Spatio-temporal validation of the Standardized Precipitation Index of the VAMOS atlas of extremes, with emphasis in drought events in Southern South America

PENALBA O. RIVERA J.

Congreso; AGU Fall Meeting; 2011

The aim of this research is to validate the spatio‐temporal variability of droughts inSouthern South America (south of 20°S) depicted by the VAMOS atlas of extremes, troughthe Standardized Precipitation Index (SPI). In the study region, droughts are a recurringclimatic phenomenon, with periodicities that vary regionally. Monitoring of these extremeevents and the development of early warning systems provide information necessary tomake timely decisions on the management of water resources.The SPI has been used to quantify deficits in water resources and as a drought monitoringtool in several regions of the world. This index quantifies, in a given time scale, thestandard deviations that the accumulated rainfall deviates from the average value.In order to build this index (hereafter SPI‐OBS), monthly rainfall data were obtained fromthe CLARIS LPB Data Base for 81 of its stations. This information was subjected to qualitycontrol procedures and has less than 10% of missing values. We used the 1981‐2008common comparison period. This permitted comparing the performance of the SPI basedon both CDC (SPI‐CDC) and CPC (SPI‐CPC) gridded observations and the MERRA (SPIMERRA)and CFSR (SPI‐CFSR) reanalysis. The validation was performed for the time scalesused in the atlas (3, 6, 12 and 24 months), which gives a broad approach of the differentprecipitation shortages. We also focused the analysis in the comparison of the extremeevents of 1988‐89 and 1995‐96. The SPI maps and time series of the atlas of extremeswere consistent with the results based on non‐gridded precipitation data. The spatialbehavior of the SPI during the selected drought events was quite accurate and capturedmuch of the variability shown in the SPI‐OBS. The performance of the SPI in the Patagoniaregion is remarkable. The time series of SPI in the East Argentina and Uruguay &Southeastern Brazil regions shows similar behaviors, but high frequency variabilities aremissing. These variabilities could be important for the forecasting of droughts on seasonalscales. Moreover, the SPI‐OBS at the time scales of 12 and 24 months showed for bothregions that the 1988‐89 drought was more severe than the 1995‐96 event. This wasverified for the Uruguay & Southeastern Brazil region but for the East Argentina region,SPI‐CPC and SPI‐CFSR showed an opposite behavior. We also plan to use the SPI‐OBS forthe validation of model outputs and other global and continental monitoring systemsbased in gridded precipitation data.