Study On ENSO Multi-model Ensemble Predictions With The Statistical Correction Method

During the last few decades,climate scientists have made tremendous advances in understanding and modeling ENSO.Although the most advanced technologies of the coupled atmosphere-ocean modeling technique have been applied to dynamical seasonal prediction,limitations and challenges still remain.It is certain that better dynamical seasonal prediction can be achieved by improving a dynamical model itself.Recent study shows that further improvement of prediction can be obtained from postprocessing procedures such as MME techniques and statistical error correction(or downscaling)methods.In this study,we have developed a new MME method called SENSM(Stepwise Pattern Projection Method Ensemble)and evaluated it using 5 dynamical models' retrospective forecasts for the period 1982-2010.The ensemble members are CFS_v2,GFDL_CM2p1,GFDL_CM2p1_aer04,GFDL_CM2p5_-FLOR_B01 and COLA_CCSM4 from Climate Forecast System,Geophysical Fluid Dynamics Laboratory and National Center for Atmospheric Research.Our study also have discussed the prediction skill of both Cold Tongue El Ni?o and Warm Pool El Ni?o.The strengths of the new method lie in a statistical error correction procedure for predictions,and a multi-model method without statistical error correction(ENSM)is conducted to compare with SENSM.The Stepwise Pattern Projection Method(SPPM)is applied to statistically correct El Ni?o-South Oscillation(ENSO)prediction in this study.The main idea of the SPPM is to produce a prediction at the predictand grid by projecting the predictor field onto its covariance pattern with the one-point predictand after selecting the predictor domain.To evaluate the prediction skill of SPPM,it is applied to sea surface temperature(SST)prediction data produced by CFS_v2.It is demonstrated that the SPPM significantly improves the performance of the prediction over the equatorial Pacific.The temporal correlation score has increase 17% in terms of Ni?o3.4 SST anomaly index with a 6-month lead.The spatial anomaly correlation coefficients for El Ni?o event predictions also increase obviously by the SPPM at most lead months.When it comes to the SST prediction for the multi-model ensemble method,we found that the skill of SENSM overmatch the single ensemble forecast and ENSM.The temporal correlation score has increased 33% compared with GFDL_CM2p5_B01 and 12% compared with ENSM in terms of Ni?o3.4 SST anomaly index with a 6-month lead.We have also examined the anomalous SST patterns related to CTI and WPI simultaneously.Results indicate that the SENSM captures the spatial patterns of SST related to Cold Tongue El Ni?o and Warm Pool El Ni?o better at most lead months then the result of ENSM and the other models.It also found that SENSM significantly improves the skills of rainfall prediction.The SENSM captures the spatial patterns of rainfall related to Cold Tongue El Ni?o and Warm Pool El Ni?o at most lead months as well.The spatial correlation scores have increased 28% and 38% compared with ENSM in terms of El Ni?o and La Ni?a precipitation anomaly composites patterns in summer.In this study,the prediction skills of ENSO are significantly improved by using several international advanced dynamical models.The proposed multi-model ensemble method will be applied in practical ENSO prediction,which cooperates the statistical correction method with the multi-model ensemble method.And these two methods will be further studied in the future.