Landslide rainfall threshold curve characterization method based on slope unit and application of landslide rainfall threshold curve characterization method

Abstract

The invention relates to a slope unit-based landslide rainfall threshold curve characterization method and application thereof, and the method comprises the steps: delimiting a research region, then obtaining the data and associated data of an underlying surface of the research region, dividing a mountain region in the research region, and obtaining a plurality of slope units; in combination with underlying surface data, obtaining a multi-dimensional characteristic variable of each slope unit, dividing a weight ratio for each characteristic variable and sorting the characteristic variables; classifying the slope units based on the sorted feature variables; for each type of slope unit, rainfall information and landslide information are extracted, a probabilistic rainfall threshold is determined according to a preset early warning probability, and a landslide rainfall threshold curve is obtained; the method is used for providing time-space coordinate information and evaluating the risk degree of the landslide disaster under a given rainfall condition when the landslide disaster is forecasted. According to the method, the early warning accuracy is improved, the interpretability is high, the time-space coordinate information of the landslide disaster can be accurately forecasted, and the risk degree of the landslide disaster under the given rainfall condition can be more accurately evaluated.

Description

Technical Field

[0001] The present invention relates to the technical field of computing, reckoning or counting, and particularly relates to a method for characterizing a landslide rainfall threshold curve based on slope units in the field of geological disaster early warning and its application. Background Art

[0002] A landslide refers to a natural phenomenon in which soil or rock mass on a slope is affected by factors such as river erosion, groundwater activity, rainwater immersion, earthquake, and artificial slope cutting, and slides downward along a certain weak surface or weak zone as a whole or dispersedly under the action of gravity. Among them, rainfall-induced landslides account for a large proportion. Rainfall-induced landslides are one of the main types of geological disasters, characterized by group occurrence and sudden occurrence, and it is difficult to prevent and control them through conventional deformation monitoring and early warning. The landslide early warning technology based on rainfall monitoring can reasonably predict disasters by determining the rainfall threshold that causes landslides, and can leave enough buffer time for local emergency response and evacuation.

[0003] Currently, the most commonly used rainfall threshold method is the regional rainfall intensity - rainfall duration threshold curve (I-D threshold curve) obtained based on historical data statistics, and its general expression is I = aD -b . Calculate the corresponding rainfall intensity threshold I based on the rainfall duration D. When the average rainfall intensity of an actual rainfall event reaches I, it is considered that this rainfall event will trigger a landslide disaster in this area.

[0004] There is an important evaluation loophole in the traditional regional I-D threshold curve, that is, a single rainfall threshold is generally adopted at the regional scale, ignoring the differences in the underlying surface environment of the slope. Therefore, it only reflects the overall comprehensive level of the study area, and the early warning lacks pertinence and is prone to misjudgment. In the prior art, many scholars have studied the influence of different underlying surface environments on rainfall thresholds. The idea is to classify historical disaster points according to the underlying surface environment where they are located, extract rainfall events that cause disasters and then plot them on a logarithmic coordinate system, and construct the lowest envelope line as the rainfall threshold curve. Compared with the traditional regional single threshold, this method refines the early warning scale of the rainfall threshold and enhances the interpretability of the early warning model. However, this method has the following disadvantages:

[0005] (1) Such deterministic rainfall thresholds only consider rainfall events that cause disasters, and do not consider a large number of non-disaster rainfall events during the characterization process, and are easily interfered by extremely individual abnormal samples;

[0006] (2) The real underlying surface environment is usually characterized as high-dimensional complex variables, while this method usually can only select a single environmental factor for classification. Summary of the Invention

[0007] The present invention solves the problems existing in the prior art and provides a method for characterizing a landslide rainfall threshold curve based on slope units and its application.

[0008] The technical concept of the present invention is based on the data of the underlying surface and associated data collected. Taking the slope unit as the early warning object, characteristic variables are extracted and their relevance to landslide occurrence is evaluated. During the evaluation process, the uncertainty of negative sample selection is fully considered, and the final relative importance is reasonably calculated through Monte Carlo experiments. Several characteristic variables with the highest final relative importance are selected, and the slope units are classified using a clustering algorithm. Considering the joint distribution characteristics of multi-dimensional characteristic variables, the scale of the study area is refined; at the same time, considering disaster-causing rainfall events and non-disaster-causing rainfall events, a probability rainfall threshold is established for each type of slope unit, so as to more accurately evaluate the risk degree of landslide disasters under given rainfall conditions.

