[GRASS-user] Problem with computation Shannon's diverzity index

[Tomas Zigo]

Hi Everyone,

I have trouble with computate Shannon's diverzity index by module
r.li.shannon [1] for configuration files created by module r.li.setup
[2].

Input layers for computation are:
1. Base raster layer representation reclassify slope with 3 m
resolution of cell size. 
2. My sampling units are centered vector site (vector layer of points
with 30 regular distribution points).

I defined my configuration file by r.li.setup as follows:

1. Insert new sampling area settings:
- Configuration file name: Shannons_diverzity_index_100_m
- Raster map name: Slope_t70
- Site file to overlay: Points

2. Define a sampling frame (region for analysis):
- Whole map layer

3. Define sampling areas:
- Sample units

4. Use keyboard to enter sampling units dimension

5. Select type of shape:
- Rectangle

6. Select Method of Sampling Unit Distribution:
- Centered over sites

7. What wide size (in cells) for each sampling unit? 
33

8. What high size (in cells) for each sampling unit? 
33

I created several (20) configuration files containing changes in
parameter wich define wide and high size in cells units for each
sampling unit (view point 7. and 8.) with centered over point vector
layer. I was increasing this parameter about value 33 cells (100 m)
from begin wide size 33 cells and high size 33 cells (100 x 100 m) to
size wide 667 cells and size high 667 cells (2000 x 2000 m).

Computation was successfull for configuration files with these
size for each sampling unit centered over points.
1. wide size 33 x high size 33 cells (100 x 100 m) 
2. wide size 67 x high size 67 cells (200 x 200 m) 
3. wide size 99 x high size 99 cells (300 x 300 m) 
...
7. wide size 233 x high size 233 cells (700 x 700 m)

However computation for configuration files with wide size 267 cells and
high size 267 cells (800 x 800 m) and 300 x 300 cells (900 x 900
m) wasn´t executed for all (30) sampling units. Computation for
sampling units bigger than 300 x 300 cells (900 x 900 m) e.g. size 333
x 333 cells (1000 x 1000 m) and more I got this error message:
r.li.worker (pid 32701) exited with abnormal status: 11.

dmesg | tail
r.li.shannon[8171]: segfault at bb600000bb5 ip 00007fbd1fe51cde sp
00007fffd67ee360 error 6 in libgrass_gis.6.4.1svn.so[7fbd1fe31000+4f000

Is any limitation for size of sampling units here?

Info:
OS: GNU/Linux
Distribution: Arch Linux
Architecture: x86_64

GRASS GIS version:
GRASS 6.4.1svn (2011) 
Revision: 43636 
Date: 2010-09-22 13:18:42 -0700 (Wed, 22 Sep 2010) 
 
GRASS 6.5.svn45714 (2011) 
Revision: 45467 
Date: 2011-02-26 12:31:24 -0800 (Sat, 26 Feb 2011) 

I uploaded my GISDBASE directory with LOCATION Zaujmove_uzemie and
MAPSET PERMANENT, which data used for computation and configuration
files of module r.li.setup and outputs of module r.li.shannon here:
http://leteckaposta.cz/935221057

Thanks,
Best regards Tomas Zigo

[1] http://grass.osgeo.org/gdp/html_grass64/r.li.shannon.html
[2] http://grass.osgeo.org/gdp/html_grass64/r.li.setup.html