentified.


59

This area contains the AVIRIS spectra and SeaWiFSmulti spectral data of the same area on the surfacewith the same observation azimuth and zenith (Table1).

The average AVIRIS spectrum from the area ofobservational and temporal coincidence was extracted.A correction factor was applied to the AVIRIS spec-tra for the transmittance from the 20 km AVIRIS alti-tude to the top of the atmosphere. This correctionfactor was calculated with the MODTRAN radiativetransfer code using Ozone values from the TOMStotal ozone archive for the area.

The SeaWiFS multi spectral band data were ex-tracted directly for this target. A graphical comparisonof the AVIRIS spectrum and SeaWiFS bands showedgood agreement. In order to quantitatively comparethe AVIRIS and SeaWiFS data, a convolution algo-rithm was developed to convolved the AVIRIS spectrato the SeaWiFS bands. The algorithm was applied tothe AVIRIS spectra for the target area and theAVIRIS predicted radiances compared to theSeaWiFS reported radiances (Table 2). These resultsare preliminary. A review of the version of SEADASused to process the SeaWiFS data is underway. Ahigh priority activity has begun to review and comparethe absolute radiometric standard used for SeaWiFScalibration and AVIRIS calibration.

The primary objective of this SIMBIOS effort hasbeen achieved. On the 2nd of October 1997, AVIRISunderflew the SeaWiFS sensor. The data set was ac-quired such that the observation geometries ofSeaWiFS and AVIRIS could be match for direct com-parison of measured radiance. A set of algorithmswere developed to calculate the surface azimuth andzenith angles for both the SeaWiFS and AVIRIS datasets.These algorithms required as input position andpointing information for the two sensors. The initialgraphical comparison of the AVIRIS spectrum andSeaWiFS bands showed good agreement for the oceantarget of overlapping observation geometry. An addi-tional set of algorithms were developed to convolvedthe AVIRIS spectra to the SeaWiFS spectral bands.Direct comparison of the AVIRIS measured andSeaWiFS measured radiance shows the agreement is

best for SeaWiFS bands 3,4,5, and 6 and least goodfor SeaWiFS bands 1, 2, 7, and 8. These are initial,preliminary results and additional review and uncer-tainty analysis is required to strengthen the basis ofcomparison. Never-the-less, these initial results vali-date the objective, approach, and activities of thisSIMBIOS Project.

14.4 WORK PLAN

In 1999 three additional underflights ofSEAWIFS are planned. The first underflight isplanned for the Spring of 1999 from the west coast. Asecond underflight will be attempted from the eastcoast in the late Spring or early Summer. The thirdunderflight is planned for the west coast in the latesummer early fall time frame. Each underflight willinclude two flight circle acquisitions to assure a matchto the SeaWiFS viewing geometry. The first flightcircle will be timed to coincide with the satellite over-pass. The second flight circle acquisition enables in-vestigation of the time sensitivity of the measure-ments. Efforts will be made to coordinate all under-flights with other participating SIMBIOS investiga-tors. Following each successful underflight, theAVIRIS runway calibration standard will be used toplace the maximum accuracy on the AVIRIS radio-metric calibration for the SeaWiFS underflight. InApril of 1999, the new SIMBIOS SXR-II transferradiometer will be used at JPL to further compare theSIMBIOS and AVIRIS radiometric standards. Resultsof all the 1999 activities of this project will be pro-vided to SIMBIOS. Work will proceed on writing upthe 1998 and 1999 results for publication.

ACKNOWLEDGMENTS

This research was carried out at the Jet PropulsionLaboratory / California Institute of Technology, Pasa-dena, California, under contract with the NationalAeronautics and Space Administration.


60

Table 1. Preliminary results of AVIRIS and SeaWiFS radiance comparison.

BAND SeaWiFS AVIRIS AVIRIS DifferenceRadiance Radiance Uncertainty

1 7.800 6.791 5.1% 12.9%2 6.852 6.311 3.5% 7.9%3 4.899 4.861 2.1% 0.8%4 3.995 3.972 2.5% 0.6%5 2.753 2.711 2.9% 1.5%6 1.248 1.193 2.3% 4.5%7 0.641 0.601 3.6% 6.2%8 0.449 0.412 3.5% 8.3%

The units of radiance are microWatts/cm2/nm/sr.

Table 2. Website locations for AVIRIS SeaWiFS match up example

1 First AVIRIS flight circle. ftp://makalu.jpl.nasa.gov/pub/rog/seawifs/pic01.jpg2 Corresponding SeaWIFS image ftp://makalu.jpl.nasa.gov/pub/rog/seawifs/pic04.jpg3 SeaWIFS azimuth ftp://makalu.jpl.nasa.gov/pub/rog/seawifs/pic05.jpg4 SeaWIFS zenith ftp://makalu.jpl.nasa.gov/pub/rog/seawifs/pic06.jpg5 AVIRIS azimuth ftp://makalu.jpl.nasa.gov/pub/rog/seawifs/pic07.jpg6 AVIRIS zenith ftp://makalu.jpl.nasa.gov/pub/rog/seawifs/pic08.jpg7 AVIRIS and SeaWIFS data measured within 5 seconds.

ftp://makalu.jpl.nasa.gov/pub/rog/seawifs/pic09.jpg


61

Chapter 15

Merging Ocean Color Data from Multiple Missions

Watson W. GreggNASA/Goddard Space Flight Center

Greenbelt, Maryland

15.1 INTRODUCTION

In response to the potential importance of phyto-plankton in the global carbon cycle and the lack ofcomprehensive data, the National Aeronautics andSpace Administration (NASA) and the internationalcommunity have established high priority satellitemissions designed to acquire and produce high qualityocean color data. Seven of the missions are routineglobal observational missions: the Ocean Color andTemperature Sensor (OCTS), the Polarization andDirectionality of the Earth's Reflectances sensor(POLDER), SeaWiFS, Moderate Resolution ImagingSpectrometer-AM (MODIS-AM), Medium ResolutionImaging Spectrometer (MERIS), Global Imager(GLI), and MODIS-PM. In addition, there are severalother missions capable of providing ocean color dataon smaller scales.

Any individual ocean color mission is limited inocean coverage due to sun glint and clouds. For ex-ample, one of the first proposed missions, theSeaWiFS, can provide about 45% coverage of theglobal ocean in four days and only about 15% in oneday (Gregg and Patt, 1994).

The proliferation of missions suggests that betterocean coverage can be obtained in less time if the dataare combined. In addition to improved, and fastercoverage, data can be taken from different local timesof day if the missions are placed in different orbits,which they are. This can potentially lead to informa-tion on diel variability of phytoplankton abundances.

We propose to investigate, develop, and test algo-rithms for merging ocean color data from multiplemissions. We seek general algorithms that are appli-cable to any retrieved ocean color data products, andthat maximize the amount of information available inthe combination of data from multiple missions. Mostimportantly, we will investigate merging methods thatproduce the most complete coverage in the smallestamount of time, nominally, global daily coverage. Wewill also assess the ability to produce fuller coveragein larger time increments, including 4-day, 8-day(weekly), monthly, seasonal and annual. Secondarily,we will investigate the ability of the missions to pro-duce coverage at different times of day, for diel vari-ability and dynamical evaluations, and develop algo-rithms to produce this information.


Work in the first two years has focused on: a)defining the problems and opportunities provided bythe existence of multiple sensors; b) analysis of pastsensors to provide insights into the characteristics,drawbacks, and advantages of individual sensor re-sponses in the context of merging; and c) analysis,development, and testing of candidate merger algo-rithms.

a) Defining the Problems

First we assessed the global coverage improve-ments possible by the six global missions. These re-sults show that significant scientific advantages canaccrue from assembling and merging data from themultiple satellite platforms proposed for ocean colorin the next decade. The principal advantages are in-creased ocean coverage in less time, and new obser-vations of the daily dynamics of phytoplankton abun-dances, resulting from different observation times ofco-located ocean areas. Data from three satellites asopposed to one, can increase ocean coverage by 58%for one day, and 45% for four days. Additional satel-lites produce diminishing returns, however. This latterpoint is not necessarily an adverse finding, since eachmission has a limited life expectancy and in-flightproblems are, unfortunately, still not rare in the satel-lite business. Using observations from pairs of mis-sions, as much as 14 hour time differences at co-located points can be realistically achieved at highlatitudes, even considering the distribution of landmasses and ice cover. Smaller time differences areobserved at lower latitudes, but 5 to 7 hour differencesare still available. Furthermore, massive numbers ofthese co-located pairs are available, suggesting thatroutine scientific studies of diel phytoplankton vari-ability can be supported (Gregg et al., 1998). Moredetailed analyses emphasized seasonal and regionalcoverage improvements and was limited to theSeaWiFS/MODIS combination, since these are thetwo missions planned for next launch. The resultsshowed that the launch of EOS AM-1 provides anopportunity to improve ocean color observations bycombining data from the SeaWiFS and MODIS mis-


62

sions. The sensors have different scanning character-istics and are flying on different platforms in differentorbits. The results suggested that very large im-provements in coverage frequency (daily to four-day)can be obtained by combining data from both sensors:40-47% increases in global coverage over SeaWiFSalone in one day, and > 100% in areas near the solardeclination. Four-day increases are slightly smallerfor global coverage, 29-35%, but meridional percentincreases are similar to the one-day case. These dif-ferences are due to reduced sun glint contaminationobtained by tilting (SeaWiFS), and scanning awayfrom the maximum glint region (MODIS),due to its10:30 AM equator crossing time, and due to the largescan width of MODIS. The results show thatSeaWiFS and MODIS are very complementary oceancolor missions that can provide more complete obser-vations of ocean processes at high frequency if dataare combined (Gregg and Woodward, 1998). A moredetailed paper with more fully defined results waspublished as a NASA TM (Woodward and Gregg,1998). Third, assessed potential capability for im-proving ocean color observations by selecting com-plementary mission orbits. Considering that observa-tions are severely hampered by cloud cover and sunglint, a possibility exists for using multiple missions toimprove the coverage of the oceans in shorter timescales. In fact, only about 10-18% of the oceans areobserved in a single day, even by so-called globalobservational missions, due to these two ocean colorcontaminants. Analysis of a 7-day period of cloudcover from the International Satellite Cloud Climatol-ogy Project (ISCCP), show that 12% new surface areais available for viewing each day. This translates toan increase of about 0.5% per hour of separation inviewing times. Thus if the ocean could be viewed by2 different satellites 4 hours apart, 2% more oceanarea could be observed. This represents a coverageincrease of about 13%. If 2 satellites were placed inEarth orbit with equator crossing times 4 hours apart,the improvement in coverage by these 2 satellites overa single one is about 60%. Furthermore, by managingthe orbits of the 2 satellites, nearly complete sun glintavoidance can be achieved, further improving theocean coverage. The best improvements can be madeby 2 satellites in the same node, whose orbital posi-tions are adjusted to view the sun glint contaminatedareas of the other. If scan edges are useful for quanti-tative ocean color observations (the validity of whichis unknown at this time), then only 2 polar-orbitingsatellites are necessary. If not, a third satellite is nec-essary, preferably in a low inclination orbit wherelosses in coverage in the tropics by the polar orbiterscan be best compensated (polar orbiters overlap cov-erage at high latitudes, and the scan edges are neces-sary only near the equator). However, the best con-figuration is 3 geostationary satellites, which providecomplete global coverage routinely with a viewing

time separation that can maximize cloud and sun glintavoidance, with a single polar orbiter to provide highlatitude coverage (Gregg, 1999a).

b) Analysis of Past Sensors

We continued investigations into OCTS data.These investigations began before the failure ofADEOS, in an effort to characterize OCTS data foruse with a merging activity with SeaWiFS. The lossof data from the sensor precludes its use in merging,but the similarity of some aspects of sensor designwith MODIS suggests that significant understandingof the capabilities and deficiencies of MODIS data canbe facilitated through a thorough analysis of OCTSdata. In fact, we have found this to be true. Initialobservations just after failure of OCTS suggested 6problem areas: band registration, image striping, cloudnoise, navigation, calibration, and bright target re-sponse. Analysis of the first three led to solutions thatimproved imagery substantially (Gregg, 1998; Gregg,1999b). Particularly the methods for reducing imagestriping is a useful method for a problem expected tooccur in MODIS imagery. Analysis of geolocation,radiometric stability and accuracy, and bright targetresponses were characterized and submitted for publi-cation (Gregg et al.,1999). The results here also haveimplications for MODIS.

c) Merging Algorithms

Efforts have began on developing merging algo-rithms. We are investigating 4 possible approaches: a)simple splicing, where 1 data set is deemed superiorand the other is patched into gaps, b) sensor weight-ing, where specific dependences and deficiencies areidentified and co-located observations are mergedusing different weighting functions for the sensors, c)the Conditional Relaxation Analysis Method, wherethe best data are selected as interior boundary condi-tions into a merged set using Poisson's equation (Rey-nolds, 1988), and d) optimum interpolation, wheremerging occurs by weighting individual sensor data tominimize spatial covariance function. The Condi-tional Relaxation Analysis Method (CRAM) was im-plemented and tested using CZCS data and in-situdata. Although these results were funded by separategrant, the results enabled us to begin to understand thestrengths and weaknesses of the method and its poten-tial use for satellite data (Gregg and Conkright, 1999).Several modifications are required before effective usewith in-situ data, but it is not clear at this time whetherthese modifications will be necessary for merging ofmultiple satellite data, since data paucity may not bean issue. It may be an issue, however, if significantdeficiencies are identified.


63

Chapter 16

Validation of Bio-optical Properties in CoastalWaters: a Joint NASA - Navy Project

Richard L. MillerNASA, Stennis Space Center, Mississippi

16.1 INTRODUCTION

This project is a joint effort between NASA’sEarth System Science Office and the Naval Oceano-graphic Office (NAVOCEANO) at the John C. Sten-nis Space Center, MS. The collaboration reflects apartnership developed by NASA and NAVOCEANOover six years to share resources and expertise in thecollection of ocean optics, hydrographic, biological,and remotely sensed data to support each agency’sgoals in coastal research. NAVOCEANO providesship time, instrumentation, and operational supportduring oceanographic surveys while NASA providesexpertise in data collection, computer programmingand modeling, and data processing and analysis. ThisSIMBIOS project was established to provide datacollected in the South China Sea, Gulf of Thailand andadjacent coastal waters. This project is designed toprovide NASA with cost-effective access to in-wateroptics and ancillary data in coastal waters representingdiverse turbid environments, and if successful, willhelp provide an effective mechanism for interagencycooperation and data exchange.


NAVOCEANO hydrographic surveys 260797 (27April - 19 May) and 260897 (23 May - 16 June) werecompleted in the South China Sea, and the Gulf ofThailand / South China Sea, respectively. The SIM-BIOS project investigators comprise the optics teamof these surveys, one of three primary operations: op-tics, bioluminescence, and basic hydrography. Verticalprofiles were acquired by each operations team at adense grid of selected stations. In addition, XBTs weredropped at regular intervals along the cruise trackbetween the profiling stations. Three optics packageswere deployed. These were designated as the day(IOPs and radiometer), night (IOPs), and the Satlanticsystem. The day and night packages (Table 1) used theMermaid II data acquisition system (Miller et al.,1995). System power and communications networkfor the day and night packages were provided byMODAPS (Wet Labs, Inc.).

Vertical Profiles and IOP

Several power and communication problems wereencountered with the MODAPS deck and subunits ofeach package. Power delivery from the subunit to theMER 1032 failed after 11 profiles and hence the com-bined package was abandoned. The subunit for theIOP package could not establish communications withthe Seabird CTD. The CTD was disconnected fromthe MODAPS system and was deployed in standardprofiling mode (internal data collection). CTD Datawere downloaded independently following each cast.Unfortunately, this approach decoupled the CTD andAC-9 data streams. A program was developed andused to interactively merge AC-9 and CTD profilesduring preprocessing and quality control. Two AC-9swere on the IOP package. The inflow tubes of oneAC-9 were fitted with 0.2 µm inline filters at severalstations to partition the total spectral absorption andbeam attenuation into dissolved and particulate frac-tions. At these stations, total (unfiltered) and dissolved(filtered) a and c profiles were obtained. A compari-sion / calibration of the AC-9s were performed whenboth AC-9s were unfiltered. The IOP package wasdeployed at 54 stations (111 profiles) and 120 stations(248 profiles), during survey 260797 and survey260897, respectively. Two PC-based programs werewritten in IDL 5.0 (castDisplay, castMapper) to applyquality control checks and routine data correction anddisplay. Temperature, salinity and scattering correc-tions were applied to all AC-9 data. Scattering correc-tions used either Method 1 of the Zaneveld Method(AC-9 User Guide, Wet Labs, Inc). A branching algo-rithm was used based on the a715 nm / b715 nm ratioto select which scattering method was applied.Method 1 was used for ratios greater than 0.35. TheZaneveld Method was used for ratio values <= 0.35.The castDisplay and castMapper (a geographic con-touring package for vertical profiles) will be used onsubsequate surveys for near-realtime processing andanalysis. All data have been translated to SeaBASSformat.


64

Satlantic Profiles

A SeaWiFS Profiling Multichannel Radiometer(SPMR) and a SeaWiFS Multichannel Surface Refer-ence (SMSR) instrument were deployed at 42 and 36stations during survey 26079 and 260897, respec-tively. In general, three profiles were acquired at eachstation. The central wavelengths for the SMPR andSMSR channels were: 405.3, 412.9, 435.1, 443.8,456.2, 490.9, 509.1, 532.7, 555.2, 590.5, 665.8, 683.8,and 700.4 nm. Satlantic’s Prosoft 4.0 (revision f) wasused to produce derived products of K, Reflectance,Rrs, Lwn, and modeled pigments. The Satlantic systemperformed without problems.

