The LASI High-Resolution Electromagnetic Sounding System

Since 1986, we have developed a series of electromagnetic (EM) sounding systems at the Laboratory for Advanced Subsurface Imaging (LASI). These may be categorized by their frequency ranges: low frequency (LF, 30 Hz to 30 kHz), medium frequency (MF, 1 kHz to 1 MHz), and high frequency (HF, 30 kHz to 30 MHz). We have applied these EM systems to a variety of targets, including: mapping of plumes of contaminated groundwater, tracing leachate for an in-situ leaching operation, locating buried hazardous waste, identifying subsidence-prone zones and detecting buried voids. These systems are also applicable to exploration for ore bodies. Currently, our sounding system uses a vertical magnetic field vector. Previous systems of this type required a slow and relatively inaccurate manual rotation of the sensing coils. The systems also used occasional calibrations of the measuring system. We have developed a 3-dimensional mathematical rotation procedure for the data, as well as a simultaneous, high-accuracy, calibration method. These new developments have greatly increased the speed of the measurements, as well as improving the accuracy of the ellipticity data. We use stepped-frequency measurements along with narrow-band filtering to allow accurate measurements, even in high-noise areas. We typically obtain ellipticity measurements at eleven frequencies and along densely sampled profile lines. We use neural networks to help interpret the measured ellipticity patterns. Neural networks can be trained to relate the measured ellipticity patterns to earth resistivities or to discrete targets buried in the earth. The neural network training may use theoretical EM responses calculated from computer codes for relatively simple structures. The theoretical EM response for many complex targets that are of interest cannot currently be reliably calculated, even with the largest supercomputers available. We have therefore chosen to do some of the training using theoretical EM responses generated in a full-scale modeling tank located at our Avra Valley Geophysical Test Site. We have also developed data-visualization software to help interpret the large data sets produced by this system.