Analysis of debris flow control effect and hazard assessment in Xinqiao Gully, Wenchuan M_s 8.0 earthquake area based on numerical simulation

1.   Introduction

On 12 May 2008, an earthquake measuring M_s 8.0 on the Richter scale occurred in the Wenchuan area of Sichuan Province, China, with a magnitude of 11 and seismic waves that circled the earth six times, affecting Thailand, Vietnam, and Pakistan. The earthquake severely damaged an area of more than 100×10³ km², including 10 counties (cities) in the severely affected area, 41 counties (cities) in the moderately affected area, and 186 counties (cities) in the general disaster area. The earthquake killed 69227 people, injured 374643, and left 17923 missing, making it the most destructive earthquake since the founding of the People’s Republic of China. Since then, landslides and mudslides have occurred frequently in the Wenchuan earthquake zone, continuing to cause loss of life and property in the area.

As a common geological disaster in mountainous regions, debris flows are famous for their sudden occurrence and high impacts (Iverson RM, 1997). Several significant debris flow events have resulted in substantial loss of life and economic damage in the Wenchuan area of Southwest China (Chen R et al., 2013; Zhang J et al., 2013; Ding MT and Huang T, 2019; Yin HQ et al., 2023). The giant debris flow events in Wenjia Gully (Ni HY et al., 2012) and Hongchun Gully (Ouyang CJ et al., 2015) resulted in severe river blockages and numerous casualties. According to statistics, the two events caused a total of 20 deaths and 66 missing persons and affected thousands of people. It is crucial to conduct prevention and disaster assessments of debris flow in mountainous areas.

Debris flow control involves two main aspects: Geotechnical engineering and biological engineering (Zhang WT et al., 2023). Geotechnical engineering includes disciplines such as blocking, protection, drainage, and crossing, which prevent soil erosion, stabilize slopes, and guide the flow (Chen X et al., 2015; Gong XL et al., 2021). Biological engineering is responsible for soil and water conservation, including agricultural and forestry soil and water conservation, as well as stabilizing material sources (Yang ZQ et al., 2013; Wang SY et al., 2017). Different measures are taken depending on the characteristics of the debris flow (Tang HM et al., 2019). A two-factor coupled assessment model, which considers the maximum flow depth and momentum, is usually used for zonal assessment of debris flow hazards (Ouyang CJ et al., 2019; Chang M et al., 2020; Cheng HL et al., 2022). This model can more accurately reflect the damage caused by debris flow impact and burial. Despite controls, failures and damage are common (Xiong MQ et al., 2016). Therefore, it is beneficial to evaluate the effectiveness of debris flow control measures and study the effectiveness of engineering solutions to enhance disaster prevention and control strategies.

Traditional research on debris flow control effects typically relies on qualitative analysis methods (Bai YJ et al., 2022; Tie YB et al., 2022). It involves summarising the characteristics and patterns of disasters through field investigations and experiments, analyzing the impact of prevention and control engineering on debris flows, and providing appropriate measures and recommendations (Haeberli W et al., 2001; Wang W et al., 2001; Imaizumi F et al., 2008; Tang C et al., 2011). However, qualitative analysis is primarily empirical and heavily influenced by human factors. To address this challenge, researchers have introduced mathematical assessment models, such as hierarchical analysis methods (Jun H et al., 2017), fuzzy mathematical methods (Fan JC et al., 2015; Li D et al., 2019), rough set theory (Pai PF et al., 2014; Li M et al., 2018), and neural network assessment models (Chang TC et al., 2010; Liu GX et al., 2013), and advanced research to a semi-quantitative stage. Nevertheless, qualitative and semi-quantitative analysis methods cannot accurately measure the specific impacts of control engineering on debris flow events, such as energy reduction and sediment interception. Consequently, quantitative analysis methods based on numerical simulation emerged.

Researchers have developed various numerical models with different assumptions. For instance, FLO-2D employs a single-block model suitable for simulating hydrodynamic-dominated dynamic processes (Nocentini M et al., 2015), while PFC, utilizing the discrete element method, is well-suited for grain flow simulations (Ray A et al., 2022). RAMMS has a depth-integral continuum model, making it suitable for simulating single-fluid models (Mikos M and Bezak N, 2021). In contrast, Massflow employed a generalized depth-integral model, primarily focusing on erosion dynamic processes (Horton AJ et al., 2019). OpenLISEM incorporates the contributions of viscous stress (Newtonian and non-Newtonian), virtual mass, generalized drag, and buoyancy in the solid-liquid two-phase fluid based on the generalized two-phase flow model, providing a broader range of physical significance compared to other models. OpenLISEM is utilized in regional debris flow hazards studies to offer quantitative information on the spatial and temporal distribution of flow velocity and depth (Pudasaini SP, 2012) and to analyze the dynamic processes. Through the above analyses, the results of the debris flow hazards assessment come out (Bastian VB et al., 2021).

