Zhen Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Zhen Zhang has a total of 47 co-authors. Here are the top co-authors by number of shared publications:
Fabian Walter
Swiss Federal Institute for Forest Snow and Landscape Research WSL Birmensdorf Switzerland
5 shared publications- Siming He
Institute of Mountain Hazards and Environment, Chinese Academy of Sciences, Chengdu, China
5 shared publications - Brian W. McArdell
Swiss Federal Institute for Forest, Snow and Landscape Research WSL, 8903 Birmensdorf, Switzerland
3 shared publications - Małgorzata Chmiel
Université Côte d’Azur, Observatoire de la Côte d’Azur, Géoazur, CNRS, IRD, Sophia Antipolis, France
3 shared publications Siming He
Chinese Academy of Sciences Institute of Mountain Hazards and Environment Chengdu China
3 shared publications- Tjalling de Haas
Department of Physical Geography, Faculty of Geosciences Utrecht University Utrecht The Netherlands
3 shared publications
Latest Publications
- Open Access
- 1 November 2026
- Published by Wiley
Inverting Debris Flow Basal Impact Forces from Seismic Waveforms via Deep Learning
Debris flows usually generate large basal impact forces because of their gravity-driven nature and moving mass, posing catastrophic threats to mountain communities. This basal impact force is a key parameter for understanding debris-flow dynamics and offers insights for hazard mitigation. However, developing cost-effective and easy-to-maintain approaches to measure or estimate this force remains challenging. While theoretical models have been developed to invert basal impact force from seismic signals, data-driven deep learning approaches offer greater flexibility and scalability, yet their performance for this task remains underexplored. Here, we train a Long Short-Term Memory (LSTM) model and an eXtended LSTM (xLSTM) model to invert basal impact force from debris-flow seismic signals. Formulating the inversion as a regression task, we used 2019 seismic and impact-force data from seven debris-flow events at Illgraben (Switzerland) for training and validation. The 2020-2023 dataset, comprising 21 events, was used to evaluate the trained models and empirical relationships between basal impact force and physical properties. The xLSTM performs similarly to the baseline LSTM across metrics, including root mean squared error and peak inverted basal impact force. Peak-force estimates achieve a mean absolute percentage error of ≤ 5%, equivalent to ≤10 kN. Using xLSTM, we further find a strong correlation (Pearson's r = 0.867) between observed debris-flow volume and cumulative inverted impact force, supporting volume estimation without direct flow measurements. This work provides insights into using seismic signals to estimate debris-flow physical properties, supporting seismic-based early-warning systems that extend beyond event detection to characterize event magnitude and impacts.
- 1 November 2026
- Published by Wiley
Open AccessInverting Debris Flow Basal Impact Forces from Seismic Waveforms via Deep Learning
Debris flows usually generate large basal impact forces because of their gravity-driven nature and moving mass, posing catastrophic threats to mountain communities. This basal impact force is a key parameter for understanding debris-flow dynamics and offers insights for hazard mitigation. However, developing cost-effective and easy-to-maintain approaches to measure or estimate this force remains challenging. While theoretical models have been developed to invert basal impact force from seismic signals, data-driven deep learning approaches offer greater flexibility and scalability, yet their performance for this task remains underexplored. Here, we train a Long Short-Term Memory (LSTM) model and an eXtended LSTM (xLSTM) model to invert basal impact force from debris-flow seismic signals. Formulating the inversion as a regression task, we used 2019 seismic and impact-force data from seven debris-flow events at Illgraben (Switzerland) for training and validation. The 2020-2023 dataset, comprising 21 events, was used to evaluate the trained models and empirical relationships between basal impact force and physical properties. The xLSTM performs similarly to the baseline LSTM across metrics, including root mean squared error and peak inverted basal impact force. Peak-force estimates achieve a mean absolute percentage error of ≤ 5%, equivalent to ≤10 kN. Using xLSTM, we further find a strong correlation (Pearson's r = 0.867) between observed debris-flow volume and cumulative inverted impact force, supporting volume estimation without direct flow measurements. This work provides insights into using seismic signals to estimate debris-flow physical properties, supporting seismic-based early-warning systems that extend beyond event detection to characterize event magnitude and impacts.
