TOO vaivlabld UDir Fa e n#%s*In the interest of early and wide dis-semination of Earth Resources SurveyProgram information and without liabilityfor any use made ,thereof."

E7.3 10 7.2 56 f / 5 30 &2,

SR. 051

F 433

A. R. N. I. C. A.

MANAGEMENT- OF NATURAL RESOURCESTHROUGHA1 lTnMATTIC CARTOGRAPHIC INVENTORY

(E73-10725) MANAGEMENT OF NATURALRESOURCES THROUGH AUTOMATIC CARTOGRAPHICINVENTORY Progress Report, Aug. 1972 -Jan. 1973 (Service de la Carte de laVegetation CNRS 80-Lp HC $6.00 CSCL 08B

' rincipai inveStsgaoL;Paul-Augustin REYDirectorService de la Carte de la Vegetation CNRS31055 TOULOUSE CEDEX France

Co-investigators:Yves GOURINARDFrancis CAMBOU

May 1973

Sponsoring AgencyCENTRE NATIONAL D'ETUDES SPATIALES129 rue de I'Universite 75007 PARIS

051 ARNICA II -1

Reproduced by

NATIONAL TECHNICALINFORMATION SERVICE

US Department of CommerceSpringfield, VA. 22151

Original photography may be purchased from3ER(OS eaga 'Center10th arnd Dakola AvenueSioux Falls, SD 57198

? TF

N73-27234

UnclasG3/13 00725

TYPE II PROGRESS REPORT

for Period August 1972 - January 1973

N o 1

Il.j

1 . 2 3.SR 051 TYPE II Report

4. A R. N C. A 5. Report DateMay 1973

MANAGEMENT OF NATURALRESOURCES THROUGH AUTOMATIC 6. Period covered:CARTOGRAPHIC INVENTORY Aug. 72 - Jan. 73

7. Principal investigator: 8. N' of Pages:

Paul-Augustin REY

9. Name and Address of Principal investigator 10. Principal investigatorOrganization Report : N° 2

Service de la Carte de la Vegetation CNRS29 rue Jeanne Marvig TOULOUSE FranceBP. 4009 31055 TOULOUSE CEDEX George ENSOR

12. Sponsoring Agency Name and Address 13. Key Words :- Natural ResourcesCENTRE NATIONAL d'ETUDES SPATIALES - Botany - Geology

129 rue de l'Universit -Botany - Geology75007 PARIS France - Management

- Inventory

14. Supplementary Notes :

Report presented with cooperation of Y. GOURINARD, F. CAMBOU,C. GUYADER, H. PICCA, G. FLOUZAT, P. GOUAUX, M. MONCHANT.

15. Abstract :

Significant results of the ARNICA program (August 1972 - January 1973)

· Establishment of image to object correspondence codes for all typesof soil use and forestry in northern Spain.

· Establishment of a transfer procedure between qualitative (remoteidentification and remote interpretation) and quantitative (numeriza-tion, storage, automatic statistical cartography) use of images.

. Organization of microdensitometric data processing and automaticcartography software.

· Development of a system for measuring reflectance simultaneous withimagery.

L

P

I

I

i

I

CONTENTS

PART I : Preface ............

- Objective .................

- Scope of activity ..........

- Significant results ........

- Conclusions, requirements ...

PART II: Results

- Chapter 1. Quz

- Chapter 2. Qu,qua

- Chapter 3. Quz

- Chapter 4. Exi

List of Illustrations

........*0*O*.

alitative studies

alitative andantitative studies

antitative studies

periments .....

; I e e X X · · X · · · · ·

ARNICA II- 1

1

2

3

5

9

10

11

27

56

68

77

051

1

PART I

PREFACE

- OBJECTIVES

- SCOPE OF ACTIVITY

- SIGNIFICANT RESULTS

- CONCLUSIONS

ARNICA II- 1051

2

OBJECTIVES

ERTS A MSS Imagery will be correlated withexixting vegetation and geological maps of sou-thern FRANCE and northern SPAIN, to developcorrespondence codes between maps units andsatellite data.

These codes will then be used in numerical formfor eventual automatic charting of natural resources.

Microclimate data from certain experimentalground stations, spectral measurements from afew meters to 2 km. using ERTS - type filtersand spectrometers, and leaf reflectance measurements will be obtained to assist in correla-tion studies.

ARNICA II- 1051

3

SCOPE OF ACTIVITY

A major part of the test-site selected for theARNICA program was clouded over in 1972. Thedocuments received for the August to Decemberperiod therefore concern essentially only mar-ginal parts of the site or parts outside.

Furthermore, except for the eastern section(Rhone delta), no repetitive cover could beassured.

The activity of the ARNICA studies group has therefore been limited to specific themes, but undergeographical conditions different from those initially defined. Certain ground activities, per-formed in November and December, could not beused as planned for correlations simultaneouswith E RTS imagery, because this imagery has notyet been received.

With these organizational problems, data werenot processed from September 1972 to Juanuary1973 in the directions explained in the TYPE IPROGRESS REPORT (Nov. 1972), in accordan-ce with the attached FLOW CHART.

See Attachment A

The present report illustrates the results acqui-red at four levels of investigation:

- Qualitative studies- Qualitative and Quantitative Studies- Quantitative studies- Experiments

ARNICA II - 1051

!ARNICA PROGRAM

FLOW CHART - Qualitative phase'

Quantitative phase- _ art

" CARTOZONINC

I Reportingt___J

.1 search for object to imagecorrespondence

ape

REMOTE IDENTIFICATION

.'. ,ma _ sonlie·t :· ,t · t * · · * '

Synthetic comparison

L' of object

and image sets with equal

' distribution

REMOTE INTERPRETATION-- - E-E -- r

Zones with equal

values of optical

density

distributionI~-

" PHOTOZONING "

.~~~~~~~

Printout of statistical values

AUTOMATIC CARTOGRAPHY

PI

C..R.S.. 5CV/C. . .S./6.7 7

L4

Analytical comparison

anr

Zones with equa

values of landsc

and

soil occupation

Computer search for image

to object

correspondence codes

and

the most significant ground

reality spectral signatures

Exploration of

i, (4 MSS images) withmicrodensitometer

checked against zonings

PHOTODENSITOME TRY

and NUME RIZATION,,7"7-r . -s

Selection of image V

to object correlation

graphs based on MSS

omm����

Verificationl

I

I

5

SIGNIFICANT RESULTS

A. QUALITATIVE STUDIES

see Part II, Chapter 1

Data: 1027-10141 MSS 4, 5, 6, 7

The excellent cover acquired by ERTS over nor-theastern Spain permitted a detailed study of thesearch for correspondences between ground usetypes and 4-band multispectral scanner imagery.

The available Cartographic Information- Forest Map of Spain, scale 1/400.000- Soil Map of the Lusitano-Iberian Penin-

sula, scale 1/500.000provided the basis for equal cartographical zo-ning which could be compared with photographiczonings established for each image.

Correspondence codes were defined as follows

1. Major landscape and soil typesCrops, Irrigated cropsForests, Arid brushlandsCalcareous soils

2. Most frequently found forestry typesQuercus lusitanicaPinus halepensisPinus silvestris

In this way the value of making basic comparisonsbetween multispectral scanner bands 5 and 7 inorder to establish correspondence codes, for phy-togeographical and geological studies, has beenconfirmed ; multispectral scanner band 6 and toa lesser. extent multispectral scanner band 4 adduseful supplementary details for finer identifica-tions .

ARNICA II- 1051

6

SIGNIFICANT RESULTS (continued)

B. QUALITATIVE and QUANTITATIVES TUDIE S

see Pa:-t II, Chapter 2

Data : 1027-10135 MSS 4, 5, 6, 7

The principal effort of the remote sensing depart-ment has developed at the " interface " level bet-ween qualitative reading operations, identificationand interpretation of data, and quantitative auto-matic information processing operations.

The work was carried out over a part of the Spa-nish Pyrenees (Aran Valley) where both very ac-curate vegetation mapping and excellent multispec-tral scanner band 4, 5, 6 and 7 images were avai-lable.

The results obtained involve the following proces-ses :

· Preprocessing of the vegetation map : Use of" Cartozoning " by remote identification offorestry known on the map and charting of" Zones with equal values of forestry".

· Preprocessing of ERTS 1 Imagery: Use of" Photozoning " by remote interpretation of" Zones with equal values of densitometricdistribution ".

· Search for densitometric correspondence codesbetween ground-reality and ERTS imagery,checked against the above equal value zones.

· Establishment of graphs of the most significantcorrespondences.

· Application to automatic statistical cartographyInventory of FIR FORESTS.

ARNICA II- 1

7

SIGNIFICANT RESULTS (continued)

C. QUANTITATIVE STUDIES

see Part II, Chapter 3

Data : 1027-10135 MSS 4, 5, 6, 7

In the first phase, the Centre d'Etudes Spatialesdes Rayonnements (C .E .S.R.) is primarily involved in defining conditions of information transferbetween qualitative visualization of the image toobject correspondence codes and automatic carto-graphic output.

The group has advanced through the following sta-ges :

- Calibration of the densitometry apparatus

- Development of available software

The examples of automatic output in visual simula-tion provide some interesting preliminary results,which can be refined as soon as two problems havebeen solved :

- data compression

- improved accuracy of geographical refe-rencing of parts of images to be numeri -zed.

Also, all of the operations will obviously be mucheasier as soon as the original digital bands becomeavailable.

ARNICA II- 1051

8

SIGNIFICANT RESULTS (end)

D. EXPERIMENTS

Two groups of experiments had been planned inassociation with the ERTS/1-ARNICA program:

- aircraft experiments

- ground experiments

reflectance and ecological measurements madesimultaneous with the imagery.

In view of the recent recording problems in thesatellite, from now on the studies will be conduc-ted in such a way as to make experimentation pos-sible in the framework of the next ERTS B pro-gram in the event that it cannot be redone forERTS A.

The present report discusses the technologicaladvances made in the air-based part of the expe-riment (development of an aircraft-borne system)The ground-based measurement system will bediscussed in a future report.

ARNICA II-1051

9

CONCLUSIONS

The work presented in this report illustrates thefundamental phases of the ARNICA program.

The method developed for the most significantimagery of the test-site, available December 31,1972, will be extended to other as a function ofthe date which can actually be used, especiallyin regions where repetitive cover can be made.

