Esri documentation indicates Contours generated from Spatial Analyst are "engineering-quality". Is this really true? I'm comparing with a Global Mapper output that indicates its only "cartographic-quality". But as you can see in the attachment, they "appear" very similar. The were both produced from a very accurate LiDAR-based DEM. For the record, the red lines are ArcGIS; the brown lines are Global Mapper. The horizontal spatial offset is noticeable. For reference, this is about 1 to 2 feet. ...Any thoughts on this distinction? Thanks.
The next sentence also should be considered
Overall contour accuracy depends on how well the data used to create the input raster represents the actual surface.
Plus all the supplemental comments in subsequent paragraphs about where things appear to go wrong.
There is even information on the use of a kernel to smooth out the dem values somewhat.
As for the 1 to 2 foot offset, it seems to be consistent-ish suggesting that a difference in datum is at play
Thanks Dan. Yeah, it seems there are technical factors and some subjectivity in attempting to define Contour quality, and especially if trying to squeeze it into "cartographic-grade vs. engineering-grade".
As for the offset in the (attached) map: The input DEM and coordinate system/datum for these two versions are the same. And even if the were different, wouldn't ArcGIS re-project on the fly if I apply the correct transformation?
That is the theory, re reproject-on-the-fly, unless minor adjustments aren't accounted for
As for the offset, I am not sure whether 'cartographic' grade adhere to the passing through cell-centers rule-ish either.
There is also the 'snap raster' and cell size issues when trying to compare two interpolated surfaces and hence their derived contours.
The only way you can definitively examine this is with constructed surfaces which I have dabble with a bit for algorithm testing
Thanks again Dan. Let me see if I can find out a little more about the Blue Marble/Global Mapper (cartographic) contour algorithm and what settings were used to create it.
check to see where and/how they define their contour placement relative to the cell as well. Also, the 'snap raster' issue. It looks like the shift is uniform so the extent of the raster may be shifted by 1/2 a cell or so in the vertical direction