Mapping Forest "Fuels"

A Tool for Better Wildfire Management

Todd Erdody, master’s degree student working with the College’s Remote Sensing and Geospatial Analysis Laboratory (RSGAL), is researching how to estimate canopy “fuels” — small-diameter branches and leaves that will burn in a wildfire.  

Erdody came to the College in Autumn 2007 from California, where he was working as a fire monitor and firefighter in Sequoia and Kings Canyon National Parks.  He says, “I spent a good part of the summer thinking about potential research topics as I ignited prescribed fires and fought, monitored, and mapped wildfires.  I realized that I wanted to build on existing research at the UW to explore the use of high-resolution, remote sensing data such as LiDAR (Light Detection and Ranging) and digital imagery to map fuels more accurately and efficiently. “

Working in the fire-prone ecosystem of eastern Washington forests dominated by Ponderosa pine, Erdody, who is funded this year by a grant from the U.S. Environmental Protection Agency, is building regression models for canopy fuel metrics. “My goal,“ he says, “is to produce maps of canopy fuel loading that can be used in similar forest types throughout the Northwest where fuel regimes have been altered as a result of fire suppression and timber management.  Land managers need precise information about fuel loading, especially at the wildland-urban interface, where wildfires can threaten the safety of people and structures and produce smoke and harmful particulate matter.”

Regression models are used to predict the behavior of wildfires based on a combination of fuels, weather, and topographic information; metrics in these models have in the past often been produced at coarse spatial resolutions using vegetation maps and satellite imagery.  Erdody’s research is looking at how well two different high-resolution remote sensing technologies (LiDAR and color-near infrared imagery) can estimate canopy fuel.  LiDAR can provide spatial, thee-dimensional information, while high-resolution imagery can provide spectral information in the visible and near-infrared bands.

About his results to date, Erdody says, “I built regression models using variables derived from LiDAR and imagery to estimate the canopy fuel metrics. I found that variables derived from LiDAR produced more accurate results over imagery-derived variables in estimating canopy fuel metrics.  Because LiDAR’s strength is its ability to measure forest structure at fine resolution, it was the optimal sensor to measure canopy fuels, which are structural in nature. I’ve also found that fusing the LiDAR and imagery datasets produce small but significant increases in accuracy over LiDAR alone.”

RSGAL research, led by Assistant Professor L. Monika Moskal, who is Erdody’s thesis advisor, applies remote sensing and geospatial tools in a trans-disciplinary approach for sustainable management solutions to environmental issues.  She says, “The applications for this kind of research are far-reaching in terms of both geography and planning.”  Adds Erdody, “I envision forest managers using these high-resolution remote sensing technologies to map fuels more effectively and TO create maps for use in wildfire and smoke modeling programs in similar ecosystems worldwide.”

The forest fuels research carried out in the RSGAL lab is part of a larger body of fire ecology research carried out in the College by Emeritus Professor Jim Agee and his graduate students, and by the College’s research scientists in the USDA Forest Service’s Fire and Environmental Research Applications Team (FERA), led by Professor Dave Peterson.  For example, one of FERA’s projects has been the development of a comprehensive software system to build, characterize, and classify fuelbeds to accurately capture the structural complexity and geographical diversity of fuel components across landscapes.

Todd Erdody by burn pile in Sequoia National Park..
Todd Erdody monitoring fire behavior and weather in Kings Canyon National Park.