To promote the application of terahertz active detection technology in cloudtarget, based on the characteristics of terahertz cloud radar system, the velocity and velocityspectral width estimation of cloud echo simulation data are studied. An self-correlationspectral moment estimation method based on frequency domain processing is proposed toimprove the estimation accuracy of velocity and velocity spectral width of cloud echo. Thesimulation results show that the trends of the velocity and velocity spectrum width obtainedby this method are roughly same with the change of the signal-to-noise ratio, compared withthe traditional spectral moment estimation method based on power spectral density. Andwhen the signal-to-noise ratio is between-5 dB and -5 dB, the velocity and velocityspectrum width estimation result is more accurate by this method.