The paper commences with an introduction to the geological conditions and environment of the study area, Xinqiao Gully. Data collection involves using remote sensing and ground investigation techniques to obtain essential information such as sediment grain size characteristics and traces of the bend super-high section. Subsequently, this data determines debris flow density, cross-section flow velocity, and depth. The OpenLISEM model is applied to analyze the Xinqiao Gully debris flow event. In addition, the model is inverted to assess the prevention and control effects. Two quantitative indexes, namely storage capacity fullness and flow intensity reduction rate of the dam, are proposed for the analysis. Finally, a two-factor model combining maximum flow depth and momentum is used to evaluate debris flow hazards at different recurrence intervals. The strategy presented in this paper aims to optimize control design and mitigate catastrophic risks.

2.   Geological setting

The Xinqiao Gully is on the right bank of the Min River in Xinqiao Village, Weizhou Town, Wenchuan County, Sichuan Province, China (Fig. 1). The gully’s outlet lies at 31°27′5.47″ N, 103°32′53.97″ E. The main channel flows through the culvert of the G213 and converges with the Min River, with Qipan Gully positioned about 800 m downstream. The gully comprises seven branch channels. Table 1 provides a detailed description of their topographic conditions. The main channel extends over a length of 4.92 km, covering a watershed area of 14.3 km². Various deposits, including flooding, collapse, landslide, and human-caused deposits, are found in the channel. Fig. 1 illustrates the implementation of prevention and control engineering measures following the Wenchuan earthquake to address the demand for disaster prevention.

Figure  1.  Geological map (a) and topographic (b) map of the study area. BC‒Branch Channel; CD‒Check Dam.

DownLoad: Full-Size Img PowerPoint

Table  1.  Topographic data in the Xinqiao Gully.

Gully	Channel length /km	Watershed area /km2	Basin relief /m	Average gully gradien /‰
XQG	4.92	14.3	1192	225
BC-1	0.90	0.35	505	564
BC-2	0.72	0.24	430	598
BC-3	1.66	1.97	798	595
BC-4	1.74	1.68	1008	600
BC-5	0.73	0.29	591	859
BC-6	0.61	0.31	388	648
BC-7	0.75	0.19	531	716
Notes: XQG refers to Xinqiao Gully, and BC refers to Branch Channel.

 | Show Table

DownLoad: CSV

The study area is in the Longmen Mountain fault zone, characterized by a complex geological structure. The maximum seismic acceleration in this region is 0.2 g, with a characteristic period of the seismic response spectrum at 0.3 s and an essential seismic intensity rated at VIII. The gully is on the upper plate of the Yingxiu fault and the Penguan anticlinorium’s southern wing. The Penguan anticlinorium, located in Wenchuan, represents the largest anticlinorium, extending 100 km in length and 35 km in width, forming an uplift in the shape of a shuttle along the NE40º to 45º axial direction, covering Yinxin, Yingxiu, and Sanjiang. The Maowen Fault, Judging Mountain Fault, and Wuzhao Mountain Fault have impacted the southwestern region of the northern wing. In contrast, the northeastern section incorporates the Upper Sinian-Permian, and the southern wing is nearly bisected by the Yingxiu Fault.

The rock outcrops in the study area primarily comprise phyllite, sandstone, and limestone. In the Quaternary layer, various types of soil with varying thicknesses are observed in different areas. Upstream, steep slopes and hilltops are covered by a 1‒2 m thick residual slope gravel soil. The midstream gentle slopes contain artificial landfill gravel soil with a 3‒10 m thickness. The gully banks support a layer of floodplain sandy gravel soil measuring 1‒4 m thick. The channel bed is encompassed by a 0.5‒3 m thick layer of sandy gravel floodplain soil. Downstream slopes are characterized by silt-filled gravelly soil, with thicknesses ranging from 2 m to 5 m at the toe and foot. The accumulation area consists of floodplain and debris flow deposits, varying from 3 m to 15 m thicknesses.

Wenchuan is affected by the Pacific’s warm high pressure in summer and the prevailing Northwest Plateau cold airflow in winter, resulting in two distinct natural climate zones. The valley area of Yingxiu and Xuankou districts south of Supaodian in Weizhou Town falls under the mountainous subtropical humid monsoon climate zone. In contrast, the valley area of Mianwu and Weizhou Towns north of Supaodian belongs to the warm-temperate continental semi-arid monsoon climate zone. The study area is in the semi-arid valley of the upper reaches of the Min River. The climate is dry with small and stable precipitation, with an annual measurement of 526.3 mm. Multi-year observations from nearby hydrological stations indicate that the 6-hour rainfall reached 53.3 mm, while the 1-hour rainfall reached 38.8 mm. The maximum daily precipitation in the area is 118.5 mm, with the highest 1-hour rainfall recorded at 42.2 mm and the maximum 10-minute rainfall at 11.5 mm.

