Impact of Climate Change on Non-stationarity of Extreme Streamflows in Godavari River Basin, India

Abstract

The frequency analysis and risk assessment of extreme precipitation and streamflow at basin scales are essential for effective hydrologic design and water resource management activities. Processes that generate streamflow are influenced by both catchment characteristics and environmental conditions. As a result of climate change, human-induced changes in land use patterns (urbanization, deforestation, encroachment of flood plains), and faulty reservoir operations, the stationarity of streamflow assumption is questionable in flood frequency analysis. The changing climate has continued to alter the intensity, duration, and frequency of extreme events in the region, and the current status of climate change impacts on hydrology calls for the evaluation of a non-stationarity approach for extremes to enhance effective planning. This study aims to investigate the non-stationarity of streamflow and hydrologic sensitivity of catchments of the Godavari River basin located in peninsular India to changing climatic circumstances using a multi-model ensemble based on CMIP6 climate models. Firstly, Mann Kendall (MK) test is performed to detect the presence of temporal trends in the observed annual maximum streamflow series at 14 gauging stations distributed across the basin. Then peak flow series are analyzed using stationary and non-stationary models assuming invariant shape parameters and linear functions as location and scale parameters with time as a covariate. A generalized extreme value (GEV) distribution coupled with downscaled climate projections is employed to assess the probability distribution of extreme events. Results show that only 2 out of 14 streamflow series show temporal trends, suggesting that using physically based covariates instead of time can provide a better fitting. The return period can be shortened by more than one-tenth of its length, and flood risk is projected to increase significantly between the historical and future periods. These findings provide insights into non-stationary extreme streamflow behaviour, emphasizing the importance of identifying dominant drivers for changes in flooding under climate change.