GPR is not a replacement for FWD, instead it acts to complement these measures. FWD gives you a deflection measurement of the pavement surface but cannot give you an accurate strength reading unless combined with accurate layer thicknesses and material type obtained from construction plans or coring samples.
GPR, especially when combined with dielectric compensation techniques, can provide accurate layer thickness, or at least an indication of where the thickness of layers are changing or are more variable (e.g. due to patching). When you combine this with the FWD results, a greater understanding of the pavement is obtained. Some unusually stiff FWD results can be attributed to patching. An increase in pavement deflection can be linked to a thinning of the asphalt thickness, instead of assuming that it is due to a material strength issue.
When rebuilding of a pavement this knowledge has significant cost advantages. For example, one case of FWD indicated there were large differences in the pavement deflection in a section of road that from the plans should have had the same construction for the whole pavement. The initial assumption therefore was that the larger deflection was due to low strength material in these sections. The suggested remedial measure was to rebuild the road with an extra 200mm of granular materials followed by a chip seal surface. When it was discovered that the material was not actually weak, but instead just thin, it allowed the design to change to 100mm of granular. The result was millions of dollars saved. For other cases GPR will show that a thicker reconstruction may be needed. While this will increase the rehabilitation cost, if this decision is made early in the reconstruction process, it can be budgeted and will result in a road that will last the required length of time. There is nothing more expensive than a road that fails a year or two out from construction.