A method and system for evaluating the risk of debris flow in a small watershed in a loess gully region

Abstract

The present application provides a kind of loess gully region small watershed debris flow risk assessment method and system, it is related to geological disaster assessment technical field, method includes determining research area, obtains the multi-source data of research area, then uses random forest model to carry out prone to evaluate screening target channel, thereafter, build the rainfall erosion physical model test of heaped loess slope, analyze erosion failure mechanism and accurately demarcate debris flow source area, finally, using MassFlow software that is inverted and calibrated parameter based on historical data to carry out single ditch debris flow numerical simulation, and based on the momentum parameter of mud depth and flow velocity, risk zoning is carried out, and the evaluation result is output;System includes prone to preliminary screening module, source analysis module, numerical simulation module and risk assessment module;The present application is combined by random forest model, physical test and numerical simulation inversion calibration, forms chain type evaluation system, improves the precision and reliability of debris flow risk assessment.

Description

Technical Field

[0001] This invention relates to the field of geological hazard assessment technology, and in particular to a method and system for assessing the risk of debris flows in small watersheds in loess gullies. Background Technology

[0002] The loess gully region has complex geological conditions and frequent human engineering activities, such as land reclamation by leveling mountains and filling gullies, resulting in a large number of loess-filled slopes. These loess-filled slopes have a loose structure and poor stability, making them extremely prone to instability under heavy rainfall conditions, and thus becoming a major source of debris flows.

[0003] Currently, the methods for assessing debris flow hazard in small watersheds of the Loess gully region have the following main shortcomings: First, regional susceptibility assessments often employ a single model (such as the information content model or logistic regression model). Since it is difficult to balance the predictive accuracy and stability of the model, the assessment results are prone to significant bias when the training data contains noise or the sample distribution is uneven.

[0004] Secondly, existing methods are insufficient for accurately delineating the source area for the risk assessment of single-channel debris flows. In particular, there is a lack of quantitative description of the initiation conditions and recharge process of special sources such as artificial landfills. The delineation of the source area often relies on empirical judgment, resulting in insufficient objectivity and repeatability of the assessment results.

[0005] Third, the selection of key parameters (such as friction coefficient and turbulence coefficient) in numerical simulations relies heavily on experience and lacks an inversion calibration method that combines indoor physical experiments and field survey data. The parameter values ​​deviate significantly from the actual situation, reducing the reliability of the numerical simulation results.

[0006] Therefore, existing debris flow hazard assessment methods still suffer from technical problems in their application to small watersheds in the Loess Gully region, such as inaccurate source delineation, strong subjectivity of model parameters, and low reliability of assessment results. Therefore, this invention proposes a debris flow hazard assessment method and system for small watersheds in the Loess Gully region to solve the problems existing in the prior art. Summary of the Invention

[0007] To address the aforementioned problems, the present invention aims to propose a method and system for assessing the hazard of debris flows in small watersheds of the Loess Gully Region. This invention has the advantage of improving the accuracy and reliability of debris flow hazard assessment in small watersheds of the Loess Gully Region, thereby solving the problems existing in the prior art.

[0008] To achieve the objectives of this invention, the invention is implemented through the following technical solution: a method for assessing the hazard of debris flows in small watersheds of loess gullies, characterized by comprising the following steps: Step 1: Determine the study area: Divide the small watersheds to be evaluated within the Loess Gully Region into different study areas; Step 2, Preliminary selection of regional susceptibility: Obtain multi-source data of the study area, and then use the random forest model to evaluate the susceptibility of debris flow in the study area, draw a susceptibility zoning map, and select typical debris flow channels in the extremely high susceptibility area as target channels based on the evaluation results. The multi-source data includes topographic factors, ecological environment factors, geological structure factors and human activity factors. Step 3: Precise delineation and mechanism analysis of the source area: For the target gullies selected in Step 2, field investigation and sampling are carried out, and a physical model test of rainfall erosion of the loess fill slope is constructed. Then, the erosion and damage mechanism of the loess fill slope is analyzed to accurately delineate the debris flow source area. Step 4: Numerical simulation and hazard assessment of debris flow in a single gully: Based on the debris flow source area delineated in Step 3, MassFlow software with parameters calibrated by inversion from historical debris flow event investigation data is used to numerically simulate the debris flow movement process. Based on the numerical simulation results, hazard zoning is performed, and the debris flow hazard assessment results of the target gully are output.

