Work

Line Spacing and Positioning Error in GPR Amplitude Maps: A Converged gprMax Benchmark Tested Against Field Data from Isthmia, Greece

Year: 2026
Type: article
Abstract: Archaeological ground-penetrating radar (GPR) maps are interpreted spatially, yet the joint effect of across-line sampling and trace-coordinate error on amplitude maps is rarely quantified. We combine a converged gprMax benchmark with field data from the Sanctuary of Poseidon, Isthmia, Greece. In the benchmark, 532 simulated 400 MHz traces (10 mm cells; gate-RMS within 1.2% of 6.67 mm cells) define dense reference maps of four idealised scenes, which were decimated (0.1–0.4 m line spacing) and perturbed with five coordinate-error models (0–0.20 m; 100 paired trials). Map error followed a power law in σ/ℓg, where ℓg is a gradient length scale of the reference map: NRMSE/std(R) = 0.58(σ/ℓg)0.77 (R² = 0.99). The same procedures were then applied to 450 MHz data from three RTK-positioned 2024 grids at Isthmia (26,700–40,500 traces each), whose gate-RMS maps reproduced Geolitix amplitude slices (Spearman ρ = 0.71–0.73). The field data followed the same power-law form, with exponents of 0.66–0.73 and prefactors 1.3–1.7 times larger. The position error at which NRMSE/std(R) reached 0.2 was 1.9–2.9 cm (0.24–0.27 ℓg) in the simulations and 1.9–5.1 cm (0.11–0.16 ℓg) in the field, against a local RTK scatter of 1–2 cm. Prediction of withheld lines was dominated by intrinsic trace-to-trace variability, and overlapping independent passes agreed only weakly (r = 0.13–0.32). Map fidelity at Isthmia was therefore limited more by line spacing and attribute repeatability than by RTK coordinates. The tolerances apply to one scalar attribute at one site and are not detection probabilities.
Language: English
Subfield: Geophysics
Open Access status: diamond
OpenAlex ID: W7236555262
FWCI:
Cites:
Cited by:
Related to: 10