he performance of the two-component extreme value distribution in regional flood frequency analysis

the performance of the regional two-component extreme value (TCEV) estimation procedure in estimating both dimensionless regional frequency curves showing the T-year flood as a multiple of the mean annual flood and site curves incorporating local and regional information was assessed using computer simulation experiments. It was found that the TCEV procedure produces estimates of both dimensionless regional curves and scaled site quantiles that are relatively unbiased but very variable when data are drawn from TCEV or Wakeby parents, although estimates are very biased when data follow a GEV distribution. The TCEV procedure is robust in the sense that its worst efforts are less bad than those of the alternatives considered, but performance declines significantly as sample sizes reduce