Discrete Water Samples

At most stations, discrete water samples werecollected at select depths (based on profiling transmis-someter and fluorometer data) using Niskin Bottlesmounted on the hydrographic CTD rosette. Thesewater samples were filtered for analysis of fluoromet-ric chlorophyll a and phaeopigments, HPLC pigments,CDOM absorption, particle absorption (QFT), andtotal suspended mater. All samples were collected andprocessed following the procedures of Mueller andAustin (1995). HPLC pigments determined were chlo-rophyll a, chlorophyll b, peridinin, fucoxanthin, 19hex, prasinoxanthin, alloxantin, lutein, zeaxanthin,diadinoxanthin, canthaxantin, and β-carotene. Ab-sorption spectra were determined for total, detrital,and phytoplankton components following the hot-methanol extraction technique (Kishino et al., 1985).Corrections for path-elongation were performed fol-lowing the method of Mitchell (1990). CDOM ab-sorption spectra were collected using a Perkin-ElmerLambda 3 spectrophotometer. The number of samplescollected for each analysis were (Survey 260797;260897): HPLC (71; 100), fluorometic chl a (86; 149),total suspended mater (50; 50), CDOM absorption (5;20), and particle absorption (0; 61).


The project was able to follow the proposed workplan with regards to data collection and processing. Ingeneral, instrument calibration and data collectionfollowed the guidelines provided by the proposal withthe exception that the MER 1032 needed to be phasedout of the instrument suite. The problems with theMODAPS created unexpected delays in preprocessingand analysis. In addition to collection problems andlimitations, considerable effort was invested to de-velop a reliable and robust method to merge CTD and

AC-9 data. Raw and intermediate data products werecreated for most profiler data. The volume of datacollected at sea exceeded our expectations. This re-sulted in a significant increase in the time required toprocess data. The NAVOCEANO provide operationalsupport to the US fleet by conducting military surveysworldwide. These surveys are conducted under theprovisions of military ordinances, and as such, do notrequire diplomatic permission from local govern-ments.

The data collected during this project were mostlyobtained in the Exclusive Economic Zone (EEZ) ofseveral countries. Therefore, as a consequence of“The Law of the Sea”, these data are currently re-stricted to DOD and DOD contractors use only.Hence, specific results and findings can not be pre-sented at this time.

Members of the project team at NASA and NA-VOCEANO are actively engaged in determining howthese data may be made available to the SIMBIOSproject office. It is the project’s intent to establish aprecedent for the release and use of this data for civil-ian projects without compromising the Navy’s inter-national agreements and mandates. If successful, thisproject may establish the means by which an impor-tant resource of oceanographic data may be madeavailable to the research community.

16.4 WORK PLAN

The project will follow the proposed project planwith the exceptions that above water estimates of Rrs(λ) will be obtained using a GER (Geophysical Envi-ronmental Research, Inc.) 1500 hand-held spectrora-diometer (300 -100 nm, effective 1 nm bandwidth)and the MER 1032 will not be used.

A survey to the South China Sea and Gulf ofThailand is planned for the summer 1999. A duplicateSatlantic system has been purchased to provide systemredundancy. A Wet Labs WetStar profiling fluorome-ter will be configured on a profiling package to directthe selection of depths for discrete water samples.SeaWiFS LAC data will be requested during filedactivites and upon processing using SeaDAS will becompared with in situ observations.

The project has acquired a AC-9 to ensure cali-bration history and performance for project deploy-ments. A MicroTops II sun photometer will be usedduring future cruises. Increased effort will be devotedto secure the release of the data for SIMBIOS projectuse.


65

Table 1. Profiling Instrument Packages.

INSTRUMENTS MEASUREMENTSDay Package

MER 1032 In-water Radiometer1 Spectral Upwelled and Downwelled Irradiance, and Spectral Up-welled Radiance

MER 21 Deck Radiometer1b Surface Incident IrradianceAC-92 Spectral absorption and beam attenuation coefficientsLS-6000 Light Scattering Sensor3 Light scattering at 880 nmSea-Bird SBE 19 CTD4 Conductivity, temperature, and depthPSA-900 Sonar Altimeter5 Height (m) of package above bottom

Night PackageAC-92 (x2) Spectral absorption, beam attenuation, backscattering coefficientsLS-6000 Light Scattering Sensor3 Light scattering at 880 nmSea-Bird SBE 19 CTD4 Conductivity, temperature, and depthPSA-900 Sonar Altimeter5 Height (m) of package above bottom

Remote - Tethered PackageSeaWiFS Profiling Multichannel Radi-ometer6

Spectral Upwelled Radiance, Spectral Downwelled Irradiance

SeaWiFS Multichannel Surface Refer-ence6

Downwelled Spectral Surface Irradiance

1Biospherical Instruments, Inc.; 1bBiospherical Instruments, Inc., MER 21 is a deck-mounted sensor;2WET Labs, Inc. (Western Environmental Technology Laboratories, Inc); 3Sea Tech, Inc.;4Sea-Bird Electronics, Inc.;5Datasonics, Inc.; 6Satlantic, Inc.


66

Chapter 17

Bio-Optical Measurement and Modelingof the California Current and Polar Oceans

B. Greg Mitchell and Piotr FlatauScripps Institution of Oceanography

University of California San Diego, La Jolla, California

17.1 INTRODUCTION

This SIMBIOS project contract supports in situoceanic optical observations in the California Currentand Southern Ocean. The principal objectives of thisresearch are to validate standard or experimentalproducts through detailed bio-optical and biogeo-chemical measurements, and to combine ocean opticalobservations with advanced radiative transfer model-ing to contribute to satellite vicarious radiometriccalibration and algorithm development.

Our sampling efforts have been directed towardsobtaining measurements in both the California Currentand Antarctic polar waters, with the purpose of gener-ating a high-quality, methodologically consistent dataset encompassing a wide-range of oceanic conditions.The combined data base includes stations which coverthe clearest oligotrophic waters to highly eutrophicblooms and red-tides, and provides a coherent set ofobservations to validate bio-optical algorithms forpigments and primary production. This unique andcomprehensive data is utilized for development ofexperimental algorithms (e.g. high-low latitude pig-ment transition; phytoplankton absorption, photosyn-thesis, cDOM.


The Southern California Bight region, from SanDiego to just north of Point Conception, has one of thelongest, most comprehensive time-series of marineobservations; the California Cooperative OceanicFisheries Investigation (CalCOFI). This region expe-riences a large dynamic range of coastal and openocean trophic structure, and has been extensivelystudied with respect to its regional optical propertiesin an effort to develop regional ocean color algorithms(e.g. Smith and Baker 1978, Mitchell and Kiefer 1988,Sosik and Mitchell 1995). During the first year of ourcontract, we participated in 4 quarterly cruises in theCalifornia Current region as part of the CalCOFI pro-gram.

The Southern Ocean is a large, remote regionwhich plays a major role in global biogeochemical

cycling. Despite evidence that bio-optical relation-ships in these waters can diverge significantly fromlower-latitude waters (e.g. Mitchell and Holm-Hansen1991), Antarctic waters have not been represented inthe databases (e.g. SeaBAM) used to formulate andtest modern ocean color algorithms. During the pastyear, we participated in 3 cruises to the SouthernOcean as part of the US JGOFS program. One cruisewas located within the Ross Sea Polyna during theannual spring phytoplankton bloom, with two subse-quent cruises covering the region of Antarctic PolarFront Zone along 170° W.

On all cruises, an integrated underwater profilingsystem was used to collect optical data and to charac-terize the water column. The system included an un-derwater radiometer (Biospherical Instruments MER-2040 or MER-2048) measuring depth, downwellingspectral irradiance (Ed) and upwelling radiance (Lu) in13 spectral bands. A MER-2041 deck-mounted refer-ence radiometer (Biospherical Instruments Inc) pro-vided simultaneous measurements of above-surfacedownwelling irradiance. Details of the profiling pro-cedure, characterization and calibration of the radi-ometers, data processing and quality control are de-scribed in Mitchell and Kahru (1998). The underwa-ter radiometer was also interfaced with 25 cm trans-missometers (SeaTech or WetLabs), a fluorometer,and SeaBird conductivity and temperature probes.When available, additional instrumentation integratedonto the profiling package included AC9 absorptionand attenuation meters (Wetlabs Inc.), and a Hydro-scat-6 backscattering meter (HobiLabs).

In conjuction with in situ optical measurements,discrete water samples were collected from a CTD-Rosette immediately before or after each profile foradditional optical and biogeochemical analyses. Pig-ment concentrations were determined flurometricallyand with HPLC. Spectral absorption coefficients(300-800 nm) of particulate material were estimatedby scanning particles concentrated onto WhatmanGF/F (Mitchell 1990) in a dual-beam spectropho-tometer (Varian Cary 1). Absorption of soluble mate-rial was measured in 10cm cuvettes after filteringseawater samples through 0.2 um pore size polycar-bonate filters.


67


Validation of satellite-retrieved normalized water-leaving radiances (LWN) was done by comparingSeaWiFS images with in situ data collected concur-rently (± 4 hours). Satellite data received at the Mon-terey Bay Research Institute, the University of Cali-fornia Santa Barbara (CalCOFI region) and McMurdoStation, Antarctica (Southern Ocean region) wereprocessed to LWN using SeaDAS 3.2 software (Fu etal. 1998). A total of 16 matching sets of LWN werefound between 2-Oct-1997 and 21-Apr-1998 for theCalCOFI region. Because of persistent cloud cover inthe Southern Ocean, only 3 matchups were possible inthe Ross Sea region (all on 1-Dec-1997). Satellitevalues were derived as averages over 3 x 3 pixel areascentered at the in situ measurement.

In both regions, these comparisons reveal signifi-cant under-estimation of the SeaWiFS-retrieved LWN

compared to in situ measurements. The differenceswere generally smallest in the 555 nm band, and larg-est at shorter wavelengths. The magnitude of under-estimation in the shorter wavlength bands increases athigh Chl concentration.

O’Reilly et al. (1998) describe the Ocean Color 2(OC2) chlorophyll algorithm that is used by NASA inthe operational processing of SeaWiFS data (Fu et al.1998). This algorithm uses the ratio of remote sensingreflectances (Rrs) at 490 and 555 nm to estimate chlo-rophyll a concentration, with the coefficients derivedby a statistical fit to a data set of 919 bio-opticalmeasurements comprising the SeaBAM data set.More recently (August 1998), NASA announced arevised version of the OC2 (OC2-v2) which was in-tended to reduce the drastic over-estimation of Chl inhigh biomass waters produced by the original OC2algorithm.

Figure 1 compares the performance of the OC2-v2 algorithm with our present data base of measure-ments from CalCOFI and the Southern Ocean. Whenapplied to the CalCOFI data, this algorithm overesti-mates chl a at very high chl a and underestimateselsewhere (with the exception of the extreme low chla). A similar pattern is seen with the Southern Oceandata, although in general the degree of underestima-tion is greater and the transition to overestimationoccurs at lower Chl. These results underscore theeventual need for specific regional empirical algo-rithms to obtain more accuate estimates of Chl andprimary production from ocean color remote sensing.We have recently developed an improved empiricalchlorophyll algorithm for the California Current(CAL-P6), which utilizes a sixth order polynomial ofthe ratio of LWN at 490 and 555nm (Kahru andMitchell, 1999).

17.4 WORK PLAN

Our participation in the quarterly CalCOFIcruises in the California Current will continuethroughout the next period. We are also initiatingfield programs in the Indian Ocean (INDOEX) and theSea of Japan (JES) to increase the regional scope ofour data base. Modeling efforts include the pursuit ofregional bio-optical algorithms for in water opticalproperties and their relationship to biogeochemicalparameters, as well as the development of semi-analytical models for the retrieval of inherent opticalproperties from satellite data. We anticipate that theseefforts will lead to an improved understanding of thevariability observed in empirical satellite algorithms.We will also initiate analyses to determine the ele-ments of the SeaWiFS processing that lead to the un-derestimates of Lwn, which is of particular concern

for high chlorophyll waters.


68

Figure 1. A comparison of the OC2-v2 algorithm (solid line in left panels) with in situ measurements ofchlorophyll a from CalCOFI and the Southern Ocean. The right panels illustrate quantile-quantile plots of thedifferences between modeled and measured chlorophyll a.


69

Chapter 18

SIMBIOS Normalized Water Leaving RadianceCalibration and Validation: Sensor Responses,

Atmospheric Corrections, Stray Light and Sun Glint

James L. MuellerCenter for Hydro-Optics and Remote Sensing

San Diego State University, San Diego California

18.1 INTRODUCTION

The objectives of research under this contract areto improve the uncertainty budgets and protocols fornormalized water leaving radiances LwN(λ) and re-mote sensing reflectance RrsN(λ) determined from insitu measurements, and to apply in situ reflectancemeasurements to validate LwN(λ) and RrsN(λ) deter-mined from radiances measured by the SIMBIOS en-semble of orbiting ocean color sensors (i.e. SeaWiFS,MODIS, GLI, POLDER and others). Our approach tothese objectives includes protocol experiments toevaluate uncertainties in reflectances derived fromabove-water and in-water radiometry, field expedi-tions to acquire in situ radiometric, optical and bio-logical ground truth data for validating SIMBIOS sat-ellite ocean color sensors and match-up analyses,combining the in situ data with satellite radiometricdata and derived parameters for vicarious radiometriccalibration, and for evaluating sun glint effects andstray light artifacts in the image data.


There is a general consensus that poorly under-stood optical variability associated with surface wavescontributes strongly to the persistently significant dis-crepancies between remote sensing reflectances de-termined from above-water and underwater radiomet-ric measurements. In above-water radiance meas-urements, wave roughness modifies the surface re-flectance relative to a flat surface. Wave rougheningalso widens the effective field-of-view of reflectedskylight components of the measured signal, and thusmay incorporate angular variability in skylight whichis not quantified by measurements using current pro-tocols (Mueller and Austin, 1995). Finally, waveslope variations repolarize reflected skylight in wayswhich may not be properly accounted for using theseand proposed alternative protocols.

In underwater radiometric profiles near the seasurface, surface waves radomly focus and defocus

incident radiance and introduce “noise” which in-creases exponentially with proximity to the surface.Additional uncertainties result from strong wave-induced accelerations on an underwater radiometricpackage, which accelerations also increase exponen-tially with proximity to the surface and introducequasi-random, difficult to measure fluctuations insensor orientation. It is very difficult to separate thesewave-related sources of uncertainty in comparing re-flectances calculated from in-water and above-watermeasurements. It is correspondingly difficult, there-fore, to partition the overall observed uncertaintybetween the two types of measurement. We areworking towards comparative outdoor tank and fieldmeasurements which attempt to separately quantifynear-surface in-water and above-water wave relateduncertainties under semi-controlled conditions.

To support the difficult measurements associatedwith these protocol experiments, we are developing aunique hyperspectral above-water and profiling spec-troradiometer and inherent optical property (IOP)system. The system is designed to allow repeatedradiometric and IOP profiling at controlled ascent anddescent rates, concurrent with measurements ofabove-water irradiance and radiance. The profilerpackage is built around a tethered variable buoyancyAutonomous Profiling Vehicle (APV) manufacturedby Ocean Sensors Inc. of San Diego. In addition tothe APV’s built-in Conductivity, Temperature andDepth (CTD) sensors, sensors are attached to the pro-filer package to measure irradiance Ed(λ,z), radianceLu(λ,z), absorption a(λ,z), beam attenuation c(λ,z)and chlorophyll a fluorescence. Ed(λ,z) and Lu(λ,z)are measured at 3.3 nm intervals from approximately380 to 800 nm using radiometers based on Zeissminiature fiber-optic spectrometers. Chlorophyll afluorescence is measured using a WETSTAR fluo-rometer, and a(λ,z) and c(λ,z) are measured at 3.3 nmintervals using a HISTAR underwater spectropho-tometer – both instruments are manufactured byWETLABS of Philomath, Oregon. (The WETSTARand HISTAR are obtained under a ONR/ECOHABresearch grant at CHORS.) Additional radiometers


70

are being integrated into the system for above-watermeasurements of irradiance Es(λ,z), total water-leaving radiance Lsfc(λ,z) and sky radiance Lsky(λ,z).

Validation Field Experiments

We are acquiring in situ data for validatingSeaWiFS (and in the future MODIS and other oceancolor sensors) data during field expeditions aboard theMexican Research Vessel Francisco de Ulloa (FdU)in the Gulf of California (GoCAL). The GoCALcruises are in collaboration with Drs. Ron Zaneveldand Scott Pegau of Oregon State University (SIM-BIOS contract NAS5-97129) and Mexican scientists(Drs. Helmut Maske, Ruben Lara-Lara, Saul Alvarez-Borrego, and their students) at CICESE (an oceanog-raphy PhD granting postgraduate university in En-senada, BC, Mexico).

The FdU is owned and operated by CICESE, andthey have thus far contributed ship time for cruises inOctober 1997 (GoCal-97), March 1998 (GoCal-98A)and Nov-Dec 1998 (GoCal-98B). Measurements oneach cruise include profiles of Ed(λ,z) and Lu(λ,z) atSeaWiFS wavelengths, fluorometric chlorophyll a,and HPLC pigments. IOP measurements are made bythe OSU investigators, and additional bio-optical andancillary measurements are contributed by CICESE.The next GoCal cruise aboard the FdU (GoCal-99A)is tentatively scheduled for 21 April – 2 May 1999.

All radiometric, optical and phytoplankton pig-ment measurement and data analysis procedures fol-low protocols in Mueller and Austin (1995). Radio-metric profiles are analysed to obtain Es(λ,0+),Ed(λ),0+), and Lu(λ,0+), using the integral method ofMueller (1995), as implemented using the CHORSinteractive Kfit routine.

We are also planning to acquire validation dataduring day-cruises near San Diego and other Califor-nia coastal sites using a small research boat. We haverecently ordered and will outfit a trailerable researchboat (a Parker-25 sport-fishing boat) to support thisand other research projects at CHORS. Electronicequipment will be installed in the boat’s enclosed pi-lothouse and forward cabin, while water sam-pling/filtering stations will be installed aft of the pi-lothouse under a permanent awning. Costs (acquisi-tion, maintenance, and operating) associated with theCHORS research boat will be shared between thiscontract, other CHORS research projects using theboat, and SDSUF cost-sharing funds.

Calibration and Characterization

To support these experiments and analyses,CHORS carries out calibrations and characterizationsof radiometers used in this project, including those ofour collaborators at OSU and CICESE. NIST trace-ability of CHORS calibrations is maintained through

tertiary standards (Optronics, Inc) and assuredthrough participation in SIRREX and independentcomparisons with other laboratories calibrating SIM-BIOS instruments.


