Modelling multi-layer fine fuel loads in temperate eucalypt forests using airborne LiDAR and inventory data

Wildfires are increasing in intensity and frequency due to climate change and land-use changes, posing critical threats to ecosystems, economies, and human safety. Fine fuels (<6 mm, such as leaves and twigs) are known key drivers of wildfire ignition and spread, particularly in temperate forests...

Full description

Saved in:
Bibliographic Details
Main Authors: Trung H. Nguyen, Simon Jones, Karin J Reinke, Mariela Soto-Berelov
Format: Article
Language:English
Published: Elsevier 2025-06-01
Series:Science of Remote Sensing
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S266601722500032X
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849329206469591040
author Trung H. Nguyen
Simon Jones
Karin J Reinke
Mariela Soto-Berelov
author_facet Trung H. Nguyen
Simon Jones
Karin J Reinke
Mariela Soto-Berelov
author_sort Trung H. Nguyen
collection DOAJ
description Wildfires are increasing in intensity and frequency due to climate change and land-use changes, posing critical threats to ecosystems, economies, and human safety. Fine fuels (<6 mm, such as leaves and twigs) are known key drivers of wildfire ignition and spread, particularly in temperate forests where high flammability increases wildfire hazard. Accurately quantifying fine fuel loads (FFL) across vertical forest layers is essential for understanding and predicting wildfire behaviour, yet past studies using Airborne Laser Scanning (ALS) have been limited to canopy fuels, overlooking surface and understorey layers that play a key role in wildfire propagation. This study addresses this gap by developing an ALS-based modelling approach to estimate FFL across four vertical layers: canopy, elevated (or ladder), near-surface, and surface. The study was conducted in eucalypt-dominated forests in Victoria, southeastern Australia. We stratified ALS point clouds into distinct layers (overstorey, intermediate, shrub, and herb), computed layer-specific structural metrics, and trained Random Forest models to predict multi-layer FFL. The models performed well, with the highest accuracy for canopy FFL (R2 = 0.74, relative RMSE = 49.42 %) and moderate accuracy for elevated, near-surface, and surface FFL (R2 = 0.42–0.56, relative RMSE = 59.77–77.57 %). Model interpretation revealed that integrating ALS metrics from multiple forest layers maximised accuracy and highlighted the complex role of vertical forest structure in predicting FFL. Prediction maps captured horizontal and vertical FFL variations across landscapes, reflecting differences in forest structure. Furthermore, pre-fire FFL, especially in surface and canopy layers, showed statistically significant associations with wildfire-induced forest loss. This study advances multi-layer FFL estimation using ALS data, offering a more comprehensive fuel information for wildfire hazard assessment and forest management. Future research should explore the scalability of this method by integrating satellite-derived data to extend FFL mapping at broader spatial scales.
format Article
id doaj-art-153fcd2f2e4943dca72e4ad4e99623f4
institution Kabale University
issn 2666-0172
language English
publishDate 2025-06-01
publisher Elsevier
record_format Article
series Science of Remote Sensing
spelling doaj-art-153fcd2f2e4943dca72e4ad4e99623f42025-08-20T03:47:20ZengElsevierScience of Remote Sensing2666-01722025-06-011110022610.1016/j.srs.2025.100226Modelling multi-layer fine fuel loads in temperate eucalypt forests using airborne LiDAR and inventory dataTrung H. Nguyen0Simon Jones1Karin J Reinke2Mariela Soto-Berelov3Mathematical and Geospatial Sciences, School of Science, STEM College, RMIT University, Melbourne, VIC, Australia; Sustainable Technology and Solution Laboratory (STAS.Lab), Thai Nguyen University of Agriculture and Forestry (TUAF), Thai Nguyen, Viet Nam; Corresponding author. Mathematical and Geospatial Sciences, School of Science, STEM College, RMIT University, Melbourne, VIC, AustraliaMathematical and Geospatial Sciences, School of Science, STEM College, RMIT