LiDAR point cloud and ICESat‐2 evaluation of 1 second global digital elevation models: Copernicus wins

• Published in: Transactions in GIS Vol. 25; no. 5; pp. 2245 - 2261
• Main Authors: Guth, Peter L., Geoffroy, Tera M.
• Format: Journal Article
• Language: English
• Published: Oxford Blackwell Publishing Ltd 01.10.2021
• Subjects: Canopies; Canopy; Digital Elevation Models; Earth surface; Elevation; Infrared detectors; Infrared radar; Lidar; Plant cover; Radar; Three dimensional models; Vegetation; Vegetation effects
• ISSN: 1361-1682; 1467-9671
• DOI: 10.1111/tgis.12825

Widely available LiDAR point clouds show that the Copernicus digital elevation model (DEM) consistently outperforms four other 1ʺ (arc second) global DEMs (ALOS, ASTER, NASA, and SRTM). Typical airborne LiDARs have 2–20 returns per square meter, or 1,500–10,000 points corresponding to each elevation posting in a 1″ DEM. The LiDAR point cloud approximates the 3D earth surface observed by visible, near‐infrared, and radar sensors used to create freely available global DEMs, and allows estimation of the canopy penetration by the sensor. Canopy is broadly defined as the range of elevations within the 1″ cell including effects from vegetation, slope, and human‐made features. In open terrain, all five DEMs approximate the ground surface. With significant canopy, the Copernicus DEM elevations cluster tightly near the center of the canopy, while the other DEMs are spread throughout and even outside the canopy defined by LiDAR. The superiority of the Copernicus DEM is evident in tests against both LiDAR and ICESat‐2 data, in all eight (high‐relief) test areas, in different vegetation types, and on both gentle and steep slopes.