Regional Flood Scaling Effect Analysis Under Changing Climate

In recent years,the impact of climate change and anthropogenic activities has changed hydrological processes,and further affected the mechanism of runoff processes.Therefore,traditional assumption of stationarity has been questioned,and nonstationarity in hydrological field is focused by many researchers nowadays.In this paper,the annual maximum flood peak(AMFP)series of eight meso-scale sub-watersheds in Daqinghe River basin were selected to conduct flood frequency analysis and flood scaling estimation under the influence of nonstationarity.The main contents and conclusions are summarized as follows:(1)The nonstationarity of AMFP series was detected.Mann-Kendall test and Spearman test were chosen to detect the trend in AMFP series,then the Non-parametric Pettitt test and Brown and Forsythe test were used to identify the change point of the series.It was found that all the AMFP series showed downward trend,and the change point was around 1979.(2)The nonstationary AMFP series were reconstructed under the environmental conditions before and after the change point(1979).Then flood frequency analysis was conducted by fitting the reconstructed stationary series using P-III distribution.Flood quantiles for the reconstructed series under the environmental conditions after1979 were lower than those before 1979.(3)Using GAMLSS model to conduct nonstationary flood frequency analysis based on nonstationary AMFP series by selecting time as covariate.The optimal distribution of Longmen was GA and the optimal distribution of Xidayang was WEI,while LOGNO fitted other six sub-watersheds best.The optimal parameters of Angezhuang and Hengshanling were constant,while the location parameters?of the other six AMFP series which were related to the mean all showed a negative correlation with time T,and scale parameters remain constant.The flood quantiles at given return period were much smaller than the results based on reconstructed AMFP series.(4)According to the flood quantiles calculated based on both stationary and nonstationary,the relationship of the drainage areas of the sub-watersheds and the flood quantiles were fitted with power function.Flood quantiles in Daqinghe River basin showed obvious scaling effect,and the scaling exponents of reconstructed stationary and observed nonstationary series increased with the return period decreasing.Based on analyzing land use changing,it was found that change ratio of forest and grassland increased with the drainage areas decreasing,which caused flood quantiles of sub-watersheds with smaller drainage area decreased more than that with larger drainage area.Therefore flood scaling exponents of reconstructed series under the environmental conditions before 1979 were smaller than those after 1979 at same return period.Then the relationship of the drainage areas,design rainfall and the flood quantiles were multiple fitted with power function whose scaling exponents were larger than the results only considering drainage area as covariant.(5)Based on the result of flood frequency analysis,the flood scaling exponent in Wangkuai sub-watershed was determined as the mean of scaling exponent of the reconstructed series under the environmental conditions before 1979.By analyzing the relation between rainfall amount and flood peak before and after 1979,the drainage area of the upstream small reservoirs and check dams was about1041.88km~2.