Impact of climate change on streamflow regime of a large Indian river basin using a novel monthly hybrid bias correction technique and a conceptual modeling framework

Abstract

The study analyses the impact of climate change on streamflow regime of the Mahanadi River basin (MRB) in projected climate scenarios obtained from CMIP5 models. Projections from a total of 9 Global Climate Models (GCMs) were utilized. Integrated MIKE 11 NAM-HD, a conceptual hydrological model was employed to simulate the streamflow at Hirakud and Mundali gauging sites of upper Mahanadi River basin (UMRB) and middle Mahanadi River basin (MMRB), respectively. Prior to generating the streamflow regimes for the projected scenarios, GCM simulated precipitation and temperature were bias corrected using India Meteorological Department (IMD) observed gridded precipitation and temperature datasets. Precipitation was corrected using quantile mapping technique employing four different approaches, i.e., seasonal, hybrid, monthly non-hybrid, and monthly hybrid approaches. Hybrid approaches explicitly correct the extreme rainfall using Generalized Extreme Value (GEV) distribution whereas Gamma distribution is used for normal rainfall. Temperature data was corrected using Gaussian distribution on daily basis employing monthwise time series. L-moments based frequency analysis method was used to estimate the 50-year return period rainfall for observed and bias corrected projected time series to assess the efficacy of bias correction methods in correcting the extreme events. Monthly hybrid approach was found to resolve rainfall climatology over both UMRB and MMRB with improved skill in extreme rainfall correction. Though, monthly non-hybrid approach produced the seasonal cycle of rainfall with highest accuracy, it performed poorly in correcting the extremes. Under projected climate scenarios mean annual rainfall is found to be increasing for BCC-CSM1.1(m), HadGEM2-AO, GFDL-CM3, and IPSL-CM5A-LR whereas other GCMs show mixed signal compared to their mean state of historic period (1976–2005). Streamflow regime in projected climate was analyzed using ensemble mean of simulated streamflow from identified GCMs. Mean monthly streamflow were overall found to be increasing towards the end of the century i.e., in far-future defined as 2070–2099. Daily high flows (defined at 5% and 10% exceedance probability) were found to be increasing in their magnitude and frequency in far-future whereas occurrences of low flows (defined at 90% and 95% exceedance probability) were found to be predominantly decreasing under projected climate.