Sang-Ho Yun
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Sang-Ho Yun has a total of 118 co-authors. Here are the top co-authors by number of shared publications:
- Jungkyo Jung
Jet Propulsion Laboratory, California Institute of Technology, CA, USA
5 shared publications Susilo Susilo
National Agency for Research and Innovation, Jakarta, Indonesia
4 shared publications- Benoit Taisne
Earth Observatory of Singapore, Nanyang Technological University, Singapore, Singapore
3 shared publications Bihan Wen
School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore, 639798, Singapore
3 shared publications- Eleanor Tennant
Earth Observatory of Singapore and Asian School of the Environment, Nanyang Technological University, Singapore, Singapore
3 shared publications - Emma M. Hill
Earth Observatory of Singapore, Nanyang Technological University, Singapore, Singapore
3 shared publications
Latest Publications
- Open Access
- 29 August 2026
- Published by Springer Nature in Communications Earth & Environment
Coulomb-like creeping segment acts as a stress sensor in Northern Sumatra
Understanding how fault creep evolves in space and time is essential for identifying zones of aseismic stress release and stress accumulation. Creeping faults typically exhibit rate-strengthening friction, in which frictional resistance increases with sliding velocity and promotes stable slip. The northern Aceh segment of the Sumatran Fault Zone is a notable exception, hosting active creep yet rupturing in a Mw 6.1 strike-slip earthquake in 2013. Here we show, using satellite observations, that creep rates decreased by ~60% from 2007–2010 to 2017–2023. Numerical models constrained by these observations reveal two creep events: one triggered by stress transfer from the distant 2004 Mw 9.2 megathrust earthquake, and a second re-acceleration driven by nearby continental earthquakes. These results suggest that the Aceh fault exhibits emergent, nearly velocity-neutral behaviour at the segment scale. In this regime, the fault behaves like a brittle Coulomb surface lacking the self-stabilizing influence of velocity- and slip-history-dependent friction, enabling long-range triggering and the coexistence of creep and earthquakes. The fault can thus act as a natural stress gauge in a region of significant seismic hazard.
- 29 August 2026
- Published by Springer Nature in Communications Earth & Environment
Open AccessCoulomb-like creeping segment acts as a stress sensor in Northern Sumatra
Understanding how fault creep evolves in space and time is essential for identifying zones of aseismic stress release and stress accumulation. Creeping faults typically exhibit rate-strengthening friction, in which frictional resistance increases with sliding velocity and promotes stable slip. The northern Aceh segment of the Sumatran Fault Zone is a notable exception, hosting active creep yet rupturing in a Mw 6.1 strike-slip earthquake in 2013. Here we show, using satellite observations, that creep rates decreased by ~60% from 2007–2010 to 2017–2023. Numerical models constrained by these observations reveal two creep events: one triggered by stress transfer from the distant 2004 Mw 9.2 megathrust earthquake, and a second re-acceleration driven by nearby continental earthquakes. These results suggest that the Aceh fault exhibits emergent, nearly velocity-neutral behaviour at the segment scale. In this regime, the fault behaves like a brittle Coulomb surface lacking the self-stabilizing influence of velocity- and slip-history-dependent friction, enabling long-range triggering and the coexistence of creep and earthquakes. The fault can thus act as a natural stress gauge in a region of significant seismic hazard.
