Classification of drought intensity in mountainous areas based on meteorological data from ground stations and aerosols using ML

Introduction. The article explores the possibility of modeling drought intensity classification in mountainous regions using meteorological data collected from ground stations and aero - probes. The problem of drought in these regions is critical due to climate change, making water resource management crucial. Materials and methods. The study used data from NASA POWER and USDM drought intensity indices. Machine learning algorithms, such as Random Forest and Decision Tree, were applied to meteorological data from ground stations and aero - probes. Results. The analysis revealed that ground station data, including temperature and pressure, were more relevant for predicting drought intensity. Models built on these data demonstrated higher accuracy compared to those using aero - probe data. Discussion. Key parameters such as ground - level pressure and temperature contributed the most to the prediction. Aero - probe data were useful for monitoring wind speed at high altitudes but had lower significance for drought intensity. Conclusion. For drought modeling, ground station data are preferable. Aero - probe data can be beneficial for more specialized tasks related to high - altitude wind monitoring. Resume. This study highlights the importance of selecting appropriate data sources for drought forecasting and confirms the relevance of ground station data in mountainous regions. Suggestions for practical applications and directions for future research. The results can be applied to develop more accurate drought forecasting models and to create hybrid models that utilize data from various sources.