Hey folks,
In order to speed up the v.rast.stats script, I've made some changes
that might be of interest. I'm relatively new to GRASS so let me know
if there is a better way to do this...
The main issue was, instead of running r.univar multiple times for
each statistic, I run it once, save to a temp file, then cat the
results for each statistic...
$ diff v.rast.stats v.rast.stats.old
< # Run once only and save to temp file
< r.univar -g $RASTER > .${TMPNAME}.univar
< n=`cat .${TMPNAME}.univar | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f1`
< min=`cat .${TMPNAME}.univar | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f2`
< max=`cat .${TMPNAME}.univar | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f3`
< range=`cat .${TMPNAME}.univar | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f4`
< mean=`cat .${TMPNAME}.univar | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f5`
< stddev=`cat .${TMPNAME}.univar | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f6`
< variance=`cat .${TMPNAME}.univar | cut -d'=' -f2 | tr '\n' ' ' |
cut -d' ' -f7`
< cf_var=`cat .${TMPNAME}.univar | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f8`
< sum=`cat .${TMPNAME}.univar | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f9`
---
n=`r.univar -g $RASTER | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f1`
min=`r.univar -g $RASTER | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f2`
max=`r.univar -g $RASTER | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f3`
range=`r.univar -g $RASTER | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f4`
mean=`r.univar -g $RASTER | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f5`
stddev=`r.univar -g $RASTER | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f6`
variance=`r.univar -g $RASTER | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f7`
cf_var=`r.univar -g $RASTER | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f8`
sum=`r.univar -g $RASTER | cut -d'=' -f2 | tr '\n' ' ' | cut -d' ' -f9`
Of course, in case the user breaks out of the script, we have to add
this to the cleanup() function:
< rm .${TMPNAME}.univar
I also noticed that it runs the whole analysis at the full extents of
the raster map. Since we're looping through categories and creating a
raster mask, we only need to be worried about the extents of that
particular vector category.
I'm not sure if this is the most elegant way to solve the issue but I
create a temp vector map of the particular category and set the region
to this vector. This speeds up the creation of the raster mask for
cases when you have very large rasters and lots of smaller polygons.
<
< # Extract the current category and set region
< v.extract input=$VECTOR output=${VECTOR}_CAT${i} list=$i >
/dev/null 2> /dev/null
< g.region vect=${VECTOR}_CAT${i} > /dev/null 2> /dev/null
<
< # Make certain we are using the proper resolution
< g.region nsres=$NSRES ewres=$EWRES -ap > /dev/null
<
< #Remove temporary vector
< g.remove vect=${VECTOR}_CAT${i} > /dev/null 2> /dev/null
<
< # generate mask .....
The one problem is that if a certain vector category extent is
smaller than the current cellsize, g.region fails. Any ideas on a more
elegant way?
For the record, this reduced operation time on my project from about
240 seconds per vector category to an average of 9 seconds.
--
Matt Perry
National Center for Ecological Analysis and Synthesis
University of California, Santa Barbara
perrygeo@gmail.com