University, Melbourne, VIC, AustraliaMathematical and Geospatial Sciences, School of Science, STEM College, RMIT University, Melbourne, VIC, AustraliaMathematical and Geospatial Sciences, School of Science, STEM College, RMIT University, Melbourne, VIC, AustraliaWildfires are increasing in intensity and frequency due to climate change and land-use changes, posing critical threats to ecosystems, economies, and human safety. Fine fuels (<6 mm, such as leaves and twigs) are known key drivers of wildfire ignition and spread, particularly in temperate forests where high flammability increases wildfire hazard. Accurately quantifying fine fuel loads (FFL) across vertical forest layers is essential for understanding and predicting wildfire behaviour, yet past studies using Airborne Laser Scanning (ALS) have been limited to canopy fuels, overlooking surface and understorey layers that play a key role in wildfire propagation. This study addresses this gap by developing an ALS-based modelling approach to estimate FFL across four vertical layers: canopy, elevated (or ladder), near-surface, and surface. The study was conducted in eucalypt-dominated forests in Victoria, southeastern Australia. We stratified ALS point clouds into distinct layers (overstorey, intermediate, shrub, and herb), computed layer-specific structural metrics, and trained Random Forest models to predict multi-layer FFL. The models performed well, with the highest accuracy for canopy FFL (R2 = 0.74, relative RMSE = 49.42 %) and moderate accuracy for elevated, near-surface, and surface FFL (R2 = 0.42–0.56, relative RMSE = 59.77–77.57 %). Model interpretation revealed that integrating ALS metrics from multiple forest layers maximised accuracy and highlighted the complex role of vertical forest structure in predicting FFL. Prediction maps captured horizontal and vertical FFL variations across landscapes, reflecting differences in forest structure. Furthermore, pre-fire FFL, especially in surface and canopy layers, showed statistically significant associations with wildfire-induced forest loss. This study advances multi-layer FFL estimation using ALS data, offering a more comprehensive fuel information for wildfire hazard assessment and forest management. Future research should explore the scalability of this method by integrating satellite-derived data to extend FFL mapping at broader spatial scales.http://www.sciencedirect.com/science/article/pii/S266601722500032XFine fuel loadAirborne LiDARVertical stratificationRandom forestWildfire
spellingShingle Trung H. Nguyen
Simon Jones
Karin J Reinke
Mariela Soto-Berelov
Modelling multi-layer fine fuel loads in temperate eucalypt forests using airborne LiDAR and inventory data
Science of Remote Sensing
Fine fuel load
Airborne LiDAR
Vertical stratification
Random forest
Wildfire
title Modelling multi-layer fine fuel loads in temperate eucalypt forests using airborne LiDAR and inventory data
title_full Modelling multi-layer fine fuel loads in temperate eucalypt forests using airborne LiDAR and inventory data
title_fullStr Modelling multi-layer fine fuel loads in temperate eucalypt forests using airborne LiDAR and inventory data
title_full_unstemmed Modelling multi-layer fine fuel loads in temperate eucalypt forests using airborne LiDAR and inventory data
title_short Modelling multi-layer fine fuel loads in temperate eucalypt forests using airborne LiDAR and inventory data
title_sort modelling multi layer fine fuel loads in temperate eucalypt forests using airborne lidar and inventory data
topic Fine fuel load
Airborne LiDAR
Vertical stratification
Random forest
Wildfire
url http://www.sciencedirect.com/science/article/pii/S266601722500032X
work_keys_str_mv AT trunghnguyen modellingmultilayerfinefuelloadsintemperateeucalyptforestsusingairbornelidarandinventorydata
AT simonjones modellingmultilayerfinefuelloadsintemperateeucalyptforestsusingairbornelidarandinventorydata
AT karinjreinke modellingmultilayerfinefuelloadsintemperateeucalyptforestsusingairbornelidarandinventorydata
AT marielasotoberelov modellingmultilayerfinefuelloadsintemperateeucalyptforestsusingairbornelidarandinventorydata