3D Trees

  1. A UNH researcher wants to use laser scanning to model individual trees.
  2. Understanding how a tree grows could lead to improved forestry management practices.
  3. He is finding significant challenges with the hardware and software that were never designed for this app.

I had lunch yesterday with a researcher from the University of New Hampshire (UNH). He was fortunate enough to acquire a laser scanner with some federal funding a few years back. Unfortunately the unit has not been performing all that well, which is a problem when you are in a remote area of northern British Columbia.

In the forestry research game it is all about studying test plots. These relatively small areas are intensely measured, sampled and monitored in some cases for the past 40 to 50 years. After all trees grow slowly and graduate student labor is cheap. The trick is to compare remote sensing and other innovative research techniques with good old fashioned manually measured facts. This establishes the ground truth.

This researcher is interested in using the scanner to create 3D models of individual trees. Why? Because if you can better understand how a tree actually grows and is affected by competition – surrounding trees, and many other variables, then you will be that much better at planting and managing a forest of trees. Seems simple enough, until you actually try to do it.

First of all there are challenges with the scanner in using it to acquire the precise dimensions of the stem and branches from the ground to the top of the tree. One that is particularly difficult to overcome is wind. The trees are moving, at least in the canopy. Let’s face it scanners were originally designed to work in a process plant where the environmental conditions are controlled. I am sure this kind of forestry application was never envisioned.

But the real hurdle is once again from the software side. Point cloud software does not like modelling trees, mainly because it does not like tapers. Back to the plant roots (I couldn’t resist), point cloud software was designed to model cylinders and pipes with a constant diameter. Now add in branches to the modeling requirement and you can see why today’s software is not going to be very useful for creating an accurate 3D model of a tree.

This is certainly not a mass market application, but it points out the need for greater software modeling flexibility, as is the case with walls having to be planar in CAD systems.

This entry was posted in Education, Research, Software and tagged , . Bookmark the permalink.

3 Responses to 3D Trees

  1. A nice article on volume estimates from trees was published in 2008. Lefsky and McHale used voxels to create sections. Really amazing stuff.

    Volume estimates of trees with complex architecture from terrestrial laser scanning
    Michael Lefsky and Melissa McHale
    Journal of Applied Remote Sensing, Vol. 2, 023521 (14 May 2008)

  2. I think your points about requiring greater modeling flexibility are well founded. I do not deal with moving tree branches, but I do deal with eroded, ancient structures whose walls are rarely right-angled and sometimes eroded mounds of mud–planar is a great luxury. This has caused many issues for me and my company.

Leave a Reply

Your email address will not be published. Required fields are marked *


This site uses Akismet to reduce spam. Learn how your comment data is processed.