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

1.6 Principal Results
The principle results of this dissertation include developing an algorithm based on
multi-spectral technique for detecting smoke and dust aerosols combining MODIS RSBs
and TEBs; discussing the feasibility of dust detection with multi-sensor measurements;
and assessing the impact of sensor calibration and characterization on detection results.
1) The multi-threshold algorithm for automatically detecting smoke and dust
aerosols is developed by combining measurements of both MODIS RSBs and TEBs. The
spectral curves of various scene types are derived statistically from large amount of
training dataset collected at the selected areas within last several years. With the spectral
and statistical analyses, the appropriate bands are selected and proper thresholds are
decided for eliminating unwanted pixels step by step. The algorithm is applied mainly to
detect smoke plumes in USA and dust storms in Asia. The results are validated not only
with MODIS true color images but also with products of OMI and CALIPSO, showing
that the algorithm works well in the given areas with small errors. Pairs of measurements
from both Terra and Aqua MODIS in consecutive days give the basic dynamic
information about smoke and dust, which are helpful for estimating the spread direction,
and magnitude change of these two types of aerosols.
2) The tentative experiment for detecting dust aerosol with multi-sensor, CALIPSO
and MODIS, is performed. Since both sensors are operated in the same orbit with similar
local equatorial crossing time, the temporal mis-registration is ignored in this dissertation.
With the spatial registration, the dust aerosol is accurately identified by jointing the
vertical information of CALIPSO measurements with MODIS thermal emissive
8