An update on the development of a self-correcting method of calibrating pavement layer depths is presented. The approach is applied to multi-offset simulations of a forthcoming 3D noise-modulated ground penetrating radar (NM-GPR) system. The expected response of a series of wide angle reflection and refraction (WARR) datasets produced by this equipment is modelled for typical road pavement configurations using the finite difference time domain (FDTD) approach. As an initial step, conventional geophysical methods and layer interface tracking algorithms are used to determine approximate layer depths and bulk permittivity values from the WARR data. One of two self-correcting analysis methods is then used to refine the layer depth predictions. The first method, interface matching, involves adjusting permittivity values within a 2D ray-path model until the corrected depth of the tracked interfaces is consistent between adjacent WARR groups. The second method, ray-path modelling, involves adjusting assumed interface depths, gradients and permittivity values within a 3D ray-path model until predicted two-way travel times match that of the tracked layer interfaces. Both methods proceed iteratively, working layer-by-layer from the top and account for inter-layer refraction in the calculations. Potential advantages of these methods include continuous non-destructive calibration of pavement layer depths and more representative estimates of bulk layer properties, particularly for lossy materials and in situations where moisture, material or compaction gradients may be present.