- Open Access
- 14 March 2026
- Published by Copernicus GmbH
Surface rupture characteristics and macroseismic effects of the 2025 Mw 7.7 Sagaing fault earthquake in central Myanmar
The 2025 M 7.7 Myanmar earthquake affected over 30 million people across Myanmar and the broader Asian region. The earthquake caused over 5,000 fatalities, injured thousands, and left several hundred people missing. Damage extended across Myanmar, Thailand, and China, with strong shaking felt throughout Southeast Asia. The rupture propagated for over 450 km, one of the longest strike-slip earthquake ruptures worldwide, cutting through densely populated and economically important regions of central Myanmar. However, the ongoing military coup and subsequent civil conflict between the central army and People’s Defence Forces (PDFs) severely limited rescue operations and ground-based field investigations. As a result, the assessment of rupture characteristics and, slip distribution, remains limited due to gaps in ground observations.In this study, we investigate rupture characteristics and coseismic offsets using ground-based field survey data integrated with remote-sensing observations and social media-derived felt reports and rupture information. Near the northern rupture termination, which coincides with an active conflict area, we mapped rupture patterns using newly updated Google Earth imagery, validated through reports of rupture posted by locals on social media (Facebook). Along the inferred 1839 M7+ rupture segment, details of the surface rupture were documented using unmanned aerial vehicle (UAV) and tape-and-compass surveys. In the restricted regions controlled by the central army, from Nay Pyi Taw to the southern rupture termination, coseismic offsets were measured using tape-and-compass methods only.Slip amounts measured from ground-based surveys south of Mandalay systematically underestimate offsets determined from remote sensing, suggesting a significant fraction of the deformation occurred beyond a few meters of the main fault zone. Nonetheless, our mapping indicates that the 2025 surface rupture partially or fully overlapped multiple earlier historical Sagaing fault ruptures, including those in 1839 (Mw 7+), 1956 (Mw 7.1), 1929 (Mw ~7.0), 1930 (Mw 7.3) and 2012 (Mw 6.8). The observed macroseismic effects are comparable to those inferred for the 1839 Ava earthquake, which was poorly understood due to limited historical data. These ground-based data provide critical insights into the rupture behaviour over multiple earthquake cycles of fault segments that, at least in 2025, are inferred to have produced supershear rupture.
- 14 March 2026
- Published by Copernicus GmbH
Open AccessSurface rupture characteristics and macroseismic effects of the 2025 Mw 7.7 Sagaing fault earthquake in central Myanmar
The 2025 M 7.7 Myanmar earthquake affected over 30 million people across Myanmar and the broader Asian region. The earthquake caused over 5,000 fatalities, injured thousands, and left several hundred people missing. Damage extended across Myanmar, Thailand, and China, with strong shaking felt throughout Southeast Asia. The rupture propagated for over 450 km, one of the longest strike-slip earthquake ruptures worldwide, cutting through densely populated and economically important regions of central Myanmar. However, the ongoing military coup and subsequent civil conflict between the central army and People’s Defence Forces (PDFs) severely limited rescue operations and ground-based field investigations. As a result, the assessment of rupture characteristics and, slip distribution, remains limited due to gaps in ground observations.In this study, we investigate rupture characteristics and coseismic offsets using ground-based field survey data integrated with remote-sensing observations and social media-derived felt reports and rupture information. Near the northern rupture termination, which coincides with an active conflict area, we mapped rupture patterns using newly updated Google Earth imagery, validated through reports of rupture posted by locals on social media (Facebook). Along the inferred 1839 M7+ rupture segment, details of the surface rupture were documented using unmanned aerial vehicle (UAV) and tape-and-compass surveys. In the restricted regions controlled by the central army, from Nay Pyi Taw to the southern rupture termination, coseismic offsets were measured using tape-and-compass methods only.Slip amounts measured from ground-based surveys south of Mandalay systematically underestimate offsets determined from remote sensing, suggesting a significant fraction of the deformation occurred beyond a few meters of the main fault zone. Nonetheless, our mapping indicates that the 2025 surface rupture partially or fully overlapped multiple earlier historical Sagaing fault ruptures, including those in 1839 (Mw 7+), 1956 (Mw 7.1), 1929 (Mw ~7.0), 1930 (Mw 7.3) and 2012 (Mw 6.8). The observed macroseismic effects are comparable to those inferred for the 1839 Ava earthquake, which was poorly understood due to limited historical data. These ground-based data provide critical insights into the rupture behaviour over multiple earthquake cycles of fault segments that, at least in 2025, are inferred to have produced supershear rupture.
