Regional Frequency Analysis of Extreme Precipitation in Louisiana

Abstract

A regional frequency analysis procedure is outlined and applied to precipitation data in Louisiana. A total of 92 rain gauges were used to generate 25 synthesized stations with long periods of record. Annual maximum series for rainfall durations of 1, 3, 6, 12, and 24 hr from the 25 synthesized stations were used for various statistical analyses. The mean annual precipitation, geographical locations, and the synoptic generating mechanisms were used to identify the three climatologically homogeneous regions in Louisiana. Using the L-moment ratios, the underlying regional probability distribution was identified to be the generalized extreme value (GEV) distribution. The regional parameters of the GEV distribution were estimated by the indexed probability weighted moments (PWM). The regional analysis procedure was tested by Monte Carlo simulation. Relative root-mean-square error (RRMSE) and relative bias (RBIAS) were computed and compared with those resulting from at-site Monte Carlo simulation. All results show that the regional procedure can substantially reduce the RRMSE and RBIAS in quantile prediction.