Neural Network Interpretation With Electromagnetic And Magnetic Data For Envlronmental Site Investigations

A feasibility study for the application of neural networks to electromagnetic and magnetic data interpretation was<br>done. The scope of \vork cntailcd three major parts: classification of data as target or background; estimation of<br>depth of targets; classification of the conductivity of the targets. The purpose of the study was to determine if<br>neural networks could be trained to find the same anomalies as a skilled human interpreter and also give<br>quantitative information that a human interpreter may or may not be able to give. We found the neural networks<br>capable of classifying the majority of the targets in the data set. Respectable depth and conductivity estimates were<br>also able to be made on these same targets.