The primary anthropogenic activities in the study area encompass farming and mining. The main residential area is downstream of the main ditch, near its confluence with the Minjiang River. On the other hand, the upstream branch ditch primarily serves as temporary housing for workers employed by Aba Mining Company. Site investigations indicate that farming activities are mainly concentrated near the residential area and have minimal impact on the surrounding vegetation. However, mining operations cause significant damage to vegetation, rock, and soil structures. The survey revealed that the slag produced during production was haphazardly piled in the trench, posing a high risk of debris flow disasters during heavy rainfall.

After investigation, the authors have gained preliminary knowledge of the historical mudslide events in this area. Before the Wenchuan earthquake, severe mudflows occurred in Xinqiao Gully in 1936, 1951, and 1998. Following the earthquake, the Department of Land and Resources of Sichuan Province conducted a post-earthquake geological disaster emergency investigation and design project, resulting in the prompt engineering of the ditch. The design parameters for the blockage project are in Table 2, and the distribution is in Fig. 1. However, the earthquake disturbed the geotechnical body, weakening its structure and resulting in numerous avalanche-slip accumulations that converged at the bottom of the slope and in the ditch. These accumulations provided the necessary material for the formation of debris flows. After the earthquake, two debris flow events occurred on July 10, 2013, and June 26, 2023. The June 26, 2023 event, which the authors investigated, destroyed conduits and blockage of culverts on the 213 National Highway flowing into the Min River. Post-event analyses concluded that the event was caused by erosion of the upstream channel headwaters by storm surface runoff resulting from short-calendar duration heavy rainfall, as shown in Fig. 2.

Table  2.  Control engineering information in the Xinqiao Gully.

Dam	Dam width/m	Dam height/m	Channel slope/‰	Deposition slope/‰	Design storage capacity/m3	Storage of sedimentation /m3
CD-1	33.5	3.5	123.0	86.0	3432.98	1142.44
CD-2	54.8	3.0	141.0	98.0	3991.43	3173.48
CD-3	21.6	3.5	176.3	123.2	2038.96	1977.64
CD-4	49.6	3.5	305.7	214.0	3006.38	2811.32
CD-5	48.6	4.5	222.0	147.0	4393.53	3506.90
CD-6	38.5	4.0	384.0	231.0	2013.07	2012.81
Notes: The data were obtained from field investigations. CD refers to Check Dam.

 | Show Table

DownLoad: CSV

Figure  2.  Characteristics of trigging rainfall in the Xinqiao Gully.

DownLoad: Full-Size Img PowerPoint

3.   Data and methods

Firstly, essential data is acquired through remote sensing interpretation, field investigation, and preliminary data organization. The parameters necessary for simulating the inversion of this debris flow event are determined through gradation tests, topographic analysis, image analysis, and empirical formulas. Following preprocessing, these parameters are input into the OpenLISEM model. Subsequently, the two-factor assessment model determines the hazard zoning situation under various recurrence intervals. The detailed process is illustrated in Fig. 3.

Figure  3.  Flow chart to analyze the preventive effect of the Xinqiao Gully debris flow and hazard assessment under different rainfall conditions.

DownLoad: Full-Size Img PowerPoint

3.1   Data and sources

Table 3 presents the data used in this study for analyzing debris flow characteristics and simulating the runout processes. The data mainly include (1) digital surface model (DSM); (2) panchromatic image (PI); (3) digital elevation model (DEM); (4) satellite image (SI); (5) lithology and fault (SL&F) data; (6) field survey measurements.

Table  3.  Data used and their sources in this study.

Name	Resolution	Source
Digital surface model	0.2 m	DJ Mavic 3 UAV equipped with an Real Time Kinematic (RTK) platform, Date: 2023.6.27
Panchromatic image	0.2 m	DJ Mavic 3 UAV equipped with an RTK platform, Date: 2023.6.27
Digital elevation model	12.5 m	ALOS (https://www.earthdata.nasa.gov/)
Satellite image	30.0 m	Geospatial Data Cloud (https://www.gscloud.cn/)
Lithology and fault	1∶50000	Chengdu Center of China Geological Survey
Note: DSM and PI data are available in the form of supplementary data.

 | Show Table

DownLoad: CSV

3.2   Models and parameters

3.2.1   OpenLISEM model

OpenLISEM, an open-source hydrological software developed by the University of Twente in the Netherlands, is proficient in assessing debris flow hazards in mountainous or urban areas. It can simulate surface runoff, floods, high-sediment flow, and debris flow in catchment areas. The simulation provides results for various parameters, including flux, velocity, flow depth, solids thickness, outflow extent, and accumulation area (Van B et al., 2021). Furthermore, the OpenLISEM model can be employed for hazard zoning and to evaluate the impact of control engineering. It encompasses a debris flow model based on Pudasaini’s generalized two-phase flow model, allowing for a smooth transition between inviscid, high-sediment, and debris flow. The governing equations (Equs. 1‒3) in OpenLISEM are as follows:

Where h is the flow depth, u and v are the flow velocities along the x and y directions, g is the gravity, S is the mass source, i.e., rainfall intensity minus infiltration intensity, S_x and S_y are the momentum sources, which are mainly generated by the action of volumetric and surface forces, the surface forces include the viscous force of the fluid, the friction between the solid phases, and the resistance between the two phases, and the volumetric force is mainly the gravitational force.

