Bushfire Egress Index: where fire can cut the way out
Shahrooz Shahparvari · RMIT University, Department of Supply Chain and Logistics
Bushfire Egress Index (BEI) · built from the open benchmark library AUFireBench 1.0.0 · 34 recorded fire seasons, 1990-91 to 2023-24
Across 1,714 rural Australian localities, the index shows where fire can cut the supply of people and goods, who is least able to leave, and how much an optimised road protection budget helps.
rural localities screened
localities below 0.9 egress continuity today
residents in those localities
of them reach 0.9 or more under a 3,000 km national protection plan
Live today
Loading the live feed
Map
Five pressures, one dot per locality
Each dot is a locality, sized by its 2021 population and coloured by the selected measure. Select a dot or a table row to see its detail card. Ringed dots are the remote and island group: shown, but not ranked.
Colour by
Weighting
lower pressure higher pressureno data (crew access is recorded in NSW, VIC, SA and TAS only)remote and island group
Ranking
Top 20 under the Equal weighting
Composite = weighted mean of the available sub-indices (0 to 100, higher means higher pressure). Switching the weighting changes who appears here; the three boxes count how many of the top 20 differ between presets.
#
Locality
Composite
Sub-indices
Remote and island group
Localities whose shortest egress route is longer than 300 km, plus offshore islands with no fixed road link, such as Palm Island, the Torres Strait communities and Kangaroo Island. Island routes are routing artefacts. The 300 km rule also takes in some regional cities, among them Mount Isa, Albany, Geraldton, Kalgoorlie, Broome, Esperance, Atherton, Mareeba, Innisfail and Port Douglas. They sit here because the model has only 117 egress destinations, so their routes are long; it does not mean they face higher fire pressure. Their scores are shown for reference and not ranked against the rest.
Locality
Residents
Shortest route
Continuity now
Composite
Optimisation
How much does an optimised road protection budget help?
Egress continuity is the share of the 34 recorded seasons in which at least one of a locality's three egress routes stays open. Choose a national budget of protected road length and see which localities move from below 0.9 to 0.9 or more.
localities below 0.9 continuity
localities moved to 0.9 or more
residents in the localities that moved
protected road edges in the plan
below 0.9 under this budgetmoved to 0.9 or more0.9 or more alreadyProtected edges by failed seasons of 34:12 to 34 to 67 or more
One national optimum, not an agency plan. The protected edges are the stored optimal solutions of the AUFireBench P2 national instances (three routes per locality, 34 seasons, budgets of 100, 1,000 and 3,000 km). The optimiser spends the budget where it lifts continuity most across the country, so a locality gains only when the plan covers its routes. A road edge counts as failed in a season when a mapped fire perimeter of that season touches it.
34 seasons
The index season by season
National index value for the ranked localities in each recorded fire season, with no protection and under each budget. Hover a season for the residents and localities with every egress route cut. The 2023-24 season is incomplete (the fire record ends on 15 October 2023), and the three plans are separate optima, so 3,000 km can sit slightly above 1,000 km in a season.
Loading the season series
no protection100 km1,000 km3,000 km
Related indices in the Lab
Corridor Watch (CWI). The same Victorian road closures, seen from the freight network: which corridors carry the disruption today.
Remote Access & Resilience Atlas (RARI). Access to services and supply chains for remote communities, many of which sit in the remote and island group here.
ICC Wildfire Response Twin. One fictional fire up close: how a single egress loss plays out for evacuation and traffic.
The live BEI counts a modelled escape route as under pressure when an active fire (at least two satellite detections from polar-orbiting sensors, within 1 km) touches it, the same rule as the 34-season history, which uses mapped fire perimeters. Official road closures for any reason, such as flooding, and severe fire weather are shown beside it as context and are not in the BEI. Routes are model routes, not road status, and the values are screening indicators, not an official rating.
How to read this
Screening only. The values are screening indicators for teaching and discussion. They are not an official risk rating and no agency uses them.
Road failure is a geometric proxy. A road edge fails in a season when a mapped fire perimeter of that season touches it. This is not observed closure.
Crew data covers four states. Crew access is computed only for New South Wales, Victoria, South Australia and Tasmania, where public station volunteer data exist. Elsewhere it shows as no data, never zero, and the composite uses the remaining sub-indices.
Remote and island localities are kept apart. Routes longer than 300 km cross a lot of fire-prone country, and offshore islands such as Palm Island and the Torres Strait communities are routing artefacts. Some regional cities fall in this group only because the model has 117 egress destinations, which makes their routes long, not because they face higher fire pressure.
Coverage. The 1,714 localities include no Northern Territory localities; the Australian Capital Territory has two.
Separate plans. The 100, 1,000 and 3,000 km plans are separate optimal solutions, not nested ones, so a larger budget can score slightly worse than a smaller one in a single season (2001-02 and 2015-16 in the chart). Each plan is never worse than no protection.
Incomplete season. The fire record ends on 15 October 2023, so 2023-24 is an incomplete season.
Residents, not visitors. Population is usual residents in 2021; visitors and seasonal peaks are not counted.
No score uses Indigenous share. It appears only in the locality detail card, as Census context.
Scaling. Fire exposure is the mean share of the 34 seasons with a mapped perimeter within 5 km and within 10 km. Egress fragility is 85 percent of (1 minus continuity today) plus 15 percent of the shortest route length capped at 300 km. Evacuation dependence, crew access and disadvantage are scaled between the 1st and 99th percentiles of the ranked localities. Each sub-index runs from 0 to 100, higher meaning higher pressure.
Contains ABS, Geoscience Australia and OpenStreetMap data; AUFireBench 1.0.0 (Shahparvari 2026). ABS and Geoscience Australia data CC BY 4.0; OpenStreetMap ODbL 1.0. Basemap tiles from Esri. Author: Shahrooz Shahparvari
About this tool
Research showcase: Bushfire Egress Index
The concept
A screening index for rural Australia that keeps five pressures apart: fire exposure, egress fragility, evacuation dependence, crew access and socio-economic disadvantage, each scaled from 0 to 100.
How it works
The page computes every sub-index in the browser from the AUFireBench 1.0.0 locality table. The composite is a weighted mean, and the optimisation panel reads the stored optimal road protection solutions of the P2 national instances.
Winning strategy
Read the sub-indices before the composite. The weighting decides who leads the ranking, while the road budget shows where one national plan lifts egress continuity and where it leaves pressure in place.
BEI = sum over k of w_k S_k / sum over k of w_k, for the available sub-indices k; each S_k in [0, 100]
