Ensemble canonical correlation prediction of precipitation over the Sahel

Authors:

Abstract
Ensemble canonical correlation (ECC) prediction method is applied to predict summer rainfall over the Sahel. The predictors are the global sea surface temperature (SST), 200‐hPa streamfunction with zonal means removed (PSI), and forecasts or simulations from the climate models. The canonical correlation analysis (CCA) is performed for each variable separately. These predicted precipitation fields form an ensemble. The ensemble mean is the equal weighted average of its members. For both the multi‐model simulations and seasonal hindcasts, the CCA correction improves forecast skill. The ensemble, which consists of the CCA precipitation forecasts based on SST, PSI and the CCA corrected model simulations gives the highest skill. Each member has forecast skill over the different parts of the Sahel. Therefore, the ensemble mean has higher skill than its individual members.

This publication has 19 references indexed in Scilit: