Wildfire Fuel Characterization in Subtropical Ecosystems Using Ground-Based SLAM LiDAR

In Hong Kong’s sub-tropical ecosystems, anthropogenic wildfires burn 5% of the territory’s natural areas every year. With projected climate change including warmer temperatures in the winter dry season, wildfires in Hong Kong may increase in frequency and intensity in the future. Increased wildfire would threaten biodiversity, water resources, reduce carbon storage, and hinder ongoing efforts to restore and rehabilitate forests. To better understand wildfire behavior and project how future climate change could affect wildfire occurrence in Hong Kong it is essential to understand the characteristics of wildfire fuels in local ecosystems. However, no information on wildfire fuels in Hong Kong is currently available.  

Here we aim to characterize wildfire fuels in Hong Kong to develop “fuel models” for the typical Hong Kong vegetation communities of grassland, shrubland, and forest. These fuel models describe wildfire fuels in terms of five derived metrics: fuel load, surface area to volume ratio, fuel bed depth, packing ratio, and bulk density. A fuel model describes how fire will behave in an ecosystem and is an important input for wildfire modeling. 

We developed fuel models for Hong Kong using ground-based Simultaneous Location and Mapping Light Detection and Ranging (SLAM LiDAR). During the winter dry season of 2022-2023, we surveyed grassland, shrubland, and forest plots at Kadoorie Farm and Botanical Gardens, New Territories, Hong Kong with an Emesent Hovermap ST SLAM LiDAR scanner. Fuel models were developed using a voxelization approach by dividing the LiDAR point clouds into uniform voxels, in which the different fuel metrics were estimated. We used field-based measurements to assess the accuracy of the LiDAR-derived wildfire fuel characteristics. Our results demonstrate the potential for SLAM LiDAR to make fast, accurate, and non-destructive characterization of wildfire fuels. The fuel models we developed will be essential for wildfire modeling, land management, and potentially for operational firefighting activities including resource allocation.