-
LIDAR –> Bare Earth
Can you get bare earth data fro LIDAR? Canyou process it in software to get a DTM?
-
10 Replies
-
Didn’t know what you meant by ‘bare earth’ so i googled it and surprisingly this was the 2nd link, might be what you are looking for.
Bare Earth Extraction from Airborne LIDAR Data in VRMesh
http://www.youtube.com/watch?v=B0x8iBcMtWA -
yes, it has been done. but its pretty sketchy from what I have seen.
-
I did a number of investigations in connection with HV transmission lines several years ago.
We did get reasonable bare earth results, but the LIDAR lads needed to do an awful lot of tweaking on the settings to separate first, intermediate and last returns (to sort out foliage, underbrush and bare earth).
There is a LOT of office work involved to get really good answers – it’s not the sort of job for anybody who isn’t working with the data 24/7/365.
It’s wise to build in good ground checks – if you are using pre-marking for the LIDAR flights then ground checks can be put in place at the same time.
-
> Can you get bare earth data fro LIDAR? Canyou process it in software to get a DTM?
I haven’t tried to extract bare earth data from the raw files. But I am fortunate enough to work in an area where that work has been done for me. Can download the bare earth files directly.
From there the data is nothing more than x,y,z data points. In some areas the data is pretty good. In other areas, not as good. As always, verify the data with your own ground truthing.
Larry P
-
I was reading on some articles from ESRI about getting DEM from LIDAR. I had the impression that LIDAR has a parameter for the earth surface. Apparently I am wrong. It was radar that could penetrate through the foliage to get earth surface values.
-
Optech had a switch in their software that eliminated trees and vegetation. It relied on the spectral signature of the chlorophyll in the plants. Each ping was cross correlated with the spectral signature of the chlorophyll (I think its called the red edge) and was eliminated if the peak of the cross correlation was above a set threshold. I used it about 9 years ago.
-
> I was reading on some articles from ESRI about getting DEM from LIDAR. I had the impression that LIDAR has a parameter for the earth surface. Apparently I am wrong. It was radar that could penetrate through the foliage to get earth surface values.
The LIDAR with which I am vaguely familiar does penetrate some vegetation. The real trick is to figure out where each return is from. Plus, I know that each type surface has its own rate of reflection. My understanding is that freshly paved parking lots are the worst. The blacktop seems to absorb most of the radar. Unless you hit a freshly painted line, you might not get any data back from those.
I freely admit that I am not an expert nor is my knowledge current. But I would be very skeptical of sales pitches claiming push one button and you have bare earth data. Last I heard it took a tremendous amount of hands on post processing to turn all returns into bare earth. Any claim of push a button and you are done sounds much like what a sales person told one of the local guys in the early days of GPS. “It won’t give you a wrong answer. If you have multipath it knows and will give you no answer at all. If you get an answer it is good to less than 1 inch.“
As you might imagine it didn’t take long to prove that claim totally false.
Larry P
-
I think that’s about right. First return should correlate with the foliage identity (unless it’s a flock of large birds – it has happened!) and last should be the ground, or other solid objects which need to be manually deleted.
Problems occur with tree trunks and branches, especially old fallen trees or ones with multi-trunks which split near ground level. Even if they get edited out they leave gaps in the bare earth surface and make it more difficult to identify “ground level” junk.
The end product “bare earth” will nearly always be higher than the real surface.
The macadam problem isn’t unique to LIDAR. Dark colours often “soak” up the beam and either produce no return, or worse, a delayed return. (Quite a common problem on swimming pool measurement, where a non-contact measurement to the dark lane centreline tiles is nearly always 1 to 2mm longer than reality.)
-
Lidar files (in LAS format) can contain a classification (number). In the example below the classification 2 shows up as yellow which I believe is ground level (I’ve never been to the site). Buildings are classification 1 (red), etc. So one could filter points based on classification and model accordingly.
-
That’s pretty close. The ASPRS standard for LiDAR classification uses numbers to identify classes of points. 0 is “created, never classified”, 1 is “unclassified”, 2 is “ground”, then comes vegetation classes, buildings, etc.
It is possible to get a good bare earth surface from LiDAR data. Most use a software called Terrascan, though there are many others. The software uses algorithms to identify ground points, vegetation, buildings, etc. Basically, it starts with the lowest points and works its way up to find points that statistically fit with their neighbors. It does several iterations to find all of the ground points. There are parameters that are set to adjust the sensitivity. The presence of buildings and terrain characteristics are the primary parameter adjustments.
As long as the LiDAR acquisition was controlled properly and captured at a high enough density, it can be reduced to a very good bare earth surface. That being said, it doesn’t replace field run topo. At least not yet.
Log in to reply.