Automated detection of dust clouds and sources in NOAA-AVHRR satellite imagery

In this paper, we introduce a new method to detect dust clouds on NOAA-AVHRR satellite images. The approach involves the use of the region growing segmentation algorithm. The region growing starts with a set of seed points obtained from the band math image, which is known to be a good indicator of dust clouds due to physical reasons, and the result will be a binary image that represents the improved dust cloud region estimate. Furthermore, we also propose a method to locate dust sources automatically. The method developed here uses corner detection and other processing applied to the boundary of the detected dust cloud region. The approaches are applied to NOAA-AVHRR satellite images from dust storm events in southwestern North America.