Radiometric data from GoCal-97 and GoCal-98A(including the results of the CHORS KFIT analysis ofOSU’s SPMR profiles) have been processed, analysedand archived in SeaBASS. Pigment data analyses andquality control have recently been completed for thesetwo cruises, and the data will be archived in SeaBASSafter appropriate reformatting and insertion of meta-data. Processing and analysis of data from GoCal-98B(24 Nov through 8 Dec 1998) are in progress.

18.4 WORK PLAN

During the current contract year we will completework in progress including:

• data processing, analysis and SeaBASS archi-val/documentation of data from GoCal-98B;

• Analysis and publication, in collaboration withinvestigators from CICESE and OSU, of datafrom completed GoCal cruises in 1995, 1996,1997, 1998A and 1998B, (including SeaWiFSdata for the 1997 and 1998 periods);

• Integration, laboratory characterization and cali-bration, and field-testing of the CHORS hyper-spectral radiometric and IOP profiling and re-flectance system;

• Continued routine calibrations of radiometersused by CHORS and our collaborators (OSU andCICESE) in this project.

During the current contract year, planned field cam-paigns include:

• GoCal-99A aboard the FdU in the Gulf of Cali-fornia between 20 April and 2 May 1999; and

• Day cruise, clear-sky validation sorties (within 20nautical miles of the California coast) aboard theCHORS research boat ( circa June 1999).

Reflectance protocol experiments will includetank and field in-water versus above-water reflectancemeasurement comparisons, as well as experimentalevaluations of plaque reflectance versus direct meas-urements of incident irradiance and instrument self-shading tests.


71

Chapter 19

Validation of Carbon Flux and Related Products forSIMBIOS: the CARIACO Continental Margin

Time Series

Frank Müller-KargerDepartment of Marine Science

University of South Florida

Ramon VarelaFundacion La Salle de Ciencias Naturales

Punta de Piedras, Estado Nueva Esparta, Venezuela

19.1 INTRODUCTION

This SIMBIOS investigation focuses on validatingocean color satellite products from waters near conti-nental margins. Specifically, our program focuses oncollecting monthly observations within a coastal up-welling site, and seasonal extreme measurementswithin the plume of the Orinoco River, which is classi-fied as having the third largest discharge in the world.

Initial focus of this project has been on setting upthe logistics for field data collection and initiatingroutine observations. The program incorporates aneffort to validate SeaWiFS products and products fromthe future MODIS (AM and PM), MISR, MERIS,GLI, and ROCSAT missions. Our goals are to evalu-ate algorithms used to compute biological and geo-physical data products, develop new products aimed atassessing carbon flux, estimate the spatial and tempo-ral errors in these products, and identify strategies thatwill help minimize such errors and eliminate possiblebiases between data sets from different missions.


The core of this SIMBIOS project is an estab-lished oceanographic time series stations in theCARIOCO basin. In November 1995, the U.S. Na-tional Science Foundation, in collaboration with theConsejo Nacional de Investigaciones Cientificas yTecnologicas (CONICIT) of Venezuela, implementeda comprehensive program of continuous observationsin the Cariaco Basin, which is located in the southeast-ern Caribbean Sea. This multidisciplinary program isreferred to as the CArbon Retention In A ColoredOcean (CARIACO) program. Specifically, the pro-gram conducts monthly observations at a station lo-cated at 10.5 N, 64.67 W.

The importance of the Cariaco Basin as a naturallaboratory to study seasonal upwelling, sediments andanoxic effects has been appreciated for over 40 years.The CARIACO Program seeks to define a budgetwhich assesses the total CO2 upwelled with deep,nutrient-rich water with respect to the annual export oforganic carbon from surface waters along this conti-nental margin. This basin is ideal for a carbon fluxstudy because it is a large depression on a continentalshelf which experiences vigorous wind-driven up-welling, and oligotrophic conditions at different timesof the year. As the circulation below 200 m in thisbasin is restricted, it forms a natural sediment trap.The Cariaco Basin is the only permanently anoxicoceanic basin, and as such serves as an oceanic analogof the Black Sea.The series station is located at 10.50N, 64.66 W and consists of:

• 1 to 3 monthly cruises with a fully-equipped,modern oceanographic vessel;

• Mooring with 4 sediment traps (200, 400, 800,1200 m; capturing bi-weekly sample integra-tions);

• Varved (laminated) sediment core analyses.

The monthly oceanographic cruises provide in-formation on:

• Composition, levels, and light absorption proper-ties of organic particulate and dissolved matter;

• Pigments, HPLC, taxonomy of phytoplankton andgeneral classification of bacteria;

• Biological productivity (phytopl. and bacteria);• The physical/chemical characteristics of the water,

including nutrient and oxygen levels;• Carbonate system.


72

Shore-based, continuous meteorological and tidegauge observations are available, and we are trying toobtain these in collaboration with the local Air Force,Navy, and Environment Ministries. This project incor-porates subsurface and above-surface spectral radi-ance, irradiance, and aerosol optical thickness meas-urements to the CARIACO series. The in situ bio-optical observations are all done following theSeaWiFS protocols. We have also participated inother SIMBIOS activities, including the first SIM-BIOS PI meeting in Salomon's Island and a cal/valcruise to the Bahamas organized under Ken Carder'sSIMBIOS program.


Each month since December 1995, bio-opticalmeasurements have been part of the suite of measure-ments made at the CARIACO site. These observationswere initially limited to above-water hyperspectralreflectance measurements. MER casts were performedonly a few months out of this period because there wasno continuous access to such an instrument.

Since October 1997, our suite of observations hasincluded MER subsurface radiance/irradiance profiles.The optical measurements are used to validateSeaWiFS. The measurements are scheduled aroundthe time of the SeaWiFS overflight. The monthlycruises last about 2 days each.

During each cruise, we collect bio-optical meas-urements using the underwater profiling BiosphericalInstruments MER2048 in conjunction with aMER2041 Deck Cell. Above water measurements aremade with a Photo Research Hyperspectral Color-imeter Model PR650, which is a calibrated, hand-heldradiometer (8 nm resolution). Derived products in-clude Lw (Water-Leaving Radiance), Rrs (Remote-sensed reflectance) and K (attenuation coefficient). Inaddition, a full suite of measurements is made whichincludes: particulate material and pigment absorption,HPLC, fluorometric determinations of Chl concentra-tion, pH, Alkalinity, Primary Productivity, POC, DOCabsorption and concentration, nutrients, sun photome-try, oxygen, and salinity.

Cruise reports for each cruise starting in Septem-ber 1997, have been submitted to the SIMBIOS Proj-ect. Approximately two years of bio-optical data havebeen submitted to the SeaBASS archive.

Preliminary results suggest marked seasonalchanges in optical properties occur in this basin asphytoplankton concentration increases and decreasesin response to coastal upwelling. Figure 1 shows theseasonal variation in surface hyperspectral remotesensing reflectance at the Cariaco station. A clear pro-gression is seen from clear water conditions duringboreal summer to blue-poor reflectances during borealwinter-spring, the time of upwelling. The decrease inblue reflectance is accompanied by a marked increase

at 683 nm, clearly defining the signal of solar-stimulated fluorescence. We plan to examine this fea-ture in relation to the primary productivity data col-lected concurrently during our CARIACO cruises.

We also find that there is a close correlation be-tween phytoplankton specific absorption coefficients,primary productivity, and chlorophyll a. These dataclearly are preliminary and need to be examined closelyto understand variations from season to season.

The Orinoco River Plume

Above-water spectral remote-sensing reflectance(Rrs 380-700 nm) was measured including correctionsfor sky-radiance. We also conducted underwatermeasurements of remote sensing reflectance at 7wavelengths (412, 443, 455, 475, 490, 510, 532, 560,589, 625, 671, 683, and 700 nm) using submersibleMER2048 instrumentation (Biospherical Instr.). Someobservations were conducted away from the coast,where the color of the plume is influenced by coloreddissolved organic matter as well as phytoplankton, butwhere suspended terrestrial material is expected to beminimal. We also collected data very close to thecoast where the plume is extremely turbid. As of Feb-ruary 1, 1999, we are currently submitting bio-opticaldata collected during the first and second 1998 Ori-noco plume cruises to SeaBASS

The Orinoco Plume data span a very broad range ofremote sensing reflectance values. The river plumedata are also being used by Joe Salisbury and CharlesVorosmarty at the University of New Hampshire forautomated modeling of river water impact off the conti-nent of South America, in a model linked to terrestrialhydrology.

19.4 WORK PLAN

We will continue to occupy the CARIACO sta-tion on a monthly basis. In addition, we plan to con-duct two cruises to the Orinoco River plume, specifi-cally in February and October 1999. Analysis of dataand samples will continue. We are currently finalizinga manuscript that describes the seasonal cycle in hy-drographic properties, phytoplankton biomass andproductivity, and vertical carbon flux at theCARIACO station. We will now initiate work on amanuscript that summarizes the bio-optical data fromthis location.

ACKNOWLEDGMENTS

This work was supported by the National ScienceFoundation (NSF Grant OCE 9216626), the NationalAeronautics and Space Administration (NASA GrantsNAG5-6448 and NAS5-97128), and the Consejo Na-


73

cional de Investigaciones Cientificas y Tecnologicas(CONICIT, Venezuela). Many people have contrib-uted substantially to the success of this program, andif they are not mentioned here it is by plain oversighton the author's part. We are indebted to the personnelof the Fundacion La Salle de Ciencias Natura-les/Estacion de Investigaciones Marinas Isla Margarita(FLASA/EDIMAR) for their enthusiasm and profes-sional support. In particular we thank FLASA's Di-rector, Dr. Hermano Gines, for his confidence in ouractivities and the crew of the R/V Hermano Gines(FLASA) for their able support at sea. Field bio-optical measurements were conducted primarily byNatasha Rondon and John Akl of FLASA. They, andYrene Astor and Ana Lucia Odriozola have processed

the majority of the data to satisfy the seabass reportformats. Jonathan Garcia (also at FLASA) has proc-essed the samples for particulate and dissolved ab-sorption coefficients. Juan Capelo and Javier Gutier-rez (FLASA) have assisted with the primary produc-tion observations. Robert Thunell and Eric Tappa (U.South Carolina) maintain the sediment trapping pro-gram and process our particulate samples (POC,PON). Mary Scranton and Gordon Taylor (SUNY)conduct bacteria productivity studies. Dissolved Or-ganic Carbon samples were processed by Ed Peltzer(formerly at the Woods Hole Oceanographic Institu-tion and currently at the Monterrey Bay AquariumResearch Institute), and lately by David J. Hirschbergof State University of New York at Stony Brook.

Seasonal Remote Sensing Reflectance

0.0000

0.0010

0.0020

0.0030

0.0040

0.0050

0.0060

0.0070

380 480 580 680 780

Wavelength [nm]

Rrs

May '96Sep '96Oct '96Nov '96Dic '96Jan '97Feb '97Mar '97Apr '97May '97

Figure 1. Remote sensing reflectance at the Cariaco station over the course of one year.


74

Chapter 20

Bermuda Bio-Optics Program (BBOP)David Siegel

University of California at Santa Barbara,Santa Barbara, California

20.1 INTRODUCTIONThe Bermuda BioOptics Project (BBOP) is a col-

laborative effort between the Institute for Computa-tional Earth System Science (ICESS) at the Universityof California at Santa Barbara (UCSB) and the Ber-muda Biological Station for Research (BBSR, D. A.Siegel [UCSB] and N. B. Nelson [BBSR] P.I.'s). Thisresearch program is designed to characterize lightavailability and utilization in the Sargasso Sea and indoing so, provide an optical link by which biogeo-chemical observations may be used to evaluate bio-optical models for pigment concentration, primaryproduction, and sinking particle fluxes from satellite-based ocean color sensors.


The Bermuda Bio-Optics Program (BBOP) col-lects detailed profiles of AOPs and IOPs in conjunc-tion with the US JGOFS Bermuda Atlantic Time-

series Study (BATS) at 31°50' N, 64° 10' W in themesotrophic waters of the Sargasso Sea. There are upto 16 cruises per year, conducted monthly with addi-tional cruises during the spring bloom period, Januarythrough May.

Continuous profiles of apparent optical properties(AOPs) are collected in the upper 200m using a Mul-tispectral Environmental Radiometer (MER-2040/2041, Biospherical Instruments Inc., San DiegoCA; Smith et al., 1984). The primary optical meas-urements are downwelling vector irradiance and up-welling radiance, Ed(z,t,λ) and Lu(z,t, λ), respec-tively. The instrument samples these properties at 12wavebands (including SeaWiFS wavelengths) plusbroadband natural fluorescence. The BBOP samplingpackage also includes sensors for temperature andconductivity (SeaBird, Bellevue WA) chlorophyllfluorescence and red beam transmission (SeaTech,Corvallis OR) and a second mast-mounted spectrora-diometer (MER-2041) with wavebands similar tothose on the underwater instrument for measuring

incident downwelling vector irradiance, Ed(0+,t, λ).Instrument deployments are planned to optimize

match-ups with the BATS primary production incuba-tions and, when possible, with SeaWiFS overpasses.

The BBOP group also collects several measure-ments of inherent optical properties (IOPs). Continu-ous profiles of spectral beam absorption and attenua-tion, a(z, λ) and c(z, λ), are measured using the WETLabs AC-9 (Brody, 1998), and discrete samples fordetermining the absorption spectra of particulates,aph(z, λ) and ad(z, λ), and CDOM (ag(z, λ)) are col-lected according to Nelson et al (1998). Spectralbeam absorption measurements have been hamperedby the fact that the baseline absorption of SargassoSea water is nearly the same as that of the deionizedwater used for the calibration. Particulate absorptionspectra are determined using the quantitative filtertechnique (Mitchell, 1990) and CDOM absorptionaccording to Nelson et al (1998). In 1997, we initi-

ated the collection of above water R+rs(l) spectra using

the Analytical Spectral Devices FieldSpec spec-trometer (ASD, Boulder Colorado). We are currently

comparing this data set with values of R+rs(l) calcu-

lated from MER data to evaluate its usefulness forSIMBIOS.

The profiling radiometers are calibrated threetimes per year at UCSB and have demonstrated re-markable calibration stability for several years, al-lowing the use of long-term averages for calibrationcoefficients (O’Brien et al., 1999). This stability,combined with streamlined procedures for data analy-sis, has made it possible for us to report many of ourderived products with ~98% reliability in near-realtime, including profiles of remote sensing reflectance(Rrs(z, λ) = Lu(z, λ)/Ed(z, λ)), and down- and up-

welled attenuation coefficients (Kd(z, λ), Ku(z, λ)).Our analysis procedures are fully documented in Sie-gel et al. (1995b). In addition to radiometer data,samples for fluorometric chlorophyll-a are collectedbefore dawn and near noon (local time) and results aredelivered to the SIMBIOS project with the AOP data,usually within one week after a cruise. Table 1 con-tains a list of data products collected by BBOP and/orBATS, which are relevant to SIMBIOS. A total of472 AOP, 16 IOP and 36 pigment profiles were deliv-


75

ered from the start of this contract (July 1997) through1998.


Since its inception in 1992, BBOP has made anumber of important contributions toward the scien-tific understanding of the relationship between lightand ocean biogeochemistry. The Sargasso Sea nearBermuda is mesotrophic, characterized by both eutro-phic and oligotrophic conditions during different timesof the year. Although low chlorophyll a stocks andprimary production rates prevail for the most of theyear, there is a short spring bloom characterized bysomewhat higher concentrations and rates which iscontrolled by winter mixing and the associated nutri-ent influx.

Our research has shown that the BATS/BBOP sitehas an interesting and important ocean color signalassociated with its colored dissolved organic material(CDOM) content. We have found that CDOM may

comprise up to 50% of the light attenuation budget(Kd(λ)) at 410 nm (Siegel et al. 1995a, Siegel andMichaels 1996) and that CDOM concentrations can bedetermined from in situ ocean color spectra (Garverand Siegel, 1997). The ratio of Kd(410) to Kd(488)can be used as an index of CDOM concentration toillustrate its time-depth dynamics (Siegel et al. 1995a,Siegel and Michaels 1996). During the summer,CDOM is simultaneously produced within the sea-sonal pycnocline and photo-oxidized in the surfacemixed layer (Figure 1c). Fall and winter mixing ho-mogenizes the profile. The seasonal distribution ofCDOM is unrelated to the those of chlorophyll or dis-solved organic carbon, however it may be that thesummer CDOM source is microbial (Siegel andMichaels 1996, Nelson et al. 1998). Since CDOM is asignificant source of light attenuation in the SargassoSea that varies independently of chlorophyll, it ap-pears that the optical definition of Case I waters whichis based solely on chlorophyll should be reconsidered.

Table 1. A partial list of measurements made by BBOP and BATS

Direct Measurements:Ed(z, λ) Downwelling vector irradiance (410,441,465,488,510,520,555,565,589,625,665 & 683 nm)Ed(0+,λ) Incident irradiance (340,390,410,441,465,488,520,545,565,589,625,665 & 683 & 350-1050 nm)Lu(z, λ) Upwelling radiance (410,441,465,488,510,520,555,565,589,625,665 & 683 nm)a(z, λ) In situ absorption spectrum using WetLabs AC-9 (410,440,490,520,565,650,676&715 nm)c(z, λ) In situ beam attenuation spectrum (same λ's as above)atp(λ) Particulate absorption spectrum by QFTad(λ) Detrital particle absorption spectrum by MeOH extractionays(λ) Colored dissolved absorption spectrumEo(z, λ) Scalar irradiance at 441 and 488 nmFf(z) Natural chlorophyll fluorescence using a broadband upwelled radiance sensorchl-fl(z) Chlorophyll fluorescence with a SeaTech fluorometerc(z,660) Beam attenuation coefficient at 660 nm with SeaTech 25 cm transmissometerT(z) & S(z) Temperature and conductivity with SeaBird probeschl-a(z) Discrete chlorophyll a determinations via Turner fluorometry (for next day delivery)

Primary Derived Products:LwN(λ) Normalized water leaving radiance (410,441,465,488,510,520,555,565,589,625,665 & 683 nm)RRS(0-,λ) In-water remote sensing reflectance (410,441,465,488,510,520,555,565,589,625,665 & 683nm)RRS(0+,λ) Above-water remote sensing reflectance (350 to 1050 nm)aph(λ) Phytoplankton absorption spectrum (= ap(λ) - adet(λ))Kd(z, λ) Attenuation coefficient for Ed(z, λ) (410,441,465,488,510,520,555,565,589,625,665 & 683 nm)KL(z, λ) Attenuation coefficient for Lu(z, λ) (410,441,465,488,510,520,555,565,589,625,665 & 683 nm)<PAR(z)> Daily mean photosynthetically available radiation at depths of the in situ C14 incubationsb(z, λ) Spectral scattering coefficient (= c(z, λ) - a(z, λ))


76

Figure 1. A-C: Time-depth contours of a) Chlorophyll-a (HPLC), b) Kd(z,t,490), c) the ratio ofKd(z,t,410):Kd(z,t,490) and D) time series of remote-sensing reflectance spectra (Rrs(t,0

-,λ)).