REQUIREMENTS

In order to extract the best part of the acquiredinformation, the following documents are neces-sary :

a) color composite images of the data1027-101351027-101411028-101931028-102501064-101901066-10294

b) original digital bands1027-101351027-10141

of the data1028-101931064-10190

ARNICA II- 1051

10

PART II

RESULTS

Chapter 1. QUALITATIVE STUDIES

- STUDY OF LAND USE PATTERN INNORTHERN SPAIN

1. Qualitative processing :Interpretation of ERTS 1 Imagery

by Mme P. GOUAUX, Department ofTeledetection, Universit6 Paul Sabatier

Chapter 2. QUALITATIVE and QUANTITATIVESTUDIES

- REMOTE SENSING OF FORESTRY(USING ERTS 1 DATA)

1. Identification

by G. FLOUZAT, Department ofTeledetection, Universite Paul Sabatier

Chapter 3. QUANTITATIVE STUDIES

- DIGITAL PROCEDURES ON REMOTESENSING IMAGERY FOR THE ARNICAERTS 1 PROGRAM

by M. MONCHANT, Centre d'EtudesSpatiales des Rayonnements

Chapter 4. EXPERIMENTS

- STUDY OF A RADIOMETER FOR VI-SIBLE WAVELENGTHS

by Laboratoire de Physique de l'Atmos-ph&re, under the direction of ProfessorPICCA, Universit6 Paul Sabatier

ARNICA II- 1051

(1)

Chapter 1.

STUDY OF LAND USE PATTERNS

IN NORTHERN SPAIN

1 - Qualitative processing :

Interpretation of ERTS 1 Imagery

P. GOUAUX

Universite Paul-Sabatier

TOULOUSE

ARNICA II-1

11

051

(2) 12

This study is included in ARNICA Project.

The ERTS 1 data available at first, inNovember-December 1972, were obtained fromcountries that could be defined in reference withthe presence of absence of the Vegetation Map.

The processed ERTS 1 data in this study areissued from countries no covered by theVegetation Map. The absence of the Vegetation Mapmisses the analysis of the preliminary workperformed for this map. So, it is not possible tofind directly the determination keys of soilunits. Their determination will be based ondocuments and cards available in the studiedcountry.

This study concerns Multispectral Scanner1027-10141 data which covers the Ebro basin fromZaragoza to the mediterranean.

I. REFERENCE MAPPING

- Forest Map of Spain-Scale 1 : 400,000

- Soil Map of the Lusitano-IberianPeninsula-Scale 1 : 1,500,000

1. The Forest Map of Spain published by theSpanish Department of Agriculture in 1966, showsthree principal features :

- Crops.

- Lightly covered soils, e.g., steppes,moorlands, grasslands, natural pasture, wasteland.

- Forests with tree species indicated.

2. Soil Map of the Lusitano-Iberian Peninsulaby H. Del Villar (1938), (Scale 1 : 500,000).Different soils are indicated in the test site :

- Alluvial soils : along the wide valleys,about 10 km. wide along the Ebro and wider alongthe Segre.

ARNICA II-1051

(3) 13

- Area of frequency of saline patches inlandfrom the Ebro.

- Saline soils : the Ebro basin is unique inthat its saline soils are of tertiary origincreating special land formations, e.g., desertsteppes ; some of this soil has been developedthrough irrigation.

- Dry and intermediate calcareous soils :covering a large part of the Ebro basin, becomingskeleton-like, especially in mountain areas.

- General occurence of xerosiallitic soils,found only in the mountanous zones.

In addition to maps, various notes have been used(see Bibliography).

II. QUALITATIVE ANALYSIS

A. Method

The same principles used for small scale dataapply to large-scale data. The data records thevarious textural areas in varying shades of grayin the different spectral bands. These texturalareas represent soil occupation units which areidentified by cartographic correlation.

B. Analysis

1) Amplitude of optical density

The four Multispectral scanner data arecharacterized by the amplitude of optical density,thus indirectly indicating the fluctuationscaused by photographic printing conditions.

2) Texture

Textural areas are delimited by analysis of theoptical density characterizing each particulararea.

Each of these textural areas groups severalcomponents ; these one being soil occupationstates.

ARNICA II-1051

(4)

AR

NIC

A

II-1

14

051

(5) 15

the boundaries of an optical level changeaccording to the spectral bands. In effect theresponses recorded for various formations (e.g.,plant, pedological, geological, hydrographical)correspond to the reflectance in each spectralband. By checking against other data, all of thesoil occupation states are known and isolated :they are thus characterized by the combination oftheir four spectral responses.

Table I shows the optical amplitude (A) andhomogeneous texture areas for each spectral band,interpreted and placed in the gray scale. Spectralband analysis separates soil occupation states by"natural selection", since this method followsfrom investigation of the various textural areas.

Thus, in the example below (cf. Table II) theknowledge of the soil occupation state F1,isolated in the Multispectral scanner 5 data,implies those of F2 (MSS 6 data) and Z1(MSS 6 data), even if the last two states are notisolated in any band.

Textural group

band Texture Number ofcomponent Componentscomponents

4 Medium 1 2 3Bgray1 2 3~~gray ~F -F2 -F

Z 1

5 Dark 1 F1gray

6 Medium 2 F-FF1-F2

gray

7 Light 3 F1F2gray- Medium 1gray

TABLE II051 ARNICA II-1

(6) 16

3) Structure

The textural character is not sufficent todetermine all the feature groups. The structuralcharacter improves the data analysis. It definesthe textural character by indicating the arrange-ment of forms which make up the textural area.It is most useful in band 5. For example,differences in structure allow one to separate 2forms of soil occupation C1 and C2 in ahomogeneous texture area C1 C2 (cf. Table III).

TABLE III

III. INTERPRETATION

By means of interpretation of the different textu-ral and structural groups using maps (I), dataanalysis (II. 1 and 2) is possible.

ARNICA II-1

Spectral Texture Structure Distinctionband

4 Medium C1C2 Same Impossiblegray homoge-

neity

5 Dark. C1C2

Averagegray- homoge- CMedium neitygray Very

homoge-neouswithin C2thelightestgrid

6 Light C 1C2

Same Impossible7 gray homoge-

neity

051

(7) 17

T:he figure, "Document ERTS 1 Interpretation",shows an interpretation of MSS 5 data.10 feature groups are distinguished.

The phytogeographic profile, between A and A' innext figure, was made from the Forest and Soilmaps.

The purpose of the profile is to locate all theformations in space and thus establish a directparallel between the vegetal group and theecological parameters such as climate and soil.There is more botanical data from forest mapsthan from interpretation of multispectral scannerdata : thus lightly covered soils and non-irriga-ted crops are not distinguishable by interpreta-tion of ERTS 1 data, whereas the arborescentformations and the irrigated crops are very clear.The pedological date, from the Soil Map, completesthe profile and clarifies soil-plant relations.

A) Crops

The place occupied by these feature groupscorresponds to the crop zone on the forest map.The soil map indicates alluvial soils and soilsaltered by irrigation, saline soils, andirrigated soils. The last two soil types areplanned for cultivation. Different crops can-bedistinguished.

1. Irrigated crops in alluvial soils

la. Crops irrigated from main river bedslb. Crops irrigated from alluvial terraces

2. Crops irrigated from saline soils

Salt serves as a limiting factor in conditioninga particular landscape characterized byuncultivated soils. With irrigation, a part ofthese soils are transformed and can be cultivated,particularly in the district of the Segre andCinca rivers.

ARNICA II-1051

aD(8) Cl

A /---. .....

DOCUMENT ERTS-1 INTERPRETATIONMSS 5 - 1027. 10141

from Forest Map of Spain - 1/400 000and Soil Map of the Lusitano-lberian Peninsula

I/1. 500 000

- Crops irrigated from main rivers bed rQuercus lusitanica

· Crops irrigated from alluvial terracesf Forests of Pinus halepensis

;-1 Crops irrigated from saline soils Pinus silvestris

C Crops at'the mouth of the Ebro m Sparse forest of Pinus silvestris

Lightly covered soils and non- I Mainly forest landscapeirrigated crops

A-A' : Direction of the profile

ARNICA II-1

b

051

(9)

19

cl ~ ~ ~

~ ~

t-.~~~~~~~~~~~~~~~t

0~

~~

~

CC

-C

A~

a::

t~ t~ \ B

00

Eo< F

-4w

( w

o

4~

~~

~~

~~

~~

-

0~

~~

~~

~

F

a -

a~

~~

¢

ti.I (.

--. .~

s

%.

-C

1~

;

D

'1

rn

to~

~~

~~

0

.0

j

2 ne~

~~

~~

-. 0

1 i

gP

i2W

481

WE

1"0

Pr~~~~~~~~~~~~~~~~~~.

U.0

~~

~~

~~

~~

Iii

.5~ ~ ~~~~~~

IA

RN

ICA

II-I

051

I

IN

(10) 20

3. Crops at the mouth of the Ebro

Plans for land use include cultivation of salinesoils. The district at the mouth of the Ebro iscomparable to the Camargue in France, a districtlocated at the mouth of the Rhone River.

Table IV (parts 1 and 2) shows the crops andtheir characteristic combinations of texturaland structural characteristics in the fourspectral bands.

TABLE IV (part 1)

ARNICA II-1

CropsSpectral

band 1 a 1 b

(Textural Medium Medium(level gray gray

4(Structure Homogeneous Average

homogeneity

(Textural Dark Medium gray/(level gray Dark gray

5(Structure Average Average

homogeneity homogeneity

(Textural Light Light gray(level gray

6(Structure Homogeneous Homogeneous

(Textural Light gray/ Light gray/(level White White(Structure Homogeneous Homogeneous(Structure Homogeneous Homogeneous

051

(11)

Spectral Cropsband 2 3

(Textural Medium gray/ Medium gray(level Light gray

4(Structure Heterogeneous Homogeneous

(Textural Medium gray Dark gray(level

5(Structure Heterogeneous Homogeneous

(Textural Light gray Light gray(level

6(Structure Average Homogeneous

homogeneity

(Textural Light gray Light gray/(level White

7(Structure Average Homogeneous

homogeneity

TABLE IV (part 2)

B) Forest

Three forest types are identified :

1. Forests of Quercus lusitanica : Quercuslusitanica is a sclerophyll hardwood. In the nearIR spectral band, sclerophyll hardwoods have thesame spectral response as conifers under imageryrecording conditions at low altitude (3000 to8000 meters) ; they absorb near IR radiationwhereas the deciduous trees reflect it. Quercuslusitanica has a spectral response in betweenresponses of deciduous hardwoods and those ofconifers. Quercus lusitanica may exhibit anidentical response at higher altitudes.

ARNICA II-1

21

051

(12) 22

2. Forests of Pinus halepensis : Pinushalepensis is an indeciduous lowland conifer.It is very common in this test site, coveringall moderate elevations.(Sierra de Alcubierre,871 m), which are favorable to its growth.

3. Forests of Pinus silvestris : Pinussilvestris is also an indeciduous conifer. Butecological conditions favorable to its growth arefound at altitudes above those favoring the growthof Pinus halepensis. It is present in only twopoints of the test site. It covers two mountainchains (Sierra de Cucalon - alt. 1491 m., theother 1201 m.)

