The impact of climate change on circulation patterns in the Eastern Mediterranean Sea upper layer using Med-ROMS model

Abstract

This paper deals with three hindcasting simulations of circulation and sea level elevation, using Med-ROMS (Mediterranean- Regional Ocean Model System, with high horizontal -resolution (8.6 km). Each simulation has started at the end of previous 100-years spin-up and control run. The model surface boundary conditions, from the corrected ERA-40 re-analyses (Pettenuzzo et al., 2010) and three momentum flux ensemble members are based on a surface wind Bayesian Hierarchical Models (BHM) - Mediterranean Forecasting System (MFS-Wind-BHM) described in Berliner and Kim (2007) and Milliff et al. (2011), cover the period January 1958 to December 2001. The results have been validated by an Empirical Orthogonal Function (EOF) analysis of surface sea height anomalies for satellite and Med-ROMS Hindcasts. The correlation coefficients between Med-ROMS (Principal Component) PC1 (Sea Surface Height anomaly) SSHa hindcasts 1, 2, 3 and AVISO satellite are 0.985, 0.97 and 0.98 respectively, all values are significant at 95% confidence limit. We have defined a Bimodal Oscillating System mechanism responsible for decadal reversals of the Ionian basin-wide circulation (BiOS) (Gacic et al., 2011, 2014) as a pattern of Eastern Mediterranean climate change. The ROMS model successfully simulated the observed general circulation pattern in the Eastern Mediterranean. Over the period from July 1997 till December 2001, the Med-ROMS has successfully simulated the north Ionian inversion; SSHa same as Satellite SSH with an averaged RMSE between them of 0.0792 m. The spatial correlation maps between wind stress curl and Med-ROMS-SSHa/EOF1-SSHa are made and found to be significantly correlated at a 95% confidence limit. The highest significant correlation coefficient for Med-ROMS-SSHa was approximately 0.8 and located off the western Turkish coast inside the Aegean Sea. This strong significant positive correlation reflects the cyclonic motion of Rhodes Gyre. A high significant correlation coefficient for Med-ROMS-EOF1-SSHa, of around 0.51, was found along the western coast of Greece in the northeast Ionian Sea. For the first time, we could correlate the North Ionian Sea inversion to wind stress curl, which was significantly correlated with the EOF1 Med-ROMS SSHa averaged hindcasts during the inversion period. In this transient period, we noticed a significant correlation coefficient, at a 95% confidence limit, between NAO indices moving on average over 12 months and Med-ROMS-SSHa -PC1 over the Eastern Mediterranean.