GPR attenuation analyses using spectral ratios of primary and multiple arrivals: examples from Welsh peat bogs

Abstract:

Ground penetrating radar (GPR) is widely applied to qualitative and quantitative interpretation of near-surface targets. Surface deployments of GPR most widely characterise physical properties in terms of some measure of GPR wavelet velocity. Wavelet amplitude is less-often considered, potentially due to difficulties in measuring this quantity: amplitudes are distorted by the anisotropic radiation pattern of antennas, and the ringy GPR wavelet can make successive events difficult to isolate. However, amplitude loss attributes could provide a useful means of estimating the physical properties of a target.

GPR energy loss is described by the bandwidth-limited quality factor Q* which, for low-loss media, is proportional to the ratio of dielectric permittivity, ε, and electrical conductivity, σ. Comparing the frequency content of two arrivals yields an estimate of interval Q*, but only if they are sufficiently distinct. There may be sufficient separation between a primary reflection and its long-path multiple (i.e. a ‘repeat path' of the primary reflection) therefore a dataset that is rich in multiples may be suitable for robust Q* analysis. The Q* between a primary and multiple arrival describes all frequency-dependent loss mechanisms in the interval between the free-surface and the multiple-generating horizon: assuming that all reflectivity is frequency-independent, Q* can be used to estimate ε and/or σ.

We measure Q* according to the spectral ratio method, for synthetic and real GPR datasets. Our simulations are performed using the finite-difference algorithm GprMax, and represent our example data of GPR acquisitions over peat bogs. These data are a series of 100 MHz GPR acquisitions over sites in the Brecon Beacons National Park of South Wales. The base of the bogs (the basal peat/mineral soil contact) is often a strong multiple-generating horizon. As an example, data from Waun Ddu bog show these events lagging by ~75 ns: GPR velocity is measured here at 0.034 m/ns (relative ε of 77.9) and spectral ratios suggest Q* of 19.9 [-6.6 +19.4]. This Q* implies that the bulk σ of the bog is 21.7 [-10.7 +10.8] mS/m. Our measurements require in situ verification (e.g. comparison with co-located electrical resistivity profiles) but our method provides a promising addition to the suite of GPR analysis tools.