[0009] A further improvement is made in the following way: In step three, the specific method for constructing the physical model test of rainfall erosion of the loess slope is as follows: Construct an indoor rainfall simulation test device, which includes a model box, artificial rainfall equipment, and volumetric water content sensors, matrix suction sensors, and pore water pressure sensors installed in the slope model. Then, the indoor rainfall simulation test device is used to simulate the erosion and damage process of the loess slope under different rainfall intensities, record the changes in the volumetric water content, matrix suction, and pore water pressure of the slope, and divide the erosion and damage process into splash erosion stage, gully erosion stage, and gully erosion and collapse damage stage.

[0010] Further improvements are made in the following: In step four, the MassFlow software, which uses historical debris flow event survey data to invert and calibrate parameters, specifically includes: establishing a digital elevation model of the target channel using UAV aerial survey, and then, based on historical debris flow event survey data, calculating and calibrating the friction coefficient and turbulence coefficient of the Voellmy model in the MassFlow software through inversion.

[0011] A further improvement lies in the following: In step four, the hazard zoning based on the numerical simulation results specifically includes: Based on the mud depth and flow velocity data output by numerical simulation, momentum parameters are calculated, where the momentum parameter is the product of mud depth and flow velocity. Then, according to the momentum parameter and mud depth data, the debris flow intensity is divided into high intensity, medium intensity and low intensity levels. Finally, combined with the preset debris flow occurrence frequency, a hazard zoning map of the target channel is generated.

[0012] The further improvement lies in that the friction coefficient of the Voellmy model is 0.1 to 0.3, and the turbulence coefficient of the Voellmy model is 100 to 300.

[0013] The further improvement lies in the fact that the friction coefficient of the Voellmy model is 0.15 and the turbulence coefficient of the Voellmy model is 200.

[0014] A debris flow hazard assessment system for small watersheds in loess gully areas includes the following modules: The susceptibility screening module is used to load the random forest model and evaluate the susceptibility of debris flows in the study area, and then screen out typical debris flow channels in the extremely high susceptibility area as target channels. The source analysis module is used to conduct rainfall erosion tests on the loess slopes in the target gully, analyze the erosion and damage mechanism of the loess slopes, and delineate the debris flow source area. The numerical simulation module is used to load MassFlow software, which uses historical debris flow event investigation data to retrieve and calibrate parameters, and to perform numerical simulations of the debris flow movement process based on the debris flow source area. The hazard assessment module receives mud depth and flow velocity data output from the numerical simulation module, and generates a hazard zoning map based on the mud depth and flow velocity data, thereby outputting the debris flow hazard assessment results for the target channel.

[0015] Further improvements are made in that: the source analysis module includes a rainfall simulation test unit and a sensor monitoring unit. The rainfall simulation test unit is used to simulate the slope erosion process under different rainfall intensities. The sensor monitoring unit includes a volumetric water content sensor, a matrix suction sensor, and a pore water pressure sensor, which are used to monitor the hydraulic parameters of the slope in real time during the rainfall erosion process.

[0016] The beneficial effects of this invention are as follows: (1) This invention uses a random forest model to evaluate regional susceptibility, which can handle multi-source high-dimensional data and reduce the risk of overfitting of a single model through an ensemble learning mechanism. Compared with traditional single models, the random forest model can still maintain high prediction stability under sample noise and imbalance, thus providing a scientific basis for the screening of typical debris flow channels.

[0017] (2) This invention targets the special material source type of artificially filled loess slopes. By constructing an indoor rainfall simulation test device, combined with a volumetric water content sensor, a matrix suction sensor and a pore water pressure sensor, the hydraulic parameters of the filled loess slopes are monitored in real time during the rainfall erosion process. The erosion and destruction process is divided into splash erosion stage, gully erosion stage, gully erosion and collapse destruction stage. Based on the above physical test results, the filled loess areas that are prone to instability and transformation into debris flow material sources can be accurately identified, which solves the problem of insufficient objectivity caused by the traditional method of relying on experience to delineate material sources.

