Traffic Noise Maps Go Probabilistic: New Model Pinpoints Urban Hotspots in Canada and Malaysia

Road traffic noise is one of the most pervasive and least visible pollutants in modern cities. Unlike smog or contaminated water, it leaves no residue, yet decades of epidemiological research have linked chronic exposure to elevated risks of cardiovascular disease, sleep disturbance, cognitive impairment in children and reduced overall quality of life. For city planners trying to meet the United Nations Sustainable Development Goal 11, which calls for cities that are inclusive, safe, resilient and sustainable, noise has long been a stubborn blind spot. Conventional noise maps tend to treat traffic as a smooth, deterministic flow, averaging away the very variability that makes intersections so unpredictable. A new study published in npj Urban Sustainability argues that this averaging is precisely where conventional approaches go wrong, and it offers a probabilistic alternative that embraces uncertainty rather than erasing it.

The research, led by Lee Hang Tan, Ming Han Lim, Hooi Ling Khoo and Yee Ling Lee of Universiti Tunku Abdul Rahman in Malaysia, together with Po-Han Chen of Concordia University in Montreal, presents an end-to-end simulation framework for mapping traffic noise at signalized crossroads. Rather than assuming fixed traffic volumes and a fixed mix of vehicle types on each approach to an intersection, the framework explicitly represents both quantities as stochastic variables, drawing on probability distributions fitted to field data. This means that every simulated traffic scenario reflects the real-world randomness of urban driving: the sudden surge of motorcycles at a Malaysian junction during rush hour, the lull in traffic late at night, the intermittent arrival of heavy trucks on a Canadian arterial road. The output is not a single deterministic noise value but a distribution of plausible sound levels at each receiver location, from which error statistics and confidence in the prediction can be derived.

The technical workflow proceeds in several connected stages. First, the team characterizes traffic probabilistically at each intersection approach, estimating the distributions of vehicle counts and of vehicle-type composition, typically separating motorcycles, passenger cars, light trucks and heavy vehicles, because each class emits noise with distinct spectral and level characteristics. Second, these stochastic inputs feed a traffic-noise prediction model that computes sound pressure levels at defined receiver points, accounting for the geometry of the intersection and the propagation of sound from multiple simultaneous sources. Third, the framework aggregates predictions across many simulation runs to generate spatial noise maps that show not only where sound levels are highest but also how confident the model is in each prediction. This end-to-end chain, from probabilistic traffic characterization through spatial prediction to map generation, is what distinguishes the approach from one-off measurements or purely deterministic modeling exercises.

Validation was carried out with field measurements at six urban intersections, three in Malaysia and three in Canada. The choice of two countries on opposite sides of the planet was deliberate and scientifically important. Malaysian intersections are dominated by high proportions of motorcycles, which produce sharp, high-frequency noise bursts and behave very differently in traffic streams than the car- and truck-heavy fleets typical of Canadian cities. If a noise-mapping framework can perform well across such contrasting vehicle fleets, climates, road geometries and driving cultures, it has a much stronger claim to general applicability than a model tuned to a single city. Across the sampled receiver locations, the model generally produced low percentage errors and small mean absolute differences between predicted and measured sound levels, providing proof-of-concept evidence that the probabilistic treatment of traffic variability translates into accurate, reliable predictions under very different urban conditions.

The resulting noise maps revealed intersection-scale hotspot patterns that would be difficult to detect with sparse spot measurements alone. Intersections are acoustic hotspots by nature: vehicles accelerate away from stops, engines labour under load, and sound from four converging arms overlaps at the corners where pedestrians wait and where storefronts, schools and clinics are often located. By resolving these patterns at the scale of a single junction, the framework allows planners to identify precisely which corners, crossings and frontages experience the highest exposures, and to compare alternative mitigation options before committing construction budgets. The authors emphasize that the maps also open the door to scenario testing: because the underlying traffic is simulated probabilistically, planners can ask what happens to noise exposure if motorcycle volumes rise, if a new bus route changes the vehicle mix, or if signal timing is adjusted to smooth acceleration patterns.