- 16 October 2025
- Published by American Geophysical Union (AGU) in Journal of Geophysical Research: Earth Surface
- Vol. 130 (10)
Linking Dynamic Parameters and Seismic Signals of Granular Flows in Different Flow Regimes: An Experimental Assessment of Effects of Particle Composition
Geophysical flows, governed by particle composition and channel slope, exhibit distinct kinematic properties and seismic responses under different flow regimes. This study examines the impact of particle composition on granular flow dynamics and seismic signal generation under various flow regimes using flume experiments under dam break conditions. By varying particle composition and flume inclination angles, we investigate the kinematic properties, seismic responses, and the relationship between flow regimes and seismic signal characteristics. The results reveal that particle composition significantly affects flow dynamics, with peak velocity exhibiting a non‐monotonic dependence on particle size, and an optimal proportion of large particles maximizing mobility. Seismic signals, including peak amplitude and power spectral density, increase with larger particle sizes and steeper inclination angles, indicating a strong coupling between flow dynamics and seismic responses. A two‐segment positive correlation between seismic signals and collisional stress highlights the role of flow regimes, with particle‐ground impacts during intense collisional interactions dominating seismic signal generation, we then introduce a dimensionless amplitude parameter and establish a unified correlation with the Savage number across flow regimes. This study advances the understanding of granular flow dynamics and seismic signatures, providing a framework for interpreting seismic data in debris flow monitoring and hazard assessment. Future work to explore the interplay of frictional and collisional mechanisms to refine models of granular flow behavior and physical interpretation of seismic data is warranted.
- 16 October 2025
- Published by American Geophysical Union (AGU) in Journal of Geophysical Research: Earth Surface
- Vol. 130 (10)
Linking Dynamic Parameters and Seismic Signals of Granular Flows in Different Flow Regimes: An Experimental Assessment of Effects of Particle Composition
Geophysical flows, governed by particle composition and channel slope, exhibit distinct kinematic properties and seismic responses under different flow regimes. This study examines the impact of particle composition on granular flow dynamics and seismic signal generation under various flow regimes using flume experiments under dam break conditions. By varying particle composition and flume inclination angles, we investigate the kinematic properties, seismic responses, and the relationship between flow regimes and seismic signal characteristics. The results reveal that particle composition significantly affects flow dynamics, with peak velocity exhibiting a non‐monotonic dependence on particle size, and an optimal proportion of large particles maximizing mobility. Seismic signals, including peak amplitude and power spectral density, increase with larger particle sizes and steeper inclination angles, indicating a strong coupling between flow dynamics and seismic responses. A two‐segment positive correlation between seismic signals and collisional stress highlights the role of flow regimes, with particle‐ground impacts during intense collisional interactions dominating seismic signal generation, we then introduce a dimensionless amplitude parameter and establish a unified correlation with the Savage number across flow regimes. This study advances the understanding of granular flow dynamics and seismic signatures, providing a framework for interpreting seismic data in debris flow monitoring and hazard assessment. Future work to explore the interplay of frictional and collisional mechanisms to refine models of granular flow behavior and physical interpretation of seismic data is warranted.
- Open Access
- 13 October 2025
- Published by American Geophysical Union (AGU) in Geophysical Research Letters
- Vol. 52 (20)
Seismic Signals Monitor Debris‐Flow Erosion and Channel Elevation
Channel erosion not only amplifies debris‐flow magnitude and impact but also reshapes local geomorphology. However, the destructive and infrequent nature of debris flows makes in situ monitoring of channel‐bed erosion processes and flow characteristics challenging. Here, we investigate seismic signals for monitoring erosion‐driven geomorphic changes, using data from 18 well‐documented debris flows at Illgraben, Switzerland, between 2019 and 2023. We find that integrated seismically derived impact forces over each event correlate with channel‐bed elevation changes, revealing erosion thresholds. Seismic peak frequencies correlate with absolute channel‐bed elevations at seismic source regions, reflecting changes in wave propagation paths due to erosion. The correlation is evident, with peak frequency shifts exceeding 15 Hz while channel‐bed elevation changes were under 4 m during the 5‐year period. These findings demonstrate the capacity of seismic signals to characterize debris‐flow erosion and track absolute channel‐bed elevations, offering new insights into geomorphic processes.
