Development of an online seismic disaster monitoring system in Yalong river

Rapid post-earthquake assessment of large dams is essential in seismically active mountain regions. We develop an online dam-safety monitoring framework for cascade hydropower stations in the Yalong River Basin. Two earthquake risk indices ($R_1$ and $R_2$), derived from a reservoir seismic network and a dam strong-motion system, trigger intensified monitoring with prioritized data acquisition and an automated preliminary report within about 15 min after shaking; a concise post-earthquake brief is produced within about 30 min when the workflow is completed. An elastoplastic deformation monitoring model is coupled with static and dynamic structural analyses to evaluate displacement/stress states and infer potential damage. A three-dimensional early-warning matrix integrates monitoring, inspection, and structural-calculation results to provide graded safety assessment. Basin-scale application demonstrates efficient decision support and helps quantify dam–valley rock-mass responses in a complex seismotectonic setting.

1 Introduction

With the continuous implementation of the dual-carbon strategy, hydropower, as a clean and renewable energy source with strong peak-shaving capability, plays an increasingly important role in optimizing the energy structure and ensuring the stable operation of power systems (Chu and Majumdar, 2012; Adams and Acheampong, 2019; Jurasz et al., 2020; Hou et al., 2023; Amiri and Gueniat, 2024). Southwest China is rich in hydropower resources and has therefore become a concentrated region for the construction of high dams and large reservoirs (Huang, 2012; Chong et al., 2016; Wang et al., 2017; Tu and Deng, 2020). At the same time, this region lies in one of China’s major earthquake zones, characterized by frequent seismicity and a high probability of strong events (Lei et al., 2017; Del Gaudio et al., 2019; Lei et al., 2019; Meng et al., 2019; Atkinson et al., 2020; Liu and Zahradník, 2020; Huang et al., 2021). For example, the Mw 8.0 Wenchuan earthquake on 12 May 2008 caused structural damage to the Zipingpu Dam and posed a potential threat to the safety of Chengdu (Yang, 2012), and the 5 September 2022 Luding Ms 6.9 earthquake induced a plastic deformation increment of up to 20.04 mm in the crown beam of the Dagangshan extra-high arch dam (Huibao et al., 2023).

Worldwide, dams have experienced seismic damage when ground motions approach or exceed design levels, and even when overall stability is maintained, local cracking and seepage may still occur—highlighting the need for timely post-earthquake assessment for emergency response and mitigation (Gao and Chen, 2001).

In mountainous river basins, earthquakes can jointly affect dams and surrounding slopes, forming a coupled disaster chain that interacts with reservoir operations (Huang, 2012; Chong et al., 2016; Wang et al., 2017; Tu and Deng, 2020). The Yalong River Basin lies along the eastern Tibetan Plateau margin in an active tectonic–geomorphic setting (Clark et al., 2004; Wissink et al., 2016; Deng et al., 2018), where frequency–magnitude screening provides basin-scale context (Salari, 2025) and robust numerical seismic-response tools are widely emphasized (Nikvand and Hajialilue Bonab, 2025).

In the field of seismic safety of hydraulic structures, extensive research has been devoted to dynamic input at dam sites, the seismic response of the dam–reservoir–bedrock system, the dynamic mechanical behavior of dam and foundation materials, and the seismic reliability and safety evaluation of structures (Zou et al., 2012; Chuhan et al., 2016; Liu et al., 2016; Ren et al., 2016; Lingao, 2017; Houqun, 2018; Jiang et al., 2018; Xu et al., 2018; Xu and Feng, 2018; 2019; Song et al., 2022; Li et al., 2023; Pang et al., 2023; Park and You, 2023). These studies have greatly advanced our understanding of dam seismic behavior and provided a sound basis for design. However, compared with analysis and design, research on post-earthquake acquisition of dam safety monitoring information, rapid structural assessment, monitoring theory and early-warning mechanisms remains relatively limited, and operation-period safety monitoring and early-warning technology for dams under earthquake conditions is still a blind spot.

The Yalong River Basin sits in an active seismotectonic setting with steep gorges and unstable slopes. At individual dam sites, foundation and abutment rock masses are commonly cut by persistent discontinuities (faults, joint sets, and locally bedding/foliation), which control deformability, potential sliding kinematics, and the coupled dam–foundation–abutment response under seismic loading. This engineering–geological context motivates a basin-wide operational framework that can rapidly link seismic triggering, monitoring interpretation and numerical diagnosis, while also clarifying that site-specific geological conditions constrain the applicability and interpretation of the system outputs.

Based on this background, this paper focuses on cascade hydropower stations in the Yalong River Basin and presents research and engineering practice on dam safety management under seismic conditions. The objectives are to improve the response efficiency and effectiveness of basin-scale dam safety information systems during earthquakes, to develop a technical framework for dam safety monitoring under seismic loading, and to propose an early-warning scheme that integrates monitoring, inspection and structural analysis results to support decision-making for earthquake-resistant dam operation and management.

2 The online monitoring system of Yalong river basin

The Yalong River, the largest tributary of the Jinsha River, has a concentrated elevation drop in its middle and lower reaches and rich hydropower resources, with a total exploitable capacity of about 30 million kilowatts. Together with planned wind and solar projects, the total installed capacity in the basin is expected to exceed 100 million kilowatts. More than 10 hydropower stations have been built or are under construction, including several world-class high dams such as Jinping, Ertan and Lianghekou (Huang et al., 2002; Chen et al., 2007; Zheng et al., 2011; Xiong et al., 2019; He et al., 2021).

The basin lies in the middle segment of the north–south seismic zone within the Sichuan–Yunnan diamond-shaped block and involves three main geological–tectonic units. Large-scale and rapid uplift of the Tibetan Plateau and intense neotectonic movement make boundary fault zones between different blocks highly active, leading to frequent earthquakes (Clark et al., 2004; Wissink et al., 2016; Gourbet et al., 2017; Zhang et al., 2017; Deng et al., 2018). Two strong earthquake belts, the Daofu–Fuhuo zone in the middle and upper reaches and the Anning River Valley zone in the lower reaches, control basin seismicity. As shown in Figure 1, the Yalong River Basin is located in a high seismic intensity zone with a high probability and frequency of strong earthquakes. The combination of high relief, weak rock units and active faults has produced numerous large landslides and unstable slopes (Huang, 2012; Wang et al., 2017), so dam safety monitoring must be closely coupled to the surrounding geological environment. From an engineering geology perspective, the cascade dams are mostly located in narrow, steep-sided gorges, where abutment and foundation rock masses exhibit pronounced spatial heterogeneity due to weathering gradients, crushed faulted zones and locally weaker interlayers. Key discontinuities—such as fault/shear zones and multiple joint sets—may provide preferential paths for deformation localization and can govern the kinematic feasibility of earthquake-induced rock-slope instabilities around reservoirs. Therefore, in practical deployment, the monitoring layout, allowable deformation thresholds and the interpretation of the monitoring–simulation results are calibrated on a station-by-station basis using site investigation and design documents, so that the framework remains operational while being constrained by local rock-mass structure and discontinuity conditions.

