Determining Asphalt Thickness Using Ground Penetrating Radar

A Comparison of Automated and Manual Methods Using Falling Weight Deflectometer Back-calculation Error Correction

Abstract

The objective of this work was to evaluate and enhance the accuracy of GPRbased
pavement thickness evaluations. GPR (Ground Penetrating Radar) data were analyzed
to determine asphalt thickness using both a traditional processing method with a trained
interpreter, and by automated processing requiring limited operator interaction. It was found
that the use of automated GPR processing significantly decreases the amount of time and the
expertise needed to analyze the pavement structure, while providing acceptable accuracy in
the estimation of pavement thickness. Since incorrect layer identification is a source of GPR
analysis error, results of Falling Weight Deflectometer (FWD) back-calculations were used
where the layer identification was suspect, and to suggest alternative layer selections. The
manual and automated processing techniques were applied to field GPR data collected from
130 FWD test locations at 26 pavement sites throughout Montana. Implementation of the
error detection and correction procedure reduced the deviation between GPR and core data by
over 30% for both manual and automated methods. Based on data from 130 cores, the average
deviation between GPR data and core data was found to be 6.2% (0.32 inches) for the manual
method vs. 7.6% (0.42 inches) for the automated method.