Table V shows that identification of this treespecies is possible from textural characteristics.The Multispectral scanner 5 data permits isolationof the two conifers (dark gray for Pinus silvestrisand medium gray for Pinus halepensis). With theMSS 6-7 data it is possible to distinguishsclerophyll hardwoods (light gray) from conifers(medium gray).

TABLE V

ARNICA II-1

Upper stratumSpectral

Spectral Quercus Pinus Pinuslusitanica halepensis silvestris

4 Medium gray Medium gray Medium gray

5 Medium gray Medium gray Dark gray

6 Light gray Medium gray Medium gray

7 Light gray Medium gray Medium gray

051

(13) 23

C) Total identification zones

Certain zones are not easily recognized. Thedifficulty is due to low contrast which does notallow one to plot the boundary between twoformations.

Two features groups can however be roughlydetermined using the forest map :

- lightly covered soils and non-irrigatedcrops which cover the lowlands.

- Mainly forest landscape covering thehighest districts. This set of features coversthe Pyrenees in the North the chains parallel tothe Mediterranean coast which then bend inland atthe level of Tortosa near the mouth of the Ebro,in the east and south.

IV. ISSUE

The data analysis makes possible to identifyquickly ten soil occupation units by means ofstructural and textural features of multispectralscanner imagery. These features can be held onlyas hypothetic interpretation basis, whenever thequantitative analysis does not involve a ground-truth phase.Only after this phase, it will be possible to usethe hypothetic basis as standards. The ground-truth phase could be made in cooperation with theEcology and Botanic Laboratories in the Universi-ty of Barcelone (Spain).

After unit identification, the second ERTS 1 dataprocessing phase, the quantitative analysis, couldbe made from results obtained during the firstqualitative phase. So, these results will besignificant quickly for the areas which have beenidentified as interesting zones.This processing phase will be included in thenext report.

ARNICA II-1051

(14) 24

The ERTS 1 processed date census, will improve theknowledge of the different units located in theNorth-East of Spain, from a physionomic andspecific point of view as well as in an economicaspect.However, a great interest can be found in the ERTS1 data, because they are up-to date, repeated.So, they present intervention possibilities, whichthey are significant from an economic point ofview.So, it will be possible to map forest and agricul-tural inventory ; and next, this one will permitto know any change hapenned in the morphology ofthe different vegetal soil-occupation units,whether is foreseeable (different phenologic stateof plant) or sudden (fire). Finally, this changewill bring to realise the different necessaryactions.

ARNICA II-1

C,

051

(15) 25

BIBLIOGRAPHY

ALDRICH (R.C.). 1971. Space Photos for Land Useand Forestry. Photogrammetric Engineering,April 1971.

CENTNER (R.M.), HIETANEN (E.D.). 1971. AutomaticPattern Recognition. PhotogrammetricEngineering, February 1971, p. 177-186.

DEFFONTAINES (P.). 1949. Le Delta de l'Ebre,etude de geographie humaine. Compte-rendu duXVIe Congres International de Geographie deLisbonne, 1949, p. 525-546.

DEFFONTAINES (P.). 1964. La progression de Lerida,ville sous-pyreneenne. Revue de Geographie desPyrenees et du Sud-Ouest, T. XXXV, fasc. 3,p. 263-273, 1964.

DUVIGNEAUD (P.), DENAEYER DE SMET (S..). 1968.Essai de classification chimique (elementsmineraux) des plantes gypsicoles du Bassin del'Ebre. Bulletin de la Societe Royale deBotanique de Belgique, T. 101, fasc. 2, p. 279-291, 1968.

GAUSSEN (H.). 1972. Les cartes de vegetation dubassin de l'Ebre. Pirineos, n° 105, p. 69-83,Jaca, 1972.

HYCKA MARUNIAK (M.). 1960. Pastizales de losMonegros y posibilidades de su mejora. Analesdel Instituto Botanico A.J. Cavanilles,T. XVIII, p. 53-67, 1960.

LAUER (D.T.). 1969. Multispectral Sensing ofForest Vegetation. Photogrammetric Engineering,April 1969, p. 346-354.

MONTSERRAT-RECODER (P.). 1956. Consideracionessobre la mejora de los prados en Seo de Urgely valles proximos. 1956. Cooperativa lecheradel Cadi. Seo de Urgel, Fondo de Obras sociales1956.

MONTSERRAT-RECODER (P.). 1966. Vegetaci6n de laCuenca del Ebro. Centro Pirenaica de Biologiaexperimental, 1-(5), 22 p., Jaca, 1966.

051 ARNICA II-1

(16) 26

RIB (H.T.), MILES (R.D.). 1969. AutomaticInterpretation of Terrain Features.Photogrammetric Engineering, February,p. 153-164.

STEINER (D.). 1970. Time Dimension for CropSurveys from Space. PhotogrammetricEngineering, Fevruary 1970, p. 187-194.

ARNICA II-1051

27(1)

Chapter 2.

REMOTE SENSING OF FORESTRY

(USING ERTS 1 DATA)

G. FLO U Z A T

Department of Teledetection

PART I - IDENTIFICATION

ARNICA II-1051

(2) 28

I. ABSTRACT

This paper discusses the use of ERTS 1 data inforestry remote sensing by correlation with thevegetation map (a known ground truth). Itdescribes the first part of an experiment on theapplication of acquired results ; furtherdescriptions will appear in two following papers.Identification of forest trees is made using partof a determination key, such as phenologicalcycles of remote sensing. With ERTS 1 data fromone pass, fir forests (Abies pectinata) areeasily detected, as are forests in which firtrees are mixed with beech (Fagus silvatica) orother pines (Pinus silvestris, Pinus uncinata).Forests of pine only are also located.

II. INTRODUCTION

The remote sensing of forestry is a fundamen-tal aspect of the ARNICA/ERTS 1 program carriedout by the Remote Sensing Group of Toulouse(France). The first stage of this study consistsof applying point correspondences research tothis type of ground object. In keeping with theARNICA project and the data analysis plan, theoperational definition of a correspondence codebetween satellite imagery and the vegetation mapinvolves searching for ground object identifica-tion keys. These phenological cycles of remotesensing are made from point calibrationsestablished in well known spectral bands duringsignificant phenological periods. The groundtruth reference is given by the Vegetation Mapof France (scale 1/200.000). Thus, calibrationof vegetal phenomena is possible everywhere inthe test site, depending only on the availabilityof ERTS 1 data.

This experiment uses data acquired by ERTS 1during Summer 1972. It will define the meaningof an image in several wavelength bands for agiven date, in relation-to the characteristicphenological cycle of remote sensing.

ARNICA II-1051.

(3) 29

III. REMOTE SENSING CONDITIONS

a) Recording Conditions :

The following experiment used ERTS 1 imagerytaken on August 19, 1972. It was the first data tobe used in the ARNICA program. Due to late arrivalin Toulouse, however, this data was not availableuntil the end of November 1972.

The characteristics of these images are givenin the following table :

002 19AUG72 C N42-49/EOO0-37 N 42-49/E000-39 MSS 4R SUN EL52 AZII34 192-0372-G-I-N-D-IL NASA ERTS E1027-10135-4 01

014 19AUG72 C N42-49/EOO0-37 N 42-49/EOOO-39 MSS 5R SUN EL52 AZII34 192-0372-G-I-N-D-IL NASA ERTS E1027-10135-5 01

026 19AUG72 C N42-49/EOO0-37 N 42-49/E000-39 MSS 6R SLUN EL52 AZII34 192-0372-G-I-N-D-IL NASA ERTS E1027-10135-6 01

038 19AUG72 C N42-49/EOO0-37 N 42-49/EOO0-39 MSS 7R SUN EL52 AZII34 192-0372-G-I-N-D-IL NASA ERTS E1027-10135-7 01

At the time these images were recorded, the Frenchside of the Pyrenees was under clouds. The Spanishwatershed and high mountains were clear.The National Weather Report for these data showsclear weather for the south side.

The data acquired over mountainous areas isapparently more reliable than that acquired overlevel country where the air layer traversed by thereflected flux was thicker. These conditions limitthe optimum region for efficient sensing to theAran Valley. However, it is necessary to known thelimits of the angular variation of the images overthe region studied. The luminous flux levels.recorded are a function of :

ARNICA II-1051

(4)ERTS trajectoryground trace

N

I. 1Ok ARAN VA10km VA

FIGURE20 m

FIGURE a

1OkmbLLEYI

9 Ikm,,ERTS 1

oK = O1

10 20

FIGURE b

ARNICA II-1

0o 30 km

30

Ik

II

051

(5) 31

- changes in the instantaneous scanningdirection

- changes in the slope and orientation ofthe detected objects

The site analysed is a 20 km. x 10 km. rectangle ;the longer sides are parallel to the ground scandirection, and the distance from the smallerwestern side to the ERTS trajectory ground projec-tion is 10 km. (Figure a). The maximum scanningangle variation for the Aran Valley is about00 10' (Figure b).

The maximum slope variation for the objectsstudied is approximately 15° . All the samples weretaken from an approximately NNW direction, allwithin 100 of one another. The distorsion of theresponses due to topography is thus negligible forthis experiment, in view of the methodology used.

b) Phytogeographic features of the site :

The experimental region contains three types ofsoil occupation units :

- agricultural areas

- woodlands

- brushwoods and high moorlands

The distribution of these units in altitude isfrom 1200 to 2500 m. Agricultural areas occupythe valleys (mostly grazing lands), while forestsare situated on watersheds down to 1900 m.The trees present consist of several vegetationseries :

- Beech series (Fagus silvatica)

- Fir series (Abies pectinata)

- Pine series (Pinus silvestris and Pinusuncinata)

ARNICA II-1051

(6) 32

In general, the afforestation density decreases asthe alpine moorlands are approached. For example,at their upper altitude limit, the pine series(Pinus uncinata) display small isolated patches ofwooded area. Pine forests are frequently unmixed,whereas fir forests are practically non-existent.Above wooded areas, moorlands cover limited areas.Alpine grass and grassy vegetation are generallyfound in the highest parts of this region.

IV. PREPROCESSING OF ERTS 1 DATA AND VEGETATION MAP

1. Preprocessing of the Vegetation Map

The-definition of identification keysis basedon quantitative correlation of informationobtained from the satellite and the map. In orderto make use of compatible data, codificationprocesses for zones and samples were applied tothe whole test site (Doc.-SCV 102572) during thepreparatory period of the program (Jan.-July 1972).

Cartographic zones are defined by five physio-nomic criteria and two specific criteria whichdefine the constant characteristics in numericalterms using the code (Conditions for the use ofthe CNRS Vegetation Map in the ARNICA Program,G. Flouzat, Doc. SCV 050472, in French). Thus eachzone has its own criteria ; those for the AranValley are given in the "Zone characteristics"columns of Table I.

