Using dogs to detect oil spills hidden in snow and ice - A new tool to detect oil in Arctic environments

All countermeasures regarding oil spilled in the environment assume that the oil spill itself is located and the polluted site is known. When working with oil spills in open water, a wide variety of remote sensing tools are available not only to locate the oil, but also to monitor the oil trajectory, the distribution of the oil and the film thickness of the oil on the sea surface. This is possible through a combination of sensors operated from satellites, airplanes, helicopters or boats. A comprehensive review of these methods is given by Fingas and Brown 2000 and Dickens et al., 2010.

Today, no proven operational system exists for detecting oil spills covered by snow and/or ice or hidden under beach sediments. The technologies with the highest potential for use with oil in ice are the fluorescence detector, ground penetration radar and ethane detectors. However, a considerable amount of R&D work and field testing is needed to verify the operational usefulness of these methods in screening large areas to detect hidden oil spills. Another common feature with these methods is the high level of technology present in this equipment. This might complicate the use of these methods in remote and often challenging Arctic areas, where external technical and logistical support and repair represent a major challenge.

One alternative to these hi-tech approaches could be to utilize the large potential in specially trained dogs to detect oil spills not visible to the naked eye or remote sensing detectors. It has long been known that dogs’ ability to detect different odors is exceptional, and this ability has been used for many purposes. Specially trained dogs are used today for related purposes such as searching for bombs, food, currency, avalanche, building collapse, drugs (Fält, 1997, USDA 2012), or for missing children (Buvik, 2003). These dogs are also used for pro-active searches and are capable of searching buildings or vehicles for drugs, or bombs, or for use in passive person scanning at airports. Mine detecting dogs have shown that dogs can work under harsh conditions and deliver reliable results. However, the methodology in which the dogs are trained and the quality of the training has a strong influence on the dog’s work performance. It is also known that dogs have been used within the petrochemical sector for detecting gas leakages in refineries and onshore pipelines (Canadian Energy Pipeline Association, 2012 and K9 Pipeline leak detection). Dogs have also been used to detect pollutants such as polyclorinatedbiphenyls (PCBs) and polyaromatic hydrocarbons (PAHs) in construction sites or old buildings, for example.

Based on the lack of reliable options for detecting oil hidden in snow and ice, SINTEF and the Trondheim Dog Training Institute initiated a cooperative study in 2006. Initial tests performed during 2007 in a laboratory environment and outdoors in Trondheim showed that dogs are able to detect and identify the smell of oil, including both weathered crude and bunker fuels. Outdoor tests in the Trondheim area have also proved that the dogs can detect the smell of oil, and can find point sources at an outdoor temperature down to −5°C. This initial work formed phase I of this project (Brandvik and Buvik. 2007).

The objective for the next phase of the work performed by SINTEF and the Trondheim Dog Training Institute was to develop a new and innovative method to detect oil spills hidden in snow, ice or beach sediments by using specially trained dogs. Extensive laboratory and field work was performed to verify the dog’s capabilities to survey beaches for oil spills (Buvik and Brandvik, 2009).

Later, the work focusing on oil detection in sediments was followed up by field trials performed in the US funded by the American Petroleum Institute (API). Owens et al. have performed several studies to evaluate current practises using oil detection dogs (API, 2013a), developed a field guide for using oil detection dogs (API, 2013b) and have formulated recommendations for further use of detection dogs (API, 2014). Owens et al., have also performed a field verification test for detecting oil spills in sediments in 2016 (API, 2016).

The work described in this paper was focused on using dogs to detect oil spills hidden in snow and ice on Svalbard (78°N). The main objective was to search larger arctic areas and indicate directions to and locations of oil spills covered in oil and ice. A possible scenario would be to detect “patches” of oil under ice or covered in snow, which imply searching over relatively large areas (multiple kilometers).