[0009] The technical solution adopted by the present invention is a method for characterizing the rainfall threshold curve of landslides based on slope units, and the method includes the following steps:

[0010] S1 Define the study area, obtain the data of the underlying surface and associated data of the study area, and divide the mountainous area in the study area to obtain several slope units;

[0011] S2 Combine the data of the underlying surface to obtain multi-dimensional characteristic variables of each slope unit, divide the weight ratio of each characteristic variable and sort them; classify the slope units based on the sorted characteristic variables;

[0012] S3 For each type of slope unit, extract rainfall information and landslide information, and determine the probability rainfall threshold according to the preset warning probability P st to obtain the landslide rainfall threshold curve. Here, the preset warning probability P st refers to the landslide probability value for issuing an alarm, such as 5%, 10%, etc.

[0013] Preferably, in S1, the data of the underlying surface of the study area includes elevation, slope, aspect, plane curvature, profile curvature, lithology, normalized difference vegetation index, land use; the land use here includes but is not limited to shrubs, forest land, cultivated land, etc.;

[0014] The associated data includes historical landslide disaster data of the study area and meteorological observation data of the study area; the historical landslide disaster data includes but is not limited to the geographical coordinates and occurrence time of the historical landslide location.

[0015] Preferably, in S1, dividing the mountainous area into several slope units based on the digital elevation model of the study area includes the following steps:

[0016] S1.1 Remove the areas with a slope less than the preset value in the digital elevation model;

[0017] S1.2 Divide slope units by the r.slopeunits method, input the predefined parameters of the algorithm, including the initial cumulative flow threshold, the minimum area threshold of slope units, the minimum circular variance of slope aspect, the reduction coefficient, and the cleaning scale threshold, and obtain the corresponding slope unit division results.

[0018] In practical applications, the minimum area threshold of slope units and the minimum circular variance of slope aspect can be determined by an optimization algorithm; specifically, preset several pairs of parameter combinations, obtain the slope unit division results corresponding to each pair of input parameters, calculate the average value of the circular variance of slope aspect within the slope units, and calculate the Moran's index of the average slope aspect of the slope units in the study area; select the most suitable parameter combination to minimize the two indicators as much as possible.

[0019] Preferably, S2 includes the following steps:

[0020] S2.1 Extract the multi-dimensional feature variables of each slope unit based on the underlying surface data.

[0021] S2.2 Based on the historical landslide disaster data of the study area and the spatial coordinate relationship of the slope units, mark the slope units where landslide events have occurred as positive samples.

[0022] S2.3 With the positive-negative sample ratio r, randomly select slope units from the slope units where no landslide events have occurred corresponding to the historical landslide disaster data of the study area and mark them as negative samples; calculate the relative importance of each feature variable in the multi-dimensional feature variables for landslide occurrence.

[0023] S2.4 Repeat S2.3 N times, calculate the average value of the relative importance of each feature variable, and sort the feature variables from largest to smallest.

[0024] S2.5 Select the top n feature variables with the highest importance, after normalization preprocessing, use the clustering algorithm to divide the slope units into m classes.

[0025] Preferably, in S2.1, the multi-dimensional feature variables are the statistical indicators of the data of each type of underlying surface of each slope unit, and the multi-dimensional feature variables also include the perimeter or area of each slope unit; calculate the statistical indicators for each type of underlying surface such as elevation and slope aspect respectively to form the multi-dimensional feature variables.

[0026] Preferably, in S2.3, the relative importance of each feature variable in the multi-dimensional feature variables for landslide occurrence is calculated using a random forest model and permutation importance analysis; in the random forest model, the number of trees, the maximum depth of the trees, and the maximum number of features considered at each split are optimized and adjusted, and the most suitable hyperparameter combination is determined based on ten-fold cross-validation and grid search; the random forest model is trained using positive and negative samples and the original accuracy of the model is calculated; for each feature variable, it is randomly permuted, the random forest model is retrained and the model accuracy is calculated; the difference between the original accuracy and the model accuracy after permutation is compared, and the larger the difference, the more important the corresponding feature; the accuracy differences of all feature variables are normalized so that their sum is 1, which is used as the relative importance.

