Begin Main Content

Lars Nerger and Watson W. Gregg

Satellites observe the color of the ocean.  These observations provide information about life in terms of plankton near the ocean surface.  In general, the greener the ocean surface looks, the more plankton are present.  However, knowing just the color of the ocean surface is insufficient to understand the ocean ecosystem, which plays an important role in the global carbon cycle.  

Ocean color observations can be assimilated into the NASA Ocean Biogeochemical Model (NOBM) which has been developed at the GMAO and simulates different kinds of plankton and nutrients.  The model can simulate the ocean biology also in regions where no observations are available, for example below the ocean surface or at locations where clouds obscure the observation.  The model by itself is only an approximation of the true ocean biology.  However, in combination with ocean color observations, the maximum information of both the model and the observations can be obtained.

This study continues earlier assimilation studies using the NOBM. Daily surface chlorophyll concentrations are estimated by assimilating chlorophyll data from the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) using an ensemble-based singular “evolutive” interpolated Kalman (SEIK) filter, which is simplified here by the use of a static error covariance matrix.  The filter operates with a localized filter analysis and is amended by an online bias correction scheme.  The performance of the filter algorithm is assessed by comparison with independent in situ data over the 7-year period 1998--2004.  The RMS log error, the bias and the correlation coefficient of logarithmic fields are considered for the performance evaluation.  The assimilation provides surface chlorophyll concentrations with about 3.3% lower error than SeaWiFS data.  While the error is only slightly lower on a global scale, the assimilation shows significantly smaller errors than SeaWiFS data in many of the major oceanic basins.  The assimilation estimate exhibits a very small bias of -0.025 log chlorophyll.  The correlation with the in situ data is increased by the assimilation from 0.52 for the free-run model to 0.83, which is slightly larger than the correlation of 0.81 of SeaWiFS data.

(click on image to enlarge)

Figure caption: Comparison of the surface chlorophyll from model and SeaWiFS with in situ data globally and separated over 12 major oceanographic basins for 1998-2004. Top: RMS log error. Middle: Mean 7-year bias of log concentrations. Bottom: Correlation coefficient. Shown are values for the free-run model (blue), the assimilation estimate (green), and SeaWiFS data (red). Below the panel at the top the number of collocation points is shown for the model and the SeaWiFS data.

Reference

Nerger, L. and W.W. Gregg, 2007: Improving Assimilation of SeaWiFS Data by the Application of Bias Correction with a Local SEIK Filter, J. Mar. Syst. (submitted).