Increased flood hazards within the Himalayan Karnali River catchment predicted for an ensemble of CMIP-6 climate change scenarios

The Himalayas are of exceptional importance for the water resources in Asia and provide fresh water for more than 1.4 billion people. However, they are also the source of frequent floods with the highest death-per-event rates in the world. The floods are caused by intense monsoon precipitation, but snow melt and glacier melt contribute to the flood peaks. Both, extreme precipitation and melt contributions are predicted to be impacted by climate change but it remains unclear how these changes will alter the flood risk in the region.

This study investigates the impact of climate change on peak runoffs in the transboundary Karnali River Basin (KRB) in Nepal / China using both hydrological and statistical modelling. The fully-distributed cryospheric-hydrological model SPHY is applied for the period 2002-2015 and calibrated and validated using the GLUE framework. The Nash-Sutcliffe efficiency, PBIAS of peak flows and extended GLUE are used as performance indicators for the selection of behavioural parameter sets. The model is run with the selected parameter sets and the outputs of 13 downscaled and bias-corrected CMIP-6 models of three different scenarios (historical, ssp245, ssp585) to quantify the climate-change-induced changes in peak flows until the end of the century. Extreme Value Analysis is then applied to estimate the exceedance probability from the simulated annual maximum flows for the climate models, scenarios and hydrological parameter sets.

The results indicate an increase in flood hazard frequency and magnitude until the end of the century. The mean magnitude of an event with 2% annual exceedance probability (AEP) increases by 23% (±19%) in the period 2020 - 2059, and 42% (±19%) in the period 2060 - 2099 for ssp245, and 28% (±23%) in the period 2020 - 2059 and 82% (±41%) in the period 2060 - 2099 for ssp585 compared to the baseline period (1975 - 2014). Flows with a 50 year return period (2% AEP) during the baseline period (10,900 m³/s) are projected to occur every 9 years in the period 2020 - 2059 and 7 years in the period 2060 - 2099 for ssp245 scenario, and every 9 and 3 years for ssp585 scenario, respectively. Glacier and snowmelt contributions are projected to change in terms of seasonality and quantity but the increase of peak flows is mainly driven by the increase in extreme precipitation.