[GRASS-dev] Reproject, change resolution and patch: doubts....

[marco.pasetti at alice.it]

Hi all,
 
I would like to submit to you the following "work plan" that I figured out to do a particular job:
 
I need to apply the module r.horizon (add-on from JRC) that computes the angular height of terrain horizon from a given point and a given DEM raster.
That DEM should be as large as possible to let the module create an horizon profile as much realistic as possible.
I have two types of GIS data of the interested area to work on:
 
A) the SRTM database, 90m resolution, WGS84
B) a local GIS database, with a resolution of 20m, Gauss-Boaga Rome40 West (preferred, because of the better resoultion)
 
all the data are in arc-info ASCII format.
Since I don't want to create confusion within different data and projects, I decided to create two main locations: a Gass-Boaga local for the local database, in which import ASCII data usign r.in.arc module, and an WGS84 location for the SRTM database (using r.in.arc here too).
 
While the SRTM data are organanized in big maps, where a single map covers all the interested area, the Local database is fragmented in multiple small (but contiguos) maps; because of that I need to patch all of them to have a sufficient working area. But that's not still enough, since some points of interest lay on the borders of such local map: that's why I need to patch the local database *big map* (all the maps patched together) with the SRTM map (only one map is needed), with the following goals:
 
1. obtain a map with the local map data, where those are present, and with the SRTM data where the local data are not present (null)
2. do not alter the original data of both locations during patching or resampling (for example for SRTM passing from 90m resolution to 20m)
 
I try to briefly summarize the steps of the procedure I figured out to reach that goals:
 
1. Create a Gauss-Boaga location (call it GB location) and import the local database (native resolution 20m) with r.in.arc
2. Set the region in order to contain all the GB maps and patch all the imported maps (contiguous) to create the *big map*
3. Create a WGS84 location (call it SRTM location) and import the needed DRTM map (native resolution 90m) with r.in.arc
4. Create another WGS84 location (all it JOB location; I create it new because I prefer to preserve the *original* locations) and set the region to fit the interested area
5. In the JOB Location: r.proj the GB big map, maintaining the orginal resolution (20m)
6. r.proj the SRTM map, setting the resolution to 20m (against the original of 90m) during the reprojection (to let me patch it with the GB map that has a 20m res)
7. r.patch the GB big map and the SRTM map, putting the GB raster as first in the input parameter to achieve the goal n°1
 
My doubt is: moving from the 90m resolution to 20m in the SRTM map I will oversample, right? but I will actually maintain the original data (that it is what I want).
Instead, if I resample the SRTM data (using one of r.resamp.* modules) I would obtain a smooth DEM, but I would also alterate the original data, introducing *dangerous* fake data, and so possible critical errors.
 
Do you agree?
 
Thanks for all,
 
Marco
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/grass-dev/attachments/20080507/c1408573/attachment.html