I have some recent data that is at the Lower Super Output Area (LSOA) level for Greater Manchester. I wrongly assumed that the data was for the latest LSOA boundaries, i.e. for the 2011 boundaries.

Acting on this assumption, I downloaded the 2011 boundaries from ONS and used mapshaper to simplify it to 1% of discardable vertices. I imported the data in to CartoDB and got this nice looking map:


Once I realised that my data did not fit this LSOA version, I tried to do the same with the 2001 ONS data. This was less successful. Even retaining 2.5% of discardable vertices still leaves a lot of holes in the map, as you can see here:


I used the following command to do the simplification and conversion before using a simple ruby script to select just the Greater Manchester features:

mapshaper -i LSOA_2001_EW_BFC_V2.shp encoding=utf8 -simplify 1% keep-shapes -o uk-2001-lsoa-boundaries-1pc.geojson format=geojson

Other notes:

  • The encoding parameter in the command above is a guess. Both sets of data required an explicit encoding and utf8 worked fine on 2011 data.
  • I think I used the same command for the 2011 data (there are not a lot of parameters to play with) but I may be wrong.
  • The results that I get with the full, clipped 2001 data look like the results I got when I unknowingly used the 20m generalised 2011 data. In fact, the first thing I did when I saw the results was to check that I had not used 20m generalised data.
  • Which is the version that you need exactly? I can help you to try to simplify it with PostGIS inside CartoDB, by using postgis.net/docs/ST_SimplifyPreserveTopology.html – iriberri May 28 '15 at 15:13
  • @iriberri Thanks for the offer. The ONS responded by another channel and I'm just going to try their suggestion first, although it seems counter-intuitive given my experiences so far. – crantok May 29 '15 at 9:32

I contacted the ONS geography team directly and they suggested using the generalised boundary data rather than the full-clipped boundary data. This worked.

I've double checked my previous work and can confirm that is the exact opposite of what worked for the 2011 boundaries. So, here is the summary:

  • If you want to simplify the 2001 lsoa boundary data, use the generalised-clipped dataset.
  • If you want to simplify the 2011 lsoa boundary data, use the full-clipped dataset.

Got it?

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.