[GRASS-user] Re: r.surf.nnbathy interpolation to whole region

[Maciej Sieczka]

>> Maciek wrote
> Carlos "Guâno" Grohmann wrote:

>>  Since nnbathy is cell-center oriented each
>> N,S,E,W extent has to be shifted half a current GRASS region resolution
>> to avoid it's output missalignment with current GRASS region.

> I always though that GRASS was cell-center oriented..

What do you mean?

Explanation of why the shift in r.surf.nnbathy in necessary:

The region:

$ g.region -g
n=404456
s=404453
w=315942
e=315945
nsres=1
ewres=1
rows=3
cols=3
cells=9

The input raster (* stands for null):

315942.5 404455.5 20
315943.5 404455.5 74
315944.5 404455.5 43
315942.5 404454.5 29
315943.5 404454.5 *
315944.5 404454.5 65
315942.5 404453.5 25
315943.5 404453.5 25
315944.5 404453.5 49

If the half-a-resolution shift IS applied (nnbathy -W 0 -P alg=nn -n
3x3 -x 315942.5 315944.5 -y 404455.5 404453.5), as described in
previous post, the output of r.surf.nnbathy in=xxx out=yyy is alligned
to the input raster:

315942.5 404455.5 20
315943.5 404455.5 74
315944.5 404455.5 43
315942.5 404454.5 29
315943.5 404454.5 48.25
315944.5 404454.5 65
315942.5 404453.5 25
315943.5 404453.5 25
315944.5 404453.5 49

If the shift IS NOT applied (nnbathy -W 0 -P alg=nn -n 3x3 -x 315942
315945 -y 404456 404453), the output is missaligned and interpolated
wrong, due to nnbathy working region set wrong:

315942 404456 NaN
315943.5 404456 NaN
315945 404456 NaN
315942 404454.5 NaN
315943.5 404454.5 48.25
315945 404454.5 NaN
315942 404453 NaN
315943.5 404453 NaN
315945 404453 NaN

Maciek