We exploited the measurement capacity of a terrestrial laser scanner to precisely characterize shrub fuel matrices in a laboratory setting, to abstract fuel elements for fire behavior modeling, and to identify strengths and limitations of TLS for these purposes. Simultaneously, we produced statistical distributions of combustion parameters for individual fuel elements by burning hundreds of individual chamise and sagebrush samples. Finally, we imaged and burned whole-shrub fuel beds in a wind-tunnel and used measurements from these experiments to further develop and validate a semi-empirical shrub model. The project was based on the principle that developing dynamic fuel models for shrub lands requires tiered experiments starting with burning of individual leaves in the laboratory, followed by combustion of whole shrubs in a wind tunnel, and culminating in burning of field plots of representative fuels. This project focused on the latter two elements.

Our experiments showed that TLS-derived 3-D shrub models can be used to detect changes in volume and biomass following combustion across a range of losses on a per shrub basis. Near constant variance across the range of mass losses showed that the TLS can detect changes in the fuel bed effectively even when only parts of the fuel bed are consumed. Although the model did not numerically depict where the loss was occurring in the shrub fuel bed due to data gaps in shrub interiors, ocular assessment of change showed that the TLS correctly identifies the locations of mass loss. The primary shortcoming of shrub geometries obtained from TLS is the shadowing that occurs interior of shrub hulls. We explored statistical methods for filling in these voids and developed L-systems fractal models for chamise and sagebrush from geometric measurements of actual shrubs to predict the locations of fuel elements.

LiDAR scan data were used as evidence to represent the local fuel density, which were combined with L-systems fractal theory during the fuel placement development in the semi-empirical shrub combustion model. The LiDAR scan data guided the L-systems approach by pointing the shrub branches to the highest possible density position, which resulted in a better combustion modeling outcome when compared to the shrub constructed by L-systems approach only. This result emphasizes the need to accurately describe fuel placement in shrubs when modeling combustion behavior. Calculated flame heights above the shrub, fraction of shrub burned, burn time, and flame propagation speed and flame path were all compared with experimental results. The modeling results suggested that the combustion behavior predicted was too intense compared to the wind tunnel experiments. One of the reasons that the simulations predicted higher combustion intensity was that individual fuel element experiments were conducted at a higher gas temperature than the local combustion zone temperature in the wind tunnel shrub experiments. Methods to correct for the effect of local temperature are currently underway.