Sensors as a risk management tool

Located on the island of Spitsbergen at 78° North, Longyearbyen serves as the administrative center of the Svalbard archipelago.

You can explore this map - and all other maps and three-dimensional scenes in this StoryMap - using the interactive controls on the map or scene.

Longyearbyen's location in the glacially-sculpted Longyeardalen valley at the foot of steep mountain slopes exposes the community to a variety of natural hazards.

Here, a view south over Longyearbyen shows the town's proximity to the mountain slopes, small glaciers at the head of the valley, and Adventfjorden (the Advent fjord) in the foreground.

Snow covers the region for up to nine months of the year at sea level, exposing Longyearbyen's residents and infrastructure to snow avalanche hazards during the long Arctic winters.

Historically, snow avalanches have regularly occurred on the steep slopes of the Gruvefjellet, Sukkertoppen, and Platåberget mountains.

This model frames climate-related risk as the manner in which climate change impacts the type, frequency, and magnitude of the natural hazards which occur in a location. The natural hazards, in turn, have rippling effects on society's infrastructure, inhabitants, and societal functions. In the respect, we can view the effects of snow avalanches as climate-related risks, where climatic changes can modify the avalanche characteristics and timing, and avalanches in turn impact socio-technical systems (e.g. infrastructure, inhabitants, and societal functions).

The risk governance process serves as a structured approach which helps societies make decisions about risks. Risk governance involves the steps of Framing, Data collection, Sensemaking, Decision-making, and Risk treatment, with each step informed by communication and collaboration between involved actors.

Combined, the model offers a framework for addressing climate-related risk from both short-term and long-term time perspectives.

The sensor work we have conducted in the ARCT-RISK project focuses primarily on the data collection phase of risk governance, but also involves sensemaking (e.g. understanding the collected data) and incorporating this information into the decision-making process.

The initial impetus for this work began when, following the 2015 and 2017 avalanche events, the University Centre in Svalbard (UNIS) received funding to install three snow monitoring stations in Longyeardalen for the 2017/2018 winter season. These sensors were intended to support snow research and site-specific avalanche forecasting services in Longyearbyen. We therefore installed these stations in know avalanche release areas to monitor snowpack condition throughout the snow season. This photo shows one of these stations installed on Sukkertoppen's western aspect (the "Lia" station) in the release area of the 2015 avalanche.

Concurrently, Telenor Svalbard decided to start exploring the possibilities of emerging Low Power Wide Area Networking (LPWAN) technologies with the goal of promoting a variety of new use cases in remote sensing and reducing data transmission costs with these low-power radio transmissions. Telenor initiated contact with the UNIS snow station project and agreed to install a low-cost prototype ultrasonic snow depth sensor onto the same mast as one of the UNIS stations to test data transmission via LPWAN, as shown in this photo.

Ultrasonic sensors have traditionally provided the majority of snow depth measurements to avalanche forecasting operations. Ultrasonic sensors measure distance based on the time-of-flight principle for a sound wave emitted from the device. This means the sensor emits a sound wave and accurately clocks the time it takes for the sound wave to reflect off the surface and return to the sensor. As the snow depth increases, the distance the sound wave must travel decreases and the time-of-flight therefore decreases as well.

Ultrasonic sensors have a relatively small measurement footprint and are most suitable for obtaining snow depth measurements for a single point-scale location.

Contrastingly, terrestrial laser scanning has historically been used in snow and avalanche research contexts and has only more recently begun to emerge as an operational tool. As an active remote sensing technology, ground-based LiDAR, or TLS, calculates the distance to a target surface (in this graphic, Gruvefjellet's western aspect serves as the target surface) by measuring the time-of-flight for a pulse of light emitted by the instrument to travel to the target surface, reflect off the surface, and travel back to the TLS. The scanner rotates through a user-defined scan window, scanning a swath of the surface and recording the position of millions of points on the surface in a three-dimensional point cloud which represents the scanned surface. We then use these point cloud data to create a digital elevation model (DEM) of the slope. The elevation difference between a scan or DEM of an area or slope with snow and the slope without snow represents the snow depth at each point on the surface.

Where an ultrasonic sensor measures snow depth at a single point, TLS technologies allow for the measurement of snow depths across an entire slope scale - or "spatially-distributed snow depths."

