Ensemble forecasting of harmful algal blooms in the Baltic Sea

Operational marine environmental modelling has been considered notoriously difficult; consequently there are very few operational models of the marine environment. Operational modelling of harmful algal blooms (HABs) requires the modelling of individual species and is therefore harder still. The separation of algal species in models requires detailed knowledge of their behaviour (survival strategy through the life cycle), and physiological ecology.We present quantitative results of an ensemble approach to HAB forecasting in the Baltic, and discuss the applicability of the forecasting method to bioheochemical modelling. Ensembles were produced by running a biogeochemical model several times and forcing it on every run with different set of seasonal weather parameters from European Centre for Medium-Range Weather Forecasts' (ECMWF) mathematically perturbed ensemble prediction forecasts. The ensembles were then analyzed by statistical methods and the median, quartiles, minimum and maximum values were calculated for estimating the probable amounts of algae. To evaluate the forecast method final results were compared against available and valid in-situ HAB data in a case study. It turns out that quantitative HAB forecasts are possible. Further verification will require expanded observational networks.