Moisture thresholds for ignition vary between types of eucalypt forests across an aridity gradient

Abstract

ContextQuantifying spatial and temporal variations in landscape flammability is important for implementing ecologically desirable prescribed burns and gauging the level of fire risk across a landscape. Yet there is a paucity of models that provide adequate spatial detail about landscape flammability for these purposes.ObjectivesOur aim was to quantify spatial and temporal variations in ignitability across a forested landscape. We asked: (1) How do fuel moisture and meteorological variables interact to affect ignitability? (2) Do fuel moisture thresholds for ignition vary across a gradient of forest types? (3) How does the spatial connectivity of ignitable fuel vary over time? (4) How could an ignitability model be used to inform fire management decision-making?MethodsWe conducted field-based ignition tests with flaming firebrands over three fire seasons. Ignitions were attempted across a range of moisture and meteorological conditions at 15 sites in eucalypt forest in south-eastern Australia. Structural equation modelling and generalized linear models were used to quantify relationships between ignitability, aridity, fuel moisture and weather.ResultsThe strongest predictors of ignitability were the moisture content of dead near surface fine fuel and in-forest vapour pressure deficit. Ignition thresholds for both varied across an aridity gradient. Dense forests (i.e., wet and damp eucalypt forests) needed drier fuel and drier in-forest atmospheric conditions to ignite than sparser forests (i.e., shrubby foothill forest).ConclusionOur modelling of ignitability could inform fire planning in south-eastern Australia and the methodology could be applied elsewhere to develop similar models for other regions. Days with consistently high ignitability across the landscape are more conducive to the development of large wildfires whereas days when ignitability is spatially variable are more suitable for prescribed burning.