Vegetation Management
Summary Description
Vegetation management (VM) incorporates many activities that provide safe and effective operations of power lines and other energy utilities. VM prevents contact between vegetation and equipment through measures suche as tree pruning, removal of unwanted plants or fragile trees, selective clearing or pest control. The potential contact, which can cause fires and outages, is controlled by maintaining clearances, but must also respect environment protection such as pesticide-free zones, protection of wildlife, control of invasive species and fire risk management. VM is a major task for electricity providers, needs to be scheduled at regular intervals and includes the monitoring of vegetation.
Climate change directly impacts VM. For example, increasing temperatures in Canada are extending the growing season for vegetation, which might increase the need for VM programs and cycle frequency. Species migration (e.g., a decrease in conifer biomass in the North and an increase of broadleaf species) might also occur due to rising temperatures and drought events may weaken vegetation. Other climate events such as wind, lightning, freezing rain and storm can also impact VM. Combined events, where two factors intensify one another, can be particularly impactful, such as strong winds during ice or wet snow accretion.
Many attributions studies show that climate change has made recent wildfires more likely, with Canadian wildfire events in 2023 being two times more likely to occur. Many factors influence forest fires as shown in the figure below: 1) climate conditions that are hot, dry and windy all of which may be more extreme and frequent due to climate change; 2) ignitions including human activity or lightning; 3) and vegetation which as discussed above may be more susceptible to fire due to climate changes like temperature increases. It is important to note that electricity assets can also be the source of ignition. In California, 10% of wildfires are ignited by powerlines due to component failures, downed lines or conductor slaps. Distribution lines are also more at risk than transmission lines when it comes to igniting fires ((Vahedi et al. (2025))).
The section below describes the role of VM in the electricity system, methods and models used linked to forest fires, a detailed discussion, gaps & recommendations.
Role in the Electricity System
VM is crucial for the electricity system as vegetation is one of the most common causes for power outages. A vulnerability assessment completed for Ontario’s distribution system showed that tree contacts are the greatest contributor to customer outages. Ontario organizations have also observed an increase in trees contact with overhead cables with wind speed higher than 70km/h.
Wildfires have major impacts on the power grid as shown in the figure below. Wildfire cause permanent damage to infrastructure which can lead to blackouts. Furthermore, the impacts are not limited to direct power line damage; impacts can also include reduced thermal ratings, line sag, and damage to wooden poles. Wildfire can also impact generation assets: solar panels can also be affected by wildfire smoke reducing their power output; wind turbine blades can be damaged by the ash and soot; and dams can also be damaged by debris found in rivers after forest fire as well as soot.
Methods and Models
The methods presented below are linked to forest fires risk.
The Fire Weather Index (FWI) is used to estimate forest fire risk (potential numeric rating) in Canada. It considers daily observations of temperature, relative humidity, wind speed and 24-hour precipitation. From these observations the field, moisture codes (FMCs) are developed has shown in the figure below. From these the fire behaviour indices: initial spread index and buildup index are estimated to evaluate the FWI.
FMC are ratings of moisture content of the forest floor as well as the dead organic content of pine forests. A higher FMC represents low moisture, hence drier conditions. - Fine Fuel Moisture Code represents the moisture content of the litter, it also indicates the potential for human caused ignition. - Duff Moisture Code represents the moisture just below the litter which is loosely compacted and indicates lightning caused ignition. - Drought Code represents the moisture of the deep contact and densely compacted which indicates the potential depth of burn.
There are four fire behaviour indices: - Initial Spread Index is the expected rate of fire spread based on wind and Fine Fuel Moisture Code, higher values indicate faster spread. - Buildup Index represents the total amount of fuel available for combustion. It relies on Duff moisture Code as well as the drought code and higher values indicates dyer fuel. - Fire Weather Index represented the intensity associated with a fire that spreads in a pine forest. It’s based on the Buildup Index and the Initial Spread Index. It indicates the fire danger in Canada. - Daily Severity Rating represents the difficulty to control fires and it’s based on the FWI.