77

Chapter 21

SIMBIOS Data Product and Algorithm Validationwith Emphasis on the Biogeochemical

and Inherent Optical Properties

J. Ronald V. Zaneveld and Andrew H. BarnardOregon State University

Corvallis, Oregon

21.1 INTRODUCTION

The purpose of the SIMBIOS program is to pro-vide routine, long-term ocean color data from thevarious international ocean color missions. Ourfunded component of this program is to address andquantify the relative accuracies of the products fromthe international missions by means of product andalgorithm validation. Our long-term objectives are toa) collect optical and biochemical data in oceanic andcoastal regions including a cooperative internationalbio-optical monitoring program in the Gulf of Califor-nia; b) provide collected data to the NASA database,SIMBIOS project office, and other interested users;and c) determine spatial and temporal error fields forthe biological and geophysical data products from thevarious ocean color missions.


Over the past 4 years we have been involved in amulti-national research effort focused on determiningthe physical and biogeochemical variability in theGulf of California. This research is done in coordina-tion with Drs. J. Mueller and C. Trees from San DiegoState University (SDSU), Drs. S Alvarez Borrego, R.Lara-Lara, G. Gaxiola and H. Maske from the Centrode Investigacion Cientifica y de Educacion Superor deEnsenada (CICESE), Ensenada, Mexico, and Dr. E.Valdez from the University of Sonora, Hermosillo,Mexico. Our component has been to measure the in-herent and apparent optical properties as well as thetraditional CTD parameters on vertical scales less than0.5 m. We have completed 5 cruises in the Gulf ofCalifornia, with 3 of these cruises occurring in the lastyear and half during our SIMBIOS contract period.Optical data collected during these cruises are cur-rently being used to validate SeaWiFS algorithms.We have also completed a cruise of opportunity off-shore of Oregon in September 1998. Data from thiscruise is currently being processed and analyzed.

Two sampling platforms are typically used duringthese cruises; a SLOW Descent Rate Optical Profiler(SLOWDROP) to measure the inherent optical prop-erties as well as the physical parameters, and a profil-ing radiometer system to measure the downwellingirradiance and upwelling radiance. Typically,SLOWDROP and radiometer profiles are made withinan hour of each other at the same location.

The SLOWDROP platform is free-falling andslightly negatively buoyant and provides 10-20 cmscale vertical resolution. A typical instrument con-figuration includes a SeaBird SBE-25 CTD for meas-uring physical parameters, a fluorometer, and twoWETLabs ac-9 (9 wavelength absorption and attenua-tion meters) for measuring particulate and dissolvedcomponent of the inherent optical properties (IOP).CTD and IOP data are collected simultaneously dur-ing a single profile using a Wet Labs, Inc. MODAPSsystem for data collection and integration. Calibra-tions of the ac-9 instruments were performed severaltimes during the contract period. Field calibrations ofthe ac-9 meters were typically done once per samplingday using a clean water standard produced using aBarnstead nanopure water system according to theSIMBIOS protocols for this instrument.

A Satlantic Inc. SPMR system is used to measureprofiles of the apparent optical properties (AOP) in-cluding downwelling irradiance and upwelling radi-ance at 412, 444, 490, 533, 555, 590, and 684 nm.The Satlantic radiometer is calibrated at SDSU byDrs. Mueller and Pegau following NIST protocols andcalibrated twice a year by Satlantic Inc. The opticalfilters on this radiometer were upgraded by the manu-facturer (Satlantic) over the past year.

One meter binned profiles of all of the data, areprovided to the SeaBASS database within 90 days ofcompletion of each cruise. Also provided are the cali-bration histories of the ac-9's and the radiometer, and adetailed logbook. In the future we plan to make thesedata sets available via the web.

The dates of each cruise and the number of pro-files made using each system are listed in Table 1.


78

Cruises in 1997 and 1998 were SIMBIOS supported.Figure 1 shows the station locations sampled duringeach cruise. Data collected during the fall cruise in1998 is currently being processed and quality con-trolled. As is such, the locations of the profiles are notshown on Figure 1. Currently, we are receiving all ofthe SeaWiFS HRPT Level 1A data for the West Coastof North America including the Gulf of Californiafrom the DAAC. This data is being archived andprocessed to various levels for use in the validation ofocean color products. We have been contracted tomaintain an optical profiling system for use in theSWIMBIOS instrument pool. The system containstwo hyperspectral absorption and attenuation meters(WET Labs Histars), a CTD (SeaBird Electronics 25),a data acquisition system (WET Labs MODAPS+),cage, and necessary cabling. We have assembled theabove system and have been testing the system invarious local environments. Currently there are stillsome issues to be resolved as to the HiStar measure-ments in comparison to the ac-9 measurements, whichat this point seems to be calibration dependent. Weare currently working closely with WET Labs to re-solve these issues. A full time technician has beenhired to accompany this system, to assist users in de-ployment and operation and to process the data. Wealso maintain 3 deionizing water filtration systems(Barnstead Nanopure) which are for use by the SIM-BIOS authorized users. These systems are used toproduce a clean water standard for ac-9 and HiStarfield calibrations. These systems are currently in handand each has been tested and is ready for general use.We expect that the calibration issues with the profilingpackage will be fully addressed and the system readyfor general usage by the SIMBIOS authorized users bylate February 1999.


Over the past contract period, we have begunanalyzing our field data collected over the past 4years. Data collected in the Gulf of California as wellas other locations were used to examine the spectralrelationships between the inherent optical properties inBarnard et al. (1998). The results of this research

were used to compare and contrast the inherent opticalproperties observed in the Gulf of California. In theshallow northwest region of the Gulf, Case II watersare consistently observed due to the resuspension ofsediments. In the central portion of the Gulf, the larg-est variability in optical properties is associated withcoastal upwelling. Away from the coasts the spectraloptical properties are fairly constant. Validation of thenormalized water leaving radiance measured bySeaWiFS in the Gulf of California during the 1997and spring 1998 cruises was examined. The results ofthis research indicated that atmospheric correction wasproblematic, often causing negative water leavingradiances at 412 nm.

In coordination with our SeaWiFS funding, wehave undertaken the development of a new remotesensing algorithm involving the triple ratio of the re-mote sensing reflectance. We have also compared thein situ radiance with the upwelling radiance resolvedby the OCTS sensor off the northeastern U.S. coast.

21.4 WORK PLAN

We will be participating in two research cruises inthe Gulf of California during 1999, tentatively plannedfor the spring and fall. We anticipate participating inother cruises in the next year, as the opportunityarises. We will also be working on the installation ofan in-line fluorometer on the trans-Gulf ferry to pro-vide a monthly transect of chlorophyll fluorescenceand beam transmission data in the Gulf of California.

In the next year, we anticipate that the hyper-spectral profiling package will receive a high amountof usage by the authorized users. We plan to maintainthese systems as well as provide technical support tothe users. Again we anticipate that this system will beavailable for general use by the end of February 1999.We plan to continue to investigate the inherent andapparent optical properties of the Gulf of California,as well as continuing our work with ocean color algo-rithm validation and development. We will also beaddressing issues of atmospheric correction over theGulf of California, which have been found to beproblematic.

Table 1 Dates and number of profiles made during each of the Gulf of California cruises.

Cruise Name Dates # IOP profiles # AOP profilesGOC95fall 11/25/95 to 12/6/95 28 69GOC96fall 10/30/96 to 11/7/96 25 25GOC97fall 10/16/97 to 10/29/97 39 24GOC98spring 3/6/98 to 3/16/98 27 7GOC98fall 11/27/98 to 12/6/98 45 11


79

Figure 1. Map of the Gulf of California with sampling locations. The circles denote sampling locationsduring the 1995 cruise, the stars the 1996 cruise, the triangles the 1997 cruise, and the squares are for the spring1998


80

Chapter 22

Validation of the SeaWiFS Atmospheric CorrectionScheme Using Measurements of

Aerosol Optical Properties

Mark A. Miller, R. Michael Reynolds and M. J. BartholomewBrookhaven National Laboratory

Upton, New York

22.1 INTRODUCTION

The focus of the research described below iscomparisons between surface and satellite measure-ments of aerosol optical properties as a medium forvalidating the assumptions made in the aerosol modelsused in the retrieval of water-leaving radiance. Tworesearch projects are being conducted as part of theSIMBIOS contract: The first project is to develop aMarine Fast Rotating Shadow-band Spectral Radi-ometer (MFRSSR) that can be used for continuous,ship-board measurements of the spectral aerosol opti-cal depth and the spectral direct-normal and spectraldiffuse irradiance. The second project is to measureand interpret aerosol properties in regions under sur-veillance by an ocean color satellite, thereby providinga data set from which to evaluate the atmospheric cor-rection algorithm. This second project entails the useof AERONET and ship-board photometers and radi-ometers. Progress toward the completion of each ofthese projects and work planned for the next period isdescribed in the subsections below.


MFRSSR

There are two primary techniques used to meas-ure the aerosol optical depth: sun photometry and ro-tating shadow-band radiometry. Just as sun photome-ters must be accurately oriented toward the sun, anobvious pitfall at sea, conventional shadow-band radi-ometers have required exact orientation. The devel-opment of fast-rotating, shadow-band radiometers, ahybrid form of the original shadow-band technique,has removed orientation requirement, whereuponshipboard use of shadow-band radiometers has beenencouraged. As compared to land-based units, ship-board fast-rotating shadow-band radiometers requiremuch faster rotation of the occulting arm, fast-response silicon detectors, higher data sampling rates,and more sophisticated data analysis. As part of a

multi-agency effort between the Department of En-ergy (DOE) and NASA SIMBIOS, the BrookhavenNational Laboratory has developed an MFRSSR. Itmeasures spectral global and spectral diffuse irradi-ance, which are used to calculate the spectral di-rect-normal irradiance as a function of solar zenithangle. From these data, the aerosol optical thicknesscan be computed during clear periods using the Lan-gley regression technique, or continuously if highquality extraterrestrial calibration coefficients areknown.

Initial data comparisons of optical depths fromhand held sunphotometers have been encouraging andan example of a time series of optical depth from theMFRSSR and MicroTops is shown in Figure 1. Inthis case, a well-defined decrease in the optical depthduring a 9-hour measurement period was observed byboth instruments. Periods of noisy optical thicknessare the result of clouds, which have not been filteredin this plot. The MFRSSR is designed to operate forlong periods with minimal attention and it is sched-uled, along with two new units, to be at sea for a totalof 7-8 months during 1999. During this period meas-urements will be taken over three oceans aboard theR/V Ron Brown and other ships, and from an island inthe Tropical Western Pacific. At the conclusion ofthese cruises, all three units will spend several weeksat Mauna Loa for calibration. An important elementof the MFRSSR design philosophy is automated op-eration and timely data analysis. Toward this end,efforts have been made to develop a cloud-filteringalgorithm that can be used to identify cloud-free con-ditions. The initial version of this algorithm wasbased on the work of Long et al. (1999), but it hassince been modified for more efficient use in a marineenvironment.

AERONET SeaWiFS Comparisons

Another approach to validating the atmosphericcorrection algorithms is to use land-based sunpho-tometers deployed in coastal locations, such asAERONET. These systems make a full range ofmeasurements that can be used to compute the atmos-


81

pheric correction parameters defined by Gordon et al.(1980). It is not possible, with current instrumenta-tion, to measure the full range of variables necessaryto compute the atmospheric correction parameters atsea. One drawback of the land-based approach, how-ever, is that near-shore measurements may not faith-fully represent conditions further from the coast,whereupon an approach that combines ship-board andAERONET measurements is probably the best alter-native. Both approaches are part of this study.

The AERONET and SeaWiFS optical propertiesfor some selected locations are being compared usingthe SeaWiFS/AERONET match-up code developed atNASA GSFC. The initial form of this code selectsdates and times of coincident SeaWiFS andAERONET measurement and extracts the aerosol op-tical depth from the respective data files, providing thefollowing are true: (1) AERONET data that havepassed a cloud filter algorithm are available during thetwo-hour period centered on the satellite overpasstime, and (2) the SeaWiFS data in each individualpixel in a scene pass all quality control checks.

The match-up code identifies coincident SeaWiFsand AERONET measurement times at several loca-tions and detailed examinations of these data are inprogress. There are several points to be made. At pre-sent, we have analyzed the matchup data for the fol-lowing locations: Dry Tortugas, Bermuda, SanNicholas Island, Lanai, Kaashidoo, and Bahrain. Pre-liminary results suggest that the pixel-to-pixel vari-ability in the aerosol optical thickness at most of thesites is unrealistically large, perhaps due to an inade-quate cloud filtering algorithm applied to theSeaWiFs data. The best match-up data sets analyzedto date are from Bermuda, Bahrain, and Kaashidoo.Comparing Bermuda and Kaashidoo, both island sites,the aerosol optical thickness as measured by bothAERONET and satellite suggests that the aerosol op-tical thickness is generally higher at Kaashidoo, whichlies in the Indian Ocean, and it tends to be more vari-able than at Bermuda. In general, more data tend topass the SeaWiFs quality control algorithms atKaashidoo, which may reflect less cloudiness morethan any other single factor. The data suggests thatthe comparison between the SeaWiFS and AERONETis somewhat better at Bermuda than at Kaashidoo.The reason for the less favorable comparison data atKaashidoo is under investigation, but some evidencesuggests that the AERONET sunphotometer calibra-tion may have impacted this result. Before this analy-sis is complete, however, the wind vector must beconsidered in match-up pixel selection. If the windspeed is greater than a few meters per second, the cur-rent two-hour sunphotometer data averaging periodsuggests that pixels outside the scene area may need tobe considered for an appropriate comparison. Con-versely, the sunphotometer averaging window could

be reduced, although it is likely that this problem isnot a factor in many of the analyses.

The largest amount of match-up data available isfrom the Dry Tortugas. After extensive analysis of theDry Tortugas data, there appear to be some problemswith data quality, most likely the result of AERONETsunphotometer calibration or filter decay. This prob-lem may be corrected in the latest match-up data set.In general, data from Bermuda suggest that the aerosoloptical thicknesses measured by AERONET andSeaWiFs are highly compatible. In contrast, match-updata from Kaashidoo suggest potential problems witheither the AERONET or SeaWiFS aerosol opticalthickness measurements, or perhaps both. It is rea-sonable to conclude on the basis of these results thatthe atmospheric correction algorithm is highly suspectin the vicinity of Kaashidoo. Data from other match-up locations are being analyzed and preliminary re-sults suggest that SeaWiFS tends to measure a higheraerosol optical thickness at almost all locations thanthe AERONET. Additional ship and satellite aerosoloptical thickness comparisons await a more extensivematch up data base, which should be in place by theend of this year.

23.3 WORK PLANNED

There are still some issues to be addressed in thedesign of the MFRSSR. Controlled experiments needto be conducted on a motion simulator. The requiredequipment for these tests has been assembled and theexperiments will be conducted when a secondMFRSSR under construction for DOE is complete. Inaddition, pitch and roll information need to be incor-porated into the data processing algorithms used toprocess MFRSSR, especially when the instrument isdeployed on small ships.

Data from the R/V Ron Brown’s 7-month cruisefor INDOEX and DOE Nauru-99 campaigns will beused to facilitate match-ups with SeaWiFS and sub-mitted to SeaBASS. In addition, two new MFRSSRunits will be deployed in support of Nauru-99: one onthe Japanese R/V Mirau and the other on the island ofNauru in the Tropical Western Pacific. These datawill include over 200 measurement days aboard theR/V Ron Brown and additional 90 measurement daysfrom the Mirai and Nauru. Several quality match-upsare anticipated from a variety of geographic locationsand aerosol regimes. At the conclusion of INDOEXand Nauru-99, in August, all three MFRSSR units (theoriginal and the two new units) will be offloaded atHawaii and taken to Mauna Loa for a three week cali-bration, along with the microtops used in these ex-periments. Before shipment back to Brookhaven Na-tional Laboratory, one MFRSSR will be used in acollaborative cruise with Dr. John Porter at the Uni-versity of Hawaii. Hence, over 300 ship-board meas-


82

urement days are planned this fiscal year and, oncethese data have been processed, satellite match-upswill be made. The AERONET/SeaWiFS matchupanalysis is continuing. Data from all of the match-upsites will be analyzed. Particular emphasis will beplaced on interpreting the match-up data from loca-

tions, such as Kaashidoo, where the analysis suggestsproblems with the atmospheric correction algorithm.In addition, attempts will be made to interpret theAERONET- SeaWiFS match-up data in the context ofthe full set of atmospheric correction parameters asdescribed by Gordon et al. (1980).

Figure 1. Spectral optical thickness measurements made with the MFRSSR at 415 nm, 500 nm, 610 nm,660 nm, and 862 nm on January 27, 1999 over the Atlantic Ocean near the equator. Indicated in the boxes areinstantaneous hand-held sun photometer measurements at 440 nm, 500 nm, 670 nm, and 870 nm.


83

Chapter 23

Measurements of Aerosol, Ocean and Sky Propertiesat the HOT Site in the Central Pacific

John N. Porter and Ricardo LetelierHawaii Institute of Geophysics and Planetology,

University of Hawaii, Honolulu, Hawaii

23.1 INTRODUCTION

Monthly cruises have been made to the HawaiiOcean Time Series (HOT) site (~100 km north ofOahu) since October 1988. The goal of these cruises isto make hydrography, chemistry, and biology obser-vations (PI: Dave Karl and Roger Lucas). Measure-ments are typically made at a near coastal station(Kahe) on the first day to test the equipment and toobtain coastal shallow water (~1500m) observations.The second and third days are spent at the HOT site.On the morning of the fourth day, measurements aremade at the HOT site and noon time measurementsare made at the Hale-Aloha station near the mooringbefore returning to port by early the next morning.The locations of the three stations are: • Kahe (21.34o N, 158.27o W)• HOT (22.75o N, 158.0oW)• Hale-ALOHA buoy (22.43oN, 158.0oW).

