Published works

Australia’s native Eucalypt forests are the most fire-prone in the world due to high rates of fuel accumulation, high flammability of fuel, and seasonally hot and dry weather conditions. Projected changes in the frequency and intensity of extreme climate and weather could increase the occurrence of ‘mega-fires’, extreme fire events with catastrophic impacts on people and the environment. Current methods for fire risk mitigation and prediction such as fire danger rating systems, fire behaviour models, and hazard reduction treatments require an accurate description of forest fuel. However, fire management authorities share a common challenge to efficiently and accurately quantify forest fuel properties (e.g. fuel load and fuel structure) at a landscape scale. A landscape includes the physical elements of geo-physically defined landforms, such as forests, grasslands, and lakes. This thesis investigates the application of the Light Detection and Ranging (LiDAR) technique in quantifying forest fuel properties, including fuel structural characteristics and litter-bed fuel load at a landscape scale.

Currently, fire fighters and land managers still rely on empirical knowledge to visually assess forest fuel characteristics of distinct fuel layers. The visual assessment method provides a subjective description of fuel properties that can lead to unreliable fire behaviour prediction and hazard estimation. This study developed a novel method to classify understorey fuel layers in order to quantify fuel structural characteristics more accurately and efficiently by integrating terrestrial LiDAR data and Geographic Information Systems (GIS). The GIS-based analysis and processing procedures allow more objective descriptions of fuel covers and depths for individual fuel layers. The more accurate forest fuel structural information derived from terrestrial LiDAR data can be used to prescribe fire hazard-reduction burns, predict fire behaviour potentials, monitor fuel growth, and conserve forest habitats and ecosystems in multilayered Eucalypt forests.

Traditionally, litter-bed fuel load is directly measured through destructive sampling, sorting, and immediate weighing after oven drying for 24 hours at 105 °C. This direct measurement of fuel load on a landscape scale requires extensive field sampling, post laboratory work and statistical analysis, which is labour intensive and time consuming. This study found new relationships among forest litter-bed fuel load, surface fuel depth, fire history and environmental factors through multiple regressions with airborne and terrestrial LiDAR data. The fuel load models established in this study indicate that litter-bed depth and fire history are the primary predictors in estimating litter-bed fuel load, while canopy density and terrain features are secondary predictors.

Current fuel models are constrained to estimate spatial variations in fuel load within homogeneous vegetation that previously experienced the same fire events. This study developed a predictive model through multiple regression to estimate the spatial distribution of litter-bed fuel load in multilayered eucalypt forests with various fire histories and forest fuel types. This model uses forest structural indices and terrain features derived from airborne LiDAR data as predictors, which can be applied when data on forest fuel types and previous fire disturbances are absent. It can be used to map the litter-bed fuel load distribution at a landscape scale to support regional wildland fire management and planning.

This study indicates that LiDAR allows a more efficient and accurate description of fuel structural characteristics and estimation of litter-bed fuel load. The results from this study can assist fire hazard assessment, fuel reduction treatment, and fire behaviour prediction, and therefore may reduce the impact to communities and environment.