- Open Access
- 13 March 2026
- Published by Copernicus GmbH
Coulomb-like creeping segment acts as a stress sensor in Northern Sumatra
Understanding the spatial and temporal evolution of creep along continental faults is key to identifying where stress is released aseismically and where it may accumulate, potentially leading to future seismic events. Typically, creeping segments exhibit rate-strengthening behaviour, where frictional resistance increases with sliding velocity, resulting in stable sliding. However, the northern Aceh segment of the Sumatran Fault Zone (SFZ), a right-lateral strike-slip fault, presents a notable exception. Here, we observe creep signals using satellite radar interferometry, capturing the temporal evolution of creep, which decreased by ~60% from the 2007–2010 period to the 2017–2023 period. Numerical modelling constrained by these observations identifies two distinct creep events. The first was triggered by stress transfer from Mw 9.2 2004 earthquake more than 150 km away, and a second localized re-acceleration due to nearby continental earthquakes. These results reveal that the fault behaviour is more consistent with a brittle Coulomb surface and lacking the usual self-stabilizing influence of velocity- and slip history-dependent friction. This central creeping section, sandwiched between two large locked domains, instead exhibits nearly velocity-neutral behaviour at the fault segment scale, making it highly sensitive to both local and regional stress changes. These findings provide evidence of long-range fault interactions, where both the subducting megathrust and possibly the oceanic mantle drive creep on a continental strike-slip fault. Located within a densely populated region, the velocity-neutral Aceh fault may participate in future earthquakes nucleating in surrounding locked segments, highlighting elevated seismic hazard in northern Sumatra.
- 13 March 2026
- Published by Copernicus GmbH
Open AccessCoulomb-like creeping segment acts as a stress sensor in Northern Sumatra
Understanding the spatial and temporal evolution of creep along continental faults is key to identifying where stress is released aseismically and where it may accumulate, potentially leading to future seismic events. Typically, creeping segments exhibit rate-strengthening behaviour, where frictional resistance increases with sliding velocity, resulting in stable sliding. However, the northern Aceh segment of the Sumatran Fault Zone (SFZ), a right-lateral strike-slip fault, presents a notable exception. Here, we observe creep signals using satellite radar interferometry, capturing the temporal evolution of creep, which decreased by ~60% from the 2007–2010 period to the 2017–2023 period. Numerical modelling constrained by these observations identifies two distinct creep events. The first was triggered by stress transfer from Mw 9.2 2004 earthquake more than 150 km away, and a second localized re-acceleration due to nearby continental earthquakes. These results reveal that the fault behaviour is more consistent with a brittle Coulomb surface and lacking the usual self-stabilizing influence of velocity- and slip history-dependent friction. This central creeping section, sandwiched between two large locked domains, instead exhibits nearly velocity-neutral behaviour at the fault segment scale, making it highly sensitive to both local and regional stress changes. These findings provide evidence of long-range fault interactions, where both the subducting megathrust and possibly the oceanic mantle drive creep on a continental strike-slip fault. Located within a densely populated region, the velocity-neutral Aceh fault may participate in future earthquakes nucleating in surrounding locked segments, highlighting elevated seismic hazard in northern Sumatra.
- Open Access
- 8 February 2026
- Published by American Geophysical Union (AGU) in Geophysical Research Letters
- Vol. 53 (3)
Coulomb Pre‐Stress Changes Modulate Coseismic Rupture Kinematics of the 2025 Mw7.7 Myanmar Earthquake Revealed by Space Geodesy
We estimate the fault geometry and coseismic slip of the 2025 Myanmar earthquake using multi‐source satellite observations, revealing a nine‐segment rupture structure that transitions from eastward‐dipping in the north to westward‐dipping in the south, with peak slip of ∼6 m and negligible shallow slip deficit. We model Coulomb pre‐stress evolution along the Sagaing Fault from 1839 to 2025, incorporating coseismic stress transfer from 10 M ≥ 6.5 earthquakes and interseismic tectonic loading, to investigate their influences on the 2025 rupture. Results show that the 1839 Ava earthquake could have cast a long‐lasting pre‐stress shadow at both ends of the 2025 rupture, provided it ruptured the same seismogenic fault. Dynamic triggering effects, fault geometry complexity, and stress localization may explain rupture within these shadow zones. Further southward rupture propagation of the 2025 rupture into the Bago segment is likely inhibited by a combination of geometric and pre‐stress barriers, preventing a larger rupture.
