Investigating the performance of dust detection indices using MODIS data and products (Case study: Khuzestan province of Iran)

Abstract

In areas with a semi-arid or arid climates, dust storms are caused by winds blowing on the surfaces with loose and dry soils. Dust storms can influence different aspects of human life, such as health, agricultural practices, urban, rural, and transportation infrastructures. Since 15 years ago, dust storms, as one of the leading environmental hazards, have occurred with increased frequency, spatial extent, and intensity in the Middle East. Several satellite-based dust-detection algorithms are introduced for identifying dust emission sources and dust plumes when rising in the atmosphere. In this research, four common algorithms, namely Brightness Temperature Difference, Normalized Difference Dust Index, Thermal-Infrared Dust Index, and D-parameter, were evaluated using MODIS Level 1B and MODIS Deep Blue AOD products in two dust storms in the Khuzestan province, Iran. Detection thresholds for the indices was derived by a comparison of dust-present versus dust-free conditions data considering different coverage of land and inspecting related periods. The detection proficiency of the algorithms was different for various events; thus previously obtained thresholds were not applicable in the algorithms performed in the Khuzestan region. Initially, dust was effectively and adequately detected by MODIS AOD. It was also revealed that MODIS thermal infrared (TIR) band algorithms or algorithms combining TIR and reflectance bands could detect dust better than reflectance-based ones. However, some commission errors were caused by substantial differences among their susceptibility to distinguish dust, cloud, and the surface. Overall, among the algorithms, TDI and D-parameter performed the best over dust sources in the Khuzestan province.