How It Works

Last updated: 16 May 2026

This page explains the inputs, methods, and limitations behind everything on SG Traffic Now — the live travel-time estimates, the forecasts, the alerts, and the guides.

Data Sources

Real-time and reference data are sourced from official providers. Each source has its own refresh cadence and its own authoritative timestamp.

Singapore

Land Transport Authority (LTA) DataMall — Singapore traffic camera feeds and speed-band data, used for live conditions and historical pattern analysis. Use of LTA data is subject to the Singapore Open Data Licence.
Immigration & Checkpoints Authority (ICA) — official advisories from ICA's verified social media accounts and media releases.

Malaysia

Lembaga Lebuhraya Malaysia (LLM) — federal highway camera feeds.
PLUS Expressways (PLUS) — North-South Expressway camera feeds.
Kuala Lumpur Command & Control Centre (KLCCC) — Kuala Lumpur metropolitan traffic cameras.
Majlis Bandaraya Johor Bahru (MBJB) — Johor Bahru municipal cameras.
Jabatan Imigresen Malaysia (JIM) — Malaysian immigration advisories.

Trip-planning data

open.er-api.com — SGD/MYR exchange rate.
data.gov.my — Malaysia petrol prices.
Official government publications — Singapore and Malaysia public holiday lists.

Refresh cadence

Camera images: pulled every 3 minutes on our side, from upstream feeds that themselves refresh every 1–5 minutes depending on the provider.
LTA speed bands: every 3 minutes.
Exchange rate: daily.
Petrol prices: weekly (mirrors the data.gov.my publication cycle).
Holiday data: refreshed when official sources publish updates.

All traffic camera images and data are used for informational purposes and remain the property of the respective providers.

Real-Time Travel-Time Estimates

The duration cards on the homepage display estimated crossing times for the immediate window — for SG → JB and JB → SG, at both Woodlands Checkpoint and Tuas Second Link.

How we currently estimate

Estimates are built from manual labelling of historical observations combined with pattern recognition. For each direction and checkpoint, we have records of typical traffic behaviour by time of day, day of week, and proximity to public holidays.

The live estimate combines that pattern with the current observed congestion signal — LTA speed bands for the approach corridors and visual inspection of upstream camera feeds.

Why not Google Maps

Google Maps is available for the same routes. For the causeway specifically, two limitations motivated us to build our own estimates:

Underestimated durations. Predicted travel times consistently fall short of observed wait times during peak windows.
No vehicle-type differentiation. Motorcycles, cars, and heavy vehicles share lane infrastructure but move at very different rates. Maps treats the road as a single class.

What's next (machine learning and computer vision)

The current approach is manual. We are working towards an automated pipeline that classifies vehicle queues directly from live camera feeds and produces faster, more accurate per-vehicle-type estimates.

The goal is to publish separate live wait times for motorcycles, cars, and heavy vehicles, and to update those figures more frequently than current upstream feeds allow.

Forecasts

The Plan Your Crossing, Woodlands Forecast, and Tuas Forecast pages show predicted congestion for upcoming days.

Method

Forecasts are pattern-based. We use at least two months of historical observations to derive typical traffic patterns by:

Direction (SG → JB and JB → SG)
Checkpoint (Woodlands and Tuas)
Time of day, at hourly resolution
Day of week
Proximity to public holidays and school holidays

Each forecast cell is labelled Low, Medium, or High based on those patterns.

Important caveat

Forecasts are forecasts. They reflect typical behaviour for the same time-of-day and day-of-week combination — not a guarantee of future conditions.

Incidents, weather, immigration-system outages, and one-off events can move actual traffic far outside the forecast band. The forecast is a planning aid, not a guarantee.

Alerts

The Alerts page lists current and recent advisories that may affect crossings.

Sources

LTA API — official LTA traffic advisories.
ICA verified social media — official ICA posts about checkpoint conditions.
Official media — Singapore and Malaysia outlets reporting on causeway-related incidents, maintenance, or scheduled closures.

How we categorise

Incidents — short-notice events (accidents, vehicle breakdowns, system outages).
Congestion — sustained heavy traffic outside normal peak windows.
Scheduled — announced closures, maintenance, and pre-publicised peak-period notices.

Alerts are curated manually before publication. We summarise the upstream source, link back where possible, and keep the alert visible until conditions return to normal or the upstream source declares the event resolved.

Guides

Articles on the site cover regulations, procedures, and crossing strategies — for example the Three-Quarter Tank Rule, Vehicle Entry Permit (VEP) requirements, QR code immigration clearance, and best-time-to-cross guides.

Editorial process

Sources are cited under each article. Regulatory claims link to their primary official source — for example, ICA, LTA, Singapore Customs, Jabatan Pengangkutan Jalan Malaysia (JPJ), or Malaysia's Ministry of Transport.

When rules change, affected articles are updated and the "Last updated" date is revised. We do not republish unverified rumours. If a piece of information cannot be tied back to an official source, we either omit it or label it explicitly as estimate or interpretation.

Known Limitations

Camera latency. Images can lag the road by 1–3 minutes. The timestamp on the image is the authoritative reference.
Camera outages. When an upstream feed is offline, the most recent retrieved image is shown until the feed resumes.
Forecast accuracy. Forecasts reflect typical conditions and do not account for one-off incidents, weather, or system failures.
Vehicle-type coverage. Current estimates are based on overall traffic patterns dominated by car traffic. Motorcycle and heavy-vehicle estimates are not yet split out separately.
Precision. Figures are rounded to keep them honest. Second-by-second precision is not provided.

Roadmap

In active development:

Machine learning and computer vision for vehicle-type classification.
Separate live and forecast estimates for motorcycles and heavy vehicles.
Faster real-time update cycles for the duration cards.

Corrections

If something on the site looks wrong, email [email protected]. We review every correction.