CoRP Symposium, 10-11August 2010, Fort Collins, CO 1 A daytime multispectral technique for detecting supercooled liquid water- topped mixed-phase clouds

CoRP Symposium, 10-11August 2010, Fort Collins, CO 1

A daytime multispectral technique for A daytime multispectral technique for detecting supercooled liquid water-detecting supercooled liquid water-

topped mixed-phase cloudstopped mixed-phase clouds

Yoo-Jeong NohYoo-Jeong NohCooperative Institute for Research in the Atmosphere /

Colorado State Universitywith

Steven D. Miller Steven D. Miller (CIRA/Colorado State University)Andrew K. Heidinger Andrew K. Heidinger (NOAA/NESDIS)

CIRA


Optically Opaque Mixed-Phase

Region (~300-500 m deep)

Precipitating Ice Region

(~0.2-2.5 km deep)

Generating Cells ~ 1-1.5 km in Length

Ice

Mixed-Phase Clouds

Significant in-flight icing hazard!

Supercooled

Liquid Water

MotivationMotivation


ObjectivesObjectives Scientific: Understand spectral reflectance

characteristics of supercooled liquid water-topped mixed-phase clouds via radiative model simulations in near IR channels

Application: Develop a multispectral satellite detection algorithm for supercooled liquid water-topped mixed-phase clouds

Operational Utility: An objective method for identifying a subset of areas where significant aircraft icing conditions may not be present through a significant depth of cloud, given a widespread field of super-cooled liquid clouds.


HypothesisHypothesis

3.0

5.55.0

km

Liquid (_liquid)

Ice (_ice)

1.6 μm

R(1.6)

2.2 μm

R(2.2)

R(2.2)/R(1.6) for a supercooled liquid

top and ice bottom cloud

R(2.2)/R(1.6) for a pristine liquid cloud

>

2.2 μm

Differential absorption properties between the liquid and ice in the near infrared Differential absorption properties between the liquid and ice in the near infrared

3.0

5.55.0

km

Liquid (_liquid)

Liquid (_liquid)

1.6 μm

R(1.6) R(2.2)

Assuming ‘all else being equal’ besides the phasephase of the cloud particles…

less reflectance more reflectance


R_COMP R_COMP We define a liquid-normalized reflectance ratio

)6.1(_/)1.2(_

)6.1(_/)1.2(__

msimRmsimR

mobsRmobsRCOMPR

Simulated for pure-liquidSimulated for pure-liquid

ObservedObserved

With stronger absorption by ice particles at 1.6 m, we expect the numerator term of R_COMP to exceed the denominator term in the case of liquid-over-ice clouds, such that R_COMP 1.


a-priori database(constructed using SBDART)

Using MOD021KM data, compute OBS Reflectance RatioR_OBS=R_obs(2.1μm)/R_obs(1.6μm)

Using MODIS optical thickness andeffective radius, for a all-liquid cloud in the database, computeR_SIM=R_sim(2.1μm)/R_sim(1.6μm)

R_COMP=R_OBS / R_SIM

• MODIS IR Cloud Phase improved by A. Heidinger• OT* : a minimum optical thickness to be detected

(a function of cloud top effective radius)• R*_COMP : a threshold for the SLW topped pixel

Liquid or Mixed phase &T_cloud_top < 273 K &Optical thickness ≥ OT*

R_COMP ≥ R*_COMP

MODIS Level 2 data:Cloud Phase, T_cloud_top,

Optical thickness, Effective radius



Flag a likely Flag a likely liquid topped liquid topped

mixed-phase pixelmixed-phase pixel



Schematics of our detection algorithmSchematics of our detection algorithm


Radiative Transfer Simulation Radiative Transfer Simulation in the Near-Infraredin the Near-Infrared

SBDART (Santa Barbara DISORT Atmospheric Radiative Transfer) model used

Compare with Terra MODIS data (GOES-R ABI in the future) Sensitivity tests for several variables A-prioriA-priori database generation database generation for idealized cloud layers

layer1: 3-5km (ice bottom), layer2: 5-5.5km (liquid top) Liquid optical thickness = 0~30 for total optical thickness = 0~30 Liquid sizes = 6, 8, 10, 12, 15, 20 μm when ice = 30 μm Ice sizes = 30, 50, 70, 100, 120 μm when liquid = 8 μm Sensor/Solar zenith angle = 0~80° Sensor azimuth angle = 0~170° Ocean and vegetation surfaces Total # of data points =15,909,696


A-prioriA-priori database databaseof R_COMPof R_COMP

Total # of data points =15,909,696 with varying sun/sensor angles, liquid/ice particle sizes, and total optical thicknesses over two different types of surfaces

Sensor zenith angle

Sensor azimuth angle

Liquid droplet size(Ice particle size)

Liquid-top cloud optical thickness Total optical thickness


Comparison of Comparison of reflectance reflectance

ratios between ratios between MODIS and MODIS and

SBDARTSBDART

For MODIS Liquid cloud pixels with

T_cld_top > 283.15,Effective radii < 20,

Optical thickness > 1

1940 UTC 29 Sept 2006

1945 UTC 29 Sept 2006

1915 UTC 05 July 2005

MASE field exp case


Determine R*_COMPDetermine R*_COMP

1.005 ~ 1.010

R_COMP_SIM = R_SIM (SLW top) / R_SIM (Liquid),where R_SIM = R_sim (2.1μm) / R_sim (1.6μm)

A particular threshold, R*_COMP (>1) gives indication

of a detectable signal for liquid-over ice clouds.

