Inputs and outputs
The tools earlier in this chapter each carry their own inputs and outputs. This page documents the one approach the handbook specifies end to end: the gravity model used to project demand across the Strategic Road Network.
Where this sits
This is the aggregate, spatial-interaction approach described in the overview, applied to a real study area rather than a synthetic population. It runs alongside the agent-based chain rather than replacing it.
The gravity model
A spatial-interaction (gravity) transport-demand model that grows observed base-year traffic in proportion to the modelled change in each site's accessibility to population and employment.
The logic is straightforward. If a sensor site becomes more accessible to people and jobs by 2045 than it was in 2024, the traffic passing it grows in proportion.
Required input data
Five datasets are needed.
1. Base-year traffic volume
| Source | National Highways WebTRIS |
| Description | Observed hourly traffic data recorded at permanent sensor sites on the Strategic Road Network, retrieved through the WebTRIS API and aggregated to annual volume (2022 to 2024) for each of the 329 sites in the study area |
2. Population projections
| Source | Office for National Statistics (ONS) mid-year population estimates |
| Description | Lower-layer Super Output Area (LSOA) resident population across England, projected to 2045 and weighted by road distance to each site to form a population accessibility index |
3. Employment projections
| Source | NOMIS Business Register and Employment Survey (BRES) |
| Description | LSOA employment across England, projected to 2045 and weighted by road distance to each site to form an employment accessibility index |
4. Planning policy
| Source | Greater Cambridge Local Plan site allocations |
| Description | The locations and capacities of allocated development, 78,527 new homes, converted into additional residents and jobs and added, area-weighted to LSOA, to the projected 2045 population and employment surface |
Why planning policy is a separate input
Trend-based projections extrapolate from what has happened. They do not know about housing that has been allocated but not yet built.
Adding local plan allocations puts committed development into the future surface explicitly, rather than hoping the trend happens to capture it.
5. Inter-zonal road distances
| Source | OpenStreetMap, routed with the Open Source Routing Machine (OSRM) |
| Description | Driving distances between every sensor site and every LSOA, used as the travel impedance, or travel disutility, in the gravity accessibility calculation |
Generated output data
Projected transport-demand map
A spatially distributed map of projected demand across the Strategic Road Network, covering each of the 329 sensor sites.
It fuses two figures:
| Figure | Year |
|---|---|
| Observed base-year volume | 2024 |
| Projected traffic volume | 2045 |
Each site's volume is grown in proportion to the change in its trip-end accessibility between 2024 and 2045.
The projected demand represents the future traffic loading on the network. It is subsequently combined with the rainfall-speed model to assess Strategic Road Network performance under future demand and climate.
Where this output goes next
The rainfall-speed pairing is the GLM Resilience Model, documented in the resilience tool chapter. This is the point where the demand projection meets the climate signal.
A heading error in the source
In the handbook, the section containing the projected transport-demand map is labelled "Required input data" for a second time, immediately after the first. From its content it is clearly the generated output, and it is presented as such here.