Predictability of summer daily extreme precipitation in Central-Eastern Argentina

PENALBA O.C, POGGI M. M., BETTOLLI M. L.,

Santiago de Chile

Conferencia; 11th International Conference on Southern Hemisphere; 2015

There is a growing need for climate information from various productive sectors in order to plan theiractivities and implement adaptation measures to reduce the impacts of climate variability, especially ofextreme events. In Argentina, having estimators or predictors of daily precipitation extremes in seasonalscale would provide more and better products and specific climatic applications required for various sectors,through climate services. The latitudinal extent of Argentina, its topography and the different productivesectors force to make climate forecast studies at different time scales, both seasonal and intraseasonal, and byregion. Therefore, for this work, the center-east of the country has been chosen as the study area. In thiscontext, the main objective was to explore climate indices as potential predictors of summer daily extremeprecipitation in this area with the purpose of statistically model the relationship found in seasonal scale. Tothis end, the intensity of seasonal and monthly mean daily extreme precipitation was used, estimated as therate between the seasonal (December, January and February) and monthly cumulative precipitation,respectively, above the daily 75th percentile and the number of days that recorded this precipitation, fromobserved daily precipitation data of 30 weather stations included in the database CLARIS-LPB, in the period1960-2012. Different climate indices (those related to the ENSO phenomenon, Antarctic Oscillation, theIndian Ocean Dipole, the sea surface temperature in regions of the Atlantic Ocean, etc.) as well as regionalindices (blocking index, index of the onset of convection in the tropics, hydric balance in previous months,etc.) were considered as potential predictors. The relationship between predictors and daily extremeprecipitation was studied using the Pearson correlation coefficient with a phase shift up to 6 months prior,considering a significance level of 5%. Subsequently, the statistical modeling of the relationship betweenpredictors and predictand was made based on singular value decomposition analysis. The first results inseasonal scale indicated significant associations between the variable and diverse global indices in differentlagged months, but with little spatial homogeneity. In some stations, indices that estimate the variable withup to 6 months in advance were found, being the ENSO phenomenon the one that provides the highest signal.Regional indices, however, showed weak correlations with summer daily extreme precipitation with evenlower homogeneity, both spatially and temporally. In monthly terms, it was found that December is themonth of the warm season with higher predictability based on the indices associated with the ENSO, withsignal of the phenomenon getting lost in January and February. In these months, the relationship withdifferent regions of the Atlantic Ocean becomes more important along with the local conditions of soilmoisture in previous months. The different signals found in the three summer months forced to makestatistical modeling per month instead of considering summer in seasonal term. The results of the modelsshow that the combination of the selected predictors in each case has a good performance in forecastingextreme precipitation events in the study region.