- 8 February 2026
- Published by American Geophysical Union (AGU) in Geophysical Research Letters
- Vol. 53 (3)
Open AccessCoulomb Pre‐Stress Changes Modulate Coseismic Rupture Kinematics of the 2025 Mw7.7 Myanmar Earthquake Revealed by Space Geodesy
We estimate the fault geometry and coseismic slip of the 2025 Myanmar earthquake using multi‐source satellite observations, revealing a nine‐segment rupture structure that transitions from eastward‐dipping in the north to westward‐dipping in the south, with peak slip of ∼6 m and negligible shallow slip deficit. We model Coulomb pre‐stress evolution along the Sagaing Fault from 1839 to 2025, incorporating coseismic stress transfer from 10 M ≥ 6.5 earthquakes and interseismic tectonic loading, to investigate their influences on the 2025 rupture. Results show that the 1839 Ava earthquake could have cast a long‐lasting pre‐stress shadow at both ends of the 2025 rupture, provided it ruptured the same seismogenic fault. Dynamic triggering effects, fault geometry complexity, and stress localization may explain rupture within these shadow zones. Further southward rupture propagation of the 2025 rupture into the Bago segment is likely inhibited by a combination of geometric and pre‐stress barriers, preventing a larger rupture.
- Open Access
- 26 November 2025
- Published by Springer Nature in Communications Earth & Environment
- Vol. 6 (1) , 971
Satellite radar advances carbon emissions accountability over tropical peat
- 26 November 2025
- Published by Springer Nature in Communications Earth & Environment
- Vol. 6 (1) , 971
Open AccessSatellite radar advances carbon emissions accountability over tropical peat
- Open Access
- 12 November 2025
- Published by American Geophysical Union (AGU) in Geophysical Research Letters
- Vol. 52 (22)
Coseismic Kinematics and Early Afterslip Evolution of the 2025 Mw ${\mathbf{M}}_{\mathbf{w}}$ 7.1 Dingri Earthquake (Southern Tibet) From Space GeodesyThe 2025 7.1 Dingri earthquake is a large normal‐faulting event within the Southern Tibet Rift System. We use space‐borne radar interferometry to investigate both coseismic rupture and ∼3‐month early afterslip. Results show that this event ruptured a west‐dipping (48°) fault, exhibiting a predominant normal dip‐slip of 4.7 m and a left‐lateral strike‐slip component of 3.2 m. Postseismic moment release ( 6.47) contributes 14.2% of the coseismic moment, predominantly (59%) from shallow afterslip (<28 km; maximum 0.3 m) exceeding the contribution from deep afterslip (<28 km; maximum 0.1 m). We find a depth‐dependent control on postseismic process: at 0–10 km, both 49% of aftershocks and substantial afterslip are driven by coseismic stress increases. Conversely, at 10–20 km depth, afterslip‐induced stress perturbations dominate, hosting 64% of aftershocks. The coseismic rupture kinematics reflect shallow crustal responses to deep tectonic processes, likely driven by the tearing of the Indian slab beneath southern Tibet.
