The effects of the nonstationarity on the regional frequency estimates

Regional frequency analysis (RFA) is widely used to estimate more reliable quantiles of extreme hydro-meteorological events. This approach requires the assumption of stationarity. In this paper, index flood method was used to analyze the nonstationary 24-hour rainfall maxima of 9 virtual sites in a region. Monte Carlo simulation was used to generate the nonstationary data based on the generalized extreme value (GEV) model with time varying location/scale parameters. To analyze the temporal change of regional frequency estimates such as index flood, growth curve, quantile, and the heterogeneity measure, the accumulation of the data from the given time step and moving window concept were applied. Results indicated that the stationarity of the data had more effects on the index flood than the growth curve. There was little effects of the shape parameter of GEV distribution on the temporal pattern of the regional frequency estimates. Temporal change of the heterogeneity measure was not significant. These results showed that the time varying parameters of the nonstationary GEV model need to be formed considering the temporal pattern of the regional frequency estimates.