Delineation of Convective and Stratiform Rainy Clouds Based on Their Spectral and Textural Features on Meteosat Data

Abstract

In the present study two schemes were developed for the delineation of convective and stratiform clouds based on the high spectral resolution of the Meteosat Second Generation (MSG). Two classification methods were proposed that use spectral cloud parameters along with textural cloud parameters. The first model is an empirical algorithm based on the estimation of the probability of convective rainfall (PCR) for each pixel of the satellite data and the second is a statistical approach (Artificial Neural Network, ANN) based on the correlation of spectral and textural parameters with convective and stratiform rain. It was found that the introduction of textural parameters as additional information tends to improve the discrimination between convective and stratiform clouds for the models in the training and validation dataset. The PCR algorithm based on spectral and textural parameters shows the best performance among all the rain classification models for the training dataset. When evaluating against the independent dataset, the ANN model based on both spectral and textural parameters produces scores significantly better than the other rain classification algorithms. All algorithms overestimate the convective rain occurrences detected by the rain stations network.