Proactive Monitoring of Thawing Permafrost on Infrastructure in Northern Canada

Under the Canadian Space Agency's smartEarth initiative, KorrAI Technologies collected and analysed precise satellite radar measurements to monitor ground movement in permafrost regions where remote communities rely on roads and airports for critical access and supplies, before infrastructure suffers severe damage. Using data from CSA’s RADARSAT Constellation Mission (RCM) and ESA's Sentinel-1, the project area covered three airports and 250 kilometres of roadways every 12 days, tracking millimetre-scale ground movement across 12,000 square kilometres of permafrost terrain.

A proprietary processing technique, UrbanSAR, identifies stable natural radar reflectors and tracks their movement over time. A companion algorithm grades the severity of each detected motion, sorting it by how far it deviates from expected behaviour. Infrastructure operators get colour-coded maps and time-series dashboards rather than raw radar data.

The practical outcome is threefold: In the Inuvik area, built-up land is subsiding at 7.2 mm per year, with clear seasonal oscillation tied to the permafrost freeze-thaw cycle; at airports, operators see the terrain subsidence developing weeks or months earlier instead of finding out a runway has settled beyond safe limits after flights are already grounded; and for remote and indigenous communities where road transport is a lifeline, early indications of terrain subsidence and timely maintenance rather than more costly after-the-fact repairs are a matter of safety, access, and basic continuity of life.

Canada's Arctic is warming far faster than the global average, so infrastructure designed for "permanent" frost is failing within its intended life. The project focused on two corridors where these risks are well documented.

The Inuvik-Tuktoyaktuk Highway, in the western Arctic, connects remote Inuvialuit communities to the regional hub of Inuvik. Along this route, the Northwest Territories (NWT) Geological Survey has been running a field experiment, compacting snow at six sites each winter to release trapped heat and slow the thaw beneath the road.

The second corridor, Highway 3 between Behchoko and Yellowknife, is the primary all-weather road from southern Canada to the NWT capital. Despite major upgrades between 1999 and 2006, this highway already shows differential settlements, pavement cracking, and embankment instability, with maintenance costs running well above projections.

The organisations responsible for safety and continuity face a persistent information gap. Point inspections and surveys are accurate when used, but they do not scale across vast, remote networks. Crews are expensive, repeat visits take time, and visible damage often appears too late.

Hence, decision-makers need monitoring that is continuous, all-season, and precise enough to flag early warning trends, not just confirm failure after the fact.

The satellite-based technique used here, Interferometric Synthetic Aperture Radar (InSAR), works by comparing radar images of the same patch of ground taken on different dates. Small differences in the returned signal reveal whether the surface has moved between passes, and by how much. The precision is striking: shifts of a few millimetres are detectable.

Under the Canadian Space Agency's smartEarth initiative, which fosters a smart use of satellite data to develop applicable solutions to key challenges on Earth, KorrAI Technologies used its Open Ground Motion Service (OGMS) platform to turn raw, satellite-based measurement capability into information people can act on.

Deformation velocity maps show the speed of ground movement at every measured point across the monitoring area. In the Inuvik corridor, the southwestern area around Inuvik shows subsidence of 9 to 45 mm per year. Along Highway 3, subsidence in the northern and central sections reaches 8 to 53 mm per year, while the eastern portion shows uplift of up to 62 mm per year. Patterns appear that nobody would spot from the ground, for example a slow, steady drop along a road embankment, or an accelerating pocket of settlement beneath a runway apron.

A hazard-grading algorithm sorts every measurement point based on both the statistical significance of movement and how far its behaviour deviates from expected patterns. The output highlights only reliable deformation signals, resulting in a simple, actionable map that allows decision-makers to quickly identify areas where intervention is needed.

Displacement time series track individual assets, such as a runway threshold, a bridge abutment, a section of embankment, over months and years. These show whether movement is accelerating, seasonal, or stagnant, and indicate safety thresholds that trigger maintenance before a problem turns into a closure.

