Passive acoustic techniques can be used to identify and quantify underwater gas release at natural sites, or at locations related to anthropogenic activities. There are still significant issues in extracting bubble signals from background noise, particularly for bubble counting and sizing techniques relying on inversion of the time-averaged acoustic spectrum. In this work we propose an adaptive single bubble identification technique, which incorporates bubble acoustic characteristics including pulsation time interval, frequency bandwidth and radiation strength. The method applies a cross-spectrogram, enabling an increase in signal-to-noise ratio resulting in a reduction of the false alarm rate on bubble identification. We demonstrate this technique using an array of hydrophones to determine the bubble size distribution and gas flux at a controlled CO₂ release site, 4 m beneath the seabed, at 120 m water depth in the central North Sea. The results show that the bubble radius, as estimated using acoustics has a distribution with a peak in the 0.15–0.3 cm range, while an estimate based on optical method suggests a range of 0.2–0.5 cm. The gas flux is acoustically estimated as 32–88 kg/day in response to a known gas injection flow rate 143 kg/day, indicating 22–62% of the injected CO₂ is emitted from the seabed in gaseous form, with the remainder being trapped, or dissolved.

被动声学技术可用于识别和量化自然地点或与人为活动有关的地点的水下气体释放。从背景噪声中提取气泡信号仍然存在重大问题，尤其是对于依赖于时间平均声谱求逆的气泡计数和大小调整技术而言。在这项工作中，我们提出了一种自适应的单气泡识别技术，该技术融合了气泡的声学特性，包括脉动时间间隔，频率带宽和辐射强度。该方法应用了交叉谱图，从而提高了信噪比，从而降低了气泡识别的误报率。在北海中部水深120 m处海床以下4 m处有_2个释放点。结果表明，使用声学方法估计的气泡半径具有在0.15-0.3 cm范围内的峰值分布，而基于光学方法的估计值则建议在0.2-0.5 cm的范围内。根据已知的143千克/天的注气流量，声学估计的气体通量为32-88千克/天，这表明22-62％的注入CO ₂以气态形式从海底排放，其余为被困或溶解。