Estimating “True” Variability of Traffic Speed Deflectometer Deflection Slope Measurements

In this paper, the difference sequence method is used to decompose traffic speed deflectometer (TSD) deflection slope measurements variability into "true" spatial variability that is due to pavement structural changes and noise variability. A robust method to evaluate the noise standard deviation is presented and the difference sequence method is validated using simulated examples. The difference sequence method is then used to calculate the noise standard deviation and "true" spatial variability of TSD deflection slope measurements. The evaluated noise standard deviation is also used to determine the optimal smoothing of TSD deflection slope measurements using an unbiased measure of the risk. This method is compared with the generalized cross validation (GCV) criterion to determine the optimal smoothing. Results suggest the unbiased risk criterion performs better than the GCV criterion. This gives an objective method, which is currently lacking, to average TSD measurements, and more generally any continuous deflection device measurements. Finally, the results in this paper are presented under the reproducible research paradigm; a MATLAB implementation is made available that can be downloaded and used to reproduce all figures and tables presented in this paper.