[GRASSLIST:3511] Re: help with integrating models?

Roger Miller rgrmill at rt66.com
Fri Apr 12 07:59:40 EDT 2002


On Friday 12 April 2002 00:57, Alex Thorn wrote:

> I am a graduate student and UC Davis, working on a
> Masters in Agronomy.  For my research project, I am
> interested in constructing a spatial model of
> waterfowl
> migrations, including their use of flooded rice fields
> as habitat.  My goal is to represent this system by
> integrating a bird behavior model into a GIS.

The only way I know of to interface a model with the GIS is through 
compatible file types.  GRASS offers "ascii" formats for raster files, vector 
files and point data.  The simplest interface requires that your model write 
data in those formats or that you have a third set of programs that converts 
data from the model format to the GIS ascii format.  You also need routines 
that move data from the GIS ascii formats to your model's formats.  A far 
better but more demanding method is to write programs that use the GRASS 
libraries and can read data in your model's output format and write it in a 
GRASS native format, and visa versa.

For a dynamic model you may need to add a second level to the  interface.  
That would be e.g. a program (possibly a script) that moves data between the 
GIS and your model, processes and displays results, exports data and reruns 
the model.  An example for a set of sensitivity runs might do something like:

run the models --> export to GIS --> process and display -->

summarize and query user --> construct new input --> export to model -->

run the models

Of course, the flow in your simulations is likely to be different from this 
simple example.


Happy computing,
Roger Miller



More information about the grass-user mailing list