The routine HOT measurements are availableduring the summer following the year of the observa-tions. The 1998 measurements will therefore becomeavailable during the summer of 1999. The data setscan be obtained at the National Oceanographic DataCenter (NODC) or from the Hawaii HOT web site(www.hahana.soest.hawaii.edu/hot/hot_jgofs.html).Selected parts of this data set will be submitted to theSeaBass archive as the data comes on line.


Microtops Sun Photometers

Beginning with the HOT89 cruise, aerosol opticaldepth measurements were made with two Microtopssun photometers. Examples of the data sets collectedare shown in Figure 1. Aerosol optical depth meas-urements were made at 380, 440, 500, 675, 870 and1020 nm. Column integrated water vapor and ozoneconcentrations were also derived. These measure-ments have been submitted to the SeaBass archive. Ingeneral we find the aerosol optical depths to be

roughly the same (~0.05) at all wavelengths for typicaltrade wind conditions where sea salt dominates. Asiandust periods also have a flat spectral aerosol opticaldepth but can have optical depths of 0.1 and higher.Under VOG (volcanic smog) conditions, the shorterwavelengths have larger optical depths such as thecase shown in Figure 1 (left panel). Calibration hasbeen a major concern and numerous calibration effortshave been carried out at Mauna Loa and Haleakala.The results of these calibrations are good and they arereported in a technical report submitted to SIMBIOS.The same report goes into detail on calibration prob-lems, measurement problems, and suggested protocolsfor using the Microtops hand held sun photometers.Comparisons between the ship sun photometer meas-urements and the SeaWiFS aerosol optical depths at870 are shown in Figure 2. In general the agreement isabout ∀0.04 with the exception of two corrispondingcases which had flat spectral optical depths higherthan 0.06, suggesting Asian dust may have been pres-ent. For VOG we found a systematic under-predictionof the aerosol optical depth by about 20%.

Marine Shadowband Radiometer

During the past year we have been working on amarine shadowband radiometer which will measureaerosol optical depths and downwelling irradiance at~20 wavelengths. This system uses a gimbaled cosineresponse detector and a rotating shadowband armwhich shadows the detector. A CVI Laser spectrome-ter is used to collect the light and a PC104 computer isused to collect the data. The system has been tested onseveral cruises and the gimbal works well. An algo-rithm has been written to collect the total and diffuselight and the direct solar component is obtained fromthe difference. A cloud screening algorithm has alsobeen developed which detects clouds by the variationin the downwelling irradiance just before and after thedetector is shadowed. Although we believe we havemade significant progress in many areas, we still havenot resolved the cosine response problem as we wouldlike near perfect cosine response with the cosine re-sponse known to 0.1%. We have tried diffusers made


84

of teflon, flashed opal, ground glass, spectralon, andintegrating spheres. We now believe that the integrat-ing sphere or the spectralon offer the best perform-ance. Several new designs are being machined.

Radiometric Calibration Facility

In the past we have carried out radiance calibra-tion of our systems at the Moby calibration facility(thanks to Dennis Clarke). Now we have purchased aNIST traceable irradiance lamp and integrating spherefrom Optronics. These systems were set up in a darkroom at the University of Hawaii for optical calibra-tion. Our plans are to maintain this facility with rou-tine checks. Unfortunately, during a SIRREX com-parison, we found our calibration was approximately5.5% too low. We could not understand this differ-ence. Therefore we have purchased a shunt resistorwhich was recently calibrated by a NIST traceablecalibration facility. We will soon use this calibratedresistor to check the current output of our power sup-ply. The Optronics power supply is the first suspect asit had a transistor failure which was replaced prior toour SIRREX comparison. We expect this problem willsoon be solved.

In Water Optics Measurements

As part of this SIMBIOS effort, in water PRR(Profiling Reflectance Radiometer) and TSRB (Teth-ered Spectral Radiometer Buoy) measurements werebegun on HOT90 cruise (Dec. 1997). PRR measure-ments are made by Karl et al. and processed by Ri-cardo Letalie. Calibration of this instrument was car-ried out and the coefficients have been submitted tothe SeaBASS archive. A description of the data andcalibration is available athttp://picasso.oce.orst.edu/users/jasmine/ORSOO/.

Aerosol Phase Function Measurements

As part of our initial SIMBIOS proposal, we pro-posed to develop a polar nephelometer which couldmeasure the aerosol phase function. Throughout the

past year we have been developing this system andhave recently collected our first measurements fromthis prototype instrument. Our design follows the sys-tem described by Winchester Jr. (in Optical Engi-neering, 1983, pg. 40). Our first measurements weremade at Bellows beach directly downwind of smallbreaking waves. We have recently purchased a pulsedlaser which will allow for phase function measure-ments at 3 wavelengths (1064, 532, 355 nm). Thesystem is being designed to install in the door of alight aircraft and we expect to hope to make columnaerosol phase function measurements in the future.

Sky and Surface Radiometer

We are developing a hand radiometer to sky radi-ance and above surface upwelling radiance. The skyradiance will be used to derive aerosol size distribu-tion information and the surface radiance will be usedto validate satellite measurements and to study thediffuse reflectance problem. This system uses a wideangle camera and a tilt meter to determine the viewinggeometry and a spectrometer to measure the sky or theaureole. The scattering angle is derived from the posi-tion of the sun in the camera image and a tilt meter.The system will be in a weather proof enclosure usinga PC/104 computer to control the two axis motor, thecamera, and the spectrometer. We had expected tohave a prototype system working by now but delayshave pushed it back to Spring 1999.

23.3 WORK PLAN

We have three goals: 1) continue collection ofexisting measurements on the HOT cruises, 2) finish-ing the construction of new instruments (shadowbandradiometer, polar nephelometer, and sky radiometer),and 3) carry out modeling efforts. The modeling effortwill focus on developing new algorithms to derivecolumn aerosol properties from surface measurementsand estimating the impact of aerosol on upwellingradiance at the top of the atmosphere.

Sean Bailey

Page 85 containing images for this chapter has been deleted due to an error in conversion to PDF.


86

Chapter 24

Assessment of the Contribution of the Atmosphere toUncertainties in Normalized Water-Leaving Radiance:

a Combined Modelling and Data Analysis Approach

Knut Stamnes and Bingquan ChenGeophysical Institute

University of Alaska Fairbanks, Fairbanks, Alaska

24.1 INTRODUCTION

Atmospheric correction, i.e. the process of deriv-ing the normalized water-leaving radiances from sat-ellite-measured imagery of the oceans, has been es-tablished as one sub-goal of the SIMBIOS programfor which the main purpose is to provide routine, longterm ocean color data from a variety of internationalocean color missions. Our research for the SIMBIOSprogram is focused on the study of atmospheric cor-rection. The overall objective is to use a sophisticated,state-of-the-art radiative transfer model for the cou-pled atmosphere-ocean system, with existing andplanned algorithms, and in conjunction with datataken by other investigators in the SIMBIOS program,to help quantify the uncertainties in the water-leavingradiance due to inadequacies in atmospheric correc-tions applicable to OCTS, POLDER, SeaWiFS,MODIS, and other ocean color sensors.

Efficient radiative transfer codes for the atmos-phere-ocean system are not generally available, andthe popular and intuitively appealing Monte Carlomethod (Morel and Gentili, 1991) is too time-consuming to be used in routine retrieval algorithms.To solve the radiative transfer equation for the cou-pled atmosphere-ocean system, we employ the dis-crete-ordinate method (Jin and Stamnes, 1994). Thismodel for computing the transfer of radiation in thecoupled atmosphere-ocean system is reliable and effi-cient. It has the accuracy of the Monte Carlo method,but overcomes its computational inefficiency.


Water-leaving Radiance at NIR Wavelengths

The main goal of this study is to assess the un-certainty or error incurred in atmospheric correctionby ignoring the water-leaving radiance caused byscattering from particles in the near-surface water atNIR wavelengths between 660-870 nm. Aerosol mul-tiple scattering is included in the forward modeling to

provide realistic simulations of the atmospheric con-tribution to the TOA radiance. We also employ a bio-optical model of ocean water with ocean particle con-centrations of 1.5, 5.0, and 10.0 mg/m3, respectively,and we employ phase functions for phytoplanktonparticles (i.e., cyanobacteria and algae, Volten et al.,1998) in the simulations conducted in this study.

Forward modeling is used to simulate or predictradiances that ocean color sensors would measure.From the deviation between the TOA radiances pre-dicted by a model of the coupled atmosphere-oceansystem (Jin and Stamnes, 1994) and the same model inwhich scattering by ocean particles (and hence water-leaving radiances) at NIR wavelengths has been ig-nored, one can determine under what conditions thewater-leaving radiance at NIR wavelengths can beignored in atmospheric corrections. For each sensorchannel we use forward modeling simulations to ac-curately predict the impact of both atmospheric aero-sols and biogenic constituents in the ocean on TOAradiances for any desired sun-satellite geometry, solarzenith angle θ 0, sensor viewing polar angle θ, andrelative azimuthal angle φ with respect to the sun.

To describe scattering by particles in the oceanwe employ the Henyey-Greenstein phase function pHG

(Henyey and Greenstein, 1941):

This phase function has the desirable feature thatit yields complete forward scattering when the asym-metry factor g=1, isotropic scattering when g=0, andcomplete backward scattering when g=-1. The linearcombination

is sometimes used to simulate a phase function withboth a forward and a backward scattering component(g>0 and g'<0). Here θ is the scattering angle, b is thefraction of forward-scattered light (0<b<1), and g and

( ) 2/3cos221

21)cos,(

Θ−+

−=Θ

gg

ggHGp

)cos,'()1()cos,()(cos Θ−+Θ=Θ gpbgbpp HGHG


87

g' are usually different. Based on laboratory measure-ments by Volten et al. (1998), we have used equation(2) with different values of g, g', and b to simulatephase functions corresponding to ocean waters withdifferent types of particles.

The measurements by Volten et al. (1998) werecarried out at 633 nm. Lacking better information weassume for our present purpose that they can be usedto approximate the phase functions at wavelengthsbetween 630 nm and 870 nm. Previous studies showthat far from sources of pollution and/or sources ofdesert aerosols the aerosol optical depth at 865 nmover the Pacific Ocean lies in the range between 0.08and 0.11 (Villevalde et al., 1994). In this study weemploy a Tropospherical aerosol model with relativehumidity of 80% (Shettle and Fenn, 1979) and use aMie code to calculate the aerosol optical properties.From forward-modeling results based on these syn-thetic phase functions we conclude that:

• The TOA radiance deviation depends on thescattering phase function of particles in the near-surface water.

• As the aerosol optical depth decreases, the TOAradiance deviation increases. When the aerosoloptical depth is large (e.g. τa=0.2), TOA radi-ances, predicted by a model in which water-leaving radiances at NIR wavelengths are ig-nored, do not deviate significantly from thoseobtained when including scattering by ocean par-ticles. Thus, at large aerosol optical depths, wemay ignore water-leaving radiances at some NIRbands (e.g. 865 nm) without introducing signifi-cant errors.

• The TOA radiance deviation increases with in-creasing ocean particle concentration. When theocean particle concentration is large and the aero-sol optical depth is low, water-leaving radiancesat NIR wavelengths contribute substantially toTOA radiances and should not be ignored.

• The TOA radiance deviation that results fromneglecting water-leaving radiances depends onthe sun-satellite geometry.

• The TOA radiance deviation at λ=765 nm islarger than that at λ =865 nm.

• The TOA radiance deviation at λ =665nm ismuch larger than that at λ =865nm. Since the 665nm wavelength has been used in atmospheric cor-rection for CZCS, this deviation can have led tosignificant uncertainties in ocean color retrievalsfrom the CZCS measurements.

To avoid the uncertainties incurred by ignoringwater-leaving radiances at NIR wavelengths retrievalsand atmospheric corrections in ocean color imageryshould be based on a radiative transfer model for thecoupled atmosphere-ocean system (Jin and Stamnes,1994).

Atmospheric Correction Algorithm

The existing atmospheric correction algorithm ismainly based on the approximate exponential relation-ship of the single scattering aerosol reflectance overthe range 412-865 nm. Two types of lookup tables forN selected aerosol models then constitute the mainpart of the algorithm. The first type of lookup table isfor the coefficient εss (λ, 865) which relates the singlescattering aerosol reflectance at wavelength λ to thatat 865 nm. The second type of lookup table is for thecoefficient k[λ, ρasλ,)] which makes a correction formultiple scattering. For each type of lookup tables,about twelve candidate aerosol models are employed;and table entries are computed for each of these mod-els for solar zenith angles between 0°and 80° in in-crements of 2.5°, for 33 values viewing zenith angles,and for eight values of the aerosol optical depth be-tween 0.05 and 0.8. The total number of separate so-lutions to the radiative transfer equation used in thepreparation of these lookup tables exceeds 33,000(Gordon, 1996). Thus, the computational effort in-volved in the construction of these lookup tables isvery large.

Based on a state-of-the-art radiative transfermodel for the coupled atmosphere-ocean system, wesimulate the process of atmospheric correction. Usingfour types of aerosol models (Tropospheric, Maritime,Coastal and Urban with RH = 50, 80 and 98 %, θ= 7°)we find that there exists an approximate exponentialrelationship of the aerosol reflectance over the range412-865 nm not only in the single scattering case, butalso in the multiple scattering case. Thus, a simple andefficient algorithm for atmospheric correction is pro-posed and discussed by comparing it with the existingalgorithm. In our algorithm, the aerosol contributionin the visible in the multiple scattering can be ex-trapolated directly from that in the near-infrared byuse of the approximate exponential relationship.Therefore it is not necessary to construct lookup tablesfor the coefficient κ[λ, ρas (λ)]. The new coefficient εms (λ, 865), which represents the ratio of the multiplescattering aerosol reflectance at λ to that at 865 nm, isintroduced in the new algorithm instead of ε ms (λ,865) used in the existing algorithm. Future testing ofthis algorithm will be carried out by use of satellitedata and field measurements. The most important con-clusion of this study is that one should examinewhether or not water-leaving radiances at NIR wave-lengths can be ignored when developing algorithmseither for the retrieval of atmosphere-ocean parame-ters or for atmospheric corrections. Otherwise, errone-ous results can be obtained if such algorithms are ap-plied to satellite-received data, even for open oceans.A simple and efficient algorithm for atmospheric cor-rection is also proposed and compared with the exist-ing algorithm.


88

Chapter 25

Validation of the Water-Leaving Radiance Data Product

Oleg V. KopelevichOcean Optics Laboratory

P.P. Shirshov Institute of Oceanology, Moscow, Russia

25.1 INTRODUCTION

The overall objective of the research is to assessan accuracy of retrieval of the water-leaving radiancefrom satellite ocean color data. Specific goals includean assessment of accuracy of the atmospheric correc-tion algorithms for variety of meteorological andoceanological conditions, effects of the rough sea sur-face reflectance and of the small-scale spatial inho-mogeneities in the subsurface layer (sub-pixel vari-ability), an accuracy of the water-leaving radiancevalues obtained from in situ measurements. Combinedapproach is used which includes direct comparisonbetween the values of water-leaving radiance derivedfrom satellite data and in situ measurements, computersimulation, theoretical analysis.


Field Methods and Data

The instrumental set constructed in theP.P.Shirshov Institute of Oceanology Russian Acad-emy of Sciences (SIO RAS) for field studies com-prises deck and floating spectroradiometers andmonitor photometer. The deck spectroradiometer is amodification of the device described before (Goldin etal. 1983); the floating spectroradiometer and themonitor photometer are new instruments (Artemyev etal., 1999). The floating spectroradiometer measuresthe upwelling radiance and downwelling irradiance inthe spectral range 390-700 nm with spectral resolution2.5 nm; the upwelling radiance is measured just be-neath the sea surface and the downwelling irradiancejust above the sea surface at a distance about 50 mfrom ship; thus the difficulties connected with influ-ence of sun glints, correction for light reflected at thesea surface, influence of ship shadow are precluded.The monitor photometer is designed to check calibra-tion of irradiance at 554 nm and measure continuouslythe surface irradiance at this wavelength during spec-tral measurements by the spectroradiometers. Theradiometric accuracy of the deck and floating spectro-radiometers and the monitor photometer is respec-tively 10%, 5% and 3%. The field testing of the float-ing spectroradiometer were carried out on the oceano-

grafic platform of the Marine Hydrophysical InstituteNASU in Katsiveli, September 1996. They showed anability of the instrument to work under bad weatherconditions (strong wind, cloudiness) when measure-ments by the deck radiometer are impracticable. The field studies under satellite ocean color sensorsSeaWiFS and MOS-IRS were performed during thescientific cruise of RV Akvanavt in the Black and Ae-gean Seas, October 6-24, 1997 (Artemyev et al.,1999). Along with measurements of spectral upwell-ing radiance and surface irradiance by the deck andfloating radiometers, the studies included measure-ments of vertical profiles of nadir upwelling radianceand downwelling irradiance by means of the sub-mersible radiometer MER-2040 (these measurementswere performed by specialists from the Polish Instituteof Oceanology, Sopot); determination of the atmos-pheric spectral optical thicknesses by means of the sunphotometer; measurements of vertical profiles of theseawater beam attenuation coefficient by the sub-mersible transmissometer; sampling and conservationof water samples for determination of phytoplanktonpigment concentration. Five drift stations were carriedout during the cruise, three in the Black Sea and twoin the Aegean Sea. The information on the stations isgiven in Table 1. Absolute radiometric calibration of the radianceand irradiance channels was performed before theexpedition. During the cruise, intercalibration of thefloating spectroradiometer and MER-2040 was carriedout; an example of intercalibration is shown in Fig. 1.Discrepancy between values of Lu and Ed measured bythese two instruments was mainly in limits of 5%. In August-September 1998 the field studies in theeastern part of the Barents Sea were carried out duringthe 13th and 14th scientific cruises of R/V AkademikSergey Vavilov. They included the measurementslisted above, excluding measurements by MER-2040,and additionally measurements of the diffuse attenua-tion coefficient at wavelength 530 nm by the Kd -meter and also determinations of the particulate matterconcentration, primary production, dissolved organiccarbon concentration, CTD profiles. The informationon the stations with optical measurements is given inTable 2.


