Construction of an Integrated Drought Monitoring Model Based on Deep Learning Algorithms

Drought is one of the major global natural disasters, and appropriate monitoring systems are essential to reveal drought trends. In this regard, deep learning a very promising approach for characterizing non-linear nature factors. We used multi-source remote sensing data such as Moderate Resolution Imaging Spectroradiometer (MODIS) Climate Hazards Group Infrared Precipitation with Station (CHIRPS) integrate impact factors precipitation, vegetation, temperature, soil moisture. The application convolutional long short-term memory (ConvLSTM) construct an integrated model was proposed tested, using Xinjiang Uygur Autonomous Region example. To better compare performance ConvLSTM models, three other classical models machine were also comparison. results show that composite index (CDI) output by had consistent high correlation rating multi-scale standardized precipitation evapotranspiration (SPEI). coefficients between CDI (SPI) all above 0.5 (p < 0.01), which highly significant, coefficient CDI-1 monthly relative humidity at 10 cm depth 0.45 well correlated. addition, spatial distribution CDI-6 simulated correlated degree expressed SPEI-6 observations stations. This study provides new regional monitoring.