RE: AI-GEOSTATS: DEM-interpolation method
- A problem with contour vertices -> point values is that most information
encoded in the contours is thrown away.
The "correct" solution is to maintain the input data as contours. For each
grid node build search lines parallel to the rows and to the columns of the
DEM, and interpolate from the nearest contour elevations found in those two
searches. If necessary, add searches offset by 45 degrees from rows and
The trick to making this practical is to note that finding all contour
crossings of, for example, a horizontal search line for a single DEM row can
be done before any node interpolation is performed. Some DEM smoothing
required. And special case software for DEM nodes inside peaks or valleys.
This method was used by USGS during the 1980s to build DEMs from contours
scanned off 1/24,000 scale quads. I suspect they are still using it. I do
not know of any algorithms outside the petroleum industry.
Dear member list
I used contour line and few elevation points to generate a DEM, the first
interpolator that I used was been the TIN interpolator but the RMS was been
not very good, so I converted all nodes of the contour lines into elevation
points and I used them to produce many model based on IDW, RBS and KRIGING
Using these interpolators I obtained a good results.
My question is:
it has meaning to convert all node of the contour lines in points and use
them to generate a DEM with different interpoators that ask points as input
[Non-text portions of this message have been removed]