In nordic countries with harsh winters like Canada, it’s sometimes hard for the public to realize the Earth is warming. That’s why we decided to do a story on how the coldest season is warming up too. We collected and analyzed snow ground data. The resulting publication is a project with four data visualization and stories of people working in industries impacted by the decreasing snow cover.
Winter is an important part of Canadian identity and culture. The Canadian public responded strongly to this story, published in both French and English (Canada is a bilingual country). This project is one of our most-read of the year.
This project also fills a gap in the academic literature. While doing our research, we noticed that many studies were focusing on snow precipitation, but almost none talked about the snow depth on the ground. On an everyday basis, it’s really hard for people to evaluate the variation in the quantity of snow falling from the sky. However, it’s very easy for them to notice the amount of it on the ground. Snow depth on the ground was a way better way to show the impact of global warming on Canadian winters, to Canadians.
This story managed to provide a rigorous analysis while vulgarizing the results in an interesting visual form at the same time.
Environment Canada is a federal agency managing more than thousands of weather stations in the country. To collect the data, we coded a web scraper in Node. For each weather station, the script checked if enough data was available for a long period of time (at least 40 years). When it was the case, the data was downloaded. We ended up with a file of 16 million rows.
We used Python to do our data analysis, with the Pandas library. We cleaned the data and did a regression analysis to find the trends. We shared all our results with one of the top Canadian expert on snow, Ross Brown, whom we also interviewed for the story.
What was the hardest part of this project?
For the temperature and precipitation data, Environment Canada provides easily downloadable files for the whole country. It’s not the case for the snow cover data that is scattered through thousands of different web pages. You need the technical skills to gather and standardize the data, which may explain why almost no experts analyzed it.
But you also need a very strong methodology with precise indicators to calculate the trends. We read academic articles to build our analysis and asked the Canadian Consortium for Regional Climatology and Adaptation to Climate Change to review it.
One of the biggest challenges was to choose indicators easily understandable by the general public while relevant for experts and to translate them visually.
What can others learn from this project?
While working on the data visualizations, we realized that we couldn’t explain all aspects of the story with just one chart, even animated. We needed several data visualizations to explain different things. We often see data-driven projects with one big main data visualizations, followed by a long text. It’s not always the best way to communicate our information. We hope other data reporters and designers will learn from this project that, often, several simpler data visualizations are better than a complex big one.