Measured, estimated or assumed
About the data
Every number on these screens is either measured in the real world or produced by our simulation. This page says which is which, where each one comes from, and how far to trust it.
How to read the tags
On every number
- REAL
- Measured in the real world: by TomTom (from probe vehicles) or by a published traffic study.
- SIMULATED
- Produced by our simulation of the roads. Only as good as its calibration.
- SAMPLE DATA
- A saved example result (the API in mock mode), not a simulation of what you asked.
On every input to the simulation
- counted / measured
- Someone measured it: a traffic count, a study, a probe speed or travel time.
- estimated
- Worked out from measured data, e.g. TomTom's vehicles-per-hour figure, which is its model's estimate.
- calibrated
- Tuned by us until the simulation's trip times match TomTom's. Fitted, not measured.
- assumed
- A reasonable guess with no data behind it yet, e.g. the split between cars and autos.
What is real: the sources
TomTom Traffic Stats REAL
Historic travel times and speeds on our route, from TomTom's probe vehicles: July 2026, every day 06:00–23:00, split into the 12 stretches between junctions. This is what the simulation is calibrated against.
TomTom Junction Analytics REAL
Minute-by-minute data for single junctions: the delay on each approach (measured from GPS trips), and the queue length and vehicles per hour, which are TomTom's model estimates, not counts. Also which way traffic turns.
Loading live data…
OpenStreetMap roads MAP
The road network the simulation drives on: lanes, speed limits, one-way streets and junction shapes, from the free, volunteer-made OpenStreetMap. It is a reference map, not a measurement: where it is wrong (a missing lane, an odd junction shape), the simulation is too.
Overture building footprints MAP
About 16,400 building outlines along the corridor (Overture Maps, release 2026-09-23.1) and 3,370 around YMCA Circle. They are only used to draw the 3D city. Most buildings have no recorded height, so heights are assumed from the footprint size (floors × 3.2 m). The simulation does not use them.
YMCA Circle traffic count, 2020 study REAL
Vehicles per hour by type on each road into YMCA Circle, from a published study (IJRAR, June 2020). Used for the YMCA close-up's traffic mix. Traffic has grown since: the 2.0× growth to 2026 is estimated by matching TomTom's speeds.
Open-Meteo rain, hour by hour REAL
Rain for every hour of July 2026 at three points on the route (a weather reanalysis, not a rain gauge), compared with TomTom's trip time for the same hour (558 one-hour slots, 06:00–24:00, every day of the month). In an hour with rain the trip was about 3% slower than the same hour on a dry day of the same weekday. Once rain had lasted an hour it was about 4% slower in light rain and about 8% in the wettest hours July had, but there were only 34 of those, so that range is wide. One check pulls the other way: rainy and dry hours of the same day took about as long, so part of this may be “rainy days are slower days”. Confidence: low to moderate. July had no cloudbursts in this data, so we cannot say what a downpour or waterlogging does.
Stretches most affected in light rain: Nallagandla → ISB Rd, Gachibowli Circle → Biodiversity jn and Khajaguda → Shaikpet. The corridor screen's Weather setting (dry, light rain, heavy rain) applies these factors, stretch by stretch, to the simulation's road speeds; traffic volume and signals stay the same (assumed).
How well the simulation matches TomTom SIMULATED vs REAL
Loading the latest calibration…
Trip time per stretch
What we tuned (calibrated knobs)
Delay at junctions with TomTom data
Traffic volume on the main road
What goes into the simulation, and what is assumed
Loading…
Words used on this page
- TomTom
- A company that makes maps and collects traffic data from millions of phones, cars and navigation devices. We use two of its products: Traffic Stats (how long trips took) and Junction Analytics (what is happening at a junction now).
- Probe vehicle
- Any vehicle that sends its anonymous GPS position to a company like TomTom. TomTom sees only a sample of all traffic: speeds and travel times from probes are reliable; vehicle counts are TomTom's estimate.
- Stretch (leg)
- The road between two neighbouring junctions, e.g. Tolichowki → Nanal Nagar. The whole trip is the sum of 12 stretches.
- Approach
- One road coming into a junction. A crossroads has four. TomTom reports delay and queue per approach.
- Simulation
- A computer copy of the roads (SUMO, an open-source traffic simulator) where thousands of virtual two-wheelers, cars, autos and buses drive, stop at signals and queue. We change the copy, not the real roads.
- Calibration
- Tuning the simulation until it behaves like the real roads: here, adjusting speed limits per stretch, signal green times and traffic volume until the simulated trip times match TomTom's. It is what makes the “what if” answers worth trusting.
- Green share
- The part of a signal's green time given to the main road. 80% means the corridor gets 80% of the green and the side roads share the rest.
- Reanalysis
- Weather data rebuilt by a weather model from satellite and station records, on a grid of about 9 km. Good for “did it rain here that hour”, less exact than a rain gauge.
- Mock mode
- The API returns saved sample results instead of running the simulator (
MOCK_SIM=1). Used while building screens and as a demo fallback. The status pill in the top bar says which mode is on.
The full glossary for the team: docs/glossary.md. Every file and its source: data/README.md.