The team travelling to Svalbard consisted of three dogs and their handlers, all of whom had been able to meet the selection criteria for this Arctic field verification. This selection was based on work performance as well as plasticity to environmental changes both for handler and dog. The project group consisted of a total of five handlers and seven dogs. The Svalbard team consisted of (see also Figure 1):

• Responsible dog trainer Turid Buvik (Trondheim Dog Training Institute) and 3.5-year-old Border Collie “Jippi”. Jippi, in addition to oil detection, was also trained for detecting PAH and PCB, and has also participating in a research project using dogs to detect human lung cancer. Jippi was also a certified sheep herding dog.

• Dog Trainer Reidun Mangrud (Trondheim Dog Training Institute) and 2-year-old Border Collie “Blues”. In addition to detecting oil, Blues is also a certified search and rescue dog.

• Project Manager Per Johan Brandvik (SINTEF) and “Tara”, a 3 year old Dachshund. Tara was trained for detecting oil in this project, and was also used for deer and fox hunting.

The ability to detect oil is only a small part of the skills needed by the dogs in mapping oil hidden in snow or ice. Both the dogs and trainers needed to be able to handle the challenging climate and logistics in Arctic areas, which were vital elements for our field training on Svalbard (78°N). Prior to the excursion to Svalbard, both the dogs and the handlers performed outdoor activities in a rough winter climate over multiple days, such as skiing, search & rescue and hunting.

The dogs must often be transported to the search area over large distances. In our case, the distance from Trondheim to Svalbard was more than 1600 km. It was important to transport the dogs to the site in a safe and effective manner. This means that the working capacity of the dogs should not be reduced due to stress, the risk of any harm to the dogs should be minimized, and in our case this meant that the dogs and their handlers traveled together for the whole journey. Helicopters are an option in some cases, but due to the long distance in this case, large and small aircraft were used together with crates on snowmobile sledges.

We obtained permission from both the Norwegian Civil Aviation Authority and the airline company SAS (Scandinavian Airline System) to transport the dogs, as service dogs, in the cabin of the plane (Boeing 373). This was done to avoid potential unnecessary stress that could be caused by freighting the dogs in crates as cargo without supervision.

On Svalbard the dogs were transported in dog crates by a small airplane (Dornier 220), but in the cabin and with good contact with the handlers. On site in Svea the dogs were strapped in crates on scooter sledges (see Figure 2 and Figure 3). When dogs were working outside in cold and windy conditions for several hours, even furry dogs needed an insulated crate and some kind of warming suit to keep warm in-between working sessions. Figure 2 shows how the dogs were kept warm between the working sessions and during transport. The snow scooters used were new vehicles with modern four-stroke engines to minimize exhaust fumes, which can be irritating and possibly harmful to both dogs and handlers.

The wind is an important factor in this approach. The dogs were mostly used to search upwind, but experienced dogs can also search downwind and crosswind. During the field work at Svalbard, we experienced unstable and changing wind conditions due to a rapidly passing low pressure front.

Wind speed and direction for the actual period are given in Figure 4. Early in the period the wind was gusting to more than 10 m/s, but later changed direction completely over a period of a few hours due to a passing low pressure front. Several times the dogs gave the search team surprising directions to the training oil spills. However, after checking the weak and variable wind, the directions indicated by the dogs proved to be correct. One of the biggest issues for this Arctic training was to see how the teams (dogs and trainers) managed to operate in the, even for Norwegian conditions, harsh and cold climate at 78° North. The temperature, in combination with strong winds, made the effective temperature below −40°C. The temperatures during the field work period are shown in the Figure 5.

The experimental oil spills were placed on the ice in van Mijien Fjord on Svalbard one week prior to the arrival of the dogs. The spills consisted of one large 10 m² oil spill (400 liters, see Figure 6), plus 16 smaller oil spills (400 ml). The smaller spills consisted of oil released into a hole in the ice (0.5 meter deep) and then covered with ice and snow (Figure 6). The 10 m² oil spill was used to simulate a larger spill giving a significant larger vapor signature on the fjord ice.

