Search

In closed hazardous waste (i.e. industrial) landfills, impermeable covers are used to seal in the waste so as to minimise water infiltration and the accumulation of leachate inside the waste. In this paper, we present a geophysical study performed on a French landfill where the cover was designed using impermeable (clay) and drainage (sand) layers along with a Geosynthetic Clay Liner (GCL) in between. As observed, the quantity of leachate increases after rain events making the leachate treatment more expensive. This could be due to areas of weakness in the cover that have developed over time or which date back to its installation. Three different geophysical methods were employed and confronted to detect and assess such damage, and, if necessary, to help choose the most suitable remediation of the cover: Automatic Resistivity Profiling (ARP) which enables a rapid mapping of the entire landfill; the self‐potential method (SP) as one of its sources is water drainage; and electrical resistivity tomography (ERT) to obtain 2D resistivity models. Based on the ARP results revealing high lateral heterogeneities of the superficial cover materials, the SP and ERT measurements were conducted on a limited area. A negative SP anomaly was observed at the top of the landfill, where the cover is the thinnest and the GCL the most damaged, suggesting preferential infiltration under the GCL. The 2D ERT profile showed both vertical and horizontal variations. Comparison with manual auger holes showed that the alternating clayey and sandy layers did not conform to plans for the original cover when it was installed at this site about twenty years ago. Variations in electrical resistivity noticed at the depth of the GCL were attributed to possible damage to the GCL and a greater than expected permeability since the GCL is normally very resistive. The results obtained in this study were important to understand the damage to the cover due to the use of different types and thicknesses of cover materials compared to what was originally planned, and deterioration of the GCL.

Electrodes and cables are sometimes embedded or buried permanently into the ground in order to minimize electrode location errors during long electrical resistivity monitoring. This procedure is efficient and useful. In order to evaluate the feasibility and suitability of this technique for a long monitoring of water content in a rock mass, electrical resistivity data have been repeatedly acquired from a buried array. Forty‐eight steel electrodes composed the array, connected to two multicore cables, which have been buried in a trench since January 2006. A fast resistivity‐meter was used to carry out dipole‐dipole electrical measurements, over a period of one year, starting in December 2006. Data acquisitions of repeated resistivity measurements have been realized in order to distinguish between variations in electrical resistivity due to noise and to changes in water content. Around once every month, successive dipole‐dipole array measurements on the same day (between three and eleven arrays with around 10 minutes for one array) have been performed. Using all the pseudosections recorded the same day, the coefficients of variation have been calculated for each data point. Results show that the variability of the data is six times greater when the data from the first array of the day‐series are taken into account. The data collected from this first array are therefore significantly noisier than the following measurements and must be removed for correct interpretation. We propose two main explanations for this effect: 1) polarization of the system array/clay at the time of the first acquisition and 2) damage encountered into the buried cables. Despite this damage, we have shown that the electrical data can be consistent and correctly exploited if the first acquisition of the day is not taken into account. In conclusion, on the basis of the results presented here, we recommend that further studies be made using buried equipment to systematically carry out several acquisitions before starting the long monitoring. We also recommend to remove data from the first acquisition whenever they are found to be significantly noisier than data of the following acquisitions.