Impact of ocean fronts on the reconstruction of vertical temperature profiles from sea surface measurements

Abstract

Temperature structures in the ocean interior are fundamental to interpret ocean dynamics. Yet their measurements are still limited, despite the increasing effort to deploy in-situ instruments over the last decades. Reconstruction of the vertical temperature profiles from sea surface variables is thus widely adopted to bridge this gap. Most of the studies deal with the reconstruction in a uniform manner without taking the presence of ocean processes into account. In this study, we evaluate the effects of ocean fronts on the reconstruction performances of vertical temperature profiles over the Northwest Pacific Ocean. Four mapping models based on linear and second-order polynomial regressions are trained by inputting the sea surface temperature (SST) and sea level anomaly (SLA). Model validations in terms of the mean bias and standard deviation (STD) are conducted based on an independent dataset excluded from the training process. The model of best performance is with inputs of SST and SLA using the polynominal regression. The spatial distribution of the bias and STD is then linked with the SST gradient to assess the front impacts. We found that the regions of high bias and STD are geographically consistent with that of large SST gradient. Both biases and STDs exhibit a quasi-linear increase trend relative to the SST gradient with slopes varying for depths. The bias reaches its maximum at around 300 m depth and decreases towards both deeper and shallower levels. These results suggest the necessity to consider fronts influence on the reconstruction of vertical temperature profiles, which shall improve the mapping accuracy.