Changes in the extreme events of precipitation and their feedbacks with temperature over South America

PABLO G. ZANINELLI; ENRIQUE SANCHEZ; ANDREA F. CARRIL; CLAUDIO G. MENENDEZ; MARTA DOMINGUEZ

Petrópolis

Workshop; CLIVAR/VAMOS Workshop on Modeling and Predicting Climate in the Americas; 2012

Laboratorio Nacional de Computação Cientifica (LNCC)

Recently, the efforts of the scientific community to simulate the regional climate of South America (SA) have been intensified. The approach for dynamical downscaling has been changed from re-initialization into continuous multi-year and climate change experiments (Solman et al. 2008, Silvestri et al. 2009, Nuñez et al. 2009, Marengo et al 2009, Alves and Marengo 2010, Marengo et al. 2010a, Pesquero et al. 2010, Sörensson et al. 2010, Chou et al. 2011, Marengo et al. 2011), and multi-model ensembles and coordinated experiments starting to became available (Roads et al. 2003, Fernandez et al. 2006; Menendez et al. 2010a, 2010b, Marengo et al. 2010b, Carril et al. 2012, Solman et al. 2012). In particular, present climate conditions of coordinated experiments conducted in the framework of CLARIS and CLARIS-LPB EU projects, have been analyzed by Menéndez et al. (2010a and 2010b), Carril et al. (2012) and Solman et al. (2012). Carril et al. (2012) have shown that the CLARIS models are able to reproduce the broad spatial pattern of precipitation, but overestimate the convective precipitation in the tropics and the orographic precipitation over highlands, while underestimate the precipitation near the monsoon core region and the daily intensity of rainfall everywhere. Solman et al. (2012) have found that CLARIS-LPB RCMs reproduce well the spatial distribution of seasonal precipitation and the inter-model spread is small over most of SA, except over the Andes region. Moreover, the ensemble mean reproduces appropriately the annual cycle of precipitation over vast regions of SA. Following Sánchez et al. (2011) and Sánchez et al. (2012), we study the climate change signal of non-precipitation indices as simulated by four RCMs. Those indices are related with the extremes in the left tail of the statistical distribution of the daily precipitation (i.e., dry spells)