3.2.2   Simulation parameters

The elevation data in TIFF format were acquired through a DJ Mavic 3 UAV equipped with an RTK platform. The parameters and terrain data in OpenLISEM were converted to MAP format using a Nutshell script, facilitating the batching process’s automation. As per the study’s requirements and to optimize computational efficiency, the grid cells in the simulation were resampled to a 2 m × 2 m resolution.

(i) Debris flow density

The sampling point was situated at the accumulation area, where the granulation curve was obtained through a gradation test, as illustrated in Fig. 4. Results revealed that the refined grains (less than 0.05 mm) constituted 10.28% of the total. The coarse grains (greater than 2.0 mm) accounted for 55.42 %, indicating that almost half of the debris flow in Xinqiao Gully consists of coarse-grained material. The debris flow density, calculated using the sediment-based method (Equ. 4) (Yu B, 2008), was determined to be 2.00 g/cm³.

Figure  4.  Separation curves of debris flow sampling.

DownLoad: Full-Size Img PowerPoint

Where P₀₅ refers to the percentage content of refined grains with a particle size less than 0.05 mm expressed as a decimal, P₂ refers to the percentage content of coarse grains with a particle size greater than 2 mm expressed as a decimal, γ_ν represents the characteristic density of the debris flow, and the value of which is taken as 2.00 g/cm³, γ₀ is taken as 1.50 g/cm³, which is the minimum density of the debris flow.

(ii) Total solid matter

Field investigations and topographic imagery analysis determined that approximately 42.2 × 10³ m³ of solid material was released during this debris flow event, primarily sourced from deposits behind the dam, in the channel, and accumulation areas. Despite an estimated total residual design capacity of approximately 18.9 × 10³ m³ for the dams based on design parameters and pre-debris event imagery, the post-dam deposition capacity identified in Table 2 was lower, around 14.6 × 10³ m³ due to the dam failure. Utilizing GIS tools, the analysis of solid volume deposited in the channel by the debris flow revealed a total of 21.9 × 10³ m³. The resulting deposited fan in the Xinqiao Gully debris flow event covered an area of approximately 4.72 × 10³ m² with an average thickness of 1.2 m, estimating the volume of the deposited fan to be approximately 5.7 × 10³ m³, as depicted in Fig. 5.

Figure  5.  Flow fan of the Xinqiao Gully debris.

DownLoad: Full-Size Img PowerPoint

(iii) Dynamic viscosity

In OpenLISEM, Pudasaini devised the rheological model for debris flow based on the analogy with Newtonian fluids (Pudasaini SP, 2012), which includes the solid concentration gradient (Equ. 5). The interaction between the solid and liquid phases during movement generates shear stress due to the concentration gradient, contributing to a more uniform distribution of concentration.

Where A(α^f ) is the mobility of the fluid at the interface, η^f is the dynamic viscosity of the water, η_d is the dynamic viscosity of the debris flow, and $\nabla \alpha ^s$ is the solid concentration gradient. The dynamic viscosity was calculated using Equ. 6 to obtain. η_d is 0.18 Pa·s.

(iv) Peak discharge

Firstly, the authors calculated the rainstorm peak discharge at 20-year, 50-year, and 100-year recurrence using the reasoning formula method (Equ. 7). Then, the rain-flood method (Equ. 8) was employed for estimating the peak discharge of the debris flow (Cui P et al., 2011b). Following this, the volume of solid material during a debris flow event was determined using Equ. 9 and Equ. 10, the outcomes of which are detailed in Table 4.

Table  4.  Results of the rain-flood method.

