Precipitation Prediction and Drought Condition Analysis in Hulunbuir Based on the Standardized Precipitation Index and LSTM Model

This study focuses on drought events in the Hulunbuir Grassland, utilizing the Standardized Precipitation Index (SPI) and the Long Short-Term Memory (LSTM) model for analysis and prediction. Natural disasters, especially droughts, pose significant challenges to sustainable development globally. Droughts have severe impacts on agriculture, animal husbandry, water resources, and ecosystems in China, with the Hulunbuir Grassland being particularly affected. The SPI was calculated to assess drought conditions, revealing that the Hulunbuir Grassland experienced alternating dry and wet periods, with an increasing frequency of short-term extreme droughts after 2003. The LSTM model was employed to predict future precipitation, showing good performance with an RMSE of 15.6257, MAE of 9.3314, and R-squared of 0.7437. The model provided precipitation projections for the next five years (2023–2027). The study concludes that the Hulunbuir Grassland has shown a trend towards drought, and the LSTM model offers a reliable prediction for future drought development.