7.D. Additional Comments on Accuracy
Note that various methods for measuring the terrain surface do not necessarily measure the same objective. In-situ surveys (such as Global Positioning System or cadastral surveys) tend to measure at a location above the ground surface, then compute the ground elevation below the "known" ground clearance of the measurement. Stereo-optical imagery may measure the top of tree canopy--which may or may not be converted to the elevation of the ground surface when the DEM is made. This could be considered an issue of "attribute accuracy." That is, are DEMs measuring what they purport to measure? In most cases, the source data sets for GLOBE have not been able to address this issue.
In addition, misrepresentations or errors in vertical datum could lead to 100+m errors in a DEM. This may happen when information is missing from a map and an analyst makes an incorrect assumption about the map. For example, the analyst may assume a particular ellipse in reprojecting data, when another ellipse, or a sphere, had been used but not documented (or incorrectly documented).