Frequency Response Properties of Forced Climatic SST Anomaly Variability in the North Atlantic

Abstract

Frequency response properties of North Atlantic (5°?57°N) sea surface temperature anomaly (Tsa) variability with periods of several months to 20 years are characterized using the Comprehensive Ocean-Atmosphere Data Set (COADS). Significant direct forcing of Tsa variability by the anomalous wind field (primarily through the resulting anomalous surface turbulent heat flux) is observed in the westerly wind and trade wind belts. To characterize properties of the large-scale climatic Tsa response to this forcing over the entire frequency band resolved, it is necessary to consider the dual role of anomalous surface heat flux as both the dominant local forcing mechanism and the dominant damping mechanism, the latter through a negative linear feedback (Newtonian relaxation). At frequencies where wind forcing is important, good agreement exists between the frequency response function estimated from data and the same function theoretically predicted by a simple stochastic forcing model where the locally forced response is damped by a negative linear feedback with a decay time scale of 3 mo. To make this comparison, the total anomalous surface heat flux represented by the standard bulk formula was decomposed into two components, one primarily representing the local wind forcing and the other primarily representing negative feedback damping. In the westerlies, wind forcing is effective over periods from several months to 8 yr, primarily 2?4 yr, and is ineffective at periods of 8?20 yr. These fluctuations are primarily forced in the western part of the basin then propagate to the east and northeast across the Atlantic at a characteristic speed of 6 km day?1. When time series of winter-only Tsa are analyzed, however, wind forcing of winter to winter Tsa variability remains significant at decadal and longer periods. In the trades, wind forcing is effective over periods from 8 mo to 13.3 yr, primarily 2-3 yr and 7?13.3 yr, and significant seasonal differences are not observed.