A method for splitting sea surface height measurements from satellite altim
etry into geoid undulations and sea surface topography is presented. The me
thod is based on a combination of the information front altimeter data and
a dynamic sea surface height model. The model consists of geoid undulations
and a quasi-geostrophic model for expressing the sea surface topography. T
he goal is the estimation of those values of the parameters of the sea surf
ace height model that provide a least-squares fit of the model to the data.
The solution is accomplished by the adjoint method which makes itse of the
adjoint model for computing the gradient of the cost function of the least
-squares adjustment and an optimization algorithm for obtaining improved pa
rameters. The estimation is applied to the North Atlantic. ERS-1 altimeter
data of the year 1993 are used. The resulting geoid agrees well with the ge
oid of the EGM96 gravity model.