Burn-P3 is a fire model commonly used in academics and practice to simulate forest fires. It is based on the SyncroSim package and maintained by the Canadian Forest Service. It evaluates the likelihood of fire or burn probability over large landscapes.
Burn-P3 is based on the Prometheus fire growth model. Prometheus is a deterministic model based on the FWI and the Fire Behaviour Prediction which simulates ignition and spread of large numbers of fires. A new version of BurnP3, BurnP3+ offers the option to use three fire growth models; Prometheus but also Cell2Fire and FireStarr. Other fire growth models can also be incorporated in BurnP3+. The newest version also offers features such as scenario comparison, flexibility to use Python or R scripts, support for multi-processing and the ability to dynamically connect to landscape change models such as ST-Sim or LANDIS-II.
Fires and ignitions are simulated in each grid cell. Therefore, Burn-P3 runs multiple model iterations to estimate fire probabilities which allows estimating the burn probability of each cell. These iterations are stochastically sampling ignition locations and other inputs, and then tallying up the results of these Monte Carlo simulations.
Many studies used Burn-P3 with climate change projections to predict future fire activity in Canada. Erni et al. (2024) used Burn-P3 to map wildfire hazards, vulnerability and risk to Canadian communities and found that over 39% of land is considered at high or very high risk of fire. For its estimates, observations from ECCC where combined with provincial meteorological stations as well as gridded observations when stations were missing or the NCEP reanalysis for missing variables such as wind. Gaboriau et al. (2023) made a similar analysis for the boreal forest in Northwest Canada using a dynamic global vegetation model, LPJ-LMfire model to project annual burns and attributes of the boreal trees. Future simulations were run with two GCMs and RCMs as well as with RCPs 4.5 and 8.5. The study showed a decrease of the burn rate in the 21st century due to a decrease of biomass in needle leaf tree species and an increase in broadleaf species. However, some grids in the South East of the studied area showed an increase in fire linked to the increase in needle tree species.
Dawe et al. (2022) used future climate projections from CanESM2, a GCM, in Burn-LP3. Gridded climate projections were downscaled with NRCANmet. Since daily fire weather conditions, which is required as an input to Burn-P3, are inferred at stations, the projections were matched to weather stations with BioSim11. Monthly normals were calculated at 29 stations and for three time periods (2020, 2070, 20120) and fed in LANDIS-II, a forest landscape model, to project vegetation (fuel). Landis-II outputs were directly translated into the Fire Behaviour Prediction System fuel types to incorporate in Burn-P3. The study was done for Saguenay, Quebec and showed an important increase in wildfire risk. The figure below present the methodology of the research.
The Burn-P3 model is often calibrated and validated with historical fire activity such as the Canadian National Fire Dataset. Burn-P3 is also used to evaluate adaptation strategies, such as the best locations to invest in caribou conservation (Dawe et al. (2022), Stockdale et al. (2018)) or mesh on poles in Manitoba (Huang & Bagen (2024)).
Vahedi et al. (2025) also proposes multiple other fire models, e.g., FireSim or FlamMap, in his paper and highlights the benefits and drawbacks of each.
Mulverhill et al. (2025) used machine learning models to project future forest fires with 8 GCMs downscaled with the ClimateNA software (30m spatial resolution) with four SSPs (1-2.6, 2-4.5, 3-7.0 and 5-8.5). Overall, ecozones showed an increase (except for the Hudson Plains under SSP1-2.6) in burned areas for all 4 SSPs by the end of the century ranging from 4% to 60% relative increase depending on the scenario and ecozone. In terms of communities, Sept-Îles was the most at risk with all 4 scenarios showing an increase over 60% of burn probability. However, Mulverhill et al. (2025) kept his biotic fixed in the fire model.