Testing single band cloud removal algorithms for SSM/I: an equatorial forest simulation approach

Recently there has been renewed interest in using low-resolution imagery from weather and ocean satellites to monitor Earth´s vegetation. The removal of cloud cover from this imagery is an important preprocessing task. This research was designed to test various heuristics for removing clouds from time sequences of 85 GHz vertically polarized SSM/I data without reference to any other SSM/1 channel. The algorithms were tested on a synthetic dataset designed to simulate the characteristics of rainfall contamination over the Amazon Basin. Nearly all of the algorithms were successful at removing cloud cover, but they differed in their sensitivity to various operator parameters