Prediction and early warning method for collapse river blocking and dam forming comprehensively induced by typhoon and seawater backflow

Abstract

The invention discloses a method for predicting and early warning collapse river blocking and dam forming comprehensively induced by typhoon and seawater backflow, and belongs to the technical field of disaster prediction and early warning. The method comprises the following specific steps: acquiring spatial information and attribute information of sea-entering river slopes, constructing a collapse potential degree calculation data set, determining a calculation model, and calculating the collapse potential degree of each slope; the typhoon and seawater backflow comprehensive induced collapse occurrence probability is calculated through the typhoon rainfall influence coefficient, the seawater backflow submerging enhancement coefficient and the crack current situation enhancement coefficient, and then the typhoon and seawater backflow comprehensive induced collapse river blocking and dam forming occurrence probability of each side slope is calculated; and performing graded early warning according to the occurrence probability of collapse, river blockage and dam formation comprehensively induced by typhoon and seawater backflow. According to the method, the particularity of collapse comprehensively induced by typhoon and seawater backflow is considered, the optimal collapse potential degree calculation model is selected, various factors of the disaster-pregnant environment are comprehensively considered, and the accuracy of river blocking and dam forming prediction and early warning of collapse comprehensively induced by typhoon and seawater backflow is improved.

Description

Technical Field

[0001] This invention relates to the field of disaster prediction and early warning technology, and in particular to a method for predicting and early warning of river collapse and dam formation induced by the combined effects of typhoons and seawater intrusion. Background Technology

[0002] Typhoons are intense tropical cyclones, typically accompanied by strong winds, torrential rains, and storm surges, which can cause severe damage to affected areas. The damage caused by the typhoon disaster chain far exceeds the damage caused by the typhoon itself. Typhoons cause seawater intrusion into rivers flowing into the sea. Combined with rainfall infiltration and existing cracks, this leads to a rise in groundwater levels on riverbank slopes, a decrease in the mechanical strength and deformation properties of the soil and rock, and an increased probability of geological disasters such as landslides. Landslides can then impound water into rivers, forming dams that block the flow and, if they break, can trigger downstream flooding. Existing landslide prediction and early warning methods have the following problems: (1) The disaster-prone environment of landslides induced by typhoons and seawater intrusion is different from that of ordinary landslides. Existing methods do not consider this particularity when selecting potential risk factors. Furthermore, existing methods have subjectivity and randomness in model selection, resulting in large errors in potential calculation results. (2) Existing methods only consider the influence of typhoon rainfall infiltration on the mechanical strength and deformation properties of slope rock and soil when analyzing the induction mechanism of typhoons. However, seawater intrusion caused by typhoon rainfall may submerge the landslide body and reduce the stability of the slope. Existing methods ignore the combined effect of typhoon rainfall and seawater intrusion. (3) Typhoons often cause continuous rainfall, and in the north, they can even cause continuous rain. Existing methods generally use process rainfall or landfall day rainfall to characterize the typhoon rainfall level, which exaggerates the role of typhoon rainfall and ignores the influence of underlying surface factors, reducing the success rate of disaster chain prediction and the efficiency of comprehensive prevention and control measures. (4) Since the development process of the collapse induced by the combined effects of typhoon and seawater intrusion is continuous and lengthy, the initial damage form is the initiation and expansion of back wall cracks and side wall cracks. However, existing methods focus on potential and typhoon rainfall, neglecting the evolution process of existing cracks and the impact of the complex disaster-prone environment that leads to crack development on the occurrence of the collapse. (5) The movement of the landslide body is affected by factors such as the type and intensity of inducing factors, the properties of the rock and soil, topography, surface roughness, and movement distance. The volume of the landslide body that may be washed into the river channel has a high degree of uncertainty. Existing methods treat it as a fixed value, which does not conform to the complex feedback mechanism between landslide blocking the river to form a dam and the disaster-prone environment. Summary of the Invention

[0003] To address the aforementioned technical problems, the purpose of this invention is to provide a method for predicting and warning of dam formation and river blockage induced by the combined effects of typhoons and seawater intrusion. The specific steps are as follows: To achieve the above objectives, this invention provides a method for predicting and warning of dam formation and river blockage induced by the combined effects of typhoons and seawater intrusion. The specific steps are as follows: Step S1: Obtain spatial and attribute information of river slopes flowing into the sea, construct a data set for calculating landslide potential, determine landslide potential hazard factors, and scale them; Step S2: Determine the collapse potential calculation model and calculate the collapse potential of each slope; Step S3: Calculate the typhoon rainfall impact coefficient, seawater intrusion enhancement coefficient, and existing crack enhancement coefficient for each slope collapse; Step S4: Calculate the probability of a landslide caused by the combined effects of typhoon and seawater intrusion. The calculation formula is as follows: ; in: The probability of a landslide caused by a combination of typhoon and seawater intrusion. This represents the slope collapse potential. The typhoon rainfall impact coefficient. The inundation enhancement coefficient is the intrusion intrusion factor. The current crack enhancement factor; Step S5: Assume the volume of the landslide that can flow into the river channel follows a sin i · k 4. V Using a Laplace distribution with mean of 2 and variance of 2, calculate the probability of typhoon and seawater intrusion combined inducing landslides and damming of rivers on each slope: ; in: The probability of a river collapse and dam formation caused by the combined effects of typhoons and seawater intrusion. This represents the cross-sectional area of ​​the riverbed at the slope. The length of the dam body. The collapse body displacement loss index, For the potential collapse volume, The angle between the direction of the landslide's movement and the river; Step S6: Conduct graded early warning based on the probability of typhoons and seawater backflow causing river collapses and dam formations.

