Which network analyst tool should I use?

960
5
09-02-2021 12:48 AM
MeganBuckland
New Contributor II

Hello everyone,

I want to find the proportion of residential polygons that can access parks within 200 metres.

- I need it to be along a network, so I cant use the buffer tool.

- I dont need to know which park is closest if there are three within 200m of one house, I simply want to find whether there is a park within 200m walking distance of a house or not.

I want to show the house polygons that have a park within 200m in one colour and the polygons that do not in another colour.

What would you suggest? Which network analyst tool would you use?

 

Notes:

I am using vertices to points for the origins and destinations so it is more accurate - the edge of each polygon to the edge of each park.

I am doing it for a whole city so I cant select the original polygons individually - I will only have lines to all the points on the edges of parks, not to the polygons.

Thanks in advance!

 

0 Kudos
5 Replies
DanPatterson
MVP Esteemed Contributor

sounds like you want a "service area"

Make Service Area Analysis Layer (Network Analyst)—ArcGIS Pro | Documentation

since you seem less interested in "how to get there".

There is a tutorial for network tools, for instance

Visualize the reachable area with Service Area—ArcGIS Pro | Documentation


... sort of retired...
0 Kudos
MelindaMorang
Esri Regular Contributor

Actually, I don't recommend Service Area for this application. Service Areas are great for visual representations of the reachable area. However, they aren't as accurate for this application here.

Origin-Destination (OD) Cost Matrix would be a better choice, particularly since you have already created point layers representing your residential polygons and your parks. Use your residential points as origins and your parks as destinations. Use a travel mode that represents walking distance, and set the cutoff to 200 meters and the number of destinations to find to 1. When you solve the analysis, the output Lines table will contain a row for each origin that found a destination within that cutoff. You will likely have some duplicates since you have multiple points representing both origins and destinations, but you can post-process the table. Ultimately, this will give you a list of residential points that have a park within the designated walk distance limit. You can join this back to your residential polygons or calculate a field or something and then symbolize using that field.

Here is a tutorial about how to do OD Cost Matrix in ArcGIS Pro: https://pro.arcgis.com/en/pro-app/latest/help/analysis/networks/od-cost-matrix-tutorial.htm

0 Kudos
DavidPike
MVP Frequent Contributor

I would probably be more tempted with Dan's service areas, then a simple spatial join to see which polygons intersect any service area. Partly because of the potential number of unplaced network locations in ODCM (although I know settings could overcome this), and the conceptual simplicity of the process and the ability to easily 'sanity check' the final result.

Just my 5 pence!

0 Kudos
MelindaMorang
Esri Regular Contributor

The problem with using Service Areas for an application like this is that the Service Area polygon geometry can change dramatically depending on the settings you use, like polygon trim, whether you use standard or high precision polygons, and even which nearby streets are and aren't reachable. The polygon generation algorithms are often more art than science. It is a visualization tool. If you want a precise answer, use OD Cost Matrix.

0 Kudos
DanPatterson
MVP Esteemed Contributor

plus considerations like

  • do you use centroids or perimeters of the parks
  • parks with entrances with stairs
  • do fences parks behave the same as unfenced parks

I remember a protracted discussion on park accessibility a number of years ago.  There were lots of things the OP hadn't considered and ended up thinking that an "close-ish" solution was probably most appropriate since they had to explain it to the public.  Perhaps adding the computational geometry aspects and how to convert the OD matrix to a geometry representation would also be prudent 


... sort of retired...
0 Kudos