Point samples are likewise located and definednumerically for these regions. A few of thesamples treated are shown in Table II in the"Sample characteristics" columns.These numericaldata make it possible, during an experiment, tomodify correlations between ERTS 1 data resultsand digitized ground truth data.

Figure 1 shows the Vegetation Map of the AranValley with boundaries and locations of some-ofthe homogeneous areas (See V,2).

ARNICA II-1051

(7) 33

2. Preprocessing of ERTS 1 Imagery

The data consist of 70 mm. positive transparen-cies in which the usable data occupy 55 mm. Thesedata are obtained by the multispectral scanner ;the four wavelength bands (4,5,6,7) are processedidentically. Photogrammetric referencing is doneby punching the centers of the four crosses oneach document. These holes are made on the filmunder a photogrammetric microscope and have a 50micron diameter ; when numerized, they have themaximum transmittance. The corresponding opticaldensities, located on the magnetic tape, are usedfor rotation and translation, to superpose the 4channels point by point.

The numerization is done with a Joyce and Loeblmicrodensitometer (SCANDIG 25) with a 50 micronstep, which provides ground deviation measurementsapproximately 160 m. apart. The digital record ofthe measured optical densities thus constitutes asub-sampling of ERTS 1 data. These data aresufficient to test the validity of the dataanalysis plan, established through the use ofsimulations. (Data Analysis Plan for ERTS 1 -ARNICA program, G. Flouzat, Doc. SCV 102572, inFrench).

The region studied is shown in Figure II. Thisdocument is taken from a photographic enlargement(scale 1/200.000) of an interesting portion of theprocessed negative (MSS band 5). The zones fromthe vegetation map are superimposed over thenumerical analysis areas.

V. METHOD OF LOCATION OF CORRESPONDING OBJECTS

The method used for the location of corres-ponding objects between satellite imagery and thevegetation map is based on the principle ofphenological cycles in remote sensing.

ARNICA II-1051

(8) 34

The determination procedure defines the relationsbetween the nature of ground objects and theirdetected responses. It is possible to apply itjust after preprocessing. The processed samplesare then accepted as valid and significantaccording to ground truth.

The establishment of such an experimentalprocedure extends the possibilities of the methodin several ways :

- by the use of a greater number of sampleswithout subjective and limitative determinationof the pureness criterion

- by a quick choice of samples withrelatively rough indications

- by the possibilities for generalizing themethod.

1. Scientific Justification

The determination keys sought for the identifica-tion of ground objects are based on the idea ofphenological cycles of remote sensing. Theserecognition models are obtained by a combinationof correspondences established between :

- the composition of a type of ground object

- the characteristics detected in samplesof that type of object.

In the case of remote sensing by the ERTS 1multispectral scanner, determination is a functionof the relationship between the nature of samplesof the same type of ground object and the form ofthe multiband relations for these samples.

This relationship is significant for only a limi-ted time.Thus, an identification key must be used onseveral significant relations of the same type ondifferent dates.

ARNICA II-1051

(9) 35

The sum of these relations defines thephenological cycle of remote sensing.In our experiment on the remote sensing offorestry with ERTS 1, the present study comprisesone part of the development of determination keys.Ground reference is given by the Vegetation Map(See IV, I ; VI, Table'I, Table II), and numericalanalyses are made from the ERTS 1 magnetic tapedata (See IV, 2).

2. Experimental approach

The experimental approach consists of selectinghomogeneous areas on the Vegetation Map.It allowes one the location of pure samples ofstudied objects. The zones containing the soiloccupation types are indexed and the area ofanalysis is narrowed down to forests. Some ofthese area white spots in figures I and II inwhich 29 of the 37 forest regions in the AranValley are analyzed.

Using these results the homogeneous sites areselected withe the statistics distributionscharacteristics which correspond at the digilizedanalysis of samples responses. The homogeneitythreshold is defined with previously analysis ofknow samples used as references. An confidenceinterval (90 °/0) defines the distribution domainof homogeneous sites (fig. III). The pure samplesare, after placed in these homogeneous areas.Using this procedure the appropriate calibrationsfor determining correspondences can be chosen.

3. Determination procedure

First this second stage determines thecorrespondence between the nature of the objectsand their multispectral response. Their responseis studied in each channel and characterized bythe statistical distribution of optical densities.

ARNICA II-1051

(10) 36

VEGETATION MAP OF ARAN VALLEY

(Zoning and samples, Scale 1/200 OOOl

Figure I G.FLOUZAT - 01.73 051 ARNICA II-1

( 1 1 ) 37 ERTS 1 DATA ON ARAN VALLEY

(Enlargement from MSS 5 negative, Scale 1/200 000) Zoning and samples from Vegetation Map

G. FL0UZAT - 01.73 051

Figure II

ARNICA II-1

(12)

MSS 4

MSS 5

MSS 6

MSS 7MSS 7

38

1 23 4 190%

1 90%

1 I 90%

0 ) o10 20 190%

Multiband relation examples for4 sites according to their I 90 0/.

Homogeneity threshold for I 90 °/oin this experiment

FIGURE III: Diagram of multibandrelation for I 90 %/

ARNICA II - 1051

(13) 39

The averages acquired in the four-dimensionalsystem define clusters of responses correspondingto different types of ground objects. Theseclusters form a part of the phenological cyclesof remote sensing of the objects studied.

Utilization and application of this determinationworks with the same procedure. Multispectralanalysis permits the identification of theresponse of an unknown object in the space ofcalibrated clusters. Its location must have twoforms :

- in a cluster area : the determination isdirectly acquired

- between several clusters. A method ofbarycentric classification defines the nature ofthe object.

4. Modification of the method

The first part of the experiment in remotesensing of forestry did not follow exactly thisprocedure :

- The pre-processing and approach werefollowed normally as described.

- The determination procedure was modifiedin this study. This change affects the order ofregrouping operations. Clusters of responses areestablished : in each spectral band.

The combination of these clusters determines thecharacteristics area of an object an a fourdimensional system.

After this simulation, several of these pre-analyses show possible areas for improvement ofthe technique :

1. The availability of a numerization witha 2 5 M step.

2. Improvements in lining up the points inthe different spectral bands.

3. The automatic inclusion of geographicalcoordinates.

4. Use of the ERTS 1 magnetic tape.

ARNICA II-1051

(14) 40

VI. RESULTS-INTERPRETATION

This study is being conducted for the purpose offorestry identification. Nevertheless, sites andsamples of zones without forests have also beentreated. The cartographic pre-processing shows29 wooded zones and 8 low-vegetation zones.

1. Experimental approach

This analysis has been carried out using themethod described in V, 2, applied to thecharacteritic sites in 3? zones. The codedetermines the areas to be treated as a functionof the dominant elements in the zone. The averagearea of the sites studied is about 150 ha. Inorder to show the successive phases of thisprocedure, examples are given on one type ofobject : Fir forests (Abies pectinata). Figure IVshows the sites selected in multispectral band 5after the establishment of a homogeneitythreshold. The confidence interval I 90 % permitsthe elimination of non-significant distributionscoming from heterogeneous areas with :

I 90 °% < 14 classes of optical density

Selection by cartographic coding in figure IVshows the homogeneous fir forest sites. Table Igives the corresponding characteristics of thisanalysis. (The areas presented here in white infigures I and II). This operation is carried outfor all the sites treated in the four availablespectral bands. The definitive selection ofhomogeneous sites is made with the fourthresholds according to the principle in figureIII. In this study of the Aran Valley thisprocedure reveals 70 homogeneous sites in whichsamples can be chosen. The cartographicpre-processing designates 50 forest sites (inthe 29 corresponding zones).From the first calibration results several thingscan be noted concerning the control phase of thisexperiment. The four spectral bands reveal twotypes of homogeneous fir sites. Figure IV, alone,suggests this distinction between analyses 11-16and 17-114.

ARNICA II-1051

(15) 41

SYMBOLS USED IN TABLES AND FIGURES

no F Sheet identification of Vegetation Mapn° 0 Zone identification in the sheetno A Numerical analysis identificationn° E Sample identification in the zone

C. UTM Square location in UTM SystemX Location in UTM System :Y X=meridian location Y=parallel location

Ar ArtificializationSbt Total woodland surfaceB/V Woodlands/Natural vegetation ratioR/S Reforestation/Woodland parcels surfaceT TopographyU.S. Soil Utilization

NM Number of measuresNMo Number of modesVm Maximum value obtainedVar VarianceMoy Statistical distribution averageT Standard deviation

D.O. Optical densities

Serie 9 Pine vegetation series (Pinus silvestris)Serie 10 Pine vegetation series (Pinus uncinata)Serie 11 Beech-tree vegetation series (Fagus

silvatica)Serie 12 Fir-tree vegetation series (Abies

pectinata)

Phy : 3 Physionomy : forestNat Nature of the vegetation seriesMel Mixing : 01 = 2nd species 10 O/;

10 = 2nd species 10 %/aE Exposition : 8 = NNW, 1 = NP Slope percent : 5 to 7 = 40 to 60 /o

ARNICA II-1051

42TA

BLE

I

U

In

In20

n In

In N N

In

IC O

I

In

In

In

O

C-

0 .-

IIn

L

n

Cn

sn

0 I n

00

In N

_ C

_

O

com

I

m

I I Z

I

LO

m

m

m

CM

cnO

In

>

f In

dM

0 r\

Nc)

InC

C

O.>

I I

I N

n

In C

N

O

t M

M

O

C

q

(O

rro L

NLO

aC

c0 N\

r0 U)

N

In Ln

CO nC

M

0 1-

C

;t n

I N

C

M0

O- n

Q- O

1

NO

N n

¢)

O)

CMN 1

0 O

In

CD

C

Cn

I n G

I O

N

N

N

E

Ln N

In

1D q

In I

0 In ) In0

q d

In>

~~

~~

f

:E

I m

N

In

Nz

O

'In

C

NIn

O

dCdM

) I

n n

CMI

' In

In CN

~-

0nI,1: O

O

O

O

O

O

O

O

O

O

O

O

O

O00000000000000

0 O

O

0-

O

O

O

O

0CO

000

0 0

000000

n 0

000

00

0 0

0

.'-

O

O

O

O

O

O

O

O

O

O

H

O

O

O

O

O

O0

O

C)

(NI 0

0 N

N

-

0n N

C

-CM

CM C

N

N

C

M

N

N C

-

In

In

I I tn

In

O

n rn

In In

In to

In

t o

n

.O

u tI

In In

.(.In

In [

t In

In In

-

In

O

O

00o-

0In O

n

In On

In In L

n000

n) Ion

IO

O

m

m

m

n N

01

N

N

CM

C

M

In In

cO

I o)

