11:15 AM - 11:30 AM
[ACG52-09] Acoustic monitoring of proglacial discharge of Qaanaaq Glacier, northwestern Greenland
Keywords:Glacier, Greenland, Acoustic observation, Signal analysis, Runoff observation
1. Background
Glaciers and ice caps along coastal Greenland have significantly contributed to sea level rise in recent decades. (Hugonnet et al., 2021) Observations of proglacial discharge are essential to estimate the impact of glacial meltwater on sea level rise. Additionally, proglacial streams can cause flooding due to extreme melting and heavy rainfall, potentially impacting local communities (Kondo et al., 2021). Therefore, continuous monitoring of proglacial discharge is crucial.
However, traditional discharge monitoring methods require the installation of expensive pressure sensors on the streambed. These sensors can become ineffective due to low discharge or sediment accumulation, leading to data gaps. As an alternative, passive acoustic methods offer a cost and labor-efficient solution, allowing non-invasive monitoring by simply placing acoustic sensors near the stream (Podolskiy et al., 2023). However, the advantages and limitations of this approach remain poorly understood.
2. Methods
In the summer of 2024, we deployed four acoustic sensors (Song Meter, Wildlife Acoustics) and three time-lapse cameras along the proglacial discharge of Qaanaaq Glacier, Greenland, to assess the potential of comprehensive acoustic monitoring for discharge estimation. The locations of the sensors are shown in Figure 1.
We conducted repeated discharge measurements using an electromagnetic current meter 32 times, to establish a relationship between continuous water level measurements (recorded by a pressure sensor) and the acoustic data. Based on the water level–discharge curve, we reconstructed continuous discharge data from the water level measurements.
3. Results and Discussion
Our results demonstrate that the sound generated by the proglacial stream correlates with discharge at all stations and can serve as a continuous proxy for runoff. We divided the long-term spectrogram data into eight frequency bands and found that all sensors strongly correlated with discharge (R ≈ 0.9) in the 94–375 Hz range, which is likely associated with the sound of flowing water.
Additionally, the correlation coefficient in the 23–47 Hz band was high for the two sensors located upstream but relatively low for the two downstream sensors (R = 0.1 and 0.5). This discrepancy was caused by background noise: at Point 3, wind interference affected the acoustic signal, while at Point 4, noise from passing vehicles on a nearby bridge introduced contamination. These findings highlight the importance of carefully selecting sensor locations to minimize background noise and implementing noise reduction strategies during deployment.
Furthermore, we analyzed the time lag of the acoustic signal between Point 1 (near the glacier terminus) and Point 4 (near the bridge). The time lag ranged from 36 to 48 minutes, and the flow velocity was estimated to be 0.65 m/s to 0.9 m/s, which was consistent with the observed value.
4. Future Work
Minimizing these interferences from background noise is essential for improving the accuracy of acoustic discharge monitoring. Additionally, traffic noise could be used to estimate the number of vehicles crossing the bridge, providing valuable data for assessing bridge usage. This information could contribute to risk management efforts related to stream flooding events.
Glaciers and ice caps along coastal Greenland have significantly contributed to sea level rise in recent decades. (Hugonnet et al., 2021) Observations of proglacial discharge are essential to estimate the impact of glacial meltwater on sea level rise. Additionally, proglacial streams can cause flooding due to extreme melting and heavy rainfall, potentially impacting local communities (Kondo et al., 2021). Therefore, continuous monitoring of proglacial discharge is crucial.
However, traditional discharge monitoring methods require the installation of expensive pressure sensors on the streambed. These sensors can become ineffective due to low discharge or sediment accumulation, leading to data gaps. As an alternative, passive acoustic methods offer a cost and labor-efficient solution, allowing non-invasive monitoring by simply placing acoustic sensors near the stream (Podolskiy et al., 2023). However, the advantages and limitations of this approach remain poorly understood.
2. Methods
In the summer of 2024, we deployed four acoustic sensors (Song Meter, Wildlife Acoustics) and three time-lapse cameras along the proglacial discharge of Qaanaaq Glacier, Greenland, to assess the potential of comprehensive acoustic monitoring for discharge estimation. The locations of the sensors are shown in Figure 1.
We conducted repeated discharge measurements using an electromagnetic current meter 32 times, to establish a relationship between continuous water level measurements (recorded by a pressure sensor) and the acoustic data. Based on the water level–discharge curve, we reconstructed continuous discharge data from the water level measurements.
3. Results and Discussion
Our results demonstrate that the sound generated by the proglacial stream correlates with discharge at all stations and can serve as a continuous proxy for runoff. We divided the long-term spectrogram data into eight frequency bands and found that all sensors strongly correlated with discharge (R ≈ 0.9) in the 94–375 Hz range, which is likely associated with the sound of flowing water.
Additionally, the correlation coefficient in the 23–47 Hz band was high for the two sensors located upstream but relatively low for the two downstream sensors (R = 0.1 and 0.5). This discrepancy was caused by background noise: at Point 3, wind interference affected the acoustic signal, while at Point 4, noise from passing vehicles on a nearby bridge introduced contamination. These findings highlight the importance of carefully selecting sensor locations to minimize background noise and implementing noise reduction strategies during deployment.
Furthermore, we analyzed the time lag of the acoustic signal between Point 1 (near the glacier terminus) and Point 4 (near the bridge). The time lag ranged from 36 to 48 minutes, and the flow velocity was estimated to be 0.65 m/s to 0.9 m/s, which was consistent with the observed value.
4. Future Work
Minimizing these interferences from background noise is essential for improving the accuracy of acoustic discharge monitoring. Additionally, traffic noise could be used to estimate the number of vehicles crossing the bridge, providing valuable data for assessing bridge usage. This information could contribute to risk management efforts related to stream flooding events.