oa Noise Reduction in a Multi-Channel SP Monitoring System

Self-potential (SP) monitoring is attracting growing interest as a tool for inferring groundwater flow, both in saturated media and in the vadose zone. Multi-channel recording systems are being used to acquire data from large arrays of electrodes with the simultaneous temporal sampling desirable for data processing and Interpretation. We present an analysis of the noise performance of one relatively common commercially available datalogger (Campbell Scientific CR1000, with AM32B multiplexer) in SP monitoring applications and illustrate the improvements realized by incorporation of passive RC filters. Our filters, consisting of a 10 kOhm resistor (Rf) in series with each input and a 10 microfarad capacitor (Cf) across each differential input pair, provide a lowpass filter with a cut-off frequency of 0.8 Hz. This corner frequency is lowered further by the source impedance Rs (sum of the electrode contact resistances) giving a lowpass corner of approximately 1/(2pi(2Rf + Rs)Cf), or 0.5 Hz for Rs = 10 kOhm. We suspect that the principal improvement arises from the attenuation of powerline noise at 60 Hz and its harmonics prior to sampling. Such noise can commonly be hundreds of times stronger than the SP signals of interest. Although the datalogger uses analog Integration over 60 Hz periods to effectively attenuate stable powerline harmonics, it remains vulnerable to instabilities in harmonic amplitude or phase, and to settling time artifacts that can arise on switching between channels with vastly differing voltages. Our limited experience to date indicates that the benefits of pre-filtering are somewhat site dependent. the filters proved critical during recent SP monitoring of water well pumping where they reduced noise levels by at least a factor of 10, allowing the identification of coherent sub-milliVolt temporal variations in SP over an array of 34 electrodes distributed around the pumping well on surface.