N- N

CM

In

X

N

N

n n

n n

n

0um II

Im

I

II

I I

I I

I I

I0

0 0

0 0

0 0

0 0

0 U

0

3i

-n

t _

O

In In

N

I O

0

-N

,I n 7--

O

0n C

N N

M

n -

-N

N

In

n CM

CM CM

In CM

-t-

-

N

N

N

N

N

NN\

N

N

N

N

NN

N

N

N

- N

- N

N

,

N

N

N

N

N

NI

N

ARNICA II-1

(16)

HInHOGH CLU

LL

0NOH

0-J

II

051

43

TABLE

II

O

N00 CO

CO

0 0

_ _

_ _

_ _

_ _

_ _

_ _

_

F)T Cn 4 (O 4

4 CT

L(T uM

't

m

mn

aCD

st

C"

0 L

,

N

N

va~

O~

fl~

N~

tflbt

C) 0 cn

-

rD

L

G

(D CO

D Lo

to

m

-r)

N(d

( -

~- v-

N

CO N \

o

o N

L

C

) Ltn Ln

N"

zt q

N

NUD O

t

D q

(,D LO

C\J D

0) N\

o CO) (

Cs

tz C C

s t4

E

m

to

e tD

eo

tv

rD

tC\

m

rn

v

9 LO

q C

n C

, 0

CN

C,

Lr q

CD L

CC C

ON*

(D

C

0 q

tl 0o

C0

N

(t z

(n cNoN

L

r O

D

-N

C

) N

O

N

0O

7-I

C- C

- C

. C

C

C

C

C

-

0 0

0 0

0 00

00

0 0

o 0

0o

Ln Lo

o o LO

O

o o L

n

L

C

1)

NC

- °

°

L

cO CD

-aC

O

aC -c

o

CC) C

DM

N

-O

o O

O

O

O

0

00

0 0

0 0

0O

O

O

O

0

00 o

-O

O

O O

Z-

zn

z L

O

o -C

Z

N

c'

N

N

~J zLL

CI-

C-

C

>

c \ C

- C,

C, N

\ cN

>

C

, C

o

C~ N

C\

~ , CC

C[

C C,~

, C

C, C

(17)

CI)

HWOI-IIz0Hi-Lo

H0'oW LL

a:I.-Li

LLw-iLU -mCO

...... ---

-

C

mC

O

o CO CO

M

CY m

C

n

CO

AR

NIC

A

iI-1051

(18) 44

The largest heterogeneity in the first group isdue to the geometric shape of the site locations:the vegetation map shows that the points of non-wooded areas overlap into the forest areas. With,the homogeneity of the second group (No. A :17-114) sites 19 and 110 cannot be distinguishedas on the vegetation map : they contain an equalmixture of beech and fir.

2. Determination procedure

a) analysis of samples :

The samples used in this second analysisare taken in homogeneous sites. They thereforehave a slightly smaller area (average about100 ha).The numerical analysis for the determinationprocedure (V, 3) is applied to these samples inthe four spectral bands (MSS 4, 5, 6, 7). Thepureness of these reference objects is definedby a variance analysis. It is clear that thestatistical distribution of measured pointsoccupies a smaller number of density levels.Figure V shows the average and standard deviationsof 14 samples taken in the sites presented infigure IV. Here, the distribution of points wherefir trees are present is found in a region + 5.5standard deviations about the mean. Thecharacteristics for these samples are given inTable II. These make it possible to define theoptical region in which the presence of firforests has a high probability. The grouping ofthe analyses for this band is thus found at :

145 + 5.5 classes of D.O. (optical density)

Identical regroupings are made in other bands forsoil occupation units (or sets of units). Thefollowing groups have been found :

- Firs and Pines ; Beech and brushwoods ;Pasture and moorlands (MSS 4)

- Firs ; Pines and Beech ; Pasture andmoorlands (MSS 5)

ARNICA II-1051

(19) 45

- Conifers, Deciduous, Pasture and moorlands(MSS 6)

- Conifers, Deciduous, Pasture and moorlands(MSS 7).

Their position on the optical analysis scale isgiven in figure VI for-various multispectralscanner bands.

b) Interpretation of clusters :

The determination is based on the combina-tion of clusters in different spectral bands.However, one cannot distinguish between 2 objectsunless their response groups are separated by onechannel. Figure VII illustrates the multibandrelation : the relative positions of the clustersrevealed by a combination of two bands are shown(MSS 5 x MSS 7).

Identifications are possible for the following :

1. Fir woodlands2. Pine woodlands3. Beech and brushwoods4. Pasture and moorlands

The use of the 4 channels makes it possible toeliminate ambiguities between groups of groundobjects. This does not improve the specificidentification in groups 2, 3, or 4, since thereis no spectral band in between. Other imagerydates are therefore needed in order to improvethese determination keys. The regions defined inthe 4 dimensional system of this multispectralcorrelation make up a part of the phenologicalcycles of remote sensing of the four sets of soiloccupation unitp.

The identifications obtained are the results ofthree different factors :

- Physiognpmy of vegetation growth.The distinction between fir and pine can be madebecause the fir forests are dense and the pineforests are sparse.

ARNICA II-1

II

051.

(20)

APPROACH PROCEDURESELECTED HOMOGENEOUS SITES

(fir woods example)

no A

A

11

12

13

1.4

15

16

17

18

19

110

111

112

113

114

140 150 160DO(optical density)

FIGURE IV

Average responses on spectral band (MSS 5). Thesecharacteristics show approximately the homogeneitythreshold by the confidence interval for this band.

46

051 ARNICA II-1

(21)

DETERMINATION PROCEDUREExample of significant sample responses

for fir woods

no A

31

42

23

54

45

46

57

48

59

410

411

412

213

214

.~~~~

. .

---- 4-

.,- c -- I

t '_ I

l *.

I- *

140 150 160

FIGURE V

This figure shows the increase in pureness of thesamples taken in homogeneous sites

051 ARNICA II-1

47

7 D.O.

(22)

EXAMPLE OF PRINCIPAL GROUPS OBTAINED

(set of point correspondences)

Pines and fir woods

F 1Beech, Brushwoods

I r i IPastures, moorlands

Fir woodsI ...... i

5 l iPines' Beech woods

5 I--qPastures, moorlands

Conifer woods

Il - IDeciduous woods,I 1

6 I I

Conifer woods

7 Deciduous

7 .... . 7 .h140 160 18C

pastures,moorlands

woods,pastures,moorlands

0.0 D.(

FIGURE VI

Groups are made, in each spectral band, with thecartographic sample code and characteristics ofthe responses analyzed on the same samples(modified process of the determination procedure).

ARNICA II-1

MSS 4

MSS 4

MSS

MSS

MSS 6

MSS

MSS

MSS

48

051

49(23)

ILLUSTRATION OF BISPECTRAL DIAGRAM

(example of simple correlation)

Principal determination areas

The use of theand ambiguity.identification

MSS

D.O

180

170

160

150

140

four channels avoids overlappingThus, the confidence of theis good.

5

160 170 180 190 D.O.MSS ?

FIGURE VII

ARNICA II-1051

(24) 50- Specific plant response : a clearer specific

identification can be made between conifers anddeciduous trees than between conifers and lowvegetation.

- Regional phytogeographical conditions : Beechis identifiable because it is the only deciduousforest tree which can grow in this region.

Thus it seems that the essential criterion foridentification is the distribution of plantmasses on the ground. These various results arein agreement with previous experiments usingsatellite data. (E.g. : NIMBUS data over theLandes forest (France, SW) ; analyses andinterpretations of ERTS 1 data over Spain (NE)(P. Gouaux 02-73)).

VII. APPLICATIONS : ANALYTICAL PLOTTING ANDINVENTORY

Using part of a determination key, the efficiencyof the preliminary study can be tested.

1. Analytical cartography

The example of the fir forest will be usedagain. The multispectral correlation defines acluster whose cartographic representation can bemade selectively. For this, the geographicaldistribution of optical densities contained inthe clusters in each band are combined for agiven ground object. A cartographical simulationof this combination is given in figure VIII(scale 1/200.000). Only the black symbol issignificant ; it represents fir forests. Theother symbols used merely indicate vegetationmasses on the ground.

2. Inventory

Surface calculations can be made using theautomatic production of this simulation.For the characteristics of the numerization, onedigitized point corresponds to a 160 m. square,or an area of 2.65 ha.

ARNICA II-1051

(25) 51

The planimetric simulation at 1/200.000 infigure VIII requires a contraction of the outputprocessing. This diagram uses one character for15 points of the basic numerical analysis. Thusthe black marking for fir forests equals 40 ha.The direct count thus provides us with anestimate of the area occupied by this naturalformation.The measurements thus obtained can only havemeaning for an entire region in which the numberof points counted provides good confidence.Thus in the study of the Aran Valley it is themeasurement of the total fir forest area that isindicative. This region contains 123 sifnificant"fir" cluster characters, or 123 x 40 - 5000 ha.measurable fir forest. However, the foreskinZone No. 9 are masked by cloud cover (see figureII). One must therefore compensate with anestimate. This adjustment applies to 1000 ha.The fir forest area detected in this region isthus a minimum of : 6000 hectares.

A parallel measurement is made on the vegetationmap over the fir forests plotted in the areastudied with the following result :

7 400 hectares

Therefore, between the inventory made by means ofremote sensing and the cartographic analysis,there is a deviation of 1400 hectares, whichrepresents about 18 '0. Part of this inconsistency(-.7,8 O/0) is due to deforestation made after thecartographic analysis date. (The sheet for FOIXwas edited in 1964). Another part (,10 O/6) is dueto the data processing method. This differencecan be reduced in two ways :

- improvements in pre-processing : using a25 p numerization (sampling x 4), processing theoriginal magnetic tape.

- repetition of the calibration experiment :definition of new correspondences on characteris-tic phenological dates in order to improve thedetermination key ; grouping according to theunmodified determination procedure.

ARNICA II-1051

52(26)

SIMULATION OF AN ANALYTIC CARTOGRAPHYOF FIR WOODS

a W covers 40 ha of fir woods_ ~ ~ ~ ~ -..-

I 4' + e. II II I. I1 e + + ..

+ ..W + .e ..* 0 m + H+ + + ..w c44* + .. 0 W f * +q +

. W + 4 . + *e 44 b. . + + 4.4l' '+ ~..... .. ee..0.. . . .. 4. 0 S**l I e e e

I4. .. m '. .e .1 *. .4 4 .· e H+ .1 ' .e *e 5 4 *e .e 4. .1'

* 00 4. 5.444i555 4+

i '+ "1 I* I * I# ' *I Igo + h N

+4.4 + 4.. 444 O bo +4 * S S *4 .4 . · * 41 4 , I I II/5 -l \.55... + *.-.44 .. , 4 +. + \

.· o ,+0 4 4.. . ( . * + +j

.. 4. ... . 4.5 . ..... ·0 e 0 4. 4. *0'1' 5 5 44. o .o I I

+1 + 0 + + + ' # - m ~ 'qYl .- " .. .+ ,+ *4 ..0 4 .I Y e

* 4.. 4.* *- -544a 5 4.+ ' e 4 .. 4. *4Z 4"' ''

*· .+.* + .* Il l v '

· +. .1 .. + . m . 4 . *. *· 44 .-. 4 + *-.4 .. m 4 . .b I I I m + .

