摘要：Obtaining horizontal edges and the buried depths of geological bodies, using potential feld tensor data directly is an outstanding question.The largest eigenvalue of the structure tensor is one of the commonly used edge detectors for delineating the horizontal edges without depth information of the potential feld tensor data.In this study, we presented a normalized largest eigenvalue of structure tensor method based on the normalized downward continuation (NDC) to invert the source location parameters without any priori information.To improve the stability and accuracy of the NDC calculation, the Chebyshev–Pade´ approximation downward continuation method was introduced to obtain the potential feld data on diferent depth levels.The new approach was tested on various models data with and without noise, which validated that it can simultaneously obtain the horizontal edges and the buried depths of the geological bodies.The satisfactory results demonstrated that the normalized largest eigenvalue of structure tensor can describe the locations of geological sources and decrease the noise interference magnifed by the downward continuation.Finally, the method was applied to the gravity data over the Humble salt dome in USA, and the nearbottom magnetic data over the Southwest Indian Ridge.The results show a good correspondence to the results of previous work.
