Daily precipitation data from about 9000 stations over South America are gridded to 1.0º to achieve more homogeneous distribution of data. Gamma distributions are fit to precipitation series at each grid point for each day of the year, in the period 1956-2002. Daily precipitation data are then replaced by their respective percentiles. Extreme events are those with a three-day average percentile above 90. The number of extreme events are computed for each month of each year. Years are classified as EN, LN, and normal years, and the mean frequency of extreme events for each month, within each category of year, is computed. Maps of the difference (and its statistical significance) between these mean frequencies for EN and normal years, and for LN and normal years show that EN and LN episodes influence significantly the frequency of extreme precipitation events in several regions of South America during certain periods.
The relationships between large-scale atmospheric perturbations and variations in the frequency of extreme precipitation events are sought through the analysis of anomalous atmospheric fields during extreme events in normal years, as well as in EN and LN episodes, in some of the regions in which there is significant change in the frequency of these events. The general features of those anomalous fields are similar for extreme events in any category of year (EN, LN or normal) close to the regions under focus, although they can be very different in remote regions for different categories of years. Over the analyzed regions, the anomalous fields show the essential ingredients for much precipitation: moisture convergence and mechanisms for lifting the air to the condensation level. In the regions where the frequency of extreme events increases (decreases) during EN or LN episodes the anomaly composites during extreme events in normal years show similarity (difference) with respect to the large-scale monthly atmospheric perturbations produced by those episodes. This indicates that the frequency of extreme events increases (decreases) when the large-scale perturbations favor (hamper) the circulation anomalies associated with them in those regions.
Besides the frequency, ENSO episodes also affect the intensity of extreme events in some regions of South America. The impact of these episodes on extreme events is even more consistent than on monthly or seasonal precipitation totals.
Supplementary URL: