A Non-stationary Hydrologic Drought Index Using Large-Scale Climate Indices as Covariates

Abstract

The dry and wet periods can be analyzed based on different drought indices. Most existing drought studies are based on stationary assumptions, and environmental changes are not considered. This study proposes a non-stationary streamflow-based drought index, incorporating large-scale climate indices to study hydrological drought for 45 years. Climate indices are used as covariates for building the non-stationary model fitted to streamflow. Correlation analysis is carried out to determine the best covariates for the streamflow in the Netravati River basin in India. The Southern Oscillation Index (SOI) exhibited a significant influence on streamflow at all time scales. The non-stationary model is compared with the stationary model, and the best model is chosen based on the Akaike information criterion (AIC). Under statistical measures, non-stationary models performed better than stationary ones at all time scales. The generalized additive model for location, scale, and shape (GAMLSS) is used for non-stationary modeling. The models are developed for short-term (3 and 6 months) and long-term (12 and 24 months) droughts. The influence of climate variables on drought classes is analyzed, and more severe drought is observed under the non-stationary scenario. The deficiency in streamflow was more than 60% in the basin in 1987 and 2002. The non-stationary drought index detected more severe drought events than the stationary index under short-term scales. Hydrological drought properties such as drought severity, duration, and peak are calculated under stationary and non-stationary scenarios, and a noticeable difference is observed. Compared to stationary models, the non-stationary model yields more logical and satisfactory findings because it effectively takes into account non-stationarities in the streamflow caused by climate change. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.