Abstract:
Every year a large amount of aerosols released from wildfires and dust storms into
the atmosphere may have potential impacts on the climate, environment, and air quality.
Detecting smoke and dust aerosols and monitoring their movements and evolutions in a
timely manner is a very meaningful task. Satellite remote sensing has been demonstrated
as an effective means for the observation of these two kinds of aerosols. In this
dissertation, an algorithm based on the multi-spectral technique for detecting smoke and
dust aerosols is developed, by combining measurements of MODoderate resolution
Imaging Spectroradiometer (MODIS) reflective solar bands and thermal emissive bands.
Data from smoke/dust events occurred during last several years are collected
visually as training data for spectral and statistical analyses. According to the spectral
curves of various scene types (aerosols, cloud, vegetation, and water et al.), a series of
spectral bands is selected jointly or separately and corresponding thresholds are defined
for scene classification step by step. The multi-spectral algorithm is applied mainly to
detect smoke plumes in the United States and dust storms in Asia. The detection results
are validated not only visually with MODIS true color images, but also quantitatively
with products of, Ozone Monitoring Instrument (OMI) and Cloud-Aerosol Lidar and
Infrared Pathfinder Satellite Observation (CALIPSO). The validations show that this
multi-spectral detection algorithm is suitable to monitor smoke and dust in the selected
study areas. The accuracy is quite good in most cases. Additionally, this algorithm can be
used to detect smoke and dust aerosols at the areas near clouds even mixed with clouds.
Detection of dust aerosol with multi-sensor satellite remote sensing measurements,
MODIS and CALIPSO, is also performed tentatively in this dissertation. After spatial
registration, the dust layers are identified combining CALIPSO Vertical Feature Mask
product and measurements of MODIS brightness temperature difference between 12 and
11-μm bands. Based on detecting results, the three-dimension information of dust
aerosols is summarized.
Additionally, the impacts of the mis-registration on the L1B data and dust aerosol
detection results are assessed. The relative errors caused by mis-registration on L1B data
are generally less than a few tenths of a percent. The impacts on dust detection results are
relative large, usually has the trend as negligible at the homogeneous and
semi-homogeneous areas, but large at the non-homogeneous areas.