Differences between observed and a coupled simulation of North Atlantic sea surface currents and temperature

Abstract

North Atlantic sea surface temperature (SST) distributions derived from observations and a coupled model from NOAA's Geophysical Fluid Dynamics Laboratory, CM2.1, are compared to evaluate the model's ability to simulate recent (1900 to the present) oceanic surface characteristics. The North Atlantic focus will limit our analyses to spatial scales less than gyre, scales usually not addressed in previous model‐observation comparisons. Identifying model differences from observations at these scales will assist modelers in identifying problems to be considered and remedies to be applied. The properties compared are the mean annual SST, standard deviation, amplitude of the annual and semiannual harmonic, decadal meridional movements of the axis of the Gulf Stream, propagation of SST anomalies along the axis of the Gulf Stream, and 100‐year trends in SST records. Because of the dependence of SST on surface currents, observed flow from surface drifters and simulated flow from 15 m fields are also compared. The model simulates the large‐scale properties of all the variables compared. However, there are areas of differences in some variables that can be related to inadequacies in the simulated current fields. For example, the model Gulf Stream (GS) axis after separation from the western boundary is located some 100 km north of the observed axis, which contributes to an area of warmer simulated SSTs. The absence of a slope current in the same region that advects colder water from the Labrador Sea in the observations also contributes to this area of higher model SSTs. The model North Atlantic Current (NAC) is located to the east of the observed NAC contributing to a large area of SST discrepancy. The patterns of the amplitude of the annual harmonic are similar with maximum amplitude off the east coast of northern North America. The semiannual harmonic exhibits relatively large amplitudes (>1°C) north of about 55°N, a signal not found in the observations. In both the model and observations, a region of increased standard deviations encompasses the GS and NAC. The model simulates north‐south migrations of the GS core but at a longer period (20 years) than observed. The model does not simulate the SST anomalies that propagate along the observed GS and NAC. The model captures both the spatial and temporal characteristics of the Atlantic Multidecadal Oscillation. Both model and observations exhibit a dipole in trends, with positive trends in the subtropical Atlantic and negative trends in the subpolar gyre. The modeled region of negative trends is limited to the western subpolar Atlantic. The observed trends extend farther to the east.