Meteorological Drought Prediction Using Long Short-Term Memory (LSTM) Model

The world is facing several natural disasters due to a drastic change in the global climate. Among all natural disasters, drought is the most devastating on the planet, directly affecting habitats and ecosystems. On the other hand, it has a direct impact on the environment. The meteorological drought monitoring and forecasting using Automatic Weather Station (AWS) is complex and error-prone due to limited spatial coverage. Therefore, in the present study, we have utilised satellite-based Moderate Resolution Imaging Spectroradiometer (MODIS) products, specifically MOD11A1 (Terra) and MYD11A1 (Aqua), as well as CHIRPS (Climate Hazards Group InfraRed Precipitation with Station data), for temperature, humidity, and precipitation analysis, respectively. The Google Earth Engine (GEE) and Google Collab were used to execute the LSTM model. Land Surface Temperature (LST) data reported an R 2 value of 0.91 for maximum temperature, 0.90 for humidity and 0.93 for precipitation analysis. This studies results can help in long-term meteorological drought management and planning for climate adaptation. The information could support policymakers, farmers, and water managers in reducing climate-related risks in the Gangapur region.