Dynamics Assessment of the Landslide–Debris Flow Hazard Chain Based on Post-Disaster Geomorphological and Depositional Evidence: A Case Study from Xujiahe, Sichuan, China

Abstract

Compound geological disaster chains pose major challenges for disaster prevention in mountainous regions due to their complex mechanisms and cascading impacts. This study investigates a landslide–debris flow–flash flood hazard chain that occurred on 21 July 2024 in the Xujia River catchment, Mianning County, Sichuan Province, China. This event is used as a representative case to improve the understanding of the formation and amplification mechanisms of breach-type debris flows through dynamic inversion constrained by sedimentary records. The objective is to reconstruct the evolution of the event and assess its downstream hazard extent. Post-disaster sedimentary and geomorphological records, including deposit distribution, channel aggradation, and flow traces, were systematically analyzed based on remote sensing interpretation, unmanned aerial vehicle surveys, and detailed field investigations. These sedimentary data were used as key constraints to estimate debris flow magnitude and mobility under different rainfall scenarios. A rainfall flood scenario-based estimation method was applied to quantify debris flow magnitude, and numerical simulations were conducted using the Rapid Mass Movement Simulation model to reproduce debris flow propagation and deposition processes. The results indicate that prolonged antecedent rainfall triggered slope failure in a tributary, leading to the accumulation of landslide-derived material and the formation of a temporary channel blockage. The subsequent breach of this blockage significantly amplified debris flow discharge, velocity, and sediment outflow, resulting in downstream hazard expansion. Simulation results constrained by sedimentary evidence show that peak discharge and solid material output under breach conditions were approximately three times higher than those of rainfall-driven scenarios under comparable rainfall frequencies. These findings demonstrate that sedimentary records provide critical constraints for the inversion of landslide debris flow disaster chain dynamics and highlight the effectiveness of post-disaster evidence based numerical assessment for hazard analysis and risk mitigation in debris flow-prone mountainous catchments.

1. Introduction

The focus of geological disaster prevention and mitigation has gradually shifted from single-type disasters to compound disaster chains as research has progressed. Compound geological disaster chains consist of two or more basic disaster types and may trigger secondary phenomena such as river damming, surge waves, and outburst flooding [1,2,3]. These disasters are characterized by coupled failure mechanisms, energy transformation, and long-distance cascading effects. They have become an important concern in disaster-prone mountainous regions, particularly in western China. Representative events include the landslide–debris flow disaster in Sanxi Village, Dujiangyan, Sichuan Province [4]; the Zhaojiagou landslide–debris flow disaster in Yunnan Province [5]; and the Xinmo rock avalanche–debris flow disaster in Sichuan Province [6].

Among various compound geological disaster chains, the landslide–debris flow type is one of the most prominent. When a single landslide transforms into a debris flow, the affected area, disaster scale, and damage intensity can increase sharply. As a result, such disaster chains pose significant risks to transportation networks, public infrastructure, and the safety of lives and property in mountainous regions [7,8]. From the perspective of motion mechanics, landslides are characterized by the overall sliding of rock–soil masses caused by the development of localized shear zones [9,10,11]. Their failure process typically involves a distinct evolutionary stage of the sliding surface. In contrast, debris flows are solid–liquid mixtures that move continuously dominated by shear flow. Their motion is commonly accompanied by intense turbulence, particle mixing, and phase transformation [12,13,14]. Because of these fundamental differences in movement mechanisms, landslides and debris flows have been extensively investigated using different approaches. Landslides are often studied with an emphasis on easily sliding strata and structural controls [15,16,17,18,19,20,21]. Debris flows, by contrast, are mostly examined using various rheological models to describe flow behavior [22,23,24,25,26,27,28,29,30,31]. Owing to their distinct mechanical characteristics, landslides and debris flows are generally classified as two different types of geological disasters. However, under certain conditions, a landslide can transform into a debris flow and continue to move downslope. This indicates the existence of a transitional stage between the two processes and increases the complexity of disaster chain evolution.

Recent studies have highlighted the importance of sedimentary archives and geomorphological signatures in reconstructing past mass-movement cascades and debris flow processes, particularly in complex chained events where multiple process domains interact. Sedimentary records not only preserve information on flow magnitude and mobility but also provide key constraints for identifying process transitions such as landslide–debris flow transformation and temporary dam formation and failure. For example, investigations of debris flow chains and multi-hazard process linkages have emphasized the role of post-event depositional patterns in understanding cascading behavior, while sedimentological studies have demonstrated how stratigraphic and facies characteristics can be used to reconstruct landslide and debris flow histories. These perspectives form an important scientific basis for the evidence-constrained dynamic inversion adopted in this study [32,33].

In post-disaster mountainous catchments, sedimentary and geomorphological records such as deposit extent, channel aggradation, transport distance of coarse clasts, and channel modification constitute the most direct physical evidence of past flow dynamics. Unlike predictive hazard modeling that relies on assumed initial conditions, event-based inversion of debris flow dynamics must be constrained by what the event actually left behind. In this context, sedimentary and geomorphological records play a critical role in bounding flow magnitude, mobility, and runout. However, many previous studies either emphasize detailed sedimentological analyses without explicitly linking them to dynamic inversion, or rely primarily on forward numerical simulations with limited integration of post event sedimentary evidence. As a result, the potential of sedimentary records to constrain the dynamic inversion of landslide debris flow disaster chains has not yet been fully exploited, particularly for quantifying amplification processes associated with temporary channel blockage and breach.

