[GRASS-user] v.buffer and g.region shift raster grid?

[Christoph Heibl]

Dear list,

I am puzzled by a phenomenon in GRASS that I cannot really explain and that spoils some of my downstream applications after preparing my data in GRASS. 

Problem: 

I have a set of vector points and some raster layers with environmental data . For each point, the environmental data of the raster cell it lies within has be to be sampled (either with v.what.rast or by the downstream application [Maxent]). So far, so good and no problem. Now imagine only a subset of the vector points is taken to sample the raster data and this is achieved by laying a buffer around this subset of points and then setting the region to this buffer. What happens now is that the vector points are shifted with respect to the raster grid (or vice versa?). This pushes those of my vector points, which lie near the coastline, into the sea and the downstream application skips them, because it cannot link them to environmental data.

These pictures may clarify the point:

http://www.christophheibl.de/points1.png: This is the default setting. The three red points represent my set of vector points.

http://www.christophheibl.de/points2.png: Now only one of the three red points is choosen as a subset and displayed, but without changing the region setting. All points keep their position relative to each other and the raster grid as in points1.png.

http://www.christophheibl.de/points3.png: The same as in points2.png, but this time a buffer was laid around the red point and this buffer was used to update the region settings. Now, as described above, the vector points are shifted with regard to the raster grid and the whole grid seems to be centered in a way that the red point comes to lie at the corner where four raster cells meet.

Question:

1. Why does this shifting occur? This is somewhat frightening for me, because I thought that the position of the grid would be fixed and given by the projection. 

2. How can I can make sure that the relationship between the raster grid and my vector points "stays the same", when I "zoom in" on subsets of the study region?

Best wishes
-- Christoph

________________________________________________________

Christoph Heibl

Systematic Botany
Ludwig-Maximilians-Universität München
Menzinger Str. 67
D-80638 München
GERMANY