HTHui Tang

Hui Tang

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Hui Tang has a total of 95 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Inverting Debris Flow Basal Impact Forces from Seismic Waveforms via Deep Learning

    by Kshitij Kar, Qi Zhou, Hui Tang, Fabian Walter, Zhen Zhang, Brian McArdell

    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.

  • Open Access

    Basal Force Probability Distributions in Thin‐Layer Granular Flows

    by Jun Fang, Hui Tang, Yifei Cui, Jens M. Turowski, Lu Jing, Clarence Edward Choi

    Extreme geophysical flows, such as granular and debris flows, can significantly shape the landscape in steep lands and generate seismic signals that can be recorded over long distances. However, direct field measurements needed to constrain the granular physics remain difficult due to the damage potential of those flows. Here, we investigate how the granular flows impact the bedrock channel using a computational fluid dynamics model coupled with the discrete‐element method, systematically varying the grain size, discharge, and water content for two flow regimes: unsteady and steady flows. These numerical experiments allow us to quantify the basal forces as probability density functions (PDFs). We test those PDFs based on several evaluation metrics, such as determination coefficient, residual sum of squares, Wasserstein distance, and information entropy for 12 selected probability distributions. We find that the t Location‐Scale and lognormal distributions are best‐performing among the tested distributions for unsteady and steady flows, respectively, rather than the exponential and generalized Pareto distributions in previous studies. Our results demonstrate that grain size, discharge, and water content strongly influence the basal‐force distribution. Furthermore, we established the relationships between flow kinematics and basal force statistics. Results indicate that compared to effective friction coefficient and volume fraction, the fitted parameters of the probability distributions exhibit monotonic relationships with Froude number, granular temperature, and dimensionless velocity. These results provide empirical constraints for modeling extreme geophysical flow incision and inversion of seismic signals generated by such flows.

  • Open Access

    Basal force distribution from steady fully-developed granular flows

    by Hui Tang, Jun Fang, Yifei Cui, Jens Turowski, Lu Jing, Yong Kong

    Understanding the impact of geophysical flows on the channel bed is essential for assessing erosion processes of bed material. In this study, the discrete element method (DEM) is used to simulate idealized, steady-state, fully-developed granular flows impacting the channel bed with systematically varying total particle number (1000-30000), grain size (2-16mm), and slope angle (28-34°) to investigate the probability distributions of the basal force. The probability density functions of the basal force, normalized to the mean force, were calculated and fitted with ten probability distributions. Four indices, namely R2, Residual Sum of Squares (RSS), Wasserstein distance, and information entropy, are introduced to evaluate the goodness of fit for each probability density distribution. By comparison, the broad probability density distribution of normalized basal force can be well-described by Gamma distributions (GD) with its shape and scale parameters. The shape parameter of the Gamma distribution is positively correlated with the total particle number and grain size, but negatively correlated with the slope angle. An opposite relationship is revealed in the scale parameter of the Gamma distribution. Additionally, we analyzed flow kinematics by calculating the coordination number, dimensionless velocity, shear rate, inertial number, and volume fraction, and linking these variables to the shape and scale parameters. The coordination number, shear rate, inertial number, and volume fraction serve as effective proxies for the shape and scale parameters, enabling interpretation of the statistical characteristics of monitored basal forces in geophysical mass flows.

  • Open Access

    Inequality and solutions in the use of artificial intelligence tools in education in China’s central and western regions

    by Hui Tang, Min Xu

    Sustainable development is a growing trend in higher education. Equity is a fundamental and inevitable requirement of education, promoting its sustainable development. However, new inequity problems were induced by the increased use of artificial intelligence (AI) tools in education. To empirically focus on under-researched regions and their clear linkage to equity concerns. A literature review and empirical investigation were conducted in this paper. First, we summarize the inequity and the solutions of using AI tools in education given in previous studies. Then, the usage of AI tools in education and the equitable awareness among teachers and students in five colleges in China’s central and western regions were studied using a questionnaire survey. The results showed that most of the college teachers and students like to use AI tools in education. However, most of them are not aware of the inequity caused by using AI tools in education. In addition, the inequality caused by using AI tools in education is real, especially for teachers in regions with different economic development levels. We advocate that governments and colleges should provide professional training about equitable awareness in using AI tools in education to teachers and students. In addition, a sound system of rules and regulations should be established to deal with rapidly evolving AI tool updates and abuses. These findings and suggestions provide critical insights for the sustainable development of higher education.

