Contaminant plumes (e.g., associated with leakages from municipal
landfills) provide a source of natural electrical potentials (or
``self-potentials'') recordable at the Earth's surface. One
contribution to these self-potentials is associated with pore water flow
(i.e., the ``streaming potential''), and the other is related to redox
conditions. A contaminant plume can be regarded as a ``geobattery'':
the source current potentially results from the degradation of the
organic matter by micro-organisms, which produces electrons. These
electrons are then carried by nanowires that connect bacteria and
thorough metallic particles that precipitate in areas of strong redox
potential gradient. In the case of the Entressen landfill (South of
France), reported here, the hydraulic head differences measured in
piezometers outside the contaminant plume is strongly linked to the
surface self-potential signals, with a correlation coefficient of -0.94.
We used a Bayesian method that combines hydraulic head and
self-potential data collected outside the contaminated area to estimate
the streaming potential component of the collected self-potential data.
Once the streaming potential contribution was removed from the measured
self-potentials, the correlation coefficient between the residual
self-potentials and the measured redox potentials in the aquifer was
0.92. The slope of this regression curve was close to 0.5, which was
fairly consistent with both finite element modelling and the proposed
gcobattery model. (C) 2007 Elsevier B.V. All rights reserved.