講演情報
[Data_Collection-13]Optimal Sampling Frequency for Network-Level Pavement Texture Parameter
*Christian Andres Sabillon-Orellana1, Danilo K.N. Inoue2, Jorge Prozzi2 (1. University of Tennessee, Knoxville (USA), 2. University of Texas at Austin (USA))
キーワード:
Pavement Texture、Network-Level Measurement、Texture Parameters、Sampling Frequency
Dynamic pavement texture measurements collected with vehicle-mounted laser sensors have enabled transportation agencies to evaluate surface characteristics efficiently at the network level. However, these systems acquire texture profiles at dense spatial and temporal sampling frequencies, producing very large datasets that are computationally intensive to process and costly to store. This study will determine the optimal sampling frequency, in both distance and Hertz, at which pavement texture parameters remain statistically equivalent to values computed from dense “ground-truth” sampling. Texture data have already been collected across multiple roadway facilities in the State of Texas using a high-frequency line-laser sensor mounted on the rear of a survey vehicle. The dataset includes distinct pavement surface types, including dense-graded asphalt, open-graded friction courses, and various concrete textures. The densely sampled profiles will serve as the reference dataset. Each profile will be resampled to progressively lower sampling frequencies, and commonly used texture parameters, such as the Mean Profile Depth and additional texture-parameters, will be computed at each sampling rate. Statistical equivalence will be evaluated using traditional statistical two-sided hypothesis testing procedures, supported by percent-error evaluation and Bland-Altman agreement analysis. The outcomes of this study should guide sampling thresholds at which texture parameters remain statistically indistinguishable from reference (denser) values, allowing agencies to right-size data collection and processing accordingly. Establishing a justified minimal sampling frequency allow more efficient network-level texture monitoring by reducing data storage and processing demands and improving scalability of texture-based pavement evaluation programs.