- 13 October 2025
- Published by American Geophysical Union (AGU) in Geophysical Research Letters
- Vol. 52 (20)
Open AccessSeismic Signals Monitor Debris‐Flow Erosion and Channel Elevation
Channel erosion not only amplifies debris‐flow magnitude and impact but also reshapes local geomorphology. However, the destructive and infrequent nature of debris flows makes in situ monitoring of channel‐bed erosion processes and flow characteristics challenging. Here, we investigate seismic signals for monitoring erosion‐driven geomorphic changes, using data from 18 well‐documented debris flows at Illgraben, Switzerland, between 2019 and 2023. We find that integrated seismically derived impact forces over each event correlate with channel‐bed elevation changes, revealing erosion thresholds. Seismic peak frequencies correlate with absolute channel‐bed elevations at seismic source regions, reflecting changes in wave propagation paths due to erosion. The correlation is evident, with peak frequency shifts exceeding 15 Hz while channel‐bed elevation changes were under 4 m during the 5‐year period. These findings demonstrate the capacity of seismic signals to characterize debris‐flow erosion and track absolute channel‐bed elevations, offering new insights into geomorphic processes.
- Open Access
- 3 May 2025
- Published by Springer Nature in Landslides
- Vol. 22 (8) , 2701-2713
Analysing dynamics of the 2000 Yigong landslide in the Tibetan plateau using seismic observations
Climate change and urbanisation increase the risk of landslides, particularly mass movements with high mobility that can travel long distances and cause significant casualties and economic losses. Despite their impact, the mechanisms controlling long-runout landslides remain poorly understood due to their complexity. This study investigates the catastrophic 2000 Yigong landslide on the Tibetan plateau, using seismic signals to invert for the force–time function during its movement and evaluate changes in the basal friction coefficient. We identified five distinct stages in the event’s dynamics, beginning with two rock collapses in the source zone during the acceleration phase. In the entrainment phase, the mass attained its maximum velocity and forces when it impacted the Zhamu Creek. Notably, the equivalent friction coefficient dropped from 0.58 to 0.06, indicating a transition from rock collapse to debris flow. Our velocity-friction analysis aligns with flash heating theory, suggesting that high-speed sliding between rough surfaces generated heat, which partially melted the glacier and introduced fluid into the sliding process, resulting in enhanced mobility and facilitated long-runout distances down the Zhamu Creek. This study highlights the value of seismic signals in understanding the physical mechanisms behind catastrophic landslides and addressing the challenges of monitoring these geohazards in remote regions.
- 3 May 2025
- Published by Springer Nature in Landslides
- Vol. 22 (8) , 2701-2713
Open AccessAnalysing dynamics of the 2000 Yigong landslide in the Tibetan plateau using seismic observations
Climate change and urbanisation increase the risk of landslides, particularly mass movements with high mobility that can travel long distances and cause significant casualties and economic losses. Despite their impact, the mechanisms controlling long-runout landslides remain poorly understood due to their complexity. This study investigates the catastrophic 2000 Yigong landslide on the Tibetan plateau, using seismic signals to invert for the force–time function during its movement and evaluate changes in the basal friction coefficient. We identified five distinct stages in the event’s dynamics, beginning with two rock collapses in the source zone during the acceleration phase. In the entrainment phase, the mass attained its maximum velocity and forces when it impacted the Zhamu Creek. Notably, the equivalent friction coefficient dropped from 0.58 to 0.06, indicating a transition from rock collapse to debris flow. Our velocity-friction analysis aligns with flash heating theory, suggesting that high-speed sliding between rough surfaces generated heat, which partially melted the glacier and introduced fluid into the sliding process, resulting in enhanced mobility and facilitated long-runout distances down the Zhamu Creek. This study highlights the value of seismic signals in understanding the physical mechanisms behind catastrophic landslides and addressing the challenges of monitoring these geohazards in remote regions.
- Open Access
- 11 April 2025
- Published by Wiley
Deep Learning Supports Cross-Regional Debris Flow Warning: A Case Study of Two Regions
Debris flows pose significant threats to human life, property, and socioeconomic development. While deep learning combined with seismic signals shows promise for warning systems, the feasibility of cross-regional applications remains unexplored. This study develops DFlowNet, a hybrid deep learning framework that integrates convolutional neural networks with Transformer architecture to achieve efficient identification of debris flow initiation signals. Trained and tested using seismic data from Switzerland, DFlowNet achieves 100% warning precision without requiring complex warning rules. Remarkably, when directly transferred to China's debris flow prone regions for four months, the pre-trained model maintains 93.3% precision while detecting all 14 debris flow events. This work provides the first empirical evidence supporting the application of deep learning for cross-regional debris flow warning, offering a potential solution for regions lacking historical event records.
