Multispectral LiDAR-Based Estimation of Surface Fuel Load in a Dense Coniferous Forest

Stefanidou, Alexandra; Gitas, Ioannis Z.; Korhonen, Lauri; Georgopoulos, Nikos; Stavrakoudis, Dimitris G.

doi:10.3390/rs12203333

Cited by 18 publications

(12 citation statements)

References 93 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…CHM followed by canopy and vegetation cover are the most important variables in estimating 1-h fuel load. This is most likely related to the fact that if the degree of coverage and the height of the canopy increase, then the biomass in terms of leaves, twigs, and small branches (e.g., fine and coarse dead fuel loads) increases as well [64][65][66]. These results are in line with those of Chen et al [67], who applied multiple regression analysis using airborne and terrestrial LIDAR for surface fuel load estimations in the open eucalyptus forests of Australia.…”

Section: Discussionsupporting

confidence: 80%

Machine Learning Techniques for Fine Dead Fuel Load Estimation Using Multi-Source Remote Sensing Data

et al. 2021

View full text Add to dashboard Cite

Fine dead fuel load is one of the most significant components of wildfires without which ignition would fail. Several studies have previously investigated 1-h fuel load using standard fuel parameters or site-specific fuel parameters estimated ad hoc for the landscape. On the one hand, these methods have a large margin of error, while on the other their production times and costs are high. In response to this gap, a set of models was developed combining multi-source remote sensing data, field data and machine learning techniques to quantitatively estimate fine dead fuel load and understand its determining factors. Therefore, the objectives of the study were to: (1) estimate 1-h fuel loads using remote sensing predictors and machine learning techniques; (2) evaluate the performance of each machine learning technique compared to traditional linear regression models; (3) assess the importance of each remote sensing predictor; and (4) map the 1-h fuel load in a pilot area of the Apulia region (southern Italy). In pursuit of the above, fine dead fuel load estimation was performed by the integration of field inventory data (251 plots), Synthetic Aperture Radar (SAR, Sentinel-1), optical (Sentinel-2), and Light Detection and Ranging (LIDAR) data applying three different algorithms: Multiple Linear regression (MLR), Random Forest (RF), and Support Vector Machine (SVM). Model performances were evaluated using Root Mean Squared Error (RMSE), Mean Squared Error (MSE), the coefficient of determination (R2) and Pearson’s correlation coefficient (r). The results showed that RF (RMSE: 0.09; MSE: 0.01; r: 0.71; R2: 0.50) had more predictive power compared to the other models, while SVM (RMSE: 0.10; MSE: 0.01; r: 0.63; R2: 0.39) and MLR (RMSE: 0.11; MSE: 0.01; r: 0.63; R2: 0.40) showed similar performances. LIDAR variables (Canopy Height Model and Canopy cover) were more important in fuel estimation than optical and radar variables. In fact, the results highlighted a positive relationship between 1-h fuel load and the presence of the tree component. Conversely, the geomorphological variables appeared to have lower predictive power. Overall, the 1-h fuel load map developed by the RF model can be a valuable tool to support decision making and can be used in regional wildfire risk management.

show abstract

Section: Discussionsupporting

confidence: 80%

Machine Learning Techniques for Fine Dead Fuel Load Estimation Using Multi-Source Remote Sensing Data

et al. 2021

View full text Add to dashboard Cite

show abstract

“…It would also be beneficial for future studies to explore the use of ALS to predict surface fuel loads (e.g. Stefanidou et al 2020).…”

Section: Discussionmentioning

confidence: 99%

Predicting black spruce fuel characteristics with Airborne Laser Scanning (ALS)

Cameron

Schroeder

Beverly

2021

Int. J. Wildland Fire

View full text Add to dashboard Cite

Wildfire decision support systems combine fuel maps with other fire environment variables to predict fire behaviour and guide management actions. Until recently, financial and technological constraints have limited provincial fuel maps to relatively coarse spatial resolutions. Airborne Laser Scanning (ALS), a remote sensing technology that uses LiDAR (Light Detection and Ranging), is becoming an increasingly affordable and pragmatic tool for mapping fuels across localised and broad areas. Few studies have used ALS in boreal forest regions to describe structural attributes such as fuel load at a fine resolution (i.e. ,100 m 2 cell resolution). We used ALS to predict five forest attributes relevant to fire behaviour in black spruce (Picea mariana) stands in Alberta, Canada: canopy bulk density, canopy fuel load, stem density, canopy height and canopy base height. Least absolute shrinkage and selection operator (lasso) regression models indicated statistically significant relationships between ALS data and the forest metrics of interest (R 2 $0.81 for all metrics except canopy base height which had a R 2 value of 0.63). Performance of the regression models was acceptable and consistent with prior studies when applied to test datasets; however, regression models presented in this study mapped stand attributes at a much finer resolution (40 m 2 ).

