Abstract: The power, mean frequency and spectrum width of the Doppler echoes are closely related to the types and characteristics of meteorological phenomena. Spectral moments estimation is very necessary. Ground clutter is a server interference to weather radar. The weather echoes and ground clutter have power spectra with shape following closely to Gaussian function, which can be used to established parametric estimation method. A parametric model for estimating spectral moments is employed and a parametric spectral moments estimator is proposed. Then RELAX is used in this method dealing with the scenes of several mixed Gaussian spectrums. Finally, experimental results with simulated weather radar signals and performance analysis are given. It is verified the present estimator can estimated the spectral moments accurately. Unlike pulse pair method, it can be used to estimate the spectrum moments of weather echoes in presence of ground clutter. And it is comparatively efficient than Maximum Likelihood method.