On forecasting permafrost thawing
Studying the impact of climate change on linear infrastructure, especially northern infrastructure, is important for many reasons. One of them is that it’s very hard to monitor thousands of kilometres of infrastructure, so remote sensing gives us the best approach—not very good in terms of resolution, but at least we can observe.
When we talk about permafrost thawing, it’s not only about climate change. It’s really about the condition of the ground inside itself. For example, the types of soils we have the ice inclusion we have in those regions. So if we find that some regions are very vulnerable to climate change but the ground itself is okay, then we don’t have to worry about them. It can deal with climate warming without any adverse consequences.
If there’s no more permafrost from a purely engineering perspective, if infrastructure in an area is good, then we don’t need to spend millions of dollars just to see how we can preserve the frozen rock. Then we can focus on identifying the regions which will be vulnerable to climate change.
On understanding which regions are most vulnerable
There are some regions that may seem to be less vulnerable to the pressures of climate change, but the ground itself is problematic. In this study, we tried to identify the regions which are actually vulnerable to permafrost thawing and consequences of that.
We identified three different locations [Hudson Bay Railway, Mackenzie Northern Railway, and Inuvik-Tuktoyaktuk and the Dempster Highway] where we have critical infrastructure. These are the areas that were identified to be problematic in decision making and we have to rethink the way that we design infrastructure in those regions.
What was interesting in our framework is that there are some regions where the adverse effects of climate change started 10 years ago, but in five years, there won’t be a problem anymore. So, if you want to design a new structure in five years, those regions will be fine. But on the other hand, there are some regions which are not really affected by climate change yet but will be in the future. So when we’re designing infrastructure, maybe we’re not thinking about the consequences of climate warming on those specific regions, but through this study, we give a warning that says, ‘Just watch out. These regions may be fine now, but in five to 10 years, they could be problematic.’
On the challenges of studying permafrost
One of the biggest issues we have up north is that relying on the field monitoring data is difficult. Sensors, for example, after two to three years, often stop working. However, when you want to develop an AI model, you need to have a continuous time series of data. If you have an interruption, you can’t use the data sets for training purposes. And this is actually the biggest issue when we want to apply AI up north. Because we just recently started gathering data about two to three years ago, and it’s not enough in terms of the length of the time series. It’s not enough to let us use it for training purposes and to use it for forecasting.
These AI models are intended for use by stakeholders, those who work as asset management engineers or urban planners, municipalities, and scientists—those who deal with data and/or decision-making.
On future projects
What we published is the first step. We have a series of very interesting papers coming up with better resolution. But it was the first step to show that when we design linear infrastructure up north, the design should not be uniform. There are some regions that are very sensitive and vulnerable to the impact of climate change and others that are not as much.
Further, what we developed is a framework that shows us the vulnerability of infrastructure for now, not from the past. This is important because if, for example, permafrost thawing is occurring now, but in five or 10 years, there’s no more permafrost, then our vision to develop northern infrastructure should take into consideration that there won’t be permafrost anymore. We have to really focus on climate change adaptation.