[GRASS-user] Exotic patch based statistics, a solution...

[Hufkens Koen]

Hi list,

just want to share some code that could be useful to some of you.

I was looking for a way to do other patch based statistics than those provided in r.statistics. Using masks and a loop in GRASS was terribly slow (certainly for very patchy areas (in my case some 100000). So this is the fastest I could make it without programming a module myself. I use R, but not the spgrass6 link, from within a bash script.

The script uses a map with all numbered patches as can be provided by r.le.patch or r.le.stats and a cover map which contains the values u want to evaluate with some exotic statistic or function.

In my setup I formatted the output as a rules file so the results could be translated back into an actual map using the patch number map as common factor. Other approaches are also possible, depending on the need of the user.

Hope some people can benefit from it.

Cheers,
Koen

-- THE BASH SCRIPT ---->

#!/bin/bash
# 
###########################################################################################################################
# 
# Grass bash script to evaluate the content of patches based upon a patch and cover map...
#
# !!! Only works within the GRASS terminal !!!
#
#
# Calculate the a map with numbered patches with r.le.patch or r.le.trace. Use this as input for the patches.number.map.
# The cover.map map is the map you want to evaluate based upon the delineated patches with another function than those
# provided by r.statistics.
#
# use: sh myscript.sh patch.number.map cover.map (name of the script depends on how you save it ofcourse)
#
###########################################################################################################################

r.out.xyz input=$1 output=- fs=' ' | awk '{print $3}' > patches
r.out.xyz input=$2 output=- fs=' ' | awk '{print $3}' > cover

#
# Make a two column matrix with first the patch number than the associated cover value
#

paste patches cover > Rinput


###########################################################################################################################
# 
# R code starts here! 
#
###########################################################################################################################
R --slave --vanilla --quiet --no-save  <<EEE


rdata <- read.table("Rinput")

for (i in 1:nrpatches){

rdatasub <- subset(rdata[,2],rdata[,1]==i)

#
# complicated patch based statistics go here...
#


}
EEE

#
# Close R session within bash GRASS script...
#

###########################################################################################################################


#
# The output of the R script is best printed according to the rules specifications so you can use the results
# directly as a reclassify rule within GRASS.
#


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/grass-user/attachments/20081213/1609953d/attachment.html