All UK environment agency lidar data now available

[Redrobes]

I have been playing about with this for a few weeks now and its pretty cool. For a while I have been petitioning people to give me some lidar data to test out my apps and do some processing on it. Some of the people involved have links to the environment agency and I was told that it was in beta that you could download the whole UK wide selection of data. It seems that this news is now out there so I thought I might as well post it here too.

https://environmentagency.blog.gov.u...-up-open-data/


EDIT - That links seems to be dead but my bookmarked link is still active:
http://environment.data.gov.uk/ds/su...ex.jsp#/survey


Its very cool. Personally I would strongly recommend that if you do want some of this data then a) go for the DSM's not the DTM's since the DTMs is *just* the ground level and has no buildings or trees in it. Its the real terrain level only. The DSMs is more raw but does contain houses, trees and pretty much all the useful stuff too. Also, b) go get the lowest res you can get first and check out the data. In most cases all the area covered does not contain lidar info and the data is in swathes of it - I presume its flight paths. It seems to me that they cover the rivers and notable features instead of going for a full block scan.

If the low res data contains the part you want then go for the higher res ones.

I got some of some hills around me which contained an iron age hill fort:


I rewrote some software I was using to be able to virtually emulate what it might look like cut with a NC mill and then cut it:


So I can cut this lidar data now too. Its a pretty expensive operation to do large but it can be done.

As for dealing with this data yourself, well its in standard ArcASCII format so I think many things will read it. Not least I would have expected the free GIS apps and maybe Wilbur amongst others.

I reckon it could be real useful to use for some kinds of real world mapping but it might be a bit overkill to use for fantasy mapping. Some of the data has usage restrictions to it. I think it might be non commercial and need attribution so check with the license pages for it.

But I have had some fun just looking at some locally known places with this.

[foremost]

Wow... this is some neat stuff. Thanks for sharing the link!

I can't say I'm familiar enough with the technical aspects to make use of this myself, but it looks like those who are able to understand that stuff should find this useful.

[Meshon]

Wow... this is some neat stuff. Thanks for sharing the link!

I can't say I'm familiar enough with the technical aspects to make use of this myself, but it looks like those who are able to understand that stuff should find this useful.

Not sure if you saw it, but Francissimo did a tutorial on using this sort of thing a little while ago. I too am unfamiliar with the technology, but the tutorial helps to clear away some of the complete mystery

[Redrobes]

I had not seen that tut and its real good too. For just DEMs generally there are loads of sources. The shuttle radar data is the most common - search for SRTM. If you want mountains and smaller scale stuff thats the ticket but the res is something like 25m, 50m often up to 1km per data sample. Which is useful in some cases but the lidar tends to be real large scale stuff like 1m so its in the realm of trees and houses etc. Thats a different kind of data to look at and it opens up some interesting possibilities. It has been real hard to get this kind of scale until recently - like this year. But this website kind of opens it right up.

I have been playing about with VisualSFM as well which is a photogrammetry program and trying to scan in some terrain using my digi camera but its worked well on some stuff but not so well on others. I am not quite at the point of doing a tut on that but by all means click on my blog where I was messing about with some similar apps a few years ago. Back then you could only get the sparse data points from it but with this new app and an extension by somebody you can get a much denser point cloud and it leads to more accurate models. And you can get the texture map from them too. If anyone has a drone with a video camera on it and is into mapping then you *definitely* need to check out that program. I was chatting with a chap from our national historic preservation trust and he told me that they are using RC drones and a similar (unnamed - possibly the very same) application to capture historic monuments.

If anyone else is doing photogrammetry then do post here and give me the heads up cos I dont always see every thread going through.

[johnvanvliet]

is this gridded level 1 or 2 data
or raw unprocessed point clouds from the lidar

for non gridded data "gdal_grid" can grid it
or
meshlab can skin it for a 3d program

VisualSFM and bundler is fun isn't it

[Redrobes]

Not sure what gridded 1 or 2 means but it is a DEM style square block of equal spaced sample heights. So probably weighted averages out of the point cloud I expect. Personally I immediately convert it into HF2 format and use all my utilities on it. Its raw in as much as the DSM's have not been processed out much to clean them up but its not a point cloud and it does not contain anything like first or last height points in it.

I have also noted that using it today as opposed to a few weeks ago, it has more data sets in it. Also, it now seems that the 2m, 1m, 0.5m and 0.25m sample spacings are not the same data sets at different res but different data sets altogether. So the coverage of them changes. It seems to me that the 1m DSM has the most coverage of them all. Its started noting down the date it was taken and it seems to go up to about 2009. So maybe in time we will get more coverage at 0.25m.

I have been playing with visual SFM and the command line extensions for it PMVS/CMVS from another third party for generating the dense point clouds. Visual SFM calls these from its UI so its easy to use within it. I have been using MeshLab to stitch the resulting dense point clouds together into an object. I havent used bundler - maybe I should try that too.

[johnvanvliet]

bundler is the original code that most Structure from Motion programs are built on
http://www.cs.cornell.edu/~snavely/bundler/

[Redrobes]

It appears that bundler generates the sparce point list just like the SIFT algorithm of VisualSFM does and both bundler and VisualSFM use the same PMVS/CMVS external code to generate the dense point cloud - which is where it gets quite exciting. So its probably not a lot in it. This Visual SFM uses multithreading to speed it up a lot tho and it really hogs that CPU. Theres a GPU version of it too but my GPU wouldnt have the requirements to cope.