[0004] Preferably, step S1 is as follows: Step S11: Obtain the number of historical disaster sites on the riverbank slopes of rivers flowing into the sea. The survey also included spatial information on the corresponding disaster sites. The survey methods included literature review, field reconnaissance, and remote sensing interpretation. The literature review covered local statistical yearbooks, news reports, and publicly published papers. The field reconnaissance used total station and UAV oblique photogrammetry. The remote sensing interpretation used object-oriented classification and SBAS-InSAR technology.

[0005] Step S12: Obtain the number of slopes with collapse risk along the riverbanks of rivers flowing into the sea. And the spatial information of the corresponding slope, and obtain the riverbed width at the corresponding slope. Riverbed depth The angle between the direction of the landslide's movement and the river. Vertical height from the normal water level to the bottom of the collapsed body and the vertical height of the collapsed body ; Calculate the cross-sectional area of ​​the riverbed at each slope. The calculation formula is as follows: ; Calculate the volume of potential collapse body The specific method is as follows: Existing cracks on each slope are identified through on-site surveys; continuous crack surfaces are drawn; and the longest distance from the rear crack to the front free surface on the drawn continuous crack surface is estimated. The longest distance between the two cracks on the depicted surface of the continuous crack. The length of the crack that has appeared on the depicted continuous crack surface from the trailing edge crack to the leading edge free surface. The length of the cracks that have appeared on the surface of the continuous crack on both sides. Through estimation , , as well as Calculate the volume of the soil and rock mass between the continuous crack surface and the free surface; the estimated volume of the soil and rock mass is the potential volume of the collapse mass. ; Step S13: From Randomly selected slopes without signs of disaster Located on the slope, and The landslide potential calculation dataset is composed of historical disaster sites. One sample; Step S14: Determine the initial selection factors for collapse potential, and eliminate their correlation and multicollinearity based on the Pearson correlation coefficient method and the variance inflation coefficient method, respectively. Factors that pass the correlation and multicollinearity tests... d Each initial selection factor is defined as a risk factor, and the risk factor values ​​for each sample are determined and scaled.

[0006] Preferably, step S14 specifically includes: The initial selection factors for landslide potential include elevation, slope, aspect, joint density, slope type, slope structure, plan curvature, profile curvature, SPI (water flow power index), STI (topographic moisture index), lithology, river distance, road distance, NDVI (homogeneous vegetation cover index), fault distance, and TWI (surface runoff accumulation). The formula for calculating the Pearson correlation coefficient is as follows: ; in: Preliminary selection factors and initial selection factors The Pearson correlation coefficient, and These are the initial selection factors. and initial selection factors The mean, and These are the initial selection factors. and initial selection factors variance Preliminary selection factors and initial selection factors covariance, E To be expected, if a preliminary factor has a Pearson correlation coefficient greater than 0.4 or not greater than -0.4 with at least two other preliminary factors, then that preliminary factor is deleted. The formula for calculating the variance inflation coefficient is as follows: ; in: Preliminary selection factors The variance inflation coefficient, Preliminary selection factors For the remaining initial factors, perform regression analysis on the multiple correlation coefficients. If the variance inflation coefficient of a certain initial factor is greater than 10, then delete that initial factor. The risk factors for each sample are scaled using a vector method. The scaling vector includes... d +1 element, before d The first element is the input item, which represents the values ​​of each risk factor in sequence. d +1 elements are output items, where, The value for historical disaster points is taken as 1. The value for slopes showing no signs of disaster is 0.

