Step 1: Source and receivers key features
The minimal field equipment for a seismoelectric experiment consists of:
Seismic source: a plate hit by a sledgehammer, a weight drop or even a vibrating source, as detailed in the next section;
Sensors: pairs of electrodes forming a dipole over which a voltage can be measured;
Cables conveying the signal back to the recording system;
Recording system;
Preamplifiers, sometimes used to buffer the signal before it is recorded.
Step 2: Example of seismoelectric data
The seismoelectric data have been recorded in a sedimentary environment by Garambois & Dietrich (2001). The data were collected in a river catchment area by the Fier river near the Annecy area. The stratigraphy of the site is marked by alluvial deposits with alternating layers of gravel, sand, and clay from the surface down to 150m depth. At the time of the experiment, the water table was located at a depth of about 1.5m. The source used here was a 200g dynamite charge buried at a depth of 1m. The data were recorded by 12 unevenly spaced dipoles, deployed on either side of the shot point, between offsets x=±5m and x=±30m. The record is dominated by coseismic Rayleigh surface waves.
Step 3: Data processing: Anthropic noise reduction
Seismoelectric data are often contaminated with anthropic noise, mainly 50 or 60Hz power-line noise, depending on the survey area. Several techniques have been proposed to reduce this power-line noise.
Figure provides an example of the block subtraction technique. This method implies recording for a few seconds while the seismic source is not active, typically just before activating the source. This "passive" recording, assumed to consist of pure harmonic noise, is used to estimate the ambient noise: it can then be duplicated over the entire duration of the actual seismoelectric recording before being subtracted from the data.
Figure displays from left to right: the raw data, the power-line noise estimate, the data after noise reduction
Step 4: Data processing: interface response extraction
The interface response can be extracted through digital filtering techniques. Most of these filtering strategies are based on the fact that the interface response, having a very high apparent velocity, appears as a horizontal event on the seismoelectric recording.
Frequency-wavenumber (f−k) filtering is currently used for interface response extraction. f-k filtering is easy to apply but suffers a number of limitations: it requires a good spatial resolution, i.e. a large number of traces, and it does not filter out the flat portions of the type I coseismic signal, such as reflection hyperbolas near the shot point. Last but not least, it may alter the amplitudes of the extracted IR.
Figure shows:
at the top: the previous data after power-line noise reduction and the associated f-k spectrum;
at the bottom: the data after filtering by a “pie-slice” filter and the associated f-k spectrum. Horizontal events are clearly visible. The events represent the interface response.
