Evaluation of Uncertainty Sources in Extreme Flow Projections under Climate Change in Watersheds with Various Climatic Features

Assessing the impacts of climate change on extreme flow is necessary for flood and drought disaster prevention and sustainable water resources management. However, large uncertainties caused by different sources need to be quantified in extreme flow projections. In this paper a modeling chain including three emission scenarios (ESs), four climate models (CMs), four statistical downscaling method (SDs), four hydrological modeling (HM) schemes and probability distributions (PDs) was established to project extreme flow changes in 3 basins with various climatic features in China for near-term and far-term periods (2021-2050 and 2051-2080) with respect to baseline period (1961-1990). Analysis of variance (ANOVA) approach is used to quantify the contribution of different uncertainty sources. Results show that the intensity and frequency of precipitation extremes in all basins will rise in response to climate change. For most signatures there is an increase of extreme high flow compares to the baseline period in all basins. Extreme low flow is expected to decrease in the Upper Ganjiang River basin change while increase in Laoha river basin. The ANOVA assessment of the relative contribution of different sources to the extreme flow projections uncertainty also shows that CMs generally are the dominate uncertainty source in extreme low flow projections. The uncertainties of precipitation prediction due to ESs gain more importance in far-term period than in near-term period. SDs are the main source of uncertainty in estimating extreme high flow variation in the Upper Ganjiang River basin. HM produces larger uncertainties in the Laoha River basin than the other basins. PDs contribute a low percentage of uncertainty in extreme flow estimates. Interaction between uncertain sources should be taken seriously and even produce larger uncertainties than individual modeling chain components partly.