02.08.2013 Views

i Detection of Smoke and Dust Aerosols Using Multi-sensor Satellite ...

i Detection of Smoke and Dust Aerosols Using Multi-sensor Satellite ...

i Detection of Smoke and Dust Aerosols Using Multi-sensor Satellite ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

6.2 Validating <strong>Dust</strong> <strong>Detection</strong> with CALIPSO<br />

The CALIPSO is an effective tool for validating the dust detection. It is capable <strong>of</strong><br />

separating airborne dust from ground dust with the vertical information <strong>of</strong> aerosol <strong>and</strong><br />

cloud layers. As illustrated in Chapter five, the CALIPSO VFM classifies various scene<br />

features into several types. Therefore, the VFM product is used for the quantitative<br />

validation <strong>of</strong> dust detection results.<br />

Fig. 6.4 shows a dust storm event over Taklimakan Desert captured by Aqua<br />

MODIS on July 26, 2006. The image is generated with measurements extracted from two<br />

swaths at UTC time 7:25 <strong>and</strong> 7:30. The dust aerosol is located at the eastern desert. The<br />

blue dash line is the footprint <strong>of</strong> CALIPSO in this area. The dust image generated from<br />

multi-spectral approach is displayed in Fig. 6.5. The VFM product <strong>of</strong> the region marked<br />

by blue dash line in Fig. 6.4 is shown in Fig. 6.6. The validation is executed by matching<br />

pixels from two <strong>sensor</strong>s with their geolocation measurements in the overlapping region. It<br />

is worth noting that the misclassification <strong>of</strong> labeling dust aerosol over desert area as cloud<br />

is observed which is mentioned in Chapter five. The dust aerosol layer marked in the<br />

black circle is misclassified as cloud in Fig. 6.6.<br />

The error analysis is performed to count the number <strong>of</strong> pixels identified (pixel is<br />

labeled as dust aerosol with both <strong>sensor</strong>s), unidentified (pixel is labeled as dust aerosol<br />

with only CALIPSO but undetected with MODIS), <strong>and</strong> misidentified (pixel is labeled as<br />

non-dust aerosol pixel with CALIPSO but detected with MODIS). According to spectral<br />

curves <strong>of</strong> dust aerosol in Figs. 4.2 <strong>and</strong> 4.3, red b<strong>and</strong> is a key spectral b<strong>and</strong> able to reflect<br />

the intensity <strong>of</strong> dust aerosol loading. Usually, the stronger the dust storm is the higher<br />

87

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!