A NOVEL METHOD FOR WEATHER NOWCASTING BASED ON SPATIAL COMPLEX FUZZY INFERENCE WITH MULTIPLE BAND INPUT DATA

Abstract

The prediction of weather changes, such as rainfall, clouds, floods, and storms, is critical in weather forecasting. There are several sources of input data for this purpose, including radar and observational data, but satellite remote sensing images are the most commonly used due to their ease of collection. In this paper, we present a novel method for weather nowcasting based on Mamdani complex fuzzy inference with multiple band input data. The proposed approach splits the process into two parts: the first part converts the multiple band satellite images into real and imaginary parts to facilitate the rule process, and the second part uses the Spatial CFIS+ algorithm to generate the predicted weather state, taking into account factors such as cloud, wind, and temperature. The use of MapReduce helps to speed up the algorithm's performance. Our experimental results show that this new method outperforms other relevant methods and demonstrates improved prediction accuracy.