6

I was wondering if it is possible to generate the floor heights of buildings from aerial lidar? We don't have terrestrial lidar or stereo imagery. I don't need the ground DEM level or the roof (DSM) level but the height above the ground where the ground floor is.

The main current requirement is -A number of buildings are on stilts and built to allow for seasonal flooding so when flood models are brought in they shouldn't be classified as high risk since they are designed to allow water to flow under them. Other building are not on stilts but raised off the ground so a max flood height of 845 for example won't affect the building as it's min floor level is 845.5.

Lots of photogrammetry software using Stereo imagery/terrestrial lidar data will allow us to automate the creation of 3D building footprints (with floor height) and build 3D models for visualizing the effect of flooding (an example workflow) BUT what can we do if the only input is aerial 4ppm Lidar and 125mm orthophotos? I looked at coupling it with google streetview but many houses are not covered as they are on private roads away from public view.

See Building extraction with LiDAR data (Improve Process) and Simulate flood to assess Asset inundation footprints for background.

We have arcgis 10.1 and ERDAS Imagine/ER Mapper and I am also testing a number of lidar analysis tools like LP360, Lastools, Lidar Analyst and E3De for this.

enter image description here

Thanks,

2
  • So what I understand from your question is that you want the height of ground floor above the surface using LiDAR data? I don't think you can do this using aerial data. You only have values at points which are visible from above and your ground floor is not visible. I am curious if someone can give some more valuable input. Cheers N
    – Naresh
    Sep 17, 2012 at 10:16
  • Yup, this is what I think as well but thought I would ask the question in case someone did have a solution.
    – GeorgeC
    Sep 17, 2012 at 12:29

1 Answer 1

8

There is no way to get floor heights from a lidar pointcloud. Lidar is captured by bouncing lasers off the groundsurface and measuring the bounced back pulses. Therefore there is no way for the lasers to 'see' through the roof of a building and return a floor height.

However, a solution to this may be to classify your las point cloud into ground and non ground data, which can be performed using lastools, which is an excellent set of tools for manipulating las data. The only issue is the licencing of the tools, which you will need if you are doing commercial work.

Once you have a ground data and building data you can use a free tool called MCC-Lidar to interpolate a ground surface underneath the buildings which will be nodata 'holes'. The MCC-lidar output will be a smooth DEM surface which will be a good approximation of the shape of the ground underneath the buildings. The algorithm was designed to create bare earth DEMs in forested areas so should operate well in your case too.

By using the non-ground las data you can also generate the roof elevations, using the lastools toolbox and perform a calculation to get the height of each building above the ground. If you know the number of floors and can estimate a mean height floor to ceiling, you could subtract these values from the total building height and the value you have left will be a good approximation of the height of the building stilts.

As a side note I would recommend compressing your data to laz format, which offers lossless compression of around 70% and will really help streamline your lidar workflows.

3
  • Thanks. las2dem using the first and last returns gives us a DTM and DEM. lasboundary then creates a rough boundary (depends on how good the original classification is). The issue is that the difference between the DTM and DEM is the total building height but not the height of the floor. Most buildings floors are at least a few inches off the ground and some are on stilts. I don't think there is a solution without stereo imagery or terrestrial lidar/side images but thought I would ask the question in case someone had a creative solution. I like your suggestion of MCC ...
    – GeorgeC
    Sep 17, 2012 at 12:39
  • ... we don't have the number of floors as it's on a large area and I am looking at a semi-automated workflow but knowing the height of the building and the fact that we don't have multistorey houses we may be able to interpret that say a roof <18ft is only slightly raised from the ground and >18ft is probably on stilts. Thanks...the whole point of the question was to generate some options. We do use LAZ when possible the issue is that many other analysis software doesn't accept Martin's Laz format.
    – GeorgeC
    Sep 17, 2012 at 12:44
  • 2
    I can sympathize with the problems of lack of data and big study areas. That sounds like the best solution to keep things automated. It looks like LAZ is becoming more and more ubiquitous, but until its adopted by everyone it can be a pain. Good luck in your analysis and if you get anything worked out I'd love to see your solution.
    – sgrieve
    Sep 17, 2012 at 12:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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