Performance metrics
Road transport is a socio-technical system, and that is the whole problem.
Acute rainfall-induced asset-level disruptions do not stay at the asset. They cascade into wider social, environmental and economic impacts. A closed road becomes a missed shift, which becomes lost income, while the traffic rerouted around it becomes emissions in a different neighbourhood.
Transport-system resilience is commonly concerned with maintaining or restoring system performance under disruptive conditions (Mattsson and Jenelius, 2015; Ganin et al., 2017). To achieve that, a comprehensive assessment of disruption impacts is essential, both for evaluating flood risk exposure and for informing a deeper understanding of road network resilience to heavy rainfall events.
Why new metrics were needed
A meaningful assessment of road transport resilience to different weather events or ageing infrastructure requires quantifiable metrics that align with the performance requirements and services central to the mission of transport agencies.
The handbook identifies two gaps that the existing metric landscape leaves open.
Gap 1: existing metrics measure normal conditions
There is a lack of well-established performance metrics specifically designed to capture the multi-dimensional impacts of rainfall-induced acute road disruptions.
A review of relevant industry and government reports in the UK indicates that most existing metrics are designed to monitor or evaluate road network performance under usual weather conditions, rather than to stress-test systems against extreme weather and quantify the impacts of severe disruptions.
Existing metrics primarily focus on:
| Focus | Typical example |
|---|---|
| Physical asset condition | Percentage of pedestrian and cycle facilities in poor or very poor condition |
| Operational performance | Delay, journey time reliability, network availability |
| Safety impact | The number of people killed or seriously injured |
| Environmental impact | Noise, biodiversity, air quality, carbon emissions |
| User satisfaction | Percentage of road users very or fairly satisfied in the national survey |
Monitoring is not stress-testing
These are perfectly good indicators for their purpose, which is tracking how a network performs across a normal year.
They are the wrong instrument for asking what happens during a one-in-thirty-year rainfall event, because they are designed to average across conditions rather than to isolate an extreme.
Gap 2: models ignore the built environment and user interaction
Existing research on road transport modelling for flood resilience analysis scarcely considers two things:
Key characteristics of the built environment, such as road type, number of lanes, maximum speed, allowed transport modes and road capacity.
The spatio-temporal interactions among transport users within the network.
These aspects are fundamental, because each road link has a limited capacity for accommodating road users in various transport modes, such as cars, buses and bicycles, within a specified time frame, typically measured on an hourly basis.
This is why the IMP is built the way it is
Both gaps point at the same requirement: an agent-based model running on a network with real link attributes.
That is precisely what transport supply and the transport model provide, which is what makes these metrics computable here.
The POESE framework
To address these limitations, the POESE framework defines multidimensional performance metrics covering physical, operational, environmental, social and economic impacts of transport disruption (Bi et al., 2026).
POESE advances a systemic perspective in transport resilience assessments by tracing the ripple effects of physical asset-level disruptions, caused by different weather conditions such as flooding, through operational performance loss towards broader environmental, social and economic consequences.
It offers a structured and comprehensive reference for quantifying transport resilience across multiple dimensions, enabling clearer interpretation of disruption impacts and more consistent benchmarking across events.
The impact chain
The metrics are captured from analysing road disruption impact chains following different weather events, such as heavy rainfall events and heatwaves. They trace how disruption propagates:
- Physical asset-level disruption
- Operational performance loss
- Broader environmental, social and economic consequences
The order is the argument
POESE is not five unrelated scorecards. It is one causal chain, and the codes follow it.
A flooded link (PH) reduces availability, which causes delay and cancellation (OP), which drives rerouting and congestion (EN), which lands unevenly on different communities (SO), which carries a monetary value (EC).
Reading a POESE result means reading down that chain, not picking one number from it.
Short-term and long-term disruption
The metrics are grouped according to which kind of disruptive event they apply to.
The full set is designed for long-term disruption scenarios, characterised by catastrophic asset damage and extended recovery periods.
For short-term disruptions without prolonged recovery, such as those triggered by rainfall but fully restored within 24 hours or several days, some impacts are expected to be minimal and can therefore be excluded from the assessment.
Per the metric table below, nine metrics apply to long-term disruption only:
| Excluded for short-term | Dimension |
|---|---|
| EN2, EN3 | Environmental |
| SO5, SO6, SO7 | Social |
| EC6, EC7, EC8, EC10 | Economic |
Why these drop out
Each excluded metric depends on sustained exposure. Air pollutant concentrations, accessibility loss, disease burden and reconstruction costs all need the disruption to persist before they register meaningfully.
