[GRASS-dev] Re: [GRASS GIS] #789: g.region option to expand the computational region of about "some" pixels?

[GRASS GIS]

#789: g.region option to expand the computational region of about "some" pixels?
--------------------------+-------------------------------------------------
  Reporter:  nikos        |       Owner:  grass-dev at lists.osgeo.org            
      Type:  enhancement  |      Status:  new                                  
  Priority:  normal       |   Milestone:  7.0.0                                
 Component:  default      |     Version:  unspecified                          
Resolution:               |    Keywords:  g.region, expand computational region
  Platform:  Unspecified  |         Cpu:  Unspecified                          
--------------------------+-------------------------------------------------
Comment (by nikos):

 > > can you provide an example using Spearfish's bugsites + elevation.dem
 or something from the NC standard dataset?
 >
 > Right. I'll try to re-produce this in Spearfish.

 {{{
 # launch grass65 in spearfish60
 nik at vertical:~$ grass65 /geo/grassdb/spearfish60/user1/
 }}}

 copy-paste-execute everything below!

 {{{
 # define a region
 g.region n=4921000 s=4915000 w=591000 e=598500 res=500 -pa

 # create a vector grid
 v.mkgrid map="tempGrid" grid=12,15 position="region" angle=0 breaks=3 --o

 # extract centroids from vector grid
 v.extract input="tempGrid" output="tempGrid_centroids" type="centroid"
 layer=1 new=-1 --o

 # draw maps
 d.mon x0 && d.vect tempGrid && d.vect tempGrid_centroids col=blue

 # convert centroids to points and draw
 v.type in=tempGrid_centroids out=tempGrid_points type=centroid,point
 d.vect tempGrid_points col=red

 # add column to be updated from raster
 v.db.addcol tempGrid_points column="landcover integer"

 # match region with "-a" indeed, it "grows" outwards (or it just stays
 like it was before!?).
 g.region vect=tempGrid_points -pa

 # check visually: nothing changed
 d.redraw

 # now try _without_ "-a"
 g.region vect=tempGrid_points -p

 # obviously the region has been altered (as expected): check (also) the
 numbers!
 d.redraw

 # now try again with "-a"
 g.region vect=tempGrid_points -pa

 # there is no "returning back" or "growing"! (only when using res=500
 again, keep reading)
 d.redraw

 # update points from landcover.30m raster and check table
 v.what.rast tempGrid_points rast=landcover.30m column=landcover
 db.select tempGrid_points
 # works fine!

 # assume that the raster was just about the same size of what the
 (grass-)region is:
 r.mapcalc landcover_cropped = landcover.30m

 # add a second column and update again
 v.db.addcol map=tempGrid_points col="landcover2 integer"
 v.what.rast vect=tempGrid_points rast=landcover_cropped col=landcover2

 # check updated table
 db.select tempGrid_points
 # works fine too!


 # now draw raster and points
 d.erase
 d.rast landcover_cropped
 d.vect tempGrid_points col=red


 # here is where the problem begins:
   # first check current region size
 g.region -p
   # for some reason I was using the "res" parameter with g.region while
 looping over several point vector maps to get them updated from rasters
   # so, let's use res=500
 g.region vect=tempGrid_points -pa res=500

 # note the changes in the region:
   # the number of rows, cols. The region growed outwards (as expected)!
   # north is now larger
   # south has become smaller (!?)
   # west has become smaller (!?)
   # east is now larger

 # now repeat v.what.rast
 v.what.rast vect=tempGrid_points rast=landcover_cropped col=landcover2

 # check updated column
 db.select tempGrid_points
 d.vect tempGrid_points col=blue where="landcover2 > 0"

 # In this example I see points at east, south not being updated (actually
 updated with NULL)!
 # Missing values only at (some) the edges! huh??
 }}}

 Nikos

-- 
Ticket URL: <http://trac.osgeo.org/grass/ticket/789#comment:4>
GRASS GIS <http://grass.osgeo.org>