Persistent and Reemergent Sea Surface Temperatures: A Recipe for Better Seasonal Climate Forecasts

Monthly tropical sea surface temperature (SST) data are used as predictors to make statistical forecasts of extended winter (November-March) precipitation and temperature for the contiguous United States. To this end, we develop and implement the combined, lagged sea surface temperature (CLSST) model. The CLSST model methodology finds predictive information not just from recent SSTs, but also from SSTs up to 18 months old. CLSST winter-season forecasts are then compared to the North American Multi-Model Ensemble (NMME) and the European Centre for Medium-Range Weather Forecasts' (ECMWF) seasonal climate model SEAS5. On average across the US, the CLSST model exhibits greater skill than either NMME and ECMWF for both precipitation and temperature. The precipitation forecast skill obtained by CLSST in parts of the Intermountain West is of particular interest. In those regions, CLSST dramatically improves the skill over that of the dynamical model ensembles, which can be attributed to a robust statistical response of precipitation in this region to SST anomalies from the previous year in the tropical Pacific.