[0007] Preferably, step S2 is as follows: Step S21: Each sample is divided into a training set and a validation set according to a set ratio. Multiple mathematical models are trained and validated using the training set and the validation set, and the goodness index and accuracy of the trained mathematical models are calculated. The goodness index is calculated as follows: ; in: For the first The goodness index of a mathematical model. and The first The overall positive rate and accuracy of the mathematical models; The formulas for calculating the overall positive rate and accuracy are as follows: ; ; in: This represents the number of samples that have experienced collapse and are predicted to collapse. This represents the number of samples that did not collapse but were predicted to collapse. The number of samples that have collapsed but are predicted not to collapse; Step S22: Determine the mathematical model with the largest goodness index as the best mathematical model. If at least two mathematical models have the same and largest goodness index, calculate the AUC value of the above at least two mathematical models and select the one with the largest AUC as the best mathematical model. If at least two mathematical models still have the same and largest AUC, select the one with the shortest ROC curve plotting time from the above at least two mathematical models as the best mathematical model. Step S23: Calculate using the selected optimal mathematical model Collapse potential of the slope .

[0008] Preferably, the mathematical models include the analytic hierarchy process (AHP) model, the deterministic coefficient model, the maximum-minimum hill-climbing model, the interpretable neural network model, the C5.0 decision tree model, the ID4.5 decision tree model, the maximum entropy model, the feedforward neural network model, the recurrent neural network model, the information content-random forest model, the AHP-information content model, the particle swarm optimization-support vector machine model, and the deterministic coefficient-random forest model.

[0009] Preferably, in step S3, the formula for calculating the typhoon rainfall impact coefficient is as follows: ; in: The formula for predicting daily available rainfall is as follows: ; in: This represents the typhoon's rainfall on that day. For the front The amount of rainfall during a typhoon is determined based on weather forecasts or actual rainfall measurements. The rainfall threshold for landslide initiation is calculated as follows: [The text abruptly ends here, likely due to an incomplete sentence or a formatting error.] Number of rain stations within the designated area of ​​the slope ,calculate Available rainfall at various rain gauge stations on the days when historical disasters occurred on the slope. The minimum available rainfall was determined using the Glikin method. Interpolation is performed to obtain The threshold for triggering rainfall is determined by the slope collapse.

[0010] Preferably, in step S3, the seawater intrusion enhancement coefficient during the collapse is... The calculation formula is as follows: ; in: The adjustment coefficient for river segment type is 1.25 for canyon meandering river segments, 1.00 for canyon straight river segments and open meandering river segments, and 0.75 for open straight river segments. The classification of river segment type shall be carried out in accordance with the relevant provisions of the "Specifications for Hydrological Survey and Design of Highway Engineering" (JTGC30-2015) and the "Specifications for Survey and Design of Highway Bridge Sites" (JTJ 062-91). This represents the estimated rise in water level at each slope due to seawater intrusion on the predicted date.

[0011] Preferably, in step S3, the current crack enhancement coefficient during the collapse is... The calculation formula is as follows: ; in: The value is the adjustment factor for the size of the landslide body, which is related to the volume of the potential landslide body. The method for determining the value is shown in Table 1.

[0012] Table 1. Method for determining the adjustment coefficient for the size of the landslide body ;

[0013] The length of the crack that has appeared on the depicted surface of the through crack, from the trailing edge crack to the leading edge free surface. The length of the cracks that have appeared on the surface of the continuous crack on both sides. This is the longest distance from the trailing edge crack to the leading edge free surface on the depicted through crack surface. This represents the longest distance between the two cracks on the depicted surface of the continuous crack.

[0014] Based on the shortest distance from the leading edge of the landslide to the riverbank and ground slope Sure The determination method is as follows: When 0≤ αWhen <35°: ; When 35°≤ α When <55°: ; When 55°≤ α When <75°: ; When 75°≤ α When ≤90°: .

[0015] Preferably, in step S5, historical data on dam bodies and rivers during dam-building disasters are statistically analyzed. The dam body data includes the dam body length. River data includes flow velocity Riverbed width Riverbed depth and cross-sectional area The river data was then normalized to obtain... , , as well as The normalization method is as follows: ; ; ; ; in: and These are the minimum and maximum values ​​of the flow velocity, respectively. and These are the minimum and maximum values ​​of the riverbed width, respectively; and These are the minimum and maximum values ​​of the riverbed depth, respectively. and These are the minimum and maximum values ​​of the riverbed cross-sectional area, respectively. The functional relationship between the dam data and the normalized river data is then determined, and the corresponding formula is as follows: ; in, , , , , , , , , , , , , , as well as All are regression coefficients; Calculate using the solved functional relationship The formula for calculating the volume of landslide required for damming a river on a slope is as follows: ; in: The volume of landslide material required to build a dam to block the river. The cross-sectional area of ​​the riverbed. This refers to the length of the dam body.

[0016] Preferred, according to median The warning levels, warning strategies, and warning measures are determined, as shown in Table 2.

