Decision tree classification model for detecting and tracking precipitating objects from series of meteorological images

Abstract

Accurate detection and identification of convective (cumulonimbus) clouds, which are potentially precipitating objects, as well as tracking cloud movement, are important tasks to locate and predict precipitation. In the present work, a Decision Tree classification model was used to locate and track precipitating objects from series of GOES-13 meteorological image sub-scenes covering the territory of Colombia, located to the northwest corner of South America. Results show that it is possible to infer a classification model that can be used repeatedly for accurately locating and tracking precipitating objects from multispectral meteorological images.This tree was obtained using the Gini impurity measure, selecting thermal-infrared (thermal-IR) and water vapour (WV) as the most