[0018] (3) This invention uses UAV aerial survey to establish a digital elevation model of the target channel, and based on historical debris flow event survey data, it calibrates the friction coefficient and turbulence coefficient of the Voellmy model in MassFlow software through inversion calculation. Thus, the closed-loop verification of numerical simulation and field survey data is realized through this parameter calibration method. Compared with the traditional empirical value method, it significantly reduces the subjectivity of parameter selection and improves the credibility of debris flow motion process simulation results.

[0019] (4) This invention forms a three-level chain evaluation system from “regional susceptibility assessment to source mechanism test to single gully fine simulation”. From the initial selection of susceptibility at the regional scale to the fine numerical simulation at the gully scale, it focuses on each level and progresses step by step. It is particularly suitable for the assessment of debris flow hazard in small watersheds of the Loess Gully Area that are strongly disturbed by human engineering activities (such as land reclamation and gully filling). Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the method steps of the present invention.

[0021] Figure 2 This is a schematic diagram of the structural framework of the indoor rainfall simulation test device of the present invention. Detailed Implementation

[0022] To enhance understanding of the present invention, the present invention will be further described in detail below with reference to embodiments. These embodiments are only used to explain the present invention and do not constitute a limitation on the scope of protection of the present invention.

[0023] Example 1 according to Figure 1 As shown in the figure, this embodiment provides a method for assessing the hazard of debris flows in small watersheds in loess gullies, including the following steps: Step 1: Determine the study area A small watershed within the Loess Gully Region was selected as the area to be evaluated. This small watershed was divided into three different study areas according to the distribution of the water system and the topographic features, which are denoted as Region A, Region B and Region C respectively.

[0024] Step 2: Preliminary selection of regional susceptibility The study area was acquired using multi-source data, which included topographic and geomorphological factors, ecological and environmental factors, geological structural factors, and human activity factors. Specifically, topographic and geomorphological factors included slope, aspect, elevation, and watershed area; ecological and environmental factors included vegetation cover and soil type; geological structural factors included distance from faults and lithology; and human activity factors included road network density, land reclamation area, and distribution of landfill disposal sites.

[0025] A random forest model was used to assess the susceptibility of debris flows in the study area. The number of decision trees in the random forest model was set to 500, and the maximum depth was set to 10. A susceptibility zoning map was output, dividing the area into extremely high susceptibility, high susceptibility, medium susceptibility, and low susceptibility zones. In this embodiment, the susceptibility zoning map shows that: area A is an extremely high susceptibility zone, area B is a high susceptibility zone, and area C is a medium susceptibility zone. Then, based on the assessment results, a typical debris flow channel within the extremely high susceptibility zone was selected as the target channel, that is, a typical debris flow channel was selected from area A as the target channel.

[0026] Step 3: Precise delineation of the source region and mechanism analysis For the target gullies selected in step two, field surveys and sampling were conducted. Uncirculated soil samples were collected from the loess fill slopes within the target gullies. The thickness, slope, aspect, and vegetation cover of the fill were recorded during the field survey.

[0027] A physical model experiment was conducted on a loess fill slope subjected to rainfall-induced erosion. An indoor rainfall simulation test device was constructed. This device included a model box, artificial rainfall equipment, and volumetric moisture content sensors, matrix suction sensors, and pore water pressure sensors deployed within the slope model. The model box measured 2 meters long, 1 meter wide, and 0.8 meters high. The artificial rainfall equipment used a nozzle-type rainfall simulator with an adjustable rainfall intensity range of 10 mm / h to 150 mm / h. Volumetric moisture content sensors, matrix suction sensors, and pore water pressure sensors were deployed at different depths (5 cm, 15 cm, and 25 cm) within the slope model.

[0028] The erosion and damage process of loess slopes under different rainfall intensities was simulated using an indoor rainfall simulation test device. Three rainfall intensities were set: 30 mm / h, 60 mm / h, and 90 mm / h. Each rainfall lasted for 120 minutes. The changes in volumetric water content, matrix suction, and pore water pressure of the slope were recorded in real time.

[0029] Observe the surface damage phenomena of the slope. The erosion and damage process is divided into three stages: splash erosion and sheet erosion stage (in the early stage of rainfall, splash pits and sheet erosion appear on the slope), gully erosion stage (with continuous rainfall, gullies appear on the slope, with a depth of 1 cm to 3 cm), and gully erosion and collapse damage stage (with increased or continuous rainfall intensity, gullies develop into gullies, with a depth of more than 5 cm, and local collapse of the slope occurs).

