Estimating Fuel Moisture in Grasslands Using UAV-Mounted Infrared and Visible Light Sensors

Area of Science:

• Ecology
• Remote Sensing
• Forestry

Background:

• Wildfire behavior prediction relies on empirical models, but physics-based models offer broader applicability.
• Physics-based fire models require detailed inputs like spatially explicit fuel characteristics, especially fuel moisture.
• Traditional fuel moisture measurement methods are time-consuming and spatially limited.

Purpose of the Study:

• To assess the effectiveness of using visible and infrared spectral reflectance for estimating grass fuel moisture.
• To develop and compare predictive models for vegetation moisture using spectral data.
• To evaluate the potential of remote sensing for rapid and flexible moisture assessment.

Main Methods:

• Collected 120 field samples (1 m²) of grassland in western Washington.
• Acquired overhead imagery in six wavelengths for the same sample areas.
• Generated predictive models using vegetation indices and principal component analysis (PCA) of spectral data.

Main Results:

• The best predictive model was a linear model incorporating PCA components and biomass, yielding an r² of 0.45.
• The model performed better when pooling dominant grass species than when analyzing each species individually.
• Modest predictive power was observed, particularly given the limited moisture range in the study.

Conclusions:

• Spectral reflectance shows potential for estimating grass moisture content, a critical wildfire fuel characteristic.
• Further research across the full fire season could enhance model effectiveness for drone-based applications.
• This remote sensing approach offers a faster, more flexible alternative to traditional destructive sampling methods.