[0017] Table 2 Relationship with warning levels, warning strategies and warning measures ;

[0018] Therefore, the present invention employs the above-mentioned method for predicting and warning of typhoon-induced collapse and dam formation caused by combined typhoon and seawater intrusion, which has the following beneficial effects: (1) When selecting the potential risk factors, the special characteristics of landslides induced by the combined effects of typhoons and seawater intrusion were considered. The correlation and multicollinearity were eliminated based on the Pearson correlation coefficient method and the variance expansion coefficient method, respectively. This not only reflects the original information of the disaster-prone environment, but also reduces the computational complexity. At the same time, this invention uses no less than 13 mathematical models to calculate the landslide potential and selects the model with the best effect based on multiple indicators, which greatly improves the accuracy of the potential calculation results.

[0019] (2) When calculating the probability of landslides caused by the combined effects of typhoons and seawater intrusion, the landslide potential, typhoon rainfall influence coefficient, seawater intrusion inundation enhancement coefficient, and crack current enhancement coefficient were introduced. This fully considers various inducing factors of the disaster-prone environment and is more in line with the disaster-causing mechanism and evolution mechanism of the "typhoon → landslide → damming" disaster chain.

[0020] (3) Due to the influence of underlying surface factors, landslides lag behind the rainfall process, and the rainfall during the process and the rainfall on the day of landfall cannot reflect the infiltration mechanism of typhoon rainfall. This invention can accurately reflect the inducing effect of typhoon rainfall on landslides by using rainfall to characterize the temporal characteristics of typhoon rainfall.

[0021] (4) By introducing the current crack enhancement coefficient, this invention improves the accuracy of predicting the collapse induced by the combined effects of typhoon and seawater intrusion, and more accurately describes the influence of underlying surface factors on the occurrence of collapse.

[0022] (5) To accurately characterize the uncertainty of the volume of the landslide body rushing into the river channel, this invention uses a probabilistic method to reveal the "node" information of the disaster chain of "typhoon → landslide → damming", and sets it to obey a sin i · k 4. V The Laplace distribution with mean of 2 and variance of 2 overcomes the inability to describe parameter discreteness when using deterministic methods to reveal the evolution mechanism of disaster chains.

[0023] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0024] Figure 1 This is a flowchart of a method for predicting and warning of typhoon-induced collapse and dam formation caused by seawater backflow, according to the present invention. Detailed Implementation

[0025] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Example This example focuses on the Futuan River, which is 71.8 km long and has a drainage area of ​​1040 km². 2 The average annual precipitation is 923.5 mm, the annual runoff depth is 347.8 mm, and the annual runoff volume is 362 million m³. 3 River network density: 0.37 km / km 2 The river mouth is located southeast of Jiacang Fourth Village, Kuishan Subdistrict, Rizhao Economic Development Zone, Shandong Province. It is 800 meters wide. The Futuan River basin experiences an average of 3.7 typhoons annually, each accompanied by seawater intrusion. The terrain along the river is rugged and complex, with well-developed geological hazards such as landslides, providing the natural conditions for the formation and evolution of a disaster chain: "typhoon → landslide → damming."

[0027] like Figure 1 As shown, step S1: Obtain spatial and attribute information of river slopes flowing into the sea; construct a data set for calculating landslide potential, determine landslide potential hazard factors, and scale them.

[0028] Futuan River and its surrounding slopes: Number of historical disaster sites =206; Latitude and longitude between: 35°16'52.07"N~35°43'09.51"N; 119°18'03.24"E~119°28'35.10"E.

[0029] Literature review: Relevant data were downloaded from the China Statistical Database. In this example, the downloaded data was the Rizhao Statistical Yearbook. In addition, 17 news reports and 25 academic papers were reviewed. Site survey: The total station used was a Topcon GTS2002, and the drone used was a DJI Mavic 3; Remote sensing interpretation: Landsat 8 satellite imagery was analyzed using an object-oriented classification method to identify disaster-prone areas; Sentinel-1 satellite imagery was processed using SBAS-InSAR technology to obtain surface subsidence data for these areas.

[0030] Number of slopes at risk of collapse =533; Latitude and longitude between: 35°16'30.49"N~35°43'46.18"N; 119°17'57.64"E~119°28'01.74"E; Riverbed width The maximum and minimum values ​​are as follows: B min =83.15m, B max =746.98m; Riverbed depth The maximum and minimum values ​​are as follows: h min =5.23m, h max =17.63m; The angle between the direction of the landslide and the river The maximum and minimum values ​​are as follows: i min = 0, i max =90°; Vertical height from the normal water level to the bottom of the collapsed body The maximum and minimum values ​​are as follows: h 1,min =0.95m, h 1,max =7.10m; Vertical height of the collapsed body The maximum and minimum values ​​are as follows: h 2,min =1.35m,h 2,max =27.95m; Riverbed cross-sectional area S The maximum and minimum values ​​are as follows: S min =684m 2 , S max =9467m 2 ; Potential collapse volume The maximum and minimum values ​​are as follows: V min =831m 3 , V max =135246m 3 ; The maximum and minimum values ​​of the crack-related dimensions are as follows: l h,min =2.35m ，l h,max =97.40m, l c,min =1.80m, l c,max =126.05m, l x,h,min =0, l x,h,max =38.15m, l x,c,min =0, l x,c,max =45.20m.

