Difference between revisions of "HorizontalAccuracy"

Revision as of 14:06, 20 April 2018

DGED

DGED elevation datasets shall conform to the predefined thresholds for horizontal accuracy. Randomly selected datasets may be tested to assure that the random horizontal error per point and relative horizontal accuracy between points meet or exceed the thresholds detailed in Table 5 ( Geographic DGED Horizontal Accuracy Requirements) and Table 6 ( UTM DGED Horizontal Accuracy Requirements). Additionally, there is a goal that the absolute horizontal accuracy meets the values established in Tables 5 and 6 (or is better). Although it is not required to meet a threshold requirement, an absolute horizontal accuracy estimate will be reported for each data set. Table 5 – DGED Horizontal Accuracy Requirements Product name Spatial resolution Random Horizontal Error Per Point (σH – meters) Relative Horizontal Accuracy between Points (RH – meters) Goal: Absolute Horizontal Accuracy CE90 (meters) Angular (arc sec.) Linear(m) L0 30 1000 <= 50m L1 3 100 <= 50m L2 1 30 <= 23m L3 0,4 12 4,4 12,4 <= 15,0m L4b 0,15 5 1,75 5,00 <= 6,00m L4 0,12 4 1,41 4,00 <= 5,00m L5 0,06 2 0,71 2,00 <= 3,00m L6 0,03 1 0,35 1,00 <= 2,00m L7 0,015 0,5 0,18 0,50 <= 1,00m L8 0,0075 0,25 0,09 0,25 <= 0,50m L9 0,00375 0,125 0,04 0,125 <= 0,25m

Horizontal Accuracy notes: 1. The random horizontal error is the random circular error at point (reported at 90% confidence). 2. The Relative Horizontal Accuracy (point-to-point) between points is calculated at the 90% confidence interval. 3. The relative (point-to-point) horizontal accuracy is based on a distance (“D”) which is determined by the size of a standard DGED dataset at a given DGED Level. “D” is defined as the maximum radial distance between any two points in the said tile. The relative error is expected to vary as a function of distance between points, and its value ranges from R0 for points within one pointspacing to the table value for points at distance “D”. 4. The absolute horizontal accuracy is a goal value, but not a requirement. Although not required to meet a threshold, the Absolute Horizontal Accuracy should be reported in the metadata and is measured at the 90% confidence interval as a CE90. 5. Relative Horizontal Accuracy and random horizontal error may be difficult to assess in the coarse datasets. In these datasets, very large targets (which may not be available) would be required for such assessment. Therefore, the relative horizontal accuracy may not be evaluated for Levels 0 to 5. NOTE: These absolute horizontal accuracy values may be beyond current capabilities.