Abstract :
Sea-viewing Wide Field-of-view Sensor (SeaWiFS) chlorophyll data were assimilated with an established three-dimensional global ocean model. The assimilation improved estimates of chlorophyll relative to a free-run (no assimilation) model. Compared to SeaWiFS, annual bias of the assimilation model was 5.5%, with an uncertainty of 10.1%. The free-run model had a bias of 21.0% and an uncertainty of 65.3%. In situ data were compared to the assimilation model over a 6-year time period from 1998 through 2003, indicating a bias of 0.1%, and an uncertainty of 33.4% for daily coincident, co-located data. SeaWiFS bias was slightly higher at − 1.3% and nearly identical uncertainty at 32.7%. The free-run bias and uncertainty at − 1.4% and 61.8%, respectively, indicated how much the assimilation improved model results. Annual primary production estimates for the 1998–2003 period produced a nearly 50% improvement by the assimilation model over the free-run model as compared to a widely used algorithm using SeaWiFS chlorophyll data. These results suggest the potential of assimilation of satellite ocean chlorophyll data for improving model results.
