Improvement of the computation of the kernel matrix with a combined method in inversion of gravity data

Modeling of gravity data includes forward and inverse modeling techniques. To solve the forward problem, we divide the half space below the ground surface into small blocks (cells) with constant density contrast and compute the kernel matrix. There are various analytical and numerical methods for calculation of the kernel matrix. Analytical methods have good accuracy but are time consuming. Numerical methods are faster and less accurate than analytical ones. The Plouff analytical method is an accurate method for computation of the kernel matrix but it is time consuming. The point mass numerical method needs less computation time than that of the Plouff method. However, for calculation of the first two rows of the model parameters, this method does not have good accuracy. In this paper, a new combined method is presented in which the Plouff and point-mass methods are combined, and as a result, the computational speed is improved while a good accuracy is also obtained. In this research, the first two rows of the model parameters are calculated analytically, and then, the next rows are calculated by point-mass numerical method. To validate the proposed method, it has been tested based on a synthetic model and a real gravity data from the Ovoid Ni- Co- Cu deposit in Canada. The obtained results show that the computation time improves significantly while the error is very small with respect to the cell dimensions.