Nader Jalali; Fazel Iranmanesh; Mohammadhadi Davoodi
Abstract
Generally, satellite images and climatological and environmental data are used simultaneously in dust storm studies. Most of the processing techniques are limited to a few indices such as brightness temperature differences, normalized dust density index, aerosol optical depth, false color composite and ...
Read More
Generally, satellite images and climatological and environmental data are used simultaneously in dust storm studies. Most of the processing techniques are limited to a few indices such as brightness temperature differences, normalized dust density index, aerosol optical depth, false color composite and visual image interpretation. Since the reliability of the above-mentioned methods are varied, so an attempt was made to increase the reliability of automatic spatial image clustering by contributing the expert knowledge via visual image interpretation. Therefore, the required satellite images were collected for a period of 2005 to 2008, in Mesopotamian, Syria and south-western provinces of Iran. For doing this, the MODIS color composite images were decomposed into their ordinary bands and then used for further analysis. In this research, spatial image clustering followed by visual image interpretation (hybrid classification) was applied for accurate mapping of geographical extents of dust storm with different intensities. The sources of dust storms were also identified through visual image interpretation by considering the interpretation criteria such as shapes and patterns. Comparison of the previous results and obtained by this research, indicated critical conditions in terms of dust storm occurrence in the region. Image clustering and hybrid classification led to daily dust persistence over the region and then those images compiled in order to prepare annual total heavy dust persistence map. The result of using brightness temperature difference method verified similarity and reliability of obtained results. Therefore, this simple method could be proposed for identification of dust sources, related plumes and their affecting areas, when raw image data are not available.
