Analysis of regional maximum precipitation

Abstract

Precipitation estimations of large return periods, which depend on short period observations, are not realistic. In such cases, regional analysis is needed. In this study, the 24-hour maximum precipitation data of the Yesilirmak River Basin in Turkey have been used. The basin can be treated as homogeneous, and is considered as a whole. The basin is divided into two subrogions in order to obtain a greater homogeneity. For both cases, regional maximum precipitation frequency analysis is performed using two-parameter Lognormal, GEV(PWM), Log Pearson Type III and Pearson Type III probability distribution functions. Dimensionless regional maximum precipitation frequency curves are obtained. Modeling studies for regional frequency analysis are carried out, which lead to robust frequency analysis of precipitation. For maximum precipitation, the GEV(PWM), Lognormal, Pearson Type III, and Log Pearson Type III distributions are selected. For each homogeneous region, regional frequency distribution curves of maximum precipitation are obtained. The curves assessed are compared to the assumed regional curve with regard to bias and variability.