That scenario-testing capability may prove to be the framework’s most consequential contribution. Traditional noise mitigation, such as barriers, quiet pavement, building insulation or traffic calming, is expensive, and retrofitting it after complaints arise is far costlier than planning for it from the start. A probabilistic noise map gives municipalities a screening tool that can be run repeatedly and cheaply, ranking intersections by expected exposure and by the uncertainty surrounding that expectation. An intersection whose predicted levels are high but uncertain might warrant additional measurement, while one with consistently high predicted levels across simulation runs becomes an obvious priority for intervention. In this way, the statistical machinery is not an academic nicety; it directly shapes how engineering resources are allocated.

The study also connects to a broader shift in how environmental scientists think about urban noise. For years, the field has recognized that noise exposure is fundamentally a stochastic phenomenon, governed by random arrivals of vehicles, variable speeds, driver behavior and weather-dependent propagation. Yet many operational noise maps used in regulatory contexts still rely on averaged inputs, partly because deterministic models are simpler to run and easier to defend in legal settings. The Canadian-Malaysian collaboration demonstrates that the computational cost of a probabilistic treatment is manageable and that the payoff, in the form of quantified reliability, is substantial. The framework’s low percentage errors and small mean absolute differences at validated receiver points suggest that the added complexity buys real accuracy rather than merely statistical sophistication.

The international dimension of the work was supported by a notable funding architecture. Activities in Malaysia were financed by the country’s Ministry of Higher Education through a Fundamental Research Grant Scheme project and a UTAR Research Fund project, while the Canadian component was funded by the Government of Canada through the Educational Exchanges for Development (SEED) program. The collaboration drew on research facilities, library resources and computing infrastructure at both Universiti Tunku Abdul Rahman and Concordia University. This kind of bilateral support is increasingly seen as essential for sustainability research, because environmental burdens such as noise do not respect national boundaries in their underlying physics, and solutions validated in only one context often fail when exported.

For the Sustainable Development Goal 11 agenda specifically, the study offers a concrete, replicable instrument. SDG 11 targets reductions in the adverse per capita environmental impact of cities, and noise is explicitly recognized in related indicator frameworks as a component of urban environmental quality. A city that adopts probabilistic intersection noise mapping gains the ability to baseline its current exposure, identify hotspots, test interventions in silico and then verify improvements with follow-up measurements, closing the loop between policy and outcome. Because the framework requires only traffic characterization data and standard noise-prediction machinery, it is within reach of mid-sized municipalities that lack the budgets for exhaustive year-round monitoring campaigns.

The authors are careful to frame the work as proof of concept rather than a finished regulatory tool. Six intersections, however well chosen, cannot represent every urban geometry, and extending the framework to larger networks, to non-signalized junctions and to sources beyond road traffic, such as railways, construction and aircraft, remains future work. The published version is also subject to the usual caveats of an early-release, open-access article that will be replaced by the final Version of Record. Nevertheless, the central message stands: treating traffic as the random process it truly is produces noise maps that are more accurate, more honest about their own uncertainty and more useful for planning. As cities worldwide grapple with densification, electrification of vehicle fleets and changing mobility patterns, the soundscape of the intersection, that deceptively ordinary patch of asphalt where four streams of traffic meet, may become one of the most closely watched indicators of urban livability. This study shows that watching it probabilistically is both feasible and worthwhile.

Subject of Research: Probabilistic road traffic noise mapping at urban intersections for sustainable city planning

Article Title: Probabilistic intersection noise mapping for urban sustainability in Canada and Malaysia

Article References: Tan, L. H., Lim, M. H., Chen, P.-H., Khoo, H. L., & Lee, Y. L. (2026). Probabilistic intersection noise mapping for urban sustainability in Canada and Malaysia. npj Urban Sustainability. https://doi.org/10.1038/s42949-026-00480-4

Image Credits: AI Generated

DOI: 10.1038/s42949-026-00480-4

Keywords: traffic noise, noise mapping, urban sustainability, intersections, probabilistic simulation, SDG 11, Malaysia, Canada, environmental health, urban planning, stochastic modeling, hotspot screening