[GRASS-dev] r.cva output file missing

GRASS GRASS developer
1

Rebecca Mayer wrote:

What is the size and x-y resolution of your mapset?

nsres: 86.08807728
ewres: 86.08807728
rows: 3793
cols: 3719
cells: 14106167

Try adjusting the cell sizes so that you have
integer cell extents.

r.cva works in this mapset when set to 'random' but
not 'sites' type of analysis. Would the cell extents
still be the problem?

Yes, but only if you are running this on Windows.

Below are the terminal outputs for three runs using
the same the dem file. In the first run I used a
binary raster map to define 1 viewpoint for a 'sites'
analysis and got a segfault. In the second run I used
a vector map to define 1 viewpoint for a 'sites'
analysis and got a segfault. The third run was set to
'random' with a sample frequency designed to produce
one viewpoint. This run completed successfully.

Terminal output:
-----------------------------------------------------
GRASS 6.2.2 (Lambert):~ > r.cva -f input=sle_demp_5pt
viewpt_mask=sle_unc3_1pt output=sle_view13 obs_elev=60
target_elev=1.6 max_dist=2000 type=sites curvc=0.13

Segmentation fault

This one crashes because you are using an unsupported combination
of options. Type=sites *must* be used in conjunction with an
input vector map of site location(s). This is really a bug in
the r.cva options parser. Instead of crashing, it should show
an appropriate error message.

GRASS 6.2.2 (Lambert):~ > r.cva -f input=sle_demp_5pt
sites=sle_unc3_1pt output=sle_view14 obs_elev=60
target_elev=1.6 max_dist=2000 type=sites curvc=0.13

100%
Creating raster map 'sle_view14'
Writing raster map 'sle_view14':
100%

Sample: requested = 10.00000% attempted = 0.00001%
actual = 0.00001%

Creating support files WARNING: unable to find
[sle_view14] in [viewtest]
WARNING: category support for [sle_view14] in mapset
[viewtest] missing
Segmentation fault

This one looks strange. The calculations actually do finish
correctly but writing the final output map fails for some reason.

Try dropping all options except for type=sites and sites=<vectmap>
and work your way from there.

Careful with option "curvc". It is basically untested and
nobody ever checked whether it gives sane results.

Which version of r.cva are you using?

Where did you download it?

Benjamin

GRASS 6.2.2 (Lambert):~ > r.cva -f input=sle_demp_5pt
output=sle_view151 obs_elev=60 target_elev=1.6
max_dist=2000 seed=1 sample=.00001 type=random
curvc=0.13

Creating raster map 'sle_view151'
Percentage complete:
100%

Writing raster map 'sle_view151':
100%

Sample: requested = 0.00001% attempted = 0.00001%
actual = 0.00001%

Creating support files
Job finished
----------------------------------------------------

--
Benjamin Ducke
Senior Applications Support and Development Officer

Oxford Archaeological Unit Limited
Janus House
Osney Mead
OX2 0ES
Oxford, U.K.

Tel.: ++44 (0)1865 263 800
benjamin.ducke@oxfordarch.co.uk

2

By going through the r.cva options I finally figured
out the key to doing 'sites' analysis without
triggering a segfault: use the -i flag. This flag
seems to tell r.cva to ignore fields in the vector map
that may define the parameters spot, offseta, offsetb,
azimuth1, azimuth2, vert1, vert2, radius1, and
radius2. (My sites map did not have fields with these
parameters.)

Example of what worked:
---------------------------------------
GRASS 6.2.2 (Lambert):~ > r.cva -f -i
input=sle_demp_5pt sites=sle_unc3_1pt
output=sle_view18 obs_elev=0 target_elev=0
max_dist=2000 type=sites

100%
Creating raster map 'sle_view18'
Writing raster map 'sle_view18':
100%

Sample: requested = 0.00004% attempted = 0.00004%
actual = 0.00004%

Creating support files
Job finished
---------------------------------------

The parameters spot, offseta, etc. when entered in the
r.cva command work as global settings, but I still got
a segfault unless -i was also used.

e.g.:
---------------------------------------
GRASS 6.2.2 (Lambert):~ > r.cva -f -h
input=sle_demp_5pt sites=sle_unc3_1pt
output=sle_view20 max_dist=2000 type=sites spot=0
offseta=0 offsetb=0 azimuth1=0 azimuth2=360 vert1=90
vert2=-90 radius1=100 radius2=1000 curvc=0.0

100%
Creating raster map 'sle_view20'
Writing raster map 'sle_view20':
100%
Segmentation fault
---------------------------------------

Thanks for your inputs and suggestions! Without them I
might have given up.

One quick note on an inconsistency I noticed between
the r.cva extension and its documentation: The help
file says "The range is -90 to 0 for 'VERT1' and 0 to
90 for 'VERT2'" whereas the r.cva dialog window sets a
range of 0 to 90 for VERT1 and -90 to 0 for VERT2.
Don't know if this affects the calculations.