- 11 April 2025
- Published by Wiley
Open AccessDeep Learning Supports Cross-Regional Debris Flow Warning: A Case Study of Two Regions
Debris flows pose significant threats to human life, property, and socioeconomic development. While deep learning combined with seismic signals shows promise for warning systems, the feasibility of cross-regional applications remains unexplored. This study develops DFlowNet, a hybrid deep learning framework that integrates convolutional neural networks with Transformer architecture to achieve efficient identification of debris flow initiation signals. Trained and tested using seismic data from Switzerland, DFlowNet achieves 100% warning precision without requiring complex warning rules. Remarkably, when directly transferred to China's debris flow prone regions for four months, the pre-trained model maintains 93.3% precision while detecting all 14 debris flow events. This work provides the first empirical evidence supporting the application of deep learning for cross-regional debris flow warning, offering a potential solution for regions lacking historical event records.
- 17 January 2025
- Published by American Geophysical Union (AGU) in Journal of Geophysical Research: Earth Surface
- Vol. 130 (1)
Evolution of Basal Force Fluctuations and Seismic Signals of Granular Flows and Their Proxy: Insights From Laboratory Flume Experiments
The forces exerted by geophysical granular flows on Earth's surface, and the resulting seismic signals, can be used to monitor natural geohazards and understand their dynamic evolution and characteristics. Substantial research has focused on linking basal force fluctuations and seismic signals to granular flow dynamics. However, the mechanisms behind the generation and evolution of seismic signals remain incompletely understood. In this study, we conducted laboratory flume experiments to gain insights into the evolution and characteristics of basal force fluctuations and seismic signals and explored their relationship with the macroscopic properties of granular flows. Our results show that the shear and normal components of basal force fluctuations exhibit different behavior during flow evolution, which are related to variations in flow velocity fluctuations. As the granular flow moves downstream, shear basal force fluctuations decrease due to weakening velocity fluctuations, whereas normal force fluctuations increase. Similar to basal force fluctuations, seismic signals follow a generalized Pareto distribution. Basal force fluctuations and seismic signals are strongly nonlinearly related to the bulk flow properties, indicating that thicker, denser and faster flows generate stronger basal force fluctuations and more intense seismic signals. However, particle size significantly influences this relationship. We demonstrate that the inertial number, characterizing the macroscopic rheological properties of granular flows, can unify basal force fluctuations and seismic signals across different particle sizes, exhibiting a negative correlation on the temporal scale. This implies that the macroscopic rheological behavior of granular flows may provide critical insights into the mechanisms of generation and evolution of seismic signals.
- 17 January 2025
- Published by American Geophysical Union (AGU) in Journal of Geophysical Research: Earth Surface
- Vol. 130 (1)
Evolution of Basal Force Fluctuations and Seismic Signals of Granular Flows and Their Proxy: Insights From Laboratory Flume Experiments
The forces exerted by geophysical granular flows on Earth's surface, and the resulting seismic signals, can be used to monitor natural geohazards and understand their dynamic evolution and characteristics. Substantial research has focused on linking basal force fluctuations and seismic signals to granular flow dynamics. However, the mechanisms behind the generation and evolution of seismic signals remain incompletely understood. In this study, we conducted laboratory flume experiments to gain insights into the evolution and characteristics of basal force fluctuations and seismic signals and explored their relationship with the macroscopic properties of granular flows. Our results show that the shear and normal components of basal force fluctuations exhibit different behavior during flow evolution, which are related to variations in flow velocity fluctuations. As the granular flow moves downstream, shear basal force fluctuations decrease due to weakening velocity fluctuations, whereas normal force fluctuations increase. Similar to basal force fluctuations, seismic signals follow a generalized Pareto distribution. Basal force fluctuations and seismic signals are strongly nonlinearly related to the bulk flow properties, indicating that thicker, denser and faster flows generate stronger basal force fluctuations and more intense seismic signals. However, particle size significantly influences this relationship. We demonstrate that the inertial number, characterizing the macroscopic rheological properties of granular flows, can unify basal force fluctuations and seismic signals across different particle sizes, exhibiting a negative correlation on the temporal scale. This implies that the macroscopic rheological behavior of granular flows may provide critical insights into the mechanisms of generation and evolution of seismic signals.
