Understanding depth-amplitude effects in assessment of GPR data from concrete bridge decks

The variation of concrete cover thickness on bridge decks has been observed to significantly affect the rebar reflection amplitude of the ground penetrating radar signal. Several depth correction approaches have been previously proposed in which it is assumed that, for any bridge, at least a portion of the deck area is sound concrete. The 90th percentile linear regression is a commonly used procedure to extract the depth-amplitude relationship of the assumed sound concrete. It is recommended herein that normalizing the depth-dependent amplitudes be divided into two components. The first component takes into account the geometric loss due to inverse-square effect and the dielectric loss caused by the dissipation of electromagnetic energy in sound concrete. The second component is the conductive loss as a result of increased free charges associated with concrete deterioration. Whereas the conventional depth correction techniques do not clearly differentiate the two components and tend to incorporate both in the regression line, they are separately addressed in this research. Specifically, while the first component was accounted for based on a library of GPR signals collected from sound areas of twenty four bare concrete bridge decks, the conductive loss caused by an increased conductivity is linearly normalized by the two-way travel time. The implementation of the proposed method in two case studies showed that, while the method significantly improves the accuracy of GPR data analysis, the conventional methods may lead to a loss of information regarding the background attenuation that would indicate the overall deterioration of bridge decks.