Variations of self-potential and unsaturated water flow with time in sandy loam and clay loam soils

Accurate assessment of soil–water fluxes is essential in soil physics due to its direct implications in environmental, agronomical or hydrological applications. Field estimations of soil–water fluxes by ‘classical’ hydraulic methods are often difficult to obtain. Moreover, water fluxes are highly variable in space and time. The obtainment of a reasonable estimate for this variable would require numerous measurement sites. However, such a requirement is rarely met. Thony et al. [CR Acad. Sci. Paris, Earth Planetary Sci. 325 (1997) 317] presented the experimental evidence of a linear relationship between the self-potential (SP) and the unsaturated soil–water flux. Therefore, this relationship would allow the indirect assessment of the water flux using electrical measurements. Such an approach would appear much more flexible and easier to perform than the current hydraulic measurements. The aim of this study is to experimentally investigate the existence and robustness of the flux–SP relationship for different soil types and pedoclimatic conditions. The soil–water fluxes and the SP were monitored in a long-term experiment involving two types of soils, contrasting in hydraulic and electric properties. The soils were placed in lysimeters which were instrumented with tensiometers and TDR probes for monitoring hydraulic heads and moisture content, respectively. Unpolarizable SP electrodes, temperature sensors and suction cups (for collecting pore water) were also installed in the lysimeters. The SP and the fluxes were measured or calculated in the 30–40cm depth section. Results show that the variations of the SP with time were clearly linked to both rainfall events and evaporation. However, in the long-term, the linear relationship between the unsaturated water flux and the SP evolves from strongly correlated to almost not correlated. The slope (sensitivity) of the flux–SP relationship varies with the soil type, decreasing with more electrically conductive soil. Taking into account a varying soil–electrode contact greatly improves the flux–SP relationship at the scale of the rainfall event, particularly when considering infiltration and drainage phases separately. Nevertheless, at the scale of a year, with alternated rainfalls and evaporation phases, the robustness of the relationship decreases (i.e. the coefficients of the relationship vary between events). This variability could be related to time variations in electrical conductivity, not so much to that of th