Auto-detection of Rain Clouds in Remote Sensing Images

The goal of our study is to create classification methods and to be applied on images gathered by meteorological remote sensing equipment. The expected effects are to boost the efficiency of these kinds of equipment to automatically identify rainy cloud cells and to extract information about precipitation describe (describing) them, and forecast how they will change throughout time. To do this, we have chosen to work with high resolution Meteosat Second Generation (MSG2) satellite images from the HRV channel representing North Africa with radar images collected in the same studied area. In this work, we used the 2D Fourier transform for the autodetection of unwanted ground echoes present on the radar images and for the automatic discrimination between the rainy clouds or not on the satellite images. We also used the textural parameters for the detection of rain cells on HRV images. The use of this method to detect rainy cells has given good results and indicates a very interesting performance.