7.A.i. Data from Raster Data Sources
In most cases, data from DTED Level 1 3" source materials (source/lineage categories 2 to 7) can be expected to have 15" or better horizontal accuracy. Thus, nearest-neighbor resamplings to 30" gridding for the MISR DEM, GTOPO30, and GLOBE should remain within the appropriate 30" grid cell. This would make no measurable horizontal error for such data.
In the case of DTED Level 0 discrete (spot) data (source/lineage category 1), the reference location is at the southwestern corner of each 30" grid cell. Errors as small as 100m in horizontal datum might misplace a DTED discrete value into an adjacent 30" grid cell, making for a maximum anticipated absolute horizontal error of one 30" grid cell. Relative horizontal inaccuracy from such data could be comparatively high--on the order of one grid cell relative error on (probably rare) occasions.
Source/lineage categories 9 to 11 probably have horizontal errors of less than 30" for almost all parts of the world.
Data from satellite altimetry, such as were used for part of Greenland (source/lineage categories 12 and 13) may be mislocated due to inaccurate determination of the satellite’s location in space. As these data were originally gridded at larger than 30" intervals, oversampling to 30" for the MISR DEM might have permitted horizontal errors of more than 30" in source/lineage categories 12 and 13 in GLOBE. We currently have no formal estimate of the horizontal accuracy of such data, other than to assign the GEOSAT gridded DEM grid cell size of 10km to the likely horizontal locational accuracy.