This study investigates the landslide debris flow disaster chain that occurred on 21 July 2024 in the Xujia River catchment, Sichuan Province, China, with the aim of elucidating the dynamic evolution and amplification mechanisms of breach type processes within a compound hazard chain. By integrating post-disaster field evidence with numerical modeling, this study reconstructs the event evolution and constrains debris flow dynamics under different triggering scenarios. Unlike previous studies that primarily focus on forward prediction or single hazard processes, this study adopts an evidence constrained inversion perspective to explicitly compare rainfall-driven and breach-induced debris flow scenarios under comparable hydrometeorological conditions. Through this approach, the relative amplification effects of temporary channel blockage and failure on downstream hazard extent are quantified, providing new insights into the role of breach processes in controlling debris flow mobility and cascading impacts in mountainous catchments.

2. Study Area and Event Description

2.1. Topography and Geomorphology

The Xujia River is a tributary of the Nanhe River and located in Gaoyang Subdistrict, Mianning County. The basin exhibits a typical erosion–accumulation geomorphic setting and covers a total drainage area of 27.83 km². As shown in Figure 1, the main gully extends approximately 11.89 km from the mountainous source area toward the downstream alluvial plain, with an average longitudinal gradient of 134.7‰.

The topography of the basin is characterized by pronounced relief, with elevation decreasing progressively from the northwestern mountainous region toward the southeastern outlet. Elevations range from approximately 3345 m along the northwestern ridge to about 1730 m at the basin outlet, resulting in a relative relief of nearly 1600 m. The slope map derived from the DEM (Figure 2a) further indicates that steep slopes are widely distributed in the upstream and midstream sections, whereas gentler terrain is mainly developed in the downstream area. This topographic configuration provides favorable conditions for rapid runoff concentration, gravitational instability on steep hillslopes, and sediment accumulation in lower-gradient downstream reaches.

Tributaries within the basin display a dendritic drainage pattern, indicating limited structural control on small-scale channel development. Field photographs taken along representative reaches of the Xujia River document the typical geomorphic and engineering impacts associated with the landslide–debris flow disaster chain in the study area. Specifically, Figure 1a shows thick mud and debris deposits covering the road surface, indicating direct burial of transportation infrastructure by debris flow materials. Figure 1b illustrates the accumulation and temporary stagnation of debris flow deposits within the channel, reflecting sediment storage and partial blockage after the event. Figure 1c presents a narrowed river cross-section caused by debris accumulation, which compressed the original flow path and reduced channel conveyance capacity. Figure 1d demonstrates the obstruction and constriction of a bridge opening by debris flow deposits, highlighting the interaction between debris flow and hydraulic structures. Figure 1e shows evident erosion and damage to the road embankment, indicating intensified lateral erosion and undercutting by post-event channel flow. Overall, these field observations reveal the combined effects of material deposition, channel constriction, and infrastructure damage along the main gully. Such features are characteristic of the evolutionary process of landslide-derived debris flows and provide direct evidence of sediment transport, temporary blockage, and subsequent channel adjustment during the disaster chain.

2.2. Stratigraphy Lithology and Structure

Mianning County is situated on the eastern margin of the Tibetan Plateau, within the northeastern segment of the Hengduan Mountains. As shown in Figure 2b, the geological framework of the study area consists primarily of Sinian and Jurassic bedrock units, which form the fundamental structural skeleton of the basin. The Sinian system is mainly exposed along the northwestern and central mountainous ridges, whereas Jurassic strata are more widely distributed in the downstream valley sectors.

Overlying the bedrock, extensive Quaternary surficial deposits of predominantly Holocene age are developed along valley bottoms, gentle slopes, and parts of the midstream reach. These deposits are composed mainly of colluvial, residual, and alluvial materials, and are generally loose and poorly consolidated. Although they do not represent the regional bedrock stratigraphy, these Holocene surficial materials provide the principal source of mobilizable sediment during intense rainfall events and play a key role in slope instability and debris flow initiation.

Structurally, the area is influenced by the Anning River deep fault zone and the Jinhe-Qinghe fault system, as indicated. These major tectonic structures control the regional geomorphic framework, including the alignment of mountain ridges and drainage systems. Long-term tectonic activity along these faults has also contributed to rock mass fracturing and slope weakening, thereby increasing the susceptibility of the basin to mass-movement hazards. According to the national seismic zonation, the study area corresponds to a seismic fortification intensity of Grade VIII, with a basic design seismic acceleration of 0.3 g and a characteristic response spectrum period of 0.45 s.

2.3. Geotechnical Characteristics of Landslide Material

Field investigations indicate that the landslide source material is mainly composed of highly weathered sandstone and mudstone fragments mixed with a silty clay matrix. The landslide mass exhibits a loose and heterogeneous structure, with poor sorting and a wide grain size distribution ranging from fine soil particles to coarse rock blocks. In the source area and along the transport channel, the material is generally matrix-supported, indicating a high proportion of fine content that promotes fluidization under intense rainfall conditions.

The spatial evolution of these materials from slope failure to final deposition is illustrated in Figure 3. In the landslide formation area, exposed slopes consist of strongly weathered bedrock and colluvial cover, where rock fragments are embedded within fine-grained soil. During downstream transport, intense channel erosion, shear disturbance, and water entrainment lead to further mixing of coarse and fine particles and an increase in moisture content. These processes reduce interparticle friction and enhance flow mobility.

In the downstream depositional zone, the resulting debris flow deposits form channel fills and an alluvial fan characterized by matrix-supported structures, poor sorting, and the coexistence of large boulders with muddy matrix. Such sedimentary characteristics indicate rapid, high-concentration flow conditions typical of debris flows rather than water-dominated floods. The observed material composition and structural features therefore provide direct field constraints on the rheological behavior and high mobility of the debris flow.

