A simple framework for estimating the annual runoff frequency distribution under a non-stationarity condition

Abstract

An increase in climate change has been leading to non-stationarity in hydrological variables. Therefore, a simple framework within the Budyko hypothesis is proposed to estimate the annual runoff frequency distribution and provide a new method for practical application in hydrological design under the non-stationarity. In this framework, the annual runoff statistic parameters, mean value (MAR) and standard deviation (STD), are derived by the Choudhury-Yang equation; at the same time, a magnification factor γ1 = 1.358 is introduced to correct the underestimation of the Budyko-based estimated STD according to observation data in 291 catchments across China. Furthermore, the P-Ш type frequency curve is selected to calculate the annual runoff on a specific design return period. Here, the impacts of non-stationarity caused by climate change was considered to be encoded in the changes in statistic parameters and frequency distribution. Based on this framework, the changes in water resources in 207 three-level basins across China during 2020–2050 are given. The results show that the average MAR will decrease by 2.7% for all basins, and the spatial distribution of MAR will be more even, i.e., it will increase in the north of China and decrease in the south of China. However, the inter-annual variability of annual runoff (shown as STD and variance coefficient, Cv) will increase in more than 70% of basins. Additionally, in the dry year, about 65% of basins show decreased runoff change, which indicates a larger reservoir capacity is need for water resource allocation. These findings offer a simple and effective way to re-examine the effects of non-stationarity on projected annual runoff.