Assessment of Climate Change Impact on River Discharge using Reduced Uncertainty Ensemble Modeling Framewor

Abstract:

A reduced uncertainty ensemble modeling framework is used to analyze the impact of changing climate on discharge variations in a sub-catchment of Mahanadi River Basin in India. An ensemble of five hydrological models, comprising of one distributed physically based and four lumped conceptual hydrological models, developed using weighted average method was chosen as the best-performing ensemble, based on categorical and temporal assessment of several ensembles developed using eight hydrological models and eight ensemble methods. The member models of the chosen ensemble were then used to simulate the river discharge over 2006 – 2050, using the projected climatic data of two regional climate models (RegCM4 and HadGEM3) under two emission scenarios (RCP 4.5 and RCP 8.5). The trend analysis of the ensemble discharge using Mann Kendall test shows that monthly peak discharge and mean monthly discharge are increasing in the first and last months of the monsoon season (June and September) and decreasing in the middle two months (July and August) in case of RCP 4.5. In case of RCP 8.5, however, the monthly peak discharge and mean monthly discharge show a decreasing trend in the starting two months (June - July) and an increasing trend in the last two months. The analysis of monthly proportion of annual yield shows that there is a persistent decrease in the percent yield after monsoon to the next monsoon in case of RCP 4.5, though the condition is less serious in case of RCP 8.5 due to alternate increasing and decreasing trend in various months. The annual yield, however, is found to be decreasing and increasing in case of RCP 4.5 and RCP 8.5 respectively. We further quantified the rate of change using Sen’s slope method followed by analysis of temporal change in dependable flow at different levels under both the emission scenarios, and found that dependable flow is increasing with atmospheric CO₂ concentration level at almost all times of exceedance.