Figure 1

Figure 1. Distribution of hydropower station and earthquakes in Yalong River Basin. (a) The hydropower stations of Yalong river. (b) Historical earthquakes with seismic intensity over magnitude.

To manage dam safety under these conditions, Yalong River Hydropower Development Company initiated the Yalong River Basin Dam Safety Information Management System (idam system) in 2011, which was commissioned in 2013 as the first basin-level dam safety information platform in China. After more than 10 years of operation, it has become a unified platform for managing dam safety information for all hydropower stations in the basin, integrating multi-source data such as safety monitoring, inspection and checking, hydrological and meteorological information, strong-motion records and flood-discharge vibration. For dam operation under seismic conditions, the monitoring and early-warning architecture based on this platform is organized into four layers: a hardware perception layer for data acquisition, a system response layer for event triggering and data transmission, a monitoring and analysis layer for automatic data processing and model calculation, and a judgment and warning layer for comprehensive evaluation and decision support for both dam structures and adjacent valley rock masses in this complex seismotectonic and geomorphic setting.

3 Key techniques for the system development

3.1 Seismic information flow system design

As the engineer’s “eyes and ears”, safety monitoring is the most direct means of capturing dam structural behavior. Hydropower stations in the Yalong River Basin are equipped with automated systems; at Jinping-I, for example, more than 18,000 monitoring points are in normal operation, covering deformation, appearance inspection, seepage and uplift pressure, stress and strain, temperature and other key indicators, as illustrated in Figure 2.

Figure 2

Figure 2. Main monitoring projects of the dam safety monitoring system at the Jinping level.

To capture seismic effects, the dams are also instrumented with dedicated strong-motion systems. In total, 73 strong-motion sensors are installed in the hub areas of seven hydropower stations. For a typical high arch dam such as Jinping-I, instruments are distributed from the crest to lower galleries and the foundation, and a free-field station is deployed near the dam site. This “structural array + free field” configuration records both the input motion and its spatial variation within the dam.

Accurate, robust acquisition of seismic information and its real-time transfer to the safety monitoring system are essential during earthquakes. Because large events may interrupt public communication networks, seismic information is obtained within the local area network (LAN) of each station and then pushed to the dam safety information management system. To avoid incomplete acquisition from a single source, the LAN integrates two components: a reservoir seismic station network and the dam strong-motion monitoring system. The seismic station network provides real-time magnitude, focal depth and distance to the dam site, and a seismic risk score $R_1$ is defined in Equation 1:

$R_1={\displaystyle \sum _{i=1}^N}{w}_m{M}_i+w_d\frac{1}{D_i+\delta }+w_l{e}^{-L_i}(1)$

where $w_m$, $w_d$, and $w_l$ are the weighting coefficients of the earthquake magnitude $M_i$, depth $D_i$, and relative distance $L_i$ from the hydropower plant, respectively, $\delta$ is a constant, and $e^{{-L}_i}$ represents the exponential attenuation of seismic influence with distance. The coefficients ($w_m$, $w_d$, $w_l$) are dimensionless weights used to balance the relative contributions of magnitude, focal depth and source-to-site distance in a transparent, engineering-oriented triggering score. The rationale is that magnitude primarily controls the potential shaking level, focal depth modulates near-field effects, and distance governs attenuation and is explicitly represented by the exponential decay term. In implementation, the weights are normalized ($w_m+w_d+w_l=1$) and are calibrated using recorded earthquakes in the basin together with expert judgement from dam operation engineers, so that the triggering behavior is consistent with the operational practice of “intensified monitoring after potentially influential events”. In practice, a station-specific threshold $T_{R1}$ is configured in the idam system; when $R_1\ge T_{R1}$, the event is regarded as potentially influential and an earthquake-linkage task is automatically created to initiate intensified monitoring.

The strong-motion system mainly records three-directional accelerations of hydraulic structures. For risk assessment, sensors are grouped into four zones. The free-field recorder constitutes zone A with weight $q_1$; sensors installed in the dam are assigned to zones B, C and D with weights $q_2$, $q_3$ and $q_4$. There are $n$ recorders in zone B, $m$ in zone C and $l$ in zone D. The strong-motion-based score $R_2$ is defined in Equation 2:

$R_2=q_{1}A+{\displaystyle \sum _{i=1}^n}{q}_2{B}_i+{\displaystyle \sum _{j=1}^m}{q}_3{C}_j+{\displaystyle \sum _{k=1}^l}{q}_4{D}_k(2)$

with

$q_1=n\times q_2+m\times q_3+l\times q_4$

Here $A$ is the representative parameter (e.g., peak ground acceleration) at the free-field station and $B_i$, $C_j$, and $D_k$ are the corresponding parameters for sensors in zones B, C and D. The zonation and weights ($q_1$, $q_2$, $q_3$, $q_4$) are introduced to reflect the representativeness of the free-field input motion (zone A) and the structural response at different dam elevations (zones B–D). This design avoids over-reliance on a single sensor and reduces the influence of sensor-level noise by within-zone aggregation. In practice, the zone weights are normalized and are selected according to the monitoring layout and the relative importance of response-sensitive regions for post-earthquake diagnosis. The representative intensity measure $P$ (e.g., PGA) follows the routine strong-motion practice in dam safety management, and the triggering threshold for $R_2$ is set to match the operational criterion for launching the rapid post-earthquake analysis workflow. When $R_2$ exceeds a predefined threshold, the dam safety information management system is triggered and a strong-motion safety analysis workflow is initiated. Meanwhile, the same trigger launches the prioritized acquisition of the important-measuring-point set and the template-based preliminary reporting described below.

Once seismic excitation is detected and the triggering conditions based on $R_1$ and $R_2$ are satisfied, the subsequent response of the monitoring system controls the timeliness and completeness of information acquisition and structural diagnosis. An earthquake-linkage inspection task is automatically created: an important-measuring-point set is inspected with priority, followed by a rapid inspection of all monitoring points. The important-measuring-point set is selected in accordance with the Technical Specification for Hydropower Dam Operation Safety Online Monitoring System (DL/T 2096–2020).

