Hi all.
We used r.le a lot, but we're not satisfied by its performances, ease of use,
and maintainability. A previous attempt at incorporating Fragstats as a grass
module (thanks Jachym!) revealed theat the code is not maintained very
actively, and probably not worth the effort. Therefore, a couple of students
have produced an interesting new C module for grass, with a rather different
approach to the problem. It is not functional yet, but it has been built with
expandability in mind, so it should be straightforward to include new
indices. I include here their note. It would be good if people could see if
the general structure and idea is ok, and help us completing it.
You can download it from:
http://www.faunalia.it/download/r.li.tar.gz
Looking forward to hear from you.
pc
Hi again, we post a development version of our work about r.li module.
The pourpose of this project is the parallel computing of focals index
on raster maps or areas.
It is based on sampling area definition of r.le but we tried to made a
more upgradable module to make definition of index indipendent from
areas definition management.
We wrote two programs:
1 - r.li.setup: this program defines the sample areas for the analysis
with a gui interface. It is based on sampling frame- and sampling area
definition of r.le.setup, but it have two novelties: the first one is the
gui, and the second one the format of configuration file that is
raster-indipendent. It means that you can define an area configuration
file on a raster map and use the same file on another map (the
dimension are saved in relative coordinates).
See the description file for more explanations.
2 - r.li.daemon is a client-server architecture that manage the
parallel computing of index on the areas defined by the configuration
file. It is a little long to explain but we'll try with few words.
The server reads the configuration file and build a list of areas, then
create a group of clients and send an area to everyone of this. The
clients calculates the index (it is passed with a function pointer) and
then returns the result to the server. Then the server stores the
result in a file (or on a raster in moving window mode that is under
development).
In this way a new index have only to take care of an area (defined with
a rectangle and a mask) and can be written only defining the C
function. To create a new index then you have only to copy the main
program and change the simple index with your index and pass your
function calling the calculateIndex function... We hope it will be
useful.
Please send comments or suggestions...
Davide Spano, Claudio Porta University of Pisa
--
Paolo Cavallini
email+jabber: cavallini@faunalia.it
www.faunalia.it
Piazza Garibaldi 5 - 56025 Pontedera (PI), Italy Tel: (+39)348-3801953
