[GRASS-user] A question about raster map resolutions

[Nikos Alexandris]

Rich Shepard wrote:

>> The LiDAR source location has x- and y-resolution of 3 feet
>> (international) and several thousand cells east-west and north-south.
>> g.region -p for this location displays:
>>
>> projection: 99 (unnamed)
>> zone:       0
>> datum:      ** unknown (default: WGS84) **
>> ellipsoid:  grs80
>> north:      1286590.24500872
>> south:      1240510.24500872
>> west:       634966.76150938
>> east:       664972.76150938
>> nsres:      3
>> ewres:      3
>> rows:       15360
>> cols:       10002
>> cells:      153630720
>>
>>    When in the project location g.region -p produces:
>>
>> projection: 99 (NAD_1983_HARN_StatePlane_Oregon_North_FIPS_3601_Feet_Intl)
>> zone:       0
>> datum:      nad83harn
>> ellipsoid:  grs80
>> north:      588219.90693215
>> south:      536993.58817534
>> west:       7524575.13218961
>> east:       7561596.20977101
>> nsres:      2561.31593784
>> ewres:      1851.05387907
>> rows:       20
>> cols:       20
>> cells:      400
>>
>>    I want to understand why the nsres is now 2561+ feet, the ewres now
>> 1851+
>> feet, and the numbers of rows and columns has decreased so drastically.
>>
>> Looking forward to learning,

Helmut Kudrnovsky:

>If it is about raster reprojection, there are hints in the r.proj manual
>about region settings before applying r.proj.
>
>E.g.
>
>A simple way to do this is to check the projected bounds of the input map in
>the current location's projection using the -p flag. The -g flag reports the
>same thing, but in a form which can be directly cut and pasted into a
>g.region command.

Some observation: `-g` prints out in shell style, suitable for
`eval`uating.  The `eval` utility, for example, constructs a
command, from the given arguments (here, the ouput of `g.region -g`)
which is then executed by the shell.  See for example
https://grasswiki.osgeo.org/wiki/Shell_scripting#Using_output_from_GRASS_modules_in_the_script.

A ready to copy-n-paste form is obtained via `-f` (this is a recent
addition by Moritz during the code sprint in Bonn).

Maybe the `-f` flag would be a useful addition to `r.proj` as well.

Nikos


>After setting the region in that way you might check the
>cell resolution with "g.region -p" then snap it to a regular grid with
>g.region's -a flag. E.g. g.region -a res=5 -p. Note that this is just a
>rough guide.
>
>A more involved, but more accurate, way to do this is to generate a vector
>"box" map of the region in the source location using v.in.region -d. This
>"box" map is then reprojected into the target location with v.proj. Next the
>region in the target location is set to the extent of the new vector map
>with g.region along with the desired raster resolution (g.region -m can be
>used in Latitude/Longitude locations to measure the geodetic length of a
>pixel). r.proj is then run for the raster map the user wants to reproject.
>In this case a little preparation goes a long way.
>
>
>
>-----
>best regards
>Helmut
>--