Analysis of measurement errors from electrical resistivity imaging investigation of First World War mining tunnels in La Boisselle, France

ABSTRACT To derive reliable electrical resistivity subsurface models using error‐weighted inversion schemes, a meaningful and correct error model is required. An over‐estimated error leads to a lack of resolution and reduced target detectability. Furthermore, biased data are easily overlooked and can lead to artefacts and significant miss‐interpretation. We carried out an electrical resistivity tomography survey to detect mining tunnels of World War I in La Boisselle, France. French, British and German troops extensively used mining warfare, such as tunnel constructions, to undermine opponents. While the location and orientation of some British tunnels are known from archaeological excavations, the exact location of the German tunnels is currently unknown. Due to systematic measurement errors resulting from a malfunction in the system, the acquired electrical resistivity tomography data in La Boisselle were significantly biased. Therefore, a detailed systematic error analysis scheme was developed. Using a workflow of systematic error examination to identify biased data such as outliers or other bias, an unbiased dataset was retrieved. Subsequently, two‐dimensional electrical resistivity tomography inversions using different error models provided a qualitative estimate of how the data errors influence the tunnel detectability within an inversion scheme. The field data from La Boisselle demonstrates the importance of correctly estimating measurement errors, especially in view of the detection of small‐scale targets, such as tunnels.