PROCESSING AND MODELLING OF AMT DATA FOR MINERAL APPLICATIONS 

Xavier Garcia, Alan G. Jones
Geological Survey of Canada
615 Booth St., Ottawa, Ontario, K1A 0E9, Canada
garcia@cg.NRCan.gc.ca

The high-frequency magnetotelluric method, AMT, is currently being 
widely used for mining exploration, especially in Canada. However, 
there are still some aspects regarding its implementation that need to 
be improved. These range from signal detection and time series 
processing to response function analysis to appreciation of 
three-dimensional effects.  The main natural electromagnetic source at 
the range of frequencies covered by mining-scale MT, audio-frequencies 
of 10 Hz - 20 kHz, is the global system of lightning. Due to the 
physical characteristics of the Earth's ionosphere and atmosphere, 
there is a minimum in the electromagnetic spectrum around 
1,000-3,000 Hz, especially during the daytime. This AMT "dead-band" is 
exactly the frequency range that is first sensitive to the presence of 
a typical economic target body. Another approach is the use of a 
remote site that can help to obtain a prediction of the magnetic field 
from the telluric fields.  Some ore deposit exploration are being 
carried out in areas where there is existing mining activity, thus the 
data can be seriously affected by noise. The classical processing 
schemes are based on either the Fourier or the windowed Fourier 
transforms, and these methods do not readily separate noise from 
signal in the actual conditions. The application of the wavelet 
transform offers an analysis of the time series at the frequency and 
time domains simultaneously. This analysis can be use to locate events 
in the time series that can be use in a posterior processing to obtain 
reliable estimates in that dead band.  In this paper we describe our 
efforts in these two aspects of processing MT data for mining 
exploration.