Pre-processing and post-processing of Large Electrical Resistivity Tomography Datasets from Hydrogeophysical Monitoring

Pre-processing and post-processing of Large Electrical Resistivity Tomography Datasets from Hydrogeophysical Monitoring

Arnaud WATLET1,*, Olivier KAUFMANN2

1 University of Mons
Department of Geology and Applied Geology
9, Rue du Houdain
7000 Mons

*Corresponding author: arnaud.watlet@umons.ac.be, +32(0)65.37.46.13

Keywords: Electrical Resistivity Tomography, monitoring, multihydrogeophysics, geophysics,

Abstract

Electrical Resistivity Monitoring (ERT) monitoring has shown its efficiency to bring valuable information in many contexts. By producing time series datasets, it allows tracking diverse evolution processes such as groundwater changes, fluid infiltration or permafrost melting. However multiple parameters often make these large time series challenging to manipulate, e.g. rainy events or instrumental breakdowns. Rainy events are primordial when looking at water infiltration processes. They can become quite bothering if looking at longer processes such as seasonal groundwater changes. Temporary instrument breakdowns or removals act also as disturbance for large datasets. They create gaps that can be problematic during time-lapse inversion procedures.

In this context, pre-processing of those large datasets becomes primordial. We propose an approach based on time series analyses of the ERT data. By fitting a realistic hydrological model on each single measurement repeated in time, we are able to filter noise and filling gaps of an ERT dataset. In order to do so, data are hierarchically formatted to allow rapid access, updates and filters on specific parameters.

To support this approach, we present a case study from a hydrogeophysical monitoring that is set up at Rochefort Cave Laboratory (RCL) site. Daily ERT data acquisition is performed at RCL since Spring 2014 resulting in a very large dataset. This site is useful in many aspects to test our pre-processing approach. First, measurements are highly sensible to rainy events given the geological context showing a fist thin layer with high conductivity. This makes this site interesting for testing filters on rainy events. Second, multiple measurement interruptions occurred due to instruments replacements or breakdowns. Filling those gaps might therefore be needed.

In parallel, synthetic datasets have also been created for taking into account diverse hydrological models and geological contexts.