Hydrological modelling for post-monsoon agricultural drought assessment and implications for the agro-economy

Reliable drought monitoring is a prerequisite for minimizing potential agricultural losses. Soil moisture is a key variable for monitoring agricultural drought assessment. This study conducted in the Bundelkhand region of Uttar Pradesh uses a macroscale variable infiltration capacity (VIC) hydrological model to simulate soil moisture and calculate soil moisture deficit index (SMDI) for agricultural drought assessment for the Rabi crop growing season, 1998-2016. Crop yield was linked with SMDI and other covariates using a random forest machine learning-based regression technique. The results show that the VIC model effectively simulated root zone soil moisture when compared with the reference data. Major droughts were identified in the years 2000-01, 2007-08, and 2015-16 in the study region. The RF-based crop yield prediction accuracy improved when irrigational factors were added. The study provides a noteworthy reference for drought assessment and prevention, water resource management, and ensuring food security.