One silver lining of the coronavirus lockdown is that many cities are now free of traffic jams. Surprisingly, this isn't the only way epidemics may smooth urban transportation, with new research showing we can model traffic congestion better using methods designed to model the spread of infectious diseases.
If limiting traffic jams seems like a trivial spin-off from the important work of disease control, remember that somewhere between 3 and 9 million people die each year from air pollution, depending on which estimates you believe. Cars are a major source of that, and the longer they are stuck bumper-to-bumper the more deadly particulates and gasses they produce.
In an age where we have developed computer models to replicate the behavior of some of the most complex phenomena on Earth, traffic has proven surprisingly resistant. The best models need vast processing power, requiring time on expensive supercomputers.
Dr Meead Saberi of the University of New South Wales and colleagues wondered if pre-existing models for the way infections spread, which have also been used to study the way ideas disperse through social media, might be re-purposed. In Nature Communications they report on their success in predicting traffic congestion in six major cities based on these models and infrastructure maps.
“Urban traffic often exhibits high spatial correlation in which links adjacent to a congested link are more likely to become congested,” the paper notes. Moreover, the authors point out, traffic jams tend to happen in different parts of a city around the same time as people rush to or from work, just like disease outbreaks.
When the work began epidemic modeling was obscure to most outside public health, but now when Saberi talks about congestion having an R0 he may get more recognition. “R0 is how fast something spreads,” he told IFLScience. “In this case how fast congestion spreads through a network.” The higher the R0, the more other streets get clogged when drivers try to avoid a known bottleneck. To their surprise, the authors found that at times when traffic was bad, each of the cities they studied had quite a similar R0, despite very different population densities and public transport systems.
A vaccine for traffic is likely to take a lot longer than one for COVID-19, but Saberi told IFLScience interim solutions are quite analogous. “We're trying to flatten the curve, encouraging people to drive earlier or later than the peak, or to mode switch by riding bicycles or catching public transport,” he said. “We hope our model can help us determine the optimum time to apply interventions,” achieving the maximum congestion reduction for the least disruption in people's lives.
The paper acknowledges one weakness, however. Infection models are based around the fact that, for most infectious diseases, people become immune to re-infection after recovery. “This clearly does not apply directly to traffic networks in which a link may recover and become congested again after a short period,” the authors note. Saberi told IFLScience one of the team's next priorities is to investigate models of spread for the rarer diseases where re-infection can occur.