4. * +0.**U* +.... 4... +.· ..0 .....*5 *~ 4 4, . , + ..

· * ... e 4. + M2~.e .~+ ' 4 .e 4' .e* 0 .

00 4..... e. e e.I. . e 4... *0· e +~ ·4 + *e .e4 - e .4e +. + 4 ... e

.· o .... . . + .... + .. 4. + .

. o+ .. .. ,+ ' . .. .5 + *-.

+ = * + *F F * * * { * * 4 + s + +

.· 4 4 *: 44 .. ... . .. .+ 4

.4*-* *4"w- .+ O .f. + .. .. 0.. 4.4 4.+ 444. 4 ·. + . 4 . 4 . .. 4 . . ..

I4 . 4 . 4 + 4.. .e . *. + .e *. .

· 5 e 55.0 .00 .e 4.. .

... . . I+ 4 4 4 m II *e .5 . .04 . .e .0 e *0 -4. 4. 5 *o I4 .5 .. e.e + . .. . +

.. ... .... +. 1-e . . . .. 0.4 4 + +

· e .e 4 ~ . '1' '1 ~4 + + ' ' '-4 00 + + I -. .. I. 4 .+

. e .. . . .* . .... e + 4+ * * .- + 4. 5 4 4. *- + 4. -

4. + ' .0 ,0 ..s4.5.* 000 .4......+ I4 + *0 4. *4 .0 4. .44* 4 *- 00 00- 00- 4 00 0

. 5 4 " .4 ~ . , s...4 * - .4.45 4 ... + .. 0 04 4 .9 4 .4 + .. 4... + .... 0 . *0 * 4. 5 .444 44... + 4.4...... .

4.4 + + *F *- * I *...................

.0 *- *0* I · * + * + 0 06. + + + .*

eS+ + * *I + + + '0 + + = +

4*

*0 I. ~ .64.-.. 4. 4. +

+4. . + 4.44. + .* + 4.

e * , + -+ · +1 0 b+ + p **

4* 4* 3 +04. *0 *

.. +00 64. .

MO wo-

a.,

(I I * nc*

ow 0 w *+. +I0 " Wu 0

I0 06 0

+ .4 W* +

.5.54.6· 4n 554.

4. *4 554..44 . 44.5

0 4 554. 5 44..04- 344 5 4.555.

6 .- .. 5 + ·· .e1,· .- +

....

*6

00 .- *

00 *0

6· 00

00 0

00 06

00 ·

.0

*0

0 *-.. 0

'4.,.0

*0.

FIGURE VIII

ARNICA II-1

F.

16

II

051

(27 ) 53The expected accuracy of this improved inventoryis about 5-6 O/o. This seems sufficient andsatisfactory for regional studies carried outwith remote sensing data obtained by satellite.

VIII. CONCLUSIONS

This first phase of the experiment in the remotesensing of forestry thus provides us with twotypes of results :

- The four main sets of soil occupation unitsin mountain areas can be identified fromcorrespondences determined by ERTS 1 imagery.

- The selection of points belonging tosignificant groups makes it possible to measurethe areas occupied by these units.

The application of this method of findingcorrespondences will provide a means of checkingthe plotting of natural ressources in the zoneschosen. The ensemble of experimentation has thefollowing structure :

- First part :

Identification and plotting of soiloccupation units.

- Second part :

Application to the actual construction ofthe Vegetation Map on the Spanishwatershed (sheet for FOIX).

- Third part :

Applying the acquired determinations tothe vegetation map (sheet for LUZ, now inpreparation).

These methodological results and their possibleapplication have a twofold importance : first,for technical reasons, the working application ofdetermination keys for ground objects is a majorobjective of the ARNICA B on ERTS B proposalsecond, for economic reasons, it serves as acheck on and helps the planning of the vegetationmapping using ERTS 1 data.

ARNICA II-1051

(28) 54BIBLIOGRAPHY

ALDRICH (R.C.). Space photos for land use andforestry. Photogrammetric Engineering, Vol.XXXVII, n° 4, April 1971.

ANUTA (P.E.), KRISTOF (S.J.), LEWANDOWSKI (D.W.),PHILLIPS (T.L.), MAC DONALD (R.B.). 1971.

Crop, soil and geological mapping fromdigitized multispectral satellite photography.LARS 061371. Purdue University.

DRAEGER (W.C.), PETTINGER (L.R.), BENSON (A.S.).The use of small scale aerial photography in aregional agricultural survey. Proceeding ofthe 7th inter-symp. on remote sensing ofenvironment. May 1971.

FLOUZAT (G.) 1972. Conditions d'exploitation dela Carte de la Vegetation (CNRS) pour leprogramme ARNICA. In "Recherches actuelles enteledetection et situation du programme ARNICA"Rapport pour le CNES (Doc. SCV 050472).

FLOUZAT (G.). 1972. Plan d'analyse des donnees deERTS 1 pour le programme ARNICA. (Doc. SCV102572).

FLOLUZAT (G.). 1973. Methodologie de tel6determi-nation des objets au sol. (Doc. SCV 020473).

GOUAUX (P.). 1973. Study of land use patterns innorthern Spain : I qualitative processinginterpretation of ERTS 1 imagery. (Doc. SCV020173).

HOLTER (M.R.). 1970. Research Needs : theinfluence of discrimination, Data processing,and system design. Remote Sensing with specialReference to Agriculture and Forestry. Nat.Acad. of Sci. Washington, D.C. 1970.

LANDGREBE (D.A.), HOFFER (R.M.), GOODRICK (F.E.)and staff. An early analysis of ERTS 1 Data.LARS information note 092972.

L.A.R.S. (Lab. Appl. Rem. Sen.) Remote multi-spectral sensing in Agriculture. ResearchBulletin 873. December 1970. LARS-AnnualReport Vol.4

051 ARNICA II-1

(29) 55

MALILA (W.A.). Multispectral techniques forimage enhancement and discrimination.Photogrammetric Engineering, Vol. XXXIV, no 6,June 1968, p. 566-575.

MONCHANT (M.). 1973. Digital procedures on remotesensing imagery for the ARNICA/ERTS 1 program.(Doc. CESR 011573).

REY (P.), GOURINARD (Y.H.), CAMBOU (F.). 1971.ARNICA PROJECT Proposal for investigationsusing data from ERTS A. (SR-051, F 433).

Personnel of the Forestry Remote SensingLaboratory. Analysis of remote sensing data forevaluating vegetation resources. School ofForestry and conservation. University ofCalifornia. Annual Progress Report 30 Septembre1971.

ARNICA II-1051

(1)

Chapter 3.

DIGITAL PROCEDURES

ON REMOTE SENSING IMAGERY

FOR THE ARNICA/ERTS 1 PROGRAM

M. MONCHANT

CENTRE D'ETUDES SPATIALESDES RAYONNEMENTS

This paper summarizes the microdensitometercalibration and defines the great lines ofavailable software. Schema of actual workis presented and figures show some examples.

ARNICA II-1

56

051

(2) 57

I. Calibration of the Apparatus

We have calibrated the densitometry apparatus

using two types of grids :

- one graduated in optical density in the

nominal density range of 0.05 - 3.05 (I)

- the other a 25 resolution grid

The numerical data were then displayed on C II10070 by means of a 3-step program :

- translation of the numerical tape into10070 characters

- calculation of histograms

- printout of these histograms

The results take into account imperfections dueto the slightly non-uniform rotation of the drumand to the positioning of the image, and haveenabled us to get a quantitative idea of thelimits of the apparatus and the necessarycorrections.

II. Procedure

First Stage

- In order to use microdensitometer data on the

CDC 6600, special data treatment, translation ofinformation, and the development of exploitation

software (COMPASS) have been necessary.

- In figure I, showing the display of the nominaldensity range grid, one observes a jitter in thenumerization of the zones of constant intensitywhich can be estimated at + 4 divisions. In orderto remedy this we have investigated :

- the dark current of the photomultiplier- the statistical fluctuations in the

recording

ARNICA II-1051

(3) 58

A correction program for dark current and for-smoothing of a line photo by a spline method hasbeen developed.

Second Stage

- Development of a sorting program permitting theselectioning of any portion of a photograph. Theuser defines a working zone in lines and columnswhich is extracted, smoothed or not according toneed, and stored, according to size and eventualuse, on disc or central memory.

- On the zone chosen, an integration is performedon an elementary surface defined by the user, asa function of the zone to be analyzed and thedimension of the printer outputs.

- A study is performed of the distorsion of theimage due to the printer and corrections.

- With a 6 line per inch carriage tape a characteris inscribed in a rectangle of 1/10 inch - 1/6inch ; if a symmetrical image exists, there mustbe a corresponding symmetry between the elementarysurface image and the printer display. This isperformed at the discretion of the user.

- Tracing of frequency histograms at the selectedarea, absolute and relative, after smoothing, withor without integration.

- Determination of the statistical characteristicsof the distribution : the average, the variance,errors. (II)

- Visual simulation on the printer. For thisdisplay the user defines

- the factor of intensity groupings- the coding scale. This has been done in such

a way as to create to the eye a visual printoutclose to that given by a photo with a density of14 levels. (III)

ARNICA II-1051

(4) 59

In order to increase flexibility, the userdefines his coding as he desires it using anynumber of these 14 levels. This provides us withthree possibilities for display :

- numerical (internal)- numerically external- simulated by grey tones. (IV)

III. Work in Progress

First Stage, partially completed

- Localization of the same zone on the 4 channelsof an ERTS photo. For this each photo isperforated with holes to define a coordinatesystem.Next an axial rotation Oi is made for each channelon each photo (i.e., [1, ... 4] ), where Oi is theangle between the axial system of the photo andthe axial system of the numerization and conse-quently of memory.

- Study of the correlations between the Ei classesofsxj (dE D1, ... n] ) land forms taking uwK(KE [1, ... m] ) samples ; the covariance andcorrelation matrices are defined enabling the userto choose the most interesting channels.

- Classification of responses by the barycentricmethod, or by the elliptic method on the mostclearly correlated responses.

- Extension of these methods to the differencesbetween the 4 D1, D2, 03, D4 channels which defineC42 = Six differences in optical density.

Second Stage

- Extension of the preceding methods to fourdimensions.- Classification of the responses on known samplesfor the construction of probability laws with 2,3, and 4 variables.- Application of the Fourier transform to thestudy of the spatial frequencies of the image.