89


The validation cruise of RV Akvanavt in theBlack and Aegean Seas

The validation of the SeaWiFS atmospheric cor-rection algorithm was performed with measured datafrom Station 2 and Station 4 where the measurementswere carried out under appropriate weather conditions.Before validation, an agreement of the values of thetop-of-the-atmosphere radiance Lt(λi) measured bySeaWiFS sensor and calculated by a Monte Carlotechnique with in situ data on ρ(λi) and τa(λi) waschecked. The procedure developed for the MonteCarlo calculation is described elsewhere (Kopelevichet al. 1998; Burenkov et al., 1999). The results ofcomparison between the measured and calculated val-ues of Lt(λi) are given in Table 3. As it is seen, theagreement is quite well for Station 2; it should beborne in mind that the observed discrepancies resultnot only from errors of the measurements but fromalso some assumption taken for the calculation. Theagreement is worse for Station 4; it can be explainedby some atmospheric inhomogeneity due to theproximity of the cloud edge to the pixel with this sta-tion. The SeaWiFS atmospheric correction algorithmwas validated by comparison of the LWN(λi) and τa(λi)values derived from SeaWiFS data and from in situmeasurements. The results of comparison are given inTable 4. As it is seen, the agreement is quite satisfac-tory both for the LWN(λi) and τa. The SeaWiFS algorithm for retrieving chlorophyllconcentration and the analytic algorithm developed bythe specialists of SIO RAS were validated by com-parison of retrieved and measured data. The SeaWiFSalgorithm overestimates significantly the chlorophyllconcentration for Case-2 waters with predominance ofabsorption by yellow substance; the analytic algorithmresults to reasonable agreement in all cases (Burenkovet al. in press).

The Akademik Sergey Vavilov Cruises in theBarens Sea

The preliminary analysis revealed some cases ofincorrect SeaWiFS data products; an example is givenin Figure 2. The negative values of LWN(443) are seenover most part of the image (left); the chlorophyllconcentration derived from SeaWiFS data agree rea-sonably with the measured values only in open part ofthe Barent Sea and differ drastically (by a factor 20and more) in the Pechora Basin (right image).

The measured spectral dependences of the radiancereflectance exhibit a wide variety of forms (Figure 3):the spectra measured in the north-eastern part of theBarents Sea have the maximum near 420 nm and theratio LWN(490)/LWN(555) about 3 (St.1209), whereas inthe southern part of the Pechora Basin they are re-spectively near 580 nm and less than 0.5 (St.1095).According to the SeaWiFS data, the chlorophyll con-centration varies over a wide range of values: fromless than 0.5 mg⋅m-3 in the northern part to more than20 mg⋅m-3 in the Pechora Basin (Figure 2, right im-age). The in situ data show much more narrow rangeof its changes. The further analysis is planned to es-tablish the reasons for the revealed errors resultedfrom the SeaWiFS algorithms.

Effects of the Sub-pixel Variability

The mathematical simulation of effects of thesmall-scale inhomogeneities in the subsurface layer(sub-pixel variability) on the retrieval results fromsatellite data has been performed. The horizontal dis-tribution of chlorophyll concentration with the sharpchange was assumed, and the averaged spectral up-welling radiance was calculated. Then the averagedspectra were used to retrieve the values of chlorophyllconcentration. Comparison between the obtained andtrue mean values has shown that discrepancies be-tween them are no more 5 % if the amplitude ofchange is less 50%, but they increase up to 15% withthe twofold change and 45% with the fourfold. Suchsharp changes are real in the coastal waters, and po-tential errors should be taken into account.

25.4 WORK PLAN

• The further analysis of data obtained during the13th and 14th cruises of R/V Akademik SergeyVavilov in the Barent Sea, August-September1998.

• Preparation of the scientific equipment for workin the 15-th cruise of R/V Akademik Sergey Vav-ilov in the Norwegian and Greenland Seas.

• The field studies in the 15th cruise of R/VAkademik Sergey Vavilov in the Norwegian andGreenland Seas, May-June 1999.

• Analysis of data obtained during the 15th cruise ofR/V Akademik Sergey Vavilov in the Norwegianand Greenland Seas, May-June 1999.

• Validation of the SeaWiFS algorithm for retriev-ing the chlorophyll concentration in the Black Seaby comparison of the satellite and available in situdata.


90

Figure 1. Results of intercalibration between the floating spectroradiometer and MER-2040 at St.2; uppercurves - surface irradiance, lower - upwelling radiance; ♦ - floating spectroradiometer, • - MER-2040.

Figure 3. Spectra of the water radiance reflectance measured in different parts of the Barents Sea duringthe 13-th and 14-th cruises of R/V Academik Sergey Vavilov: St.1209 - the north-eastern part of the Barents Sea;St.1088 and St.1095 - the northern and southern parts of the Pechora Basin (see Table 2).


91

Figure 2. Spatial distributions of values of LWN (443) and Chl a in the eastern part of the Barents Sea onAugust 25, 1998


92

Table 1. The information on the stations of Akvanavt’97.

Date St. Coordinates (degrees)at the measurement

moments

Localtime

Difference withGMT, h

10/07/97 1 * 42.5 N, 39.5 E 09:20 - 14:40 +310/08/97 2 *,** 43.0 N, 35.6 E 10:05 - 15:40 +310/09/97 3 42.9 N, 31.6 E 10:48 - 15:11 +310/11/97 4 *, ** 39.3 N, 25.1 E 11:30 - 14:55 +310/16/97 5 ** 39.6 N, 25.8 E 10:19 - 14:58 +3* - under SeaWiFS; ** - under MOS-IRS.

Table 2. The information on the stations of R/V “Akademik Sergey Vavilov” 13 and 14 cruises.

Date St. Coordinates (degrees)at the measurements

moments

Local time Difference withGMT, h

08/11/98 1088 70.42 N, 47.58 E 10:20 - 14:55 +408/12/98 1090* 70.18 N, 52.42 E 13:40 - 15:00 +408/14/98 1095 68.97 N, 58.47 E 08:15 - 10:30 +408/14/98 1097 69.07 N, 58.42 E 16:20 - 17:30 +408/15/98 1099 69.10 N, 58.03 E 09:00 - 10:30 +408/19/98 1112* 69.09 N, 58.29 E 11:30 - 12:50 +408/19/98 1112* 69.09 N, 58.29 E 13:55 - 14:25 +408/23/98 1123 69.06 N, 58.16 E 07:00 - 10:00 +408/23/98 1125* 69.50 N, 57.25 E 13:10 - 14:20 +408/23/98 1126 69.67 N, 57.24 E 16:25 - 17:30 +408/24/98 1131* 69.77 N, 56.28 E 14:10 - 16:00 +408/25/98 1133 70.33 N, 55.30 E 10:00 - 14:40 +408/29/98 1150 69.33 N, 50.12 E 08:10 - 09:30 +408/30/98 1154* 69.62 N, 50.57 E 10:30 - 14:50 +408/31/98 1157* 70.54 N, 52.79 E 12:30 - 14:30 +409/02/98 1164 68.40 N, 50.70 E 08:20 - 10:20 +409/10/98 1174 69.25 N, 41.00 E 08:40 - 10:40 +409/11/98 1183 71.50 N, 41.00 E 09:00 - 11:00 +409/12/98 1190 73.25 N, 41.00 E 10:00 - 12:00 +409/13/98 1196 74.75 N, 41.00 E 09:30 - 11:30 +409/15/98 1209 78.00 N, 41.00 E 10:20 - 13:00 +409/27/98 1281 76.00 N, 42.27 E 10:20 - 12:40 +4* - under SeaWiFS

Table 3. Comparison between the SeaWiFS measured and Monte Carlo calculated values of the top-of-the-atmosphere-radiance Lt(λi); ∆ is the relative discrepancy.

Station 2 Station 4λ,nm

Lt SeaWiFS,µW cm-2 sr-1 nm-1

Lt MC,µW cm-2 sr-1 nm-1

∆,%

Lt SeaWiFS,µW cm-2 sr-1 nm-1

Lt MC,µW cm-2 sr-1 nm-1

∆,%

412 7.40 7.42 0.3 8.10 7.89 2.6443 6.52 6.57 0.7 7.24 6.86 5.0490 4.84 4.83 0.2 5.35 5.16 3.5510 4.12 4.08 1.0 4.50 4.21 6.4555 2.93 2.80 4.4 3.18 2.96 6.9670 1.25 1.16 7.2 1.55 1.44 7.1765 0.55 0.59 7.2 0.77 0.73 4.8865 0.36 0.33 8.3 0.55 0.43 22


93

Table 4. Comparison between values derived from SeaWiFS and in situ data for the normalized water-leaving radiance LWN(λi) and the aerosol optical thickness τa(865); ∆ is the relative discrepancy.

Station 2 Station 4λ, nm LWN SeaWiFS,

µW cm-2 sr-1 nm-1LWN measured,µW cm-2 sr-1 nm-1

∆,%

LWN SeaWiFS,µW cm-2 sr-1 nm-1

LWN measured,µW cm-2 sr-1 nm-1

∆,%

412 0.536 0.492 9 0.770 0.732 5443 0.752 0.708 6 0.886 0.891 0.6490 1.036 0.998 4 0.882 0.784 6510 0.991 0.925 7 0.592 0.537 10555 0.770 0.772 0.3 0.271 0.272 0.4τa(865) 0.065 0.095 31 0.160 0.142 13


94

Chapter 26

OCI Data Validation in the Waters Adjacent to Taiwan

Hsien-Wen Li, Chung-Ru Ho and Nan-Jung Kuo

Department of Oceanography

National Taiwan Ocean University, Keelung, Taiwan

26.1 INTRODUCTION

Taiwan successfully launched her first experi-mental satellite, ROCSAT-1, on January 27 1999. It isa three-axis stabilized, low-earth orbit satellite with a

35° inclination. ROCSAT-1 is designed to carry outthree scientific experiments: ocean color imaging,ionospheric plasma and electrodynamics, and Ka-bandcommunication.

The payload instrument for ocean color imagingon ROCSAT-1 is called Ocean Color Imager (OCI).OCI is a multi-spectral push-broom imager, which isdesigned to map six reflected spectral radiances fromocean surfaces. The six spectral bands are identical tothe six of the eight spectral bands on SeaWiFS. Theband characteristic comparison of OCI and SeaWiFSis listed in Table 1.

In order to have successful operation of OCI,there are two scientific groups under the OCI project,the Science Team (ST) and the Science Data Distribu-tion Center (SDDC). The purposes of the ST are tocalibrate the sensor, to correct for the atmosphericinfluence, to develop bio-optical algorithms, and tocalibrate and validate OCI data. The SDDC at Na-tional Taiwan Ocean University (NTOU) is responsi-ble for processing OCI data and then distributing it tousers.

OCI will perform imaging at any time between9:00 and 15:00 local time. The normal mission opera-tion of OCI is to map ocean surface pigments when-ever weather permits.

Ideally, OCI will acquire ocean surface pigmentdata when cloud coverage is less than 50% of the areaalong the track. The actual OCI operation will ulti-mately be decided by the OCI-ST. A preliminary op-eration priority has been established based on geo-graphic locations, request sequences, weather condi-tions, and cross calibration and field validation areas.

For calibration and validation purposes, imagesare schedule to be taken when the OCI tracks overlapwith SeaWiFS, or the calibration buoy location orlocation of the experimental cruises.


It is believed that the OCI calibration at eachwavelength will change in some unpredictable man-ners as a function of time. Experience with previoussensors, such as Coastal Zone Color Scanner (CZCS),has shown that it is very difficult to determine a sen-sor’s calibration once it has been launched (Muellerand Austin, 1992). Unlike the SeaWiFS instrument,OCI has no on-board solar calibrator. Therefore, across-calibration program with SeaWiFS will be per-formed after OCI is available. To validate OCI andSeaWiFS data, field measurements of optical proper-ties are carried out with a Tethered Spectral Radi-ometer Buoy (TSRB-II) and a SeaWiFS ProfilingMulti-channel Radiometer (SPMR) made by SatlanticCompany. TSRB-II is an optical buoy, which canmeasure the in-water upwelling radiance just beneath

the sea surface ),0( λ−uL and the incident spectral

irradiance above the sea surface )(λsE . SPMR is aprofiling radiometer that can measure the in-water

upwelling radiance ),( λZLu and downwelling

spectral irradiance ),( λZEd with depth Z.

To obtain )(λwL that is measured by the OCI it

is necessary to propagate ),0( λ−uL upward through

the sea surface as

)(

),(1),0()(

2 λθλρ

λλw

uwn

LL−

= − (1)

where ),( θλρ and )(λwn are the Fresnel reflec-tance at the solar zenith angle and the refractive indexfor seawater, respectively. In order to remove the in-fluence of view angle, sun angle and the solar irradi-ance, the water-leaving radiance is generally trans-ferred to normalized water-leaving radiance Lwn as

)(

)()()( 0

λλ

λλs

wwn E

FLL = (2)


95

where )(0 λF denotes the mean extraterrestrial solar

irradiance (Neckel and Labs,1984). For calculating theremote sensing reflectance just above the sea surface

),0( λ+rsR the following equation was used

)(/),0(54.0),0( λλλ surs ELR −+ = (3)

Since the band characteristics of OCI are similarto those of SeaWiFS, some of bio-optical algorithmsfrom the SeaWiFS Bio-optical Algorithms Mini-Workshop (SeaBAM) are evaluated using the data setsthat are collected in the water adjacent to Taiwan. Theevaluated algorithms are listed in Table 2. We havecollected a total of 29 in situ data including opticalproperties and water samples in order to derive therelationship between optical properties and chloro-phyll a concentration. All samples, except the surfaceseawater, were collected by Go-Flo 2.5/5 L samplerwhich was attached on the rosette multi-sampler. Sur-face seawater was collected with a bucket. One liter ofseawater was immediately filtered by Whatman GF/F(25 mm) filter, under pressure (<0.2 bar). To deter-mine the chlorophyll a concentration, the filteredsample was frozen in liquid nitrogen on board, whilein the laboratory the N, N-dimethylformamide (DMF)extracting solvent was added in darkness at -20°.Compared with the acetone extracting solvent, DMFhas some advantages (Suzuki and Fujita, 1986; Porraet al., 1989; Suzuki and Ishimaru, 1990) and wastherefore used in this study. The measurement of chlo-rophyll a was undertaken in the laboratory accordingto the fluorescence method described by Stricklandand Parsons (1972).


A total of 29 bio-optical data were made in thetime period of 1998 in the Kuroshio region near Tai-wan for testing the SeaBAM algorithms. The field

measurements include )(λsE , ),( λZLu ,

),( λZEd , and water sample for determining the

chlorophyll a concentration. The Lwn is computedusing Eqs. (1), and (2). The Rrs is obtained from Eq.(3). Most of the water samples were taken in the CaseI water. The range of measured chlorophyll a concen-tration is from 0.08 ug/l to 1.25 ug/l in this data set.The statistical results between calculated chlorophyll aconcentration using some of SeaBAM algorithms andmeasured chlorophyll a concentration from water

samples are listed in Table 3. The results show that R2

of Aiken-C, CalCOFI 2-band linear, CalCOFI 2-bandcubic, Morel 2, and SeaWiFS algorithms are higherthan 0.8. The RMS is smaller than 0.015 for thesemethods. This suggests that the SeaWiFS algorithm isalso suitable for the waters adjacent to Taiwan.

26.4 WORK PLAN

In situ measurements of optical properties andchlorophyll a concentration will be continuous in thewaters adjacent to Taiwan, especially in Taiwan Straitand East China Sea. The schedule for normal opera-tion of OCI is from April 1999. Therefore the inter-comparison between SeaWiFS and OCI will be themost important plan for next period. Some ofSeaWiFS project members will also take bio-opticaldata for validating OCI data. OCI data will be avail-able in April 1999.

ACKNOWLEDGMENTS

Many thanks go to captains and crew of R/VOcean Researcher II and R/V Ocean Researcher III forshipboard assistance. We appreciate the assistance ofthe GSFC/DAAC of NASA for providing theSeaWiFS data. This work is supported by the NationalSpace Program Office (NSPO) of Taiwan under grantsNSC83-NSPO-A-RDD-019-002, NSC84-NSPO-(A)-OCI-001-02, NSC85-NSPO-(A)-OCI-019-02, NSC86-NSPO-(A)-OCI-019-01, NSC86-NSPO-(A)-OCI-019-02, NSC87-NSPO-(A)-OCI-019-01, NSC87-NSPO-(A)-OCI-019-02, NSC88-NSPO-(A)-OCI-019-01, andNSC88-NSPO-(A)-OCI-019-02.

Table 3. Summary of statistical results of algorithm evaluations.

Method 1 2 3 4 5 6 7 8 9

R2 0.859 0.833 0.816 0.772 0.825 0.769 0.794 0.792 0.848

RMS 0.012 0.014 0.015 0.019 0.015 0.020 0.017 0.018 0.013


96

Table 1. A Comparison of characteristics between SeaWiFS and OCI.SeaWiFS OCI

Inclination 98.25° 35°

Altitude (km) 705 600

Period (min) 98.9 96.6

Orbital repeat time (days) 16 52

Spectral bands (nm) B1 402-422B2 433-453B3 480-500B4 500-520B5 545-565B6 660-680B7 745-785B8 845-885

B1 433-453B2 480-500B3 500-520B4 545-565B5 660-680B6 845-885B7 545-565

Nadir pixel (m2) 1130 x 1130 800 x 800

Swath width (km) 2801 702

Redundancy No 555 nm

Color sensing Scanner push broom

Crossing equator time (local time)

12:00 9:00 ~ 15:00

Bits 10 12

Tilt -20°, 0°, 20° No

Launch date (year/month) 1997/8 1999/1

Table 2. Bio-optical algorithms performed for the data taken in the waters adjacent to Taiwan.

