This report results from a joint research and development project between Aarhus University and DHI with the objective to (1) operationalize a new method for optimization of a satellite-based chlorophyll product for complex and near-coastal marine water bodies, as well as (2) test and develop routines for assimilation of this optimized chlorophyll product into DHI’s mechanistic biogeochemical models.
The method for optimization of satellite-based chlorophyll monitoring uses Sentinel-3’s chlorophyll product for complex and near-coastal marine water bodies together with in situ chlorophyll data acquired in the frame of the NOVANA program and a new distance-weighted regression model. The regression model generates spatial layers of two scaling factors, which are used to calculate chlorophyll concentrations based on estimates of pigment absorption in the Sentinel-3 chlorophyll product. The resulting optimized chlorophyll product significantly improves the relationship with chlorophyll concentrations acquired from traditional in situ samples for the Danish marine water bodies. Furthermore, the scaling factors show clear relationships with local ecological conditions and phytoplankton communities. The optimization method has been operationalized in a way that new satellite and in situ data can henceforth routinely be integrated and the optimization method applied automatically. This operational system can likewise generate chlorophyll data suitable for assimilation within DHI’s model.
Data assimilation is a complex task, however, at the example of a local model for the Northern Belt Sea, we could show that the developed routines using ensemble Kalman filter have the potential to improve the agreement between modelled and in situ acquired chlorophyll concentrations. This improvement is not only limited to areas with available satellite data for assimilation, but also in areas without or with less coverage, such as e.g. Vejle Fjord, the agreement between model results and in situ data gets better. Data assimilation does not only affect the single assimilated variable, but also other related variables. In comparison with in situ observations, the agreement for light attenuation was improved a bit; however, for nutrients we did not observe the same positive tendencies for all reference stations.
In general, the project’s results show that satellite-based data has great potential for monitoring of chlorophyll concentrations and phytoplankton distributions in Danish marine waters. Furthermore, we show that assimilation of these data into mechanistic models can considerably improve their performance. The project also identifies a couple of potential focus areas for further research with respect to optimizing the added value of integrating these new technologies into the national marine monitoring program for the Danish Seas.