[0030] Based on the experimental results and combined with field surveys, the loess fill areas in the target gullies that are prone to instability and transformation into debris flow sources were accurately identified. These areas were then vectorized in GIS software to create a precise debris flow distribution map.

[0031] Step 4: Numerical Simulation and Hazard Assessment of Single-Ditch Debris Flow A digital elevation model of the target trench was established using UAV aerial surveying. The UAV was equipped with a lidar, flew at an altitude of 150 meters, and had a point cloud density of 10 points per square meter. The generated digital elevation model had a grid resolution of 1 meter.

[0032] Based on historical debris flow event survey data, the friction coefficient and turbulence coefficient of the Voellmy model in MassFlow software were calibrated through inversion calculation. The historical debris flow event survey data included the debris flow occurrence time, deposition range, maximum deposition thickness, and estimated flow velocity. A trial-and-error method was used for inversion: the initial value of the friction coefficient was set to 0.2, and the initial value of the turbulence coefficient was set to 200. MassFlow software was run to simulate debris flow movement, and the simulated deposition range was compared with the historically surveyed deposition range. Parameters were adjusted until the two matched. After inversion calculation, the friction coefficient was calibrated to 0.15, and the turbulence coefficient was calibrated to 200.

[0033] Based on the calibrated Voellmy model and a preset 100-year rainfall frequency (150 mm of rainfall in 24 hours), the entire process of debris flow is simulated. Velocity distribution maps and mud depth distribution maps are output with a spatial resolution of 1 meter.

[0034] Based on the mud depth and flow velocity data output from the numerical simulation, momentum parameters are calculated. The formula for calculating momentum parameters is: hν = h × ν In the formula, h represents mud depth and ν represents flow velocity. Based on momentum parameters and mud depth data, debris flow intensity is divided into three levels: high intensity (hν > 5 m² / s and h > 2 m), medium intensity (hν 1 to 5 m² / s and h 0.5 to 2 m), and low intensity (hν < 1 m² / s and h < 0.5 m). Combining this with the frequency of debris flows occurring once every 100 years, a hazard zoning map of the target channel is generated. The hazard zoning map includes high-risk, medium-risk, and low-risk areas. The debris flow hazard assessment results for the target channel are output.

[0035] Furthermore, the above-mentioned grading thresholds are determined based on the characteristic parameters of the target gully in this embodiment. In practical applications, they can be adjusted according to the specific topographic conditions of the gully and the investigation data of historical debris flow events.

[0036] Furthermore, the accuracy of the evaluation results of the random forest model was verified using receiver operating characteristic (ROC) curves. The specific verification process is as follows: The first step was to construct a sample dataset: 58 historical debris flow event locations within the study area were collected as positive samples, and 58 locations were randomly selected from non-debris flow development areas (i.e., low-risk and extremely low-risk areas in the susceptibility zoning map) within the study area as negative samples. The feature vector of each sample point includes 12 feature variables: topographic factors, ecological environment factors, geological structure factors, and human activity factors.

[0037] The second step is to divide the data into training and validation sets: The 116 sample points are randomly divided into training and validation sets in a 7:3 ratio. The training set contains 81 samples, including 41 positive samples and 40 negative samples. The validation set contains 35 samples, including 17 positive samples and 18 negative samples.

[0038] The third step is to train the model: Use the training set to train the random forest model. Set the number of decision trees in the random forest model to 500, the maximum depth to 10, and use the default values ​​for the other parameters.

[0039] The fourth step is to predict the class probability of the validation set samples: use the trained model to predict the 35 samples in the validation set and output the predicted probability value of each sample belonging to the positive sample (mudslide point).

[0040] Step 5: Plot the Receiver Operating Characteristic (ROC) curve: Sort the validation set samples from highest to lowest predicted probability. Using each predicted probability value as a classification threshold, calculate the true positive rate and false positive rate for each threshold. The true positive rate is calculated as: True positive rate = Number of correctly predicted positive samples / Total number of actual positive samples. The false positive rate is calculated as: False positive rate = Number of incorrectly predicted positive samples / Total number of actual negative samples. Plot the ROC curve with the false positive rate on the x-axis and the true positive rate on the y-axis.

