On the methods of evaluation of apparent resistivity under conditions of low message-to-noise ratio

In geoelectrical prospecting, especially for geothermal purposes, different methods of spectral analysis of noise-degraded voltage records have been published in recent years. In this paper we study the confidence limits of the methods by utilizing records obtained with a small dipole array at a fixed spacing. By progressive drops in the intensity of the energizing current, we simulated what really happens in the field at the largest spacings: a progressive deterioration of the message-to-noise ratio (s/r). However, by that small array we were able to determine the true apparent resistivity with the maximum available current intensity. The comparison with the true value allowed us to establish that: (i) as far as the message is “physically” present, all methods appear equivalent and sufficiently reliable. By virtue of its simplicity, stacking is the method which best fits for a ready analysis directly in the field; (ii) when the s/r drops below a certain threshold, the mathematically extracted message loses its physical meaning. The results derived from the diverse methods not only do not conform with each other, but are also far different from the true value. The progressive maximum likelihood is at present the only method that in the worst cases pinpoints the unreliability of the result.