The research aims to develop and evaluate machine learning models to predict the intensity of drought in the United States using meteorological data collected from satellites. It is assumed that machine learning models trained on meteorological data can predict the intensity of drought with high accuracy. Two models were used to predict drought: CART (Classification and Regression Trees) and random forest. The data was preprocessed, including filling in the missing values with the median method and normalization. Both models showed good classification accuracy. The random forest model achieved 94.2% accuracy, and the CART model achieved 93% accuracy. The analysis of the importance of the signs revealed that the pressure on the surface, the temperature range at an altitude of 2 m and the temperature of the earth’s surface are the key meteorological parameters affecting the intensity of drought. This study demonstrates that machine learning models can be used to accurately predict drought, which makes it possible to develop more effective strategies for managing water resources and adapting to climate change, especially in agriculture.