ARNICA II-1051

(5)

, s

OmaINm1E SEMmI ON RFM

YES

N

: 5 1*

.LSO 6

M.MONCHANT 01.73ARNICA TI- E I

60

051

(6)

cn

r

LU

w 0(i) L

U

C-)

~i 0

LL

<

Z tF0

4-

cU)

wa

zo -0oZ 8

U-w wU

l-

F-

0LI..U

)u L4

I-QU-

0Z0t53 ZOOo)o0U

Iu

z0~ (Jo

cnVWV

U

o a S

O L

L

F5 u

ZIO

LL8 S

(n L

t <

0L

w u

i w

~lc

WO Q 6

u Q-

e

wU,

fiaL

L

8V

nIw

at =

a. Q~ LL

fa zctt m

~4!i

-a:MHC~U

AR

NIC

A

I1-1 61

051

Ir-

(7) oL

7 Jr I

'tos

cJ ((

A -CO FIGURE

4c 01 o-.. A I--.

051 ARNICA II-I

I

I

)1r.

it7*a .e

2 :

zI,

o it*°

la -

= e

e >-

CI

i t

e

-sJS

-ast =5,

VI

L 7

-J

G'

e~

2

LSO

11el

C....

4

itl

4 ODur

e~

l

r

ee.

H zm

oSX

S*tv E*FSzJCe2

t2itCC14Q****

a t

e e

~66

I Sr.0

aaa

6 65S5

^I

5 .550.s

5 5

t

5 5.5

C

C

5 5.5

.5

5 5

5 5

5 s

5 s

10:5_________________

_

c* _

5 K5

..

£5 _

_ c

_

I 05.0

.0:.5

..5..5

..5I

5..5

..0055.5

5.0

:.5.

I 6~

0

6~

~~

lt 5.

..

..

..

..

.5 5

5 S

..

..

.5

O C

- rF

. 4 U

4P

ac a

C

-4w

I P

W

a C

0:0

,

"I"

!_ !w

_!L

W_,;_

___illl_

____

I;

WrILCa2'.z i 7ritI./' i=I,I,,I=.$e$$

..e

:o

' L

4l.e

I .4 *

I

S

U-II

VI I N

U'

IEC,

Ua,'40

.0C$

zCe

r

*=

CI-1: 4It,4¢

II

IIiIIIIIIIIIIIIiIIIIIIIIIIIIIIIIIIIIIIIi-III

s

-w

C

I I

(9) 64

EXAMPLE OF A 10-LEVELS GRAY SCALE

(for visual simulation ouput)

... 3...3..~3..33... ,33133 *3*33333****33f33333*333*3383333333....................1 3333

340..........3.3.... 4***3 33333333*3333*l*333*133333*3319*33*333** .............. 3333................................ lnI , .!.. ..... I.l.... ...........H............ ....... ............ 33.333

.................................................................................. 3.33

3388933*433333*33463334338.............................................................

·····~~·· e I ·1~1·~~·.....·..........

·I·e·ia ·~e·im·Baj·e· ee· I eI meu·e·e·jiil·iH inl..........ee1mn 3333jl·eme1ai1e66n·s·je1 m..l.. . ........ ....... u·uHlsee 33H

................... ............. +H1nnHHB1eH·n~H 333........... ...... ......... ........ ..IH~UHHHII( 1 eBI.... nlnl~ nnRINB#~UnB33

~H ·u~lauuu Hd~Hs~H~#H~BB~·uS~H~M.....................H 3#4333B3 a4 dH3H334 BH343330 uu344334333.343ii333n133n34..................

,3.1..*..I~t4%33441..t.131*4341.fl.41.* .....4..3..113....................... tI . 1311

1411*33 . I .It . *33.... .. I*1

I. . t*.l~t,*t.**l,.,,,3**.*3331*lflt~*..t3........I. I.....1*11

4**33# 3*344n *4*4 33# 44*44**43#B 3**433.................* ..............

...................... .................

~: : 1 : : 1 : : : i 1 1 1 1 1 1 1 1 : 1 1:........................................... ........ ~ .......

FIGURE M]

051 ARNICA II-1

3*,3,4*334*343* *,434,3 433;*4;4343*434334*34343t*33*4*433***333*3343334**3*

*34*34443,34**4*3*34.3*43*33**3*3*****4*433*3*.44 *333**343344*4*33.*34443*3

*4,,4*3 **3*,343*443*,343*343*4*3*433444**44*4*433**3 *43334,3*4*43 4*44433*4.444444*4444.*444444*44**34 *,*44444443434,4444444343*44*4*444,4444*4,*44,4

-4 * 44*4444 44*44 * *34444*3**3***44443****43344*44444*44 * 44,4*444 *4*444*4**44**3 *444444* 44*444*443*4*34444, 4,33*43*4*4444341*444,,*4*4*4* *444444*433*., 4, 34444* 4 44*4434444444344434 4,**34444,444.444 444*,*4**4**4 * 4***4*44*44*4414*1llllltlt ~ *44*444 144444*,43 1444**444

-, * 4,3*4434*,4,*4 444 4444444444 44444*4 14 444*4 4*3*44*4441 4*,,**434 444* 44***444434**, 4t 4444444 44*44*443444443**441 44 443,*44444444* 444*,44,4

44* *43434344334* 43 434*3*4*44443343444444434341 44*444444444344444*3**4*4

*44*4433*44*4***,*4,444334 444*34*444*3344443444443*3433*4*4*444443443*344*

444*4444444.*444**,4, 4,4444**444444*34444*444444444444443,444,4443444444,4

.* !,?****V?*t*?**?!**t((~() (((

:::FIGUR III?::·····?!:??!: !1 ?:I~::::::::! 105t:1:::?::::::::1:!:: ARNICA 11-1:?llI~l·!:::::

(10) 650............. ............ ---Y----________

VAL D A1AN CODAGE DE LOCALISATION 2 MSS S

CAORAGF El MISE FN FORME DE LIMA'iE

150 520

* 4

+ 171 4+ 4

…-----I-----lo ---- e*-- I

lo-ooo*- ---e-oolI

I* . .....------- Ifhl- I -II

ELFMFNT DE SURFACE

I

051ARNICA II-1

FIGURE IV

TABLE DE CODAGE

119 - 139 I - X140 - 140 X141 - 141 A142 - 14? U =143 - 143 U - X144 - 144 I =

1 4b - 146 E X14? - 147 E - A1)4 - 14A F xA149 - 149 F AISO - 150 E . x)ql - 151 * = 11C2 - 1i t = I1J3 - 153 t = 115 - 154 = I155 - 155 = IIo - 16 t = I15 - L57 : I'8 - !b + 4 -1I¥ - 159 + : -16O - 160 *, : -161 - 161 : -Ih - 162 + :13 - 163 ; :164 - 164 ; :165 - 165 ; :16-1 -1 :17 - 167 : :16 - IhS161l9 - 169 ..I 70 - 170 .171 - 171 .172 - 17 1.7- L73 ,

VVV

VV

FF

FFF

O 140 140 14O 140 140o 14

. I0101010In10

A8a8A

666

666444444

22

(11) 66seet '*t. *'.t *...t.3. u* *---i*... .... ...

IIH.-'!IT'''e'T''T. ltl l I!Itllll' ·- ,··o·(. ... . .. ·· ' * ·'·4.

HI., *t i... *·· * ·.·t **·**te ·e-. ...* *.t :. -*w-- . -............m ·, .,

.,lltt* *ittll*,* lelmtI' II'l* - l*ot .* .. .· . .......

I44..····llIi. 11111S..· . "wrll, .· · ' .· *·l

,.O......~t+. .t....gg_.4e+i 2titi te...... ... *,t

t+**2t-~t1t+· ·lNI(l*'1tl',1· ~l--·I+ll-- ' · ' -· .. . ... · · ··--t-'l+,t4llt9 *l#, O -HII'*··tM··t .l··.' .*····* .·.· IIl·t..ee· duOOlegt g jt ,e· ... ..e... ll,.......· .. ,,

toodel e* ·4I"l, _ l*t le ll^ll-,lfilltfl. ~I",-,,,", tIl(j1 l . .***. ,·'·'*'*·llll

·ti*..li.I···*·lSlll!,eINlS l!*, · *ti-t--··t·to ,4

D ,ee......eo *·llHHlslu,, ... ,..'.:'.ma et .....*..... t ' ".

.... . .... , ,,uu., , ,.,,, .. . ......I ei.....S. . ... ......

...*00 so ·. !,, ,, Os , .IM , Ht........ I ........ ·..... . · ..."'*·OO* ' '·* '°,-,111,--,---.' !i---u ..i........ .......... t ··

:.ntsgsgg~emegge,,.t*e ~............. .. .....

It* 't ' 0 * 0 ' * *figto*. * f o*ot.t· .*..t. t ...· ·.......,,,,,·············· ,., ,, lt ,,,I.*t ....... "Itt-ttHs" " t ttev '*I It.. t I...... * I,.tt ·..' IIftII*M 00

rt.*.e1 _..... o..,...ttN, * **..l ....... .· ··...·l 14H 1 ·* Ils.·*lll44lttjjjjllltjjjl,,,l*O···-*· *- * ·t- I 1·tltI *-*** .· *

"*4+ vj g nrlewg§§s n g *+ + g9,,... . . ......Iag..

tlIIII*___1_*e --- * -- *.. .... ·. *'t..ttg

?tl+.e+**** ·**·* .... **14 IIt* l t*U, *-----*

too#w*itt n*t ..t. .e,.ingg...t..g.,t.,.. t. .eus.,b .... l.·

14t*:::01t .it.lltlBIt01lI i elll4 ......... - *i e*--

k4.it.egHUSgttNt¢.__.I. : - .t ,heMk,-...auff 1,041 ,' ..... ..

4tion"ti *4tII... t it.tt.. .. 9iUOO....O..t. **tt .t_

..0jo.tittott..It ......... U tt....---- -l-t.it

*t-ljlllN*ll-tl +tJ t *+llt *-- -- l*l· - f-·*'·'··'tt- l*''*''·*l_

ew.ttt....... t . . . .......... ... .. **

· t*l·tll*l .t ..... ...ltti..i .tllll·ttti·-t*·· · 4.··

lll it*ll5...t**. t ..**tlt..... .gsllg!* t·t.i ·*IttI.

*lt.IIfl .t.ttl... ·..t. tI·..tl*· Ntttt_ tt0 ..·...tlt

·ell1e**t1it*ttt·ltll ***t.t·- ..ttt· . .ttt **·U*··.·..tlefldlllll

*20*.1.1 1 1 . tiii..ttI ItY gtggUlitltHt .

g ,- ## . .. ..* ...... i.. tttte.**

et It(Ilj. ltlit.. s.l . *tj1,. ijl..... ItttllB41f_·s.·....1iII i -. ·

t.*o*ttt .t. tt ..... i . .. t| .

