Building simplified uncertainty models of object depth measurement by ground penetrating radar

Abstract

Ground Penetrating Radar (GPR) is often regarded a prospecting tool of underground targets rather than rigorous measurement tool where measurement uncertainty must be accompanied. This paper attempts to make use of GPR as a measurement tool by developing a simple model and evaluating the uncertainty of underground target’s depth estimation in two ways: (1) prior to GPR survey and (2) after GPR survey, for practical use by the partitioners. For (1), the expected uncertainty prior to GPR survey was modelled based on the estimation of the assumed host material’s dielectric permittivity and the assumed two-way travel time subject to required target depth. For (2), the expanded uncertainty of the estimated cover depth by GPR was propagated from the errors in the measured pairs of (xi,ti) of hyperbolas in the radargram. This effort attempts to simplify the complex computation in Xie et al. (2021a, 2021b) by assuming negligible uncertainties of buried target diameter and distance of GPR antenna separation. Experiments were carried out on various host materials to validate the model with a common offset antenna of nominal centre frequency at 400 MHz. The experimental results shows that (1) the expected uncertainty prior to survey was strongly dependent on the increase of the target depth. But in the evaluated uncertainty after the GPR survey, this strong dependency is replaced by the scattering noise in the reflection data (xi,ti). (2) The maximum percentage SD (i.e. error) is 11.5% of detected depth at 95% confidence level. This percentage is smaller than the the most stringent accuracy requirement in PAS 128 (15% of the detected depth) (ICE, 2014). This work contributes to the accuracy evaluation of the estimated depths of underground targets. The expected uncertainty of depth prior to GPR survey can provide a reference for the practitioners and clients during engineering, e.g. underground trenchless tunneling project). And the evaluated uncertainty after GPR survey turns a uncertain survey to a serious measurement reflecting the quality and reliability of the estimation results by GPR.