Errors in the winter temperature response to ENSO over North America in seasonal forecast models

Abstract

AbstractThis study presents the ability of seasonal forecast models to represent the observed mid-latitude teleconnection associated with El Niño-Southern Oscillation (ENSO) events over the North American region for the winter months of December, January, and February. Further, the impacts of the associated errors on regional forecast performance for winter temperatures are evaluated, with a focus on one-month lead time forecasts. In most models, there exists a strong linear relationship of temperature anomalies with ENSO and, thus, a clear anomaly sign separation between both ENSO phases persists throughout the winter, whereas linear relationships are weak in observations. This leads to a difference in the temperature forecast performance between the two ENSO phases. Forecast verification scores show that the winter season warming (cooling) events during El Niño in northern (southern) North America are more correctly forecasted in the models than the cooling (warming) events during La Niña. One possible reason for this result is that the remote atmospheric teleconnection pattern in the models is almost linear or symmetric between the El Niño and La Niña phases. The strong linear atmospheric teleconnection appears to be associated with the models’ failure in simulating the westward shift of the tropical Pacific rainfall response for the La Niña phase compared to that for the El Niño phase, which is attributed to the warmer central tropical Pacific in the models. This study highlights that understanding how the predictive performance of climate models varies according to El Niño or La Niña phases is very important when utilizing predictive information from seasonal forecast models.