0

I have a weighted overlay that identifies the most important areas in a corridor scaled from 0 - 100. I would like to be able to click on a cell and be able to identify which input rasters contributed to the weighted overlay.

I think I should convert all inputs into weighted overlay to a value of 1, do a combine and then combine with the weighted overlay?

Any ideas?

I used Combine in my related question Cell statistics sum resulting cells to indicate original input rasters

| improve this question | |
  • Links to other questions are automatically translated to the question title. Otherwise you can use the format [text of link](url) to insert a link. If some of your rasters have no data areas and you want to figure out which ones do or don't factor into the overlay, then your first part works (input factors to binary, Combine them). But you can only get what you want from that raster. If you then Combine that with the overlay result you're likely to have a tremendous number of unique values which will kind of make it useless. You may be looking at a raster attribute table or multiple bands. – Chris W May 21 '15 at 5:06
  • Actually even a raster attribute table won't do it. You're trying to store two unique values in a single raster with a single band (if you want a one click id of both overlay value and contributing info). Without some coded value trickery, this isn't possible. – Chris W May 21 '15 at 5:18
  • what does coded value trickery mean? something complicated i am guessing? – Amanda May 21 '15 at 11:15
  • Not so much complicated in the execution as in use. Assume you have few enough inputs there are a maximum of 999 possible combinations. You'd take your overlay output values times 1000, and then run the whole Combine thing on the binary versions of the inputs. Then you'd add those two results together. Your final values would have values where the one/ten/hundred digits coded the input combination and the 1k/10k/100k digits were the overlay rating. You'd have to have a key for those input values, and know that the raster value was actually two values, nevermind symbolizing it. – Chris W May 21 '15 at 18:21
0

The only way I can think of doing this in a raster environment is to take each raster that contributed to the weighting and reclassify into multiples of 10 then add all these together. At this point I am assuming your weighting rasters were essentially binary in nature.

For example you had 4 weighting rasters A, B, C and D and for each raster it had a value or zero. You then reclassify these rasters into multiples of 10:

A = 1 B = 10 C = 100 D = 1000

So add all these together to "create a who contributed raster". You can click on any cell and find out who had contributed to it. If you clicked on a cell and it returns a value 1100 then only C & D contributed. If it 1001 then only A & D contributed.

This approach would work with a small number of weighting rasters. You do not say how many you have so if it was it hundreds I would imagine you exceed the computers capability to store such a number.

| improve this answer | |
  • some input rasters are binary but some have scores associated with categorical classes that are then weighted. there are 7 input rasters. in the above, do you mean do the weighted overlay and then add to the "who contributed raster" because the weighted overlay identifies the most important areas by assigning 100 to the best pixel, this would affect the result. if i am understanding properly. but maybe i can use a version of this approach.... – Amanda May 21 '15 at 11:14
  • You would create 7 new rasters from your 7 weighted rasters. You reclassify them into the coding scheme I suggested and add those together. That gives you your who contributed raster. You then use your original 7 rasters in your weighting analysis as you have. So you have two final rasters one your analysis and the other I'm suggesting. You do not do your weighting analysis with the reclassified rasters. – Hornbydd May 21 '15 at 16:12
  • This is essentially another way of doing what the Combine tool does. It would take the binary rasters and give you a new raster that had a unique value for each input combination. Your method just assigns each raster a digit in a value and adds them. The problem, as I understand it, is the asker wants both the contributing value and the actual overlay analysis value in the same cell value somehow. As you say, I don't see a way to do this and have it useable/make sense. You have to have two separate rasters for the two different values (contributors and overlay rating). – Chris W May 21 '15 at 18:29
  • @ChrisW - yeah that was how I was envisaging it, 2 rasters and when they use the info tool they query both layers at the same time. – Hornbydd May 21 '15 at 20:14
  • 1
    @user23828 Because rasters don't store information that way. A raster cell has a single value. If you look at an attribute table for a raster you'll see one row for every value and a count of how many cells there are of that value. Even if you added a raster attribute table, all cells with the same value share the same attributes. I suggest giving this help file a read: resources.arcgis.com/en/help/main/10.1/index.html#//… If you Combine the results of your contribute and overlay, think about the number of possible unique combinations - 100 to the nth (inputs). – Chris W May 22 '15 at 6:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.