[0041] Step 6: Calculate the AUC value. Integrate the area under the receiver operating characteristic curve to obtain an AUC value of 0.97.

[0042] Example 2 This embodiment provides another method for assessing the risk of debris flow in small watersheds in the Loess Gully Region. The difference between this method and Embodiment 1 is that the values ​​of the friction coefficient and turbulence coefficient in the Voellmy model are different.

[0043] Specifically, based on historical debris flow event survey data, the friction coefficient and turbulence coefficient of the Voellmy model in MassFlow software were calibrated through inversion calculation. The inversion calibration process is as follows: the initial value of the friction coefficient was set to 0.3, and the initial value of the turbulence coefficient was set to 120. The MassFlow software was run to simulate debris flow movement, and the simulated deposition range was compared with the deposition range of historical surveys. The parameters were adjusted until the two matched. After inversion calculation, the friction coefficient was calibrated to 0.25, and the turbulence coefficient was calibrated to 150.

[0044] Simulation results show that when the friction coefficient is 0.25 and the turbulence coefficient is 150, the simulation results are comparable in accuracy to those with a friction coefficient of 0.15 and a turbulence coefficient of 200, both meeting the requirements for engineering applications.

[0045] Example 3 according to Figure 2 As shown, this embodiment provides a debris flow hazard assessment system for small watersheds in the Loess gully region, including the following modules: a susceptibility screening module, a source analysis module, a numerical simulation module, and a hazard assessment module. Specifically: The susceptibility screening module loads a random forest model to evaluate the susceptibility of debris flows in the study area and selects typical debris flow channels in extremely high-susceptibility areas as target channels. The random forest model can be implemented using various programming languages ​​and machine learning frameworks, such as Python with the scikit-learn library.

[0046] The sediment source analysis module includes a rainfall simulation test unit and a sensor monitoring unit. The rainfall simulation test unit simulates slope erosion processes under different rainfall intensities. It includes a model box and artificial rainfall equipment. The sensor monitoring unit includes a volumetric water content sensor, a matrix suction sensor, and a pore water pressure sensor, used to monitor the hydraulic parameters of the slope in real time during rainfall erosion. Sensor data is acquired once per second and transmitted to the host computer via a data acquisition card.

[0047] The numerical simulation module loads MassFlow software with parameters calibrated by inversion from historical debris flow event investigation data, and performs numerical simulations of the debris flow movement process based on the debris flow source region. The MassFlow software uses the Voellmy model.

[0048] The hazard assessment module receives mud depth and flow velocity data output from the numerical simulation module, generates a hazard zoning map based on the mud depth and flow velocity data, and outputs the debris flow hazard assessment results for the target channel.

[0049] The system operates as follows: First, the susceptibility screening module outputs the range of the target channel. Then, the source analysis module delineates the precise source area through rainfall simulation experiments. Next, the numerical simulation module calls MassFlow software to perform numerical simulation. Finally, the hazard assessment module generates a hazard zoning map and outputs the assessment results.

[0050] Example 4 This embodiment provides another debris flow hazard assessment system for small watersheds in the Loess gully region. The difference from Embodiment 3 lies in the friction coefficient (0.15) and turbulence coefficient (200) of the Voellmy model in the MassFlow software used in the numerical simulation module. Other modules are the same as in Embodiment 3. The operational results show that using a friction coefficient of 0.15 and a turbulence coefficient of 200 yields the highest agreement between the numerical simulation results and the actual debris flow deposition range identified in the field survey, with an overlap rate of 92%. Compared to systems using other parameter values, the simulation results of this embodiment show the highest agreement with the actual debris flow deposition range identified in the field survey.

