[GRASS5] raster inaccuracy
roger at spinn.net
Tue Nov 9 20:11:19 EST 2004
The model region is defined by the authors of the model. The model origin is
at a more-or-less arbitrary location and the boundaries are such that it
divides equally into 188X256 rows and columns, each 5280 feet wide.
I'm not sure what the actual boundaries of the region were when I produced the
illustration. That region was set by zooming in on the corner then panning
to the northwest to get a clear view of the corner. The actual size of the
mismatch varies with the location of the origin and with the resolution of
the region. I can make the raster representation accurate by going to very
fine resolutions. Otherwise, if there is any situation in which the raster
map is accurate then it would probably -- as you suggest -- depend on a
getting the right combination of origin, resolution and cell size.
The view in the X monitor is essentially identical to the illustration
provided by the png monitor.
This is not a problem with importing data. The vector representation of the
model grid was constructed as a Grass ascii vector file. Using d.where the
grid looks correctly located and dimensioned. The raster file in this case
is not a model file. The raster file was created directly off the vector
representation of the grid using v.to.rast. The raster file and the vector
file represent the same data, but the raster representation is inaccurate,
with the size of the inaccuracy varying with the region resolution and
On Tuesday 09 November 2004 16:45, you wrote:
> > I use Grass in part to support finite difference ground water models.
> > There is a substantial inaccuracy in Grass' raster representation that
> > I've known of for some time. The problem surfaced again recently,
> > caused errors in my reported results and required me to repeat several
> > hours of work at substantial costs. I think this problem has been
> > present from early betas of 5.0 through 5.3 (our current installation).
> > The png file at
> > http://members.spinn.net/~roger/mismatch.png
> is your region something like this:
> or more to the point is your input data like the above and your region
> as a test, do
> g.region res=25 # or res="some common denominator"
> and see if things start to line up in the XDRIVER.
> (the PNG driver is much less tested than XDRIVER, maybe other bugs..)
> I've had the same trouble with importing model data; setting the region
> resolution to half, or setting the bounds as the first example above
> (25-75) fixed it.. but you have to be careful it doesn't inadvertently
> switch back, also when you do overlays with maps which use a 0-50 bounds
> settings the computer is doing what you thing it is doing...
> maybe it?
More information about the grass-dev