Effectiveness of automated fuelsticks for predicting the moisture content of dead fuels in Eucalyptus forests

The fuel moisture content (FMC) of forest litter strongly affects fire ignition and behaviour, and is a key factor in planning when and where to burn. Planned burns are safest and most efficient when FMC lies within a narrow range (9 % - 17 %). Drier fuels (< 9% FMC) can burn too intensely and fires may be difficult to control, while wetter fuels (> 17% FMC) may fail to ignite and/or burn poorly. Planned burning is associated with real risks of escape, and large annual operational costs (e.g. in the order of several $100M's per year in several Australian states), but in the absence of reliable estimates of FMC, fire behaviour predictions are compromised, and many burning opportunities may be missed, leading to the inefficient allocation of resources and compromising strategic burn planning. The aim of this research is to quantitatively evaluate the performance of Campbell Scientific 10-hour fuel moisture sticks for the prediction of the FMC of dead surface and elevated fuels in a wide variety of Eucalyptus forests. Fuelsticks can be installed and monitored remotely, providing continuous real-time information to fire managers on fuel conditions. The performance of the sensors was evaluated in the context of the needs of planned burn managers, specifically, i) daily FMC trend information to assist in the efficient scheduling and resource allocation in the weeks and days prior to the burn, and ii) hourly FMC prediction to build an understanding of how FMC is responding to current conditions. The primary research objectives were to, a) quantify the relationship between Campbell 10-hour fuel moisture stick readings and the in situ gravimetric FMC of surface and elevated fuels collected from 40m x 40m plots, b) use the collected data to develop site-specific corrections to improve fuelstick FMC predictions, and c) evaluate the capacity of the uncorrected and corrected fuel moisture stick FMC readings to improve operational planned burning decision making. The fuelsticks were installed in eight contrasting forested locations across south east Australia and evaluated from December 2014 to June 2015. The results showed that the uncorrected 10-hour fuelstick readings resulted in prediction errors of sufficient magnitude (RMSE of 8-17% FMC) that limits the utility of this method (as applied in this study) for planned burning decision making at the scale of days to weeks, resulting in correct planning decisions only about 50% of the time. However with site-specific corrections and the use of addition microclimate data the fuelsticks enabled correct planning decisions 75% of the time at the scale of days to weeks. Additional fuelstick replication and the use of fuelsticks with shorter response times could further improve the predictive ability of this method. At sub-daily scales the 10-hour fuelsticks were poorly suited predicting current FMC conditions, as the FMC of fuels changed more rapidly than FMC of fuelsticks. The results also indicated that the use of EMC type models to predict the FMC of the surface and profile litter is likely to be unsuccessful due to slow response time of the fuels relative to the rate of change of the atmospheric forcing during the day, and due to the lack of representation of recent precipitation on FMC. Overall, these initial results indicate that locally-calibrated remotely monitored fuelsticks can be used to identify FMC trends which substantially improve planned burning decision making in the weeks and days prior to burning. However the 10-hour fuelsticks tested were unsuited to prediction at sub-daily scales, and more generally, further refinement of the fuelstick and microclimate FMC monitoring methodology is recommended in order to fully exploit the potential of this technology to improve operational decision making.