Cloud bursts are significant natural hazards that can release vast volumes of rain and may lead to flash floods in the rivers of affected regions. This phenomenon often results in catastrophic events, especially in mountainous and hilly areas. Rapid changes in weather necessitate a fast and efficient weather prediction system that can provide early warnings, allowing for prompt action to be taken. This study refers to a machine learning-based approach to identify cloudbursts from satellite imagery. This process leverages texture-based image features (Haralick features) and machine learning algorithms to classify cloudburst events or not. First, we collect satellite images of various cloudbursts that have occurred over India in recent years, along with images of clouds that did not result in such events. After preprocessing these images, Haralick features are extracted. These features are then evaluated using various tree-based machine learning classifiers. It has been observed that the Random Forest classifier performs better than other tree-based classifiers in predicting cloud bursts.