Method Algorithms Band ratio(R), Coefficient(s)1. Aiken-C C=EXP(a0+a1*ln(R))

C=(R+a2)/(a3+a4*R)R=Lwn490/Lwn555a=[0.464,-1.989,-5.29,0.719,-4.23]

2. CalCOFI 2- band Linear C=10^(a0+a1*R) R=log(Rrs490/Rrs555)a=[0.444,-2.431]

3. CalCOFI 2- band Cubic C=10^(a0+a1*R+a2*R^2+a3*R^3) R=log(Rrs490/Rrs555)a=[0.450,-2.860,0.996,-0.3674]

4. Morel 1 C=10^(a0+a1*R) R=Log(Rrs443/Rrs555)a=[0.2492,-1.768]

5. Morel 2 C=EXP(a0+a1*R) R=Ln(Rrs490/Rrs555)a=[1.077835,-2.542605]

6. Morel 3 C=10^(a0+a1*R+a2*R^2+a3*R^3) R=Log(Rrs443/Rrs555)a=[0.20766,-1.82876,0.75885,-0.73979]

7. Power law model C=EXP(a0+a1*R) R=Ln(Lwn443/Lwn555)a=[-0.0177,-1.233]

8. Carder Ratio Log(C)=a0+(a1+a2*log(R))*log(R) R=Rrs490/Rrs555a=[0.0469,-1.0129,-0.7416]

9. SeaWiFS C=10^(a0-a1*R+a2*R^2+a3*R^3)+a4 R=log(Rrs490/Rrs555)a=[0.2974,-2.2429,0.8358,-0.0077,-0.0929]


97

GLOSSARY

ACE Aerosol Characterization Experiment

ADEOS Advanced Earth Observation Satellite

(Japan)

AERONET Aerosol Robotic Network

AM-1 Not a acronym, used to designate the

morning platform of EOS

AOP Apparent Optical Properties

AOT Aerosol Optical Thickness

APV Autonomous Profiling Vehicle

ARGOS Not an acronym, but the name given to

the data collection and location system

on the NOAA Operational Satellites.

ASCII American Standard Code for Information

Interchange

AVHRR Advanced Very High Resolution

Radiometer

AVIRIS Advanced Visible and Infrared Imaging

Spectrometer

BATS Bermuda Atlantic Time-series Study

BBOP Bermuda Bio-Optics Profiler

BBSR Bermuda Biological Station for Research

BNL Brookhaven National Laboratory

BTM Bermuda Test Mooring

Cal/Val Calibration and Validation

CalCOFI California Cooperative Oceanic

Fisheries Investigation

CALVAL Calibration Validation

CARICO Carbon Retention in a Colored Ocean

Case-1 Water whose reflectance is determined

by absorption.

Case-2 Water whose reflectance is significantly

influenced by scattering.

CCD Charge-Coupled Device

CDOM Chromophoric Dissolved Organic Matter

CHN Carbon, Hydrogen and Nitrogen

CHORS Center for Hydro-Optics and Remote

Sensing (San Diego State University)

CICESE Centro de Investigation Cientifica y de

Edcacion Superior de Ensenada

(Mexico)

CIMEL The name of a sun photometer manu-

facturer

CNES Centre National d’Edudes Spatialle

CONICIT Consejo Nacional de Investigaciones

Cientifica y Technologicas (Venezuela)

CRAM Conditional Relaxation Analysis Method

CTD Conductivity-Temperature-Depth

CZCS Coastal Zone Color Scanner

DAAC Distributed Active Archive Center

DLR Deutshe Forshungsanstalt f r Luft-und

Raumfarhrt (German Aerospace Center)

DMF Dimethylformamide

DOC Dissolved Organic Carbon

DoD Department of Defense

DOE Department of Energy

ECOHAB Ecology of Harmful Algal Blooms

EEZ Exclusive Economic Zone

EORC Earth Observation Research Center

EOS Earth Observing System

FFP Firm-Fixed Price

FOV Field of View

ftp File transfer protocol

FWHM Full Width Half Maximum

GAC Global Area Coverage, coarse resolution

satellite data with a nominal ground re-

solution at nadir of approximately 4 Km

GB Gigabyte, or about one billion bytes

GF/F Not an acronym, but a specific type of

glass fiber manufactured by Whatman.

GLI Global Imager

GoCal Gulf of California

GPS Global Positioning System


98

GSFC Goddard Space Flight Center

HIVE High-Latitude Intercomparison and

Validation Experiment

HOBI Hydro-Optics, Biology and

Instrumentation

HOT Hawaii Ocean Time series

HPLC High Performance Liquid

Chromatography

HQ Headquarters

HRPT High Resolution Picture Transmission

ICESS Institute for Computational Earth

Science System

IDL Interactive Data Language

INDOEX Indian Ocean Experiment

IOCCG International Ocean Color Coordinating

Group

IOP Inherent Optical Properties

IR Infrared

IRS Indian Remote Sensing Satellite

ISCCP International Satellite Cloud

Climatology Project

ISPO Interim SIMBIOS Project Office

JGOFS Joint Global Ocean Flux Study

JPL Jet Propulsion Laboratory

LAC Local Area Coverage, fine resolution

satellite data with a nominal ground re-

solution at nadir of approximately 1Km

LIDAR Light Detection and Ranging Instrument

LED Light Emitting Dicode

LOA Laboratorie d’Optique Atmospherique

MARMAP Marine Resources Monitoring, Assess-

ment and Prediction

MBARI Monterey Bay Aquarium Research In-

stitute

MB Megabyte, or about one million bytes

MER Multispectral Environmental Radiometer

MERIS Medium Resolution Imaging Spec-

trometer

MFRSSR Marine Fast Rotating Shadow-band

Spectral Radiometer

MISR Multi-angle Imaging SpectroRadiometer

MOBY Marine Optical Buoy

MODAPS Modular Ocean Data and Power System

MODIS Moderate Resolution Imaging

Spectroradiometer

MOS Modular Optoelectronic Scanner

MSl12 Multi-Sensor level-1B to level-2 code

MTPE Mission to Planet Earth

NASA National Aeronautics and Space

Administration

NASDA National Space Development Agency

of Japan

NAVOCEANO Naval Oceanographic Office

NDBC National Data Buoy Center

NIMBUS Not an acronym, but a series of NASA

experimental weather satellites contain-

ing a wide variety of atmosphere, ice,

and ocean sensors.

NIST National Institute of Standards and

Technology

NOAA National Oceanic and Atmospheric Ad-

ministration

NODC National Oceanographic Data Center

NOPP NIMBUS Observation Processing Sys-

tem

NRA NASA Research Announcement

NSF National Science Foundation

NSPO National Space Program Office

NTOU National Taiwan Ocean University

OCI Ocean Color Imager

OCTS Ocean Color and Temperature Sensor

ONR Office of Naval Research

ORCA Optical Research Consortium


99

of the Arctic

OSC Orbital Sciences Corporation

OSU Oregon State University

PAR Photosynthetically Available Radiation

PHILLS Hyperspectral Imager for Low Light

Spectroscopy

PI Principal Investigator

PM-1 Not a acronym, used to designate the

afternoon platform of EOS

POLDER Polarization Detecting Environmental

Radiometer (France)

PRR Profiling Reflectance Radiometer

R&D Research and Development

R/V Research Vessel

ROCSAT Republic of China (Taiwan) Satellite

platform for the OCI sensor.

RR Round-Robin

SAB South Atlantic Bight

SBE Sea-Bird Electronics

SCOR Scientific Committee on Oceanic Re-

search

SDDC Science Data Distribution Center

SeaOPS SeaWiFS Optical Profiling System

SeaBAM SeaWiFS Bio-optical Algorithms Mini-

workshop

SeaBASS SeaWiFS Bio-optical Archive and Stor-

age System

SeaDAS SeaWiFs Data Analysis System

SeaWiFS Sea-viewing Wide Field-of-view Sensor

SIMBAD The name of a sun photometer

SIMBIOS Sensor Intercomparison and Merger for

Biological and Interdisciplinary Oceanic

Studies

SIO Scripps Institution of Oceanography

SGI Silicon Graphics, Inc.

SIRREX SeaWiFS Intercalibration Round-Robin

Experiment

SLOWDROP Slow Descendent Rate Optical Profiler

SMSR SeaWiFS Multispectral Surface

Reference

SOOP Ship of Opportunity Program

SOW Statement of Work

SPMR SeaWiFS Profiling Multi-channel

Radiometer

SQM SeaWiFS Quality Monitor

ST Science Team

TB Terabytes

TOA Top of the Atmosphere

TOMS Total Ozone Mapping Spectrometer

TOTO Tongue of the Ocean

TSM Total Suspended Matter

TSRB-II Tethered Spectral Radiometer Buoy - II

XBT Expendable Bathythermograph

REFERENCES

Artemyev, V. A., V. I. Burenkov, M. I. Vortman, A.V. Grigoryev, O. V. Kopelevich, and N. Khrapko,1999a: Sub-satellite measurements of oceancolor: new floating spectroradiometer and its me-trology. Oceanol. (in press).

Artemyev, V. A., V. I. Burenkov, S. Woznyak, A. V.Grigoryev, M. Daretsky, A. Demidov, O. V. Ko-pelevich, O. N. Frantsuzov, and A. N. Khrapko,1999b: “Sub-satellite measurements of oceancolor: the field studies in the Black and AegeanSeas”. Oceanol., (in press).

Balch, W. M., D. T. Drapeau, T. L. Cucci, and R. D.Vaillancourt, K. A. Kilpatrick, and J. J. Fritz,1999: Optical backscattering by calcifying algae-Separating the contribution by particulate inor-ganic and organic carbon fractions. J. Geophys.Res., (in press).

Barnard, A. H., W. S. Pegau, and J. R. V. Zaneveld,1998a: Global relationships of the inherent opticalproperties of the oceans, J. Geophys. Res., 103,24,955-24,968.

Bisagni, J. J., D. J. Gifford, and C. M. Ruhsam, 1995:The spatial and temporal distribution of the MaineCoastal Current during 1982. Cont. Shelf Res., 1,1-24.


100

Boyd, C. M., 1985: Is secondary production in theGulf of Maine limited by the availability of food?Arch. Hydrobiol. Beih., 21, 57-65.

Brody, E. A., 1998: An In Situ Assessment of InherentOptical Property Variability in the Sargasso Sea.M.A. Thesis, University of California Santa Bar-bara, Santa Barbara, California, 70pp.

Burenkov, V. I., O. V. Kopelevich, S. V. Sheberstov,S. V. Ershova, M. A. and A. Evdoshenko, 1999:“Bio-optical characteristics of the Aegean Sea re-trieved from satellite ocean color data.” In: TheEastern Mediterranean as a Laboratory Basin forthe Assessment of Contrasting Ecosystems, P.Malanotte-Rizzoli, Ed., (in press).

Carder, K. L., F. R. Chen, Z. P. Lee, S. K. Hawes, D.Kamykowski, 1999: Semianalytic MODIS algo-rithms for chlorophyll a and absorption with bio-optical domains based on nitrogen-depletion tem-peratures, J. Geophys. Res., (in press).

Chavez, F. P., K. R. Buck, R. R. Bidigare, D.M. Karl,D. Hebel, M. Latasa, L. Campbell, and J. Newton,1995: On the chlorophyll a retention properties ofglass-fiber GF/F filters. Limnol. Oceanogr., 40,428-433.

Chavez, F. P., P. G. Strutton, and M. J. McPhaden,1998: Biological-physical coupling in the centralequatorial Pacific during the onset of the 1997-98El Niño, Geophys. Res. Let., 25, 3,543-3,546.

Clark, D. K., H. R. Gordon, K. J. Voss, Y. Ge, W.Broenkow, and C. Trees, 1997: Validation of at-mospheric correction over the oceans, J. Geophys.Res., 102, 17,209-17,217.

Culver, M. E., A. Subramaniam, J. C. Brock, M. E.Geesey, G. R. DiTullio, and D. Jones, 1998:NOAA CSC/CRS Cruise NOV97SAR: SargassoSea Cruise. CSC Technical Report CSC/9-98/002,NOAA Coastal Services Center, Charleston, SC.

Dickey, T., 1991: The emergence of concurrent high-resolution physical and bio-optical measurementsin the upper ocean, Rev. of Geophys., 383-413.

Dickey, T., D. Frye, H. Jannasch, E. Boyle, D. Manov,D. Sigurdson, J. McNeil, M. Stramska, A.Michaels, N. Nelson, D. Siegel, G. Chang, J. Wu,and A. Knap, 1998: Initial results from the Ber-muda Testbed Mooring Program, Deep-Sea Res.,45, 771-794.

Esaias, W., G. Feldman, R. Frouin, W. Gregg, S.Hooker, and C. McClain, 1995: Ocean Color-

Multisensor Data Evaluation and Utilization Plan,Report to NASA Headquarters, 72 pp, availablevia www at URL: http://simbios.gsfc.nasa.gov

Federal Acquisition Regulation (FAR), Published byGSA’s FAR Secretariat on Behalf of DoD, GSA,and NASA available via www at URL:http://www.arnet.gov/far/index.htm.

Fernandez, E., P. Boyd, P. M. Holligan, and D. S.Harbour, 1993: Production of organic and inor-ganic carbon within a large scale coccolithophorebloom in the North Atlantic Ocean. Mar. Ecol.Prog. Ser., 97, 271-285.

Fougnie, B., 1998: Contribution a l’observation de lacouleur de l’ocean a partir du capteur spatialPOLDER. Ph.D. Thesis, No. 2342, University ofLille, Lille, France, 208 pp.

Fougnie, B., P.Y. Deschamps, and R. Frouin, 1999a:Vicarious calibration of the POLDER ocean colorspectral bands using in-situ measurements. IEEETrans. Geosci. Remote Sens., (in press).

Fougnie, B., R. Frouin, P. Lecomte, and P.-Y Des-champs, 1999b: Reduction of skylight reflectioneffects in the above-water measurement of diffusemarine reflectance. Appl. Opt., (in press).

Frouin, R., E. Pouliquen, and F.-M. Breon, 1994:Ocean color remote sensing using polarizationproperties of reflected sublight. Proc. 6th Int.Coll. On “Physical Measurements and SpectralSignatures in Remote Sensing,” Jan. 17-21, 1994,val d’Isere, France, 665-674.

Fu, G., K. Settle, and C. R. McClain, 1998: SeaDAS:The SeaWiFS Data Analysis System. Proc.Fourth Pacific Ocean Remote Sens. Conf. Qing-dao, China, 7/28/98, 73-77.

Fukushima, H., A. Higurashi, Y. Mitomi, T. Naka-jima, T. Noguchi, T. Tanala, M. Toratani, 1998:Correction of atmospheric effect onADEOS/OCTS ocean color data: algorithm de-scription and evaluation of its performance, J. ofOceanogr., 54, 417-430.

Garver, S. A., and D. A. Siegel, 1997: Inherent opticalproperty inversion of ocean color spectra and itsbiogeochemical interpretation . 1. Time seriesfrom the Sargasso Sea. J. Geophys. Res.–Oceans,102, 18,607-18,625.

Goldin, Y. A., B. F. Kelbalikhanov, and V. N.Pelevin, 1983: Measurements of the light field pa-


101

rameters in the ocean. Ocean Opt., Vol. 1, A. S.Monin Ed., Nauka, Moscow, 236-248.

Gordon, H. R., 1996: MODIS normalized water-leaving radiance algorithm theoretical basisdocument (MOD 18). Version 3, 110 pp atftp://eospso.gsfc.nasa.gov/ATBD/REVIEW/MODIS/ATBD-MOD-17/atbd-mod-17.pdf.

Gordon, H. R., and D. K. Clark, 1981: Clear waterradiances for atmospheric correction of coastalzone color scanner imagery, Appl. Opt., 20,4,175-4,180.

Gordon, H.R., and M. Wang, 1994a: Influence of oce-anic whitecaps on atmospheric correction ofocean-color Sensor, Appl. Opt., 33, 7,754-7,763.

Gordon, H. R., and M. Wang, 1994b: Retrieval ofwater-leaving radiance and aerosol optical thick-ness over the oceans with SeaWiFS: A prelimi-nary algorithm, Appl. Opt., 33, 443-452.

Gordon, H. R., D. K. Clark, J. L. Mueller, and W. A.Hovis, 1980: Phytoplankton pigments derivedfrom the Nimbus-7 CZCS: initial comparisonswith surface measurements. Science, 210, 63-66.

Gregg, W. W., 1998: Initial assessment of ocean colordata from the OCTS mission. EOS, Trans. Am.Geophys. U. 79, 28.

Gregg, W. W., 1999a: Considerations for future oceancolor sensors. Proc. 8th Japan/US WorkingGroup on Ocean Color (JUWOC), J. Ishizaka andJ. Campbell, Eds., Kona, Hawaii, Nov. 1998.

Gregg, W. W., 1999b: Initial analysis of ocean colordata from OCTS. I. Imagery analysis. Appl. Opt.,38, 3,476-485.

Gregg, W. W., and F. S. Patt, 1994: Assessment of tiltcapability for spaceborne global ocean color sen-sors. IEEE Trans. Geosci. Remote Sens., 32, 866-877.

Gregg, W. W., and M. E. Conkright, 1999: Initial re-sults from a global blended ocean chlorophyllanalysis. Proc. 79th AMS Annual Meeting, 10thSymposium on Global Change Studies. AmericanMeteorological Society, Boston, Massachusetts.

Gregg, W. W., and R. H. Woodward, 1998: Improve-ments in high frequency ocean color observa-tions: Combining data from SeaWiFS andMODIS. IEEE Trans. Geosci. Remote Sens., 36,1, 350-1,353.

Gregg, W. W., F. S. Patt, and W. E. Esaias, 1999:Initial analysis of ocean color data from OCTS.II. Geometric and radiometric analysis. Appl.Opt., (submitted).

Gregg, W. W., W. E. Esaias, G. C. Feldman, R.Frouin, S. B. Hooker, C. R. McClain, and R. H.Woodward, 1998: Coverage opportunities forglobal ocean color in a multi-mission era. IEEETrans. Geosci. Remote Sens., 36, 1,620-1,627.

Henyey, L.C., and J.L. Greenstein, 1941: Diffuse ra-diation in the galaxy. J. Astrophys., 93, 70-83.

