Connecting ocean observations with prediction

<strong class="journal-contentHeaderColor">Abstract.</strong> Ocean prediction relies on the integration between models and satellite and in situ observations through data assimilation techniques. Nowadays, satellites offer high-resolution observations of essential ocean variables at the surface, widely adopted in combination with precise but sparse in situ measurements that, from the surface to the deep ocean, can constrain large-scale variability in models. Moreover, observations are a valuable source of information for validating and assessing model products, for improving them, and for developing the next generation of machine learning algorithms aimed at enhancing the accuracy and scope of ocean forecasts. The authors discuss the role of observations in operational ocean forecasting systems, describing the state of the art of satellite and in situ observing networks and defining the paths for addressing multi-scale monitoring and forecasting.