NeSPReSO: A Neural Approach for Generating Synthetic Ocean Profiles Using ARGO Data, with an Accessible API for the Gulf of Mexico

Accurate circulation modeling in the Gulf of Mexico (GoM) is hampered by the limited availability of in-situ subsurface data, leading to inaccuracies in subsurface representations. These inaccuracies diminish the reliability of ocean models and constrain the duration of dependable forecasts. To address this, we present the latest version of the Neural Synthetic Profiles from Remote Sensing and Observations (NeSPReSO), a data-driven approach designed to efficiently and accurately estimate subsurface temperature and salinity profiles using satellite-derived surface data as input. This method provides a robust alternative to conventional synthetic data generation techniques. 

NeSPReSO applies Principal Component Analysis (PCA) to extract dominant features from temperature and salinity profiles in an Argo dataset. A neural network is then trained to predict these features using inputs such as time, location, and satellite-derived variables, including absolute dynamic topography, sea surface temperature, and sea surface salinity. The model's performance was rigorously evaluated using independent Argo profiles and glider data collected in the GoM, demonstrating better performance compared to traditional methods such as Gravest Empirical Modes (GEM), Multiple Linear Regression (MLR), and Improved Synthetic Ocean Profile (ISOP). Results show reductions in root mean square error and bias, indicating that NeSPReSO effectively captures the primary variability of subsurface fields. Furthermore, the synthetic profiles generated by NeSPReSO align well with observed data, accurately representing key oceanographic features such as thermoclines, haloclines, and the region's temperature-salinity structure.

To facilitate widespread application, we have developed an API that allows users to generate synthetic profiles for any location in the Gulf of Mexico at varying spatial and temporal resolutions. This resource offers the broader scientific community a valuable tool for estimating quantities such as the region's heat content and enhancing oceanographic research and forecasting capabilities.