METEOSAT SECOND GENERATION ( MSG ) : CAPABILITIES AND APPLICATIONS

In this paper an algorithm to predict the short-term evolution of cloud formations with high rainfall probability by using image sequences coming from meteorological satellites (Meteosat images) is described. The proposed algorithm consists of four steps: the first step performs image processing procedures (thresholding and relaxation, edge following, frequency filtering) adapted to a specific environmental application; the second step is dedicated to solve the correspondence problem in different images while the third one deals with the problem of modelling the parameters defining the time evolution of a cloud formation and investigates how these parameters can be estimated. The last step concerns the motion prediction of processed clouds. The main goals of this work are to test the effectiveness of the proposed procedure with different values of the filtering parameters, and to verify the importance and the influence of the mentioned parameters in the prediction mechanism. Some experimental results obtained by using real image sequences coming from Meteosat satellites are shown.