- 12 November 2025
- Published by American Geophysical Union (AGU) in Geophysical Research Letters
- Vol. 52 (22)
Open AccessCoseismic Kinematics and Early Afterslip Evolution of the 2025 Mw ${\mathbf{M}}_{\mathbf{w}}$ 7.1 Dingri Earthquake (Southern Tibet) From Space GeodesyThe 2025 7.1 Dingri earthquake is a large normal‐faulting event within the Southern Tibet Rift System. We use space‐borne radar interferometry to investigate both coseismic rupture and ∼3‐month early afterslip. Results show that this event ruptured a west‐dipping (48°) fault, exhibiting a predominant normal dip‐slip of 4.7 m and a left‐lateral strike‐slip component of 3.2 m. Postseismic moment release ( 6.47) contributes 14.2% of the coseismic moment, predominantly (59%) from shallow afterslip (<28 km; maximum 0.3 m) exceeding the contribution from deep afterslip (<28 km; maximum 0.1 m). We find a depth‐dependent control on postseismic process: at 0–10 km, both 49% of aftershocks and substantial afterslip are driven by coseismic stress increases. Conversely, at 10–20 km depth, afterslip‐induced stress perturbations dominate, hosting 64% of aftershocks. The coseismic rupture kinematics reflect shallow crustal responses to deep tectonic processes, likely driven by the tearing of the Indian slab beneath southern Tibet.
- Open Access
- 5 August 2025
- Published by Springer Nature in Communications Earth & Environment
- Vol. 6 (1) , 1-11
Earthquake damage mapped more comprehensively and accurately by radar satellites than optical imagery
Summary: Following a disaster, responders need to rapidly assess the extent of the damage. The prevailing view is that very-high-resolution optical satellite images can provide more accurate estimates of building damage than lower-resolution synthetic aperture radar (SAR). However, we demonstrate that SAR-based damage proxy maps we produced after the 2023 Kahramanmaraş Türkiye earthquakes outperformed comparable maps derived from optical imagery. This finding held true for both a state-of-the art machine learning method and for visual interpretation. The SAR-based maps achieved an F1 performance score approximately twice as high as the optical-based maps (0.47 vs 0.24, 0.23 and 0.15). Additionally, SAR offers established advantages in both coverage and timeliness: SAR was able to image the entire affected area within ten days, whereas the very-high-resolution optical dataset covered only 5.4% of the SAR-covered area during the same timeframe. Using the largest ground dataset that we know of for any event, we show that the damage distribution captured by our SAR-based maps strongly correlates with the ground observations over a wide range of spatial scales from neighbourhoods (~1 km2) to provinces (~10,000 km2). Based on this experience, we argue that any reliable remote sensing-based damage assessment system should incorporate radar to complement other techniques.
- 5 August 2025
- Published by Springer Nature in Communications Earth & Environment
- Vol. 6 (1) , 1-11
Open AccessEarthquake damage mapped more comprehensively and accurately by radar satellites than optical imagery
Summary: Following a disaster, responders need to rapidly assess the extent of the damage. The prevailing view is that very-high-resolution optical satellite images can provide more accurate estimates of building damage than lower-resolution synthetic aperture radar (SAR). However, we demonstrate that SAR-based damage proxy maps we produced after the 2023 Kahramanmaraş Türkiye earthquakes outperformed comparable maps derived from optical imagery. This finding held true for both a state-of-the art machine learning method and for visual interpretation. The SAR-based maps achieved an F1 performance score approximately twice as high as the optical-based maps (0.47 vs 0.24, 0.23 and 0.15). Additionally, SAR offers established advantages in both coverage and timeliness: SAR was able to image the entire affected area within ten days, whereas the very-high-resolution optical dataset covered only 5.4% of the SAR-covered area during the same timeframe. Using the largest ground dataset that we know of for any event, we show that the damage distribution captured by our SAR-based maps strongly correlates with the ground observations over a wide range of spatial scales from neighbourhoods (~1 km2) to provinces (~10,000 km2). Based on this experience, we argue that any reliable remote sensing-based damage assessment system should incorporate radar to complement other techniques.
