Inter-System Concordance and Short-Term Survey Repeatability of Two Ground-Penetrating Radar Acquisition–Interpretation Chains in Forensic Search: An Exploratory Field Study

Abstract

Background/Objectives: Comparative evidence for GPR system selection in forensic search is scarce. This exploratory study assesses the short-term repeatability and inter-system concordance of two GPR acquisition–interpretation chains: the multi-frequency Cobra CBD and the lightweight single-frequency Zond Aero 500 NG. Methods: At a forensic test facility containing nine graves (six pig carcasses, three controls; burial depths 30–90 cm), of which five (one carcass grave per depth class) were traversed and scored, record windows (50, 100, 200 ns) were compared under harmonized settings. At an anonymized operational site (≈9 × 7 m), a cart-based grid protocol was executed in duplicate with both systems. Independently interpreted anomalies were compared via one-to-one centroid matching, endpoint offsets, and repeat-survey variability. Results: All five traversed graves were visually confirmed by both systems at 50 and 100 ns and none at 200 ns (unblinded assessment at known markers). One-to-one matching paired every Zond Aero 500 NG segment with a Cobra CBD segment within 1.5 m (medians 0.48 m longitudinal, 0.72 m transverse); surplus Cobra CBD segments (27–36%) reflect finer segmentation. Consensus re-interpretation of all datasets altered segmentation (33–27; 22–31 segments) more than localization. Between-system endpoint offsets (0.6–3.2 m) resembled repeat-survey offsets (1.1–2.9 m). Conclusions: The concordance characterizes complete acquisition–interpretation chains: sensor, survey, and interpreter effects cannot be separated, and without ground truth they quantify concordance, not detection accuracy. Within the tested cart-based configurations, no localization difference exceeding repeat-survey variability was resolved. Broader equipment-selection claims require ground-truthed, multi-site studies, and the record window should match expected burial depth.