*40 tItl..w l* * * °,etl, *. *. : ' 'u,.

et .**g..ts... , S...·-.t1 S.l41,.t.1.t.It el, S... .

0 .Itltel

.

, @4 .?.·l.l. -,*tto 4 tU ***i ll*tl'"4*t-l- *

· .5441 ..J.illi 4 l'_***'t·11***l ',*i*0,1..tl4+ l.*·'....00**'l l1* , llt *·*... .. . l

.!t|t··l| ltltl. llt*· **cl.-- ltlll$111*tllol··t ·*fHtll* '*·*H·s,,I1· ll1tllll-| *' *** ltll _H -|tl''t-+ttsts**'l*· ........ · * *, . Il· · *

05 1 ·t.t· * t FIGUREI· ARAtt IN ttI-.· lIt**·ll

' · '- '' · '' 'Irtl ctl Nt ** ' * *-It' .l* *j·llll .* , ., ·· ··.

t *1tt' * *·I-- *t·' ·'14H1,

.05lttf * ·*,',,"1 FIGURE IV ARNICA'II-1

· ·ljJ · **'**lJ0·* 'l' ll' '! ''~~~~~~~~~~~~~~~~~~~

(12) 67

VISUAL SIMULATION

(with recorded gray scale)

'' I1 ' i 'q 4iiill!i i ! t ;ii

i,,,!.l ;,: ,,, ........... · ',,, ,, - ....

, ,, .: . - , ........ . ..... . :..

,lii~i,, . :i- i !, ,iii .. :.. ... -II!!.11t

... .tifl! !1tm li: 'iF:;i .......

:tljl ... !::' . ...... i u {: ?ril i i: · . ..... .. '

:iiiih i ?:l .iiiiiii i :}:iiii; 'ii i ,i

:i , . iiiii j.-,, 1,,!iiiiii - "- . ' j i .: ' . . · '

. i. :, -... ,

: .: - ::-.i i: ! . . .. - -....... ..

·;.i:4 iii~ :::

i a .' . i, ? ' I, · . - ,

' 05i;ii il iiiill h l ' ! iiii--i :

' :-i li ii 'l;' r ':'. '-:~-il;--hi ! |ii-i;ii,;i :.j!jj,ii -i: :r, iii" :,ii,i iiit::::. .:;. r ! ' -l:: ..

'' .· , .': .....-. . ... !- .. ... . .

H~~~~~'' Ii

: :!iiiii i' ' : i ! i ' ' -

: :'; ,~,,,,-,,iiiiiL · ,.: niiiiii~i i : lii' l; i .i .i:ii

051 :RNI~jA II ' ... "'qlj 'i:-:1i'~iiiiiiijiiiT:

051 ARNICA II~-1:::::-

(1) 68

Chapter 4.

STUDY OF A RADIOMETER FOR VISIBLE WAVELENGTHS

Laboratoire de PHYSIQUE de l'ATMOSPHERE

Service du Professeur PICCA

UNIVERSITE PAUL-SABATIER

TO ULO USE

ARNICA II-1051

(2) 69

I. ANALYSIS OF THE EXPERIMENT

A) PURPOSE

To measure the energy flux received from theground at various wavelengths in the visiblerange.

8) USES

1. Study of the effect of lower atmospheretransparency by recording the energy fluxreceived from the ground at various altitudes.

2. Study of the influence of atmospherictransparency between plane and satellite levels.

3. Study of radiation received from theground in conjunction with weather conditions ;study of land and vegetation characteristicsobserved from the air ; all of these in threebands of the visible spectrum.

4. Evolution of radiation received from theground during the year.

NOTE : The measurements are affected byground illumination (solar radiation),re-emission from the ground (land and vegetationcharacteristics) and transmission of radiationthrough the lower atmosphere. These three effectscan interfere with each other. In carrying outthese experimental flights the relativeinfluence of each of these factors will have tobe determined as accurately as possible.

C) THEORETICAL STUDY

Illumination E produced on a screen with areaS' illuminated at normal incidence by a sourcewith area S at distance r and luminous energy Lis :

E = L r (1)

ARNICA II-1051

(3) 70

If S' is the image of source S in a lens with afocal length f whose aperture is limited by acircular diaphragm with diameter D (see figure 1)and if the source is far enough away, formula (1)becomes :

L AT (D)2

D= relative aperture of the lens.

f

II. EXPERIMENTAL APPROACH

A) ORDER OF MAGNITUDE OF THE MEASUREMENTS

The total solar radiation received by theground, in the visible range (between 0 .4 M andO.7M ) is an average of 400 Watts/m2 in a clearsky. In the following calculation the energyflux is considered to be uniformly distributedin this spectral band.

The terrestrial albedo is highly variable (froma few O/o to 70 or 80 % for snow). It is generallyestimated to be 10 O/o. The filters used have abandwidth of approximately o.1 J and theirabsorption in this range is 30 °/.

The luminous flux transmitted through eachfilter is then of the order of 9 Watts/m2.

The area of the source is 104 m2 and thealtitude of the sensor is r = 103 m.

Without a lens, equation (1) gives :

E = 90 mW/m2

With a lens having a relative aperture ofD I and neglecting its own absorption :f 3,5

E = 500 mW/m2

ARNICA II-1051

(4) 71

8) CHOICE OF EQUIPMENT

Our choice was influenced by two conditions :

1. To limit the angular field of view sothat measurements are made on a ground area ofI hectare, which corresponds to the recordingresolution of the satellite.

2. The detector must be able to record lowillumination. In order to sight the same area atdifferent altitudes, we have chosen SOLIGORlenses whose focal length varies between 55 and135 mm ; with the addition of a focal lengthextender a 1 hectare area on the ground can besighted from altitudes of 550-2700 meters.

The detectors must be capable of measuringillumination much lower than 100 mW/m2, sincethis experiment is being performed to a largeextent for the comparative study of thereflectivity of various land areas. Alsoexperimental conditions require light materials ;therefore, we have eliminated photomultipliers,whose power sources and associated electronicsare too bulky. We have chosen a photo diodecoupled to an I.C. amplifier a UDT 500 fromUnited Detector Technology, Inc. Figure 2 listsits principal features.

The filters, made by Metallisation et Traitementsoptiques, were chosen to have bandwidth whichdiffer as little as possible from those of thefilters aboard the satellite. They are twofilters in the SPECIVEX series ; the DA 531C andthe DA 620C' whose transparency maxima are 0.531Mand 0.620 , respectively ; they work undernormal incidence and are obtained by combiningabsorbent materials and thin layers multi-dielectric or not. The third filter A 680b' isa Fabry-Derot interference filter its maximum is0o.680

ARNICA II-1051

(5) 72

In order to relate recorded signals to thenature of the ground sighted, the Service desActivitds Aeriennes of the Institut GeographiqueNational has put at our disposal a type T 232 MK7 35 mm camera equipped with a 50 mm lens.Measurements and photographs are coordinated bymeans of an intervallometer adapted to thecamera.

All of the results are stored in a MulticorderWatanabe 4 channel chart recorder, seriesMC 611 H ; three channels are used for recordingsignals transmitted through the photo diodes ;the fourth is used for peaks coming from theintervallometer on the camera. The sensitivityis 1 mV to 100 volts full scale ; the chartspeeds of the 25 cm-wide paper are 300, 600 and1200 mm per minute or hour ; each pen uses theentire width of the paper. This apparatus workson 220 v. AC ; the measuring system is completedby a SD0 501 SODILEC DC to AC converter whichconverts the 27 volts DC available in the planeinto 220 volts AC.

Photo 3 shows the detector ; figure 4 shows theexperimental system.

In conclusion; this first report consists of apreliminary study of the experiment. The entiresystem has been operational since January 15,1973. The technical description and the variouscalibrations will be presented in the nextreport.

ARNICA II-1051

(6)

Screen with area S'

Cir diaphragm wit. ciaphragwith diameter O

Lens With focal length f

Distant Source with area S

uire'1

051

73

ARNICA II_1I

(7)

74

7//0)

-P~

~~

~~

~~

~~

~c

Ln,

a)a

/ /

/,

·-

(1) O

_

LL c

1 L

a, r,

,ra/

r c,

I00,

~~~L0

-u~~~40c

4LL

~

~~

~~

~~

~(d

O

I-- D cn

~~~~~~~~~~~~~~~~~ -(..0D

uqu

/// /-./,-/

0)~

~

)O

C\J

4-3

a) ~

III0

oo

OO

0

(Am) @5eqIO^ qnd~n0

u

// 1/'

" o;///,

L0* LOT-

4'4·e 1··0·

> c

L

/ /."/

//'CO ~

~ '-"

-O+

1

·-

tm

c-(A

UJ)~

~~

~~

~~

~~

~~

~~

O

.rO{0 r,~n

> ~.

· C

# 0

.~4 Q

,,,o , o

'8

o~.

U) N

U)

@ n I/n

- -H

I -i..---

O.

r- 0o

4 >

E

'0o 0Q

E

W

00I-.~

-__O

---0+

-4-0)0)

2S~osadWfl~U Ew

flT~

q\

00W

OE

(MA/VA

r

ARNICA II-1

051

(8)

75

u m

m -H

T

3 C(_

O

-H

•H

r-H

"D

Q_

I E

Q

[0

-P

O

13

J= C

Q

- (0

1—1

[0 o

•H

-P

a a •P

.C

en

c m

r-l

H id

o

o

LL

(H

a) -p

i-t •H

4

-1H

a) TS

C

03

-P

X

03

03

_Q

[0 r

•H

-P

U

D3

(0 C

>

03

-p

(0 Q

O

0

-

Figure 3 : Detector components

051

ARNICA II-1

(9)

AR

NIC

A

II-1

76

051

77

LIST OF ILLUSTRATIONS

PagePART I:

- Flow chart for the ARNICA program 4

PART II

Chapter 1.

- Homogeneous texture areas ...... 14- ERTS 1 Document interpretation .. 18- Phytogeographic profile ......... 19

Chapter 2.

- Aran Valley: ERTS trajectory ... 30- Vegetation map of Aran Valley .... 36- ERTS data on Aran Valley ....... 37- Diagram of multiband relation .... 38- Procedure for selected homogeneous

sites ........ ................. 46- Determination procedure ......... 47- Example of principal groups ...... 48- Illustration of bispectral diagram .. 49- Simulation of analytic cartographyof fir forests ................... 52

Chapter 3.

- Flow chart ................. ..... 60- Software ............ .. .... .. 61- Densitometric reponses .......... 62- Optical density histogram ........ 63- Example of a 10-level gray-scale .. 64- Localization coding .. .......... 65- Automatic cartography ........... 66- Visual simulation .-............ 67

Chapter 4.

- Equipment diagram .............. 73- UDT-500 Photodiode ........*...* 74- Detector components ......... 75- The experimental system .*[***.... 76

ARNICA II- 1051