SAFER Tool
The Statewide Analysis of Fire Evacuation Risk (SAFER) tool, formerly referred to as the Statewide Evacuability Model (SEM), measures the combined risk imposed by multiple fires across the full geographic extent of an Australian state. The tool computes several evacuability risk-metrics from a given set of input data comprising a road network, a population dataset, and the geographical extents of wildfire scenarios. The algorithm relies on the maximum-flow method, which after comparison with other approaches has proved to be the most suitable for solving the statewide evacuability problem. The model potentially could be applied to any kind of natural disaster for which the geographical extent of an event can be predicted or otherwise estimated.
Inputs and Outputs
The SAFER model requires three type of inputs and produce three type of risk metrics as outputs. These outputs will be later uploaded into a web based tools for further analysis of these risk layers.
Input data
- Road network (Built using Open Street Maps)
- Population (Census based)
- Fire (Phoenix, Aurora)
Output data
- Ignition point risk layer :
This estimates the risk that a given fire poses to the population and road network lying within its perimeter.
- Community risk layer :
This estimates the risk that the aggregate of fire scenarios poses to a given community.
- Impact risk layer :
This estimates the risk that the aggregate of fire scenarios poses to a given critical road-link.
Below is a snapshot of SAFER web tool showing ignition point risk across Victoria
Application of Maximum-Flow method in bushfire evacuation
Below image demonstrates the maximum-flow calculations on a realistic road network. The orange node (50) is the ignition point, while the fire extent is represented by the yellow nodes. The light-green nodes represent possible exit-nodes (just outside the fire’s perimeter), while the dark-green nodes are the selected destination nodes (located in beeline distance away from the fire). The blue node (61) is the population node; the flow of vehicles will be calculated from this node to any of the dark-green safe nodes (nodes 95, 102-105 on the diagram). The arrows represent all possible routes between adjacent nodes, with the dark arrows indicating the weakest links.
The outputs provided by a maximum-flow algorithm can be readily processed to give numerical-valued risk-metrics, for example evacuation time per population, that yield best-case estimates of the risk posed by each individual fire scenario, and by all fires in aggregate, to the population nodes and road links.
Although the maximum-flow method is based on a simplified (idealised) representation of the evacuation problem, it is sufficient to calculate reliable evacuability risk-metrics for the whole state and for many fire scenarios in a computationally efficient manner.
Flows between source-destination pairs can be computed independently of the fire and of the population. It would therefore be potentially possible to pre-process the state’s entire network and pre-compute the maximum flow (and weakest links) between any pair of nodes.