Nonlinear inversion of gravity data for simple geometry anomalies using the Centred-Progressive Particle Swarm Optimisation algorithm

Depth estimation of gravity anomalies is one of the most important geophysical problems in the exploration of mineral deposits. In the present paper, we try to estimate mass anomaly depth by using an artificial intelligence method called the Centred-Progressive Particle Swarm Optimisation (CP-PSO) with sample shapes such as a sphere, horizontal cylinder and vertical cylinder, which simulate the shape of most causative bodies. Using an artificial intelligence method is common for the case in ore bodies detection and delineation. When modelling gravity data, we estimate the depth and shape factor; therefore, we suppose depth (z) and shape factor (q) as particles in this algorithm. This technique was tested for synthetic models contaminated with random noise and the results are quite acceptable and promising. The proposed method was also successfully applied to real mineral exploration data. The desired location is near one of the outcroppings of the Safo manganese mine located in north-western Iran. The results show that the estimated depth and the shape factor model were in good agreement with the results obtained through the Euler method and drilling.