科研之友
微信
新浪微博
Facebook
分享链接
摘要
Roadway grade significantly affects onroad speed and acceleration, vehicle fuel consumption, vehicle emissions, driver behavior, traffic safety, roadway capacity, and congestion. However, it has always been challenging to obtain grade data at sufficient spatial resolution to use grade as an explanatory variable in transportation models. This paper aims to address this problem by proposing a method to obtain high-accuracy roadway grade data from the Digital Elevation Model (DEM), a nation-wide open data source from the U.S. Geological Survey (USGS). Although DEM data cover most of the nation, data resolution and the presence of roadway cut and fill sections affects spatial grade accuracy and requires a solid strategy to remove or infill these segments. Cubic smoothing spline is applied to minimize the impact of noisy data, and improve grade estimation accuracy. The selection of the key parameter lambda, in the spline method is also discussed to balance between smoothing out noisy elevation data, and retaining vertical fluctuations along the road. In general, lambda. is recommended 100-1000 for local roads, and 1000-10,000 for highways to maintain small average estimation error. The relationship between optimum lambda. that minimizes root-mean-square error (RMSE) and road fluctuations is also explored, which can be used for lambda. selection. Using real-world measurements as ground truth, the grade results generated from the DEM achieve an average estimation error of 0.5-0.58% for local roads, and 0.21%-0.23% for highways, depending on the resolution of the DEM data used. The results demonstrated the validity and applicability of DEM in generating high-accuracy roadway grade data.
- 出版日期2018-7
