FR3.TO5.4.pptx

RETRIEVING BRDF OF DESERT USING TIME SERIES OF MODIS IMAGERY

Haixia Huang, Bo Zhong, Qinhuo Liu, and Lin Sun

Presented by Bo Zhong

[email protected]

Institute of Remote Sensing Applications, Chinese Academy of Sciences

IGRSS 2011, Vancouver , Canda

Outline

Background

Methodology

Preliminary results

Applicatoins

Conclusions

Background

BRDF is the key parameter for:

Quantitative remote sensing

Erath radiation budget

More

Desert is one of the main landcover types

Strongly reflecting the solar radiation

More

Problem

• There is no “good” BRDF product of desert

Methodology-flowchartMODIS imagery

Converting DN to TOA reflectance

Identifying the “clearest” of each

observations

Retrieving reflectance of “clearest”

observations

Fitting to Staylor-Suttles BRDF

model

Lookup Tables

BRDF of desert

Methodology- site choosing

Location of the experimental site(MODIS imagery color composite)

Cole view of the site(TM imagery color composite)

Methodology- site choosing

It is stable, so it can be seen as an invariant

object;

There are a lot of lakes within the calibration

site, which are seldom polluted, so the lowest

AOD of calibration site can be determined by

Dark Object (DO) method using Landsat TM

and ETM+ data.

8

（a）Mar. 3, 2000 （b）Feb. 3, 2010

Methodology- site choosing

AOD retrieval using DO method

40 60 80 1000

1

2

3

Aerosol

opti

cal

depth

Apparent radiance(w/m2/sr/mic)

Aerosol optical depth and radiance curve

Corresponding path radiance and aerosol optical depth

52.87

0.43

ETM+

imaging

date

Aerosol

optical

depth

TM

imaging

date

Aerosol

optical

depth

2000.03.03 0.1543 2006.09.20 0

2000.04.29 0 2006.10.31 0.0440

2001.10.16 0.0218 2007.05.18 0.1536

2001.11.17 0 2007.06.03 0

2002.01.04 0.0657 2009.06.17 0.4279

2002.03.18 0.2801 2009.08.11 0.05444

2002.05.28 0.3274 2009.08.27 0.0043

2002.09.17 0.1058 2009.09.28 0.0096

2002.11.04 0.3633 2010.02.03 0.07676

2002.11.13 0.0286 2010.06.04 0.254

2002.12.15 0.1195 2010.07.29 0.3552

2003.03.28 0.4273 2010.08.14 0

2010.08.23 0.2178

Original method

Time series of

MODIS imagery

Identifying clear

pixels

Reflectance of

clear pixels

BRDF fitting

Reflectance of

hazy pixels

AOD of hazy

pixels

LUT

MODIS surface

reflectance

Modifications for the original method

AOD determination for the “clearest” days;

Shrinking the use of the algorithm from globe to the

desert calibration site, which is stable;

Identifying the “clearest” observations for every 10

degrees in view zenith angles from 0-50 degree (0-10, 11-

20, 21-30, 31-40, and 41-50);

Using Staylor-Suttles BRDF model instead of Walthall BRDF.

MODIS-B3: Staylor-Suttles coefficients

Preliminary results

MODIS-B1: Staylor-Suttles coefficients

MODIS-B2: Staylor-Suttles coefficients

Comparison with MODIS products

R2 much higher

RMSE is lower

Applications I: inter-calibration of AVHRR using retrieve BRDF

Spectral matching of AVHRR and MODIS

AVHRR data simulation using the new method

Inter-calibration

Validation

Spectral matching

AVHRR 1 (0.645 μm) AVHRR 2 (0.865 μm) AVHRR 3 (1.6 μm)

ai 0.9885 1.0105 1.0004

Applications II: global desert BRDF retrieval

Mapping of the desert

BRDF and AOD retrieval simultaneously using

the new method

Preliminary validation

The chosen desert sites

The geolocations of the deserts

Desert

Name

Lat（°） Lon（°） Altitude

（m）

Duration

(yyyy.mm.d)

Taklimakan 39.0°N-40°N 84°E-85°E 1050 2009.10.1-

210.10.1

Rabal-

Khali

18.8°N-19.8°N 45.5°E-46.5°E 700 2009.10.1-

210.10.1

Lybia 24°N-25°N 12°E-13°E 740 2009.10.1-

210.10.1

Sahara 19.5°N-20.5°N 8°W-9°W 260 2009.10.1-

210.10.1

Taklimakan desert

Rabal-Khali desert

Lybia desert

Sahara desert

Conclusions

The new method is able to catch the BRDF

characterization of deserts

This method can be used for inter-calibration of

reflective bands of moderate satellite data like

AVHRR

This method is helpful for researches on earth

radiation budget

Thank you for your attention!