Improved Annual Average Daily Traffic Estimation on Coverage Count Segments from Volume Correlations with Multiple Automatic Traffic Recorder-Equipped Segments: Empirical Results from Ohio Highways

On most highway segments, annual average daily traffic (AADT) is presently estimated from a small sample of daily volumes obtained from coverage counts that are being collected when a more extensive, almost complete census is conducted on a few segments equipped with automatic traffic recorders (ATRs). Origin-destination flows (OD) that use both the coverage count segment and a specific ATR-equipped segment induce correlation between their daily volumes. A regression-based estimator of AADT on coverage count segments that exploits the OD-flow-induced correlation was developed previously. Using simulated data, this regression-based estimator was shown to perform better than the traditional estimator. However, volumes are typically collected on multiple ATR-equipped segments on the days that volumes are collected on the coverage count segment. By appropriately combining information from all these segments, one can expect even better performance, while requiring no additional data collection. In this paper, a simple estimator is suggested to address the multiple ATR case encountered in practice. The estimator averages the regression-based estimates corresponding to each of the multiple ATR-equipped segments. Using real data from 12 Ohio highway segments, this estimator markedly outperforms the traditional estimator on the three performance measures considered. The empirical performance tends to improve with increasing pair-wise correlation between the coverage count and ATR-equipped segments. Possible improvements to this estimator are also discussed.