- Open Access
- 18 March 2025
- Published by Copernicus GmbH
Validating L-Band InSAR for Tropical Peat Motion Monitoring: A Transformative Approach for Peat Measurement, Reporting, and Verification
Peatlands are vital carbon stores but risk turning into carbon sources due to anthropogenic disturbance. In Southeast Asia, an estimated 65% of tropical peatlands have been drained, primarily for agriculture and plantation development. This drainage results in drying, subsidence, and oxidation of the peat, leading to substantial CO₂ emissions stemming from the belowground biomass. Quantifying carbon emissions from peatlands in this region is therefore crucial and growing in importance for peat Measurement, Reporting, and Verification (MRV). However, measurements remain sparse due to the vast, swampy, densely vegetated, and often inaccessible nature of these tropical landscapes. By measuring peat motion, which is a proxy for peat CO₂ emissions, L-band Interferometric Synthetic Aperture Radar (InSAR) offers a cost-effective monitoring solution with the ability to penetrate vegetation and cover large areas at millimeter-scale precision. Despite its potential, the accuracy of L-band InSAR has not been rigorously validated in Southeast Asia and results can vary significantly depending on the InSAR processing strategies taken.Here, we address this gap by using a unique combination of L-band ALOS-2 InSAR and high-accuracy, high-rate ground truth data from peat cameras to assess the InSAR accuracy for measuring peat motion. We use 5-year long ground measurements from 8 peat cameras distributed across oil palm plantations, croplands, scrublands, and forests in Central Kalimantan, Indonesia, to provide unprecedented evaluation of the accuracy of InSAR both temporally and spatially. We also mitigate major noise sources in ALOS-2 InSAR time series of ground motion stemming from the ionosphere, troposphere, and closure phase bias to improve the InSAR accuracy.Our results demonstrate consistency of InSAR with the long-term velocities of peat motion from ground truth data across all sites with a mean RMSE of < 5 mm/year. We also show high accuracy of InSAR (RMSE ~1.6 cm and Pearson’s correlation coefficient ~0.7) in detecting sub-annual fluctuations of peat displacements over plantations and croplands, and poorer performance over some scrublands and forests but that can be identified based on the InSAR coherence. The 5-year span of the InSAR and peat camera datasets also reveal subsidence trends driven by drainage and accelerated during El Niño events. These results show the suitability of L-band for both multi-annual and sub-annual monitoring of peat motion in the tropics. In addition, these findings demonstrate that L-band InSAR can be scaled up to provide regional to global coverage, especially with the advent of new and upcoming L-band satellites (e.g. ALOS-4 and NISAR). Frequent data acquisitions and accurate data processing and analysis are essential for MRV of peat carbon. Without reliable data, verifying carbon credits and restoration efforts becomes impossible, undermining market credibility and effective mitigation of climate change impact. This study thus validates InSAR as a cost-effective and scalable tool and contributes to the robust assessment of ongoing climate action.