Holben B. N., T. F. Eck, I. Slutsker, D.Tanre, J. P.Buis, A. Setzer, E. Vermote, J. A. Reagan, Y. J.Kaufman, T. Nakajima, F. Lavenu, I. Jankowiak,A. Smirnov, 1998: AERONET - A Federated In-strument Network and Data Archive for AerosolCharacterization. - Remote Sens. Environ., 66, 1-16.

IOCCG 1998: Minimum Requirements for an Opera-tional Ocean-Colour Sensor for the Open Ocean,IOCCG Report No. 1, IOCCG Project Office,Bedford, Nova Scotia, 46pp.

Iqbal, M., 1983: An Introduction to Solar Radiation,Academic Press, San Diego, California, 390 pp.

Ivanov, A., 1975: Introduction a l’Oceanographie:Tome II. Vuibert, Paris, 208 pp.

JGOFS (Joint Global Ocean Flux Study), 1991:JGOFS Core Measurement Protocols, JGOFSReport No. 6, Scientific Committee on OceanicResearch, 40pp.

JGOFS, 1996: Protocols for the Joint Global OceanFlux Study (JGOFS) Core Measurements. ReportNo. 19, Joint Global Ocean Flux Study, ScientificCommittee on Oceanic Research, InternationalCouncil of Scientific Unions, IntergovernmentalOceanographic Commission, Bergen, Norway,170 pp.

Jin, Z., and K. Stamnes, 1994: Radiative transfer innonuniformly refracting layered media: atmos-phere-ocean system. Appl. Opt., 33, 431-442.

Johnson B. C., J.B. Fowler, and C. L. Cromer, 1998:Volume 1, The SeaWiFS Transfer Radiometer(SXR), NASA Tech. Memo. 1998-206892, S. B.Hooker and E.R. Firestone, Eds., NASA GoddardSpace Flight Center, Greenblet, Maryland, 58 pp.


102

Kahru, M., and B. G. Mitchell, 1999: Empirical chlo-rophyll algoirthm for the California Current, Int.J. Remote Sens., (in press).

Kasten, F., and A. T. Young, 1989: Revised optical airmass tables an approximation formula, Appl.Opt., 28, 4,735-4738.

Kawamura, H., 1998: OCTS mission overview, J. ofOceanogr., 54, 383-399.

Kishino, M., N. Okami, and S. Ichimura, 1985: Esti-mation of the spectral absorption coefficients ofphytoplankton in the sea. Bull. Mar. Sci., 37, 634-642.

Kishino, M., T. Ishimaru, K. Furuya, T., Oishi, K.,Kawasaki, 1998: In-water algorithms forADEOS/OCTS, J. of Oceanogr., 54, 431-436.

Kopelevich O.V., D. B. Rogozkin, and S. V. Sheber-stov, 1998: Results of application of the new at-mospheric correction method to the Pacific Oceancolor simulated data set. Proc. PORSEC’98, Vol.1, M-X. He and G. Chen, Eds., Secretariat, Qing-dao, China, 54-58.

London, J., R. D. Bojkov, S. Oltmans, and J. I. Kelley,1976: Atlas of the global distribution of totalozone July 1957 –June 1967, NCAR TechnicalNote 133+STR, National Center for AtmosphericResearch, Boulder, Colorado, 276 pp.

Long, C. N., and T. P. Ackerman, 1999: Identificationof clear skies from broadband pyranometer meas-urements and calculation of downwelling short-wave cloud effects. J. Geophys. Res., (in press).

Miller, R. L., K. Davis, O. von Zweck, V. G. Sprague,Jr., and J. Sommers, 1995: Shipboard data acqui-sition, processing and analysis: An integratedsoftware approach. Proc. MTS/IEEE Oceans ‘95,San Diego, California, 1,942-1,947.

Mitchell, B. G., 1990: Algorithms for determining theabsorption coefficients of aquatic particulates us-ing the quantitative filter technique (QFT). SPIEOcean Opt. X, 1,302, 137-148.

Mitchell, B. G., and D. A. Kiefer, 1988: Variability inpigment specific particulate fluorescence and ab-sorption spectra in the northeastern Pacific Ocean.Deep-Sea Res. I, 35, 665-689.

Mitchell, B. G., and M. Kahru, 1998: Algorithms forSeaWiFS standard products developed with theCalCOFI bio-optical data set. Cal. Coop. OceanFish. Invest. Res., 39, 133-147.

Mitchell, B. G., and O. Holm-Hansen, 1991: Bio-optical properties of Antarctic Peninsula waters:Differentiation from temperate ocean models.Deep-Sea Res. I, 38, 1,009-1,028.

Morel, A. and B. Gentili, 1991: Diffuse reflectance ofoceanic waters: its dependence on Sun angle asinfluenced by the molecular-scattering contribu-tion. Appl. Opt., 30, 4,427-4,438.

Morel, A., 1988: Optical modeling of the upper oceanin relation to its biogenous matter content (Case Iwaters), J. Geophys. Res., 93, 10,749-10,768.

Mueller, J. L., and R. W. Austin, 1992: Ocean OpticsProtocols for SeaWiFS Validation. NASA Tech.Memo. 104566, Vol. 5, S.B. Hooker and E.R.Firestone, Eds., NASA Goddard Space FlightCenter, Greenbelt, Maryland, 43 pp.

Mueller, J. L., and R. W. Austin, 1995: Ocean OpticsProtocols for Validation, Revision 1, NASA Tech.Memo. 104566, Vol 25., S.B. Hooker, E. R.Firestone, and J. G. Acker, Eds., NASA GoddardSpace Flight Center, Greenbelt, Maryland, 67 pp.

Nakajima, T., A. Higurashi, K. Aoki, T. Endoh, F. H.Fukushima, M. Toratani, Y. Mitomi, B. G.Mitchell, and R. Frouin, 1999: Early phase analy-sis of OCTS radiance data for aerosol remotesensing. IEEE Trans. Geosci. Remote Sens., (inpress).

NASDA, 1994: ADEOS ground station OCTS Level 0format specification: Version 1.3, NASDA Tech.Memo., 127 pp.

National Aeronautics and Space Administration(NASA), 1996: Sensor Intercomparison andMerger for Biological and Interdisciplinary OceanStudies, (SIMBIOS), July 15, 1996, NASA Re-search Announcement (NRA) 96-MTPE-04, Na-tional Aeronautics and Space Administration,Washington, DC, 31 pp.

Neckel, H. and D. Labs, 1984: The solar radiationbetween 3,300 and 12,500 A. Solar Phys., 90,205-258.

Nelson, N. B., D. A. Siegel, and A. F. Michaels, 1998:Seasonal dynamics of colored dissolved materialin the Sargasso Sea, Deep-Sea Res. I, 45, 931-957.

O’Brien, M. C., D. W. Menzies, D. A. Siegel, and R.C. Smith, 1999: Long-term calibration history ofseveral multispectral environmental radiometers(MER), NASA Tech. Memo. 1999-206892, S. B.


103

Hooker and E. R. Firestone, Eds., NASA God-dard Space Flight Center, Greenbelt, Maryland,(in press).

O’Reilly, J. E., C. Evans-Zetlin, and D. A. Busch,1987: Primary production. In: Georges Bank, R.H. Backus and D. W. Bourne, Eds,, Massachu-setts Institute of Technology, Cambridge, Massa-chusetts, 220-233.

O’Reilly, J. E., S. Maritorena, B. G. Mitchell, D. A.Siegel, K. L. Carder, S. A. Garver, and C. R.McClain, 1998: Ocean color chlorophyll algo-rithms for SeaWiFS. J. Geophys. Res., 103,24,937-24, 953.

Patt, F. S. and W.W. Gregg, 1994: Exact closed-formgeolocation algorithm for earth survey sensors,Int. J. Remote Sens., Vol.15, No. 18, 3,719-3,734.

Patt, F. S., R.H. Woodward, W.W. Gregg, 1997: Anautomated method for navigation assessment forearth survey sensors using island targets, Int. J.Remote Sens., 18 (16), 3,311-3,336.

Penndorf, R., 1957: Tables of the refractive index forstandard air and the Rayleigh scattering coeffi-cient for the spectral region between 0.2 and 20.0microns and their application to atmospheric op-tics, J. Opt. Soc. Am., 47, 176-182.

Porra, R. J., W. A. Thompson, and P. E. Kriedemann,1989: Determination of accurate coefficients andsimultaneous equations for assaying chlorophyllsa and b extracted with four different solvents:verification of the concentration of chlorophyllstandards by atomic absorption spectroscopy.Biochem. Biophys. Acta, 975, 384-394.

Reinersman, P. R., K. L. Carder, and F. R. Chen,1998: Satellite-sensor calibration verification withthe cloud-shadow method. Appl. Opt., 37, 5,541-5,549.

Reynolds, R. W., 1988: A real-time global sea surfacetemperature analysis. J. Clim. 1, 75-86.

Riley, T., and S. Bailey 1998: The SixthSeaWiFS/SIMBIOS Intercalibration Round-Robin Experiment (SIRREX-6) August-December 1997, NASA Tech. Memo. 1998-206878, Eds., NASA Goddard Space Flight Cen-ter, Goddard, Maryland, 26pp.

Sathyendranath, S., G. F. Cota, V. Stuart, H. Maass,and T. Platt, 1999a: Remote sensing of phyto-plankton pigments: A comparison of empirical

and theoretical approaches. Intl. J. Remote Sens.,(submitted).

Sathyendranath, S., T. Platt, V. Stuart, H. Maass, andG.F. Cota, 1999b: Remote sensing of phyto-plankton biomass and primary production. Aqua.Biol. (in press).

Schwindling M., P. Deschamps, and R. Frouin, 1998:Verification of aerosol models for satellite oceancolor remote sensing. J. Geophys. Res., 103,24,919 - 24,935.

Sharp, J. H., 1991: Total mass and particulate car-bon, nitrogen and phosphorus. In: Marine Parti-cles: Analysis and Characterization, D.C. Hurdand D.W. Spencer, Eds., American GeophysicalUnion, Washington, DC, 87-90.

Sharp, J. H., 1974: Improved analysis for particulateorganic carbon and nitrogen from sea water.Limnol. Oceanogr. 19, 984-989.

Shettle, E. P., and R. W. Fenn, 1979: Models for theAerosols of Lower Atmosphere and the Effects ofHumidity Variations on their Optical Properties.AFGL-TR-79-0214. Air Force Geophysics Labo-ratory, Hanscomb AFB, Massachusetts.

Shimada, M., 1998a: A cal/val report on the OCTSversion 3 products, NASDA Tech. Memo., 20pp.

Shimada, M., 1998b: Calibration and validation of theocean color version-3 products fromADEOS/OCTS. J. of Oceanogr., 54, 401-416.

Siegel, D. A., A. F. Michaels, J. C. Sorensen, M.O’Brien, and M. A. Hammer, 1995a, Seasonalvariability of light availability and utilization inthe Sargasso Sea, J. Geophys. Res., 100, 8,675-8,713.

Siegel, D. A., and A. F. Michaels, 1996: Quantifica-tion of non-algal light attenuation in the SargassoSea: implications for biogeochemistry and remotesensing, Deep-Sea Res. II, 43, 321-345.

Siegel, D. A., M. C. O’Brien, J. C. Sorensen, D. A.Konnoff, and E. Fields, 1995b: BBOP DataProcessing and Sampling Procedures. Vol: 19,Institute for Computational Earth System Science,University of California Santa Barbara, SantaBarbara, California, 23 pp.

Siegel, D. A., M. C. O’Brien, J. C. Sorensen, D. A.Konoff, E. A. Brody, J. L. Mueller, C. O. Davis,W. J. Rhea, and S. B. Hooker, 1995c: Results ofthe SeaWiFS Data Analysis Round-Robin, July


104

1994 (DARR-94), NASA Tech. Memo. 1994-104566, Vol. 26, S. B. Hooker and E.R. Firestone,Eds., NASA Goddard Space Flight Center,Greenbelt, Maryland, 58 pp.

Siegel, D. A., M. C. O’Brien, J. Sorensen, D. Konnoff,and E. Fields, 1995b: BBOP Sampling and DataProcessing Protocols. U.S. JGOFS Planning Re-port Number 19, U.S. JGOFS Planning Office,Woods Hole, Massachusetts, 77 pp.

Smith R.C., C.R. Booth, and J.L. Starr, 1984:Oceanographic bio-optical profiling system, Appl.Opt., 23, 2,791-2,797.

Smith, R. C., and K. S. Baker, 1978: The bio-opticalstate of ocean waters and remote sensing. Limnol.Oceanogr., 23, 247-259.

Smith, R. C., K. J. Waters, and K. S. Baker, 1991:Optical variability and pigment biomass in theSargasso Sea as determined using deep-sea opti-cal mooring data, J. Geophys. Res., 96, 8,665-8,686.

Sosik, H. M., and B. G. Mitchell, 1995: Light absorp-tion by phytoplankton, photosynthetic pigments,and detritus in the California Current System.Deep-Sea Res. I, 42, 1,717-1,748.

Strickand, J. D. H., and T. R. Parsons, 1972: A Practi-cal Handbook of Sea Water Analysis. 2nd edition,Fish. Res. Board of Canada, 167, 310 pp.

Subramaniam, A., E. J. Carpenter, and P. G. Falkow-ski, 1999: Optical properties of the marine diazo-trophic cyanobacteria Trichodesmium spp.; II – Areflectance model for remote-sensing. Limnol.Oceanogr., (in press).

Subramaniam, A., E. M. Armstrong, K. J. Waters, J.C. Brock, P. A. Tester, and E. Haugen, 1997a:NOAA CSC/CRS Cruise MAR97OCC: OCTSCalibration Cruise. CSC Tech. Report CSC/5-97/001, NOAA Coastal Services Center,Charleston, South Carolina, 18 pp.

Subramaniam, A., E. M. Armstrong, K. J. Waters, J.C. Brock, A. W. Meredith, and R. O. Ranheim,1997b: NOAA CSC/CRS Cruise MAY96NY:New Bight Apex Cruise. CSC Tech. ReportCSC/6-97/001, NOAA Coastal Services Center,Charleston South Carolina, 33 pp.

Subramaniam, A., K. J. Waters, A. W. Meredith, E.M. Armstrong, R. M. Bohne, W. G. Keull, J. R.Nelson, G. R. DiTullio, and J. C. Brock, 1997c:NOAA CSC/CRS Cruise APR96FER: Gray’s

Reef Cruise. CSC Tech. Report CSC/7-97/001.NOAA Coastal Services Center, Charleston,South Carolina, 33 pp.

Subramaniam, A., J. C. Brock, P. A. Tester, E.Haugen, and R. P. Stumpf, 1998: NOAA NMFSCruise MAY97OB: Onslow Bay and PamlicoSound Cruise. CSC Tech. Report CSC/2-98/001,NOAA Coastal Services Center, Charleston,South Carolina, 54 pp.

Suzuki, R. and Y. Fujita, 1986: Chlorophyll decompo-sition in Skeletonema costatum: a problem inchlorophyll determination of water samples. Mar.Ecol. Prog. Ser., 28, 81-85.

Suzuki, R., and T. Ishimaru, 1990: An improvedmethod for the determination of phytoplanktionchlorophyll using N, N-Dimethylformamide. J.Oceanogr. Soc. Japan, 46, 190-194.

Utermöhl, H., 1931: Neue Wege in der quantitativenErfassung des Planktons (mit besonderer Beruck-sichtigung des Ultraplanktons). Verh. int. Ver.theor. angew. Limnol., 5, 567-596.

Utermöhl, H., 1958: Zur Vervollkommnung derquantitaiven Phytoplankton-Methodik. Mitt. int.Ver. theor. angew. Limnol., 9,1-38.

Villevalde, Y. V., A. V. Smirnov, N. T. O’Neill, S. P.Smyshlyaev, and V. V. Yakovlev, 1994: Meas-urement of aerosol optical depth in the PacificOcean and North Atlantic. J. Geophys. Res., 99,20,983-20,988.

Volten, H., J. F. de Haan, J. W. Hovenier, R. Schreurs,W. Vassen, A. G. Dekker, H. J. Hoogenboom, F.Charlton, and R. Wouts, 1998: Laboratory meas-urements of angular distributions of light scat-tered by phytoplankton and silt. Limnol.Oceanogr., 43, 1,180-1,197.

Wang, M. and B. A. Franz, 1998: A vicarious inter-calibration between MOS and SeaWiFS, Proc. of2nd Int. Workshop on MOS-IRS and Ocean Col-our, Berlin, June 10-12, 95-102.

Wang, M. and B. A. Franz, 1999: Comparing theocean color measurements between MOS andSeaWiFS: a vicarious intercalibration approachfor MOS, IEEE Trans. Geosci. Remote Sens., (ac-cepted).

Wang, M., 1999: A sensitivity study of the SeaWiFSatmospheric correction algorithm: effects ofspectral band variations, Remote Sens. Environ.,67, 348-359.


105

Wang, M., S. Bailey, C. R. McClain, C. Pietras, T.Riley, 1999: Remote Sensing of the Aerosol Op-tical Thickness from SeaWiFS in Comparisonwith the In Situ Measurements, ALPS99 Work-shop, Méribel, France.

Woodward, R. H., and W. W. Gregg, 1998: An as-sessment of SeaWiFS and MODIS ocean cover-age. NASA Tech. Memo. 1998-208607, NASAGoddard Space Flight Center, Greenbelt, Mary-land, 45 pp.

Wright, S. W., S. W. Jeffrey, R. F. C. Mantoura, C. A.Llewellyn, T. Bjornland, D. Repeta, and N.Welschmeyer, 1991: Improved HPLC method for

the analysis of chlorophylls and carotenoids frommarine phytoplankton. Mar. Ecol. Prog. Ser., 77,183-196.

Yentsch, C.S., and D.W. Menzel, 1963: A method fordetermination of phytoplankton chlorophyll andphaeophytin by fluorescence. Deep-Sea Res., 10,221-231.

Zimmermann, G., and A. Neumann, 1997: The space-borne imaging spectrometer MOS for ocean re-mote sensing, Proc. of the 1st Int. Workshop onMOS-IRS and Ocean Colour, Berlin, April 28-30,1-9.