Inquiry of using sDNA in analysis of park accessibility in an island and its accuracy

Question

I have a question  whether sDNA closeness analysis could be used in any region/ city. I am working on an analysis of  the accessibility of parks in Manhattan. You know, Manhattan is a narrow linear island so the core area with high integrity (closeness) sometimes is pretty close to the boundary of the island with low integrity.  It resulted in low accessibility of the parks in the edge of Manhattan when performing calculation with the sDNA model. However, the pedestrian flows are high in parks along boundaries according to observation data. So I am wondering if the methodology is more appropriate to use in continental rather than island sites.
Have you done any research about this limitation?
Attached are the diagrams I drew with sDNA and GIS. They are local closeness at R 800m, accessibility of park at R 800m (join with the closeness of road). The other diagram with red dots shows the times of crowd observed at those parks. You can tell many dots distributed in the parks along the boundaries, which can't be explained by the closeness of the road system.
I am wondering if closeness/integrity is a suitable  method to explain the crowd data.
Update: The closeness I mean in the analysis is actually NQPDA, which is the model you said better to describe accessibility. So the first and second suggestions are not working. Adding scenic weighting/ program weighting may work since their distribution reflects such apparent patterns.
How to add weighting in integral analysis in sDNA?
I think I may use the "learn" and "predict" function in sDNA.  I plan to use three weighting factors (NQPDA, population density nearby and number of programs) to run a predicted model for estimating the possibility of crowds.
How to apply these three  factors in "learn" and "predict" function in sDNA?
I looked at the sDNA manual but it didn't introduce details. https://sdna.cardiff.ac.uk/sdna/wp-content/downloads/documentation/manual/sDNA_manual_v3_0_alpha1/guide_to_individual_tools.html#learn

Sideshow Bob · Answer

Three thoughts spring to mind looking at these maps.

The parks appear to be situated just outside of the the 'cores' you mark on the first map, possibly increasing the radius to 1600 metres may give you closeness that better matches the footfall in parks. If this is not sufficient then,

Yes, there are alternatives to closeness. See my comments here Why NQPD is better than inverse of MAD/SAD in closeness measurement? TLDR - closeness measures quality of access, Network density (in sDNA, Links, Length or Weight within radius) measures quantity of access, and a gravity model (in sDNA, NQPD) combines both. So you may find a better fit with Links or NQPD (though you may have to calibrate the latter)

If all of this fails to give a good fit, I don't know Manhatten but could it be that some parks on the edge are more scenic (sea views, etc)? If some are inherently more attractive than others you could always apply destination weighting to reflect this.

Tiana · Answer

Thank you so much. I agree with you that asking questions publicly can benefit more people. 
The closeness I mean in the analysis is actually NQPDA, which is the model you said better to describe accessibility. So the first and second suggestions are not working. Adding scenic weighting/ program weighting may work since their distribution reflects such apparent patterns. How to add weighting in integral analysis in sDNA? I think I may use the "learn" and "predict" function in sDNA. 
I plan to use three weighting factors (NQPDA, population density nearby and number of programs) to run a predicted model for estimating the possibility of crowds. How to apply these three  factors in "learn" and "predict" function in sDNA? I looked at the sDNA manual but it didn't introduce details. https://sdna.cardiff.ac.uk/sdna/wp-content/downloads/documentation/manual/sDNA_manual_v3_0_alpha1/guide_to_individual_tools.html#learn Would you like to explain more information?
Thank you.

Inquiry of using sDNA in analysis of park accessibility in an island and its accuracy

2 Answers

Add your own answers!

Ask a Question