May 29, .2012: BM-Seminar

13:30 Seeburg seminar room

Abstract:

Estimates of the Mean Dynamic Topography (MDT) are obtained from the combination of various satellite observations. The MDT provides valuable information for ocean models: via the principle of geostrophy, the MDT is equivalent to surface velocity information. Along with several other data sets, e.g. temperature and salinity data, the MDT is assimilated into the inverse ocean model IFEOM in a weighted least squares sense. In theory, the inverse error covariance matrix of the MDT data represents the appropriate weight for the MDT data within the optimization. However, the available error estimate is a formal one and requires additional tuning by a weighting factor. Different approaches for the determination of this weighting factor are presented. The inverse error covariance matrix is used to derive a smoothing operator S. An identity matrix I = S-1S is introduced into the satellite observation equations. This leads to a modified diagonal weighting matrix and to a modified MDT data set S·MDT in the normal equations. When the modified MDT data S·MDT is assimilated into the ocean model using the modified weighting matrix, the ocean model result is changed significantly. The smoothing operator S represents a natural filter for the MDT data, and no more or less arbitrary choice of filter width is required any longer.