Abstract
Eighteen 30-day integrations with the NMC global atmospheric model (T40 resolution) were performed in order to test the impact of sea surface temperature anomalies (SSTAs) on 30-day forecasts for the Northern Hemisphere early summer. The years considered—1987, 1988, and 1989—correspond to a warm El Niño-Southern Oscillation (ENSO) event, a cold ENSO event, and a normal (non-ENSO year), respectively. For each year, 30-day forecasts were started on three successive days around 22 May, using climatological SSTs, and repeated using SSTAs fixed at their initial values.
The results indicate that SSTAs have a clear positive impact on the tropical forecasts and surface fluxes. The impacts on the extratropical forecasts, on the other hand, tend to be positive but small. Larger positive impacts in midlatitudes are obtained only in a case in which the atmospheric anomalous circulation is apparently driven by the ocean anomalies. A simple rule of thumb to distinguish whether quasi-stationary atmospheric anomalies are the cause or the result of SSTAs is discussed. It is also found that ensemble averaging results in a modest improvement in forecast skill. Moreover, in areas where the ensemble forecast anomalies are found to be significantly different from zero in a statistical sense, the anomalies tend to verify well, suggesting a method to estimate a priori regional skill. Overall, the Southern Hemisphere forecasts are more skillful than those in the Northern Hemisphere, perhaps because of a seasonal effect.
