Sunday, 3 November 2013

Releasing data really works! Part IV

[Update: this was summarised into a story map here posted also on this blog here]

Two years then one year ago I QC'd UK Ordnance Survey data for East Anglia, and sent the polyline spike and kickback errors to the Agency, who posted the corrections this year. They noted the errors I reported fell below their own QC criteria, but they invited me to retest their updated dataset. This issue is topical as  posted on GISlounge on the same topic proposing other tools.

1SpatialCloud work on a fremium basis, but since I tested their predecessor Socium, they kindly let me re-test both old and new datasets in their new web tools. I first tested for geometric, network and polygon checks on May 2010 Ordnance Survey Boundary Line polygons for Parish boundaries:
click to enlarge
 If results were very good in 2010, with 25 errors out of 1777 polygons...
Click to enlarge
...they were even better in the 2013 update at only 1 spike out of 1779 polygons!
click to enlarge
And how to better illustrate this clean-up than on the map posted on Arcgis Online (updated)?
click to enlarge
Again, making public data available does help spur on data improvements, and online data validation helps identify errors quickly and efficiently. This makes it easier for the public to communicate, and for data custodians to high-grade their holdings.