As a backup under the centralized management mode of cascade dams, a remote call-measurement function is provided. Through a unified remote collection control protocol, commands can be issued from the basin-level idam system to each station-level monitoring system, which then starts local acquisition services and returns data. This mechanism enhances robustness when local automation services are interrupted and supports rapid, basin-wide data retrieval after earthquakes.

Following the “1 dam, one policy” principle, each station pre-defines strong-motion response point sets and report templates, and the overall linkage flow is shown in Figure 3. Once triggering conditions are met, the basin-level idam platform initiates an earthquake-linkage task with prioritized acquisition of the important-measuring-point set, rapid collection of remaining points, and a remote call-measurement fallback; data are then uploaded and processed automatically to generate a concise post-earthquake brief within the prescribed window. Virtual earthquake tests across operating stations show that, except for one station with hardware limitations, most stations can return key monitoring information within about 15 min and generate a concise operation/safety brief within about 30 min, providing operational evidence for the end-to-end timeliness and completeness of the linkage chain, complementary to the field demonstration in Section 4.

Figure 3

Figure 3. System linkage information flow.

3.2 Online numerical simulation technology

Under seismic conditions, rapid evaluation of dam structural behavior requires timely characterization of the deformation field and identification of whether different parts of the structure remain elastic or have experienced plastic damage. To this end, a deformation monitoring model is established to separate the main components of dam deformation and to provide boundary conditions for online numerical simulation.

Under ground motion loading, dams may exhibit pronounced elastoplastic deformation. The elastic component recovers within a relatively short time and is difficult to capture by conventional monitoring, whereas the plastic component, also referred to as “residual deformation” or “additional long-lasting deformation” (Shengwu and Dewen, 2009; Xuexing et al., 2010; Huibao et al., 2023), is irreversible and therefore crucial for structural diagnosis. In this work, the earthquake-induced deformation term ${\delta }_v$ refers to the post-event residual (permanent) deformation increment inferred after the transient elastic vibration decays, rather than the full time-history vibrational response. This definition is consistent with the practical objective of rapid post-earthquake diagnosis, where irreversible deformation is the key indicator for potential damage. High-frequency elastic vibration is instead characterized by the strong-motion monitoring system and is further examined through the subsequent three-dimensional dynamic analyses. Accordingly, two complementary analysis streams are implemented after triggering: a nonlinear static updating analysis that uses the monitoring-inferred residual displacement field to update the post-event stress/deformation state of the dam–foundation–abutment system, and a nonlinear dynamic analysis that uses recorded accelerograms from the strong-motion system to examine transient structural responses. The measured dam deformation $\delta$ is decomposed into four components: hydraulic deformation ${\delta }_H$, temperature-induced deformation ${\delta }_T$, time-dependent deformation ${\delta }_{\theta }$ and an earthquake-induced deformation ${\delta }_v$, as shown in Equation 3:

$\delta ={\delta }_H+{\delta }_T+{\delta }_{\theta }+{\delta }_v(3)$

The first three components correspond to conventional statistical models of dam deformation, while the seismic component ${\delta }_v$ represents the plastic deformation increment. Here “plastic” is used in an operational sense to denote the irreversible increment observable in routine deformation monitoring; the step-like representation is adopted to capture the typical post-earthquake jump in residual deformation that remains after shaking. Because this increment typically shows a step-like evolution in time, a time-segmented function $f\left(x\right)$ is introduced to describe the step process, as shown in Equation 4. For the $k$-th significant earthquake occurring at time ${\theta }_k$,

$f\left(x\right)=\left\{\begin{array}{c}0x<0\\ 1x\ge 0\end{array}\right.(4)$

and the cumulative seismic component is expressed as shown in Equation 5:

${\delta }_v={\displaystyle \sum _k}{b}_{k}f\left(\theta -{\theta }_k\right)(5)$

where $\theta$ and ${\theta }_k$ are the cumulative numbers of days from the initial measurement date to the monitoring date and to the date of the $k$-th strongest earthquake, respectively, both divided by 100, and $b_k$ are regression coefficients representing the plastic deformation increments from each event.

Combining the hydraulic, thermal, temporal and seismic contributions, the deformation model at a given monitoring point is written in Equation 6:

$\begin{array}{ll}\hfill \delta =& {\displaystyle \sum _i}\left[a_{1i}\left(H_u^i-H_{\mathrm{u}0}^i\right)\right]+{\displaystyle \sum _{j=1}^2}\left[b_{1j}\left(\mathit{sin}\frac{2\pi jt}{365}-\mathit{sin}\frac{2\pi j{t}_0}{365}\right)+b_{2j}\left(\mathit{cos}\frac{2\pi jt}{365}-\mathit{cos}\frac{2\pi j{t}_0}{365}\right)\right]\hfill \\ \hfill & +c_1\left(\theta -{\theta }_0\right)+c_2\left(\mathit{ln}\theta -{\mathit{ln}\theta }_0\right)+{\displaystyle \sum _k}{b}_{k}f\left(\theta -{\theta }_k\right)+a_0\hfill \end{array}(6)$

where $H_u$ and $H_{\mathrm{u}0}$ are the reservoir levels at the monitoring day and the initial measurement day, respectively; $t$ is the cumulative number of days at the monitoring date, and $t_0$ is the cumulative number of days at the beginning of the modelling interval; $\theta =t/100$, ${\theta }_0=t_0/100$; and $a_{1i},b_{1j},b_{2j},c_1,c_2,a_0$ are regression parameters. The deformation monitoring model is calibrated separately for each dam and each monitoring point using multi-year time series, so the effects of site-specific material properties and boundary conditions are implicitly embedded in the fitted coefficients of the hydraulic, thermal and time-dependent terms. Under seismic conditions, the earthquake-induced component is identified as a residual increment after removing these baseline contributions, which reduces reliance on explicitly prescribing uncertain constitutive parameters in a purely forward simulation.

Because strong earthquakes at a given dam site are episodic, model validation in an operational setting should combine real-event demonstrations with controlled tests under known inputs. Accordingly, we use “virtual earthquake” residual deformation increments as prescribed inputs to provide a verifiable benchmark for the monitoring model under seismic conditions. To verify the capability of the monitoring model under seismic conditions, two earthquakes are simulated as virtual plastic deformation inputs to the crown beam at the crest of the Jinping-I arch dam. Virtual Earthquake two is referenced to the maximum radial displacement increment of the Dagangshan arch dam during the 5 September 2022 Luding Ms 6.8 event. The imposed plastic deformation increments are 8.96 mm and 20.04 mm, while the model estimates 8.68 mm and 19.87 mm, giving a total seismic deformation increment error of less than 0.5 mm. From the initial impoundment to April 2024, the model also separates 13.49 mm of time-dependent deformation at the crown beam location, yielding a total plastic deformation of 42.04 mm. On this basis, plastic deformation monitoring indices can be defined using multi-year variability or multi-year cumulative time-dependent deformation as reference levels. When an earthquake occurs, the model is activated at the system level to isolate earthquake-induced plastic deformation and compare it with these indices, thereby supporting comprehensive safety assessment and operational decision-making.