Due to prolonged antecedent rainfall prior to the event, the landslide materials were in a near-saturated state at the time of failure. Field evidence, including softened soil texture, surface seepage, and remolded shear zones, suggests a significant reduction in shear strength and an increase in pore-water pressure during initiation. These hydro-mechanical conditions are favorable for the rapid transition from slope failure to debris flow motion. Although detailed laboratory testing of geotechnical parameters was not available for this event, the observed material characteristics, sedimentary structures, and moisture conditions provide important qualitative constraints on debris flow mobility and transformation behavior. These field-based observations were therefore used to guide the selection of rheological parameters and the interpretation of numerical simulation results.

2.4. Meteorological and Hydrological Characteristics

Mianning County experiences abundant precipitation under the influence of the regional monsoon climate. Statistical records from the past eleven years indicate an average annual rainfall of 1083.3 mm, with annual totals ranging from 661.2 mm to 1411.0 mm.

Rainfall monitoring data from the Pingba Station in Hui’an Township (102.11° E, 28.37° N) further illustrate the temporal distribution of precipitation in the study area. The mean monthly precipitation is 106.0 mm, and the mean annual precipitation over the past five years reaches 1271.5 mm. As shown in Figure 4, rainfall exhibits a pronounced seasonal pattern characterized by a strong summer concentration. Monthly precipitation increases rapidly from May, reaches peak values in June and July, and remains at a high level through August and September before declining sharply in October. In particular, July records the highest mean monthly rainfall, exceeding 260 mm, indicating a pronounced rainfall peak during the mid-summer period. Such concentrated and high-magnitude precipitation provides favorable hydrometeorological conditions for rapid runoff generation and sediment mobilization within the basin.

During extreme events, short-duration rainfall intensity can be particularly high. The maximum recorded rainfall amounts are 147.2 mm within 24 h, 53.2 mm within 6 h, 41.0 mm within 1 h, and 16.7 mm within 10 min. Such rainfall characteristics indicate a high potential for rapid runoff generation and sharp increases in channel discharge within the basin.

3. Data

3.1. Rainfall Data Prior to the Event

To characterize the antecedent rainfall conditions prior to the landslide–debris flow event, daily precipitation data covering approximately 6–7 months before the disaster were collected and analyzed. Meteorological data were obtained from the ERA5-Land reanalysis dataset provided by the Copernicus Climate Data Store, which offers spatially continuous and temporally consistent precipitation records and has been widely applied in hydrometeorological studies.

Based on the geographic location of the study area, daily rainfall data from June to July 2024 were extracted and statistically analyzed. Figure 4 presents the temporal variation in daily rainfall and cumulative rainfall during this period. The results indicate that rainfall occurred frequently and persistently, with a total of 49 rainy days recorded between June and July, accounting for approximately 80.3% of the entire period. From July 1 to July 20 alone, rainfall was observed on 17 days, representing about 85% of the time span.

Several pronounced rainfall peaks can be identified in Figure 5. Notably, daily precipitation reached 72 mm on June 28 and 73.5 mm on July 20. Correspondingly, the cumulative rainfall increased rapidly during these periods, reaching 496.5 mm by July 21. The continuous accumulation of rainfall during a relatively short time window suggests persistently high antecedent moisture conditions prior to the disaster.

To evaluate the reliability of the ERA5-Land reanalysis data, precipitation characteristics were further compared with records from multiple local rain gauge stations within Mianning County. Observations from these stations show annual precipitation ranging approximately from 700 mm to over 1500 mm, with July consistently representing the month of maximum rainfall in

Table 1. Summary of precipitation records from representative rain gauge stations in Mianning County from 2010 to 2020.

Station	Mean Annual Precipitation (mm)	Maximum Monthly Precipitation in July (mm)
Mianning County Town	1083.3	514.6
Fuxing Town	1003.5	468.1
Hebian Town	1108.3	473.8
Hongmo Town	713.9	332.6
Hui’an Town	1553.9	647.2
Lingshan Temple	1379.1	491.5
Ruoshui Town	863.1	399.6

. The maximum monthly rainfall recorded at several stations exceeds 470–650 mm, which is comparable to the cumulative rainfall magnitude derived from ERA5-Land during the pre-event period. Moreover, the seasonal rainfall pattern and the occurrence of heavy rainfall peaks are consistent between station observations and reanalysis data. This agreement indicates that ERA5-Land data reasonably capture both the magnitude and temporal evolution of precipitation in the study region and are suitable for representing antecedent rainfall conditions prior to the event.

3.2. Field Investigation Data

A detailed field investigation was conducted in the Xujia River catchment following the debris flow-flash flood event of 21 July 2024. The investigation focused on identifying the source area of the debris flow, characterizing slope and gully geomorphology, and documenting sediment deposition and flood impact in downstream areas. Field surveys were carried out using a combination of visual inspection, tape measurements, handheld GPS positioning, and post-disaster unmanned aerial vehicle (UAV) imagery.

3.2.1. Characteristics of the Debris Flow Source Area

Based on detailed field investigations and interpretation of UAV imagery, the debris flow source area in Tributary No.1 shown in Figure 1 can be subdivided into several geomorphological units with distinct material characteristics and roles in the debris flow initiation process.

As shown in Figure 6, the upper part of the catchment is characterized by a clear water collection area, where surface runoff is concentrated but sediment supply is limited. Downslope, the debris source area is composed of multiple unstable slope units (Zones I₁ and I₂), which are mainly distributed on both sides of the gully. Zone I₁ covers an area of approximately 49,038 m² with an average thickness of about 1.5 m, while Zone I₂ is separated from Zone I₁ by a small upslope gully. Field observations indicate that these areas are mainly composed of colluvial and residual deposits, with the main components being a mixture of blocky rocks and clay. The structure is loose and poorly consolidated.

