Occassional thoughts about orienteering
Wednesday, June 13, 2012
More fun with lidar and vegetation
Lidar data includes information on "density" of objects. While I'm not exactly sure what that means, I think it gives information that can help understand how thick a forested area might be. The map below shows an image where I've used different shades of green to identify regions of different object density. The darkest green areas are areas of higher object density. The white areas are areas of the lowest density.
To get this image, I used OL Laser to assign the shades to different object densities and I excluded any points that were within one meter of the ground.
My previous experiments with trying to map runnability were based on object heights. Those experiments were moderately successful. I was able to use the information to help understand runnability. But I still had to do a lot of interpretation in the field and make a lot of changes. I'm hoping that the density information will be more precise, requiring fewer changes.
I'm looking forward to comparing the map above with the actual terrain and see how well the density data maps vegetation.
Back to okansas.blogspot.com. posted by Michael | 8:53 PM
I want to check out if the vertical vegetation density profile will correlate better with actual runability.
I.e. not just the maximum canopy height but the profile/distribution of intermediate returns.
I have a large test area that's intended for the 2015 JWOC that we hope to host in Telemark, Norway.
Parts of the area have recently updated current maps, so I do have something to calibrate against before going out in the forest to field-check.
The maps I posted on June 11 are based on canopy height and they do a good job of showing the mix of open and wooded areas. You can see how well they match the photo. I think the object density information looks like it has the possibility of providing a good look at the runnability of the forested areas. I'm looking forward to checking out the terrain and seeing how well the data showed changes within the forest.
Just like you I have used canopy height up to now, the problem is the huge difference between high canopy forest which suffocates anything below it, leaving a very runnable area, vs another patch with the same top heights, but which also has significant intermediate returns in the most interesting 1-3 meter range.
Due to the top cover, the density of the low/brush returns will be much lower than in an area of pure brush, but the runnability might be very similar or even worse.
I spent a little bit of time in the terrain yesterday. I looked at some of the vegetation differences that show up in the image in the map in the blog post. I spotted some of the differences, but they didn't really reflect differences in runnability so much as differences in how the trees looked. Part of the issue is that the forested terrain in this area is almost all very nice and open. Part of the issue is that I haven't figured out the ideal way of pulling the useful information from the lidar.Post a Comment
I plan to keep experimenting with the data and see if I can find something a little more useful.
I think getting useful runnability information from the data is going to be tricky.
I think that the density information will prove more useful than the height information. But it is going to take some more practice to figure out the best solution.
I will keep trying!