A road that reopens by teatime does not rebuild anyone's asthma risk.
The source contradicts itself on EC9 and EC10
The handbook's metric table and its surrounding prose disagree about which economic metrics apply to short-term disruption.
| Metric | Metric table says | Prose exclusion list says |
|---|---|---|
| EC9 (cost of injuries and fatalities) | Short and long-term | Excluded from short-term, as part of "EC6-9" |
| EC10 (cost of social inequity) | Long-term only | Not listed as excluded, implying it applies to short-term |
This page follows the metric table, which is the more specific and more authoritative statement, and which is also the more defensible reading: injuries occur during a short disruption and should be costed, whereas social inequity arises from sustained limitation.
The discrepancy should be resolved in the source document.
The full metric set
Thirty-two metrics across five dimensions.
| Code | Performance metric | Applies to |
|---|---|---|
| PH1 | Number of road links affected, for example by flooding | Short and long-term |
| PH2 | Total length of road links affected (km) | Short and long-term |
| PH3 | Total length of critical roads, such as A-Roads in the UK, affected at varying depths (km) | Short and long-term |
| OP1 | Total delay (min) | Short and long-term |
| OP2 | Average delay (min) | Short and long-term |
| OP3 | Morning-peak (7 to 9 AM) average delay (min) | Short and long-term |
| OP4 | Afternoon-peak (4 to 6 PM) average delay (min) | Short and long-term |
| OP5 | Number of cancelled journeys by trip purpose | Short and long-term |
| OP6 | Number or percentage of cancelled bus service | Short and long-term |
| EN1 | Increase in GHG emission (kgCO2e) | Short and long-term |
| EN2 | Increase in NOx emission (kg or μg/m³) | Long-term only |
| EN3 | Increase in PM2.5 emission (kg or μg/m³) | Long-term only |
| SO1 | Percentage of individuals experiencing delayed journeys by demographic group (%) | Short and long-term |
| SO2 | Percentage of individuals experiencing cancelled journeys by demographic group (%) | Short and long-term |
| SO3 | Geographical distribution of individuals experiencing delayed journeys by demographic group | Short and long-term |
| SO4 | Geographical distribution of individuals experiencing cancelled journeys by demographic group | Short and long-term |
| SO5 | Decrease in the number or percentage of a population group able to access a type of destination from home within a given proximity by a specified transport mode | Long-term only |
| SO6 | Geospatial distribution of increased air pollutant emission across demographic groups (μg/m³) | Long-term only |
| SO7 | Increased burden of disease | Long-term only |
| SO8 | Number of slightly injured individuals | Short and long-term |
| SO9 | Number of seriously injured individuals | Short and long-term |
| SO10 | Number of fatalities | Short and long-term |
| EC1 | Cost of journey delay (£) | Short and long-term |
| EC2 | Cost of journey cancellation (£) | Short and long-term |
| EC3 | Public transport revenue loss (£) | Short and long-term |
| EC4 | Cost of emergency response (£) | Short and long-term |
| EC5 | Cost of asset repair, including labour costs (£) | Short and long-term |
| EC6 | Cost of asset reconstruction, including labour costs (£) | Long-term only |
| EC7 | Cost of increased GHG emission (£) | Long-term only |
| EC8 | Cost of increased air pollutant emission (£) | Long-term only |
| EC9 | Cost of injuries and fatalities (£) | Short and long-term |
| EC10 | Cost of social inequity (£) | Long-term only |
Implementation
Code location is internal
A Python implementation estimating performance metrics for short-term disruptions is noted as available on the DARe Server. This is an internal location and should be replaced with a citable or publicly accessible reference before publication.
What POESE produces
The framework provides a set of multidimensional metrics for quantifying the impact of different extreme weather conditions, such as flood-induced disruption and heatwaves, or of ageing infrastructure in road transport systems.
By explicitly capturing all five dimensions, the POESE metrics move beyond existing indicators that either monitor annual system performance under typical weather conditions, or consider only network connectivity. The result is a structured framework for assessing road transport resilience.
What this chapter covers
| Page | Metrics | What you will find |
|---|---|---|
| Physical impacts | PH1 to PH3 | Network availability lost to the hazard |
| Operational impacts | OP1 to OP6 | Delay and cancellation |
| Environmental impacts | EN1 to EN3 | Emissions from congestion and rerouting |
| Social impacts | SO1 to SO10 | Equity, accessibility, health and safety |
| Economic impacts | EC1 to EC10 | User, owner and societal costs |
| Further reading | The source paper and its demonstration |