  • Role of Liquefied Deposition Layers in Modulating Seismic Wave Generation in Surge‐Type Debris Flows

    by Fengrun Jiang, Dongri Song, Xiaoyu Li, Wei Zhong, Junfeng Li, Sunil Poudyal, Qi Zhou, Hui Tang

    Surge‐type debris flows propagate as a sequence of surges, forming gradually thickening in situ deposition layers between surges that dynamically alter channel‐bed conditions. Seismic recordings from Jiangjia Ravine reveal a progressive attenuation of ground motion amplitude with surge sequence, despite comparable flow magnitudes—indicating a decoupling between flow scale and seismic response. We attribute this to the accumulation and liquefaction of inter‐surge deposition layers, rather than pre‐existing deposits. To quantify this mechanism, we adopt an effective transmission parameter ( ξ ) within a fluvial seismology‐based framework, and propose a sigmoid function linking ξ to normalized deposition layer thickness ( H *). This formulation significantly improves the prediction of seismic power spectral density (PSD) across surges and provides a transferable approach to characterize subsurface flow–bed interactions. Our findings underscore the critical role of bed structure evolution during flow in modulating debris‐flow‐induced seismic signals, with implications for real‐time monitoring and early warning in sediment‐rich catchments.


  • Open Access

    A Flow Hydrograph Estimation Method Based on Sediment‐Water Relationship in Channelized Debris Floods

    by Xiaojun Guo, Jianyi Cheng, Hui Tang, Yong Li, Siling Zhang

    Debris flood, a major hazard in mountainous areas, is characterized as a mixture of water and sediment primarily driven by hydrological processes. However, the relationship between water and sediment remains unclear, and accurately describing the hydrograph of such flows continues to pose significant challenges due to the complexity of the debris flood process and the difficulties associated with field monitoring. This study utilizes real‐time monitoring data from the Ergou catchment, China, combined with hydrological modeling, to investigate the relationship between water discharge and sediment influx during debris floods. Our results show that the instantaneous sediment flux exhibits considerable fluctuations at real‐time resolution, consistent with original measurements, following a Pareto distribution. As the temporal interval increases to half, one, and 5 min, the relationship gradually emerges as an exponential function, eventually revealing a near‐linear correlation between the total volumes of sediment and water throughout the flood. Based on these observations, we propose a framework integrating hydrological simulation of water flow, sediment‐water relationships to estimate the continuous hydrograph of debris floods. The errors for the peak discharge and the total volumes decrease with lower temporal resolution. Specifically, the errors of peak discharge at the 2‐min resolution are confined within the ranges of [−26%, 6%]. The findings contribute to improve understanding of debris flood processes on a watershed scale.

  • Open Access

    A Stream Power Based Sediment Entrainment Model Across Geophysical Flows Informed by Machine Learning

    by Xueqiang Lu, Gordon G. D. Zhou, Jens M. Turowski, Kahlil F. E. Cui, Hui Tang, Bo Cao, Yunxu Xie, Alessandro Pasuto, Yuting Zhao

    Sediment entrainment marks the initiation of particle motion on the bed surface and plays a crucial role in quantifying sediment transport. While the entrainment behavior might vary among different geophysical flows, the underlying mechanisms are often similar. To explore the shared dynamics, we compiled a global database of diverse geophysical flows: stream flow (SF), hyperconcentrated flow (HF), and debris flow (DF), obtained from field observations and laboratory experiments. We first validate existing Shield's number based entrainment frameworks but find them inadequate to account for entrainment fluxes of all flow types, particularly for the HF and DF, across a wide range of excess Shields number ( θ / θ c ) where θ c represents the critical value. Utilizing the random forest regression algorithm, we then proposed a new stream power ( ω ) dependent bursting area formula ( A P ) for all types of mass flows considered, resulting in a unified ω ‐based entrainment model. The revised model achieves an R 2 of 0.924, which is more than twice that of the θ ‐based model ( R 2 = 0.427) when applied to the same compiled database. This work provides valuable insights for improving sediment transport modeling, which are essential for developing effective river management strategies and optimizing the design of related infrastructure systems.


