Hello everyone,
I have a question concerning the suitability values that are displayed under my suitability map in the Contents Pane. I am using ArcGIS Pro Vers. 2.9 and have chosen a suitability scale from 1-10 in the model. However, the values of the displayed suitability map randomly change. I have attached a screenshot where the suitability map and the values are displayed and are ranging from 0.41 to 10. From my understanding, this should be from 1-10 due to the scale that I have set beforehand. When I change certain things in the transformation pane this number changes.
Does anyone know what this range actually displays and why it changes?
Thanks,
Eva
Solved! Go to Solution.
I see from the image in the correspondence with Steve you are using the percent weighting method. In the Suitability Modeler, when you specify weights as percents, we multiply the criterion values by the identified percent. That is, if the weight for a criterion is 25%, we multiply each value in the input criterion by .25.
Why does your minimum value not equal 1? Let’s say you have three input criteria. The percent weights you have specified for the 3 criteria are 25%, 25% and 50%. For a particular cell location, the value for each criterion are: criterion 1 = 1, criterion 2 = 1, and criterion 3 = 1. The final suitability value for that cell will be 1 * .25 + 1 * .25 + 1 * .5 = .25 + .25 + .5 = 1.
But then let’s say for another cell, the values for the criteria are (1, 3, 1). The final suitability value for that cell will be 1 * .25 + 3 * .25 + 1 * .5 = .25 + .75 + .5 = 1.5.
In the Suitability Modeler we output the actual weighted sum value from the percent weighting. In your situation, you probably have no location in your study area that has all 1’s for each criterion, therefore the minimum will not be 1. This same logic holds why your maximum value in not 10; you probably have no locations that have all 10’s.
However, if you are familiar with the Spatial Analyst Weighted Overlay tool, the actual weighted sum values are linearly transformed (in this case stretched) to the evaluation scale (the suitability scale). As a result, the range of the output is the same as the suitability scale being used for the transformations.
We suggest using the actual weighted sum values but from this discussion, we may add a check box to the Suitability tab indicating whether you want us to normalize your values to the suitability scale. You will then have a choice to do either.
What format is your input criterion, is it a .tiff or a fgdb raster or an Esri grid?
What happens if you create a full resolution Suitability raster (hit run at the bottom of the Suitability pane)
Please post a screen shot of the transformation pane showing the 2 graphs.
I have 13 input criterion and they are all in the fgdb raster format.
When creating a full resolution raster the range has changed to 2.11-9.29. So it doesn't display the full suitability scale.
I have attached a screenshot with the transformation pane of one of my input criterion.
Eva
Would it be possible for you to zip the fgdb and email it to me together with the weight values?
Thanks
-Steve
I see from the image in the correspondence with Steve you are using the percent weighting method. In the Suitability Modeler, when you specify weights as percents, we multiply the criterion values by the identified percent. That is, if the weight for a criterion is 25%, we multiply each value in the input criterion by .25.
Why does your minimum value not equal 1? Let’s say you have three input criteria. The percent weights you have specified for the 3 criteria are 25%, 25% and 50%. For a particular cell location, the value for each criterion are: criterion 1 = 1, criterion 2 = 1, and criterion 3 = 1. The final suitability value for that cell will be 1 * .25 + 1 * .25 + 1 * .5 = .25 + .25 + .5 = 1.
But then let’s say for another cell, the values for the criteria are (1, 3, 1). The final suitability value for that cell will be 1 * .25 + 3 * .25 + 1 * .5 = .25 + .75 + .5 = 1.5.
In the Suitability Modeler we output the actual weighted sum value from the percent weighting. In your situation, you probably have no location in your study area that has all 1’s for each criterion, therefore the minimum will not be 1. This same logic holds why your maximum value in not 10; you probably have no locations that have all 10’s.
However, if you are familiar with the Spatial Analyst Weighted Overlay tool, the actual weighted sum values are linearly transformed (in this case stretched) to the evaluation scale (the suitability scale). As a result, the range of the output is the same as the suitability scale being used for the transformations.
We suggest using the actual weighted sum values but from this discussion, we may add a check box to the Suitability tab indicating whether you want us to normalize your values to the suitability scale. You will then have a choice to do either.
Thanks for your detailed answer Kevin!
This all makes sense and I agree that it would be beneficial to include a check box with the possibility to normalize the values to the suitability scale. I have read a few papers about the evaluation of suitability models and one step of the process was to calibrate the model to the whole range of the suitability scale. This was mentioned as important when comparing the results of different sites or models (e.g. Brooks, Robert (1997): Improving Habitat Suitability Index models. In: Wildlife Society Bulletin 25, S. 163–167 or Theuerkauf, Seth J.; Lipcius, Romuald N. (2016): Quantitative Validation of a Habitat Suitability Index for Oyster Restoration. In: Frontiers in Marine Science 3, S. 64.)