Victorian bushfires 2026: Longwood fire post-event field survey
In early January 2026, Victoria experienced some of the most dangerous fire weather conditions since the 2019 – 2020 Black Summer fires. Prolonged extreme heat, desiccated landscapes, and forecast winds exceeding 100 km/h in parts of western Victoria combined to produce catastrophic fire danger conditions. Temperatures remained well over 40 degrees for consecutive days, prompting total fire bans across Victoria and parts of New South Wales.
The Longwood bushfire commenced on 7th January 2026 under these extreme conditions and was characterised by fast-moving, high-intensity fire behaviour. By 12th January 2026, hundreds of homes and other buildings had been destroyed, more than 136,000 hectares had reportedly burned and the fire had tragically claimed one life. The fire was believed to have been caused by sparks from a trailer travelling along the Hume Highway and wasn’t declared as contained until the 21st January 2026.
The fire footprint extended across open grazing and pastoral land interspersed with bushland, rural settlements, and dispersed agricultural properties (Figure 1). The scale of damage included the destruction of hundreds of residential and agricultural buildings, hundreds of kilometres of fencing and significant livestock losses.
Purpose of the Survey
Between 18th and 20th February 2026, a Risk Frontiers field team visited the Longwood area to document impacts, focusing on the locations reported to be most affected. The team surveyed the townships and surrounding areas of Ruffy, Whiteheads Creek, Gooram, Molesworth, Terip Terip, Alexandra and Yarck, as well as rural areas between them, to:
- Assess building-level damage within the defined bushfire footprint
- Evaluate building exposure representation using Microsoft Bing Maps Building Footprint data
- Examine the relationship between structural damage, proximity to bushland, and ember and grassfire attack
The survey focused exclusively on buildings. Damage to fencing, livestock, and non-building infrastructure was not assessed but was observed. In total, the survey identified 291 bushfire affected properties, of which the majority were destroyed. This report complements an earlier Risk Frontiers briefing on the Victorian bushfires (Risk Frontiers, 2026).
Targeting and Area Selection
Using the spatial boundary of the Longwood fire, areas prioritised to investigate in the survey were determined using multiple intelligence sources, primarily media sources which were supplemented by ‘on the ground’ information from sources including recovery centres, the Country Fire Authority (CFA), road crews and tree clearance teams operating in the area and local residents and farmers spoken to during the survey.
These sources provided situational awareness of heavily impacted towns, key transport corridors, and areas of concentrated structural loss. Road and clearance teams offered real-time insight into road closures, access constraints and roadside damage, while local farmers provided property-level accounts of fire behaviour, ember attack, and personal structural losses.
Observations of destruction and damage
Each structure was categorised according to the Risk Frontiers three-tier damage classification system:
- Destroyed / Major damage – Total or near-total structural loss
- Partially damaged – Minor to moderate structural damage
- Undamaged – No visible structural damage
Buildings were classified as primary or secondary structures. In rural contexts, primary structures typically refer to residential dwellings serving as the principal place of habitation. Secondary structures include sheds, workshops, machinery storage buildings, garages, and other ancillary outbuildings.
Building identification
In total,162 unique G-NAF addresses were inspected allowing for observation and categorisation of 291 individual structures. Individual properties were identified using the Geocoded National Address File (G-NAF). G-NAF enabled consistent linkage between building footprints, field observations, and the fire footprint mapping, allowing clear differentiation between building-level and property-level impacts.
Microsoft Bing Maps Building Footprint data was used to establish a spatial inventory of structures within the fire footprint. The dataset provides building geometry polygons, enabling identification of individual structures rather than simple address points which might not accurately represent the location of affected buildings.
This approach was used in the Longwood survey because many rural properties contain multiple structures. Primary and secondary structures are often clustered together, and some structures on a property may be impacted when others may not. Building footprints enabled building-level damage attribution.
Damage classification and apportioning
Of the 291 buildings identified and categorised, 103 structures were classified as primary structures (residences, community centre / restaurant) and 188 were classified as secondary structures (sheds and outbuildings). Only buildings safely observable from public roads were assessed and the survey team made no attempt at entry onto private property. Georeferenced photographic evidence was collected at each location
In terms of damage inventory:
- 271 structures were categorised as Destroyed / Major damage
- 5 structures were categorised as Partially damaged, and
- 15 structures were categorised as Undamaged
Although buildings sometimes appeared to have different levels of damage from the outside, the team observed numerous examples of damage that compromised structural integrity. In many cases, the external shell, often made of corrugated iron or tin was warped, buckled, or deformed by heat, making the structure unusable even if parts of the exterior remained standing. Examples can be seen in Figure 2, where all the buildings were considered to have been destroyed.
Once structural integrity is lost, the building is unsafe to occupy and no longer functional. As a result, in practical terms the survey categories “Destroyed” and “Major Damage” often represent the same outcome: a structure that cannot be safely used.
