Wind-Stress Induced Multi-Year Predictability of Subpolar North Atlantic Sea Surface Temperature
Wind-Stress Induced Multi-Year Predictability of Subpolar North Atlantic Sea Surface Temperature
Abstract:
Wind-stress is an important driver of ocean and climate variability. It mostly affects variability on seasonal to interannual timescales, as in the El Niño/Southern Oscillation. However, wind-stress is also suggested to influence decadal timescale variability. For example, it was shown that coupled climate model simulations forced by observed tropical Pacific wind-stress reproduce the recent global warming hiatus. Furthermore, similar experiments could predict the decadal climate shifts in the Pacific. Thus, wind-stress forcing of the ocean is essential to understand climate predictability on timescales from seasonal to decadal.
In this study, we assess the value of wind-stress initialization for multi-year sea surface temperature (SST) prediction. For this purpose, we perform decadal hindcasts (historical predictions) with the Kiel Climate Model (KCM) initialized from “initialization runs” with prescribed wind-stress anomalies from the ERA-Interim reanalysis. The skill of the hindcasts in predicting observed SST anomalies, as defined by anomaly correlations with annual observed anomalies, is quantified.
We find extended predictability of subpolar North Atlantic SSTs for lead times of several years, even beyond 8 years where the auto-correlation (persistence) becomes zero. The enhanced skill is due to wind-driven changes in the horizontal and vertical ocean circulation. In the “initialization runs” with prescribed wind-stress anomalies, variabilities in the wind-stress curl drive a meridional displacement of the subpolar gyre. This results in anomalous heat and salinity advection into the subpolar region, also affecting the Atlantic Meridional Overturning Circulation and the associated northward heat transport, causing multi-year variability in the subpolar SSTs. The adjustment of the ocean circulation together with the subsurface heat storage enables skillful multi-year North Atlantic SST predictions.
In this study, we assess the value of wind-stress initialization for multi-year sea surface temperature (SST) prediction. For this purpose, we perform decadal hindcasts (historical predictions) with the Kiel Climate Model (KCM) initialized from “initialization runs” with prescribed wind-stress anomalies from the ERA-Interim reanalysis. The skill of the hindcasts in predicting observed SST anomalies, as defined by anomaly correlations with annual observed anomalies, is quantified.
We find extended predictability of subpolar North Atlantic SSTs for lead times of several years, even beyond 8 years where the auto-correlation (persistence) becomes zero. The enhanced skill is due to wind-driven changes in the horizontal and vertical ocean circulation. In the “initialization runs” with prescribed wind-stress anomalies, variabilities in the wind-stress curl drive a meridional displacement of the subpolar gyre. This results in anomalous heat and salinity advection into the subpolar region, also affecting the Atlantic Meridional Overturning Circulation and the associated northward heat transport, causing multi-year variability in the subpolar SSTs. The adjustment of the ocean circulation together with the subsurface heat storage enables skillful multi-year North Atlantic SST predictions.