To better understand what the TLS "sees", we can compare photos from the scanning location at times without snow (left) and with snow (right). By calculating the difference between the DEMs from the scans taken at each time, we can determine the snow depth at each point on the surface.

Certain locations in Longyeardalen develop recognizable patterns of avalanche activity based on the area's specific topography and common meteorological patterns. On Gruvefjellet, the primary avalanche hazard results from the cornices which build up on the edge of the plateau and slab avalanches which release on the slope below.

Similarly, cornices and slab avalanches releasing in the avalanche paths on Platåberget threaten the Huset building and Vei 300.

Avalanche hazard on Sukkertoppen, as exemplified by the destructive avalanches in 2015 and 2017, primarily stems for easterly winds depositing wind-drifted snow in the more open avalanche paths on Sukkertoppen's western aspect.

In this concept, the pictured communication device transmits the data acquired by the ultrasonic sensor to a cloud-based data management system. The communication device transmits data via Telenor's Narrow Band Internet of Things (NB IoT) technology on their Low Power Wide Area Network (LPWAN). This technology allows data transmission anywhere with cellular coverage and significantly reduces the stations' power requirements.

From the data management system, users can access the data in real-time via a website data portal. This allows avalanche forecasters, snow observers, and other risk managers to view snow depth changes and make decisions based on the most up-to-date data.

...we added stations for better spatial coverage and to provide redundancy in the event of sensor failure. By the 2021/2022 season, we therefore had paired, redundant sensors installed on each of the mountain slopes of primary concern for avalanche forecasting in Longyearbyen. The relatively low sensor cost allowed us to place multiple sensors directly in avalanche paths, where possible sensor destruction might prevent placement of more expensive sensors.

We removed the sensors on Sukkertoppen prior to the completion of the avalanche protection fences in the summer of 2022. However, for the 2022/2023 winter season we added a sensor in one of the slushflow paths on the northern aspect of Platåberget:

While we did not have TLS data for this area, we used our knowledge of snow depths in this location from manual snow surveys (during the slushflow forecasting season) to choose this sensor's location.

On Gruvefjellet the current pair of sensors (Nybyen North and Nybyen South) replace the original single Nybyen sensor.

Sensor relocation in this location occurred primarily to improve access to the two sensors, as the original sensor could not be accessed throughout the majority of the winter due to avalanche hazard.

Now, Nybyen South can be accessed during periods of low avalanche hazard via the indicated route up under the cliffs above Nybyen.

However, data acquired in the new station location, while representative of the middle of the avalanche paths, is less representative of the upper release areas directly under the cornices. We therefore selected the location of Nybyen North in an area more representative of these upper release areas...

These graphs display the wind and precipitation conditions observed at the Svalbard Airport ("Lufthavn" in Norwegian) located just west of Longyearbyen from February 5th - February 10th, 2023.

The data show the period began with relatively strong easterly winds on February 5th. Wind speeds decreased throughout evening of the 5th, and precipitation began to fall around midday on the 6th (red arrow on the precipitation graph). A few hours later, the winds quickly shifted from an easterly direction to a westerly direction (red arrow on the wind direction graph). Westerly winds continued to increase in speed (red arrow on the wind speed graph) the evening of the 6th and remained elevated throughout the day on 7th.

In general, we expect the west-facing aspect of Gruvefjellet to accumulate snow when easterly winds place this slope in the lee. Similarly, we expect the east-facing aspect of Platåberget to accumulate snow under westerly winds.

We see these patterns reflected in the snow depths measured by the snow sensors on Platåberget (Huset High and Huset Low) and Gruvefjellet (Nybyen North and Nybyen South). Snow depths begin to increase midday on February 6th as it began to snow (black arrows). Later that evening when the winds abruptly switched from easterly to westerly and gained strength, however, snow depths on Platåberget began to increase as the wind transported snow into these slopes (red arrows). At the same time, snow depths on Gruvefjellet decreased dramatically as the westerly winds eroded snow from these west-facing, windward slopes.

Snow continued to accumulate under both the Huset High and Huset Low sensors throughout the day on February 7th ( dashed lines indicate missing data from the sensors) such that when the wind died on February 8th, at least 25 - 50 cm of snow had accumulated on Platåberget, while Gruvefjellet lost a similar amount of snow.

Compiled and available in real-time, these data can help avalanche forecasters and other hazard managers develop a "picture" of the snow situation on the slopes above Longyearbyen.