Initialized Seasonal-to-Interannual Forecasting without Initalization

Seasonal to interannual forecasts made by coupled general circulation models (CGCMs) undergo strong climate drift and initialization shock, driving the model state away from its long-term attractor. Here we explore initializing directly on a model’s own attractor, using an analog approach in which model states close to the observed initial state are drawn from a “library” obtained from prior uninitialized CGCM simulations. The subsequent evolution of those “model-analogs” yields an ensemble forecast, without additional model integration. This technique is applied to four CGCMs from the multi-model ensemble used operationally by NCEP, 29 CGCMs from the CMIP5 archive, and the 40-member CESM1 “Large Ensemble” (LENS), by selecting from prior long control and/or historical (externally forced) runs those model states whose monthly SST and SSH anomalies best resemble the observations at initialization time. Hindcasts are then made for leads of 1-24 months during 1958-2019. Deterministic and probabilistic skill measures of these model-analog hindcasts are comparable to, and in some regions better than, initialized CGCM hindcasts, for both the individual models and the multi-model ensemble. Notably, we find that the model-analogs have Year 2 (months 13-24) hindcast skill that is comparable to the initialized CGCM hindcasts made from the Decadal Prediction Large Ensemble (DPLE). Where model-analogs have greater skill, it suggests that CGCM improvement in seasonal to interannual skill is possible; where the model-analog skill is higher, it suggests that CGCM skill is degraded by initialization shock. The ability of model-analogs to be cheaply initialized monthly also is used to illustrate the importance of seasonality to Year-2 skill, and suggests possible changes to forecast protocols to aid in improved Year 2 forecasts from the numerical models as well.

This study suggests that with little additional effort, sufficiently realistic and long CGCM simulations may offer skillful seasonal to interannual forecasts of global SST anomalies, even without sophisticated data assimilation or additional ensemble forecast integrations. The model-analog method could provide a baseline for forecast skill when developing future models and forecast systems.