Abstract: Design low-flow analysis under nonstationary conditions is a critical consideration in water resources management. The design low-flow varying from one year to the next obtained by using the time-varying moment method is hard to be applied to practical application. This paper is aimed to improve the characterization of nonstationary design low-flow under the expected number of exceedances (ENE) interpretation of return period by employing meteorological covariates in the nonstationary frequency analysis. The method of using time as the only covariate is also applied for comparison. Both methods are applied to the annual minimum monthly streamflow series of the Weihe River, China. Results demonstrate that the nonstationary design low-flow results of both methods are significantly different from the stationary case. The nonstationary design low-flow result using temperature and precipitation as covariates is found more reasonable and advisable than that of the case using time as covariate. It is concluded that nonstationary design low-flow analysis can be helpful to water resources management during dry seasons exacerbated by climate change.

Key words: design low-flow, nonstationarity, time-varying moment method, return period, expected number of exceedances (ENE), covariates, Weihe River Basin