Smart AUVs for detection and quantification of greenhouse gas seepage in the oceans

Greenhouse gas seepage into the oceans is a major environmental challenge. As ocean temperatures continue to rise, significant amounts of methane (CH4) currently trapped in permafrost are expected to be released into the oceans. Further, carbon capture and storage (CCS) is a rapidly emerging technology aimed at reducing CO2 emissions and reaching international climate goals by injecting and permanently storing large amounts of CO2 in geological reservoirs below the seabed. Finally, there are numerous legacy oil and gas wells offshore potentially acting as conduits of CH4 from shallow geological layers to the marine environment. Monitoring and documenting these processes using current technology is challenging and, in many cases, economically prohibitive. Autonomous underwater vehicles (AUVs) are highly suitable for mapping large ocean areas and can be equipped with a range of chemical and acoustic sensors for detailed mapping of both the seabed and the water column. Currently, these vehicles are autonomous in the sense that they can follow a pre-programmed route and collect data for later analysis and interpretation. Smart AUVs aims to enhance these vehicles' ability to detect and map greenhouse gases in the ocean by utilizing advanced data analysis, including Artificial Intelligence (AI), allowing the vehicle to make intelligent decisions in real-time based on sensor readings. In this way, the vehicle will continuously optimize both its route and data collection parameters, ensuring that features of interest are mapped in detail while avoiding spending unnecessary time in areas where there are no signs of leakage. The project brings together expertise in applied mathematics, AI, signal processing, autonomy, CCS, and oceanography to achieve this goal. The project aims to develop new methods for detecting and mapping CO2 and CH4 in the oceans, as well as strategies for how the vehicle should move to optimally map potential leaks. In WP1, the project conducted a study to identify and describe relevant emission scenarios to focus on, including emission source, leakage rate, type of discharge, ocean depth, etc.). Furthermore, in WP2, we have simulated these representative emission scenarios in high spatial and temporal resolution, to create an understanding of how such emissions behave in a realistic ocean environment and to generate a comprehensive simulated data set for the development of intelligent algorithms. In the spring of 2024, we conducted extensive controlled releases at a depth of 70 meters, 450 meters from the shore. Small to medium amounts of CO2 and CH4 in gas form (bubbles) were released at the seabed over a period of 5 days, using a specially designed lander created for this purpose. The release was controlled from land, with digital flow meters at the release point ensuring full control of the emissions. During the sea trials, a HUGIN AUV was used to map the emissions in detail, both acoustically and chemically. The AUV had been modified for this purpose in advance, with relevant chemical sensors (CO2, CH4, O2, pH), in addition to acoustic sensors (HISAS, EM2040, and EK80). The EK80 sensor is commonly used in the fisheries industry and has not previously been used on a HUGIN AUV in this way. It was integrated on the HUGIN AUV in this project because it has the potential to acoustically quantify gas emissions. Although the main goal of this year's field trials was to test different AUV data acquisition patterns and their suitability for mapping CO2 and CH4 emissions, we have also laid the foundation to test adaptive path planning in 2025. An algorithm developed for automatic detection of gas bubbles using the EM2040 sonar was optimized and integrated into the HUGIN AUV as part of the project. The algorithm was run in real-time on one of HUGIN's internal computers during data acquisition, and when detecting a potential leak, a metafile describing the findings was stored on a local disk. The next step, which we hope to take in 2025, will be to allow the output from this and other algorithms to send messages to the control system in the HUGIN AUV, allowing the AUV to form an "understanding" of its surroundings and make intelligent real-time decisions accordingly. In the planned field trials in 2025, we hope to use an Eelume AUV in addition to HUGIN. The project has had several summer students contributing to data analysis and visualization, and the project's two PhD students are working on both acoustic detection of leaks and adaptive path planning.

Monitoring the marine environment is a vital part of ensuring safe and sustainable marine operations and understanding the dynamics of the global carbon cycle. AUVs play a key role in marine monitoring because of their ability to cover large areas and use sensors tailored to the monitoring needs. Currently these vehicles have limited decision autonomy and therefore follow a pre-defined travel path. SmartAUVs will dramatically improve the monitoring capabilities of AUVs, by applying artificial intelligence (AI) in concert with specialized signal processing techniques to enable AUV decision autonomy, i.e., the ability to take intelligent action in real time based on sensor input. The aim of WPs 1 and 2 is to understand critical CO2/CH4 emission scenarios and simulate these to provide detailed knowledge about plume properties. These insights determine the desired AUV behaviour including travel path and sensor usage. In WP3 we develop intelligent algorithms for automatic leak detection; converting raw sensor data to information based on which the AUV can act (e.g., sonar data converted to information about seepage). In WP4 we develop AUV autonomy including situational awareness and optimized travel path. Algorithms for leak detection (WP3) and adaptive behaviour (WP4) will be implemented in the HUGIN AUV processing unit, and full autonomy demonstrated during field trials (WP5). The Eelume AUV will also be used during the field trials, and a basic level of decision autonomy developed and demonstrated. The HUGIN and Eelume AUVs have fundamentally different properties and complementary monitoring capabilities. SmartAUVs will contribute to improved monitoring of the oceans, which in turn will enable sound management of marine activities, strengthened confidence in safe CO2 storage, and insights into the amount of CH4 entering the oceans through natural- and industry related processes. Active industry engagement will enable efficient benefits realization.