Potential for tsunami detection via CCTV cameras in northeastern

Leveraging Coastal Cameras to Enhance Tsunami Monitoring and Early Warning

Tsunamis pose a significant threat to coastal areas, making advancements in real-time detection and understanding of their generation mechanisms crucial. While various technologies have been developed for observing sea-level changes, such as GPS buoys, seafloor pressure gauges, and offshore sensor networks, the coverage and resolution of the observational network can still be improved.

One promising approach is to leverage the extensive network of Closed-Circuit Television (CCTV) cameras maintained by the Japanese government along the coasts, rivers, and roads. These cameras, part of a disaster prevention infrastructure, can transmit real-time images to local authorities during emergencies, aiding in damage assessment. However, effectively utilizing this vast video footage for operational purposes, such as tsunami detection, remains a challenge.

This study explores the potential of CCTV networks in northeastern Toyama Prefecture, Japan, as a new data source for tsunami detection following the 2024 Noto Peninsula earthquake. By analyzing CCTV footage, the researchers were able to extract time-series water level fluctuations and identify several long-period peaks (more than 100 seconds) in the power spectral density, suggesting the presence of tsunami components.

Key Findings:

• Spectral Consistency Across Locations: The CCTV footage revealed several long-period spectral peaks (100-200 s, 200-300 s, 400-500 s, and 885 s) that were consistent across multiple locations, including offshore wave observation points. This suggests these frequencies may represent the spectral characteristics of the tsunami event in the northeastern Toyama region.

• Observed Run-up and Water Level Deviations: At the Yokoyama CCTV location, a maximum run-up of approximately 3 meters was observed around 16:28. Water level deviations were also detected at the Shimoiino and Ekko CCTV sites, though identifying clear tsunami components proved more challenging due to their smaller magnitudes compared to other wave components.

• Potential for Real-Time Monitoring: The study demonstrates the potential of CCTV networks for tsunami detection, complementing traditional offshore sensor-based methods. Further research is needed to achieve real-time detection capabilities and overcome challenges in distinguishing tsunami signals from other wave disturbances.

Extracting Tsunami Signatures from CCTV Footage

The researchers employed a multi-step approach to extract tsunami-related information from the CCTV footage:

1. Image Preprocessing: The original CCTV footage (29.97 fps, 1920 x 1080 resolution) was decomposed into individual frames. Time-averaging of 29 frames (approximately 1 second) was used to reduce noise and data processing costs while maintaining sufficient stability for edge detection.

2. Shoreline Detection: The Canny edge detector was applied to the grayscale, time-averaged images to identify the shoreline. Additional filtering steps were implemented to remove noise and enhance the robustness of the shoreline position detection.

3. Waveform Extraction: The time-series data of shoreline position changes were used to derive water level deviation (WLD) waveforms. Multiple transects were averaged to offset noise effects and extract more reliable waveforms.

4. Spectral Analysis: Spectral analysis of the WLD waveforms was conducted using the Welch method, with a low-pass filter applied to eliminate the influences of wind waves and swells.

Validating Tsunami Signatures through Offshore Observations

To validate the CCTV-derived waveforms, the researchers compared them to offshore wave observations at the Tanaka and Toyama monitoring stations. The analysis revealed several key findings:

• Waveform Consistency: The WLD waveforms obtained from the Yokoyama CCTV footage exhibited similar spectral characteristics (e.g., peaks at 100-200 s, 200-300 s, 400-500 s, and 885 s) to the offshore observations, suggesting a common tsunami signal.

• Temporal Alignment: The timing of significant water level deviations observed at Yokoyama aligned well with the offshore measurements at Tanaka, with a slight delay due to the shoreline location.

• Local Amplification: The gentler sea surface elevations observed at the Toyama station, located further south from Yokoyama and Tanaka, suggest that the more pronounced fluctuations at the other locations could be attributed to tsunami components.

While the Shimoiino and Ekko CCTV locations also detected potential tsunami components, the smaller magnitudes of the water level deviations made it challenging to clearly distinguish them from other wave disturbances using the current methods.

Overcoming Challenges for Real-Time Tsunami Detection

The primary challenge in leveraging CCTV footage for real-time tsunami detection lies in the reliable differentiation of tsunami signals from other wave disturbances. The researchers identified two main sources of noise:

1. Wave-originated Errors: These include the effects of topography, wave setup, or wave reflections, which can introduce complexities in the waveforms.

2. Shoreline Detection Inaccuracies: Issues such as seismic motion blurring and post-inundation wet conditions can affect the accuracy of shoreline detection, leading to potential errors of up to 1 meter in the pre-tsunami water levels.

The limited pre-earthquake data (16:00-16:10) also restricted a comprehensive analysis of non-tsunami conditions, making it challenging to remove these noise sources effectively.

To address these challenges, the researchers emphasized the need for further investigation into refinement of the detection methods, ensuring a balance between minimizing errors and maintaining versatility across various sites. Developing techniques for automatic tsunami signal detection, while acknowledging the current reliance on some subjective interpretation, will be crucial for realizing the full potential of CCTV-based tsunami monitoring.

Towards Real-Time Tsunami Monitoring and Early Warning

The ultimate goal of utilizing CCTV cameras for tsunami detection is twofold:

1. Real-Time Monitoring and Early Warning: By analyzing waveforms and run-up heights from CCTV footage, the accuracy and timeliness of tsunami alerts can be enhanced, supporting prompt evacuation efforts.

2. Post-Tsunami Impact Assessment: CCTV data can facilitate detailed post-tsunami analyses, enabling a thorough understanding of the event’s impact and the effectiveness of existing disaster prevention measures.

These approaches are expected to strengthen immediate tsunami response measures and contribute to the formulation of long-term tsunami disaster prevention strategies.

While the initial findings of this study demonstrate the feasibility of CCTV-based tsunami detection, further advancements are needed to achieve real-time operational capabilities. Ongoing research and collaborations between researchers, local authorities, and technology providers will be crucial in realizing the full potential of this emerging approach to coastal hazard monitoring and early warning.

To support the continued development of this technology, the researchers have committed to making a sample code for the CCTV footage analysis used in this study available on their GitHub repository (https://github.com/Chuocoastlabo) by next summer in Japan. However, the CCTV footage itself cannot be immediately shared due to the need for permission from the provider.

By leveraging the extensive CCTV network infrastructure already in place, the potential for enhanced tsunami detection and early warning systems holds promise for coastal communities around the world. As researchers and practitioners continue to explore and refine these methods, the integration of CCTV-based monitoring into comprehensive disaster risk reduction strategies can play a vital role in building resilience against the devastating impacts of tsunamis.