USGS DEM file geolocation questions [message #85041] |
Wed, 26 June 2013 06:24 |
Kenneth Bowman
Messages: 86 Registered: November 2006
|
Member |
|
|
Does anyone understand the details of the geolocation metadata in USGS
DEM (e.g., GRIDFLOAT files) files? I have looked in the documentation
I can find online, but it is unclear.
First, do the xllcorner and yllcorner parameters in the header files
(see example below) refer to the location of the first data *point* in
the lower-left corner of each tile, or to the lower-left corner of the
lower-left grid *cell* in the tile (with the cell center offset by
cellsize/2)? Should I think of the elevation data as point data or as
averages over the grid cells? With the lat and lon numbers given, the
latter seems to make more sense to me, but I am not familiar with the
conventions used with DEM data.
ncols 3612
nrows 3612
xllcorner -106.0016666667
yllcorner 37.99833333333
cellsize 0.000277777777778
Second, the 1 arcsecond files contain 3612 x 3612 values, which
includes 6 rows or columns of overlap on each side. When I compare,
for example, the elevations of the 6 columns outside a 1 x 1 deg tile
with the first 6 columns inside the adjacent tile, which I think should
match exactly in location, the numbers are close, but not the same.
That seems to defeat the whole point of having a seamless national DEM.
For the two sample tiles that I picked (Pikes Peak, which contains
steep terrain), the elevations differ by 10 - 20 m.
I can just ignore the overlap rows and columns, but I fear I am making
some more fundamental error.
Thanks, Ken Bowman
|
|
|