[0031] From 533 slopes, 206 slopes without signs of disaster were randomly selected and combined with 206 historical disaster sites to form a landslide potential calculation dataset, including 412 samples. The Pearson correlation coefficients between slope type and profile curvature and road distance were 0.427 and -0.509, respectively. The Pearson correlation coefficients of any two other initially selected factors were both > -0.4 and < 0.4. (TWI) VIF It is 11.024.

[0032] Therefore, after removing the slope type and TWI, a total of d =14 risk factors, as shown in Table 3.

[0033] Table 3. 14 risk factors obtained ;

[0034] The hazard factors of 412 samples were scaled using a vector method: the scaling vector of each sample consisted of 15 elements, with the first 14 elements being the input items, which were the values ​​of each hazard factor, and the 15th element being the output item. Among them, the value of 206 historical disaster points was 1, and the value of 206 slopes without disaster signs was 0. The hazard factor scaling method is shown in Table 4.

[0035] Table 4 Hazard Factor Scale ;

[0036] Step S2: Determine the collapse potential calculation model and calculate the collapse potential of each slope. Divide the 412 samples into an average of [number missing]. m =4 sets, each set containing 103 samples. 3 sets (309 samples) were randomly selected as training samples and 1 set (103 samples) as validation samples. The 13 selected mathematical models were trained based on the training samples, and the goodness index of each model to the validation samples was calculated. The calculation results are shown in Table 5.

[0037] Table 5. Goodness-of-performance indices of each model ;

[0038] The interpretable neural network (DNN) model and the deterministic coefficient-random forest (CF-RF) model have the largest and the same goodness index, both at 0.942. The output values ​​of both models are: TP =48、 FP =4、 TN =49、 FN =2、 TPR =0.960, Precision =0.923. Further calculation of the AUC for the two models shows that the AUC of the interpretable neural network model is 0.912, and the AUC of the coefficient of determination-random forest model is 0.915. Therefore, the coefficient of determination-random forest model is selected as the optimal model. This model was used to calculate the collapse potential of 533 slope locations. The maximum and minimum values ​​of the calculated results are: k 0,min =0.021, k 0,max =0.967.

[0039] Step S3: Calculate the typhoon rainfall impact coefficient, seawater intrusion enhancement coefficient, and existing crack enhancement coefficient for each slope collapse.

[0040] Number of rain gauges within 30km of 533 slopes =8. Investigate the available rainfall at each rain gauge station on the day of 206 historical disasters. Q k And statistics Qk,min The results showed: Jiacang River Estuary Monitoring Station Q k,min = 182.64mm, Xiheya Monitoring Station Q k,min =173.24mm, Gaojiacun Monitoring Station Q k,min =175.18mm, Xiaoguzhen West Monitoring Station Q k,min =186.27mm, Caijiatan Tidal Flat Monitoring Station Q k,min =199.52mm, Platform Ridge Monitoring Station Q k,min =196.25mm, Feijiacun Monitoring Station Q k,min =185.28mm, Dongsun Village Monitoring Station Q k,min =186.37mm.

[0041] The Glikin interpolation method based on the spatial analysis function of ArcGIS 10.2 was used to analyze data from eight rain gauge stations. Q k,min Interpolation yielded the rainfall threshold for triggering slope collapse at 533 locations. Q y The results showed: Q y,min =170.13mm, Q y,max =205.08mm.

[0042] Taking Typhoon Gemi, the third typhoon of 2024, which affected Rizhao City, Shandong Province, as an example, the typhoon rainfall impact coefficient of the landslide was calculated. k 1. The results show: k 1,min =0.165, k 1,max =0.581.

[0043] Based on engineering experience and weather forecasts, the estimated water level rise Δ at each slope caused by seawater intrusion is calculated. h Taking Typhoon "Gemi" as an example, the Δ value of 533 slopes h min =0, Δ h max =0.65m. Field surveys determined the river section type at each slope, showing: 63 meandering river sections in canyons, 96 straight river sections in canyons, 143 open meandering river sections, and 231 open straight river sections. Taking Typhoon "Kammuri" as an example, the seawater intrusion enhancement coefficient for each slope collapse was calculated, and the results show:k 2,min =0, k 2,max =0.119.

[0044] Of the 533 slopes, V ≤1000m 3 291 locations, 1000m 3 < V ≤10000m 3 99 locations, 10000m 3 < V ≤100000m 3 82 locations, V >100000m 3 61 locations, or 291 slope locations g =0.4, 99 slopes g =0.6, 82 slopes g =0.8, 61 slopes g =1.0. Calculate the current crack enhancement factor for each slope collapse. k 3. The results show: k 3,min =0, k 3,max =0.635.

