Regional flood frequency analysis in the central-western river basins of Argentina

Abstract

Knowing the probability of occurrence of a flood event is an important issue for water resources planning. At-site probability models require a long extension of hydrological data for robust estimation of low-frequency events. As the mean record length of 25 gauge stations in western river basins of Argentina is 49 years (until 2010), regional models are an interesting tool to determine mountain rivers system dynamics. This study aims to estimate low-frequency quantiles of annual maximum flow in Argentinean western river basins (28°S–37°S) applying regional frequency analysis based on the L-moments method. Besides, mean annual maximum flow of 75 gauge stations (22°S–52°S) was analysed. First, an exploratory data analysis was performed; normality, independence, and randomness were accepted in the 27%, 87%, and 91% of cases, respectively. Increasing trends in annual maximum flows in the north-western and central-western rivers of Argentina were detected, whereas decreasing trends in annual maximum flow in the Patagonian Andes were identified. Base on at-site characteristics and at-site statistics, a homogeneous region of 12 stations with a record period of 568 years was formed. General extreme value was the most appropriate distribution for this homogeneous study region. Estimation accuracy using Monte Carlo simulations was performed. The error bounds were set at 90%, the mean square error was 9.23%, and the relative bias was 1.6%. The regional method performed better than the at-site estimation.