Zone II, located upslope of the main gully and bounded by ridge lines, represents the principal material source area of the debris flow. This zone covers an area of approximately 56,260 m², with an average thickness of 1.5 m and an estimated material volume of about 84,390 m³. The slope gradient in this zone generally exceeds 40°, providing favorable topographic conditions for gravitational failure once material strength is reduced.

Downstream of the source area, the gully enters the debris flow formation and transport zone. As shown in Figure 6, failed materials from Zones I and II entered the gully channel, resulting in significant channel widening and incision. Field measurements show that the channel width locally reaches approximately 20.0 m, with a maximum erosion depth of up to 5.5 m. The accumulation of failed materials in the lower part of the gully formed a temporary blockage (Zone III₂), which subsequently experienced breaching under concentrated runoff conditions.

The spatial correspondence between the source areas, transport channel, and breach location shown in Figure 6 and Figure 7 provides direct field evidence for the landslide–debris flow transformation process in Tributary No. 1 and serves as an important basis for subsequent numerical modeling.

3.2.2. Channel Cross-Sectional Characteristics and Geomorphic Response

To quantify channel deformation and geomorphic response induced by the landslide–debris flow event, representative cross-sections were extracted along Tributary No. 1 and its downstream reach, as shown in Figure 3 and Figure 7. The selected cross-sections (2-2′ to 4-4′) correspond to the upstream formation area, midstream transport section, and downstream accumulation zone.

Cross-section 1-1′, located within the landslide–debris flow formation area, exhibits limited channel incision, indicating that geomorphic modification in this section was dominated by slope material supply rather than channel erosion. In contrast, cross-sections 2-2′ and 3-3′, situated within the circulation (transport) area, show pronounced channel widening and deep incision. Field measurements indicate that the maximum incision depth in this section reaches approximately 5.5 m, accompanied by significant lateral erosion of the channel banks. At cross-section 4-4′, located in the downstream accumulation area, the channel geometry is characterized by substantial infilling due to the deposition of debris flow materials. The comparison between the original channel surface and the post-event affected region reveals those large volumes of solid material accumulated in this section, forming a temporary blockage that altered the local channel profile.

Overall, the spatial variation in cross-sectional morphology along Tributary No. 1 reflects a clear transition from material supply and initiation in the upstream area, to intensive erosion and transport in the midstream section, and finally to deposition and blockage formation in the downstream area. These cross-sectional characteristics provide essential geometric constraints for subsequent hydrological assessment and numerical modeling of debris flow and dam breach processes.

3.3. Event Scale Physical Parameter Constraints

To provide independent physical constraints on the magnitude of debris flow mobilization, bulk density and discharge-related parameters were estimated using field-based reconstruction and hydrodynamic formulations.

Field slurry reconstruction tests conducted in the deposition zone yielded a debris flow bulk density (γ_c) between 1.60 and 1.68 t/m³, with an average value of 1.65 t/m³. This density is consistent with a matrix-supported debris flow, as indicated by field evidence of high fine particle content, weak sorting, and near-saturated material conditions.

Peak debris flow discharge was linked to rainfall-induced flood peaks through a bulk amplification relationship:

$$Q_c=(1+\varphi )Q_p{D}_c$$

(1)

where Q_p is the clear-water flood peak discharge, 1 + ϕ is the sediment concentration correction factor related to bulk density, and D_c is a blockage/amplification coefficient reflecting channel obstruction effects observed in the field. These parameters are physically constrained by field observations of channel narrowing, deposit texture, and the independently reconstructed bulk density.

The total event scale sediment outflow (Q_s) was then approximated through a mass-volume relationship:

$$Q_s=Q_{c}T{\displaystyle \frac{{\gamma }_c-{\gamma }_{\omega }}{{\gamma }_s-{\gamma }_{\omega }}}$$

(2)

where T is the effective flow duration, γ_ω the density of water, and γ_s the density of solid particles. The resulting sediment mobilization is in the order of 10⁴ m³, which agrees with post-event geomorphic evidence, including channel infilling thickness, outlet deposition scale, and downstream sediment accumulation.

These calculations are not intended as precise engineering design values but serve as order-of-magnitude physical constraints linking material properties, flow hydraulics, and geomorphic observations. The density–discharge–volume consistency indicates that the sediment magnitude adopted in the simulations lies within a physically reasonable range.

3.4. Estimation of Debris Flow Deposit Volume Based on Geomorphic Evidence

To quantify the volume of solid material mobilized during the event, the deposit volume was estimated based on post-event geomorphic mapping rather than solely relying on hydrodynamic back-calculation. The deposition area was delineated through interpretation of UAV imagery and field investigation, where the spatial extent of fresh debris flow deposits was clearly identifiable from surface texture, color contrast, and vegetation burial features. Representative deposit thicknesses were measured in the field using tape measurements at exposed sections, channel margins, and depositional lobes. Because thickness varies spatially, the deposition zone was subdivided into several geomorphologically homogeneous subareas according to topographic position (channel bed, overbank area, lobe front, and temporary blockage zone). An average deposit thickness was assigned to each subarea based on multiple field observations.

The total deposit volume was then calculated using an area–thickness integration approach:

$$V_{geom}={\displaystyle \sum _{i=1}^n{A}_i{h}_i}$$

(3)

where A_i is the area of subregion i derived from UAV mapping, and h_i is the representative average deposit thickness obtained from field measurements.