Recurrence P/%	Maximum one-hour rainfall s/(mm/h)	Confluence time τ /h	Peak runoff coefficient ψ	Rainstorm peak discharge Qp /(m3/s)	Debris flow peak discharge of Qc/(m3/s)	Total flow of a debris flow event Q/m3	Solid material volume of a debris flow event QH/m3
5	29.85	2.92	0.83	46.00	140.11	13.32	8.07
2	36.30	2.72	0.87	61.37	186.94	17.77	10.77
1	41.10	2.60	0.89	73.22	223.05	21.20	12.85

 | Show Table

DownLoad: CSV

Where Q_p is the rainstorm peak discharge (m³/s), ψ is the peak runoff coefficient, s is the maximum one-hour rainfall (mm/h), τ is the confluence time (h), F is the catchment area (km³), taken as 14.3, Q_c is the peak discharge of debris flow (m³/s), Q is the total flow of a debris flow process (m³), Q_H is the solid material wash-out volume of a debris flow event (m³), D_c is the blocking coefficient, which is taken as 1.20 according to the field investigation results, γ_c is the density of debris flow (g/cm³), calculated by Equ. 4 and taken as 2.00, γ_w is the density of clear water (g/cm³), taken as 1.00, γ_h is the debris flow solid material density (g/cm³), taken as 2.65, ϕ is the correction coefficient for the bulk weight of the debris-flow, which is calculated by Equ. 8 and taken as 1.54, T is the duration of debris flow (s), according to the survey, take 3600.

3.3   Verification of simulation results

The accuracy of indicators such as the debris flow fan, flow velocity, flow depth, and run-out volume is often utilized for validation (Ding XY et al., 2023; Wang ZF et al., 2023). Flow velocity and depth are derived through inverse estimation utilizing the bend super-high section (Scheidl C et al., 2015), where the flow depth is estimated using the measured average flow depth, and the flow velocity is calculated using Equ. 11, with results detailed in Table 5. Moreover, the accuracy of the debris flow fan, Ad, is computed using Equ. 12, with a value range of [0,1], where a value closer to 1 indicates higher accuracy of the simulation results, and conversely, values nearing 0 denote less accurate simulation results. The run-out volume is estimated in section 3.2.2.

Table  5.  Statistics of the results of the measured section.

No.	Sectional width B /m	Average flow depth h /m	centerline radius Rc /m	Superelevation Δh /m	angle of the channel θc	empirical correction factor C	Debris flow velocity V /(m/s)
1	5.68	3.27	41	1.64	0.24	5.85	4.39
2	6.06	3.15	46	1.90	0.27	4.16	5.72
3	5.78	2.92	49	2.05	0.20	4.36	6.19

 | Show Table

DownLoad: CSV

In Equ. 11, V is the velocity, g is the gravity, θ_c is the angle of the channel, C is the empirical correction factor, B is the channel width, R_c is the centerline radius, h is the average depth of flow, and Δh is the superelevation. In Equ. 12, A_d is the accuracy of the debris flow fan, S₀² is the area of overlap within the fan, S_n is the area of the numerically modeled fan, and S_m is the area of the field surveyed fan.

Field investigation and analysis of satellite images (Figs. 6a, b) identified the initiation zone in BC-1 and BC-2. The resulting debris flow followed the main channel, causing erosion. Figs. 6c‒d illustrate the process of debris flow movement, primarily resulting in dam failure, road destruction, and siltation. The simulation of this event is centered on BC-1 and BC-2 as the initiation area.

Figure  6.  Satellite images before (a) and after (b) the debris flow, and disaster characteristic map (c) and (d).

DownLoad: Full-Size Img PowerPoint

Table 6 summarizes the simulated results (Fig.7) alongside the measured outcomes. The average relative errors for cross-section flow depth and flow velocity are 11.2 % and 8.5 %, respectively. The debris flow fan and run-out volume accuracy are reported at 70.2 % and 21.1 %, respectively. Notably, these errors stem from the debris flow’s traversal of the National Highway 213 through a road culvert, which needs to be accounted for in the DEM, resulting in a discrepancy between the outlet and the actual terrain. Nevertheless, the simulation accuracy for flow depth and flow velocity in the investigated section remains high, effectively capturing the movement process of the Xinqiao Gully debris flow.

Table  6.  Summary of verification methods.

Indicator	Measured values			Simulated values			Average relative errors /%
flow depth /m	3.27	3.15	2.92	2.88	2.91	2.51	11.2
flow velocity /(m/s)	4.39	5.72	6.19	4.62	4.95	5.77	8.5
run-out volume /104m3	0.57			0.69			21.1
accuracy of the debris flow fan /%	area of the field surveyed fan /103m3			area of the numerically modeled fan /103m3			area of overlap within the fan	accuracy
	4.72			6.47			4.63	70.2

 | Show Table

DownLoad: CSV

Figure  7.  Verification analysis of debris flow simulation. The numbers 1, 2 and 3 represent the measurement section numbers.

