The sea spectrum proposed by Elfouhaily et al. is widely employed in ocean remote sensing. In this paper, we find that there exists some disagreements between the results estimated using the Elfouhaily’s angular spreading function and the empirical model NSCAT2 (which is regarded as a reliable reference in this paper). Firstly, the peak value of (k) calculated with Elfouhaily’s method monotone increases with wind speed. While the peak value of (k) estimated with the NSCAT2 first increases and then decreases with wind speed. Secondly, the wavenumber that corresponds to the peak value of (k) isn’t fixed at 363rad/m. To make the Elfouhaily’s angular spreading function agrees well with the empirical model, some modifications are made on it with numerical fitting in this paper. The numerical simulations have shown that the modified angular spreading function could provide better predictions.