Weather Extremes and Urban Traffic
What do the results tell us?
The impact of disturbances in the transport sector caused by extreme weather events for a medium sized city tend to be modest. We find that the cost ranges from around €7.1 million to €32.5 million per year for Zurich and €20.3 million to €94.5 million for Switzerland. Here we have to note that these figures are based on 1 day disturbances or disruptions, and only relate to passenger transport. Therefore the real costs of disruptions due to extreme precipitation events can be significantly higher as one should add multi-day disruptions, extra logistic costs and costs of damage to infrastructure.
The impact of passenger information in case of extreme events, using integrated weather route information at travellers can reduce aggregate disruption costs with about one third. This means that providing transport information at low cost can significantly reduce the cost of extreme events.. We find that the most significant gains (about 70%) of the cost reductions result from allowing users to make changes to their daily travel and activity patterns, such as: leaving earlier or later from work or taking a day-of. Taking alternative routes and other transport modes are effective, but especially in the case of rerouting travellers may overreact to new traffic information.
What can we do with the results?
The case study outcomes can be used to better understand the impact of weather extremes on traffic.To some degree, the impacts for a medium sized city tend to be modest as is explained above. In the type of events we considered, with return periods of 10 years and higher, this is not particularly surprising. The total expected damage from a highly frequent event with relatively low damage however, may be similar to a low frequency event with a high damage. For example, over a period of 50 years, an event with a return period of 10 years causing damages around 20 million euros is (statistically) similar to a once in 50 year event with 90-100 million euros of damages.
Besides this, the events that we modelled also create extra logistic costs and will to some extent cause physical damage. These impacts could not be directly quantified in this case, but may add significantly to the cost. While not treated here, we can refer to the case of ‘Flooding' in London, which treats the damage recovery and economy wide impacts of extreme events.
As an operational adaptation measure, integrated weather-route information aimed at travellers at various stages of trip planning and implementation can be cost-effective, especially when users can be made aware of the disruption at the time of their ‘trip scheduling' phase. On-route information is generally less performant and may risk causing overreaction at the side of the travellers.
It is clear that optimal information provision would reduce the disruption cost. Development cost of such services may still be significant. It is therefore likely that such innovations would best pay off by developing them at a European scale (where the avoided costs are many times the Swiss avoided cost level. As such joint, i.e. multi-country, development may help to reduce the development and supply cost per country.
The results of this case-theme do not reach beyond 2030 and do not distinguish between RCP's and SSP scenarios. The MATSIM model is validated for the population projections and infrastructure capacity projections for the Canton Zürich up to 2030, but this does not match directly with SSP scenarios. Furthermore, while the climate change projections for the area produced by ToPDAd make predictions on overall precipitation, these are not useful to determine the possible occurrence of extreme precipitation events.
How are the results obtained?
On the basis of an agent-based microsimulation model (MATSIM), well validated for the Zürich Canton, a set-up was tested for different degrees of disturbances (reduced capacity/speed) and disruptions (temporary inaccessible links and areas). To test the impact of weather information as an operational adaptation strategy, we assumed that travellers had access to different levels of travel information. In the optimal case, they were fully aware of the disturbance (or disruption) on the network and were able to optimise their travel behaviour and optimally replan their activities and travel patterns. In a series of alternative cases, we manipulated agent behaviour to simulate non-optimal behaviour based on partial on-route information supplied to a fraction of the agents in the model.
This larger set of simulations was later matched with various real-world precipitation events with current return times of 2, 5 and 10 years (based on data of Meteo Swiss and the longer term GIS data supplied by the ToPDad project). Subsequently, for the three levels of return times the consequences of disturbances for travel behaviour and overall utility (value) per traveller are calculated. Finally, the results for the Zürich Canton are scaled up by projecting the same responsiveness by mode for 7 other urban regions while correcting for total travel volume and modal split of each Swiss urban region.
What are the broader applications?
Tentative upscaling of the Swiss figures would imply, based on Switzerland's GDP as compared to the GDP of EU28 (~2%) that annual time cost in the EU due to extreme precipitation induced disruption can amount to € 1 billion to € 4.7 billion. This ranges above earlier damage estimates of € 0.5 ~ 1.0 billion in the EWENT study (Nokkala et al 2012).This means that the cost on European scale is much more significant than for individual countries, which would also justify a European based research strategy towards improving prediction methods.
Though our results focus on urban passenger transport, similar results as those obtained in this case could be obtained for operational adaptation measures for other sectors. Weather prediction may be essential as a first line in increasing resilience of the transport sector. In the road sector, developers of GPS navigation software have become aware of the potential of including weather related phenomena. Similarly stakeholders in rail, air and inland waterway transport indicate an increased interest in the integration of weather prediction as an operational measure for transport. On longer term this may not only effect transport rerouting (which has a limited impact on overall cost), but also scheduling and planning of transport trips, as well as investments in critical infrastructure.
Key Messages and Conclusions
The impact of passenger information in case of extreme events, using integrated weather route information at travellers can reduce aggregate disruption costs with about one third. Given the relatively low implementation cost of these options, they are cost-effective ways to increase the resiliency of the transport sector.
Rescheduling activities is more effective at reducing overall cost of extreme events
We group adaptive behaviour to extreme events in two large groups:
The ‘Best' response is a combination of both strategies. We compared this response with a situation of ‘Full' on-route travel informedness, but without the ability to reschedule activities. From this we conclude that rescheduling alone is responsible for about 70% of the total potential cost reduction of travel information.
Cost of extreme events compared to best (adapted) response for different levels of disturbance.
Providing full information to the whole population is not always optimal
Supplying on-route information to travellers is beneficial until the moment when about half of them are informed. Informing more may no longer have an effect, or even counteract possible benefits. The reason is simply that travellers may overreact on the new information, such that the alternative route(s) also becomes overloaded. This is particularly the case when the information provision is very efficient, but the information available is not sufficiently up-to-date. This is more likely when the disturbance is relatively minor. From a cost-benefit standpoint, it may therefore not be necessary to fully inform all travellers, but aim at a moderate well-informed audience.
(Click on chart image to explore the data)
Impact of informedness on travellers for different levels of disturbance.