By Helen Dacre, Professor in the Department of Meteorology
When severe weather is forecast in the UK, a frequently asked question arises: Do people actually change their plans in response to the warnings issued?
Weather warnings are designed to help us prepare and stay safe but measuring whether they influence behaviour has always been difficult. Surveys can tell us what people say they would do, but not necessarily what they actually do.
By analysing anonymised mobile phone data from over 100 million road journeys, our study reveals how people responded to snow warnings issued by the Met Office on 31 January 2019, and the findings are both encouraging and surprising.
The Met Office’s National Severe Weather Warning Service (NSWWS) has issued impact-based warnings since 2011, covering hazards such as snow, wind and rain. These warnings are designed not just to forecast weather, but to communicate risk and prompt action. Yet understanding their real-world impact has been challenging.
This study uses anonymised mobility data from the O2 network, which captures patterns of movement across England and Wales. Unlike surveys, this approach shows how people actually travelled, where they went, how far, and when.
The focus was a winter storm on 31 January 2019, when yellow and amber warnings for snow and ice were issued across parts of the UK. Heavy snow was forecast to hit southwest England and south Wales later in the day.
Fewer journeys overall
Across England and Wales, total road travel dropped by about 5% compared with a typical winter Thursday. That equates to 5.7 million fewer trips.
But the timing of these changes is key.
- Early morning (before snow fell): Total travel levels were broadly similar to normal.
- Midday to evening: Total travel dropped significantly, by over 6% overall, with some areas seeing reductions of up to 25%.
- Where snow fell: The largest decreases were seen in regions already experiencing wintry conditions.
At first glance, this suggests that people reacted mainly to the snow itself, rather than the warnings. But a deeper look reveals a more nuanced story.
A striking drop in commuting before the snow
One of the clearest signals of behaviour change comes from work-related travel. Between 7am and 10am, before snow began falling in most areas, commuting trips fell by 20% nationwide. This was offset by an increase in non-work-related trips.
This is a substantial shift. It indicates that many people altered their plans in advance of the weather, likely in response to forecasts and warnings. Interestingly, this reduction wasn’t limited to areas under official warnings. Commuting dropped across much of England and Wales. This suggests that:
- People may respond to national messaging rather than localised warnings
- Employers and institutions (e.g. schools) may play a key role in driving decisions
- Awareness of regional differences in risk may be limited
Short trips up, long trips down
Another clear behavioural change lies in how far people travelled. Long journeys decreased by around 14% in some regions and short, local trips increased by about 5%. In some areas, short trips rose even before the snow arrived, particularly within warning zones.
Figure: Relative difference in the number of long road trips between 0700-1000 UTC on 31 January 209 compared to a baseline average. Blue colours indicate a decrease, and red colours indicate an increase. Yellow and black contours indicate the position of the rain/snow on the occluded front at the start and end of each period. Dacre et al. (2026)
This pattern suggests a shift in priorities: People avoided longer, potentially risky journeys. At the same time, they made local trips, perhaps to stock up on food or essential. In other words, people weren’t simply staying home, they were adapting their behaviour.
Heading home early
The data also hints at another response: people cutting their day short. Between late morning and early afternoon, there was a noticeable increase in journeys away from workplaces in some areas before the snow reached them. This aligns with the timing of an upgraded amber warning issued mid-morning, which highlighted the risk of travel disruption and stranded vehicles. The implication is that warnings may not just affect whether people travel, but when they travel.
Fewer reported accidents: a surprising result
Snow typically increases the risk of road accidents. But on this day, the number of police-reported collisions causing injury was 10% lower than average. The most likely explanation is reduced exposure. Fewer vehicles on the road means fewer opportunities for collisions and people avoiding long trips may reduce the likelihood of serious accidents
It’s also possible that while minor incidents increased, they weren’t reported to police. Either way, the findings suggest that behavioural changes may help reduce risk at a population level.
A communication challenge
One of the most intriguing findings is that behavioural changes, especially reductions in commuting, were not confined to warning areas. This raises an important question: are people responding to the detail of weather warnings, or to a broader perception of risk? If people outside warning zones are changing their behaviour just as much as those inside them, it may indicate:
- Limited understanding of geographic differences in risk
- Over-generalisation of warnings
- Or strong amplification of messages through media and social networks
For forecasters, this highlights a potential challenge: how to communicate localised risk more clearly, without diluting the overall message.
Why this matters
Early warning systems are central to disaster risk reduction, and initiatives like the UN’s “Early Warnings for All” are pushing for global access to effective systems.
But warnings only work if they lead to action. This study provides large-scale evidence that people do respond to weather warnings, often before hazards occur. They adapt their behaviour in specific ways (avoiding long trips, reducing commuting). These changes may reduce risk, including road accidents. At the same time, it shows that communication can be improved, particularly in helping people understand where risks are greatest.
Looking ahead
This analysis focuses on a pre-pandemic event. With the rise of hybrid working, future responses to weather warnings may look quite different. There is also scope to explore other hazards, such as heavy rain or wind, different warning levels and how people interpret them, and the combination of mobility data with surveys, to better understand why people act. Because ultimately, behaviour during extreme weather is complex. It is shaped not just by forecasts, but by trust, perception, personal circumstances, and practical constraints.
In conclusion
Weather warnings do influence behaviour, but not always in the ways we might expect. Understanding those responses is key to making warnings more effective, targeted, and ultimately, lifesaving.
Discussion about this post