NON-PARAMETRIC SEA-STATE BIAS MODELS AND THEIR RELEVANCE TO SEA LEVEL CHANGE STUDIES

The direct estimation of sea-state bias (SSB) from sea height residuals is extended with a parametric tting process and a successive smoothing of the remaining residuals. This hybrid method essentially produces a nonparametric SSB model in the form of a smooth grid in a 2dimensional space determined by signicant wave height and backscatter coefcient. The hybrid method allows a much higher resolution than parametric models, without the disadvantage of the direct method’s limited wind speed / wave height domain. The use of sea height residuals as input data allows estimation of a realistic SSB model with only a few months of data. With state-of-the-art geophysical corrections the hybrid SSB model can be constructed with a high level of accuracy. The impact of errors in the mean sea surface model and tide models appears marginal. Errors in the ionospheric correction, which has a geographical distribution similar to wind speed and wave height, tend to leak into the SSB model. The wave height and backscatter both show trends as a function of time. Wave heights appear to drop over time, but backscatter shows trends either way. This results in trends in SSB and hence sea level that may not be ireali. The different trends in SSB between TOPEX Side A and Side B, as well as between the CSR SSB model and our hybrid model, poses challenges on the estimation of sea level change to better than 0.2 mm/yr.