Suitability Modeling - Having issues combining Continuous Functions with Unique Categories (ie Landuse)

[MarkGroom]

Hi all, I am attempting a Suitability Model for suitable Industry locations within our County boundaries. I am running ArcGIS 2.9.1. 

I am having issues outputting a viable Suitability model with metrics involving both Continuous Functions (Euclidean distance rasters derived from Road, Rail, Fiber, and POIs distances) & Unique Categories (Feature to Rasterized Zoning, Brownfields, and Parcel).

In the Suitability Modeler pane, I transform and apply weighting/function type for both type of raster datasets - but when I run the model, it outputs a Suitability Map of a single Raster cell or a raster with no information.

I have run successful Suitability models for just the Euclidean distance rasters, but when I try to model both Euclidean Distance + Unique Categories (landuse, brownfields, parcels) - I get no results.

I value any feedback or comment, still relatively new to this! Thanks for your time!

edit: for the Unique Values raster datasets, I have tried both Rasterizing the entire Parcel + Zoning dataset (and assigning weights to each individual zoning/parcel category) as well as filtering out the desired Zoning + Parcel attributes (industrially zoned, vacant, etc.), and then rasterizing such those scattered features.

[SteveLynch]

Screenshots of

1) Transformation panes of each one of the Unique categories

2) suitability map

will be useful.

[MarkGroom]

Thanks for your help Steve. I can't get into the VPN to access my work computer at the moment, but as soon as I do on Sunday/Monday, I will take screenshots and update the post. I appreciate your input.

Thanks, and have a good rest of your day.

[MarkGroom]

Hi Steve, following up on the thread from late last week. I have ran through my model quickly, in the same workflow as I would normally, in order to take screenshots. You can see here, I can add in my criteria metrics, which are a combination of unique categorical data (land use, zoning, brownfield sites) and continuous data (Distance from Highways, POIs, Fiber lines, etc.)

Below are screenshots of my categorical data (transformation pane), as well as an example of one of my continuous datasets for reference.

You can see the zoning and parcels are features that, before rasterizing, I queried to only have the criteria I want (and then rasterized before adding into the model). Which is why they are scattered, individual pieces. Please let me know if it is better to have the entire dataset in and rasterized, instead of scattered pieces. Below is the resulting output of my model - it is just one small polygon.

When running the model with only the Euclidean distance rasters, the model outputs a beautiful, suitability model (results that one would expect - though I did not attach any here). 

Are there any obvious faults I have with my inputs? Or weighting issues you see, etc.? I value any feedback, comments, or ideas you have. Thanks for your time!

[SteveLynch]

Mark

Your IndustrialZoning layer has a lot of NoData and this means that no suitability values are created in those areas. I.e if any of the input criteria have NoData in a particular area implies that the suitability at that location will also be NoData.

[MarkGroom]

Interesting, thanks for the information. Did not think about NoData fields. I will cleanly re-rasterize the Zoning, and add the input back into the model and update you. Thanks for your help.

[MarkGroom]

Hi Steve, do you have a workflow or solution on how to remove the NoData from my rasters? You can see in this sample Zoning raster, in which I just converted from vector to raster, and applied a mask, the file has NoData values (in black). Which through your initial answer, are not desired for Suitability Modeler. Or if not, do you have a solution or workflow in which you would apply categorical data into the model, such as zoning or parcel data. Thanks for your time!

Zoning (NoData values in black)Zoomed in, can still see NoData values (in black)

Thanks for your time!

[SteveLynch]

Mark

Are you

1) assigning different suitability values to each of the zones? or

2) do you only want suitability to be calculated in non-NoData areas?

for #1 I'd suggest you look at Con

and for #2, don't use the zones as a criterion but rather set it as a mask in the Environment.

[MarkGroom]

Hi Steve,

Ideally I would assigning different suitability weights for each Zoning category (Industrial weighted higher, and Residential weighted lower). Since there are 100+ Zoning categories, that is why I initially queried out the industrially zoned areas and only rasterized those select criteria (which is why there were only 3-4 categories, all weighted high, in the initial set of screenshots I sent early in the day). 

So I think I would do a little bit of #1 and #2. I would assign different suitability values for each zone, as well as suitability not calculated for NoData areas.