Static and dynamic structural analyses are then used as an effective numerical tool for evaluating dam safety during earthquakes. Taking the Jinping-I dam as an example, online automatic three-dimensional nonlinear analyses are performed with boundary conditions derived from monitoring data. Because the boundary conditions are derived from monitoring-inferred permanent displacements rather than assumed constraints, the online analyses are constrained by observed response of the dam–foundation–abutment system, which helps reduce sensitivity to uncertainty in boundary-condition specification. Permanent deformations at key seismic control points are first obtained from the deformation model, and the limited observations are then spatially densified using an interpolation algorithm based on radial basis functions (RBFs) (Wenzhuo et al., 2023) to generate continuous displacement boundary conditions along the abutments and foundation, capturing the influence of steep valley geometry and heterogeneous rock masses. Here, the RBF-interpolated displacement field is used purely as a kinematic boundary condition to spatially densify sparse monitoring observations along the abutments and foundation. The monitoring-derived permanent displacements already reflect the integrated mechanical response of the dam–foundation–abutment system under the local geological setting (e.g., lithology, jointing, faults and valley morphology). Therefore, the interpolation itself is not intended to reconstruct geological heterogeneity; instead, geological complexity is represented in the three-dimensional numerical model through zoned material parameters and, where applicable, explicit discontinuity/interface descriptions.

In the RBF method, the basis function at a point $\left(x,y\right)$ is expressed as $h_k\left(x,y\right)=h\left(d_k\right)$, where $d_k$ is the distance from $\left(x,y\right)$ to the $k$-th data point. To improve polynomial accuracy, the interpolant is written as shown in Equation 7:

$F\left(x,y\right)={\displaystyle \sum _{k=1}^n}{a}_k{h}_k\left(x,y\right)+{\displaystyle \sum _{k=1}^m}{b}_k{q}_k\left(x,y\right)(7)$

where $q_k\left(x,y\right)$ is a polynomial basis of degree less than $m$, and $a_k,b_k$ are coefficients. Incorporating the spatial coordinates of the monitoring points and extending the independent variables yields a three-dimensional interpolation function suitable for variables $\left(x,y,z\right)$, as shown in Equation 8:

$F\left(x,y,z\right)={\displaystyle \sum _{j=1}^n}{a}_j{\left[{\left(x-x_j\right)}^2+{\left(y-y_j\right)}^2+{\left(z-z_j\right)}^2+c^2\right]}^{1/2},i=1,2,\cdots n(8)$

where $c$ is a shape parameter. Prior studies indicate that excessively large values of $c$ relative to the spacing between data points may lead to an ill-conditioned coefficient matrix and inaccurate solutions, whereas small values generally yield more stable results. In this study, $c$ is taken as one-tenth of the minimum distance between monitoring points. Using the interpolated permanent deformation field as boundary conditions, the three-dimensional nonlinear static and dynamic calculations of the dam structure can be updated online to evaluate post-earthquake stress and deformation states of the integrated dam–foundation–abutment rock mass system. This spacing-based scaling links the shape parameter to the local data density on the abutment/foundation surfaces and helps avoid over-parameterization when the monitoring points are unevenly distributed along a steep valley. Nevertheless, in a heterogeneous gorge setting the interpolated field may smooth localized deformation gradients associated with major discontinuities or stiffness contrasts, and the interpolation accuracy is sensitive to point layout and any extrapolation beyond the instrumented region. To reduce such effects, the interpolation is applied only to generate boundary conditions within the monitored abutment/foundation domain, and the subsequent three-dimensional analysis accounts for heterogeneity through the constitutive zoning and structural/discontinuity representation of the dam–foundation–abutment system.

3.3 Structural diagnosis and early warning methods

After an earthquake, information for evaluating dam structural safety mainly comes from three sources: dam safety monitoring data, inspection and checking results, and three-dimensional numerical structural calculations (including static and dynamic analyses). These three sources are mapped onto a three-dimensional coordinate system and combined in an early-warning matrix, which feeds diagnosis results back to dam safety managers in a unified manner.

3.3.1 Monitoring and inspection axes

Through the system described above, hydropower stations in the Yalong River Basin can obtain safety monitoring data immediately after an earthquake, and the deformation monitoring model under seismic conditions provides online indicators for detecting abnormalities. According to the response characteristics of monitoring data, deformation results are classified into local and overall anomalies.

For single monitoring points, a local anomaly index $m_1$ is defined in Equation 9:

$m_1=\frac{\text{Plastic}\text{deformation}\text{increment}{\delta }_v}{\text{Plasticity}\text{allowable}\text{deformation}\text{value}{\delta }_{v.\mathit{max}}}(9)$

where ${\delta }_v$ is the earthquake-induced residual (permanent) deformation increment identified by the monitoring model and ${\delta }_{v.\mathit{max}}$ is the allowable plastic deformation at that point.

To characterize the overall state of the deformation field, a multi-point model is established by subregions or using a panel model (Jimin et al., 2020). Suppose there are $h$ monitoring points describing the deformation field. According to their importance, weights ${\eta }_1,{\eta }_2,{...,\eta }_h$ are assigned, and ${\sum }_{l=1}^h{\eta }_l=1$. The overall anomaly index $m_2$ is then defined in Equation 10:

$\left\{\begin{array}{l}{m}_2=2\times \frac{{\sum }_{l=1}^h{\eta }_l{\delta }_v^l}{{\sum }_{l=1}^h{\eta }_l{\delta }_{v.\mathit{max}}^l}\\ {\sum }_{l=1}^h{\eta }_l=1\end{array}\right.(10)$

where ${\delta }_v^l$ and ${\delta }_{v.\mathit{max}}^l$ are the plastic deformation increment and allowable plastic deformation at the $l$-th point. By combining $m_1$ and $m_2$, the monitoring information is classified into four levels: normal (A), slight abnormality (B), controllable abnormality (C) and serious abnormality (D).

Hydraulic inspection is another key means to obtain the operational status of dams and other hydraulic structures. According to DL/T2204-2020, when a significant tremor occurs at the dam site, especially when an earthquake of magnitude five or above occurs within 100 km, a special inspection should be carried out immediately. Inspection information (records, text, photos and videos) is standardized and structured through the basin-level dam inspection system, and results are divided into four grades: normal (A), slight abnormality (B), general abnormality (C) and severe abnormality (D). These are mapped to an inspection index $p$ for use in the early-warning matrix.

