Investigators:
Dr. Randall L. Guensler, Georgia Tech
Dr. Michael O. Rodgers, Georgia Tech
Haobing Liu, Ph.D. Student, Georgia Tech
Project Overview:
The Digital Elevation Model (DEM) is a digital cartographic/geographic dataset of elevations maintained by U.S. Geological Survey (USGS). The primary source of DEM is 3-Dimensional Elevation Program (3DEP) operational, LIDAR and interferometric synthetic aperture radar data (USGS, 2016). DEM contains multiple resolutions, providing elevation values in every grid of 1×1 meter (1/27 arc-second, the highest resolution), 3×3 meter (1/9 arc-second), 10×10 meter (1/3 arc-second), and 30×30 meter (1 arc-second). Elevations are pixel-centered in raster datasets representing the value at the center. Currently, elevation with 10×10 meter is the most refined dataset that covers the whole United States. Elevation with 1×1 meter resolution is available in less 5% of the country, and data with 3×3 meter resolution covers 25%. Part of Alaska is currently covered by 30×30 resolution data, but 5×5 meter will eventually become the highest-resolution elevation dataset over the state (USGS, 2016). The USGS is continuously expanding the coverage of elevation data with 1×1 meter resolution, and a dataset with national coverage of such high resolution in the United States will be supported (Sugarbaker et al., 2014; Wood et al., 2015). Considering its wide coverage, there is of great value in using it to estimate road grade, and append the grade information onto the vehicle GPS trajectories to analyze its influence on MARTA transit activity, energy use and emissions from macro to micro-scale.
The Georgia Tech Team has developed a streamlined method to extract roadway elevation profile from DEM database, clean and refill erroneous elevation data, and apply cubic smoothing spline to generate road grade with high-resolutions. Based on the road grade validation compared with field measurement, the proposed method can generate highly accurate road grade, with root mean-square error (RMSE) of estimated road grades 0.2-0.23% for highways and 0.5-0.6% for local roads. Figure below shows the estimated vs. measured results for highways and local roads with DEM 10×10 meter applied. The terrain elevations and grades for ground positions are sampled at 20-30 foot spaced horizontal intervals along the roads, which can be easily matched with vehicle second-by-second GPS data for energy and emission calculation, or implementation of eco-driving strategy. The team has integrated the method in the PACE computing cluster and is currently generating the elevation profile for the Atlanta network.
