[GRASSLIST:2383] Re: Fun with gdal - Follow-up

[SWlab]

[Wednesday 28 January 2004 18:23] From Hamish

[I'm putting the list in copy to get insights of our listmates :)]

> > I'm far more concerned with the first question, though: should I (1)
> > import maps directly in the current location with r.in.gdal ? (2)
> > Create temp locations for each quad map and reproject those maps to my
> > current location by r.proj ? I should get the same result, right (viz,
> > if the projection info is the same for the quad maps and the current
> > location...) ? Why do I have such large errors (15m is abig deal...)

> Perhaps you used different map resolutions somewhere along the line? Are
> all the quads the same resolution? Maybe it is 'snapping' a quad to a
> lower resolution somewhere which is introducing the error. Do some maps
> use the coordinate as the corner of the grid cell and others the
> centroid?

The four DEM maps come from the same source, and have been processed the same 
way. They're standard USGS DEM, viz one line file with a header, a series of 
profile, and an accuracy record. Basically, there shouldn't be any pb, and I 
would assume that a "r.in.gdal i=demfile o=dem lo=newtmplocation" would give 
the proper results. 

Then, the problem would come from the importation in the current region. Could 
you point  me to a doc describing how "r.proj" and "r.in.gdal" work ? 

BTW: the resolution of each DEM is 10mx10mx0.1m. My region is defined from the 
most northern boundary to the most southern and from most western to the most 
eastern. Because of the slight tilts of the maps, the estearn boundary comes 
from the SE quad map, western from NW, southern from SW, and northern from 
NE. My grid size is 10m.

> (in GRASS the region boundaries are the outside edge of the cells, the
> coordinates of each cell represent the center of the cell, if I recall
> correctly)
> Then a 30m quad would be transposed 15m north and 15m east, for example?

I was thinking about something like this. From the two options, then, which 
one would give the smallest error ? importing directly, or reprojecting ?

Did anybody have the same problem ? Is there a "standard" way to import quad 
maps in a larger map ? It's a pretty trivial operation, but I'm not a GIS 
specialist at all, and don't have real clues.

Thanks for your interest
P.

-- 
Soil & Water Laboratory
Dept. of Biological & Environmental Engineering
Cornell University
ITHACA, NY 14853
Tel: (607)255.2463