Affected by external environment, flight path of airborne Synthetic Aperture Radar(SAR) is always inaccurate. Under the circumstance of limited accuracy of navigation system, estimation of Doppler centroid frequency from airborne SAR data is necessary. Error of Doppler centroid frequency estimation decides the linear phase error, affects the estimation of high order phase error and azimuth compression quality. In airborne terahertz video-SAR system, submillimeter motion compensation accuracy is required, which tremendously increases the importance of Doppler centroid frequency estimation. Traditional Doppler center frequency estimation methods usually have good results in the frontal view or small squint mode of SAR system, but the deviation is often large in the mode of high squint angle. In order to effectively complete range walk correction in multiple modes, based on the measured data of THz video-SAR and starting from traditional envelope estimation method, an improved envelope estimation method is promoted. Compared with two traditional methods, in front view mode, all methods mentioned can give good results, but in scan SAR mode, 97% images can be well-estimated and focused with promoted method better than with other methods. By comparison, the proposed method exhibits higher robustness and efficiency. This work can help research on SAR imaging under high frequency.