Reduction of geoid gradient error in ocean variability from satellite altimetry

Abstract

Altimeter measurements of sea‐level variability have errors due to the altimeter not repeatedly sampling the same point on the ocean surface. The errors are proportional to the local slope of the mean sea surface. Accurate removal of geoid error is essential if altimeter data are to be used to study the relationship between geostrophic turbulence and bathymetry. The error can be reduced by using an accurate model of the mean surface. We use the multiyear TOPEX altimeter data set to develop a model for the mean sea surface along the TOPEX/POSEIDON ground track by estimating the coefficients of a local plane centered on every 2 km x 7 km bin sampled by the altimeter. We have evaluated the ability of this model. compared against two global mean sea‐surface models, to reduce the error associated with steep gradients. The two global models are the Center for Space Research 1995 model and the Ohio State University 1995 model. The three models show similar variability over the oceans, and none shows the large residual errors that can be seen in collinear analysis near some seamounts and trenches. The standard deviation of the variability using the plane model, however, is consistently smaller in low‐variability, high‐geoid‐gradient areas, indicating a slightly better performance than the two global models.