[0045] Step S4: Calculate the probability of a landslide caused by the combined effects of typhoon and seawater intrusion.

[0046] The probability of landslides caused by the combined effects of typhoons and seawater intrusion was calculated for each slope. The results show that: p l,min =0, p l,max =0.634.

[0047] Step S5: Calculate the probability of typhoon and seawater backflow induced by the combined collapse of each slope to block the river and form a dam.

[0048] This study analyzed historical data on dam bodies and rivers during dam-building disasters, identifying 249 such disasters. , , , as well as The data was then normalized to obtain... , , as well as Calculated and , , as well as The functional relationship is as follows: .

[0049] Calculate the dam length at 533 slopes. and the volume of debris flow required to form a dam. V g The results showed: L b,min =3.94m, L b,max =15.27m, V g,min =2987m 3 , V g,max =116444m 3 .

[0050] Determine the collapse mass transport loss index for 533 slope locations. k 4. The results show that: 147 slopes k 4=1.0, slope at point 102 k 4 = 0.9, slope at point 75 k 4 = 0.8, 64 slopes k 4 = 0.7, 52 slopes k 4 = 0.6, 48 slopes k 4 = 0.5, 33 slopes k 4 = 0.4, 12 slopes k 4 = 0.3.

[0051] The average volume of landslides that could be washed into the riverbed at 533 slope locations was calculated, and the results showed that the minimum value was 1107 m³. 3 The maximum value is 135724m 3 Taking Typhoon "Gemi" as an example, the probability of landslides and dam formation caused by the combined effects of typhoon and seawater intrusion on various slopes was calculated. The results show that: p cb,min =0, p cb,max =0.602.

[0052] Step S6: Conduct graded early warning based on the probability of typhoons and seawater backflow causing river collapses and dam formations.

[0053] Due to 533 slopes p cb median p cb,m =0.382, therefore a yellow alert should be issued for the combined effects of typhoon and seawater intrusion inducing river collapse and dam formation. The alert strategy and measures are shown in Table 6.

[0054] Table 6. Warning Levels, Warning Strategies, and Warning Measures in this Embodiment ;

[0055] Based on the above methods and ArcGIS Engine, a predictive and early warning system for typhoon-induced riverbank collapses and damming along rivers flowing into the sea has been developed. This system enables real-time map lookup of early warning levels, strategies, and measures. It also allows querying of all spatial and attribute information for the Futuan River, 206 historical disaster sites, 533 slopes, and 8 rain gauge stations. Spatial information includes the river's orientation and the latitude and longitude of historical disaster sites, slopes, and rain gauge stations. Attribute information includes river segment type, etc. n 1. n 2. B , h , i , h 1. h 2. S , V , l h , l c , l x,h , l x,c , r i,j , u i , u j , s i , s j 、cov( i , j ), VIF i , d , m , TP , FP , FN , TN , k 0、 Y , TPR Precision, AUC FPR , n 3. Q k , Q k,min , Q y , Q y,k , k 1. Q 0、 Qi 、D h 、 k 2、 or 、 k 3、 p l 、 k 3、 g 、 L b 、 v b 、 B y 、 h y 、 S y 、 v b,1 、 B y,1 、 h y,1 、 S y,1 、 V g 、 v b,min 、 B y,min 、 h y,min 、 S y,min 、 v b,max 、 B y,max 、 h y,max 、 S y,max 、 t 0、 t 1、 t 2、 t 3、 t 4、 t 5、 t 6、 t 7、 t 8、 t 9、 t 10 、 t 11 、 t 12 、 t 13 、 t 14 、 L q 、 α 、 k 4、 p cb 、 pcb,m .

[0056] Therefore, this invention employs the aforementioned method for predicting and warning of landslides and dam formations caused by a combination of typhoons and seawater intrusion. The warning levels are divided into four categories: blue, yellow, orange, and red. This method reveals the feedback mechanism between landslide occurrence and disaster-prone environmental characteristics such as potential, typhoon rainfall, seawater intrusion, and crack development. It is integrated forward with disaster risk assessment and monitoring, and backward with engineering management decisions, significantly reducing casualties and economic losses caused by the "typhoon → landslide → dam formation" disaster chain, and contributing to the formation of a comprehensive disaster risk prevention and control system.