- 18 March 2025
- Published by Copernicus GmbH
Open AccessValidating L-Band InSAR for Tropical Peat Motion Monitoring: A Transformative Approach for Peat Measurement, Reporting, and Verification
Peatlands are vital carbon stores but risk turning into carbon sources due to anthropogenic disturbance. In Southeast Asia, an estimated 65% of tropical peatlands have been drained, primarily for agriculture and plantation development. This drainage results in drying, subsidence, and oxidation of the peat, leading to substantial CO₂ emissions stemming from the belowground biomass. Quantifying carbon emissions from peatlands in this region is therefore crucial and growing in importance for peat Measurement, Reporting, and Verification (MRV). However, measurements remain sparse due to the vast, swampy, densely vegetated, and often inaccessible nature of these tropical landscapes. By measuring peat motion, which is a proxy for peat CO₂ emissions, L-band Interferometric Synthetic Aperture Radar (InSAR) offers a cost-effective monitoring solution with the ability to penetrate vegetation and cover large areas at millimeter-scale precision. Despite its potential, the accuracy of L-band InSAR has not been rigorously validated in Southeast Asia and results can vary significantly depending on the InSAR processing strategies taken.Here, we address this gap by using a unique combination of L-band ALOS-2 InSAR and high-accuracy, high-rate ground truth data from peat cameras to assess the InSAR accuracy for measuring peat motion. We use 5-year long ground measurements from 8 peat cameras distributed across oil palm plantations, croplands, scrublands, and forests in Central Kalimantan, Indonesia, to provide unprecedented evaluation of the accuracy of InSAR both temporally and spatially. We also mitigate major noise sources in ALOS-2 InSAR time series of ground motion stemming from the ionosphere, troposphere, and closure phase bias to improve the InSAR accuracy.Our results demonstrate consistency of InSAR with the long-term velocities of peat motion from ground truth data across all sites with a mean RMSE of < 5 mm/year. We also show high accuracy of InSAR (RMSE ~1.6 cm and Pearson’s correlation coefficient ~0.7) in detecting sub-annual fluctuations of peat displacements over plantations and croplands, and poorer performance over some scrublands and forests but that can be identified based on the InSAR coherence. The 5-year span of the InSAR and peat camera datasets also reveal subsidence trends driven by drainage and accelerated during El Niño events. These results show the suitability of L-band for both multi-annual and sub-annual monitoring of peat motion in the tropics. In addition, these findings demonstrate that L-band InSAR can be scaled up to provide regional to global coverage, especially with the advent of new and upcoming L-band satellites (e.g. ALOS-4 and NISAR). Frequent data acquisitions and accurate data processing and analysis are essential for MRV of peat carbon. Without reliable data, verifying carbon credits and restoration efforts becomes impossible, undermining market credibility and effective mitigation of climate change impact. This study thus validates InSAR as a cost-effective and scalable tool and contributes to the robust assessment of ongoing climate action.
- Open Access
- 12 December 2024
- Published by Copernicus GmbH in Natural Hazards and Earth System Sciences
- Vol. 24 (12) , 4585-4608
Automating tephra fall building damage assessment using deep learning
In the wake of a volcanic eruption, the rapid assessment of building damage is paramount for effective response and recovery planning. Uncrewed aerial vehicles, UAVs, offer a unique opportunity for assessing damage after a volcanic eruption, with the ability to collect on-demand imagery safely and rapidly from multiple perspectives at high resolutions. In this work, we established a UAV-appropriate tephra fall building damage state framework and used it to label ∼50 000 building bounding boxes around ∼2000 individual buildings in 2811 optical images collected during surveys conducted after the 2021 eruption of La Soufrière volcano, St Vincent and the Grenadines. We used these labelled data to train convolutional neural networks (CNNs) for (1) building localisation (average precision equals 0.728) and (2) damage classification into two levels of granularity: no damage vs. damage (F1 score = 0.809) and moderate damage vs. major damage (F1 score = 0.838) (1 is the maximum obtainable for both metrics). The trained models were incorporated into a pipeline along with all the necessary image processing steps to generate spatial data (a georeferenced vector with damage state attributes) for rapid tephra fall building damage mapping. Using our pipeline, we assessed tephra fall building damage for the town of Owia, finding that 22 % of buildings that received 50–90 mm of tephra accumulation experienced at least moderate damage. The pipeline is expected to perform well across other volcanic islands in the Caribbean where building types are similar, though it would benefit from additional testing. Through cross-validation, we found that the UAV look angle had a minor effect on the performance of damage classification models, while for the building localisation model, the performance was affected by both the look angle and the size of the buildings in images. These observations were used to develop a set of recommendations for data collection during future UAV tephra fall building damage surveys. This is the first attempt to automate tephra fall building damage assessment solely using post-event data. We expect that incorporating additional training data from future eruptions will further refine our model and improve its applicability worldwide. To facilitate continued development and collaboration all trained models and the pipeline code can be downloaded from GitHub.
