The Influence of ENSO in the Flows of the Upper Paraná River of South America over the Past 100 Years

Abstract

The Upper Paraná River is the main tributary of the La Plata River basin, the second largest in South America, contributing with an annual mean flow of 12 000 m3 s?1 to more than one-half of the total water flowing in the La Plata River system. The Paraná River has a relevant importance in the region for transportation and hydroelectricity generation. This paper studies the influence of ENSO on the monthly flows measured at Posadas (27°23?S, 55°53?W) during the period of 1901?97. The original data are converted into standardized monthly anomalies, and the annual cycle is removed. Two data subsets are generated: a first group includes the years of warm ENSO events, or El Niño, and the second group includes the years of cold ENSO events, or La Niña. The elements of the subsets are composites of 24 consecutive months starting in January of the year when the ENSO event begins and ending in December of the following year. The results show that the averaged flows observed during El Niño events are always larger than those observed during La Niña events. The difference between the two groups peaks during November and December of the year when the ENSO event starts. During the following year, the difference reduces rapidly, and by May both La Niña and El Niño flows are above the average. A categorical contingency analysis between Pacific Ocean SST in the Niño-3 region and riverflow terciles during November?December of 1950?97 shows a statistically significant relationship. The relationship is such that upper-tercile (lower tercile) Niño-3 SST coincide with upper-tercile (lower tercile) riverflows in 37.5% of the cases, and only 2% of the cases account for the opposed situation of the type above?below and below?above. Also, a statistically significant positive correlation between November?December mean riverflows and 2-month-averaged Niño-3 SST starts as early as March?April and remains during the following months. Because the 1901?97 time series displays a variable long-term trend, a linear detrending is applied by means of a least squares regression of the original monthly time series. Also, a piecewise linear detrending is applied for three different periods that were identified with relatively steady long-term trend, namely, 1901?43, 1944?70, and 1971?97. The composite of ENSO events of the detrended time series shows no difference with the composite obtained with the original time series of monthly values. The conclusion is that the moderate trend observed in the Upper Paraná River flows over the past 100 years is not sufficient to modify the detected ENSO influence in the monthly mean flows.