Biogeosciences Research Spotlight

Detecting Gas Leaks with Autonomous Underwater Vehicles

A Norwegian team develops an improved, cost-effective method to detect chemical discharges under the sea.

Source: Journal of Geophysical Research: Oceans

By Sarah Witman

On the ocean floor, at this very minute, fleets of robots roam free—independent of human operators. These machines, called autonomous underwater vehicles, or AUVs, have battery-powered sensors on board to detect and take instant measurements of various oceanic properties, providing data for scientific studies. Many AUV surveys take place each year in areas of interest, such as deep-sea vents, cracks in the seafloor that expose the hot, volcanic rock below.

One important use of AUVs is in detecting anomalies to help pinpoint the location of harmful chemical discharges under the sea. For example, oil wells bored into Earth’s surface or natural gas stored underground for later use are susceptible to leaks. Also, fish farms (such as the massive, circular pens housing salmon and trout in the fjords of Norway) may leak organic waste and other pollutants into marine ecosystems.

Detecting these anomalies is a tough job, however. AUVs have limited battery life, so simply covering large distances to try to locate discharges or a false alarm comes at a cost. Plus, ocean dynamics—tides, currents, and winds—change constantly. This change affects the movement and mixing of seawater, making the detection of potential leaks even more challenging.

Here Alendal suggests some strategies for designing AUV surveys to better detect anomalies in the ocean and to do so in a cost-effective way. The strategies involve creating a map of potential discharge locations in a given area and assigning a probability of them leaking. The author also suggests defining a small background probability to account for leaks from unknown locations, for example, when gas ascends to the seafloor through lateral conduits that create discharges farther afield.

In addition, predictions on how the discharge will be transported and diluted in the water column are needed, along with specs that describe an instrument’s capacity to detect a leak, because these will define the area in the neighborhood of a leak in which the leak may be detected. Then, once AUVs are deployed, the author envisions a system in which each measurement failing to detect a seep will update the underlying map of probable discharge locations, using Bayes’s theorem, and the location for the next measurement with the highest probability of detecting a discharge can be selected. Within such an operating plan, the AUV’s efforts can be maximized, allowing operators to balance the relative costs between making measurements and traveling long distances.

The author tested these strategies using simulations of the North Sea. For the test, 15 possible discharge locations were chosen randomly in a 130-square-kilometer area of seafloor, representing oil and gas wells perforating the seafloor with the highest risk of leaking the stored gas.

In the demonstration, the author used a general circulation model to predict the gas transport and dilution of a leak away from a source. Prior research has used the same framework of simulations to design programs that monitor leaks with fixed instrumentation. Here the author demonstrates how such simulations can assist in defining survey paths for moving platforms, such as AUVs.

This study provides a pathway toward improving AUV surveys of areas of interest along the seafloor. The method it outlines has the potential not only to allow these vehicles to locate gas leaks and other targets with greater precision but to do so at a lower cost to private companies, universities, and other organizations. (Journal of Geophysical Research: Oceans, https://doi.org/10.1002/2016JC012655, 2017)

—Sarah Witman, Freelance Writer

Citation: Witman, S. (2017), Detecting gas leaks with autonomous underwater vehicles, Eos, 98, https://doi.org/10.1029/2017EO080597. Published on 29 August 2017.
© 2017. The authors. CC BY-NC-ND 3.0