Mapping fire risk using LiDAR

Wildfire risk is not evenly distributed across a forest. Two neighbouring stands can experience very different fire behaviour because their trees, understory vegetation, canopy structure and fuels are different. At the same time, fire can move beyond individual stands, meaning that effective prevention requires information at the landscape level. The challenge is obtaining sufficiently detailed information over hundreds or thousands of hectares. A study in the Model Forest of Urbión in central Spain explored how airborne LiDAR can help bridge this gap, transforming measurements of forest structure into spatially explicit maps of potential fire behaviour.

The approach combined airborne LiDAR, field forest inventory and wildfire simulation. Two forest areas of 992.7 and 221.7 ha were analysed using 160 field plots as ground reference data and airborne laser scanning at approximately two pulses per square metre. The LiDAR measurements were used to estimate 13 forest variables, including tree density, basal area, dominant height, timber volume and different components of biomass. These variables were then mapped continuously across the forest landscape and used to characterise the fuels available for a potential wildfire. The complete methodological framework is described in the original study.

Some forest characteristics could be estimated remarkably well from the airborne measurements. The model for dominant tree height reached an R² of 0.906, while timber volume reached 0.871 and several biomass and structural variables also showed strong relationships with the LiDAR information. The results were less precise for the understory: shrub cover produced an R² of only 0.167. This difference is important for fire applications. Laser pulses can describe the upper forest canopy very effectively, but the trees themselves can intercept part of the signal before it reaches vegetation close to the ground. As a result, LiDAR may provide an excellent description of many aspects of forest structure while some of the most difficult fuel components remain below the canopy.

The next step was to translate forest structure into fuel types. When individual standard fuel models were predicted, agreement with field observations was relatively modest, at around 39%. However, when fuels were grouped into broader categories according to their fire behaviour — such as grass-, shrub- and tree-dominated fuels — agreement increased to 64.5%. This illustrates an important point when remote sensing is used operationally: the most detailed classification is not always the most useful one. For landscape fire planning, broader categories that capture meaningful differences in fire behaviour can sometimes provide more robust information than attempting to reproduce every individual fuel class.

These spatial forest and fuel maps were then introduced into the FlamMap fire-behaviour model. Four combinations of fuel moisture and wind speed were evaluated, ranging from normal summer conditions to a dry scenario with winds of 32 km h⁻¹. The model estimated variables including flame length, fire intensity and crown-fire activity across the landscape. In addition, 500 randomly distributed ignition points were simulated to identify areas that were repeatedly reached by modelled fires. The result was not simply a map of vegetation or fuels, but a spatial representation of how the existing forest structure could translate into different levels of potential fire behaviour.

The contrast between scenarios was substantial. Under normal fuel moisture and winds of 16 km h⁻¹, approximately 21.8 ha of the 992.7-ha forest were classified as potentially experiencing active crown fire. Under dry conditions and winds of 32 km h⁻¹, this increased to approximately 170.9 ha. The area with simulated flame lengths above two metres increased from about 61.6 ha to 253.1 ha. Similarly, the area falling within the highest modelled burn-probability category increased from 21.5 to 178.1 ha. These maps demonstrate how the same forest landscape can present very different potential fire behaviour as fuel moisture and wind conditions change.

The practical value of this approach lies in connecting forest resources and fire risk on the same map. Forest managers do not only need to know where intense fire behaviour could occur; they also need to know what is located there. By combining information on timber volume, biomass, canopy structure and fuels with simulated fire behaviour, areas containing valuable forest resources can be evaluated together with their exposure to wildfire. This makes it possible to consider where fuel treatments, thinning or other preventive measures could have the greatest effect, rather than treating fire prevention and timber management as separate planning problems.

The study also highlights an important limitation. A fire-risk map should not be interpreted as a prediction of exactly where the next wildfire will occur. The simulations describe potential fire behaviour under specified forest, fuel and weather conditions, while the burn-probability component was derived from random simulated ignitions rather than a detailed historical ignition model. In addition, the Urbión study area was dominated by relatively well-managed, even-aged forest, and modelling understory vegetation remained comparatively difficult. The framework is therefore best understood as a decision-support tool for spatial forest planning: a way to identify where hazardous fire behaviour and valuable forest resources may coincide and where management interventions could consequently be prioritised.

More broadly, the study demonstrates why wildfire management needs to move beyond the individual forest stand. Fire spreads through landscapes, responding to the spatial arrangement of fuels as well as weather and topography. Remote sensing provides a means of describing that landscape continuously, while fire-behaviour models translate the resulting forest structure into possible fire outcomes. Combining these two sources of information provides a practical framework for integrating wildfire risk into operational forest management.

Further information: Research on forest risks, forest management, biomass production and remote sensing is available through the Biomass Production research group at the University of Eastern Finland. Further publications and activities can also be found at sites.uef.fi/biopro.

Reference

González-Olabarria, J.-R., Rodríguez, F., Fernández-Landa, A., & Mola-Yudego, B. (2012). Mapping fire risk in the Model Forest of Urbión (Spain) based on airborne LiDAR measurements. Forest Ecology and Management, 282, 149–156. https://doi.org/10.1016/j.foreco.2012.06.056

For related research, visit the Biomass Production research group, University of Eastern Finland.

Abstract
The present study sets a methodological framework to combine LiDAR derived data with fire behaviour models in order to assess fire risk at landscape level for forest management and planning. Two forest areas of the Model Forest in Urbión, Soria (Central Spain) were analyzed, covering 992.7 ha and 221.7 ha. The modelling phase was based in 160 field sample plots as ground data, and the LiDAR data had a density of first returns of 2 pulses/m², which were used to construct 13 models for stand variables (e.g. basal area, stem volume, branch biomass). The coefficients of determination ranged from 0.167 for shrub cover, to 0.906 for dominant height. The modelled variables were used for a classification of fuel types compatible with the continuous data. The simulation phase was performed using the spatialized data on FlamMap in order to assess the potential fire behaviour resulting across the whole landscape for four scenarios of moisture and wind conditions. The results showed maps of fire intensity and probability of fire occurrence, based on the simulation of 500 random ignition points, which allowed the analysis of the spatial relation between the initial state and allocation of forest resources and their risk of fire. The methodology proposed, as well as the results of this research are directly applicable for operational forest planning at landscape level.

Highlights
► Fire behaviour and occurrence are estimated with defined scenarios in a forest area. ► We present a methodology that combines forest fire simulators and LiDAR data. ► It provides 13 models for stand level variables and its spatialization. ► A set of rules to classify fuel types compatible with the continuous data is presented. ► The methodology IS applicable for operational planning purposes at landscape level.

Keywords
Airborne LiDAR; Forest inventory; Fire risk assessment; Mediterranean model forest



Find the paper in Science Direct.
ResearchGate link