- Open Access
- 8 April 2024
- Published by Springer Nature in Nature Communications
- Vol. 15 (1) , 1-10
Landslide hazard cascades can trigger earthquakes
While earthquakes are well-known to trigger surface hazards and initiate hazard cascades, whether surface hazards can instead trigger earthquakes remains underexplored. In 2018, two landslides on the Tibetan plateau created landslide-dammed lakes which subsequently breached and caused catastrophic outburst floods. Here we build an earthquake catalog using machine-learning and cross-correlation-based methods which shows there was a statistically significant increase in earthquake activity (local magnitude ≤ 2.6) as the landslide-dammed lake approached peak water level which returned to the background level after dam breach. We further find that ~90% of the seismicity occurred where Coulomb stress increased due to the combined effect of direct loading and pore pressure diffusion. The close spatial and temporal correlation between the calculated Coulomb stress increase and earthquake activity suggests that the earthquakes were triggered by these landslide hazard cascades. Finally, our Coulomb stress modeling considering the properties of landslide-dammed lakes and reservoir-induced earthquakes globally suggests that earthquake triggering by landslide-dammed lakes and similar structures may be a ubiquitous phenomenon. Therefore, we propose that earthquake-surface hazard interaction can include bidirectional triggering which should be properly accounted for during geological hazard assessment and management in mountainous regions.
- 8 April 2024
- Published by Springer Nature in Nature Communications
- Vol. 15 (1) , 1-10
Open AccessLandslide hazard cascades can trigger earthquakes
While earthquakes are well-known to trigger surface hazards and initiate hazard cascades, whether surface hazards can instead trigger earthquakes remains underexplored. In 2018, two landslides on the Tibetan plateau created landslide-dammed lakes which subsequently breached and caused catastrophic outburst floods. Here we build an earthquake catalog using machine-learning and cross-correlation-based methods which shows there was a statistically significant increase in earthquake activity (local magnitude ≤ 2.6) as the landslide-dammed lake approached peak water level which returned to the background level after dam breach. We further find that ~90% of the seismicity occurred where Coulomb stress increased due to the combined effect of direct loading and pore pressure diffusion. The close spatial and temporal correlation between the calculated Coulomb stress increase and earthquake activity suggests that the earthquakes were triggered by these landslide hazard cascades. Finally, our Coulomb stress modeling considering the properties of landslide-dammed lakes and reservoir-induced earthquakes globally suggests that earthquake triggering by landslide-dammed lakes and similar structures may be a ubiquitous phenomenon. Therefore, we propose that earthquake-surface hazard interaction can include bidirectional triggering which should be properly accounted for during geological hazard assessment and management in mountainous regions.
- Open Access
- 1 February 2024
- Published by American Geophysical Union (AGU) in Journal of Geophysical Research: Earth Surface
- Vol. 129 (2)
Seismic Monitoring and Geomorphic Impacts of the Catastrophic 2018 Baige Landslide Hazard Cascades in the Tibetan Plateau
Surface hazards can form hazard cascades which expand their reach. However, our understanding of their complex dynamics and ability to mitigate their impacts remain limited. In 2018, two landslides dammed the Jinsha River in the Tibetan plateau and formed landslide‐dammed lakes. Subsequent dam breaches prompted the evacuation of >120,000 people. An early warning system for floods using a regional seismic network has been proposed on the basis of catastrophic floods having been detected ∼100 km away, with seismic energy proportional to discharge. Surprisingly, we find that this catastrophic outburst flood was undetectable beyond a few kilometers, with peak seismic energy preceding peak discharge. We propose that river channel stability also controls seismic energy generation and should be considered for accurate monitoring of catastrophic floods. In contrast, we find that the various processes during dam breach can be well‐characterized seismically further away and provide warning ∼60 min before discharge exceeds monsoon flood levels. We also show that numerical modeling of dam breaches which typically lacks in situ measurements can benefit from incorporating seismic data as constraints. Finally, we show that this event drastically increased sediment fluxes ∼670 km downstream for years and may significantly reduce the capacity of hydropower plants. Our results reveal ways to improve early warning of catastrophic outburst floods and the need to consider surface hazards' long‐term impact when managing infrastructure in mountainous regions.