3.3.2 Structural analysis and early-warning matrix

According to DL/T2096-2020, the structural performance of dams should be reviewed under extraordinary operating conditions and relevant load combinations. Therefore, the outputs of three-dimensional static and dynamic structural calculations are an important information source for structural diagnosis when larger earthquakes occur. At the system level, the results obtained through Section 3.2 are categorized into five grades from normal to severely abnormal, denoted as A, B, C, D and E, and mapped to a structural calculation index $n$. Specifically, this index is derived from the monitoring-constrained nonlinear static updating and the record-based nonlinear dynamic evaluation performed in Section 3.2, which together provide complementary evidence for post-earthquake structural diagnosis.

By combining the monitoring axis, inspection axis and structural calculation axis, a three-dimensional spatial coordinate system is constructed for comprehensive judgment of dam safety. Based on this coordinate system, an early-warning matrix is defined, in which the post-earthquake warning value $V$ is expressed as a function $G$ of the monitoring index $m$, inspection index $p$ and structural calculation index $n$, as shown in Equation 11:

$V=G\left(m,p,n\right)(11)$

In this study, we adopt an interpretable and implementation-oriented fusion rule in the discrete grade space, i.e., $V=m+n+p$. The rationale is that the three axes (monitoring, inspection and structural calculations) serve as complementary screening channels in post-earthquake operations, and a transparent rule is preferred for rapid triage and traceable decision-making. The additive rule is intentionally conservative in the sense that an elevated level in any channel increases $V$ and therefore cannot be fully masked by the others. We note that relative importance and uncertainty may vary across dams, events and data quality; therefore, the framework is readily extensible to a weighted form $V=w_{M}m+w_{I}n+w_{S}p$ ($w_M+w_I+w_S=1$) or to probabilistic/knowledge-based fusion. Given the limited number of labeled post-earthquake cases available for robust training and calibration at present, we use the equal-weight additive rule in this paper, while the above extensions are left for future work. The three indices are converted from their respective grades by assigning A = 1, B = 2, C = 3, D = 4 and E = 4. According to the value of $V$, the post-earthquake warning level is blue for $V\le 5$, yellow for $5<V\le 8$, orange for $8<V\le 11$, and red for $11<V\le 13$. In practical implementation, these thresholds are indexed in the three-dimensional coordinate system to obtain the corresponding warning level.

In principle, additional geological indicators such as slope displacement, rockfall inventories or mapped landslide activity around the reservoir can be incorporated into the same framework, enabling joint assessment of earthquake-induced damage to both dams and surrounding geological units.

4 Performance

Validation of an online post-earthquake monitoring framework is inherently event-dependent, since strong ground motions at a specific dam site may not occur within a limited observation period. In this paper, we therefore present a multi-level validation strategy: (i) basin-wide virtual-earthquake linkage tests to verify the reliability and timeliness of the triggering–acquisition–reporting chain across operating stations; (ii) controlled verification of the elastoplastic deformation monitoring model using prescribed “virtual earthquake” residual deformation increments with known magnitudes; and (iii) an end-to-end field demonstration during the 27 May 2024 M5.0 event, which confirms that the system can be automatically triggered and can deliver consistent early-warning outputs under real operational constraints. The following case study focuses on the field demonstration, while Sections 3.1–3.2 provide the complementary verification evidence for the linkage and modelling components.

On 27 May 2024, an M5.0 earthquake occurred in Muli County, Liangshan Prefecture, Sichuan Province, China (28.25°N, 100.70°E), with a focal depth of 8 km. The epicenter was located approximately 91 km from the Jinping-I Hydropower Station. Because of this distance, the seismic wave energy was significantly attenuated when it reached the dam site. The strong-motion monitoring system recorded a peak ground acceleration (PGA) of 1.93 gal at station SM17, with a shaking duration of 31.75 s and a predominant frequency of 3.32 Hz. The corresponding displacement time history and response spectrum at SM17 are shown in Figure 4a. The reference seismic intensity at the dam site was evaluated as II on the Chinese Seismic Intensity Scale, indicating a weak tremor. The event occurred within the tectonically active mountain belt along the eastern margin of the Tibetan Plateau, where steep rock slopes and thick valley fill characterize the local geological setting (Wissink et al., 2016; Deng et al., 2018).

Figure 4

Figure 4. (a) Displacement and response spectrum at strong-motion station SM17 during the 27 May 2024 M5.0 Muli earthquake. (b) Three-dimensional dam safety early-warning matrix under seismic conditions.

For this event, the strong-motion information was automatically collected and processed by the basin-level platform. According to the predefined triggering and evaluation procedures, the monitoring, inspection and structural analysis indices place the post-earthquake state of the Jinping-I dam at a blue (safe) level in the three-dimensional dam safety assessment and early-warning matrix, as shown in Figure 4b. No earthquake-induced damage or abnormal deformation was identified and no emergency intervention was required. This case thus serves as an event-level verification of the matrix decision logic under real monitoring, inspection and analysis inputs.

5 Discussion

5.1 Methodological advantages and implications for seismotectonic and geohazard environments

The framework developed in this study couples basin-scale monitoring, data-driven modelling and three-dimensional numerical analysis into a unified workflow for earthquake-related dam safety management. Compared with practices that focus mainly on single-dam response or design-stage analyses, the present system introduces two complementary seismic risk indices, $R_1$ and $R_2$, which jointly exploit a reservoir seismic station network and dam strong-motion records for event triggering. Combined with intensified post-earthquake patrols and centralized data acquisition on the idam platform, this dual-source design enables cascade hydropower stations in the Yalong River Basin to obtain key monitoring information within about 15 min after an event and to rapidly update the status of critical indicators.

The proposed workflow is transferable in principle to other dam–reservoir systems where seismic/strong-motion and dam monitoring are available; however, site-specific calibration is required for key-point selection, $R_1$/ $R_2$ thresholds/weights, allowable deformation limits, and early-warning mapping rules.

On the structural side, the elastoplastic deformation monitoring model separates earthquake-induced plastic deformation from hydraulic, thermal and time-dependent components, linking long-term monitoring data to potential damage in the dam–foundation–abutment rock mass system. By interpolating permanent deformation at control points using radial basis functions, continuous displacement boundary conditions are generated along steep abutments and deeply incised valley walls and used to drive three-dimensional static and dynamic calculations. This monitoring–simulation chain allows post-earthquake stress and deformation fields to be updated online rather than relying solely on pre-computed scenarios, providing a practical route for numerical “back analysis” of earthquake effects during operation. The 27 May 2024 M5.0 Muli earthquake case shows that, even for relatively weak shaking at the dam site, the system can automatically capture the event, process the data and confirm a blue (safe) state in the early-warning matrix without additional manual intervention.

