Skip to main content

How GIS and citizen science can unmask Canadian tornadoes

Canadians often think of tornadoes as a uniquely American occurrence but annually there are significantly more tornado appearances than are officially reported in Canada. In fact, an average of 60 tornadoes are recorded, but the actual number is likely closer to 230. Why is there a large margin of error in detecting tornadoes? And how can we identify tornadoes that have already happened and use this information to predict and make everyone safer? The Northern Tornadoes Project integrates GIS and citizen science to help answer these questions.

Tornadoes are a natural phenomenon that scare people and which Hollywood has exploited in films like Twister and the Wizard of Oz. But they have always been a curiosity to storm chasers and meteorologists. They struggle to know how and why tornadoes form, and more importantly, how to predict where one will happen. Incredibly, our current meteorological systems often cannot even detect a tornado. This is where the Northern Tornadoes Project (NTP) shines.

The NTP is a diverse partnership which includes: participants from Western University’s engineering department, Environment and Climate Change Canada (ECCC), and ImpactWx. They share one ambitious goal: to capture all tornado occurrences in Canada. As one of the project’s leaders, Professor Greg Kopp told Western News, they want “to find, assess, store data, and learn from each event. It’s a big goal and it’s a big country.”

To capture usually unrecorded tornado events, the NTP team developed a system that collects imagery data from ground, drone, aerial, and satellite resources. Additionally, they added a citizen science component that enables members of the public to report wind damage and/or tornado occurrences using Survey123 for ArcGIS. This is what the form looks like:

Reporting a tornado form

The data received from the public are verified by the researchers, and stored and accessed by the NTP team through their Open Data site, created by Liz Sutherland – a GIS specialist at Western Libraries. She will present the project at the Esri UK Scottish Conference as the site is an excellent example of how GIS services using Esri software and applied citizen science amplify efforts to track tornado events. This same approach can easily be applied to track other types of natural disasters.

What will the geospatial data be used for?

The retained data are intended for use by both members of the public and research communities. The public can view a map of verified tornado events from 2014 to 2019 and click on icons representing recorded tornadoes. Information about the tornado or an image displaying the damage of its track and site, are shown in a pop-up. For the 2019 season, five unknown occurrences have been validated from 500 observations and 1300 collected ground photos.

Fig. 2 – Map of Verified Tornado Events (2014 – 2019)

Anyone can also view information in three story maps representing Ontario, Québec, and Manitoba. Each story map displays specific geographic areas with a temporal comparison of before and after wind damage and geotagged images from the sites. Furthermore, anyone with ArcGIS Online/Enterprise credentials can view up-to-date tornado occurrences with pictures on their Operations Dashboard app.

The NTP is using the collected geospatial data to address these long-term goals: in order to save more Canadian lives, reduce property damage and costs, and increase knowledge of severe storm activities

  1. Continue enhancing the detection of tornadoes throughout Canada

  2. Augment severe and abnormal weather predictions

  3. Delve into the unprecedented consequences from climate change

Because the project is at its relative infancy, finding a rough estimate in how many lives are saved and amount of insurance savings is unknown; however, for sure in the long term more lives will be protected and tornado insurance costs will go down.

Looking more deeply into the collected data, NTP researchers hope to investigate why one supercell generates a tornado while another one, which may be in a similar environment and structure, does not. Understanding that, and using proper machine learning algorithms, the next generation of weather instruments may allow much improved prediction of when and where a tornado is about to occur and inform the nearby populations.