Department of Defense SERDP Grant - The focus for our sensitivity research has been the process of developing 3D models of samplings from LiDAR data that was acquired as part of sampling burn experiments conducted at the Fire Lab in the summer of 2021. The problem is non-trivial because while LiDAR data is rich and detailed, transferring a full LiDAR data set directly into a fire behavior model like Fire Dynamics Simulator (FDS) would be computationally prohibitive; there is just too much detail.

 To remedy the problem we’ve developed algorithms that reduce the resolution of the LiDAR to whatever scale is desired. The following figure shows that process, which ranges from 1 cm cubed on the left to 8 cm cubed on the right.

Burns are then simulated in FDS for each of the resolutions:

The burns having the greatest parity with observations are used to guide the process of selecting the resolution the LiDAR data is re-sampled at. The primary observation we concern ourselves with is the mass of the burning sapling as a function of time. A typical result is seen in the following figure.

As a result of this research we are able to state that:

1. The Fire Dynamics Simulator model is capable of faithfully reproducing the mass lost during combustion in laboratory experiments.​
2. Spatial distribution and resolution of fuels represented in models have the most significant impact on model results.​
3. Moisture content of the fuel can also have a significant impact on results and if poorly constrained can be more significant that geometric considerations.
4. LiDAR data can be aggregated into cells as large as 6 cm cubed. Beyond that, the representation is too coarse and it is no longer possible to match observations with computer simulation.