This repository contains the official implementation of the methods described in the paper, "Convolutional neural networks for sea surface data assimilation in operational ocean models: test case in the Gulf of Mexico," now published in EGUsphere. In this work, convolutional neural networks (CNNs) are trained to correct model forecasts of sea surface temperature (SST) and sea surface height (SSH) using real satellite observations (GHRSST and altimeter data), model outputs from a high-resolution (1/25°) HYCOM run, and the Tendral Statistical Interpolation System (T-SIS) increments. We perform five controlled experiments to evaluate how different CNN architectures, input datasets, assimilation fields, training windows, and boundary conditions affect the assimilation process in a complex, operational setting. By integrating this approach into full primitive-equation models, we demonstrate that CNN-based data assimilation can deliver faster, more reliable ocean predictions, paving the way for improvements in environmental monitoring and maritime operations. For further details, please refer to the published paper in EGUsphere