1

I converted a polygon shapefile which defined ED boundaries to raster. The value i based the raster surface on was a population count for each ED. The values ranged from 568 and 18160, as expected, and was automatically displayed using a color ramp.

EDs in raster format

However, these values must be normalized to the range of [0,1] for an MCE, but i am not getting the results i want. I have been using the following formula in the raster calculator to normalize:

  • ("Louth_EDs"/17592) + 1

...17592 being the input range of values, and the formula inverted because i want the EDs with the lowest population to have a higher value in the MCE. The results are as below:

EDs normalized normalized values

The inversion seems to have worked: the ED with the highest population is ranked lowest at 0, but every value outside this area has a value of 1 and nothing lower. I was hoping to make this criteria a factor in the MCE i.e. i wanted a true range of values between 0 and 1 rather than a a simply yes/no output - this criteria is to determine degrees of suitability.

How would i introduce a range of values to the normalization for the MCE? I am not adept at programming languages.

PolyGeo
  • 65,136
  • 29
  • 109
  • 338
Milo
  • 137
  • 2
  • 3
  • 11
  • 1
    Does ("Louth_EDs"/17592.0) + 1.0 work? I don't know Arc but I guess it could be treating this as integer division, hence only 0 or 1. – Steven Kay Jan 05 '16 at 17:19
  • I've just run that in the calculator now, just adding a minus to the beginning of your formula. It seems to have worked well, thank you very much. I'd include images but i'm not sure how to do that in a comment. The maximum value is 0.967..., but the lowest value is -0.032... You woudn't happen to know why the minimum would be below zero, and should this affect the MCE if all other rasters are between [0,1]? – Milo Jan 05 '16 at 17:33
  • 2
    normally, I'd use the formula ((value-minimum)/(maximum-minimum)) to normalise the range [minimum,maximum] to [0,1]. To reverse this, subtract it from 1. – Steven Kay Jan 05 '16 at 17:47
  • Thanks Steven. I tried that now (using 1.0 again) and it is inverted properly but it gives the same results as before (Boolean surface, 0 or 1). – Milo Jan 05 '16 at 19:18
  • 2
    You need to float the data http://gis.stackexchange.com/questions/174412/standardising-dem-raster-for-a-mce – If you do not know- just GIS Jan 05 '16 at 20:00
  • Perfect. The float was just giving me a value of 1 across the entire area until i tweaked where the brackets fell and used the format given in the other post:

    Float(“DEM” – "DEM".minimum) / (“DEM”.maximum – "DEM".minimum)

    ...and simply inverting that formula the way i've been doing. There is variation across the raster and it all falls perfectly within [0.1].

    I believe my question is answered. Thanks to you all, it's been very helpful and given me insight, especially in how ArcMap can sometimes treat the figures as integers. I have used float before but never knew its purpose until now!

     – Milo Jan 05 '16 at 20:42

0 Answers0