[0051] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the present invention without departing from its framework and scope of application, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for assessing the hazard of debris flows in small watersheds of loess gullies, characterized in that: Includes the following steps: Step 1: Determine the study area: Divide the small watersheds to be evaluated within the Loess Gully Region into different study areas; Step 2, Preliminary selection of regional susceptibility: Obtain multi-source data of the study area, and then use the random forest model to evaluate the susceptibility of debris flow in the study area, draw a susceptibility zoning map, and select typical debris flow channels in the extremely high susceptibility area as target channels based on the evaluation results. The multi-source data includes topographic factors, ecological environment factors, geological structure factors and human activity factors. Step 3: Precise delineation and mechanism analysis of the source area: For the target gullies selected in Step 2, field investigation and sampling are carried out, and a physical model test of rainfall erosion of the loess fill slope is constructed. Then, the erosion and damage mechanism of the loess fill slope is analyzed to accurately delineate the debris flow source area. Step 4: Numerical simulation and hazard assessment of debris flow in a single gully: Based on the debris flow source area delineated in Step 3, MassFlow software with parameters calibrated by inversion from historical debris flow event investigation data is used to numerically simulate the debris flow movement process. Based on the numerical simulation results, hazard zoning is performed, and the debris flow hazard assessment results of the target gully are output.

2. The method for assessing the hazard of debris flows in small watersheds of loess gullies according to claim 1, characterized in that: In step three, the specific method for constructing a physical model test of rainfall erosion on a loess-filled slope is as follows: an indoor rainfall simulation test device is constructed, which includes a model box, artificial rainfall equipment, and volumetric water content sensors, matrix suction sensors, and pore water pressure sensors installed in the slope model. Then, the erosion and damage process of the loess-filled slope under different rainfall intensities is simulated through the indoor rainfall simulation test device. The changes in volumetric water content, matrix suction, and pore water pressure of the slope are recorded, and the erosion and damage process is divided into splash erosion stage, gully erosion stage, and gully erosion and collapse damage stage.

3. The method for assessing the hazard of debris flows in small watersheds of loess gullies according to claim 1, characterized in that: In step four, the MassFlow software, which uses historical debris flow event survey data to invert and calibrate parameters, specifically includes: establishing a digital elevation model of the target channel using UAV aerial surveys, and then, based on historical debris flow event survey data, calculating and calibrating the friction coefficient and turbulence coefficient of the Voellmy model in the MassFlow software through inversion.

4. The method for assessing the hazard of debris flows in small watersheds of loess gullies according to claim 1, characterized in that: Step four, which involves hazard zoning based on numerical simulation results, specifically includes: Based on the mud depth and flow velocity data output by numerical simulation, momentum parameters are calculated, where the momentum parameter is the product of mud depth and flow velocity. Then, according to the momentum parameter and mud depth data, the debris flow intensity is divided into high intensity, medium intensity and low intensity levels. Finally, combined with the preset debris flow occurrence frequency, a hazard zoning map of the target channel is generated.

5. The method for assessing the hazard of debris flows in small watersheds of loess gullies according to claim 3, characterized in that: The friction coefficient of the Voellmy model is 0.1 to 0.3, and the turbulence coefficient of the Voellmy model is 100 to 300.

6. The method for assessing the hazard of debris flows in small watersheds of loess gullies according to claim 5, characterized in that: The friction coefficient of the Voellmy model is 0.15, and the turbulence coefficient of the Voellmy model is 200.

7. A debris flow hazard assessment system for small watersheds in loess gully areas, characterized in that: Includes the following modules: The susceptibility screening module is used to load the random forest model and evaluate the susceptibility of debris flows in the study area, and then screen out typical debris flow channels in the extremely high susceptibility area as target channels. The source analysis module is used to conduct rainfall erosion tests on the loess slopes in the target gully, analyze the erosion and damage mechanism of the loess slopes, and delineate the debris flow source area. The numerical simulation module is used to load MassFlow software, which uses historical debris flow event investigation data to retrieve and calibrate parameters, and to perform numerical simulations of the debris flow movement process based on the debris flow source area. The hazard assessment module receives mud depth and flow velocity data output from the numerical simulation module, and generates a hazard zoning map based on the mud depth and flow velocity data, thereby outputting the debris flow hazard assessment results for the target channel.

8. The debris flow hazard assessment system for small watersheds in loess gullies according to claim 7, characterized in that: The source analysis module includes a rainfall simulation test unit and a sensor monitoring unit. The rainfall simulation test unit is used to simulate the slope erosion process under different rainfall intensities. The sensor monitoring unit includes a volumetric water content sensor, a matrix suction sensor, and a pore water pressure sensor, which are used to monitor the hydraulic parameters of the slope in real time during the rainfall erosion process.

Images