Measurements of normalized radar cross sections of wind-generated waves were made at X band for both vertical and horizontal polarization for incidence angles of 10¿, 28¿, 48¿, and 68¿. The study, conducted in the Naval Research Laboratory wind-wave facility, sought to measure effects on the backscatter of varying water temperature, wind speed, and wind stress. The cross-section measurements were averaged for 2.13 min simultaneously with wind speed and wind stress. Air and water temperature were measured periodically with mercury thermometers. The results were compared with the empirical model functions developed for the Seasat-A satellite scatterometer, SASS I, and SASS II, and with the physically based models of Durden and Vesecky, Plant, and Donelan and Pierson. In order to use the SASS I and SASS II models for these comparisons, differences between Ku and X bands were assumed to be small; where possible, the other models were evaluated at X band. It was found that none of the models was consistently accurate at all wind speeds, incidence angles, and polarizations. Part of the inconsistency can be assigned to the fact that the models were developed for open ocean conditions with much higher sea states.

However, Plant's model can easily be adjusted to account for this effect by using a relationship between mean squared slope and wind stress appropriate for tank conditions. When this was done, the model did fit the data better but the improvement was not dramatic. This indicates that inaccuracies in the models are probably due to other factors as well. When plotted versus 19.5-m winds on a log--log scale, the measured cross sections do not fall on straight lines. Thus a power law dependence of cross section on wind speed is not a good representation of that relationship over our whole wind speed range. The data exhibit large variations at low wind speeds, however, which are not related to system noise. This may indicate that the statistics of backscatter depend markedly on wind speed. When these low wind speed data are omitted, a power law was found to fit the remaining data rather well. Although the water temperature was varied from 9¿C to 36¿C when the measurements were made at a 48¿ incidence angle, no temperature dependence was detectable above the low-wind-speed variability. The wave tank data compare well enough with 10 GHz, 3.0 cm (X band) aircraft measurements, and with the 14.6 GHz, 2.1 cm (Ku band) satellite data used in the SASS II model to cast doubt on the hypothesis that cross section depends on antenna altitude. ¿ American Geophysical Union 1992