Description: The effect of satellite sea surface temperature assimilation on the forecast quality of the coastal ocean-biogeochemical model HBM-ERGOM in the North- and Baltic Seas is studied. The HBM-ERGOM model is currently used operationally without data assimilation by the German Federal Maritime and Hydrographic Agency (BSH). The model is configured with nested grids with a resolution of 5 km in the North- and Baltic Seas and a resolution of 900 m in the German coastal waters. The biogeochemical model ERGOM contains three phytoplankton groups (Cyanobacteria, Flagellates, Diatoms) and two zooplankton size groups to simulated the biogeochemical cycling in the coastal seas. To improve the predictions of the HBM-ERGOM model, data assimilation was added by coupling the model to the parallel data assimilation framework (PDAF, http://pdaf.awi.de). The ensemble-based error-subspace transform Kalman filter (ESTKF) is applied for the data assimilation. As a first step to improve the biogeochemical forecasts, before the planned assimilation of ocean color data products, the impact of assimilating satellite sea surface temperature data is assessed. Two cases are considered. First, the impact of weakly coupled data assimilation. In this case, the assimilation of temperature only directly influences the physical model variables in the analysis step while the biogeochemical fields react dynamically to the changed physical model state during the ensemble forecasts using the coupled model. The second case is the strongly-coupled data assimilation in which next to the physical model fields also the biogeochemical fields are directly updated in the analysis step through the multivariate covariances estimated by the joined physical-biogeochemical ensemble of model states. Here, it is assessed whether these covariances are sufficiently well estimated to result in an improvement of the biogeochemical fields.

Description: Data assimilation combines observational data with numerical simulation models. The methodology allos to improve the initialization of model predictions, determining model deficiencies, but also to enhance data sets by augmenting the data with dynamical information from numerical models simulating e.g. ocean physics or biogeochemistry. This combination can fill data gaps by an interpolation which accounts for the dynamical information provided by the numerical model. Further the observed information can be used to improve unobserved variables, and even fluxes. This is accomplished through the use of dynamically estimated cross-covariances between the observed and unobserved variables. The assimilation can result in data sets which, at the resolution of the model, exhibit smaller errors than using the observations or the model alone. I will discuss the method of ensemble-based data assimilation on the example of ocean-biogoechemical modeling with the assimilation of satellite ocean color data.

Description: Modellvorhersagen helfen die Datenbasis für die behördlichen Berichtspflichen für die Meeresstrategierahmenrichtlinie zu verbessern. Um die Qualität der Modellvorhersagen zu weiter zu verbessern kann der Modellzustand mit Beobachtungsdaten kombiniert werden. Dieses wird in quantitativer Weise durch Methoden der Datenassimilation vorgenommen. Im Rahmen des Projektes MeRamo wurde das Vorhersagemodell des Bundesamtes für Seeschifffahrt und Hydrographie für den kombinierten ozean-ökosystem Zustand in der Nord- und Ostsee mit Beobachtungsdaten des Satelliten Sentinel-3a sowie Satelliten der amerikanischen Behörde NOAA mit Hilfe der Datenassimilation kombiniert. Hierdurch wird die Simulation sowohl des physikalische Zustands (wie Temperatur und Salzgehalt) als auch ökologischer Größen wie Nährstoffe oder Planktonkonzentrationen beeinflusst. Im Vortrag wird die verwendete Datenassimilationsmethodik diskutiert und der Einfluss der Assimilation auf den Meereszustand, vor allem in Hinblick auf mögliche Indikatoren für die Meeresstrategierahmenrichtlinie betrachtet.

Description: The effect of the assimilation of satellite sea surface temperature onto the forecast quality of the coastal ocean-biogeochemical model HBM-ERGOM in the North and Baltic Seas is studied. The HBM-ERGOM model is currently used pre-operationally, without data assimilation, by the Germany Federal Maritime and Hydrographic Agency (BSH). The model is configured with nested grids with a resolution of 5km in the North and Baltic Seas and a resolution of 900m in the German coastal waters. To improve the predictions of the HBM-ERGOM model, data assimilation was added by coupling the model to the parallel data assimilation framework (PDAF, http://pdaf.awi.de). The ensemble-based local error-subspace transform Kalman filter (LESTKF) is applied for the data assimilation. It is studied how the biogeochemical model fields are impacted by the assimilation of sea surface temperature (SST) data from the Sentinel 3a and NOAA satellites. Two cases are considered. First, the impact of weakly coupled data assimilation. In this case, the assimilation of temperature only directly influences the physical model variables in the analysis step while the biogeochemical fields react dynamically to the changed physical model state during the ensemble forecasts using the coupled model. The second case is the strongly-coupled data assimilation in which next to the physical model fields also the biogeochemical fields are directly updated in the analysis step through the multivariate covariances estimated by the joined physical-biogeochemical ensemble of model states. Here, it is assessed whether these covariances are sufficiently well estimated to result in an improvement of the biogeochemical fields. For the weakly-coupled assimilation it is found that while the biogeochemical model fields are influenced by the SST data assimilation, the averaged deviation from in situ data remains almost constant and small improvements, but also deterioration, can occur. In case of the strongly coupled assimilation, the vertical effect of the assimilation has to be constrained to avoid deterioration. This effect is caused by the ensemble-estimated correlations between SST and biogeochemical fields in lower layers of the model.

Description: Satellite data of both physical properties as well as ocean colour can be assimilated into coupled ocean-biogeochemical models with the aim to improve the model state. The physical observations like sea surface temperature usually have smaller errors than ocean colour, but it is unclear how far they can also constrain the biogeochemical model variables. Here, the effect of assimilating satellite sea surface temperature into the coastal ocean-biogeochemical model HBM-ERGOM with nested model grids in the North and Baltic Seas is investigated. A weakly and strongly coupled assimilation is performed with an ensemble Kalman filter. For the weakly coupled assimilation, the assimilation only directly influences the physical variables, while the biogeochemical variables react only dynamically during the 12-hour forecast phases in between the assimilation times. For the strongly coupled assimilation, both the physical and biogeochemical variables are directly updated by the assimilation. The strongly coupled assimilation is assessed in two variants using the actual concentrations and the common approach to use the logarithm of the concentrations of the biogeochemical fields. In this coastal domain, both the weakly and strongly coupled assimilation are stable, but only if the actual concentrations are used for the strongly coupled case. Compared to the weakly coupled assimilation, the strongly coupled assimilation leads to stronger changes of the biogeochemical model fields. Validating the resulting field estimates with independent in situ data shows only a clear improvement for the temperature and for oxygen concentrations, while no clear improvement of other biogeochemical fields was found. The oxygen concentrations were more strongly improved with strongly coupled than weakly coupled assimilation. The experiments further indicate that for the strongly coupled assimilation of physical observations the biogeochemical fields should be used with their actual concentrations rather than the logarithmic concentrations.