GIS

[Ned Gardiner]

Hello Alex,

We did not get a chance to meet this summer when I was attending
the summer school, but better late than never.  Here are some
thoughts for your discussion.  I work with GIS
for ecological analysis with the Long Term Ecological Site 
called Coweeta (in North Carolina).  I am not a cartographer;
rather, I have been slowly picking up spatial statistics while
immersing myself in a proprietary software package called Arc/Info
which, I am sure you know, is put out by ESRI.  The ESRI product
is a very involved piece of software written in billions of little
macros which a user can piece together, either using AML, an older
standard designed for the main products of the past, or AVENUE, 
another scripting tool, but this one object-based and for use
with ArcView.  I have had neither the desire nor need to learn
ArcView to this point, but it represents the future of ESRI's 
GIS software.  AML and AVENUE allow one to make system calls to 
run user-written software, in addition to facilitating complex
analysis using only ESRI code.  All that is to say that your 
first objective ("low-level integration" or whatever exact
term you used) is pretty well accessible right now.

My own experience with modeling and statistical analysis using
GIS is that GIS is not a powerful tool for the big jobs of 
analyzing pattern.  The reason is that many GIS are relational
databases and nothing more.  So this data structure is not efficient
for computation.  I usually export data, do something to it, and
re-import it for map display.  The exceptions to this rule would
not impress you, given your objectives with the SWARM project.

Back to AVENUE.  ESRI is pushing the object thing on NT.  I am 
bracing at this and its commercial implications, so I applaud your
efforts to develop SWARM for modelers.  If you look into fragstats,
you'll see that the GRASS interface is not out of reach.  However,
GRASS is no longer freeware (at least the latest versions).  There 
are a couple of educational packages you might want to look at.

* DMS- desktop mapping system; this is a remote sensing/photogrammetry
package for DOS which does loads of stuff.  Unfortunately written
in fortran and not portable at all right now.  You can contact Tom
Jordan by address@hidden if you are interested.

* Idrisi- This is my favorite GIS package (even though I NEVER use it
since I have Arc/Info on my Sparc 5 which is pretty hard to 
turn your nose at) because a student can by the thing for about 
$150.  It is designed for and by researchers, so there are very 
quick and good spatial stats built in.  I can't tell you more 
since the last time I used it was about 4 years ago.  

I was very impressed by your GIS incorporation with the archaeological
data for the Four Corners region.  I look forward to learning swarm 
and implementing some models using GIS.  Ultimately, I would be inter-
ested in the highly integrated options you point to, but I would be 
more than psyched to see a facile way to simply import/export data.
I have been using Splus (MathSoft/AT&T) and its spatial stats/GISLink
modules to bring data in and out of Arc/Info.  That is an object-based
language worth looking at for ideas since they have developed statistical
routines already.  

Good luck.  
-- 
Ned Gardiner
Coweeta LTER GIS Coordinator
Institute of Ecology
University of Georgia
Athens, GA  30602-2202
(706) 542-5691
(706) 542-6040 FAX

                  ==================================
   Swarm-GIS is for discussion of Geographical Information Systems (GIS)
   and their integration or use with Swarm.  For list administration
   needs (esp. [un]subscribing), please send a message to
   <address@hidden> with "help" in the body of the message.
                  ==================================