From an Earth science perspective, the Yalong River Basin is an active tectonic–geomorphic setting with unstable slopes and landslides (Clark et al., 2004; Wissink et al., 2016; Deng et al., 2018). While the current implementation focuses on dam safety, the three-axis early-warning matrix can be extended to include surrounding geohazard indicators (e.g., slope displacement or landslide activity), providing a practical prototype for integrated, basin-scale post-earthquake assessment.

5.2 Limitations and future work

Real-earthquake validation remains limited: the reported case is an M5.0 event with weak shaking at the dam site, and the virtual scenarios in the deformation model use only a small set of assumed plastic increments. System behavior under stronger shaking and more complex loading histories has not yet been verified with field data, and allowable plastic-deformation thresholds still partly rely on engineering judgement; therefore, broader calibration/validation using additional recorded events (including higher-intensity cases at dam sites and other cascade reservoirs) is needed.

Data gaps, limited sensor coverage, and computational burden may delay the structural-calculation axis in real time. The framework therefore prioritizes conservative monitoring/inspection-based triage and updates the warning state as additional data and numerical results become available.

Methodologically, several components adopt simplified formulations. The risk scores $R_1$ and $R_2$ use linear weighting with fixed thresholds and do not explicitly quantify uncertainties in source parameters, attenuation, or instrument performance. The deformation model assumes step-like plastic increments at discrete earthquake times, so progressive or multi-event cumulative damage (e.g., cyclic degradation or creep–damage interaction) may be underestimated. In the numerical simulation, RBF interpolation currently uses a fixed shape parameter and lacks explicit error quantification in heterogeneous abutments and irregular valley geometries, especially where discontinuities dominate deformation. Future work will (i) conduct sensitivity analyses for weights/thresholds using more recorded events, and (ii) explore uncertainty-aware fusion (e.g., Bayesian updating or evidence-theory-based fusion) and geology-constrained interpolation with uncertainty envelopes (e.g., cross-validation/resampling).

Looking forward, the framework could be extended by using data-driven methods to calibrate weights/thresholds in the risk scores and warning matrix, integrating additional geohazard indicators (e.g., GNSS/InSAR slope displacement and landslide/rockfall information), and embedding the workflow into a basin-scale digital twin that couples monitoring with real-time operation data for scenario-based emergency decision support.

6 Conclusion

This study proposes and implements a basin-wide framework for earthquake-resistant dam safety management in the Yalong River Basin, integrating seismic monitoring, deformation modelling and numerical simulation within a unified early-warning scheme. The main conclusions are:

1. By jointly using a reservoir seismic station network and dam strong-motion monitoring within the local area network, the framework provides dual-source seismic triggering and rapid post-earthquake data acquisition, enabling automated intensified post-earthquake patrols and allowing cascade hydropower stations to obtain key safety monitoring information within minutes after an event.

2. The elastoplastic deformation monitoring model, combined with radial-basis-function interpolation and three-dimensional static–dynamic structural analysis, establishes a monitoring–simulation chain that separates earthquake-induced plastic deformation from other components and updates dam–foundation–abutment rock mass stress and deformation fields online to support identification of potential damage.

3. By structuring monitoring, inspection and structural calculation results into a three-axis early-warning matrix, the framework achieves multi-source information fusion and graded post-earthquake safety assessment. Applied to the tectonically active Yalong River Basin and demonstrated through the 27 May 2024 event, it provides an operational prototype linking dam engineering practice with regional assessment of earthquake-induced geological hazards in mountainous river systems.

Data availability statement

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

Author contributions

JL: Software, Writing – original draft, Methodology, Resources, Investigation, Validation, Project administration, Writing – review and editing, Formal Analysis. PZ: Writing – review and editing, Investigation, Data curation, Methodology, Conceptualization, Resources, Writing – original draft, Validation, Visualization. CZ: Investigation, Writing – review and editing, Writing – original draft, Data curation, Formal Analysis, Resources, Methodology, Validation. Q-xM: Investigation, Writing – original draft, Methodology, Conceptualization, Project administration, Supervision, Writing – review and editing. ZH: Writing – review and editing, Resources, Funding acquisition, Writing – original draft, Project administration, Visualization, Investigation, Methodology, Software. TF: Visualization, Validation, Resources, Writing – review and editing, Software.

Funding

The author(s) declared that financial support was received for this work and/or its publication. Supported by Postgraduate Research & Practice Innovation Program of Jiangsu Province (SJCX25_0268).

Conflict of interest

Authors JL, PZ, and CZ were employed by Yalong River Hydropower Development Company Limited. Author TF was employed by PowerChina Huadong Engineering Corporation Limited.

The remaining author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

Adams, S., and Acheampong, A. O. (2019). Reducing carbon emissions: the role of renewable energy and democracy. J. Clean. Prod. 240, 118245. doi:10.1016/j.jclepro.2019.118245

CrossRef Full Text | Google Scholar

Amiri, M. H. N., and Gueniat, F. (2024). Towards a framework for measurements of power systems resiliency: comprehensive review and development of graph and vector-based resilience metrics. Sustain. Cities Soc. 109. doi:10.1016/j.scs.2024.105517

CrossRef Full Text | Google Scholar

Atkinson, G. M., Eaton, D. W., and Igonin, N. (2020). Developments in understanding seismicity triggered by hydraulic fracturing. Nat. Rev. Earth and Environ. 1 (5), 264–277. doi:10.1038/s43017-020-0049-7

CrossRef Full Text | Google Scholar

Chen, T., Cui, P., and Chen, X. (2007). Prediction of soil erosion on different underlaying surface in construction period of xichang to Panzhihua expressway. Wuhan Univ. J. Nat. Sci. 12 (4), 699–704. doi:10.1007/s11859-006-0321-9

CrossRef Full Text | Google Scholar

Chong, S., De-jie, L., Kai-hua, C., and Jia-wen, Z. (2016). Failure mechanism and stability analysis of the Zhenggang landslide in Yunnan Province of China using 3D particle flow code simulation. J. Mt. Sci. 13 (5), 891–905. doi:10.1007/s11629-014-3399-0

CrossRef Full Text | Google Scholar

Chu, S., and Majumdar, A. (2012). Opportunities and challenges for a sustainable energy future. Nature 488 (7411), 294–303. doi:10.1038/nature11475

PubMed Abstract | CrossRef Full Text | Google Scholar

Chuhan, Z., Feng, J., Jinting, W., Xu, Y., and Jianwen, P. (2016). Key issues and research progress in seismic safety evaluation of high concrete dams. J. Water Resour. 47 (03), 253–264. doi:10.13243/j.cnki.slxb.20151204

