Influence of Fuel Heterogeneity and a Novel Fuel Rendering Technique on Fire Spread Predictions
Dr. Eric Rowell | Tall Timbers Research Station
The overarching objective of this project is to quantify the sensitivity of fire spread predictions to the level of spatial detail found in novel 3D fine-scale surface fuels measurements. This is important because numerical-based coupled fire-atmosphere models produce highly resolved predictions of fire behavior, but have been limited by oversimplified fuels data and fuel models. Using high resolution 3D fuelbeds both from a new technique (i.e. virtual fuelbed renderings) and from terrestrial laser scanning, the project aims to determine the effect of fuel 3D resolution and quantification methods on outputs of coupled fire-atmosphere models. The specific objectives are to:
- Determine the 3D resolution at which surface fuel characteristics, namely bulk density, surface area-to-volume ratio, and fuel height, will affect fire behavior simulation outputs. The project will quantify the differences across scales and identify thresholds or gradients of little to significant differences.
- Determine how the method of quantifying these fuel characteristics, such as aggregation approaches (e.g., recalculation of bulk density at different scales, averaging across scales), will affect fire behavior simulation outputs.
- Determine the measurable differences of how the method of 3D fuels characterization (e.g. terrestrial laser scanning [TLS] vs. virtual fuelbed renderings) affects fire simulations due to the type of input characterization (voxels vs. objects). This will be tested across resolutions (O1) and aggregation methods (O2) of input fuel characteristics.
- Use information theory to determine how the scale, method, and technique affect the representation of fire behavior phenomena (e.g. adjacent fuel interactions in 3D space, fireline interactions) within the model.
The technical approach is centered on the formulation of fine-scale fuelbeds in 3D space. The researchers will use existing terrestrial laser scanning datasets from two southeastern U.S. pinelands, with an approach to creating virtual fuelbed renderings and associated field data to produce inputs for use in a coupled fire-atmosphere model. As all of the objectives are directly associated with fuel representation within the model, assessing each objective will be done through an iterative process. This will provide for identification and quantification of single element (e.g. changing resolution) and interacting element (e.g. changing resolution x 3D fuelbed technique) effects on fire behavior model outputs. Shannon’s Entropy will measure entropy of the 3D input data as compared to the spatial model outputs of fire behavior.
This project seeks to find the optimal balance in level of detail in 3D representation of fuels to yield the most information from coupled fire-atmosphere models as they move from research tools into the operational arena ( RC-2643). Using an optimal 3D representation of fuels will maximize the potential accuracy of fire behavior predictions, improving estimates of emissions and heat release for use in smoke transport models. This will improve the understanding of complex fire interactions affecting organization of convective structures which reduces risk for prescribed fire management of Department of Defense lands. Examining the potential risk of utilizing this novel 3D rendering technique for creating virtual fuelbeds is critical for advancing fuels characterization, particularly because the approach may either revolutionize the way fuels data is represented or even collected, or may prove unnecessary based on the results of this project. If successful, this approach can be tested and utilized in other surface fire regimes across various Department of Defense installations with minimal field data collection.