In this paper, we introduce a new method called Modular Feed-forward Neural Network (MFNN)

Tilt angle filter is an interpretation method that is used to determine the source borders locations from potential fields data. Moreover, the tilt angle is applied for estimation of the anomaly source depth, such as contact-depth method and tilt-depth method. In this paper an application of the tilt angle technique obtained from the first vertical and horizontal gradients of the gravity anomaly from semi-infinite vertical cylindrical source is described. The technique is based on the tilt angle and derivatives ratio. In this approach the depth estimates are proportional to the computed tilt angles and their distances from the cross section center of the anomaly cause on the surface. This new method is termed the tilt-distance-depth (TDD). The method is demonstrated using synthetic gravity data, with and without random noise, and real gravity data from Iran. The results are also compared with the solutions from Euler deconvolution technique and inverse modelling using Modelvision software.

No abstract

Determination of potential fields' anomaly borders is a useful help to their interpretation. There is various technique of edge detecting that is applied in image processing. In this paper, the canny edge detection (CED) method has been proposed as boundary enhancement of the magnetic and gravity potential field data. The Canny operator works in a multi-stage process. This method is based on the characteristic of intensity values of considered pixel. The edge detector should have a good signal-to-noise ratio, so that edges can be found even if potential field data quality is poor. For 2-dimensional bounds, residual potential field map is first smoothed by using a 2-D Gaussian filter. Afterwards, computing the horizontal gradients of the smoothed map and then using the gradient magnitude and direction to estimate borders strength and direction at every pixel. The Canny edge detection algorithm uses double threshold for edges revelation. In this research, a new procedure to define the thresholds has been suggested. The results obtained from the synthetic data set, with and without random noise, have been discussed. The method is demonstrated on real gravity and magnetic data set surveyed from Iran. The CED results are compared with three common methods as edge detector, namely the analytic signal, tilt angle and total horizontal derivative of the tilt angle.

In this paper, an inversion method based on the Marquardt's algorithm is presented to invert the gravity anomaly of the simple geometric shapes. The inversion outputs are the depth and radius parameters. We investigate three different shapes, i.e. the sphere, infinite horizontal cylinder and semi-infinite vertical cylinder for modeling. The proposed method is used for analyzing the gravity anomalies from assumed models with different initial parameters in all cases as the synthetic data are without noise and also corrupted with noise to evaluate the ability of the procedure. We also employ this approach for modeling the gravity anomaly due to a chromite deposit mass, situated east of Sabzevar, Iran. The lowest error between the theoretical anomaly and computed anomaly from inverted parameters, determine the shape of the causative mass. The inversion using different initial models for the theoretical gravity and also for real gravity data yields approximately consistent solutions. According to the interpreted parameters, the best shape that can imagine for the gravity anomaly source is the vertical cylinder with a depth to top of 7.4 m and a radius of 11.7 m.