Gravity gradient tensor (GGT) is used in many interpretation methods. GGT are sensitive to noise. Therefore, noise reduction is vital to calculate Gradient’s. GGT are used to edge detection by curvature gravity gradient tensor (CGGT) method. The gravity data should as far as possible be clear the noise before use in the interpretation. We use singular value decomposition (SVD) to reduce noise in gravity data. In order to consider reduce noise we used GGT and edge detection CGGT after the noise reduction. We used SVD method in real data and synthetic data with noise. The results of GGT and CGGT show that SVD method has been effective in reducing noise.