I am rather new at GIS and I would like to have some guidance on what follows.

I have some DEMs derived from LiDAR data. These DEMs contain building and other human features (e.g., roads).
I was wondering what the (negative?) consequence would be if contour lines are extracted from these DEMs. In other words, how the presence of human features will affect the calculated z.values (i.e., elevations.).

The DEMs I got have the horizontal resolution of 1 meter. The program I am working with is ArcGIS, and I am a bit familiar with QGis as well.


According to the ASPRS specification buildings are class 6. To make a DEM from LiDAR data you must only use class 2 (Ground) and 8 (Model key Point). ArcGis gives the option to create a terrain from a subset of points. Level 1 classification is ground/non-ground, level 2 introduces low/medium/high vegetation and so on.. so it depends on what level your LiDAR is classified to as to what you can derive from it.

If you have included all the points regardless of class then you do not have a DEM, you have a DSM (Digital Surface Model).

If your LiDAR data is unclassified then you have a real problem deriving a good surface from it; LiDAR typically contains a very small amount of noise which is assigned to a class (1 = unclassified or 3 = Low Veg/noise) and needs to be filtered out from surfaces otherwise you will end up with sharp spires or shafts depending if the noise is greatly below or above the surface.

If your rasters already contain buildings then you've got a problem, try to get the original LiDAR data and make the DEM yourself or the contours will be rubbish! A method that I have used to get good DEM from LiDAR is using LAS to multipoint, create a terrain and then use terrain to raster. From 10.1 ESRI introduced the LAS Dataset which should be able to rasterize direct from LiDAR to DEM - I have not explored this yet so I can't comment on the effectiveness of it.

There are a set of tools called las tools that has some unusual licensing, and in these tools is LAS2DEM that I have heard does a good job at creating DEM raster from LiDAR files, again having not explored this avenue I can't comment any further.

  • Hello! Thank toy for your informative reply. And sorry for my delayed reply. I am interested in building a DEM (i.e., bare earth). I am using ArcGis 10.1. I would like to know if, in order to get the DEM from Lidar dataset, besides selecting class 2 (ground), I have to choose "any return" or "last return" instead?. Thank you. Best – NewAtGis Jun 9 '14 at 9:29
  • No, first and last return do not matter for this case. LiDAR pulses can have multiple returns. In the case of bare fields/roads then only one return is likely. If there are trees then many returns are possible as the beam passes through foliage. If you don't have class then last return may be a suitable proxy, but that must be a last resort. – Michael Stimson Jun 9 '14 at 21:54
  • Thanks a lot. Very informative. And, apologies for these basic questions of mine, but I am rather new a GIS and Lidar. Cheers – NewAtGis Jun 10 '14 at 6:27
  • You're welcome. No need to apologize, I struggled with the same concepts when I started with LiDAR about 5 years ago. – Michael Stimson Jun 10 '14 at 21:21

Derivatives of LiDAR are only as good as the point cloud itself. If buildings are left in the point cloud when first derivatives (DEMs) are created, the elevation values will reflect the heights of buildings, as well as ground. This is useful if you are determining the elevation of building rooftops. Second derivatives (contour lines) created from a DEM with rooftops, again, will reflect the values of the rooftops.

Look at this presentation. It explains and visualizes the difference between bare earth and first return DEMs, as well as DEMs used to visualize rooftops.

If it is rooftop height, bridge height, etc that you are after, then these surface models are necessary. However, if you require bare earth values (for a flood analysis, line of sight, etc), then a bare earth DEM may be required.

All this said, if you want elevation contours of buildings, then what you have will work. Be warned that contours of vertical surfaces will be dense if you generate small contour intervals (such as 1 or 2 feet). With ArcMap you can use the Contour tool to generate contours from your DEMs. With 1 meter cells, you are able to create accurate contours down to 1 meter. You can do smaller intervals, but the interpolated values will not be as accurate.

  • Hello! Thank you for your prompt reply and for the information provided. No, I am not interested in elevation of rooftops. So, if I am not mistakenly interpreting your reply, I should go to Lidar data and filter the buildings out, right? In this way, contour lines should not be affected by buildings height, am I right? On this issue, I guess I will post a question on how to filter the lidar data and creating a bare earth DEM (using ArcGIS). Thank you. And if you have any other suggestion, feel free to reply. Cheers. – NewAtGis Jun 6 '14 at 18:12

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.