GLM Resilience Model
The GLM Resilience Model estimates, for every link on the Strategic Road Network, how much travel speed falls as rainfall rises.
What it estimates
| Scope | The Strategic Road Network |
| Unit of estimation | Each individual road link, separately |
| Relationship modelled | Rainfall intensity against travel speed |
| Basis | Observed data, 2022 to 2024 |
| Principal quantity | The resilience coefficient, per link |
The model builds on established evidence that rainfall systematically reduces traffic speed (Hranač et al., 2006; Becker et al., 2026). What it adds is specificity: rather than applying a general relationship, it derives one per link, directly from what was measured there.
The Explain stage
The handbook refers to an "Explain" stage of the framework, in which link-level attributes are used to account for why the sensitivity of speed to rainfall varies between links.
The example given is instructive:
Links with more lanes and higher speed limits are found to be more resilient
The handbook's phrasing, that such links "are found to be" more resilient, indicates this emerged from the data rather than being assumed.
The handbook does not say why. One plausible reading is that a wider road has more capacity to absorb cautious driving before flow breaks down, but the mechanism is not established in the source and should not be treated as settled.
This stage is what turns a set of coefficients into something transferable. Knowing that a link is sensitive supports prioritisation; knowing why supports predicting sensitivity on links without sensors, and designing interventions that address the cause.
The other framework stages are not described
"Explain" is the only stage the handbook names, which implies the framework has others, most likely a preceding estimation or prediction stage.
The full set of stages and what each does should be documented here before publication.
How it fits the wider IMP
The resilience tool runs as a parallel, empirical track alongside the simulation chain.
| Simulation chain | Resilience tool | |
|---|---|---|
| Hazard | Modelled by CityCAT | Observed rainfall at gauges |
| Network response | Simulated agent by agent in MATSim | Estimated statistically per link |
| Strength | Can explore unobserved conditions and adaptation options | Grounded in measured behaviour |
| Coverage | The modelled study area | The Strategic Road Network |
It shares one important input with the simulation side. The projected future demand it uses comes from the gravity model, the same spatial-interaction model documented in the transport demand chapter.
The join between demand and climate
The gravity model supplies projected traffic volume; the GLM supplies the rainfall-speed sensitivity. Combining them is what produces a projection of network performance under both future demand growth and future climate.
Neither input alone would support that.