The Scottish Government is making a substantial set of high-resolution (mostly 50cm) LiDAR data available. The data set does NOT cover the entire Scotland; There is a coverage map (anything that’s highlighted or gray).
The primary data sets were captured in 6 phases, between 2011 and 2022.
The data seems accurate and very well aligned (georeferenced). There are some artifacts, though I might be able to minimize this further by rearranging the sometimes overlapping imagery. It seems to be generally very high quality.
Versions
It comes in two versions:
Digital Surface Model (DSM): Any top surface. Includes trees, buildings, bridges etc.
I find this type of data useful for, among other things:
Tracing natural features such as cliffs, streams, etc. These features often stand out well let you to trace more accurately than e.g. from aerial imagery.
Tracing features obscured by tree canopy in aerial imagery. The LiDAR data is detailed enough that trails, walls, fences etc are often clearly visible. The Terrain Model (DTM) is the choice for this, as it has tree canopy removed.
An additional layer to improve general accuracy, even where features are visible in aerial imagery.
Is there interest in me trying to make the full data set available? That might require a different hosting arrangement, but I’m sure that could be worked out.
There is also similar data for England (which is served by the Environment Agency’s WMS and is already available in iD), and Wales (which doesn’t seem to have a tile server either).
I wonder if it might be worth hosting these layers for Scotland and Wales under OSM UK?
If I’m not mistaken, OSM UK hosts the cadastral parcel layer already, so perhaps a lot of the infrastructure in place already. If someone know knows more about what it would take, please chime in!
The original LiDAR dataset for Scotland is 791 GB.
The example is just a set of pre-rendered z/x/y tiles. Since this is a limited area, and the data rarely changes, pre-rendering the full set is probably feasible. The example goes down to z17, though since the source is 50cm at best there’s probably little additional detail beyond z16, so the latter might be enough.
Preprocessing is with GDAL: hillshade source data tiles with gdaldem, create virtual raster (vrt), then render z/x/y tiles from that using gdal2tiles.
The example tile set is about 10 GB. Let me run a test set of the full area (limited levels) and extrapolate from that. That should give us an idea of the size of the full tile set if we were to go that route.
Alternatively, if we were to render tiles on the fly, the hillshaded source data is just under 300 GB, though some of the data is overlapping so it is possible that creating a single source raster would reduce that size.
I just tried that patch and it works great in JOSM, so hopefully that will get accepted and take care of Wales.
While the Scottish Remote Sensing Portal does have a WMS endpoint as well, it unfortunately doesn’t seem as useful for this specific case: Those datasets are broken out by individual phases, and there is no hillshade layer available. If I’m missing something, please let me know.
I rendered a full set of tiles to level 14, and that came in around 1.1 GB. With each subsequent level roughly 4 times the size of the previous, and level 14 being about 805 MB, that would make z15 about 3.2 GB and z16 about 12.8 GB. So, with two layers (DSM and DTM), I’d estimate somewhere around 35 GB tiles down to z16.
Also, I also looked a bit more at the data, and Phase 5 seems to have a lot of artifacts around the edges. By excluding that (most of that area is covered by other phases anyway) that seems to have removed the artifacts I was seeing previously.
