Modelling temperature effect in time lapse DC monitoring experiments through inversion of thermal diffusivity

In the recent years, time-lapse surveys have been performed widely to monitor, for instance, hydrogeological tracer experiments (Cassiani et al., 2006), groundwater watershed characterization (Miller et al., 2008; Deiana et al., 2018), seasonal variations (Hiblich et al., 2011; Musgrave and Binley 2011), landslide behaviour and evolution (Cassiani et al., 2009, Wilkinson et al., 2010), and so on. One of the main concerns, when resistivity surveys are performed, is to be sure to impute the variations to the right phenomena, distinguishing the electrical changes of interest from all the others, which are assumable as noise. Temperature variations might represent the main noise source in the time-lapse conductivity surveys since temperature has a strong impact on the resistivity parameters, hence on the inversion results. For example, seasonal temperature trends could mask the conductivity variations, and thus leading to misleading interpretations, up to the depths from the surface that can be reached by external fluctuations. Haley et al. (2007; 2009; 2010) have pointed out the importance of considering the temperature variations in time-lapse geoelectrical surveys, including in the inversion procedure a correction for this effect. In this study we intend to disentangle the temperature effect from resistivity variations inverting for the thermal diffusivity of the medium in a simultaneous time-lapse inversion that does not require direct temperature measurements below ground, both on a synthetic dataset and on-field experiments.