Characterisation of concealed road distress severity based on FDTD forward simulation of ground-penetrating radar

Concealed pavement distresses often develop after surface damage and can significantly impair pavement structural performance. Ground-penetrating radar (GPR) is widely used for detection, but assessing distress severity from image features remains difficult. To clarify how GPR image features vary with concealed distress severity and support distress classification, feature analysis was conducted using finite-difference time-domain (FDTD) forward simulation. Concealed crack models with different widths and heights, dry base-layer models with varying void ratios, and base-layer models with identical severity but different moisture contents were constructed, and synthetic B-scan images were generated.Parabolic fitting was applied to crack reflection curves, and the quadratic coefficient α was used to establish relationships with crack width and height. Positive exponential correlations were obtained, with average correlation coefficients of 0.946 and 0.958, respectively. As the severity of dry base loose areas increased, waveform textures in B-scan images became more disordered. Texture metrics from the grey-level co-occurrence matrix (GLCM) showed strong correlations with distress severity, with R² values of 0.976, 0.996, 0.979, and 0.995, among which average homogeneity showed the highest correlation. Increasing void moisture content also produced more disordered textures, with R² values of 0.936, 0.914, 0.911, and 0.871, where mean contrast showed the strongest correlation.