All the small samples were tagged with a small cord wire as shown in Figure 7. The cords were very discrete and none of these cords were visually detected by the teams (dog/handler) during the training period. The cords offered no assistance in detecting the oil spills due to the visual detection of the cord wires by either the dogs or trainers. However, the cords together with the GPS coordinates were used to find the locations for removing the oil at the end of the field work. The oil spills were distributed in the inner part of the van Mijen Fjord outside SINTEFs Arctic field research station at Svea on Svalbard, see Figure 6 and Figure 7.

The dogs were equipped with 2 different GPS positioning devices, a Trackstick and a Garmin 220 tracking device. This was used to track the dog’s search pattern and compare it with the oil spill positions, wind directions etc. Only the Garmin system gave the necessary accuracy and updating frequency (1–3 meters, updating every other second). The Garmin system also offered real-time updating using a built-in UVF transmitter, while the Trackstick system only logged the positions in the memory stick. Originally the Garmin system was designed to give real-time tracking of dogs during hunting, while the Trackstick is mostly used to track vehicles.

This system made it possible to track the dogs during field training and to study each individual track in relation to oil and wind at the subsequent debriefing. Parameters such as distances, average search speeds etc. were displayed and calculated using the Garmin Mapsource software ver. 6.11. See examples of tracks in Figure 8

The objective with this initial training was to show how the dogs were able to detect the smell of oil starting with an easy target (spill no. 18, 400 ml drilled down into the ice). All the dogs started from the shore, and went onto the ice in a straight line 90 degrees crosswind/downwind from oil sample no. 18. All three dogs gave a clear indication after approximately 400 meters that oil was upwind, as seen in Figure 8. The search speed in such open terrain was in the 3–5 km/h range for the long-legged dogs and in the range of 1–2 km/h for more detailed search or with the short-legged dogs.

The main objective of this part of the field training was to estimate the dimensions or approximate size of an oil spill by using oil detecting dogs. For this purpose we used two series of small oil spills with a spacing of 10 m to indicate a larger oil spill (oil spill no. 07–13 + 14–17 in Figure 10).

Triangulation was used to determine the range of findings and oil samples. In the example in Figure 10 an initial search was made downwind to find the start of the oil spill (red track) and then entering from opposite sides to make indications of range (red, green and yellow track). Using the wind, the west/east range of this oil spill was estimated to approx. 100 meters. The dogs stopped and marked (sitting or laying down) when detection oil in the ice. During training, the distance from the marking to the actual oil spill was typically 50–100 meters. This “safety distance” is an important part of the training to avoid the dogs entering the oiled area. This “safety distance” could be as small as 20 meters when there was little wind. No north/south delineation were performed during this test, but would have been needed for a more precise two-dimensional mapping of the oiled area.

One of the main objective of this training was to differentiate between and locate different point sources. For this purpose we used point sources with larger spacing (100–200 meters), see Figure 9. The wind was very weak and variable during this training day (10 April, see Figure 4), which made the work more challenging. The dogs, especially the more experienced Jippi (yellow track), managed to correctly locate the different point sources (spill 4, 5 & 6 in Figure 9) and search both down- and upwind. Also, the two other dogs located most of the point sources, the green dog located spill 4 & 5 and the red dog located spill 5 (see Figure 9).

Another objective of this training was to determine how far downwind the dogs were able to detect the oil. In this case, the large oil spill (10 m²) consisting of 400 liters of Statfjord crude was used (see Figure 6). The team used the dogs at three different distances downwind (approximately 800, 3000 and 5000 meters) and at approximately 200 meters on the upwind side of the oil spill. The tracks are given in Figure 11. Searches were only performed on one side of the assumed oil fume plume on the downwind side of the spill site. If these oil detections or indications the oil plume, are mirrored along the assumed centerline straight downwind from the oil spill, it was possible to indicate the dimensions and direction of the plume (see Figure 11, Figure 12 and Figure 13). The maximum distance for oil plume detection during this training was measured to approximately 5 km with no indication of this being the maximal detecting distance. Time limitations prohibited further testing to determine final limits.

