Ghent population density issue and proposed solution

[carlhiggs]

Hi @global-healthy-liveable-cities/ghscic-study-executive and in particular @gboeing , @shiqin-liu @duelran and @VuokkoH

Our Belgian collaborator Delfien van Dyck contacted us regarding concerns about Ghent's population density estimates, where 0% of the population were estimated to meet the population density threshold (somewhat contradictory of Ester's threshold analysis using IPEN study data based on official sources).

The short of it is, the older GHS R2019a 2015 population estimates were coarse (250m) and wrong for Ghent in particular: it averaged over variation, probably due to relying on coarse administrative data, resulting in artificially flat population distribution --- lacking peaks and troughs which should be there (and resulting in a max density estimate of 3200 persons per sqkm).

The good news is I checked out the new GHS R2022 estimates for 2015 (but which go up to 2030, at 100m resolution) and they capture data which seems more reflective of Ghent collaborator expectations -- ie. some places in Ghent have population density greater than 7000 persons per sqkm.

I proposed that we should consider prioritising an update for Ghent, trialling use of the new data. This could also be an opportunity to make some additional tweaks before larger refactoring -- eg

• remove between-city walkability comparison (as discussed)
• consolidate configuration (should be possible to re-work things so city-configuration is only done once in one place in format both human- and machine-readable)
• use 100m population grid
  • and consider whether we could attribute our indicators to this directly, rather than abstract a 250m hexagonal grid? I think should be do-able, would minimise error and hopefully maximise compatibility with others projects (same format as the GHSL data product

The proposed larger re-factoring involves integrating the policy tools, cleaning up the architecture and making the process overall more user-friendly to get up and running and apply to new areas/indicators.

I have my PhD milestone coming up, and will be travelling/presenting then on leave, so not sure how much that will be progressed before December.

But what do you think of the above?

I have other thoughts about updating our tools, but keen to get broader group feelings on this before acting. I'm waiting on my supervisors to sign off on a questionnaire for this before I ask people to participate in a collaborative software development activity (focused on clarifying expectations/priorities). Hopefully I can act on this soon.

ps. should I write the above up as enhancements on GitHub? (technically the GHS data issue isn't an issue with our code; using their new data would be an enhancement!)