DownLoad: Full-Size Img PowerPoint

4.   Control effectiveness analysis and debris flow hazard evaluation

4.1   Control effectiveness analysis

The results of the inversion of the Xinqiao Gully debris flow event are presented in Fig. 8. Figs. 8a‒b depict the prominent post-dam siltation and pre-dam acceleration experienced after the debris flow passes through the check dam (Yang HQ et al., 2021). Post-dam siltation occurs as solid material accumulates behind a dam due to the interception of debris flow, leading to a significant increase in sediment depth compared to before the dam. On the other hand, pre-dam acceleration refers to the acceleration that follows the passage of a debris flow over a dam. This acceleration is attributed to the dam’s obstruction of flow, resulting in the conversion of kinetic energy into gravitational potential energy (Chen Z et al., 2022). Once the potential energy reaches a certain threshold, it is transformed back into kinetic energy, leading to the acceleration of the flow. This paper proposes to employ the storage capacity fullness rate and flow intensity reduction rate as indices for assessing the impact of interception and energy reduction in dam projects. The indices are expressed as follows (Equs, 13, 14):

Figure  8.  Simulation results of the Xinqiao Gully debris flow. a‒maximum depth; b‒maximum velocity.

DownLoad: Full-Size Img PowerPoint

Where I_F is the flow intensity reduction rate, and the result is expressed as a % symbol, HV_U is the flow intensity upstream of the dam (m²/s), HV_D is the flow intensity downstream of the dam (m²/s), H is the flow depth (m), and V is the flow velocity (m/s). R is the dam’s storage capacity fullness rate, Q_A is sedimentation storage, and Q_D is the design storage capacity.

The Table 7 summary reveals that the check dam achieves an average flow intensity reduction of 41.05% to 64.61%. Notably, CD-6 experiences a lower flow intensity reduction rate than other dams, potentially due to its proximity to the debris flow source and rapid burial during the debris flow initiation. Consequently, this dam’s effectiveness in mitigating subsequent debris flows is limited, underscoring the impact of blockage when dams remain uncleared post-debris flow events. The storage capacity fullness rate of the check dam progressively decreases from the upstream to the mouth of the ditch, with CD-5 being an exception due to a localized failure, indicating successful interception and silting during the flow process. However, CD-2, situated near the residential area at the gully outlet, retains a high storage capacity fullness rate, suggesting a need for increased control of potential large-scale debris flows. It is important to note that the indicators proposed in this study aimed to quantify the control effectiveness on expected debris flows rather than after the event, in contrast to similar indicators such as the flow reduction rate. Analysis of hydro-meteorological data established the debris flow’s recurrence period to be between 10 to 20 years, warranting detailed hazard analysis under more severe working conditions.

Table  7.  Calculated results of flow intensity reduction rate.

Dam	Flow intensity upstream of the dam HVU /(m2/s)	Flow intensity downstream of the dam HVD /(m2/s)	Flow intensity reduction rate IF /%	Storage capacity fullness rate R /%
CD-1	10.81	4.31	60.13	33.28
CD-2	19.84	7.02	64.61	79.51
CD-3	10.60	6.25	41.05	96.99
CD-4	18.05	9.65	46.56	93.51
CD-5	39.03	15.09	61.33	79.82
CD-6	23.71	21.43	9.60	99.99

 | Show Table

DownLoad: CSV

4.2   Indexes of debris flow hazard zoning

The hazard zoning of debris flows involves considering intensity and recurrence, as detailed in Tables 8, 9 (Ouyang CJ et al., 2015). Debris flow intensity is typically related to the flow’s depth and impact force, generally expressed as a function of the flow intensity (Yan Y et al., 2023). Therefore, it is customary to utilize a two-factor model incorporating maximum flow depth and momentum coupling to classify the intensity of debris flows. These debris flow hazards are classified into various scenarios based on recurrence, including 20-year, 50-year, and 100-year cases.

Table  8.  Standards of debris flow intensity zoning.

Intensity	Depth h /m	Logic	Flow intensity hv /(m2/s)
high	h $\geqslant$ 3.0	∪	hv $\geqslant$ 8.0
medium	1.5 $\leqslant$ h < 3.0	∩	2.0 $\leqslant$ hv < 8.0
low	0.0 $\leqslant$ h < 1.5	∩	0.0 $\leqslant$ hv < 2.0

 | Show Table

DownLoad: CSV

Table  9.  Hazard classification by intensity and probability of occurrence.

Recurrence Intensity	[0,20) high	[20,50] medium	(50,100] low
high	high	high	medium
medium	high	medium	low
low	medium	low	low

 | Show Table

DownLoad: CSV

4.3   Results of debris flow hazard assessment

The results of debris flow intensity zoning, based on Table 8, are visualized in Fig. 9. The red area, signifying high intensity, predominantly encompasses the inner part of the ditch, while the yellow area, indicating medium intensity, is concentrated on both sides of the nearby ditch. The green area, representing low intensity, is predominantly located in the upstream Aba Mining Company's building complex and the downstream accumulation area, situated away from the residential area near the ditch. Statistical findings (Table 10) indicate that as rainfall intensity rises, the area of high-intensity rainfall expands from 161.72 × 10³ m² to 231.54 × 10³ m², with its percentage increasing from 53.37 % to 57.49 %. Although the medium- and low-intensity rainfall areas similarly increase significantly, the percentage changes are negligible. The primary areas at risk include the ditches and banks.

