UBC scientists generate weather models which are then used by local transportation authorities. (Istockphoto)

How scientists and planners work together to keep winter roads safe

Tucked away at UBC’s Vancouver campus, atmospheric scientists armed with computer clusters are helping smooth out British Columbian’s winter commute.

By Timothy Chui and Rosie Howard, UBC Department of Earth, Ocean and Atmospheric Sciences

Few things bring Vancouverites together more quickly than collectively grumbling over a centimetre or two of snow. It’s an occurrence that that can bring the city — and its vehicles — to a screeching halt. Further inland, winter weather can strand travellers on highways for hours or result in deadly accidents.

Weather can make or break travellers’ plans, so being able to predict the timing and impact of weather is important for commuter and transportation planners. However, weather forecasting is challenging, and lousy forecasts come up frequently as ‘discussion topics’ (read: complaints), second only to bad weather itself! At UBC, the Weather Forecast Research Team led by atmospheric scientist Roland Stull, is dedicated to improving forecasts.

Because the atmosphere is such a complicated system, it is challenging to make even simple-seeming predictions — like how cold tomorrow will be — using pen-and-paper. Today, we use computer models and large computer clusters to do the number crunching. It’s a process called numerical weather prediction.

Numerical weather prediction involves solving a complex set of equations on a grid that represents sections of the atmosphere. Some physical processes, like the collision of rain droplets in a thunderstorm, can’t be seen on the scale of the grid because they are too small. So we estimate how these processes affect weather based on prior experiments or observations, called (physics) parameterizations.

JMA Numerical Weather Prediction. We use similar models at UBC to predict the weather. (JMA)

Once the models finish their forecast, we can remove some sources of error by looking at how well our models have predicted the weather in the past. And instead of relying on a single model’s results, we can combine several models to form an ensemble forecast. Ensemble forecasts are typically more accurate than individual forecasts. They also provide likelihoods of different scenarios — like the probability the temperature will drop below freezing — to help in decision-making.

Two transportation organizations use our winter forecasts — Coast Mountain Bus Company (you probably know them as TransLink) and Caribou Road Services, in northern BC. Our team creates high-resolution weather forecasts for TransLink and maintains several automated weather stations to inform them about conditions in real time. This helps TransLink decide when to start salting roads, defrosting trolley-bus wires, planning alternate bus routes, or increasing service during snowy weather. In the north, Caribou Road Services maintains highways and roads and uses our forecasts to determine when to deploy snowplows and salting equipment.

UBC during a snowy day. (Jamil Rhajiak, UBC Communications & Marketing)

After the terrible winter of 2017/2018 — which resulted in buses having a lot of trouble climbing Burnaby Mountain to SFU — CMBC asked us to install a weather station at the top of Burnaby Mountain. However, this was no ordinary weather station. Because it is so easy for ice and snow to accumulate and cause traffic issues on the mountain we equipped this station with the ability to provide freezing level forecasts. When the next snowstorm hits Metro Vancouver, our forecasts will provide guidance the CMBC’s bus schedules and contingency plans. If they see that the freezing level hit at the top of the mountain, they’ll be ready to take action.

A couple of decades ago, forecasters could predict about three days of weather in advance with a good degree of accuracy. But better models and technological improvements have helped extend that timespan to five days and sometimes even a week.

Bus going down Main Street in Vancouver. (Richard Eriksson, Flickr)


We’re continuing to find ways to improve forecasting — from leveraging cloud-computing facilities for added computer resources to testing different types of atmospheric grids. We’re even experimenting with machine-learning to automatically remove errors and create ensemble forecasts. Hydro and wind-power companies, transportation organizations, and government health agencies concerned with forest-fire smoke can then use our forecasts to plan for — and respond to — severe weather events.

With ever-increasing computer power, and more sources of data from surface weather stations and satellites than ever before, weather forecasting continues to evolve. These advancements can be used to help governments, industry and the public anticipate bad weather and react in time, saving money and lives.

Hopefully one day, as you’re waiting for the bus in a snowstorm, you’ll be able to say you totally saw that coming.

Follow UBC Science on Twitter, Facebook, Instagram and subscribe to our newsletter.