This approach ensures that the sediment volume estimation is directly constrained by observable geomorphic evidence, independent of the hydrodynamic model.

4. Methods

4.1. Rain–Flood Method for Debris Flow Discharge Estimation and Volume Consistency Check

To estimate debris flow characteristic parameters in Tributary No.1 in different rainfall scenarios, a rainfall–flood scenario-based engineering method was adopted. The method establishes a linkage between extreme rainfall conditions and debris flow magnitude and is commonly applied in mountainous catchments [34].

Extreme rainfall parameters corresponding to different return periods were derived from long-term regional rainfall records and transformed into clear-water flood peak discharge (Q_p) using hydrological relationships. These flood peaks serve as the hydraulic baseline for debris flow estimation. Debris flow peak discharge (Q_c) was then calculated using the bulk amplification relationship described in Section 3.3:

$$Q_c=\lambda Q_p$$

(4)

where λ represents the combined effect of sediment concentration and channel blockage. The range of λ values was not arbitrarily assigned but selected within the physically constrained bounds established from field evidence and density reconstruction (Section 3.3). Larger λ values were adopted for dam breach scenarios to represent the sudden release of temporarily stored sediment.

Based on the estimated Q_c, the total mixture volume transported during a debris flow surge was estimated as

$$V_{dyn}=Q_{c}T$$

(5)

where T is the effective duration of the debris flow peak stage. The corresponding solid material volume was then obtained by converting the mixture volume using the reconstructed debris flow bulk density. These calculations provide an independent hydrodynamic estimate of the order of magnitude of sediment transport during the event.

It should be emphasized that this rainfall–flood method is not used as the primary source of deposit volume determination. Instead, the deposit volume is principally constrained by geomorphic mapping and field-based thickness measurements (Section 3.4). The hydrodynamic calculations presented here serve as a physical consistency check: the dynamically estimated sediment volume is of the same order of magnitude as the mapped deposit volume, indicating that the inferred event scale is hydraulically feasible under the observed rainfall conditions.

Therefore, this scenario-based framework is employed to analyze the relative influence of rainfall intensity and temporary blockage/breach processes on debris flow magnitude, while simultaneously providing a cross-validation between hydrodynamic capacity and geomorphically observed sediment volume.

4.2. Rapid Mass Movement Simulation (RMMS) Model for Debris Flow Numerical Simulation

The numerical simulation of debris flow motion was conducted using the debris flow module of RAMMS. RAMMS is a physically based numerical modeling framework developed jointly by the Swiss Federal Institute for Forest, Snow and Landscape Research and the Swiss Institute for Snow and Avalanche Research. The model is widely applied to simulate the propagation, deposition, and impact of debris flows in complex mountainous terrain [35]. In particular, the Voellmy-type rheology implemented in RAMMS is suitable for reproducing high-concentration, surge-type debris flows triggered by sudden releases, such as dam breach events, which are a key process in the present study.

In RAMMS, debris flow is treated as a depth-averaged continuum and described as a non-stationary and non-homogeneous flow. The governing equations are derived from the principles of mass and momentum conservation. The state of the debris flow is characterized by the flow depth H(x, y, t) and the depth-averaged velocity components U_x(x, y, t) and U_y(x, y, t). The mass conservation equation is expressed as

$${\displaystyle \frac{\partial H}{\partial t}}+{\displaystyle \frac{\partial (H{U}_x)}{\partial x}}+{\displaystyle \frac{\partial (H{U}_y)}{\partial y}}=Q(x,y,t)$$

(6)

where H denotes the flow depth, U_x and U_y are the velocity components in the horizontal directions, and Q(x, y, t) represents the mass source term.

The momentum conservation equations in the x and y directions are given by

$${\displaystyle \frac{\partial (H{U}_x)}{\partial t}}+{\displaystyle \frac{\partial }{\partial x}}\left(C_{x}H{U}_x^2+{\displaystyle \frac{1}{2}}{g}_z{k}_{a/p}{H}^2\right)+{\displaystyle \frac{\partial (H{U}_x{U}_y)}{\partial y}}=S_{gx}-S_{fx}$$

(7)

$${\displaystyle \frac{\partial (H{U}_y)}{\partial t}}+{\displaystyle \frac{\partial }{\partial y}}\left(C_{y}H{U}_y^2+{\displaystyle \frac{1}{2}}{g}_z{k}_{a/p}{H}^2\right)+{\displaystyle \frac{\partial (H{U}_x{U}_y)}{\partial x}}=S_{gy}-S_{fy}$$

(8)

where C_x and C_y are the shape coefficients, g_z is the gravitational acceleration in the vertical direction, and k_a/p is the earth pressure coefficient. The terms S_gx and S_gy represent the gravitational driving forces, while S_fx and S_fy denote the basal resistance forces.

Basal resistance is described using the Voellmy-type rheological formulation, which combines Coulomb friction and velocity-dependent turbulent resistance. The basal shear resistance S_f is expressed as

$$S_f=\mu N+(1-\mu )N_0-(1-\mu )N_0{e}^{-{\displaystyle \frac{N}{N_0}}}+{\displaystyle \frac{\rho g{U}^2}{\xi }}$$

(9)

where μ is the friction coefficient, N is the normal stress acting on the basal surface, N₀ is the yield stress of the debris material, ρ is the bulk density of the debris flow, g is gravitational acceleration, U is the depth-averaged velocity magnitude, and ξ is the turbulence coefficient.