- 1 February 2024
- Published by American Geophysical Union (AGU) in Journal of Geophysical Research: Earth Surface
- Vol. 129 (2)
Open AccessSeismic Monitoring and Geomorphic Impacts of the Catastrophic 2018 Baige Landslide Hazard Cascades in the Tibetan Plateau
Surface hazards can form hazard cascades which expand their reach. However, our understanding of their complex dynamics and ability to mitigate their impacts remain limited. In 2018, two landslides dammed the Jinsha River in the Tibetan plateau and formed landslide‐dammed lakes. Subsequent dam breaches prompted the evacuation of >120,000 people. An early warning system for floods using a regional seismic network has been proposed on the basis of catastrophic floods having been detected ∼100 km away, with seismic energy proportional to discharge. Surprisingly, we find that this catastrophic outburst flood was undetectable beyond a few kilometers, with peak seismic energy preceding peak discharge. We propose that river channel stability also controls seismic energy generation and should be considered for accurate monitoring of catastrophic floods. In contrast, we find that the various processes during dam breach can be well‐characterized seismically further away and provide warning ∼60 min before discharge exceeds monsoon flood levels. We also show that numerical modeling of dam breaches which typically lacks in situ measurements can benefit from incorporating seismic data as constraints. Finally, we show that this event drastically increased sediment fluxes ∼670 km downstream for years and may significantly reduce the capacity of hydropower plants. Our results reveal ways to improve early warning of catastrophic outburst floods and the need to consider surface hazards' long‐term impact when managing infrastructure in mountainous regions.
- Open Access
- 27 March 2022
- Published by Copernicus GmbH
Linking characteristics of debris flows to their high frequency seismic signature: insights from field measurements and model predictions
Measuring debris-flow properties remains a significant challenge in studies of natural hazards. Recent works suggest that the seismic signals generated by debris flows can help analyze flow dynamics, but theoretical details for estimating bulk flow properties from seismic signals are not fully understood or comprehensively tested. Here, we invert basal force fluctuations on the torrent bed using high frequency seismic signals generated by 6 well-documented debris flows at Illgraben, Switzerland. Verified against independent measurements, our seismically-derived basal force fluctuations match well with the measured basal fluctuations at a force plate and correlate with the bulk flow properties, including flow depth and weight. We propose a physical model employing the multi-particle force chains and random single-particle impacts within a debris flow to simulate the generation of high frequency seismic signals. We find that the random impacts of single particles and of multi-particle force chains are active at the same time, and together they control the debris-flow’s basal force fluctuations. For different events and different positions within events, the relative contributions of single particle impacts and of multi-particle force chains dominating the basal fluctuations vary significantly and control the non-linear relation between the high-frequency seismic signal strength and the bulk flow characteristics. According to our model, fluctuating basal forces are strongly controlled by particle sizes and flow depth. Our results open new perspectives for the understanding of high frequency seismic signals generated by debris flows and the estimation of bulk flow characteristics, such as flow depth and weight.
- 27 March 2022
- Published by Copernicus GmbH
Open AccessLinking characteristics of debris flows to their high frequency seismic signature: insights from field measurements and model predictions
Measuring debris-flow properties remains a significant challenge in studies of natural hazards. Recent works suggest that the seismic signals generated by debris flows can help analyze flow dynamics, but theoretical details for estimating bulk flow properties from seismic signals are not fully understood or comprehensively tested. Here, we invert basal force fluctuations on the torrent bed using high frequency seismic signals generated by 6 well-documented debris flows at Illgraben, Switzerland. Verified against independent measurements, our seismically-derived basal force fluctuations match well with the measured basal fluctuations at a force plate and correlate with the bulk flow properties, including flow depth and weight. We propose a physical model employing the multi-particle force chains and random single-particle impacts within a debris flow to simulate the generation of high frequency seismic signals. We find that the random impacts of single particles and of multi-particle force chains are active at the same time, and together they control the debris-flow’s basal force fluctuations. For different events and different positions within events, the relative contributions of single particle impacts and of multi-particle force chains dominating the basal fluctuations vary significantly and control the non-linear relation between the high-frequency seismic signal strength and the bulk flow characteristics. According to our model, fluctuating basal forces are strongly controlled by particle sizes and flow depth. Our results open new perspectives for the understanding of high frequency seismic signals generated by debris flows and the estimation of bulk flow characteristics, such as flow depth and weight.