Moreover, the time-series data revealed something the velocity maps alone could not: seasonality. Warmer months produce upward peaks as the active layer expands; winter months produce dips as the ground contracts and refreezes. In Inuvik's built-up areas, the net trend beneath the seasonal oscillation is a subsidence rate of 7.2 mm per year, based on 1,068 radar reflector points over two years.

At the heart of the processing is UrbanSAR, a proprietary technique that identifies stable natural radar reflectors (exposed rock, gravel pads, infrastructure surfaces) and tracks their motion over time. It works well in sparsely vegetated terrain of Canada’s North that lacks the dense built-up surfaces typical of conventional InSAR applications.

By comparing the InSAR displacement measurements with land cover maps and surficial geology maps, the project also identifies surface types that are particularly vulnerable. In the Inuvik corridor, built-up land and fine-grained surficial materials showed the strongest thaw-related ground movement, aligning with published research and earlier Mackenzie Valley borehole observations.

The primary data source is ESA's Sentinel-1 C-band SAR satellite that can image the Earth's surface regardless of cloud cover at regular intervals. The RADARSAT Constellation Mission (RCM), operated by the Canadian Space Agency, adds further radar acquisitions and flexible imaging modes tailored to Canadian priorities. Between the two missions, radar image time series of the study area can be updated roughly every twelve days.

These radar observations were combined with two auxiliary data layers. The Canada Centre for Remote Sensing's 30-metre land cover map (2020, Landsat-derived, 86.9% accuracy) offered further detail regarding the distribution of forest, sparse vegetation, built-up land, bare land, and wetlands. Surficial geology maps from the Geological Survey of Canada (1:125,000 and 1:250,000 scale) provided critical subsurface information. Linking displacement to these layers turned raw motion data into an understanding of why certain areas move the way they do.

Absolute accuracy comes from calibrating against GNSS reference stations, anchoring relative satellite measurements to a fixed geodetic frame for consistent multi-year comparisons. Cloud computing automates ingestion, point detection, noise filtering, hazard grading, and product delivery.

The system can also utilize over two decades of archived RADARSAT-1 data to extend trends before the current monitoring window. Next, KorrAI plans to ingest Sentinel-1C and NISAR data; NISAR’s L-band penetrates deeper into vegetation and soil, improving performance where C-band can be limited.

The OGMS platform now covers 12,000 square kilometres of Northern Canada, including three airports and 250 kilometres of roadways, with ground-movement assessments delivered every twelve days at millimetre precision. It empowers stakeholders to shift from reactive to proactive risk assessment and infrastructure maintenance.

For communities, early detection of instability helps prevent catastrophic damage and associated costs. Timely alerts reduce risk to residents and workers relying on vulnerable infrastructure.

For infrastructure operators and engineers, emergency repair costs can vastly exceed those of scheduled work. With satellite-based information on which assets are moving fastest, operators can direct budgets to the right places at the right time. The project also showed that satellite observations can be turned into a planning tool for field operations.

For government decision-makers, the data fills a critical gap. Territorial and federal agencies can now see which portions of corridors and assets are under the most stress at a scale that would be physically impossible with ground crews alone. That changes the quality of decisions around where to invest, where to redesign, and where to plan alternatives.

The ground deformation capability developed through OGMS has since been integrated into TRAIL, KorrAI's broader desktop study platform. Operators can now access hazard grades and displacement data inside a workspace that also pulls in geology, site history, drainage risk, and natural catastrophe exposure. That consolidation matters because permafrost thaw rarely acts alone; water, geology, and construction history all shape how a site responds to warming. Having those layers in one place means the people making maintenance and investment decisions see the full picture, not just the ground deformation data in isolation.

The approach is inherently scalable. The same satellites that watch Northern Canada monitor every landmass on Earth.

The work supports UN Sustainable Development Goals 9 (Industry, Innovation, and Infrastructure), 11 (Sustainable Cities and Communities), and 13 (Climate Action).