How well do operational climate models reproduce ENSO precipitation in Southern South America?

Analyses of historical data shows that much of Southern South America (SSA) experiences consistent and strong precipitation anomalies associated with the various phases of El Nino and La Nina episodes. The timing of the anomalies varies throughout SSA due to different processes leading to precipitation anomalies. We examine the ability of current generation climate models i.e., those currently used for seasonal forecast purposes at the IRI to reproduce these consistent anomalies when forced by observed SST anomalies. Precipitation data from two models are examined in slightly different ways. The 8-member ensemble mean precipitation from the ECHAM3 model forced with observed SST for the period 1956-1992 is compared to observed data from a set of 134 stations in SSA. Composites of precipitation percentiles for every month within the El Nino/La Nina cycle are generated, along with the assessment of the consistency of the anomalies, given by a test based on the hypergeometric distribution. This analysis shows that the ECHAM3 model best reproduces the main features of the observed ENSO cycle precipitation anomalies during La Nina episodes.

ENSO-related shifts in the 13-member ensemble mean precipitation distributions from the NCEP GCM are also examined for broad seasons. The composites of circulation anomalies from the model outputs and NCEP/NCAR are compared for November of El Nino years to diagnose differences between the modeled and observed rainfall anomalies.