Figure  9.  Debris flow intensity zoning of different recurrence intervals. a‒20-year recurrence interval; b‒50-year recurrence interval; c‒100-year recurrence interval. SH‒Secondary highway; RUC‒Road under construction.

DownLoad: Full-Size Img PowerPoint

Table  10.  Areas classified as relatively low, medium, and high intensity.

Recurrence P /%	High-intensity area			Medium-intensity area			Low-intensity area
	Area /103m2	Proportion /%		Area /103m2	Proportion /%		Area /103m2	Proportion /%
5%	161.72	53.37		57.73	19.05		83.57	27.58
2%	204.70	56.40		72.83	20.07		85.39	23.53
1%	231.54	57.49		76.58	19.02		94.59	23.49

 | Show Table

DownLoad: CSV

Relying on the intensity of debris flow to delineate hazardous areas has limitations (Luna BQ et al., 2011). Combining intensity and occurrence probability levels (Table 9) is necessary to establish a hazard map of Xinqiao Gully debris flow. This process involves the utilization of three hazard grades: High, medium, and low (Fig. 10), overlaying the zoning situation of different recurrences. The high-hazard zone, which covers an area of 229.07×10³ m², represents 56.67 % of the total threatened area and is primarily located in the fast and deep areas within the inner part of the ditch. The medium-hazard zone encompasses an area of 95.46×10³ m², accounting for 23.62 % of the total threatened area, predominantly situated on either side of the ditch, with a fraction adjacent to the highway near the ditch bank. The low danger zone covers an area of 79.67×10³ m², constituting 19.71 % of the total threatened area (Table 11), and is primarily situated in the accumulation area at the mouth of the ditch and on gently sloping terrain. Notably, the area of low-hazard zones for debris flow disasters with a 5% recurrence and the area of high-hazard zones for debris flow disasters with a 1% recurrence are both zero due to their infrequency.

Figure  10.  Debris flow hazard zoning of different recurrence intervals. a‒20-year recurrence interval; b‒50-year recurrence interval; c‒100-year recurrence interval; d‒hazard overlying map.

DownLoad: Full-Size Img PowerPoint

Table  11.  Areas classified as relatively low, medium, and high hazard.

Recurrence P /%	High-hazard area			Medium-hazard area			Low- hazard area
	area /103m2	proportion /%		area /103m2	proportion /%		area /103m2	proportion /%
5	219.45	72.42		83.57	27.58		0	0
2	204.70	56.40		72.83	20.07		85.39	23.53
1	0	0		231.54	57.49		171.18	42.51
overlying	229.07	56.67		95.46	23.62		79.67	19.71

 | Show Table

DownLoad: CSV

The Xinqiao Gully debris flow exhibits consistent characteristics that can lead to disasters during various occurrences, primarily impacting the channel and its surrounding areas. It is crucial to closely monitor the construction area of the upstream Aba Mining Company and the residential area at the downstream gully mouth. Furthermore, debris flow in certain gully sections can result in road damage and siltation, hindering emergency investigations and rescue operations post-disaster. Additionally, with the intensification of rainfall, the high-hazard area for debris flow expands beyond the ditches, posing an increasing threat to residents living along the banks and highway facilities. To mitigate these risks, we propose improving the blocking project between CD-3 and CD-5 in the upstream section of the ditch. Moreover, raising the height of the downstream drainage channel and performing anti-blockage reconstruction on the culvert at the outlet of the G213 can help minimize the threat of debris flow to the safety of people’s lives and public infrastructure.

5.   Discussion

This study provides a practical approach for quantitatively evaluating the effectiveness of debris flow control projects in similar areas. The findings reveal a notable decrease in flow intensity during the active phase of the Xinqiao Gully debris flow control project, affirming the effectiveness of the engineering measures. Furthermore, considering intensity and recurrence interval, the hazard zoning assessment offers a comprehensive understanding of the potential risks associated with debris flow in the region. Additionally, using the OpenLISEM model for quantitative analysis highlights the vital importance of implementing effective control measures to mitigate the impact of debris flow on the region’s livelihood and public infrastructure.

5.1   Applicability of the OpenLISEM Model

Debris flow can be categorized into slope debris flow and gully debris flow based on watershed geomorphology. The crucial factor for the numerical simulation of slope debris flow is determining the distribution of debris sources (Ni HY et al., 2014). In contrast, gully debris flow primarily involves identifying the position of drainage points and the flow discharge curve (Chang M et al., 2017). In the case of the Xinqiao Gully debris flow event, the initiation area, characterized by a short, indistinct valley form and high position, can be considered a process of slope erosion starts to form debris flow.

