Cruise-day traffic forecast

1. The network is Southampton's. one-way road links from OpenStreetMap, with lanes and speed limits, from motorways down to unclassified roads.
2. Zones and gateways come from geography. zones built from OpenStreetMap places. Each zone's homes and jobs are Census 2021 residents aged 16+ and workplace counts (ODWP01EW, per MSOA), placed within each MSOA by nearby residential, commercial, industrial and port land; its shops come from retail land. county gateways sit where the M27, M3, A35, A33 and other roads cross into the area. Where cutting the map to the area broke the M27 and M3, and where OpenStreetMap lacks the M3 to M27 eastbound link, the model joins them up.
3. Everyday traffic is generated from data. Commuters from the county and back, residents working outside, trips between zones (more between big, close zones), through traffic, county visitors to the shops, and lorries to and from the port (fitted to DfT lorry counts on the M271). Each has its own pattern over the day, different at weekends. The school run adds drop-off and pick-up peaks on term-time weekdays (city council half-term 26–30 Oct). Match days add fans to St Mary's.
4. Counted roads use the counts. On Portswood Road and Bevois Hill, where the model's own traffic falls short of the July 2024 counts, capacity is the counted peak-hour flow plus : peak speeds there were close to the all-day average and buses ran 0–4 minutes behind timetable, so the roads were busy but not full. Highfield Lane and St Denys Road, where the model already carries the counted traffic, keep their road-class capacity. That margin makes modelled bus run times on Portswood Road match the council's on a July-like weekday. Local traffic the zones miss is added so each link's flow matches the counts, then in university term (21 Sep to 12 Dec 2026): the counts were taken in July, outside term, and few students drive (assumed).
5. Cruise traffic is added ship by ship. Car drop-offs, park and cruise, taxis and coaches, timed by disembarkation and boarding and sent to the terminal's nearest dock gate.
6. Drivers choose routes as roads fill. Every 15 minutes each of the origin–destination flows splits over up to three real routes and shifts towards the quickest. Traffic that cannot get through a full road waits for the next 15 minutes, so queues build and clear.
7. Days vary. Drivers misjudge speeds together; everyday traffic swings by day, trip type, area and hour; each ship's passengers share its delay and travel mix. Wet days and incidents occur at their usual rates, and dates further ahead get wider ranges.
8. Many simulated days. Each date is simulated times. A road's chance of congestion is the share of those days on which it goes over 85% of its capacity. Each simulated day is also run without the ships, so the ships' effect is measured on the same day.

Scaled . Commuters from the county and residents working outside are Census 2021 car drivers crossing the edge of the area (). Through traffic and other county trips come from DfT counts on the roads into the area; on the M27 and M3 only the drop to the next count inside can leave for the area, since consecutive motorway counts see mostly the same vehicles. Port freight is in car-equivalents (a lorry counts as 2.5 cars on a busy road), fitted to DfT lorry counts. Visitors, trips within the area and the school run are invented totals; trips within the area are fitted to council counts, and everything crossing the edge of the area to DfT counts, because census day fell in lockdown. Cruise travel split: .

Trips within the area are scaled so that, summed over four counting sites around Portswood, modelled weekday flows match what was counted. Saturday is a check, not fitted. Single roads are less accurate, because each zone loads the network at one point. Source: .

The same report gives bus run times on Portswood Road (Table 3-5): in July 2024 services ran 0:24 to 4:22 minutes behind timetable in the peaks, and a quarter of Bluestar 2 southbound buses were 5 or more minutes late. Modelled delays on Portswood Road should sit in that range out of term and above it in term.

County commuters, visitors and through traffic are scaled together so that modelled average daily two-way flows match the DfT annual average daily flow (all motor vehicles) at these count points. Modelled flows are 05:00–20:00 on a weekday, Saturday and Sunday, weighted 5:1:1 and grossed up to 24 hours ( of traffic assumed between 05:00 and 20:00). Estimated points are DfT figures factored from an older count and count half in the fit. Itchen Bridge is tolled and the model has no tolls, so some of its traffic sits on Northam Bridge instead; both are checks, not fitted. Modelled port lorries count once here, not as 2.5 cars. Source: , downloaded .

Lorries run between the county gateways (shared by each gateway road's DfT lorry count) and the port: . Their number is scaled so that the M271 between the A35 and junction 1, which leads only to the Western Docks and Redbridge, carries of its DfT lorry count (assumed port share). The other points also carry lorries that are not going to the port, so modelled port lorries should sit below their counts. Weekend days run at of a weekday, and lorries follow the hour-by-hour pattern of articulated lorries leaving the docks on the M271 (National Highways WebTRIS site 19974, 410 complete days, Oct 2024 to Jun 2026). The port's own gate counts would replace these assumptions. Source: .

• Queues blocking junctions upstream (queues wait at the bottleneck only)
• School locations: school trips follow where people live for now
• Signal timings at individual junctions
• Rail, and people switching mode when roads are busy (buses run on the modelled roads in the Vehicles view but do not add to the traffic)
• Traffic inside the port
• The Itchen Bridge toll
• Commuters' routes outside the area (each picks a gateway by a straight-line estimate)

Only roads with sensors can be checked against reality. Every other road on this page is a model output.