ArcGIS Pro method for simple area weighting interpolation

445
4
02-18-2021 03:00 AM
Labels (1)
Bethpc9
New Contributor II

What method (using ArcGIS Pro 2.7) would you propose for carrying out simple area weighting interpolation (also called areal weighted interpolation)? 

I do not want the intricate Areal Interpolation with models as is offered through the Geostatistical Wizard. My project is comparing a simple AWI to a dasymetric method; transferring population count from census tract to a raster format (after which it can be applied to any areal unit).

So I simply need to create a raster output whereby each grid cell is assigned a population value based on the percentage area of the host/source areal unit (census tract). 

0 Kudos
4 Replies
DanPatterson
MVP Esteemed Contributor

not sure but

Tabulate Intersection (Analysis)—ArcGIS Pro | Documentation

or one of the similar tools in that toolset


... sort of retired...
Bethpc9
New Contributor II

Thanks Dan - I think you're definitely onto something there. I couldn't quite figure out how to apply it so have gone a long way around: 

  • Converted my census polygons to raster
  • Used Zonal Statistics as Table to count the number of pixels in each polygon 
  • Divided each census polygon 'population' by number of pixels to get population per pixel
  • Used Zonal Statistics again, this time using a different set of polygons (my target zones) to 'sum' the values of those population pixels. 

This was probably a very clumsy way around. Please correct me if this could have been done a different way. 

0 Kudos
DanPatterson
MVP Esteemed Contributor

The only concern would be your 3rd step.  You have spread the population amongst all the pixels which may not reflect the original distribution (aka, density)


... sort of retired...
0 Kudos
Bethpc9
New Contributor II

Thanks Dan, definitely a warranted concern. Fortunately my data is population 'count', and the project is trying to provide a better alternative to simple area weighting interpolation to avoid this concern. But I'm currently trying to do this more simple areal weighting as a base to compare methods to. 

Hope this makes a bit more sense in context?

0 Kudos