The OpenLISEM model has been extensively utilized for quantitative analysis and assessment of debris flow disasters (Van B et al., 2021). This study has demonstrated its effectiveness in simulating debris flow events and evaluating the impact of engineering control measures. However, it is important to recognize certain limitations and drawbacks of the model. As the current version lacks the function to set drainage points and flow discharge curves, OpenLISEM is unsuitable for gully debris flow and can be replaced by Flo-2D (Kurovskaia VA et al., 2022) and Massflow (Horton AJ et al., 2019) software. Despite providing valuable quantitative analysis capabilities, the accuracy and reliability of OpenLISEM may be influenced by factors such as setting simulation boundary conditions and grid accuracy. Therefore, future research should address these limitations by utilizing local grid refinement techniques.

5.2   Evaluation of Debris Flow Control Effectiveness

Previous studies evaluating the effectiveness of debris flow control have primarily relied on post-disaster qualitative or semi-quantitative analyses, overlooking the assessment of control measures (Tang C et al., 2011; Li M et al., 2018). Hence, this paper conducts a quantitative study on the effectiveness of debris flow control engineering, aiming to improve the understanding of the efficacy of engineering measures in mitigating debris flow hazards, which can, in turn, inform disaster emergency management. The results show that through the careful implementation of control engineering, the intensity of debris flow can be significantly reduced, consequently mitigating the activity intensity of debris flow and confirming the effectiveness of such measures in minimizing potential hazards.

5.3   Disaster prevention and mitigation recommendations

Mountain disaster prevention and control in southwest China boast a rich historical background (Cui P et al., 2003; Cui P et al., 2011a; Chen X et al., 2015a). Grounded in the three fundamental principles of prevention, control, and avoidance, a range of methods, including geotechnical engineering and bioengineering, are employed to establish systems focusing on preventing debris flow occurrence (SPDO), controlling debris flow movement (SCDM), and averting debris flow damage (SPDD) (Cui P et al., 2011a; Chen X et al., 2015). Specifically, the approach adopted in the Xinqiao Gully centers on SCDM, involving blocking engineering in the upper and middle gully sections and drainage engineering in the downstream accumulation area.

However, Current research indicates that under the most perilous conditions of the Xinqiao Gully debris flow, further optimization and enhancement of control measures are necessary to ensure effective mitigation of debris flow hazards. It is recommended to bolster solutions such as increasing interception projects and rebuilding drainage channels to address potential shortcomings and enhance disaster prevention and reduction capabilities. Moreover, future research could emphasize the development of comprehensive risk management strategies for debris flow events within the broader area. It may involve utilizing advanced modeling techniques, integrating real-time monitoring systems, and exploring innovative engineering solutions to fortify the region’s safety and disaster resilience.

6.   Conclusions

In general, secondary geological hazards such as landslides and debris flows will continue to occur for several years, decades, and even centuries after a mega-earthquake. These hazards cause secondary damage to the local population. Geologists have a responsibility to go deep into the disaster area to study the hazards and make management recommendations to reduce future losses.

In this study, investigating the debris flow event in the Xinqiao Gully involved a quantitative analysis of the effectiveness of debris flow control measures. The simulation obtains the distribution of maximum flow velocity and depth of debris flow with recurrence intervals of 20, 50, and 100 years to predict the debris flow disaster under different rainfall frequencies. Subsequently, the intensity and hazard zoning were determined for various rainfall frequencies. Finally, a hazard overlay map was generated utilizing a two-factor evaluation model. The conclusions drawn from the research are as follows.

(i) Field investigation revealed that the source of the debris flow originates in areas BC-1 and BC-2, following the process of rainfall, confluence, initiation of runoff scour, debris flow, accumulation of blocked silt, overtopping, and movement along the ditch into the river.

(ii) The impact of control engineering on the debris flow process in this event was analyzed based on the OpenLISEM model and field investigation. It was suggested to utilize indicators such as the rate of reduction in flow intensity and the rate of fullness of the dam’s storage capacity for measuring the impact of control engineering. The results show a decrease in Xinqiao Gully debris flow intensity ranging from 41.05 % to 64.61 %, with the storage capacity of dams decreasing from the upstream to the outlet. It suggests successful interception of debris flow and prevention of siltation during the movement process. However, it is essential to ensure timely removal of sediment buildup behind the dam.

(iii) The two-factor model was utilized to evaluate the hazard of debris flow zoning. Findings revealed that the high-hazard area is concentrated in the ditch channel, while the medium-hazard area is located on both sides. The low-hazard area is primarily situated in the construction area of Aba Mining Company and the residential area downstream of the ditch mouth. It was noted that as rainfall intensity increases, the hazardous area expands to the surrounding areas.

(iv) Based on investigation and simulation prediction analysis, increasing the height of the CD-5 and CD-6 dams upstream and constructing additional barriers between the CD-3 and CD-5 is recommended. Moreover, consider increasing the height or width of the drainage ditch or installing additional barrier structures for roads or residential areas near the ditch.