Effects of Magnetic Soils on Magnetometry in Uxo Discrimination Problems

Magnetometry has emerged as an effective tool for UXO discrimination. The current approach is based upon the magnetic dipole moment recovered from the inversion of surface total-field magnetic anomalies. However, the reliability of inversion will be strongly influenced by the noise in the data such as that produced by magnetic soils in the background geology. We examine this problem to understand under what conditions the magnetic dipole inversion will be severely affected and, when such conditions occur, how to alleviate the effect. We achieve this by evaluating the errors of the dipole inversion due to different soil susceptibilities. The magnetic susceptibility of soils is modeled as a correlated random process whose spectral property is estimated using the TEM data from Kaho’olawe, Hawaii. The UXO response is modeled as that of a dipole embedded in the soil. Inverting the total-field response using a large number of soil realizations yields a reliable estimate of the errors. To remove the soil effect, we propose to preprocess the data before applying inversion. We show that a Wiener optimal filter can be used as a preprocessing tool to alleviate the effect of the soil response and reduce the errors of inverted dipole parameters, thus providing better confidence in discrimination.