5.B.i.f. Conterminous United States and Vicinity
Data coverage for the conterminous United States and some neighboring areas was available from AMS/DMA/NIMA sources, with several lineages:
One 30" version was contributed directly from NIMA to NGDC (now NCEI) for GLOBE. This version (DTED Level 0) includes spot, minimum, mean, and maximum 3" values for each 30" grid cell. However, independent peer review led to the discovery
that the "mean" values were frequently lower than the "minimum" values, and that the spot values appeared to be the most accurate representative elevation values. These data are the newest and sometimes contain revisions to
source data since DMA contributed data to NGDC and to USGS.
A second version for the conterminous U.S. was previously reprocessed to 30" by DMA and contributed to NGDC for public distribution. This version was contributed to NGDC before DTED Level 0 (noted above) was designed, but after the 3"
(noted below) data were contributed to USGS. Thus, this DMA/NGDC data set may have benefited from enhancements to the topographic data since DMA’s contribution to USGS. Both discrete (spot, nearest-neighbor) and average data were
contributed to NGDC. As the other data in the region used nearest-neighbor resampling, the spot version of the DMA/NGDC data were considered more appropriate for use in GLOBE Version 1.0.
A third version at 3" gridding for most of the U.S. except Hawaii was contributed by DMA to USGS for public distribution. (USGS calls this version "USGS 3-arc-second DEMs.") This was a pioneering DEM, perhaps the first gridded DEM
ever developed. USGS resampled these 3" data to 30" for inclusion into GTOPO30, using a nearest-neighbor technique. DMA/NGDC data may have received editorial improvements, while NIMA/DTED Level 0 data clearly (according to peer review
input) have many such enhancements.
Because of the lineage of enhancements, NIMA/DTED Level 0 were selected where available. Second priority went to DMA/NGDC data. Where neither of these data were available, DMA/USGS data were used.
In several cases in coastal areas, NIMA discrete (spot) grid cells had values of 0 (sea level) where NGDC or USGS data had values above zero. Inspection of these areas led to the observation that the single 3" value sampled to make DTED Level
0 discrete data may have been zero, but the samples used to make the NGDC and/or USGS versions were non-zero. In such cases, NGDC or USGS non-zero values were allowed to replace sea level DTED Level 0 discrete values. A mask was built to prevent
inland values of zero elevation from being overwritten by subsequent non-sea-level values. Non-sea-level values were allowed to overwrite values of zero elevation in near-coastal areas. Inspection of the three candidate data sets in near-coastal
and inland areas suggested this as the most appropriate procedure for this version of GLOBE. The order of priority was (1) DTED Level 0 discrete value, (2) NGDC value if DTED Level 0 discrete value was zero or ocean mask value, and (3) GTOPO30
value if the previous two sources were both zero or ocean mask values.
Return to Report Introduction