Topographical data have significant influence on the high frequency component in gravimetry data. The high precision of terrain data forward method can basically reflect local gravity characteristics. In order to obtain accurate gradient data and build the gravity gradient reference database quickly, the fast frequency method for digital elevation model forward gravity gradient was investigated in detail. A novel Parker’s forward formula based on cosine transform has been deduced to calculate the full tensor gravity gradient, and then discusses accuracy and calculation time of the frequency method by numerical experiment. The simulation tests and analysis show that, the mean abstract error of gradient data forwarded by our Parker’s forward method is estimated to be less than 0.5E precision. The errors of this solution decreased by 3dB in comparison to Fourier transform Parker’s forward method. Comparing with spatial domain rectangular prism numerical methods, it gives better apparent advantages over calculating scale and efficiency.
