We present an optmization to Regularized Gravimetric Inversion, proposed by Ramos (2013), to obtain estimates of the basement relief of sedimentary basin computationally more efficient. This procedure proposed performs nonlinear inversion using similar formulation to the linear inversion at each iteration. In this procedure is used a similar matrix to sensitivity matrix and our computational optimization simply consist in not evaluate the elements of this matrix that would receive value close to zero, assigning zero value to these elements, thus eliminating costly calculations. We show in numerical tests that the significant influence of the gravitational anomaly is only generated by the proof mass if it is located in some distance from the observation point. This numerical observation allowed us to establish which elements of the matrix used should be evaluated and which can simply receive zero value. With this built matrix, we evaluated in synthetic tests the effectiveness of the estimation in recovering the true relief. We compare the computational time to estimate the basement relief of synthetic models between the original procedure proposed by Ramos (2013) and the optimized procedure that we proposed. The results showed computational time savings, although if the number of parameters is small, demonstrating the efficiency of the proposal.
|
