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 ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
values <strong>of</strong> dark surface is smallest, generally less than 20 K. In Figs. 4.4d <strong>and</strong> 4.4f, the<br />
logarithm <strong>of</strong> single reflectance <strong>of</strong> red b<strong>and</strong> is introduced to separate dust from surface<br />
scenes. <strong>Using</strong> logarithm can enhances the resolving power <strong>of</strong> this index, especially in<br />
small reflectance range. Obviously, the Ln (R1) <strong>of</strong> most dust pixels is larger than -1.2<br />
over bright surface, <strong>and</strong> -1.6 over dark surface.<br />
The detail thresholds <strong>of</strong> each index are summarized in the Table 4.4. The thresholds<br />
are calculated statistically from all training data, namely selected dust events in China<br />
during 2001-2007. The errors caused by using the defined thresholds <strong>of</strong> each test are also<br />
listed in Table 4.4. The largest error is appeared at the separation <strong>of</strong> dust from cloud over<br />
dark surface in reflective solar spectrum, which is up to 12.484%. This big error caused<br />
possibly by light dust pixels suspended over water. Considering that some pixels can be<br />
counted repeatedly, the total errors <strong>of</strong> both over bright <strong>and</strong> dark surfaces are not equal to<br />
the summation <strong>of</strong> errors from each test.<br />
4.2.2 Flowchart<br />
The flowchart <strong>of</strong> dust detection algorithm is shown in Fig. 4.5. The L1B<br />
measurements are input into cloud module to filter out cloud first by employing BTD (12,<br />
11) <strong>and</strong> NDDI. The remaining pixels are then divided into two branches based on<br />
pre-stored surface type information: dust over bright surface <strong>and</strong> over dark surface. The<br />
BTD (3.7, 11) <strong>and</strong> the logarithm <strong>of</strong> reflectance at b<strong>and</strong> 1 are applied in both branches<br />
with respective thresholds. At the end, a same “noise-elimination” process is executed to<br />
filter out discrete pixels.<br />
54