The Doppler spectrum modeling of the sea clutter is of great significance for using the Doppler radar to suppress the sea clutter and detect the target on sea surface. Taking into account the characteristics of the Doppler spectrum components corresponding to three scattering mechanisms, the Bragg, whitecap, and breaking-wave, we propose a time-varying Doppler spectrum model of radar sea clutter in this paper. The Doppler spectrum shifts and bandwidths of the three scattering mechanisms are considered respectively, and a Doppler shift of an added wave speed is introduced into this model. Because the spectrum intensity is defined as a random variable being a function of the measuring time of the sea clutter series, the model has the ability to model both the long-time averaged Doppler spectra and the short-time Doppler spectra. By modeling the short-time Doppler spectra and the long-time averaged Doppler spectra of the shore-based radar sea clutter measured in the China Yellow Sea at P and S bands, it is indicated that the proposed model has a higher precision than the conventional model, and especially, the modeling errors decrease significantly in the cases of the short-time Doppler spectra with long measuring time and the long-time averaged Doppler spectra with complex shapes.