Date of Award


Document Type


Degree Name

Master of Science (MS)



First Advisor

Kathryn A. Yurkonis


Fire is a critical physical and chemical process required to sustain many grassland ecosystems. In North America, observations of grassland fire behavior in warm-season, southern grasslands are commonly used in fire behavior modeling efforts across the Great Plains. However, grasslands of the northern Great Plains contain a greater component of cool-season vegetation that may generate different fire behavior. To further our understanding of prescribed fire behavior in North Dakota grasslands, we quantified fuel, weather, and fire behavior characteristics associated with 27 prescribed fires conducted across three sites in North Dakota. We sampled 27 points on each fire arranged into a Sierpinski triangle sampling scheme with three fractally nested spatial scales. Field station are climatologically and vegetatively different sites, yet fuel and weather characteristics associated with the fires were similar. Ultimately, fire behavior was similar between the stations having burned under similar fuel bed properties and weather conditions. Fire behavior averaged 227.26 ± 94.74 °C (maximum temperature), 0.4 ± 0.3 m (flame height), 4.47 ± 3.82 m/min (0.07 ± 0.064 m/s; rate of spread). Maximum temperature and flame height were best explained by fuel moisture, relative humidity, and quantity of the last rainfall event. Rate of spread was best explained by dew point, wind speed, and quantity of last rainfall. However, increased fuel moisture and relative humidity suppressed fire behavior. To quantify spatial heterogeneity, we assessed fuel bed properties (fuel load, soil and fuel moisture) prior to ignition and the resulting fire behavior (maximum temperature, flame height, and rate of fire spread) on 26 prescribed fires. We used a hierarchical Restricted Maximum Likelihood (REML) variance component analysis with the full 27-point dataset to assess how each sample scale (100 m, 10 m, 1 m) contributed to the variance in the fuel and fire behavior responses. Fuel loads (LAI) were most variable at higher scales (100 m, 10 m) but most similar at the 1 m scale. Fuel moisture contrasts fuel load in that it was the most variable at 1 m but similar at 100 m. Soil moisture variance was not dependent on the sample scale. Assessing relationships between fuel explanatory and fire response variables, we found similar effects of heterogeneity in fuel load and fuel moisture on maximum temperature and flame height. Maximum temperature and flame height exhibit the most variation when fuel load is most consistent and when fuel moisture variance is high. Rate of spread has a limited dataset and did not relate to the variation in fuel load and fuel moisture. Understanding the spatial variability within the fuel bed and its contribution to fire behavior will aid fire practitioners will better guide future planning efforts and provide a greater understanding of ecological fire effects.