[0057] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for predicting and warning of damming and river collapses induced by a combination of typhoons and seawater intrusion, characterized in that, The specific steps are as follows: Step S1: Obtain spatial and attribute information of river slopes flowing into the sea, construct a data set for calculating landslide potential, determine landslide potential hazard factors, and scale them; Step S2: Determine the collapse potential calculation model and calculate the collapse potential of each slope; Step S3: Calculate the typhoon rainfall impact coefficient, seawater intrusion enhancement coefficient, and existing crack enhancement coefficient for each slope collapse; Step S4: Calculate the probability of a landslide caused by the combined effects of typhoon and seawater intrusion. The calculation formula is as follows: ； in: The probability of a landslide caused by a combination of typhoon and seawater intrusion. This represents the slope collapse potential. The typhoon rainfall impact coefficient. The inundation enhancement coefficient is the intrusion intrusion factor. The current crack enhancement factor; Step S5: Assume the volume of the landslide that can flow into the river channel follows a sin θ · k 4. V Using a Laplace distribution with mean of 2 and variance of 2, calculate the probability of typhoon and seawater intrusion combined inducing landslides and damming of rivers on each slope: ； in: The probability of a river collapse and dam formation caused by the combined effects of typhoons and seawater intrusion. This represents the cross-sectional area of ​​the riverbed at the slope. The length of the dam body. The collapse body displacement loss index, For the potential collapse volume, The angle between the direction of the landslide's movement and the river; Step S6: Conduct graded early warning based on the probability of typhoons and seawater backflow causing river collapses and dam formations.

2. The method for predicting and warning of typhoon-induced river collapse and dam formation caused by combined typhoon and seawater intrusion as described in claim 1, characterized in that, Step S1 is as follows: Step S11: Obtain the number of historical disaster points on the riverbank slopes along rivers flowing into the sea. And spatial information of the corresponding disaster sites; Step S12: Obtain the number of slopes with collapse risk along the riverbanks of rivers flowing into the sea. And the spatial information of the corresponding slope, and obtain the riverbed width at the corresponding slope. Riverbed depth The angle between the direction of the landslide's movement and the river. Vertical height from the normal water level to the bottom of the collapsed body and the vertical height of the collapsed body ; Calculate the cross-sectional area of ​​the riverbed at each slope. The calculation formula is as follows: ； Calculate the volume of potential collapse body The specific method is as follows: Existing cracks on each slope are identified through on-site surveys; continuous crack surfaces are drawn; and the longest distance from the rear crack to the front free surface on the drawn continuous crack surface is estimated. The longest distance between the two cracks on the depicted surface of the continuous crack. The length of the crack that has appeared on the depicted continuous crack surface from the trailing edge crack to the leading edge free surface. The length of the cracks that have appeared on the surface of the continuous crack on both sides. Through estimation , , as well as Calculate the volume of the soil and rock mass between the continuous crack surface and the free surface; the estimated volume of the soil and rock mass is the potential volume of the collapse mass. ; Step S13: From Randomly selected slopes without signs of disaster Located on the slope, and The landslide potential calculation dataset is composed of historical disaster sites. One sample; Step S14: Determine the initial selection factors for collapse potential, and eliminate their correlation and multicollinearity based on the Pearson correlation coefficient method and the variance inflation coefficient method, respectively. Factors that pass the correlation and multicollinearity tests... d Each initial selection factor is defined as a risk factor, and the risk factor values ​​for each sample are determined and scaled.

3. The method for predicting and warning of typhoon-induced landslides and dam formations caused by combined typhoon and seawater intrusion, as described in claim 2, is characterized in that... Step S14 is as follows: The initial selection factors for landslide potential include elevation, slope, aspect, joint density, slope type, slope structure, plan curvature, profile curvature, water flow power index, topographic humidity index, lithology, river distance, road distance, homogenized vegetation cover index, fault distance, and surface runoff accumulation. The formula for calculating the Pearson correlation coefficient is as follows: ； in: Preliminary selection factors and initial selection factors The Pearson correlation coefficient, and These are the initial selection factors. and initial selection factors The mean, and These are the initial selection factors. and initial selection factors variance Preliminary selection factors and initial selection factors covariance, E As a result, if the Pearson correlation coefficient between a preliminary factor and at least two other preliminary factors is greater than 0.4 or less than -0.4, then that preliminary factor is deleted. The formula for calculating the variance inflation coefficient is as follows: ； in: Preliminary selection factors The variance inflation coefficient, Preliminary selection factors For the remaining initial factors, perform regression analysis on the multiple correlation coefficients. If the variance inflation coefficient of the initial factors is greater than 10, then delete the initial factors. The risk factors for each sample are scaled using a vector method. The scaling vector includes... d +1 element, before d The first element is the input item, which represents the values ​​of each risk factor in sequence. d +1 elements are output items, where, The value for historical disaster points is taken as 1. The value for slopes showing no signs of disaster is 0.

