Modifications to the [14] model for specular ocean emissivity have recently been suggested by [7] in order to improve the performance at high microwave frequencies. The work presented here tests both the original and modified models using a set of satellite and ground-based observations that is designed to eliminate as much as possible the dependence of the test on parameters other than the surface emission itself. Clear sky, low humidity, and low wind conditions were used exclusively to reduce

more »

... the dependence of the test on atmospheric and wind-roughened sea models. Radiosonde observations (RaObs) coincident with TOPEX satellite overpasses were used to reduce errors due to inexact knowledge of the atmosphere. Our tests confirm the superior performance of the Ellison model at higher frequencies. In an effort to remove the residual bias between the models and the observations, we also suggest a parameterized modification to both models that "best fits" the models to the data. In this case, the modified Ellison model maintains its superior performance at high frequencies, suggesting that it has an inherently more accurate frequency dependence. The root mean-squared (RMS) error in the modified Ellison emissivity model, over the range of 18-40 GHz, is found to be 0.0037, which translates into a model error of approximately 1 K in terms of brightness temperatures. Index Terms-Microwave remote sensing, ocean emissivity.