5:15 PM - 7:15 PM
[SSS04-P03] Along-Slope Currents Driving Gas Bubble Emissions in the Northern South China Sea
Keywords:along-slope currents, seafloor gas emission, seismic bubble waveform, machine learning, pressure depression, Bernoulli’s principle
Methane seepage from the seafloor and gas hydrate accumulation beneath subsurface caps have been observed globally, responding to changes in gas hydrate stability and methane emissions. Seepage systems are highly sensitive to pressure and temperature variations, with tidal fluctuations proposed as a key factor in methane release. However, studies suggest that sea level changes influence hydrate stability over millions of years.
This study analyzed high-resolution seismic data from three ocean-bottom seismometer (OBS) surveys conducted in 2007, 2008, and 2011 in the northern South China Sea (SCS), covering depths of 200–3,800 m. Bubble signals recorded by OBSs are localized energy onsets detected only at the nearest station. A single bubble waveform exhibits a high-frequency resonant wave chain lasting under two seconds, identified as a depressing burst signal when methane ascends from sediment fractures to the free surface of a non-Newtonian fluid. These signals are better recorded by geophones than hydrophones.
The SCS is known for abundant gas hydrates, with the Shenhu Basin developed for commercial gas hydrate mining. Across our three OBS surveys, bubble signals are prevalent along continuous seismograms. Given their similarity in time and frequency domains, machine learning is a promising tool to filter signals efficiently. We employ a recurrent neural network (RNN) with a long-short-term memory (LSTM) architecture for signal detection. The model is trained using features from time-series envelopes and frequency patterns derived from wavelet spectrograms, identifying signal sections based on attenuation duration and frequency content. Post-processing statistically compiles each bubble event across three geophone channels.
Bubble detections reveal spatial and temporal trends. The number of detected bubbles decreases with increasing depth, but occurrences are rare at depths shallower than 1,000 m. The timing of bubble cluster emergence varies across stations, forming an along-slope pattern, particularly for the OBS array northeast of the SCS in 2007. Stations in the Shenhu basin exhibited a higher number of bubble detections with little spatial variation. We infer that gas hydrate deposits primarily control bubble emissions, but external forces triggering hydrate dissociation also contribute.
Comparing local tidal models with the spatiotemporal distribution of bubble clusters, we find that local tide-induced water level variations are not the primary mechanism for gas hydrate dissociation. Instead, we propose that along-slope currents—shaped by surface tides and bathymetric features—are the key trigger for methane emissions. According to Bernoulli’s principle, bottom currents passing over the seafloor cause localized pressure depression, which facilitates gas hydrate dissociation or the upward migration of methane bubbles.
Based on bubble cluster shifts, we estimate along-slope current speeds in the northeastern SCS at 0.93–2.84 m/s, comparable to prior studies on internal wave propagation near Dongsha Atoll, where speeds of 2.94 m/s were recorded near the OBS array in 2007. Our findings suggest that bottom current-induced pressure depression is a significant driving force for seafloor gas emissions.
This study analyzed high-resolution seismic data from three ocean-bottom seismometer (OBS) surveys conducted in 2007, 2008, and 2011 in the northern South China Sea (SCS), covering depths of 200–3,800 m. Bubble signals recorded by OBSs are localized energy onsets detected only at the nearest station. A single bubble waveform exhibits a high-frequency resonant wave chain lasting under two seconds, identified as a depressing burst signal when methane ascends from sediment fractures to the free surface of a non-Newtonian fluid. These signals are better recorded by geophones than hydrophones.
The SCS is known for abundant gas hydrates, with the Shenhu Basin developed for commercial gas hydrate mining. Across our three OBS surveys, bubble signals are prevalent along continuous seismograms. Given their similarity in time and frequency domains, machine learning is a promising tool to filter signals efficiently. We employ a recurrent neural network (RNN) with a long-short-term memory (LSTM) architecture for signal detection. The model is trained using features from time-series envelopes and frequency patterns derived from wavelet spectrograms, identifying signal sections based on attenuation duration and frequency content. Post-processing statistically compiles each bubble event across three geophone channels.
Bubble detections reveal spatial and temporal trends. The number of detected bubbles decreases with increasing depth, but occurrences are rare at depths shallower than 1,000 m. The timing of bubble cluster emergence varies across stations, forming an along-slope pattern, particularly for the OBS array northeast of the SCS in 2007. Stations in the Shenhu basin exhibited a higher number of bubble detections with little spatial variation. We infer that gas hydrate deposits primarily control bubble emissions, but external forces triggering hydrate dissociation also contribute.
Comparing local tidal models with the spatiotemporal distribution of bubble clusters, we find that local tide-induced water level variations are not the primary mechanism for gas hydrate dissociation. Instead, we propose that along-slope currents—shaped by surface tides and bathymetric features—are the key trigger for methane emissions. According to Bernoulli’s principle, bottom currents passing over the seafloor cause localized pressure depression, which facilitates gas hydrate dissociation or the upward migration of methane bubbles.
Based on bubble cluster shifts, we estimate along-slope current speeds in the northeastern SCS at 0.93–2.84 m/s, comparable to prior studies on internal wave propagation near Dongsha Atoll, where speeds of 2.94 m/s were recorded near the OBS array in 2007. Our findings suggest that bottom current-induced pressure depression is a significant driving force for seafloor gas emissions.