These are datasets that have been useful. I did not create most of them. Other people are encouraged to test their programs against these files, so that we have a common basis for comparison.
| usgs1201/ |
usgs-bin.tz 24 sample level 1 DEMs. It has the alphabetically the first DEM starting with each letter of the alphabet (at the time the selection was made). The intention is that this be an unbiased sample of DEMs. The header info has been deleted, leaving only 1201x1201 elevations in each file. They are stored as 2-byte binary ints. Each file is 1201x1201x2 = 2884802 bytes before compression. To get a feel for the cells, see montage0.jpg and montage1.jpg, the two montages of the colored shaded cells. Most of the terrain is not mountainous at all. usgs1201-ascii-part.tar Tar file of the gzipped ASCII versions of the first 11 cells. Each file has 12012 integer elevations. Why do I include both binary and ascii versions? The binary versions are much faster to read, while the ascii versions are easier to read, display, and understand. There is also one more cell, lake-champlain-w-bin.gz and lake-champlain-w-ascii.gz, containing a region from Mt Marcy to Lake Champlain in NYS. The ascii file has 12012 ascii numbers, and the bin version, 12012 2-byte binary ints. To understand the terrain, see the color image. Note that the binary files have a endian problem if you port between Sun and Intel. These files are in the Intel order. | ||||||||
| etopo5/ |
Elevation and bathymetry data for the whole earth, at a 5 arc-min resolution. The files are ascii ints. See the etopo5.README file, altho it is slightly obsolescent. | ||||||||
| contour/ |
Raster contour elevation files, from Mike Gousie, to test terrain elevation interpolation. Each file is a square matrix of ascii numbers, suitable for reading by Matlab. A '0' is an unknown elevation. square.dat My small sample contour file, whose square corners make interpolation hard. | ||||||||
| tin_tests/ |
Artificial test cases of matrices of elevation posts, used to test TIN. In each case, the .png.gz file is a gzipped ASCII list of 12012 elevations. makeXXX.cc created the XXX test case. heada.pgm Prepending this file to an uncompressed 12012 ASCII elevation file makes an ASCII PGM file. ridge.{gz,png} This is a dataset with major valleys in the 4 cardinal directions and ridges at 45 degrees, to test if tin finds these lines automatically. circles.{gz,png} Two overlapping circular hills. This tests whether the TIN can identify the ridge and intersections to put edges there. sharpcircles.{gz,png} Two overlapping circular hills with sharp ridges. This is another test of whether the TIN can identify the ridge and intersections to put edges there. It is quite similar to circles. smoothcircles.{gz,png} Two overlapping circular hills with very smooth ridges. This is another test of whether the TIN can identify the ridge and intersections to put edges there. railway.{gz,png} Two overlapping circular hills, cut by a railway. The vertical sides of the cut are hard to interpolate. | ||||||||
| polygon/ |
Some sample planar graph (map) data containing polygons, to test map overlays. They are in a simplified form of the Harvard Odyssey format, with non-geometric info deleted. They are intended to be overlaid against each other in pairs.
|