This section summarizes the conclusions and recommendations resulting from this fieldwork. Generally, the dogs performed very well, and better results than expected were achieved.

During ordinary passenger flights (Boing 737) the dogs were able to handle the stress at check-in, crowds/queues and security check very well. They also coped well with lying under the aircraft seat for extended periods (2–3 hours), and during takeoff and landing. There were no negative comments from the other passengers or airport staff.

In the small fixed wing aircraft (Dornier 220) the dogs were transported in their crates in the cabin with the handlers. All the dogs handled this very well, with little stress and no complaints.

The transport by snow scooter sledge was challenging. The dogs handled the bumpy and noisy rides very well, without showing any lack of concentration or large stress response. However, the snow surface was rather smooth due to favorable snow conditions prior to this field work. Other more challenging snow or terrain conditions could make scooter transport difficult and create a possible need for helicopter transport.

This field work showed that the temperature stress (10 m/s wind and −15°C) was manageable for both the dogs and handlers. The work was organized in two periods of four hours each, a total of eight hours per day. This could have been extended (due to 24 hours of sunlight), but time was also needed to evaluate the daily training, adjust plans for the next day and to rest the teams.

The dogs were able to concentrate on the oil search and ignored the local wildlife. One search was performed with several seals laying on the ice only 20 meters away, and fresh polar bear tracks were also ignored. Prior training and motivation on-site helped the dogs to ignore the smell from the local wildlife. This can especially be challenging for dogs that are also used for tracking and hunting.

The documentation from this 4 day Arctic training (oil properties, GPS-tracks, video and photos) was extensive. The dogs managed to:

1. Locate smaller oil spills (400 ml of weathered oil, 50 cm into the ice, covered in snow and left for a week before it was tracked by the dogs).

2. Search over large areas with a high speed and areal coverage (3–4 km/h) and endurance (4 × 2 hours a day). This was dependent on terrain and environmental conditions.

3. Determine the dimensions of larger oil spills by indicating the borders of clusters of smaller oil spills (“10 meter spacing”).

4. Locate a larger oil spill (400 L, on top of ice covered in snow) based on the triangulation of detected plume dimensions. The oil plume from the oil was clearly detected by the dogs up to 5 km downwind of the spill location.

During this training the dogs were not used to pinpoint the exact location of each oil spill and no real blind searches were performed, since the handlers knew the approximate location of the oil. However, identifying exact positions and blind searches have been documented by both laboratory and field training previously (Buvik and Brandvik, 2008).

The main objective in this study was to search larger arctic areas and indicate directions to and locations of oil spills covered in snow and ice. The technique was also very efficient to screen large areas and show that they were was free of hidden patches of oil. A possible scenario would be to detect “patches” of oil under ice or covered in snow in an Arctic fjord like van Mijien on Svalbard, which imply searching an area of 5 by 35 kilometers.

The results from this field training have shown that dogs can detect even small amounts (milliliters) of weathered oil that has been covered by snow and ice for a week. Larger amounts of weathered oil (400 liters) can be tracked from distances of several kilometers.

To be able to utilize dogs as an operational tool detecting oil spills hidden by snow and ice, the following tasks must be accomplished:

1. Discuss with national authorities the documented capabilities of trained dogs to detect oil spills hidden in snow and ice, with subsequent consideration for inclusion in oil spill contingency planning.

2. Establish guidelines training and certification of for Arctic dog equipages (handlers & dogs).

3. Establish operational procedures for search in an arctic environment.

4. Establish agreements between dog training institutes and oil spill contingency organizations for operational use of such dogs.

This work has been an integrated part of an Oil-in-Ice Joint Industry Program (2005 – 2010) and the authors wish to thank the sponsors of this program; the Norwegian Research Council, Agip KCO, ConocoPhillips, Chevron, Shell, Statoil and Total.