[0027] Preferably, r is 1:2 to 5; N is 1000 to 2000.

[0028] Preferably, S3 includes the following steps:

[0029] S3.1 For each type of slope unit, extract the meteorological observation data of the surrounding study area corresponding to rainfall events.

[0030] S3.2 Based on the historical landslide disaster data of the study area, classify the rainfall events in S3.1 into disaster-causing rainfall events and non-disaster-causing rainfall events.

[0031] S3.3 According to the preset warning probability P st Determine the probability rainfall threshold and obtain the landslide rainfall threshold curve.

[0032] Preferably, the preset warning probability P st is determined according to the threshold level; the threshold levels include red warning, orange warning, yellow warning, and blue warning, and P st gradually decreases.

[0033] An application of the landslide rainfall threshold curve characterization method based on slope units as described above, which is applied to provide spatio-temporal coordinate information and evaluate the risk degree of landslide disasters under given rainfall conditions when predicting landslide disasters.

[0034] The present invention relates to a landslide rainfall threshold curve characterization method based on slope units and its application. After delimiting the study area, the data of the underlying surface and associated data of the study area are obtained, and the mountainous areas in the study area are divided to obtain a number of slope units; in combination with the data of the underlying surface, the multi-dimensional feature variables of each slope unit are obtained, the weight ratios of each feature variable are divided and sorted; the slope units are classified based on the sorted feature variables; for each type of slope unit, rainfall information and landslide information are extracted, and according to the preset warning probability P stDetermine the probability rainfall threshold and obtain the landslide rainfall threshold curve; the method is applied to provide spatio-temporal coordinate information and evaluate the risk degree of landslide disasters under given rainfall conditions when forecasting landslide disasters.

[0035] The beneficial effects of the present invention are as follows:

[0036] (1) Compared with the traditional single threshold in the region, the present invention further divides the research area, refines the warning scale of the rainfall threshold, and improves the warning accuracy;

[0037] (2) Compared with the traditional empirical rainfall threshold, the present invention considers the influence of the underlying surface environment and its synergistic effect on the occurrence of landslides, and has strong interpretability;

[0038] (3) Taking the slope unit as the warning object and using the probability rainfall threshold to forecast landslide events is conducive to accurately forecasting the spatio-temporal coordinate information of landslide disasters and more accurately evaluating the risk degree of landslide disasters under given rainfall conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is the flowchart of the method of the present invention;

[0040] Figure 2 is a partial schematic diagram of the slope unit division result in the example area of the present invention;

[0041] Figure 3 is a schematic diagram of the spatial distribution of historical landslide disaster points in the example area of the present invention;

[0042] Figure 4 is a bar chart of the final relative importance of some characteristic variables in the embodiment;

[0043] Figure 5 is a relationship curve of the number of characteristic variables n with the optimal clustering number m and the weighted average AUC;

[0044] Figure 6 is the distribution of different types of slope units in the research area;

[0045] Figure 7 is the result of the probability rainfall threshold curve of different types of slope units in the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0046] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0047] The present invention relates to a method for characterizing the rainfall threshold curve of landslides based on slope units, and the method includes the following steps:

[0048] S1 Define the study area, obtain the underlying surface data and associated data of the study area, and divide the mountainous areas in the study area to obtain a number of slope units;

[0049] In S1, the underlying surface data of the study area includes elevation, slope, aspect, plane curvature, profile curvature, lithology, normalized difference vegetation index, land use;

[0050] The associated data includes the historical landslide disaster data of the study area and the meteorological observation data of the study area.

[0051] In S1, dividing the mountainous area into a number of slope units based on the digital elevation model of the study area includes the following steps:

[0052] S1.1 Remove the areas with a slope less than the preset value in the digital elevation model;

[0053] S1.2 Select the r.slopeunits method to divide the slope units, input the predefined parameters of the algorithm, including the initial cumulative flow threshold, the minimum area threshold of the slope unit, the minimum circular variance of the aspect, the reduction coefficient, and the cleaning scale threshold, and obtain the corresponding slope unit division result.

