analysis of copernicus baltic sea model to understand and predict intense upwelling events in the finland gulf. characterise upwelling by one or more structures in SST (EOFs?) and correlate with wind. can we do the same with ferrybox data? how much of the fb sst variabilit is captured in sat sst and in the model? does the assimilation do well when there is no good sat picture? can we make better forecasts of somer upwelling metric from a simple linear model based on obs than the model?