show abstract

“…In published articles, much research demonstrates that the performance in the estimation of surface fuel loads using lidar data and optical images varies dramatically. From the perspective of PRMSE, the integrated approach of diverse remote sensing data was around 20-38% for a dense coniferous forest located in central Greece [33] and 37-98% for a bark beetle-affected forest in eastern Grand County in north-central Colorado [57]. A better accuracy of PRMSE around 5-47% was achieved for an upland oak-dominated forest in Kentucky using small footprint full-waveform lidar data [58].…”

Section: An Examination Of the Appropriateness Of The Cokriging-based Surface Fuel Mapping Methodsmentioning

confidence: 99%

“…According to the remote sensing-based IPCC method, a canopy fuel map can be derived using an ALS canopy height model by segmenting every single tree, using, for example, mathematical morphology-based watershed segmentation [26][27][28][29], Multilevel Morphological Active Contour (MMAC) [30] or Multilevel Slicing And Coding (MSAC) techniques [31]. Moreover, the latest development of lidar sensing enables precise inventories of surface fuel and canopy fuel using mobile terrestrial lidar instruments [32][33][34][35]. To overcome the high cost of high-density point cloud ALS data, alternate methods for surface fuel loading (SFL) estimation can be based on mathematical/empirical models using inventory data such as vegetation/species maps and related environmental factors [18,36], satellite fullwaveform lidar data [37,38], and photon lidar data [39].…”

Section: Introductionmentioning

confidence: 99%

Generating a Baseline Map of Surface Fuel Loading Using Stratified Random Sampling Inventory Data through Cokriging and Multiple Linear Regression Methods

Lin

Huang

et al. 2021

Remote Sensing

View full text Add to dashboard Cite

Surface fuel loading is a key factor in controlling wildfires and planning sustainable forest management. Spatially explicit maps of surface fuel loading can highlight the risks of a forest fire. Geospatial information is critical in enabling careful use of deliberate fire setting and also helps to minimize the possibility of heat conduction over forest lands. In contrast to lidar sensing and/or optical sensing based methods, an approach of integrating in-situ fuel inventory data, geospatial interpolation techniques, and multiple linear regression methods provides an alternative approach to surface fuel load estimation and mapping over mountainous forests. Using a stratified random sampling based inventory and cokriging analysis, surface fuel loading data of 120 plots distributed over four kinds of fuel types were collected in order to develop a total surface fuel loading model (lntSFL-BioTopo model) and a fine surface fuel model (lnfSFL-BioTopo model) for generating tSFL and fSFL maps. Results showed that the combination of topographic parameters such as slope, aspect, and their cross products and the fuel types such as pine stand, non-pine conifer stand, broadleaf stand, and conifer–broadleaf mixed stand was able to appropriately describe the changes in surface fuel loads over a forest with diverse terrain morphology. Based on a cross-validation method, the estimation of tSFL and fSFL of the study site had an RMSE of 3.476 tons/ha and 3.384 tons/ha, respectively. In contrast to the average loading of all inventory plots, the estimation for tSFL and fSFL had a relative error of 38% (PRMSE). The reciprocal of estimation bias of both SFL-BioTopo models tended to be an exponential growth function of the amount of surface fuel load, indicating that the estimation accuracy of the proposed method is likely to be improved with further study. In the regression modeling, a natural logarithm transformation of the surface fuel loading prevented the outcome of negative estimates and thus improved the estimation. Based on the results, this paper defined a minimum sampling unit (MSU) as the area for collecting surface fuels for interpolation using a cokriging model. Allocating the MSUs at the boundary and center of a plot improved surface fuel load prediction and mapping.

show abstract

Multispectral LiDAR-Based Estimation of Surface Fuel Load in a Dense Coniferous Forest

Cited by 18 publications

References 93 publications

Machine Learning Techniques for Fine Dead Fuel Load Estimation Using Multi-Source Remote Sensing Data

Machine Learning Techniques for Fine Dead Fuel Load Estimation Using Multi-Source Remote Sensing Data

Predicting black spruce fuel characteristics with Airborne Laser Scanning (ALS)

Generating a Baseline Map of Surface Fuel Loading Using Stratified Random Sampling Inventory Data through Cokriging and Multiple Linear Regression Methods

Contact Info

Product

Resources

About