Comparison with previous bushfire research
A key focus of the field survey is to compare the observed outcomes with findings from previous Risk Frontiers post-fire research. Figure 3 provides a comparison of the proportion of destroyed and partially damaged structures from the Longwood bushfire against previous risk Frontiers field surveys on the South Coast and Rappville in 2020, and from Tathra in 2018.
An overwhelming finding from the Longwood field survey is that 98% of properties impacted by the fire were ultimately determined to have been destroyed. The survey results show that once buildings became involved, they were almost always destroyed or damaged to the extent that they were categorised as destroyed and were no longer structurally sound.
These findings are consistent with previous Risk Frontiers research from the South Coast and Rappville (2020) and Tathra (2018), which similarly found that once a building catches fire – regardless of construction material, it is highly likely to be totally destroyed. The survey team saw numerous examples of structures made from corrugated iron, brick, and timber that had been consumed in the blaze. Research from Tathra (2018) indicate that 68% of fire-affected premises were ultimately destroyed, while data collected from the South Coast and Rappville (2020) provides an even stronger indication of this trend, with 92% and 100% respectively of observed buildings destroyed.
The survey team frequently observed that when one building ignited, multiple nearby structures on the same property were also destroyed (Figure 4). In many cases, vehicles located close to buildings were also burned.
Ember Attack, Grass Fire Spread and Distance to Bushland
Previous Risk Frontiers research has shown that proximity to bushland is a key determinant of building vulnerability during bushfires. Observations from the Longwood fire reinforce our previous research demonstrating that building losses can extend well beyond the immediate bushland interface, reflecting the combined effects of extreme fire weather, long-range ember attack and grassfire spread. The Longwood curve in Figure 5 (orange dashed line) shows the relationship between the proportion of destroyed buildings and their distance from adjacent bushland, compared with other major bushfires distance from adjacent bushland, compared with other major bushfires.
The distribution highlights how the spatial pattern of losses in the Longwood fire differed from more forest-dominated fire events such as Kinglake and Maryville in 2009.
To interpret Figure 5 we have provided some points of interest. Approximately 30% of destroyed buildings at Longwood were within 1m of bushland. Approximately 40% of destroyed buildings were located within 10 m from bushland, indicating that many losses occurred close to vegetation where exposure to radiant heat and ember attack is greatest. 50% of destroyed buildings were located approximately 20 m from bushland, meaning that half of all destroyed buildings were within about 20 m of vegetation. For 80% of destroyed buildings the distance increases to approximately 70–80 m, showing that most losses still occurred relatively close to bushland, but that the influence of embers likely played less of a role in lieu of the surrounding grassland which would have been transporting the fire front. Overall, the pattern indicates that building losses in the Longwood fire were concentrated near bushland, with more than the half of the surveyed destroyed buildings located within 20–30 m of bushland.
However, destruction extended out to around 80 m or more, consistent with the effects of ember attack and grassfire spread rather than direct flame contact alone. Field observations also showed many destroyed buildings located in open paddocks, often hundreds of metres from continuous bushland or forest edges. In some cases, structures were completely destroyed despite treed vegetation being several hundred metres away (Figure 6).
Compared with fires such as Kinglake or Marysville, observations from the Longwood fire suggest that building losses were spread further away from dense forest, reflecting the observed mix of grassland and bushland terrain. Field observations support that grassfires made an important contribution to the damage pattern. Dry grass and other materials across paddocks allowed fires to spread quickly once ignited. Embers often appear to have started fires in grass near buildings and secondary items such as round hay bales, which then spread across open areas and contributed to impacting isolated structures. Strong winds also carried embers long distances, increasing the chance of fires starting well away from the main fire front.
Building Inventory Discrepancies
Microsoft Bing Maps Building Footprint data was used to create a spatial inventory of structures. However, field observations showed that the dataset was not always accurate which is an understandable shortcoming as the data have not been updated since 2018. The survey team frequently found buildings on the ground that were not represented in digital exposure datasets.
On several occasions, the survey team spoke with local Council representatives conducting secondary damage assessments who reported similar issues, noting that several buildings across multiple properties did not appear in official Council records or were not of the permitted type.
These discrepancies likely arise from incremental or unauthorised development, farm outbuildings, sheds, and other informal structures that are not captured in official development records or national building footprint datasets.
Incomplete or outdated building inventories can lead to an underestimation of structural exposure and total damage costs after hazard events. Buildings that are missing from baseline datasets may also be excluded from post-disaster loss assessments.
In rural and remote areas, where secondary dwellings and informal structures are more common, these gaps may also lead to an underestimation of resident populations. This can affect emergency planning, evacuation modelling, and recovery efforts, and in some cases create uncertainty in fatality or missing-person assessments and potentially leading to underestimation of losses and misallocation of mitigation or recovery resources.