How I can get Digital Height Model (only object heights, DSM – DTM = DHM)?
Many applications can do this, but they convert it to GRID or TIN format. I was wondering if it was possible to keep the data as point clouds.
[GIS] Normalizing point cloud data
demlidarnormalizepoint
Related Solutions
The current option to import las files in ArcGIS is the LAS dataset data type. According to ESRI:
The LAS dataset provides fast access to large volumes of lidar and surface data without the need for data conversion.
If the surface (raster) you are trying to create is a bare-earth DEM refer to the following posts:
One option is the LAS dataset to Raster tool, but only when it is possible to filter1 the data by its ground returns. On the contrary, using the Cell Assignment Type argument equal Average or Minimum on unclassified data could (probably, 'would') result in bad DEMs. Mostly, the result will depend on the point cloud density, the terrain cover complexity, and the quality required in the output. It is important to (double) validate the DEM if going through this way.
1 Pay attention that to filter the LiDAR files in a LAS dataset it is necessary to generate a LAS dataset layer first (through the Make LAS dataset layer tool), or LAS dataset to Raster will consider all types of returns (just filtering using the LAS dataset properties for visualization won't work when processing data).
- Selecting LiDAR return point type in LiDAR dataset with ArcPy?
- Specifying LAS dataset filters in ArcGIS ModelBuilder?
Another option for creating DEMs in ArcGis is to use the combination of Las to Multipoint tool for filtering ground points, and then, Point to Raster (it is probably less efficient than the first method, though).
Other workflows for creating DEMs from LiDAR data in ArcGIS are described in the post below. It also has insights about aspects which should be taken into consideration, when generating DEMs.
For working on DEMs starting from unclassified .las data I would suggest using more robust algorithms not yet available within the ArcGIS platform. For example:
On the other hand, if the surface (raster) is a Digital Surface Model - DSM refer to the following posts:
Input of .las/.laz files and conversion to .asc DSM (LASzip and Fusion).
Input of .las files and conversion to .asc DSM (LAStools).
Another option to import .las files in ArcGIS would be to convert them into ASCII (.txt, for example). Unless there is a specific reason for using plain text, previous options are better. The main differences working with .las or ascii files are approached in this post:
So, an alternative would be to use the command line LDA2ASCII from Fusion.
The LDA2ASCII syntax is:
LDA2ASCII InputFile OutputFile format
Suppose the following:
- Fusion is stored under
c:\Fusion. - the las file name is
project.las. project.lasis stored in:c:\Lidar.- the new txt file will be saved at:
c:\LiDAR. - its name will be:
project.txt
Hint: use an text editor to write Fusion's program commands (e.g. Notepad++). Save it with .bat extension. Then, write:
Retrieve the following information: X, Y and Elevation.
c:\Fusion\LDA2ASCII c:\Lidar\project.las c:\Lidar\project.txt 0
Or if the .las files contain other information such as: Pulse, Return, X, Y, Elevation, Nadir and Intensity:
c:\Fusion\LDA2ASCII c:\Lidar\project.las c:\Lidar\project.txt 1
Save the .bat file, and then execute it (shortcut = F5)
Open project.txt in ArcGIS.
- DEM = digital elevation model
- DTM = digital terrain (topography?) model
- DSM = digital surface model
Sort of the same thing - distinctions can be made if considering just terrain or also objects on that terrain. See What is the difference between DEM, DSM and DTM?
As for Grid Format, as mkennedy says, my guess would be the native raster format Esri uses. You would have to speak with your client to confirm. To create one you can export your DEM tiff. When specifying an output location, just point to a folder. Whenever exporting a raster in Arc, if you do not specify a file extension the result is a GRID file. You may also be able to choose GRID in a 'Format' dropdown depending on what dialog you're looking at.
Note that this format isn't as self-contained as a tif would be. If you create a new, blank folder and select that as the output location, you'll get two subfolders - one with the filename you specified and an INFO folder. There may also be .xml and .ovr files outside those two folders. It is important that all of this be included when transmitted - particularly that INFO folder. It's a required part of the raster.
Best Answer
An option to normalize* LiDAR point clouds (and keep it as a point cloud) is Fusion. One will need the command line
ClipDatatogether with the switches:dtm:file, which is the bare-earth model (DTM), andheight.ClipData description says:
When the switches
dtm:fileandheightare added, each return in the lidar cloud will have the elevation subtracted by the elevation in the corresponding pixel in the DTM. The output file will be of type.las, where the z coordinates will be heights.It will also work with .laz files (compressed .las) in Fusion's version 3.4 and above, but needs LAStools installation too.
The ClipData syntax to perform such analysis would be the following:
ClipDatais the command line itself./heightand/dtm:fileare the switches necessary to normalize the cloud.InputSpecifieris the original .las file,SampleFileis the output file (.las file).For example: let's assume our lidar file has name
gross.lasand it is stored under the directory C:/LiDAR. The DTM is stored in the same directory with namebare_earth.dtm1. The bounding box UTM coordinates ofgross.lasare:7100000 7200000 730000 740000. The normalized cloud will be namednormalized.lasand it will be stored in the same directory as the other files. Fusion is installed under the directoryC:. Type this:1. One needs to have the bare-earth model with the .dtm Fusion's format to run ClipData. Refer to this thread to learn how to generate a DTM starting from a non classified lidar cloud. Then, use Fusion's
ASCII2DTMtool to convert the DTM from .asc format to .dtm extension.*Scheme about the normalizing process.