4.3. Parameterization and Scenario Design

4.3.1. Physically Constrained Calibration of Rheological Parameters

The predictive performance of the RAMMS debris flow module is highly sensitive to the rheological parameters [8], particularly the friction coefficient (μ) and the turbulence coefficient (ξ). In this study, parameter determination was conducted using a geomorphology-constrained inverse approach, rather than subjective visual fitting.

Post-event field investigations revealed several key dynamic characteristics of the debris flow event in Tributary No. 1: (i) a runout distance that is long relative to the catchment length, (ii) matrix-supported muddy deposits with weak sorting, (iii) long-distance transport of meter-scale boulders, (iv) pronounced channel scour and lateral erosion, and (v) surge-type behavior associated with temporary blockage and breach processes. These features collectively indicate a high-mobility, surge-type debris flow as opposed to a friction-dominated granular slide.

Based on debris flow regime classification and previously reported parameter ranges for similar flow types, the admissible parameter space was first bounded. Friction-dominated granular flows typically require μ > 0.25 and relatively small ξ values (<150 m s⁻²), whereas high-mobility surge-type debris flows are characterized by lower effective friction (μ ≈ 0.10–0.18) and higher turbulence coefficients (ξ ≈ 200–400 m s⁻²). The observed geomorphic and sedimentological evidence clearly places the present event within the latter category, and parameter testing was therefore restricted to this physically plausible domain.

Within this bounded parameter space, simulations were evaluated against multiple independent geomorphic constraints rather than a single visual criterion. Three key observational controls were applied: (1) the mapped runout limit of the event, which constrains overall flow mobility; (2) the spatial position of the main deposition and temporary blockage zones identified from UAV mapping and field surveys; and (3) the order of magnitude of flow depth, which must be compatible with observed channel incision (up to ~5.5 m) and deposit thicknesses. Parameter combinations that reproduced the runout distance but generated unrealistic over-spreading, or that matched deposition extent only under implausibly low velocities or flow depths, were systematically excluded.

The parameter pair μ = 0.15 and ξ = 250 m s⁻² was identified as the only combination that simultaneously satisfies all three geomorphic constraints while remaining within the physically admissible range for high-concentration, surge-type debris flows. These values are also consistent with published ranges for debris flows in steep mountainous gullies.

Accordingly, the rheological parameters adopted in this study should be interpreted as physically constrained inverse estimates derived from geomorphic evidence, rather than empirical tuning constants.

4.3.2. Topographic Data and Model Termination Criteria

High-resolution topographic data are essential for accurately representing debris flow propagation paths and deposition patterns. In this study, a digital elevation model derived from UAV-based orthophotos was used as the topographic input for all simulations. The DEM captures detailed terrain features of the gully and downstream channel and provides a reliable basis for numerical modeling.

The simulations were automatically terminated when the kinetic energy of the debris flow decreased to less than 5% of its maximum value. This termination criterion ensures that the final deposition state is reached while avoiding unnecessary computational cost.

4.3.3. Scenario Design

To investigate the effects of rainfall intensity and dam breach conditions on debris flow behavior, a set of representative simulation scenarios was designed. Two flow conditions were considered for Tributary No.1, including normal conditions and dam breach conditions. For each flow condition, simulations were performed under rainfall frequencies of 5% and 2%, which correspond to medium and extreme rainfall scenarios in the study area.

The initial debris flow discharge and solid material volume for each scenario were derived from the rainfall–flood scenario-based estimation described in Section 4.1. These parameters were used as input conditions for the numerical simulations. In addition, to assess the downstream hazard extent of the flash flood–debris flow disaster chain in the Xujia River, further simulations were conducted under rainfall frequencies of 10%, 5%, 2%, and 1%. This scenario design enables a systematic comparison of debris flow dynamics and downstream impacts under different rainfall intensities and breach conditions while maintaining consistency in model parameters and numerical settings.

Detailed grain size statistics along and across the flow path were not available for this post-event investigation. As a result, sediment segregation and effective friction could not be explicitly constrained based on grain size distributions. Instead, qualitative sedimentary indicators, including long-distance transport of large boulders, matrix-supported muddy deposits, and extensive channel scour, were used as proxies to infer a highly mobile debris flow regime and to guide the selection of rheological parameters in the numerical simulations.

5. Results

5.1. Estimated Debris Flow Magnitude Under Rainfall–Flood Scenarios

Table 2. Estimated debris flow characteristic parameters of Tributary No.1 under different rainfall and breach scenarios.

Different Scenarios	Peak Flood Discharge, Qp (m3/s)	Gross Bulking Factor, λ	Debris Flow Peak Discharge, Qc (m3/s)	Total Sediment Volume per Event (104 m3)	Solid Material Outflow Volume (104 m3)
Normal condition at 20%	7.14	1.5	14.14	0.112	0.044
Normal condition at 10%	9.57	1.5	20.59	0.326	0.134
Normal condition at 5%	11.97	2.0	37.63	0.596	0.224
Normal condition at 2%	15.21	2.0	48.74	0.893	0.290
Dam-breach condition at 5%	11.97	4.0	86.84	3.439	1.542
Dam-breach condition at 2%	15.21	4.5	144.80	5.734	3.023

summarizes the estimated debris flow characteristic parameters of Tributary No.1 under different rainfall frequencies and flow conditions. Under normal conditions, debris flow peak discharge increased progressively with decreasing rainfall frequency, from 14.14 m³/s at the 20% rainfall frequency to 48.74 m³/s at the 2% rainfall frequency. Correspondingly, the estimated total sediment volume and solid material outflow exhibited a similar increasing trend.

