The accuracy of satellite altimetric measurements of sea level is limited in part by the influence of ocean waves on the altimeter signal reflected from the sea surface. The difference between the mean reflecting surface and mean sea level is the electromagnetic bias. The bias is poorly known, yet for such altimetric satellite missions as the Topography Experiment (TOPEX)/Poseidon it is the largest source of error exclusive of those resulting from calculation of the satellite's ephemeris. Previous observations of electromagnetic bias have had a large, apparently random scattering in the range of 1--5% of significant wave height; these observations are inconsistent with theoretical calculations of the bias. To obtain a better understanding of the bias, we have measured it directly using a 14-GHz scatterometer on the Chesapeake Bay Light Tower. We find that the bias is a quadratic funcition of significant wave height H1/3. The normalized bias β, defined as the bias divided by the significant wave height, is strongly correlated with wind speed at 10 m, U10, and much less strongly with significant wave height. The mean value for β is -0.034, and the standard deviation of the variability about the mean is ¿0.0097. The standard deviation of the variability after removing the influence of wind and waves is ¿0.0051=0.51%. The results are based on data collected over a 24-day period during the Synthetic Aperture Radar and X-Band Ocean Nonlinearities (SAXON) experiment from September 19 to October 12, 1988. During the experiment, hourly averaged values of wind speed ranged from 0.2 to 15.3 m/s, significant wave height ranged from 0.3 to 2.9 m, and air minus sea temperature ranged from -10.2¿ to 5.4 ¿C. Because U10 can be calculated from the scattering cross section per unit area &sgr;0 of the sea measured by spaceborne altimeters, we investigated the usefulness of &sgr;0 for calculating bias.

We find that β is strongly correlated with &sgr;0 and much less strongly with H1/3. The standard deviation of the variability after removing the influence of the radio cross section and waves is ¿0.0065=0.65%. The results indicate that electromagnetic bias in radar altimetry may be reduced to the level required by the TOPEX/Poseidon mission using only altimetric data. We find, furthermore, that the relationship between &sgr;0 and wind speed agrees with previously published power law relationships within the accuracy of the measurement. The mean value of β, its variability, and the sensitivity of β to wind speed all agree well with previous measurements made using a 10-GHz radar carried on a low-flying aircraft. The mean value of β, its variability, and the sensitivity to wind were all significantly larger than previous measurements made using a 39-GHz radar also carried on a low-flying aircraft. All experiments included a similar range of wind speeds and wave heights. The SAXON data were, however, much more extensive, and the statistical relationships correspondingly more significant. The mean value of β is very close to the mean value determined from global measurements of sea level made by Geosat. ¿American Geophysical Union 1991