Can Optimal Projection Improve Dynamical Model Forecasts?

An optimal projection for improving the skill of dynamical model forecasts is proposed. The proposed method uses statistical optimization techniques to identify the most skillful or most predictable patterns, and then projects forecasts onto these patterns. Applying the method to seasonal mean 2-m temperature from the Ensemble-Based Predictions of Climate Changes and Their Impacts (ENSEMBLES) multimodel hindcast dataset reveals that the method improves skill only in South America and Africa, suggesting that the benefit of optimal projection is limited to certain regions, but can be substantial. Further investigation reveals that the improvement in skill comes not from optimal projection itself, but from the EOF prefiltering that is done to reduce the dimension of the optimization space. Thus, much of the improvement attributable to optimal projection can be achieved by suitable EOF filtering. Interestingly, models are found to generate patterns that project only weakly on observational datasets but are strongly correlated between models. An important by-product of the method is a concise summary of the skillful or predictable structures in a given forecast. For the ENSEMBLES dataset, the method convincingly demonstrates that most of the seasonal prediction skill over continents comes from two components, ENSO and the global warming trend. In addition, the method can be used to determine whether a pattern exists that is well predicted by one model but not by another model (complementary skill).

The Predictive Skill and the Most Predictable Pattern in the Tropical Atlantic: The Effect of ENSO

This work investigates the predictive skill and most predictable pattern in the NCEP Climate Forecast System (CFS) in the tropical Atlantic Ocean. The skill is measured by the sea surface temperature (SST) anomaly correlation between the predictions and the corresponding analyses, and the most predictable patterns are isolated by an empirical orthogonal function analysis with a maximized signal-to-noise ratio. On average, for predictions with initial conditions (ICs) of all months, the predictability of SST is higher in the west than in the east. The highest skill is near the tropical Brazilian coast and in the Caribbean Sea, and the lowest skill occurs in the eastern coast. Seasonally, the skill is higher for predictions with ICs in summer or autumn and lower for those with ICs in spring. The CFS poorly predicts the meridional gradient in the tropical Atlantic Ocean. The superiority of the CFS predictions to the persistence forecasts depends on IC month, region, and lead time. The CFS prediction is generally better than the corresponding persistence forecast when the lead time is longer than 3 months. The most predictable pattern of SST in March has the same sign in almost the whole tropical Atlantic. The corresponding pattern in March is dominated by the same sign for geopotential height at 200 hPa in most of the domain and by significant opposite variation for precipitation between the northwestern tropical North Atlantic and the regions from tropical South America to the southwestern tropical North Atlantic. These predictable signals mainly result from the influence of the El Niño–Southern Oscillation (ENSO). The significant values in the most predictable pattern of precipitation in the regions from tropical South America to the southwestern tropical North Atlantic in March are associated with excessive divergence (convergence) at low (high) levels over these regions in the CFS. For the CFS, the predictive skill in the tropical Atlantic Ocean is largely determined by its ability to predict ENSO. This is due to the strong connection between ENSO and the most predictable patterns in the tropical Atlantic Ocean in the model. The higher predictive skill of tropical North Atlantic SST is consistent with the ability of the CFS to predict ENSO on interseasonal time scales, particularly for the ICs in warm months from March to October. In the southeastern ocean, the systematic warm bias is a crucial factor leading to the low skill in this region.