> Andrew,
> The simplest scripting approach would be using the GeoFormula process
> in TNTmips, where you could use the Multi-Criteria Analysis Model
> option from the GeoFormula Creation Wizard. This option lets you set
> up a multi-criteria decision analysis model using any number of input
> spatial objects. It uses a weighted linear combination form of MCDA,
> where each input object represents one of the decision factors, and
> you assign a relative importance weighting factor to each object. You
> then use the cell values in a raster object or the database attributes
> of polygons in a vector object to set up a table of "scores" for each
> input object that quantify the spatial variation in that factor. For
> each output cell in the GeoFormula result, the cell value is computed
> by multiplying the score at that location in each object by the
> weighting factor for that object and summing these products.
> In your application this linear combination approach isn't ideal,
> because most of your factors are not gradational in value, but yes/no,
> in/out criteria. But you could assign scores of 0 and 100 for these
> binary criteria (for example), such as 100 for inside the 500-m road
> buffer and 0 for outside, give each of the 5 factors an equal
> weighting value (20%), and then only output cells with a value of 100
> would satisfy the siting requirement for all factors.
> There are several Technical Guides on the Microimages web site
> describing MCDA and the GeoFormula process:
> www.microimages.com/documentation/cplates/65mcaplate.pdf
> www.microimages.com/documentation/cplates/65gfwizard.pdf
> Given the binary nature of the criteria, though, Kevin's region
> intersection approach might be simpler to set up and carry out.
> --- Randy at MicroImages
> On Jul 25, 1:42 am, Andrew Terhorst <andrew.terhorst.cs...@gmail.com>
> wrote:
> > I have to site 14 automatic weather stations in a 3350km2 river
> > catchment. We intend using these for real-time rainfall-runoff
> > modelling. I want to use TNTmips to locate optimal sites for the
> > automatic weather stations.
> > I have certain restrictions where I site these stations:
> > 1. They need to fall within a GSM-based mobile telephone reception
> > area (data is communicated form the automatic weather stations to a
> > central database via a GSM-based mobile telephone network).
> > 2. The stations must be located in a 1 hectare clear space (i.e. no
> > vegetation canopy)
> > 3. The stations cannot be located in a national park (too much red-
> > tape involved seeking permission to do so).
> > 4. The stations must be accessible by road (vehicle) i.e. located
> > within 500m of a road.
> > 5. The stations must be located in those parts of the river catchment
> > where flow accumulation is highest i.e. those parts of the catchment
> > that contribute most run-off.
> > I have the following spatial data at my disposal
> > A. DEM
> > B. Mobile telephone coverage map
> > C. Land ownership
> > D. Land cover
> > E. Road network
> > F. Climate surfaces (annual and monthly gridded rainfall/evaporation/
> > radiation surfaces)
> > I can encode business rules in some script that employs raster algebra
> > I guess and maybe play around with the watershed modelling tool.
> > What advice can you give me?
> > Cheers
