Assessment of Systematic Errors in the Computation of Gravity Gradients from Satellite Altimeter Data

Abstract

With satellite radar altimetry, the oceanic geoid can be determined with high precision and resolution. Double differentiation of these data along satellite altimeter ground tracks yields along-track gravity gradients that can be used to compute vertical gravity gradients at ground track crossovers. One way to counteract the noise amplification due to the differentiation is to smooth the data using smoothing splines. Although the effect of satellite altimeter data noise has been investigated to some extent, the associated systematic errors have not been assessed so far. Here we show that some of the systematic errors cannot be neglected. In particular, we found that the negligence of the dynamic ocean topography (DOT) may introduce errors that are greater than the measurement noise induced errors. If the gravity gradients are to be used for GOCE validation, then also in this case the DOT may not be neglected as the signal at GOCE altitude of 260 km may be above the GOCE requirements. In addition, we show that the altimetry derived gravity gradients cannot be compared one-to-one with those in a local Cartesian frame. The differences are small compared with the total signal, but they may be larger than the satellite altimetry induced stochastic errors and may be above the GOCE requirements. The cubic splines second derivative truncation error requires the use of 10 Hz altimeter data for the computation of gravity gradients at the Earth's surface, while 1 Hz data are sufficient for validation at GOCE altitude.