Spatial Analysis - Reclassify: assigning a measurable scale across layers?

[AllisonBorges]

Hi everyone!

Please excuse my lack of etiquette with using appropriate terms. Hopefully my question will still make sense.

I'm working on a habitat suitability model. I'm combining 6 layers where all of the old values (or range of values) have been rated on a scale of 0-100; 0 being non-habitat with increasing habitat value up to 100. I've been working with the reclass tools to assign new values, but am running into trouble with creating a natural 0-100 scale because not all of my newly assigned values are cleanly ranked from 0-100. For example (the photos below) show two factors that have been rated on a scale of 0-100. The landcover layer contains all of the values, but the distance to road layer does not; therefore, when I reclassify each layer (assigning new values) in my model they aren't measured on the same scale; I can't seem to find an effective way to input a new value of 0 for the distance to roads factor to make sure it's measured on a 0-100 scale even though it's lowest value is 30. Is there a way to re-scale the values range when assigning new values so that it is the same between all of my layers even if they don't all naturally range from 0-100? Hopefully that makes sense. I apologize if this isn't clear. Thank you for any and all help!

[DanPatterson_Retired]

And you are reluctant to reclassify roads to a 0 -X scale (perhaps by normlizing them on some basis) before combining them?

If they are Rasters, then you can use the Combine tool, but if your classes are on a 0-100 scale do the increments actually represent an interval scale (step between the numbers represent a real increment) or are you using the 0-100 scale as a nominal scale (bad, good, better, best).  Nominal data using numbers as classes comes with the warning that the actual numeric values mean nothing except a representation of the classes making up the composite. (ie your cougar file, what does 0 represent and 100 and 80? is the difference between 0 and 80 represent a 4x magnitude range compared to the 80-100 range?)

Clarifying what and how you are combining the variables would help, especially if you have a mix of nominal, ordinal and interval or ratio observations.  You can't move up a scale, you have to move down to the lowest common denominator, hence my reluctance to suggest definitely to normalize your roads data to a 0 based scale rather than leaving 30 as the lowest