Wind Stress‐Induced Multiyear Predictability of Annual Extratropical North Atlantic Sea Surface Temperature Anomalies

A Reintges,W Park,M Latif,M H Bordbar

doi:10.1029/2020gl087031

Abstract

AbstractLong‐term predictability of the North Atlantic sea surface temperature (SST) is commonly attributed to buoyancy‐forced changes of the Atlantic Meridional Overturning Circulation. Here we investigate the role of surface wind stress forcing in decadal hindcasts as another source of extratropical North Atlantic SST predictability. For this purpose, a global climate model is forced by reanalysis (ERA‐interim) wind stress anomalies over the period 1979–2017. The simulated climate states serve as initial conditions for decadal hindcasts. Significant skill in predicting detrended observed annual SST anomalies is observed over the extratropical central North Atlantic with anomaly correlation coefficients exceeding 0.6 at lead times of 4 to 7 yrs. The skill is insensitive to the calendar month of initialization and primarily linked to upper ocean heat content anomalies that lead anomalous SSTs by several years.

Highlights

The potential of multiyear to decadal climate predictability arises from two sources: external forcing and long‐term internal climate variability (Meehl et al, 2009)
The aim of this study is to show how specification of wind stress anomalies in a climate model influences multiyear predictions of extratropical North Atlantic sea surface temperature (SST)
This study shows that prescribing reanalysis wind stress anomalies in the KCM, next to the skill arising from long‐term trends, is an additional source for skillful forecasts of the extratropical North Atlantic SST at lead times of several years

Summary

Introduction

The potential of multiyear to decadal climate predictability arises from two sources: external forcing and long‐term internal climate variability (Meehl et al, 2009). In the North Atlantic, initialized climate hindcasts (retrospective forecasts) show additional multiyear skill compared to uninitialized simulations that only consider external forcing (Boer et al, 2016; Meehl et al, 2014; Smith et al, 2019). The basis for multiyear predictability due to initialization is the long‐term oceanic memory for short‐term atmospheric fluctuations. This enables multiyear predictions of climatic parameters and marine ecosystems such as fish production (Årthun, Bogstad, et al, 2018; Mantua et al, 1997; Tommasi et al, 2017). Important climatic applications are Atlantic hurricane activity (e.g., Smith et al, 2010), terrestrial precipitation (e.g., Delworth et al, 2015; Newman et al, 2016; Ruprich‐Robert et al, 2018; Yang et al, 2019), and land surface temperatures (e.g., Årthun et al, 2017; Årthun, Kolstad, et al, 2018; Knight et al, 2014; Liu et al, 2019; Monerie et al, 2018; Ruprich‐Robert et al, 2018)

Objectives

Results

Conclusion