Regional Frequency Analysis for the Prediction of Maximum Flows in Ungauged Basins of the Peruvian Amazon

Abstract

The estimation of the maximum design flow is important for flood management. However, the limited existence of gauged sites and the scarcity of hydrological measurements make it impossible to estimate them in ungauged basins. In this study, the regional frequency analysis (RFA) was carried out for the prediction of maximum flows in ungauged basins of the Peruvian Amazon. The methodology consisted of the identification of homogeneous regions, selection of the regional distribution function, estimation of regional quantiles, regionalization of the index-flood, and the prediction of maximum flows in ungauged basins. The results have identified a well-defined homogeneous region called region 1. The generalized extreme value (GEV) distribution proved to be more adequate to represent the data sample of region 1, and the basin area explained the variability of the index-flood in 99.4% (R2 = 0.994). The prediction of maximum flows in ungauged basins presented wide ranges of uncertainty, mainly for high return periods. It is concluded that the RFA provides reliable estimates for the prediction of maximum flows as long as the uncertainty ranges are considered at each frequency.