A particular threshold, R*_COMP (>1) gives indication

of a detectable signal for liquid-over ice clouds.


where x=liquid droplet effective radius and y=minimum optical thickness.

Minimum Optical Thickness (OT*)Minimum Optical Thickness (OT*)

Use SBDART simulated database focusing on top liquid droplet sizes. Currently, surface types and

ice particle sizes are not considered. The impact of ice sizes can be neglected compared with liquid sizes.

using R_SIM(2.1/1.6) and _liquid=1 intersections,

Varying liquid droplet sizes

5

10

15

20

25

5 10 15 20

Effective raidus (m)

Min

imu

m _

tota

l0.135917.34189.0011.0 23 xxxy


a-priori database(constructed using SBDART)

Using MOD021KM data, compute OBS Reflectance RatioR_OBS=R_obs(2.1μm)/R_obs(1.6μm)

Using MODIS optical thickness andeffective radius, for a all-liquid cloud in the database, computeR_SIM=R_sim(2.1μm)/R_sim(1.6μm)

R_COMP=R_OBS / R_SIM

• MODIS IR Cloud Phase improved by A. Heidinger• OT* : a minimum optical thickness to be detected

(a function of cloud top effective radius)• R*_COMP : a threshold for the SLW topped pixel

Liquid or Mixed phase &T_cloud_top < 273 K &Optical thickness ≥ OT*

R_COMP ≥ R*_COMP









Schematics of our detection algorithmSchematics of our detection algorithm


Apply to Terra MODIS dataApply to Terra MODIS data


Terra MODIS L1B (MOD021KM) Data on 31 Oct. 2006Terra MODIS L1B (MOD021KM) Data on 31 Oct. 2006

Data scan started at 1625 UTC


Terra MODIS L2 (MOD06) products on 31 Oct. 2006Terra MODIS L2 (MOD06) products on 31 Oct. 2006


1625 UTC31 Oct 2006

Likely liquid topped mixed-phase pixels in red

R*_comp = 1.005

R*_comp = 1.010 R*_comp = 1.100

R*_comp Detected supercooled liquid top pixels

1.0050 766

1.0075 741

1.0100 715

1.0500 447

1.1000 251

Cyan colorCyan color means pixels having temperatures below 273K and also either water or mixed-phase (6,695 points out of total 20,571 pixels in the domain)


Preliminary Validation Preliminary Validation ExercisesExercises


C3VP/CLEX-10 Field ExperimentC3VP/CLEX-10 Field Experiment CLEX (Cloud Layer Experiment)

is a series of field experiments funded by the Department of Defense's Center for Geosciences/Atmospheric Research at CIRA/Colorado State University for non-precipitating, mid-level, mixed-phase clouds since 1996.

CLEX-10 collaborated with the Canadian CloudSat/CALIPSO Validation Project (C3VP) that took place from 31 October 2006 to 1 March 2007 over Southern Ontario and Quebec.

CARE Ground

Site

Sample CloudSat Ground track

C3VP/CLEX-10 C3VP/CLEX-10 Target regionTarget region


1625 UTC19 Jan 2007


R*_comp = 1.005

R*_comp = 1.010


1.0050 2218

1.0075 2165

1.0100 2121

1.0500 1415

1.1000 824


- 12°C


1625 UTC20 Feb 2007


R*_comp = 1.005

R*_comp = 1.010


1.0050 821

1.0075 792

1.0100 760

1.0500 491

1.1000 304


- 8°C


ConclusionsConclusions A daytime multispectral algorithm for distinguishing

between pristine liquid and liquid-topped ice clouds is in development.

The approach takes advantage of differential absorption properties between liquid and ice cloud particles in the near infrared.

The technique, applied here to Terra MODIS, is designed with an eye toward the future GOES-R Advanced Baseline Imager.

Preliminary case study results show signals near regions of observed liquid-over-ice. The algorithm fails in cases of overriding cirrus.


Future WorkFuture Work

The algorithm will be tested and validated for more cases with quantitative uncertainty estimates.

Additional constraints using various channel combinations to clearly exclude ice phase clouds will be studied.

More detailed analysis and simulations using the 2.25 μm will continue in preparation for applications to GOES-R ABI data.


THANK YOU!THANK YOU!

CIRA

Documents

CoRP Symposium, 10-11August 2010, Fort Collins, CO 1 A daytime multispectral technique for detecting supercooled liquid water- topped mixed-phase clouds