  • Open Access

    Enhancing Debris Flow Warning via Machine Learning Feature Reduction and Model Selection

    by Qi Zhou, Hui Tang, Clément Hibert, Małgorzata Chmiel, Fabian Walter, Michael Dietze, Jens M. Turowski

    The advent of machine learning has significantly improved the accuracy of identifying mass movements through the seismic waves they generate, making it possible to implement real‐time early warning systems for debris flows. However, we lack a profound understanding of the effective seismic features and the limitations of different machine learning models. In this work, we investigate eighty seismic features and three machine learning models for single‐station‐based binary debris flow classification and multi‐station‐based warning tasks. These seismic features, derived from physical and statistical knowledge of impact sources, are grouped into five sets: Benford's law, waveform, spectra, spectrogram, and network. The machine learning models belong to two families: two ensemble models, Random Forest and eXtreme Gradient Boosting (XGBoost); one recurrent neural network model, Long Short‐Term Memory (LSTM). We analyzed feature importance from the ensemble models and found that the number and even the types of seismic features are not critical for training an effective binary classifier for debris flow. When using models designed to capture patterns in sequential data rather than focusing on information only in one given window, using the LSTM does not significantly improve the performance of binary debris flow classification task over Random Forest and XGBoost. For the multi‐station‐based debris flow warning task, the LSTM model predicts debris flow probability more consistently and provides longer warning times. Our proposed framework simplifies machine learning‐driven debris flow classification and lays the foundation for affordable seismic signal‐driven early warning using a sparse seismic network.


  • Link Between the Impact Mechanisms of Granular Flow‐Dam Interaction and the Generated Seismic Signal: Insights From Laboratory Experiments

    by Xinzhi Zhou, Yifei Cui, Jun Fang, Hui Tang, Zhen Zhang, Shuofan Wang

    To manage debris flows, it is critical to both recognize potential impact mechanisms and evaluate, via physical models, associated impact forces. Seismic signals can be used to detect flows and infer flow properties, but inferring flow characteristics in channels with check dams and upslope deposits remains challenging. In this study, we conducted laboratory flume experiments to investigate the influence of varying flume inclination and pre‐retained material (deposition upslope of the barrier) height on flow characteristics and impact mechanisms, and analyzed their correlation with seismic signals. The objective was to examine the impact of deposition upslope of the barrier on interactions between debris flows and check dams by combining seismic and dynamic parameters. We found that a frequency domain feature‐based method can successfully filter out noisy signals. Results showed that it is possible to distinguish the impact mechanism of a granular flow in the presence of deposition upslope of the barrier from the seismic signals' attributes, that is, the peak signal amplitude envelope before granular flow overflow and its ratio to the peak amplitude of the control test without the barrier. Furthermore, the presence of check dams and upslope deposits in the debris flow channel significantly reduces the sensitivity of seismic signals to flow velocity. A key advantage of using the peak amplitude of the seismic signals generated by debris flows for identifying the impact mechanism is the potential for enhanced safety and cost‐effectiveness compared with contact monitoring instruments such as force plates.


  • Open Access

    When Geomorphology Meets Quantum Computing: a Quantum Machine Learning Model for Extreme Flows Classification 

    by Hui Tang

    Landslides, debris flows, hyperconcentrated flows, and floods are among the most dangerous natural hazards worldwide. One of the fundamental tasks for geomorphologists is to classify and identify the kinds of processes they observe in the field. The task is more challenging than it sounds, especially considering high-damage processes like debris flows and landslides. Meanwhile, multiple dimensionless numbers (e.g., Reynolds number and Einstein number) based on first-principle physics have been widely used to describe these natural flows. When we use these dimensionless numbers and datasets to classify the flow, we face a long-standing challenge in machine learning: the curse of dimensionality. One of the expertise for quantum machine learning methods (e.g., Quantum Support Vector Machine, QSVM) is to deal with such a high-dimensional dataset. Therefore, based on Quantum machine learning methods, we develop a framework to objectively define the type of natural flows using the dimensionless number. Our preliminary results show that the QSVM method has very similar outputs compared to classical SVM, but it is relatively slower than classical ones. Meanwhile, for the high-dimensional k-mean cluster, the Quantum K-mean model has shown different clusters compared to the classical version. In the future, we will develop a hybrid version combining classical K-mean with Quantum acceleration to understand different flow types.