Contours from LiDAR data not matching SHP file contours

[rafaelaustin]

I recently received contours from a county agency in SHP format that looked good and seemed to represent the actual existing topography. But when I tried using the LAS LiDAR files that they had for the whole county, they did not match the contours I got when I imported the SHP files; they were very "choppy".

I'm using C3D 2014 and I imported the SHP files through "Surfaces > Create Surface from GIS Data".  The LiDAR was imported through Prospector by creating a Point Cloud from the LAS file, then Adding Points from that Point Cloud to a new surface. I even tried Editing the Surface and usign the Simplify Surface command, but that didn't help.

When I search for this topic I see many references to poorly processed contour lines that produce profiles that are jagged. However, I see the contour lines that were produced (through ArcGIS?) and exported as SHP files that looks ok. Is there a better way to process this data in C3D? Is it better to import the LAS into ArcGIS and export SHP files? I was recommended a third party tool like Global Mapper, but I don't understand (or convinced) that it does anything different than C3D. 

I have all this LiDAR data that I feel is no good.  😞 

Thanks!

[Neilw_05]

I am still using 2013 so I don't know if there have been any improvements in the point cloud engine, but I posted a thread about issues with surfaces built from point clouds here: http://forums.autodesk.com/t5/autocad-civil-3d-general/clip-or-filter-point-cloud/m-p/3144490/highli...

Basically the problem is surfaces built from point clouds may be highly decimated (not all the points are used), so you may be seeing a poor correlation between your shape contours vs. the surface contours due to the surface not utilizing all the data in the point cloud.

[KirkNoonan]

Depending on the terrain, using just the point cloud to create your surface will likely result in jagged contours. Properly processed LiDAR derived surfaces will also utilize breaklines that are generated from specific return types, aerial digitizing or ground topo. If someone has already given you the shape files of the post-processed LiDAR you should consider using those. You can create a surface from the shape file by selecting "Create Surface From GIS Data" under surfaces on the home tab.

[rafaelaustin]

Thanks Neilw (and others), that confirms some of what I'm looking for.  So the short-term workaround is to zoom in on a small area and piece together several surfaces? I used to do that with Google Earth surfaces; it seemed to produce a more detailed surface if I tiled several together. Maybe for the same reasons? I will try this process this afternoon.

The story about the data:  The County contracted a firm to fly the LiDAR scan. The raw data went to aconsultant who generated an ESRII Geodatabase (GDB) file and gave that and the LAS files to the County. The County then gave me a SHP file for a specific location. Then I found out that my project expanded and I needed more data. I then had them copy the LAS files to an external drive and that's what I was hoping to use for the area not covered by the initial SHP file.

The second part of my quest is: How did the consultant process the files that eventually yeilded a good SHP file?  Is there something in the way ArcGIS processes the LAS data that is better? Does it actually look at ALL 1.5M points per LAS file and generate contours based on that? Will Global Mapper also look at all 1.5M points and produce contours based on that?

Our AutoCAD vendor has graciously offered to generate contours for me, but I can't ask that several times a month. Or go back to the County to ask for a section at a time. I'd really like to be able to process a usable surface from the LiDAR data that I have and may collect in the future.

Thoughts?

[Neilw_05]

As I noted in the thread I mentioned, I could not get a full resolution surface from the point cloud, even when zooming to tile it in smaller pieces. Instead I added the points file directly to the surface definition which gives 100% fidelity. You'll also need to verify whether any breaklines were used to build the surface that generated the contours. If so you will need to obtain them and add them to the surface. You will want to use a clip boundary to limit the extents of the data that is added to the surface. Otherwise you will likely overwhelm C3D trying to process all of it.

[rafaelaustin]

Thanks for clarifiying that. Unfortunately, I don't have any other data (break lines) other than the LAS files. I'd have to generate my own break lines and sect points. Definitely cheaper than sending the survey crew out, but...  ugh! I've got about 9 miles of roadway to topo.

When you say "3) Add point files to surface definition", what steps are you doing to create that from an LAS file?

Thanks for your help!

[Neilw_05]

In my case I had an XYZ ascii points file. I don't know if you can add LAS files to a surface definition, sorry. If not I would look for tools to convert LAS to XYZ or some other workaround, maybe Infraworks if you have it.

[rafaelaustin]

I was afraid of that.  Unfortunately, the section of the county that my project is located does not have XYZ files.

I will look into the Global Mapper program that my vendor uses or try to look into a solution using ArcGIS.  I'll post back in a couple of days.

[AllenJessup]

I concur with what Neil has said. That's basically the way I found was the best way to create the Surface. I was lucky enough that I already had a set of breaklines to use.

One way I haven't tried with Lidar yet is that if you have IDSP you will have Recap. theoretically you can import the Lidar into Recap and export an RCP file and use it to create a surface in Civil 3D. I haven't tried that yet.

Allen

[rafaelaustin]

Thanks, I'll look into Recap too.

[autoMick]

I recently went through a few point cloud dramas, I think some of it might get written up in the knowledge base soon by the Autodesk support people. I had a number of issues with my LAS files, but the relevant part of it for this discussion was that I had to set the POINTCLOUDPOINTMAX variable to 10 million (default is 1.5M - so beware if you have a slow computer) and then zoom in on specific areas and only add the points from the LAS file to those extents. Break it up into a number of tiles if you have a long area to do. I found then that the point cloud generated very acceptable surface contours.

BTW, I hadn't even opened RECAP prior to these latest problems - anyone working with point clouds should have a look at that program.

Cheers

- Mick

[Neilw_05]

You can export the point cloud to an XYZ text file if you want to get around the point cloud problems. There are also several free utilities that will convert LAS to XYZ.

[smay8399]

I use this process to create smooth contours:

1. Use Point cloud to create surface.  (Call it EG-LIDAR)

2. Use point  "create points" to create COGO points manually or on a grid every 10 to 50 feet.

3.  Add extra points manually in areas of interest.

4.  Create new surface from all these COGO points (Call it EG-LIDAR-SAMPLED)

Check the contours to make sure they match the lidar raw contours. 

[rafaelaustin]

Today's update...  I asked our GIS department to look at the LAS data. After a couple of attempts, the contours they generated looked like the ones I generated in C3D.  So my next step is to get the GDB file from the County and export SHP files in house.  I'm a little disappointed with that route because we just paid $400 to acquire some other LiDAR data and now I'm not so sure it's useful without going through 2 or 3 extra steps.

The green is the SHP generated contours, the white is the LAS from the GIS guys.  Big difference.  I was able to get a better correlation with C3D, but it was still not very close.

[rafaelaustin]

As a comparison, this is what C3D generated:

[smay8399]

The green is more what I would expect from LIDAR. Maybe there is something wrong with your LAS files

[Neilw_05]

It is apparent to me that you missing breakline data. This is most obvious when looking at the road crossing the site in the east-west direction.

No doubt there are other issues but it is obvious there needs to be a better correlation.

[smay8399]

Just as a comparison, this is what I produce out of C3D from LIDAR.  No sampling on this surface at all.

The data is from the Puget Sound LIDAR Consortium

[rafaelaustin]

If there is breakline data associated with the LAS files, are they seperate files? What extension would they be?

I'm surprised that the consultant that prepared the GDB didn't include the break lines with the LAS data. Unless I just don't recognize it, or the guy at the county.

[rafaelaustin]

Yes SMAY, that's what I was hoping to be able to extract with C3D.  Still hopeful I'll find a way to use the data we have.