Improving SST Anomaly Simulations in a Layer Ocean Model with an Embedded Entrainment Temperature Submodel

Abstract

Abstract In this study, an improved sea surface temperature (SST) anomaly (SSTA) solution for the tropical Pacific is presented by explicitly embedding into a layer ocean general circulation model (OGCM) a separate SSTA submodel with an empirical parameterization for the temperature of subsurface water entrained into the ocean mixed layer (Te). Instead of using subsurface temperature directly from the OGCM, Te anomalies for the embedded SSTA submodel are calculated from a historical data-based empirical procedure in terms of sea level (SL) anomalies simulated from the OGCM. An inverse modeling approach is first adopted to estimate Te anomalies from the SSTA equation using observed SST and simulated upper-ocean currents from the OGCM. A relationship between Te and SL anomalies is then obtained by utilizing an empirical orthogonal function (EOF) analysis technique. The empirical Te parameterization optimally leads to a better balanced depiction of the subsurface effect on SST variability by the mean upwelling of anomalous subsurface temperature and vertical mixing in the equatorial Pacific. As compared with a standard OGCM simulation, SSTA simulations from the embedded submodel exhibit more realistic variability, with significantly increased correlation and reduced SSTA errors due to the optimized empirical Te parameterization. In the Niño-3 region (5°S–5°N, 150°–90°W), the anomaly correlation and root-mean-square (RMS) error of the simulated SST anomalies for the period 1963–96 from the standard OGCM are 0.74° and 0.58°C, while from the embedded SSTA submodel they are 0.94° and 0.29°C in the Te-dependent experiment, and 0.86° and 0.41°C in the experiment with one-dependent-year data excluded, respectively. Cross validation and sensitivity experiments to training periods for building the Te parameterization are made to illustrate the robustness and effectiveness of the approach. Moreover, the impact on simulations of SST anomalies and El Niño are examined in hybrid coupled atmosphere–ocean models (HCMs) consisting of the OGCM and a statistical atmospheric wind stress anomaly model that is constructed from a singular value decomposition (SVD) analysis. Results from coupled runs with and without embedding the SSTA submodel are compared. It is demonstrated that incorporating the embedded SSTA submodel in the context of an OGCM has a significant impact on performance of the HCMs and the behavior of the coupled system, with more realistic simulations of interannual SST anomalies (e.g., the amplitude and structure) in the tropical Pacific.