CrossRef Full Text | Google Scholar

Clark, M., Schoenbohm, L., Royden, L., Whipple, K., Burchfiel, B., Zhang, X., et al. (2004). Surface uplift, tectonics, and erosion of eastern Tibet from large-scale drainage patterns. Tectonics 23 (1). doi:10.1029/2002tc001402

CrossRef Full Text | Google Scholar

Del Gaudio, V., Zhao, B., Luo, Y., Wang, Y., and Wasowski, J. (2019). Seismic response of steep slopes inferred from ambient noise and accelerometer recordings: the case of Dadu River valley, China. Eng. Geol. 259, 105197. doi:10.1016/j.enggeo.2019.105197

CrossRef Full Text | Google Scholar

Deng, B., Chew, D. M., Jiang, L., Mark, C., Cogn, N., Wang, Z., et al. (2018). “Heavy mineral analysis and detrital U-Pb ages of the intracontinental Paleo-Yangzte basin, eastern margin of Tibetan Plateau,” in Heavy mineral analysis and detrital U-Pb ages of the intracontinental Paleo-Yangzte basin, eastern margin of Tibetan Plateau (Bremen, Germany: PANGAEA – Data Publisher for Earth & Environmental Science). doi:10.1594/PANGAEA.886236

CrossRef Full Text | Google Scholar

Gao, L., and Chen, J. (2001). Seismic safety evaluation of concrete dams. J. Water Resour. (02), 8–15. doi:10.13243/j.cnki.slxb.2001.02.002

CrossRef Full Text | Google Scholar

Gourbet, L., Leloup, P. H., Paquette, J.-L., Sorrel, P., Maheo, G., Wang, G., et al. (2017). Reappraisal of the Jianchuan Cenozoic basin stratigraphy and its implications on the SE Tibetan plateau evolution. Tectonophysics 700, 162–179. doi:10.1016/j.tecto.2017.02.007

CrossRef Full Text | Google Scholar

He, Y., Huang, D., Li, S., Shi, L., Sun, W., Sanford, R. A., et al. (2021). Profiling of microbial communities in the sediments of Jinsha River watershed exposed to different levels of impacts by the vanadium industry, Panzhihua, China. Microb. Ecol. 82 (3), 623–637. doi:10.1007/s00248-021-01708-9

PubMed Abstract | CrossRef Full Text | Google Scholar

Hou, G., Muraleetharan, K. K., Panchalogaranjan, V., Moses, P., Javid, A., Al-Dakheeli, H., et al. (2023). Resilience assessment and enhancement evaluation of power distribution systems subjected to ice storms. Reliab. Eng. and Syst. Saf. 230, 108964. doi:10.1016/j.ress.2022.108964

CrossRef Full Text | Google Scholar

Houqun, C. (2018). 60 years of seismic research on hydraulic concrete structures. J. China Inst. Water Resour. Hydropower Res. 16 (05), 322–330. doi:10.13244/j.cnki.jiwhr.2018.05.001

CrossRef Full Text | Google Scholar

Huang, R. (2012). Mechanisms of large-scale landslides in China. Bull. Eng. Geol. Environ. 71 (1), 161–170. doi:10.1007/s10064-011-0403-6

CrossRef Full Text | Google Scholar

Huang, J., Zhao, D., and Zheng, S. (2002). Lithospheric structure and its relationship to seismic and volcanic activity in southwest China. J. Geophys. Res. Solid Earth 107 (B10), 3–14. doi:10.1029/2000JB000137

CrossRef Full Text | Google Scholar

Huang, H., Dai, S., and Xie, F. (2021). Monitoring in-situ seismic response on rock slopes using ambient noise interferometry: application to the 2019 changning (Mw 5.7) earthquake, China. Front. Earth Sci. 8, 8–2020. doi:10.3389/feart.2020.610181

CrossRef Full Text | Google Scholar

Huibao, H., Gang, C., and Dejun, J. (2023). Characterization of the deformation of Dagangshan extra-high arch dam during the Sichuan Luding Ms6.8 earthquake. J. Water Resour. 54 (05), 599–609. doi:10.13243/j.cnki.slxb.20230017

CrossRef Full Text | Google Scholar

Jiang, S.-H., Papaioannou, I., and Straub, D. (2018). Bayesian updating of slope reliability in spatially variable soils with in-situ measurements. Eng. Geol. 239, 310–320. doi:10.1016/j.enggeo.2018.03.021

CrossRef Full Text | Google Scholar

Jimin, W., Chongshi, G., Chen, Z., and Jian, L. (2020). Analysis of deformation behavior of Jinping first-level dam based on panel spatio-temporal modeling. J. Hydropower Generation 39 (11), 21–30.

Google Scholar

Jurasz, J., Canales, F. A., Kies, A., Guezgouz, M., and Beluco, A. (2020). A review on the complementarity of renewable energy sources: concept, metrics, application and future research directions. Sol. Energy 195, 703–724. doi:10.1016/j.solener.2019.11.087

CrossRef Full Text | Google Scholar

Lei, X., Huang, D., Su, J., Jiang, G., Wang, X., Wang, H., et al. (2017). Fault reactivation and earthquakes with magnitudes of up to Mw4.7 induced by shale-gas hydraulic fracturing in Sichuan Basin, China. Sci. Rep. 7 (1), 7971. doi:10.1038/s41598-017-08557-y

PubMed Abstract | CrossRef Full Text | Google Scholar

Lei, X., Wang, Z., and Su, J. (2019). The December 2018 ML 5.7 and January 2019 ML 5.3 earthquakes in South Sichuan Basin Induced by Shale Gas Hydraulic Fracturing. Seismol. Res. Lett. 90 (3), 1099–1110. doi:10.1785/0220190029

CrossRef Full Text | Google Scholar

Li, Y., Pang, R., and Xu, B. (2023). Dynamic reliability analysis of three-dimensional slopes considering the spatial variability in soil parameters. Structures 56, 104977. doi:10.1016/j.istruc.2023.104977

CrossRef Full Text | Google Scholar

Lingao (2017). Seismic analysis and safety evaluation of dams. Hydropower Pumped Storage 3 (02), 14–27.

