[STATSGRASS] gstat problem on Linux

[Edzer J. Pebesma]

Hi Thomas, I remember trying to look at your data about a
month ago, but it required more GRASS expertise than I could manage.
So my main question is: can you make a small, reproducable example
that I could investigate (on linux only)?

Do both your GRASS installations have the same version?
--
Edzer

Thomas Adams wrote:

> I have been using GSTAT with GRASS successfully on a Mac OS X system 
> with no problems. When I move the **IDENTICAL** files to a Linux 
> system (RedHat Linux 7.3) the GSTAT variogram code works fine, but 
> when I attempt to generate a predicted map using ordinary kriging I 
> get a singled valued map, which is completely different from the 
> results with the same data on my Mac OS X system.
>
> I have attached 2 gif files that show the difference; the 
> "precip.200301_pr.map.gif" is generated on Linux (the one that is bad 
> - the single value everywhere is 1.528851), the "2003.pr.gif" was 
> generated on Mac OS X, which is correct.
>
> gstat kriging output:
>
> GRASS:/grass/data > gstat gstat.krige.200301.precip
> gstat: Linux version 2.4.3 (04 July 2003)
> Copyright (C) 1992, 2003 Edzer J. Pebesma
> using Marsaglia's random number generator
> name|sta.precip.200301.15k
> desc|s.in.ascii sites=sta.precip.200301.15k 
> input=sta.precip.200301.grass fs='|'
> GRASS site list sta.precip.200301.15k: 0 cat, 2 dim, 0 str, 1 dbl.
> data(precip.200301): gstat/grass: 414 sites read successfully.
> sta.precip.200301.15k     (GRASS site list)
> attribute:             col[1]     [x:] x_1        : [    
> 193192,1.74381e+06]
> n:                        414     [y:] y_2        : [    
> 992152,2.21716e+06]
> sample mean:          1.64865     sample std.:          0.87102    
> [using ordinary kriging]
> ncols 305
> nrows 242
> initializing maps ..
> 84% done
> CREATING SUPPORT FILES FOR precip.200301_pr.map
> CREATING SUPPORT FILES FOR precip.200301_var.map
> 100% done
>
> One twist is that during the gstat build I get the following:
>
> .
> .
> .
> gcc -c -O  -DHAVE_CONFIG_H hsehldr.c
> ar ru meschach.a lufactor.o chfactor.o qrfactor.o solve.o norm.o 
> hsehldr.o
> ranlib meschach.a
> make[1]: warning:  Clock skew detected.  Your build may be incomplete.
> make[1]: Leaving directory `/grass/gstat-2.4.3/meschach'
> cp meschach/meschach.a lib/libmeschach.a
> (cd src; make)
> make[1]: Entering directory `/grass/gstat-2.4.3/src'
> .
> .
> .
>
> I have found on the internet that this type of warning can occur with 
> older versions of Gnu make, but I am using version 3.79.1.
>
> Any and all ideas are welcomed!
>
> Regards,
> Tom
>
>
> ------------------------------------------------------------------------
>
>
> ------------------------------------------------------------------------
>
-------------- next part --------------
Skipped content of type multipart/related