Aggregating Sea Surface Hydrodynamic Forecasts From Multi-Models for European Seas

Abstract

: Maritime information services supporting European agencies such as the FRONTEX require European ‐ wide forecast solutions. Following a consistent approach, regional and global forecasts of the sea surface conditions from Copernicus Marine Service and national met ‐ ocean services are aggregated in space and time to provide a European ‐ wide forecast service on a common grid for the assistance of Search and Rescue operations. The best regional oceanographic model solutions are selected in regional seas with seamless transition to the global products covering the Atlantic Ocean. The regional forecast models cover the Black Sea, Mediterranean Sea, Baltic Sea, North Sea and combine the North Sea – Baltic Sea at the Danish straits. Two global models have been added to cover the entire model domain, including the regional models. The aggregated product is required to have an update frequency of 4 times a day and a forecasting range of 7 days, which most of the regional models do not provide. Therefore, smooth transition in time, from the shorter time ‐ range, regional forecast models to the global model with longer forecast range are applied. The set of parameter required for Search and Rescue operations include sea surface temperature and currents, waves and winds. The current version of the aggregation method was developed for surface temperature and surface currents but it will be extended to waves in latter stages. The method relies on the calculation of aggregation weights for individual models. For sea surface temperature (SST), near real ‐ time satellite data at clear ‐ sky locations for the past days is used to determine the aggregation weights of individual forecast models. A more complicated method is to use a weighted multi ‐ model ensemble (MME) approach based on best forecast features of individual models and possibly including near real time observations. The developed method explores how satellite observations can be used to assess spatially varying, near real time weights of different forecasts. The results showed that, although