Convective Clouds Extraction from Insat3d Satellite Images

Abstract

Convective clouds are well-known for bringing torrential rainfall, high winds, and other bad thunderstorms. Early identification of convection can significantly minimize loss. The auto extraction of convective clouds is crucial. To tackle this situation, we collect large-scale cloud data from remote sensing images that allow us to detect convective clouds in various methods. The overall process starts with INSAT-3D satellite imagery as an input. The system then processes and analyzes it with the past records to understand the presence of Convective clouds. There are various critical applications which require the use of this. One major day-to-day use is for airplanes. Convective clouds can cost human lives if gotten in contact with airplanes. Thus, early detection here can help save human lives and material costs. Many approaches for studying clouds have been proposed. Different methods in use can produce different results and efficiency of the solution may differ with each method. We proposed a model based on thresholding technique for the segmentation and GAN architecture for extracting the convective clouds from the satellite images.