Google Scholar

Liu, J., and Zahradník, J. (2020). The 2019 MW 5.7 changning earthquake, Sichuan Basin, China: a shallow doublet with different faulting styles. Geophys. Res. Lett. 47 (4), e2019GL085408. doi:10.1029/2019GL085408

CrossRef Full Text | Google Scholar

Liu, J., Liu, F., Kong, X., and Yu, L. (2016). Large-scale shaking table model tests on seismically induced failure of Concrete-Faced Rockfill Dams. Soil Dyn. Earthq. Eng. 82, 11–23. doi:10.1016/j.soildyn.2015.11.009

CrossRef Full Text | Google Scholar

Meng, L., McGarr, A., Zhou, L., and Zang, Y. (2019). An investigation of seismicity induced by hydraulic fracturing in the Sichuan Basin of China based on data from a temporary seismic network. Bull. Seismol. Soc. Am. 109 (1), 348–357. doi:10.1785/0120180310

CrossRef Full Text | Google Scholar

Nikvand, V., and Hajialilue Bonab, M. (2025). Seismic response analysis of tunnels using a coupled polygonal finite element method. Civ. Geoengin. Lett. 1, 100009–100010. doi:10.22034/CGEL.1.1.e100009

CrossRef Full Text | Google Scholar

Pang, R., Zhou, Y., Chen, G., Jing, M., and Yang, D. (2023). Stochastic mainshock-aftershock simulation and its applications in dynamic reliability of structural systems via DPIM. J. Eng. Mech. 149 (1), 04022096. doi:10.1061/(asce)em.1943-7889.0002176

CrossRef Full Text | Google Scholar

Park, D., and You, H. (2023). A digital twin dam and watershed management platform. Water 15 (11), 2106. doi:10.3390/w15112106

CrossRef Full Text | Google Scholar

Ren, A., Wang, Y., Chen, Z., Song, Y., and Sun, P. (2016). Performance of the reinforced right abutment slope of Zipingpu Dam during magnitude 8.0 earthquake, Wenchuan, China. Q. J. Eng. Geol. Hydrogeology 49 (4), 298–307. doi:10.1144/qjegh2015-033

CrossRef Full Text | Google Scholar

Salari, F. (2025). Seismic hazard assessment of Minab City using the preliminary Gutenberg-Richter method. Civ. Geoengin. Lett. 2, e100044. doi:10.22034/CGEL.2.2.e100044

CrossRef Full Text | Google Scholar

Shengwu, S., and Dewen, C. (2009). Seismic phenomena and deformation monitoring analysis of Zipingpu concrete panel rockfill dam in Wenchuan earthquake. J. Rock Mech. Eng. 28 (04), 840–849.

Google Scholar

Song, L., Pang, R., Xu, B., and Zhou, Y. (2022). Dynamic reliability analysis of the stability of the 3D slope of a rockfill dam based on the Copula function and generalized probability density evolution method. Int. J. Geomechanics 22 (5), 04022038. doi:10.1061/(asce)gm.1943-5622.0002341

CrossRef Full Text | Google Scholar

Tu, G., and Deng, H. (2020). Formation and evolution of a successive landslide dam by the erosion of river: a case study of the Gendakan landslide dam on the Lancang River, China. Bull. Eng. Geol. Environ. 79 (6), 2747–2761. doi:10.1007/s10064-020-01743-9

CrossRef Full Text | Google Scholar

Wang, S. N., Xu, W. Y., Shi, C., and Chen, H. J. (2017). Run-out prediction and failure mechanism analysis of the Zhenggang deposit in southwestern China. Landslides 14 (2), 719–726. doi:10.1007/s10346-016-0770-y

CrossRef Full Text | Google Scholar

Wenzhuo, K., Fei, K., and Zhi, M. (2023). Intelligent optimized radial basis function network model for multi-parameter inversion of extra-high arch dams. Hydropower Energy Sci. 41 (11), 69–72. doi:10.20040/j.cnki.1000-7709.2023.20230213

CrossRef Full Text | Google Scholar

Wissink, G. K., Hoke, G. D., Garzione, C. N., and Liu-Zeng, J. (2016). Temporal and spatial patterns of sediment routing across the southeast margin of the Tibetan Plateau: insights from detrital zircon. Tectonics 35 (11), 2538–2563. doi:10.1002/2016tc004252

CrossRef Full Text | Google Scholar

Xiong, F., Guo, S., Liu, P., Xu, C.-Y., Zhong, Y., Yin, J., et al. (2019). A general framework of design flood estimation for cascade reservoirs in operation period. J. Hydrology 577, 124003. doi:10.1016/j.jhydrol.2019.124003

CrossRef Full Text | Google Scholar

Xu, J., and Feng, D.-C. (2018). Seismic response analysis of nonlinear structures with Uncertain parameters under stochastic ground motions. Soil Dyn. Earthq. Eng. 111, 149–159. doi:10.1016/j.soildyn.2018.04.023

CrossRef Full Text | Google Scholar

Xu, J., and Feng, D.-C. (2019). Stochastic dynamic response analysis and reliability assessment of non-linear structures under fully non-stationary ground motions. Struct. Saf. 79, 94–106. doi:10.1016/j.strusafe.2019.03.002

CrossRef Full Text | Google Scholar

Xu, B., Wang, X., Pang, R., and Zhou, Y. (2018). Influence of strong motion duration on the seismic performance of high CFRDs based on elastoplastic analysis. Soil Dyn. Earthq. Eng. 114, 438–447. doi:10.1016/j.soildyn.2018.08.004

CrossRef Full Text | Google Scholar

Xuexing, C., Yunlong, H., Koon, X., and Bin, L. (2010). Impact analysis of Wenchuan earthquake on Yeler dam. Geotechnics 31 (11), 3542–3548. doi:10.16285/j.rsm.2010.11.021

CrossRef Full Text | Google Scholar

Yang, Z. (2012). Seismic damage analysis of Zipingpu panel rockfill dam and panel seismic countermeasures for Wenchuan earthquake.

Google Scholar

Zhang, Z., Daly, J. S., Li, C. a., Tyrrell, S., Sun, X., and Yan, Y. (2017). Sedimentary provenance constraints on drainage evolution models for SE Tibet: evidence from detrital K-feldspar. Geophys. Res. Lett. 44 (9), 4064–4073. doi:10.1002/2017GL073185

CrossRef Full Text | Google Scholar

Zheng, H., Zhao, X., Zhao, T., Chen, F., Xu, W., Duan, X., et al. (2011). Spatial-temporal variations of methane emissions from the Ertan hydroelectric reservoir in southwest China. Hydrol. Process. 25 (9), 1391–1396. doi:10.1002/hyp.7903

CrossRef Full Text | Google Scholar

Zou, D., Zhou, Y., Ling, H. I., Kong, X., and Xu, B. (2012). Dislocation of face-slabs of Zipingpu concrete face rockfill dam during Wenchuan earthquake. J. Earthq. Tsunami 6 (2), 1250007. doi:10.1142/s1793431112500078

CrossRef Full Text | Google Scholar