Deep Learning for Monitoring Agricultural Drought in South Asia Using Remote Sensing Data

Drought, a climate-related disaster impacting variety of sectors, poses challenges for millions people in South Asia. Accurate and complete drought information with proper monitoring system is very important revealing the complex nature its associated factors. In this regard, deep learning promising approach delineating non-linear characteristics Therefore, study aims to monitor by employing remote sensing data over Asia from 2001–2016. We considered precipitation, vegetation, soil factors forwarded neural network (DFNN) as model input parameters. The evaluated agricultural using moisture deficit index (SMDI) response variable during three crop phenology stages. For better comparison performance, we adopted two machine models, distributed random forest (DRF) gradient boosting (GBM). Results show that DFNN outperformed other models SMDI prediction. Furthermore, results indicated captured pattern high spatial variability across penology Additionally, showed good stability cross-validated training phase, estimated had correlation coefficient R2 ranges 0.57~0.90, 0.52~0.94, 0.49~0.82 start season (SOS), length (LOS), end (EOS) respectively. between inter-annual in-situ SPEI (standardized precipitation evapotranspiration index) was almost similar SPEI. provides comprehensive producing consistent distribution which establishes applicability monitoring.