Assimilation of SeaWiFS data in a coupled physical-biological model of the Adriatic Sea
In: Journal of Marine Systems. Elsevier: Tokyo; Oxford; New York; Amsterdam. ISSN 0924-7963; e-ISSN 1879-1573, more
Also appears in:Grégoire, M.; Brasseur, P.; Lermusiaux, P.F.J. (Ed.) (2003). The use of data assimilation in coupled hydrodynamic, ecological and bio-geo-chemical models of the ocean. Selected papers from the 33rd International Liege Colloquium on Ocean Dynamics, held in Liege, Belgium on May 7-11th, 2001. Journal of Marine Systems, 40-41. Elsevier: Amsterdam. 1-406 pp., more
|  |
| Keywords |
Aquatic communities > Plankton > Phytoplankton
Aquatic communities > Plankton > Zooplankton
Models
MED, Adriatic [Marine Regions]
Marine/Coastal |
| Author keywords |
SeaWiFS; 3-D Nutrient-Phytoplankton-Zooplankton (NPZ) model; AdriaticSea |
| Authors | | Top |
- Garcia-Gorriz, E.
- Hoepffner, N.
- Ouberdous, M., more
|
|
|
| Abstract |
We use the variational adjoint method to assimilate chlorophyll concentrations derived from daily Sea-viewing Wide Field-of-view Sensor (SeaWiFS) maps in a coupled physical–biological model of the Adriatic Sea. The model consists of a 3-D Nutrient–Phytoplankton–Zooplankton (NPZ) model coupled with a primitive-equation physical model. We estimate the optimal values of a subset of four parameters obtained for two groups of assimilation experiments for January and June 1998 separately. We find that after assimilation of SeaWiFS data the biogeochemical model describes better the bloom conditions in January 1998 than the non-bloom conditions in June 1998. In January 1998, we recover a common parameter set for the entire Adriatic, the north and south basin experiments. In contrast, for June 1998, the misfits associated are larger for the entire Adriatic Sea and the north basin, whereas the south basin presents the lowest misfits of our experiments. Additionally, we obtain a common parameter set for both January and June 1998 in the north basin, consistent with results by other authors in the area. We also evaluate with twin experiments the uncertainty induced in the recovered control variables by several sources of error associated with the assimilation of daily SeaWiFS images. The SeaWiFS sampling is the main source of uncertainty in the retrieval of a control variable (11%) when the sources of error are tested individually. The correct value of the rest of parameters in the biogeochemical model is more critical to the retrieval of the true value of a control variable than the error in the assimilated data. When the sources of error are combined and the four control variables of the SeaWiFS assimilation experiments are recovered simultaneously, the highest uncertainty for a control variable is 21%, that corresponds to the herbivorous grazing. |
|
