Vendredi 30 novembre - ENSO representation in climate models from CMIP3 to CMIP5
de 14:00 à 15:00
|Où ?||Salle Jules Verne|
Hugo Bellenger - LOCEAN-IPSL
|Ajouter un événement au calendrier||
Titre : ENSO representation in climate models from CMIP3 to CMIP5
The ability of coupled ocean-atmosphere general circulation models (CGCMs) to simulate the El Niño Southern Oscillation (ENSO) has improved over the past decades. Nevertheless, the diversity of model simulations of present-day ENSO characteristics indicates current limitations in our ability to model this climate phenomenon and anticipate properly changes on short and long time scales. Recent multi-model studies of ENSO in CMIP models show that standard evaluation via metrics and the process-based evaluation are both required. Here we apply a series of ENSO metrics mainly proposed by the CLIVAR Pacific Panel to both CMIP3 and CMIP5 simulations. It is found that ENSO characteristics are not radically different in CMIP5 versus CMIP3 simulations. The large diversity in the CMIP3 amplitude of Sea Surface Temperature (SST) variability is however reduced in CMIP5. If ENSO seasonal phase lock is better represented in CMIP5, other ENSO fundamental characteristics as SST anomalies spectrum and associated central Pacific precipitation variability remain difficult to represent. Large errors in the tropical Pacific mean state simulation moreover remain in CMIP5. The Bjerknes and the shortwave feedbacks, which are the central Pacific zonal wind stress and eastern Pacific shortwave flux response to SST anomalies in the eastern Pacific, remain important sources of uncertainty in CGCMs. In particular, the shortwave feedback is nonlinear as its sign depends on the atmospheric regime that switches seasonally from subsident to convective in the observations. However, only one third of CMIP3 and CMIP5 models can reproduce this shift of regime. Interestingly, there is a strong relationship between the shortwave feedback nonlinearity and the modeled ENSO amplitude. The causality remains unclear, however it has to be noted that the ability of a model to change its atmospheric regime is linked with its ability to reproduce the tropical Pacific mean state. Finally this study provides a synthesis on the ability of each CMIP3 and CMIP5 model to simulate ENSO and an example of use of the metrics to sum up the information on the simulation of ENSO and its associated processes.