Wednesday, August 4, 2010
Exhibit Hall A, David L Lawrence Convention Center
Background/Question/Methods Recurring fires in tropical landscapes dominated by nonnative invasive grasses such as guinea grass (Urochloa maxima) pose a significant threat to surrounding ecosystems and developed areas. To manage fire in tropical ecosystems, better predictions of fire behavior and spread are needed. Predicting fire behavior is commonly done with fire models (e.g., BehavePlus), but their accurate and reliable use by land managers requires a better understanding of fine-scale temporal changes in the parameters that most drive fires – fuels and weather. Live and dead fuel moisture, both critical determinants of fire behavior (e.g., probability of ignition and spread rates), are typically needed to parameterize fire models. However, many current fire models do not correspond well to real, on-the-ground fuel moisture values due to an inability to accurately estimate changes in fuel moisture over fine temporal scales without intensive sampling. Environmental parameters such as temperature, relative humidity, precipitation, and soil moisture are easily and commonly measured with weather stations on a continuous basis, and potentially provide a means to estimate fuel moisture. Therefore, the potential exists to greatly improve the accuracy of fire models and provide an important tool to land managers if correlations between weather variables and live and dead fuel moisture can be identified. The objective of this study was to quantify relationships between environmental variables and fuel moisture in tropical grasslands dominated by the nonnative grass Urochloa maxima in three sites on the Waianae Coast and North Shore of the Island of Oahu, Hawaii. We hypothesized that because precipitation, soil moisture, and temperature largely determine plant water availability, they would be strong predictors of fuel moisture. To address this hypothesis, we measured soil and fuel moistures and a suite of weather parameters over 9 months to determine whether these variables can be used to predict live and dead fuel moistures accurately over fine temporal scales.
Results/Conclusions
A positive linear relationship existed between soil moisture and live (r2=0.72; P<0.01), standing dead (r2=0.84; P<0.01), and litter (r2=0.70; P<0.01) fuel moisture. Additionally, 24 hour-cumulative rainfall was positively and linearly correlated with live fuel moisture (r2=0.64, P<0.01). These results suggest that easily obtainable weather parameters can be used to predict live and dead fuel moisture to more accurately parameterize fire models. More accurate fuel moisture inputs will greatly improve the capacity for land managers to make real-time predictions of fire ignition and spread rates in guinea grass ecosystems in Hawaii and throughout the tropics.
