[seek-beam] GARP pipeline diagrams

[Ricardo Scachetti Pereira]

      Dear, SEEK BEAM WG members,

    I believe that this is the first message posted to the SEEK BEAM 
mailing list, so welcome you all!!

    As I promised during our last meeting in New Mexico, I produced 3 
pipeline diagrams showing the data flow and the analytical steps 
involved in GARP Analyses.
    The diagrams are available at SEEK CVS repository, at

    http://cvs.ecoinformatics.org/cvs/cvsweb.cgi/seek/projects/#dirlist    

    From there, follow the links through /beam/niche modeling/garp 
directories to get to the diagrams.

    These are first drafts and I would like you guys to take a look and 
make suggestions and comments.
    There is a diagram for a native prediction, one for invasive species 
analysis and the last one is for climate change analysis.   
    The numbers in parenthesis on each process box identify unique 
processes (repeated number refer to the same analytical step, but with 
different input data). Each data flow is also marked with one letter. 
Same letter refers to same data flowing (not only type but identical 
data). Should those letters refer to data flow types rather than unique 
data sets flowing? I think so, because that (data type) is what will be 
the target of SMS & KR WGs.
    The diagrams are a little bit oversimplified, regarding two 
procedures that are frequently used in GARP analysis. They are: i) 
jackknife of env layers to define which ones are more appropriate for 
analysis (between analytical step 1, 2 and 3), and the best subset 
procedure, in which we generate many GARP models and then sum them up to 
create a probability map instead of a boolean presence/absence 
prediction map. Those procedures might make the pipeline more 
complicated as they might require for loops and other complex 
constructions. So let's leave them out for now. I can add those 
complexities later on the diagrams for information purposes, if required.
    If the pipelines look good, I can add more detailed specs on each 
process and each data flow.
    Feel free to send comments and ask questions.

    That is about it for now.
    All the best,

Ricardo

PS.  I'm (B)CC'ing this message to all Kansas KDI developers, also. We 
had some discussions about SEEK after the NM meeting, and they might be 
interested in following up. I would also like to invite the Kansas croud 
to join the mailing list, as I will use that as the primary 
communication means regarding BEAM activities.
 

-- 
Ricardo Scachetti Pereira
Research Manager
Centro de Referência em Informação Ambiental - CRIA
http://www.cria.org.br/



--
This message has been scanned for viruses and
dangerous content and is believed to be clean.