Abstract:
Drought is one of the significant natural disasters that incurs billion dollars of economic damage every year. Among all, agricultural drought needs critical attention for drought monitoring due to its direct effect on crop yield and management of irrigation water. Most of the previous studies focused on regionalizing drought using k-means, hierarchal, fuzzy, and entropy-based clustering techniques. However, these techniques are not suitable where the clusters are not separated distinctively, and the number of clusters cannot be estimated automatically. In this study, we have developed agricultural drought hotspot maps using soil moisture deficit index (smdi) and the regional severity (s), duration (d), and frequency (f) curves using complex network algorithm for the future warming climate (2041–2070) of the mahanadi river basin (mrb) in india. We have used a modified dynamic budyko (db) hydrological model to simulate daily soil moisture at a spatial scale of 0.25° × 0.25° using input from four gcms for the rcp 4.5 scenario. The modified db model was calibrated and validated for the study area. The model proved to be capable of simulating the soil moisture dynamics over the basin and also effectively captured the historical droughts occurred in the basin. The drought hotspot maps of the basin suggest that the northern, south-eastern, and central parts of the basins are going to experience more number of droughts. The results suggest that for most of the clusters, the regional s-d-f curve can be utilized to understand the future drought characteristics at site-specific as well as regional scale, as the confidence band is found to be very narrow. Overall, our study provides a framework to develop regional s-d-f curve. © 2021 elsevier b.v.