Discrete analysis for the America west coast rainfall predictability using El Niño/Southern Oscillation relationships

Abstract

The objective of the study was to determine the probability of occurrence of wet or dry season events, based on the phase of El Niño/Southern Oscillation (ENSO) phenomenon using multinomial response logit and logit regression models. The study used monthly time series of the Pacific equatorial sea surface temperature (SST), a sea level pressure index (SOI) and rainfall anomalies over a 2.5x2.5 degrees grid along the west coast of Central and South America, for latitudes starting at 25°N, through 45°S, since 1951 to 2011. An ENSO index (NSO) was defined as predictor and rainfall as response. Series were first transformed into trimesters, replacing data by the three months average of the seasons DJF – MAM –JJA - SON. Data was categorized into terciles to construct non symmetrical three way contingency tables, including a time lagged categorization of the predictor variable (NSO). Two types of latitudinal profiles of the predictability (association), for the West Coast of Central and South America, using ENSO as predictor were generated as results. One using the categorized NSO index, using a multinomial response logit model to estimate the probabilities at the corners, of the contingency tables, representing the chances of extreme events of rainfall, given the El Niño and La Niña events and a second using multinomial response logit regression models, to estimate the same rainfall events, but using the NSO as a continuous predictor in which the NSO index was used as continuous predictor. The analysis was performed for all the contingency tables, particularly for the “corner cells” and the results were plotted as latitudinal profiles.