4. The method for predicting and warning of typhoon-induced river collapse and dam formation caused by combined typhoon and seawater intrusion as described in claim 2, characterized in that, Step S2 is specifically as follows: Step S21: Each sample is divided into a training set and a validation set according to a set ratio. Multiple mathematical models are trained and validated using the training set and the validation set, and the goodness index and accuracy of the trained mathematical models are calculated. The goodness index is calculated as follows: ； in: For the first The goodness index of a mathematical model. and The first The overall positive rate and accuracy of the mathematical models; The formulas for calculating the overall positive rate and accuracy are as follows: ； ； in: This represents the number of samples that have experienced collapse and are predicted to collapse. This represents the number of samples that did not collapse but were predicted to collapse. The number of samples that have collapsed but are predicted not to collapse; Step S22: Determine the mathematical model with the largest goodness index as the best mathematical model. If at least two mathematical models have the same and largest goodness index, calculate the AUC value of the above at least two mathematical models and select the one with the largest AUC as the best mathematical model. If at least two mathematical models still have the same and largest AUC, select the one with the shortest ROC curve plotting time from the above at least two mathematical models as the best mathematical model. Step S23: Calculate using the selected optimal mathematical model Collapse potential of the slope .

5. The method for predicting and warning of typhoon-induced river collapse and dam formation caused by combined typhoon and seawater intrusion as described in claim 4, characterized in that, Mathematical models include the analytic hierarchy process (AHP), the deterministic coefficient model, the maximum-minimum hill-climbing model, the interpretable neural network model, the C5.0 decision tree model, the ID4.5 decision tree model, the maximum entropy model, the feedforward neural network model, the recurrent neural network model, the information-random forest model, the AHP-information model, the particle swarm optimization-support vector machine model, and the deterministic coefficient-random forest model.

6. The method for predicting and warning of typhoon-induced river collapse and dam formation caused by combined typhoon and seawater intrusion as described in claim 1, characterized in that, In step S3, the formula for calculating the typhoon rainfall impact coefficient is as follows: ； in: The formula for predicting daily available rainfall is as follows: ； in: This represents the typhoon's rainfall on that day. For the front The amount of rainfall during a typhoon is determined based on weather forecasts or actual rainfall measurements. The rainfall threshold for landslide initiation is calculated as follows: [The text abruptly ends here, likely due to an incomplete sentence or a formatting error.] Number of rain stations within the designated area of ​​the slope ,calculate Available rainfall at various rain gauge stations on the days when historical disasters occurred on the slope. The minimum available rainfall was determined using the Glikin method. Interpolation is performed to obtain The threshold for triggering rainfall is determined by the slope collapse.

7. The method for predicting and warning of typhoon-induced river collapse and dam formation caused by combined typhoon and seawater intrusion as described in claim 1, characterized in that, In step S3, the seawater intrusion enhancement factor during the collapse is... The calculation formula is as follows: ； in: For river section type adjustment coefficient, This represents the estimated rise in water level at each slope due to seawater intrusion on the predicted date.

8. The method for predicting and warning of typhoon-induced landslides and dam formations caused by combined typhoon and seawater intrusion, as described in claim 1, is characterized in that: In step S3, the current crack enhancement coefficient during the collapse is... The calculation formula is as follows: ； in: This is a collapse mass adjustment factor, which is related to the potential collapse mass volume; This is the longest distance from the trailing edge crack to the leading edge free surface on the depicted through crack surface. This represents the longest distance between the two cracks on the depicted surface of the continuous crack; The length of the crack that has appeared on the depicted surface of the through crack, from the trailing edge crack to the leading edge free surface. The length of the cracks that have appeared on the surface of the continuous crack on both sides; Collapsed body transport loss index The shortest distance from the leading edge of the landslide to the riverbank and ground slope Related; When 0≤ α When <35°: ； When 35°≤ α When <55°: ； When 55°≤ α When <75°: ； When 75°≤ α When ≤90°: 。 9. The method for predicting and warning of typhoon-induced landslides and dam formations caused by combined typhoon and seawater intrusion, as described in claim 1, is characterized in that... In step S5, historical data on dam bodies and rivers during dam-building disasters are statistically analyzed. The dam body data includes the dam length. River data includes flow velocity Riverbed width Riverbed depth and cross-sectional area The river data was then normalized to obtain... , , as well as The normalization method is as follows: ； ； ； ； in: and These are the minimum and maximum values ​​of the flow velocity, respectively. and These are the minimum and maximum values ​​of the riverbed width, respectively; and These are the minimum and maximum values ​​of the riverbed depth, respectively. and These are the minimum and maximum values ​​of the riverbed cross-sectional area, respectively. The functional relationship between the dam data and the normalized river data was then determined, and the constructed functional relationship is as follows: ； in, , , , , , , , , , , , , , as well as All are regression coefficients; Calculate using the solved functional relationship The formula for calculating the volume of landslide required for damming a river on a slope is as follows: ； in: The volume of landslide material required to build a dam to block the river. The cross-sectional area of ​​the riverbed. This refers to the length of the dam body.

10. The method for predicting and warning of typhoon-induced landslides and dam formations caused by combined typhoon and seawater intrusion, as described in claim 1, is characterized in that: according to median The warning levels, warning strategies, and warning measures are determined. There are four warning levels, and each warning level corresponds to a specific warning strategy and warning measures.

Images