Impact of ECCO ocean-state estimates on the initialization of seasonal climate forecasts

The impact of ocean-state estimates generated by the consortium for Estimating the Circulation and Climate of the Ocean (ECCO) on the initialization of a coupled general circulation model (CGCM) for seasonal climate forecasts is examined. The CGCM consists of the University of California, Los Angeles, Atmospheric GCM (UCLA AGCM) and an ECCO ocean configuration of the Massachusetts Institute of Technology GCM (MITgcm). The forecasts correspond to ensemble seasonal hindcasts for the period 1993-2001. For the forecasts, the ocean component of the CGCM is initialized in either early March or in early June using ocean states provided either by an unconstrained forward ocean integration of the MITgcm (the "baseline" hindcasts) or by data-constrained ECCO results (the "ECCO" hindcasts). Forecast skill for both the baseline and the ECCO hindcasts is significantly higher than persistence and compares well with the skill of other state-of-the art CGCM forecast systems. For March initial conditions, the standard errors of sea surface temperature (SST) anomalies in ECCO hindcasts (relative to observed anomalies) are up to 1 C smaller than in the baseline hindcasts over the central and eastern equatorial Pacific (150 degrees-120 degrees W). For June initial conditions, the errors of ECCO hindcasts are up to 0.5 degrees C smaller than in the baseline hindcasts. The smaller standard error of the ECCO hindcasts is, in part, due to a more realistic equatorial thermocline structure of the ECCO initial conditions. This study confirms the value of physically consistent ocean-state estimation for the initialization of seasonal climate forecasts.