- 12 December 2024
- Published by Copernicus GmbH in Natural Hazards and Earth System Sciences
- Vol. 24 (12) , 4585-4608
Open AccessAutomating tephra fall building damage assessment using deep learning
In the wake of a volcanic eruption, the rapid assessment of building damage is paramount for effective response and recovery planning. Uncrewed aerial vehicles, UAVs, offer a unique opportunity for assessing damage after a volcanic eruption, with the ability to collect on-demand imagery safely and rapidly from multiple perspectives at high resolutions. In this work, we established a UAV-appropriate tephra fall building damage state framework and used it to label ∼50 000 building bounding boxes around ∼2000 individual buildings in 2811 optical images collected during surveys conducted after the 2021 eruption of La Soufrière volcano, St Vincent and the Grenadines. We used these labelled data to train convolutional neural networks (CNNs) for (1) building localisation (average precision equals 0.728) and (2) damage classification into two levels of granularity: no damage vs. damage (F1 score = 0.809) and moderate damage vs. major damage (F1 score = 0.838) (1 is the maximum obtainable for both metrics). The trained models were incorporated into a pipeline along with all the necessary image processing steps to generate spatial data (a georeferenced vector with damage state attributes) for rapid tephra fall building damage mapping. Using our pipeline, we assessed tephra fall building damage for the town of Owia, finding that 22 % of buildings that received 50–90 mm of tephra accumulation experienced at least moderate damage. The pipeline is expected to perform well across other volcanic islands in the Caribbean where building types are similar, though it would benefit from additional testing. Through cross-validation, we found that the UAV look angle had a minor effect on the performance of damage classification models, while for the building localisation model, the performance was affected by both the look angle and the size of the buildings in images. These observations were used to develop a set of recommendations for data collection during future UAV tephra fall building damage surveys. This is the first attempt to automate tephra fall building damage assessment solely using post-event data. We expect that incorporating additional training data from future eruptions will further refine our model and improve its applicability worldwide. To facilitate continued development and collaboration all trained models and the pipeline code can be downloaded from GitHub.
- Top Cited Papers
- 23 August 2024
- Published by American Association for the Advancement of Science (AAAS) in Science
- Vol. 385 (6711) , 866-871
Slow rupture in a fluid-rich fault zone initiated the 2024 M w 7.5 Noto earthquake
The 2024 moment magnitude 7.5 Noto Peninsula (Japan) earthquake caused devastation to communities and was generated by a complex rupture process. Using space geodetic and seismic observations, we have shown that the event deformed the peninsula with a peak uplift reaching 5 meters at the west coast. Shallow slip exceeded 10 meters on an offshore fault. Peak stress drop was greater than 10 megapascals. This devastating event began with a slow rupture propagation lasting 15 to 20 seconds near its hypocenter, where seismic swarms had surged since 2020 because of lower-crust fluid supply. The slow start was accompanied by intense high-frequency seismic radiation. These observations suggest a distinct coseismic slip mode reflecting high heterogeneity in fault properties within a fluid-rich fault zone.
- 23 August 2024
- Published by American Association for the Advancement of Science (AAAS) in Science
- Vol. 385 (6711) , 866-871
Top Cited PapersSlow rupture in a fluid-rich fault zone initiated the 2024 M w 7.5 Noto earthquake
The 2024 moment magnitude 7.5 Noto Peninsula (Japan) earthquake caused devastation to communities and was generated by a complex rupture process. Using space geodetic and seismic observations, we have shown that the event deformed the peninsula with a peak uplift reaching 5 meters at the west coast. Shallow slip exceeded 10 meters on an offshore fault. Peak stress drop was greater than 10 megapascals. This devastating event began with a slow rupture propagation lasting 15 to 20 seconds near its hypocenter, where seismic swarms had surged since 2020 because of lower-crust fluid supply. The slow start was accompanied by intense high-frequency seismic radiation. These observations suggest a distinct coseismic slip mode reflecting high heterogeneity in fault properties within a fluid-rich fault zone.