How to calculate % of population that overlaps another polygon?

3958
11
10-22-2016 07:15 AM
KathrynAngleton
New Contributor III

I am trying to figure out what percentage of my City's population falls within a certain distance of their parks. I created a 10-min walkshed via network analyst, but my population data is limited. The best I have is Census Block Groups (2014-5 year estimates). Since I don't have points for population or even just household points with household sizes in the attribute table, I'm relying on the BG polygon. I know I can intersect the BG poly with the parks network analyst output poly, but how do I calculate the percentage of the population (a field within the BG attribute table) overlaps with the parks output? Is there a better way to do this? Any ideas, input, or advice is greatly appreciated. Also, I do not have BAO or BA Desktop, I know those tools can generate population numbers based on polygon inputs.

11 Replies
RobertScheitlin__GISP
MVP Emeritus

Kathryn

  Calculation the percentage of the original polygon that intersects the parks walk distance polygon is not to hard (clip the bg poly using the park polygon and some math to determine clipped portions SqFt vs the original BG poly SqFt), but then the population number (say 43) is assigned to that polygon so you would have to assume even distribution and divide that pop by the percentage of the BG poly that intersects.

KathrynAngleton
New Contributor III

Thanks for the tip! If split doesn't work for me, I will try this.

0 Kudos
KathrynAngleton
New Contributor III

Robert, would you say tabulate intersection does this for me? It appears to have created the output I was looking for.

RobertScheitlin__GISP
MVP Emeritus

Kathryn,

   That sounds like exactly what you need. Back years ago (the last time I had to do this) that tool did not exist.

DanPatterson_Retired
MVP Emeritus

An approach which is similar to Robert's clip and calculate is to do a Split—Help | ArcGIS for Desktop but it requires an Advanced License, but it does it all at once

KathrynAngleton
New Contributor III

Hi Dan,

Thanks for the advice.  Would I want to have use ratio policy enabled on the population field within the BG attribute table? What I had originally done was had that enabled before I intersected the two polygons, but the percentage of pop I got back didn't seem right to me knowing the city and parks. 

0 Kudos
KathrynAngleton
New Contributor III

Dan, I have advanced. Split only works with string fields. Since population is numerical, how would split do what I want? I've tried tabulate intersection as an alternative.

0 Kudos
MicahBabinski
Occasional Contributor III

Hi Kathryn,

Expanding on that idea, you could consider pre-processing your block groups by erasing water and right of way polygons (if you can get them). Then, you could try to obtain address points or building footprints for your area. You could divide the block group population by number of buildings and assign each building that amount of people.

It just depends on what data you have available and how sophisticated an estimate you want to go for. I'd recommend you set up your process in Python script or ModelBuilder so you can evaluate results and refine your estimate.

Good luck! I did a similar project in my first GIS job and it was a lot of fun.

KathrynAngleton
New Contributor III

Hi Micah,

This is an interesting approach. I do have building footprints, address points, ROW polygons, and waterways. My concern about using addresses or footprints is that I would be putting residents in non-residential buildings. Land uses are mixed/integrated here, so I couldn't rely on assuming certain BGs would be 100% residential. 

To add, my Python skills are limited and I've never used model builder. Should I stay away from those processes then? I usually do more cartography than actual geography or heavy geoprocessing. 

0 Kudos