Under dam breach conditions, debris flow magnitude increased markedly compared with normal conditions at the same rainfall frequency. At the 5% rainfall frequency, the debris flow peak discharge increased from 37.63 m³/s under normal conditions to 86.84 m³/s under dam breach conditions. At the 2% rainfall frequency, the peak discharge further increased to 144.80 m³/s. The estimated solid material outflow under dam breach conditions was approximately an order of magnitude higher than that under normal conditions. These estimated parameters were used as input conditions for subsequent numerical simulations.

5.2. Simulation Results of Breach-Type Debris Flow in Tributary No.1

Given the absence of a deposit thickness map, the simulated runout distances are interpreted as falling within the observed inundation and deposition extent, rather than as an exact reconstruction of the delivered sediment volume. Numerical simulations were conducted using the debris flow module of RAMMS to obtain the spatial distributions of maximum flow depth and maximum flow velocity under different rainfall frequencies and flow conditions. The simulation results are summarized in

Table 3. Simulation results of debris flow caused by No.1 breach gully.

Different Scenarios	Runout Distance (m)	Deposition Area (km2)	Flow Depth (m)	Flow Velocity (m/s)
Normal condition at 5%	187	0.017	1.31	4.93
Dam breach condition at 5%	698	0.095	2.48	5.88
Normal condition at 2%	278	0.028	1.84	4.90
Dam breach condition at 2%	708	0.12	3.15	6.05

and illustrated in Figure 8 and Figure 9.

Under normal conditions, the simulated debris flows exhibited relatively limited runout distances and deposition areas. For rainfall frequencies of 5% and 2%, the maximum flow depth ranged from 1.31 m to 1.84 m, and the maximum flow velocity remained below 5.0 m/s. Differences in runout distance and deposition area between the two rainfall scenarios were minor.

In contrast, dam breach conditions resulted in a substantial increase in debris flow mobility. For the 5% rainfall frequency, the runout distance increased from 187 m to 698 m, and the deposition area expanded from 0.017 km² to 0.095 km². Similar trends were observed under the 2% rainfall frequency, with the maximum flow depth reaching 3.15 m and the maximum velocity exceeding 6.0 m/s. Overall, the simulations show that dam breach conditions significantly enhance debris flow velocity, flow depth, runout distance, and deposition extent.

5.3. Predicted Hazard Extent of the Flash Flood–Debris Flow Disaster Chain

Based on the estimated debris flow inputs and numerical simulations, the hazard extent of the flash flood–debris flow disaster chain in the Xujia River was predicted under different rainfall frequencies. The calculated flash flood and debris flow parameters are summarized in

Table 4. Calculated results of Xujia River flash flood and debris flow under different rainfall frequencies.

Different Rainfall Frequencies	Flood Flow QP (m3/s)	Debris Peak Flow Qc (m3/s)	Total Siltation Amount of One-Time Debris Flow Q (104 m3)	Amount of Solid Material Outflow QH (104 m3)
10%	92.61	92.61	14.934	4.98
5%	116.11	212.86	12.86	4.29
2%	147.63	270.65	10.12	3.37
1%	171.42	314.27	8.07	2.44

, and the spatial hazard ranges are shown in Figure 10.

In a 10% rainfall frequency scenario, the affected length of the flash flood–debris flow extended approximately 2.16 km downstream from the confluence of Tributary No.1 and the Xujia River, with an influenced area of about 0.30 km². As rainfall frequency decreased, both the affected length and area increased. In the 1% rainfall frequency scenario, the affected length reached approximately 2.37 km, and the affected area expanded to about 0.52 km². The simulated hazard extent shows a clear asymmetry across the river channel, with more severe impacts on the left bank due to local topographic constraints.

6. Discussion

6.1. Reconstruction of the Breach-Type Debris Flow Based on Field Evidence

Field investigations provide essential constraints for reconstructing the debris flow event in Tributary No.1 and for interpreting the numerical simulation results. The downstream transport section of the tributary is relatively straight and morphologically stable, which allows a more reliable estimation of debris flow discharge based on depositional features and inundation marks. The estimated peak debris flow discharge of approximately 136.2 m³/s and the solid material deposition at the gully outlet of about 3000 m³ indicate a high-magnitude event.

Meteorological records show that the 24 h rainfall on 20 July 2024 reached only 73.5 mm, which is significantly lower than the 20% rainfall frequency threshold. However, the observed debris flow discharge and solid material outflow were comparable to those estimated under the 2% rainfall frequency scenario. This apparent inconsistency suggests that rainfall intensity alone cannot explain the magnitude of the event and implies the involvement of additional amplification mechanisms.

Field evidence indicates that a temporary blockage formed in the upstream channel, leading to the accumulation of solid materials. The subsequent failure of this blockage resulted in a sudden release of water–sediment mixtures, significantly increasing debris flow discharge and solid material transport capacity. Such breach-type processes have been widely reported as a critical factor in enhancing debris flow magnitude under moderate rainfall conditions. The present case provides a clear example of this mechanism in a small mountainous catchment.

The lack of a spatially continuous deposit thickness map inevitably introduces uncertainty in constraining the absolute delivered sediment volume and runout. However, the objective of this study is not to precisely reconstruct the total sediment yield but to assess the relative amplification effect of dam breach processes compared to rainfall-driven debris flow scenarios under bounded sediment volume conditions. Within this framework, the combination of channel incision depth, erosion patterns, deposition extent, and transport distance of coarse materials provides sufficient first-order constraints to evaluate differences in flow mobility and downstream impact between scenarios.

