Detection of cavities and tunnels from gravity data using a neural network

Abstract

We have developed a simple approach to determine the depth and radius of subsurface cavities from the microgravity data. Horizontal location of cavity center is picked up by an algorithm as the projection of the minimum of gravity anomaly. Depth to the cavity center is determined using back propagation neural network. The cavity radius can be then calculated using the determined parameters if the density contrast between the host rock and the cavity filling materials is known or assumed according to the geological background. The present method is tested by several synthetic data sets and showed high ability to determine the cavity parameters in presence of natural noise. Applying the method to field data from Medford cavity site, Florida, the estimated cavity parameters are coincident the boring results. The method is proved to be fast and robust and can be used in field situation.