Unusual range of values in LiDAR DEM

Question

I downloaded the data in LAS file format and converted the LAS file to "multipoint" and then used "point to raster" tool to make it as raster, for missing data I used the raster calculator. I got the DEM but the problem is in the range of values (i.e, the lowest elevation and the highest elevation) which I am getting in my DEM, which I know that it is way more than what it should be, but not sure what are the factors which affect the range of the elevation values. What can be the possible reasons for this and how can I correct it (by looking for the solution I found somewhere that I should use some filter but I have no idea how to use a filter in this case)?

I want to keep the most of the elevation values as it, just want to remove the few points which are very different from the majority of elevation points and is responsible for making the range of values much bigger.

Answer 1

It could be many things. For example, is the .las data composed by only ground returns? If it is not, then the point to raster tool would be using non ground points into the DEM, causing the distortions. In this case, you would need to classify ground points first. See:

• Determining bare earth DEM from unclassified LAS file?

Visualizing the DEM would be sufficient to verify if it is the case. In ArcGIS you can use ArcScene, but there are other options.

In addition to Aaron's advice, be aware that ArcGIS has already a better workflow for what you are trying to achieve; create a las dataset and convert it directly to raster (without needing LAS to multipoint). See:

• Converting LiDAR data to raster (DEM/DSM) for ArcGIS input?

Another issue could be the presence of outliers in the las point cloud, for example, non ground points assigned as ground (class 2). Remove them before generating the DEM. See:

• Finding outliers with wrong Z values in LIDAR data or
• Deleting anomalous points in *.lasd using ArcGIS for Desktop?

If you want to remove outliers/noise directly from the DEM, there are many approaches you could try depending on the situation. For examples, see:

• Filtering noise out from bare-earth DEM with terrace fields? or
• How to de-noise a DSM.

Last, always try to get to know better the data you are working with before processing: check the metadata for coherency and try to apply some quality control to see if everything is ok.