6.2. Consistency Between Estimated Results and Numerical Simulations

The numerical simulations conducted using the RAMMS model reproduced the main characteristics of the debris flow process under different rainfall–flood scenarios. The absence of quantitative grain size statistics along and across the flow path limits a direct assessment of sediment segregation and its influence on effective friction and flow mobility. However, field observations such as the sustained transport of large boulders over distances exceeding 3.5 km, the widespread presence of matrix-supported muddy deposits, and pronounced channel incision collectively indicate a highly mobile debris flow regime rather than a hyperconcentrated or water-dominated flow. Within this context, the rheological parameters adopted in the numerical simulations should be interpreted as effective bulk representations of flow behavior. They do not represent grain size-specific physical constants but are appropriate for evaluating relative differences between dam breach and rainfall-driven scenarios under comparable uncertainty conditions.

In contrast, simulations assuming rainfall-driven debris flows without blockage failure tend to underestimate both the peak discharge and the extent of downstream impacts. This comparison highlights the importance of explicitly considering breach-type scenarios when modeling debris flow hazards in gullies prone to temporary dam formation. The agreement between simulated results and observed damage patterns supports the rationality of the adopted parameterization and scenario design.

It should be noted that uncertainties remain in the estimation of rheological parameters and entrainment processes. Nevertheless, the consistency between independent field evidence and numerical outcomes suggests that the simulations capture the dominant physical controls governing debris flow propagation in this event.

6.3. Implications for Flash Flood–Debris Flow Disaster Chains

The present event demonstrates the strong coupling between debris flows and downstream flash floods. Part of the solid material discharged from Tributary No.1 was deposited at the gully outlet, while the remaining portion entered the Xujia River and was transported downstream together with floodwaters. This interaction resulted in channel aggradation, reduction of flow capacity, blockage of bridges and culverts, and widespread overbank flooding.

As shown in Figure 11, satellite images before and after the event reveal that the river channel has widened significantly, and the damaged area along the river has expanded. These observations illustrate how localized debris flow events can trigger cascading hazards at the watershed scale. Such disaster chains pose a substantial threat to downstream infrastructure and communities, even when rainfall intensity does not reach extreme levels.

From a hazard assessment perspective, the results emphasize that conventional rainfall thresholds may be insufficient for early warning in catchments susceptible to blockage formation and sudden breach. Incorporating breach-type debris flow scenarios into numerical simulations and risk assessments is therefore essential for improving the reliability of hazard zoning and emergency planning.

6.4. Uncertainty and Parameter Sensitivity

Despite the integration of field evidence, geomorphic constraints, and physically based modeling, several sources of uncertainty remain in the reconstruction and simulation of the debris flow event. These uncertainties mainly relate to rainfall input, sediment volume estimation, and rheological parameter selection, but they do not alter the main physical interpretation of the event [36].

Rainfall data prior to and during the event were derived primarily from the ERA5-Land reanalysis dataset. Although reanalysis products provide spatially continuous precipitation fields, they may smooth localized convective peaks in complex mountainous terrain. Comparison with available rain gauge records within Mianning County shows consistent seasonal patterns and comparable magnitudes of heavy rainfall, suggesting that the reanalysis captures regional rainfall characteristics reasonably well. Nevertheless, localized rainfall intensity near the catchment may deviate from the reanalysis value by approximately 10% to 20%, which propagates into the estimation of flood peak discharge. Since the core finding of this study concerns the discrepancy between rainfall-based expectations and the observed debris flow magnitude rather than the exact rainfall value, the interpretation of a breach-amplified process remains robust under this level of rainfall uncertainty.

Uncertainty also exists in estimating the total sediment volume delivered during the event. The deposit volume was derived from geomorphic mapping and representative thickness measurements instead of a spatially continuous thickness map. Given the spatial heterogeneity of deposition and field measurement variability, the total solid material volume is estimated to have an uncertainty of approximately 30%. This uncertainty affects the absolute magnitude of sediment yield but has limited influence on the comparative analysis between rainfall-driven and breach-type scenarios, because both are evaluated within the same bounded sediment volume framework and are primarily used for order-of-magnitude consistency checks.

The rheological parameters μ and ξ used in the numerical simulations represent effective bulk flow resistance rather than grain-scale physical constants. Although these parameters were determined using a geomorphology-constrained inverse approach, a degree of non-uniqueness remains. Sensitivity analyses indicate that increasing μ mainly reduces runout distance and deposition extent, whereas decreasing ξ primarily lowers peak velocity and dynamic impact with a weaker effect on overall runout. Within the physically admissible range for high-mobility surge-type debris flows, the qualitative contrast between rainfall-driven and dam breach scenarios remains unchanged. Therefore, the central conclusion that temporary blockage and subsequent breach processes strongly amplify debris flow magnitude is insensitive to moderate variations in rheological parameters.

7. Conclusions

This study assessed the dynamic evolution of a landslide–debris flow–flash flood hazard chain that occurred in the Xujia River catchment, Sichuan Province, by integrating post-disaster geomorphological and sedimentary evidence with rainfall–flood scenario analysis and physically based numerical simulations. Field investigations and modeling results jointly demonstrate that breach-type processes can significantly amplify debris flow magnitude and downstream impacts, even under moderate rainfall conditions, producing hazard intensities comparable to those associated with much rarer rainfall events. Numerical simulations using RAMMS reproduced the observed patterns of flow depth, velocity, runout distance, and asymmetric downstream hazard distribution, indicating that post-event evidence provides effective constraints for reconstructing debris flow dynamics in small mountainous catchments. By focusing on post-disaster reconstruction rather than long-term predictive modeling, this work highlights the critical role of temporary blockage formation and failure in compound hazard chains and provides practical insights for hazard assessment and risk mitigation in debris flow-prone regions.