[0054] The minimum area threshold of the slope unit and the minimum circular variance of the aspect can be determined by an optimization algorithm. Specifically, a number of pairs of parameter combinations are preset, the slope unit division results corresponding to each pair of input parameters are obtained, the average circular variance of the aspect within the slope unit is calculated, and the Moran's index of the average aspect of the slope units in the study area is calculated; the most suitable parameter combination is selected to minimize the two indicators as much as possible.

[0055] S2 Combine the underlying surface data, obtain the multi-dimensional characteristic variables of each slope unit, divide the weight ratio of each characteristic variable and sort them; classify the slope units based on the sorted characteristic variables;

[0056] S2 includes the following steps:

[0057] S2.1 Extract the multi-dimensional characteristic variables of each slope unit based on the underlying surface data;

[0058] In S2.1, the multi-dimensional characteristic variables are the statistical indicators of each type of underlying surface data of each slope unit, and the multi-dimensional characteristic variables also include the perimeter or area of each slope unit.

[0059] S2.2 Based on the historical landslide disaster data of the study area and the spatial coordinate relationship of the slope units, mark the slope units where landslide events have occurred as positive samples;

[0060] S2.3 Randomly select slope units from the slope units corresponding to the historical landslide disaster data in the study area that have not experienced landslide events as negative samples according to the positive-negative sample ratio r; calculate the relative importance of each feature variable in the multi-dimensional feature variables for landslide occurrence.

[0061] In S2.3, calculate the relative importance of each feature variable in the multi-dimensional feature variables for landslide occurrence based on the random forest model and permutation importance analysis.

[0062] In the random forest model, optimize and adjust the number of trees, the maximum depth of the trees, and the maximum number of features considered at each split, and determine the most suitable combination of hyperparameters based on ten-fold cross-validation and grid search method; use positive and negative samples to train the random forest model and calculate the original accuracy of the model.

[0063] For each feature variable, randomly permute it, retrain the random forest model and calculate the model accuracy; compare the difference between the original accuracy and the model accuracy after permutation, the larger the difference indicates the more important the corresponding feature; normalize the accuracy differences of all feature variables so that their sum is 1 as the relative importance.

[0064] r is 1:2 to 5.

[0065] N is 1000 to 2000.

[0066] Repeat S2.3 for N times, calculate the average value of the relative importance of each feature variable, and sort the feature variables from largest to smallest.

[0067] Select the top n feature variables with the highest importance, after normalization preprocessing, use the clustering algorithm to divide the slope units into m classes.

[0068] In S2.5, the value of the number of feature variables n is determined through the following steps:

[0069] First, set the value range of n; for each value of n, use the elbow method or other optimization methods to determine the optimal number of clusters m, and use the clustering algorithm to divide the slope units into m classes.

[0070] For each class of slope units, extract the meteorological observation data of its surrounding study area, corresponding to rainfall events; based on the historical landslide disaster data of the study area, classify the rainfall events into disaster-causing rainfall events and non-disaster-causing rainfall events.

[0071] Subsequently, for each class of slope units, draw the receiver operating characteristic curve of rainfall threshold predicting landslide events and calculate the area under the curve AUC; calculate the weighted average AUC of all classes of slope units.

[0072] Finally, plot the relationship curve between n and the weighted average AUC, select the value of n corresponding to the maximum weighted average AUC as the optimal number of feature variables, and use the clustering result of the slope units at this time as the final clustering result.

[0073] Correspondingly, the value of the number of clusters m is determined by the elbow method or other corresponding optimization methods. The specific steps are as follows:

[0074] First, set the value range of the number of clusters m;

[0075] For each value, use the selected clustering algorithm to cluster the data set and calculate the Bayesian Information Criterion (BIC) of the clustering result;

[0076] Plot the relationship curve between m and BIC, observe the inflection point of the curve, and the corresponding value of m is the optimal number of clusters.

[0077] In the implementation process of the present invention, usually r is set to 1:3, N is set to 1000, and n is set to 3 - 10.

[0078] In the present invention, the statistical indicators of the multi-dimensional feature variables in S2.1 include but are not limited to mean, standard deviation, mode, range, etc.

[0079] S3 For each type of slope unit, extract rainfall information and landslide information, and determine the probability rainfall threshold according to the preset warning probability P st to obtain the landslide rainfall threshold curve.

[0080] S3 includes the following steps:

[0081] S3.1 For each type of slope unit, extract the meteorological observation data of the surrounding research area corresponding to the rainfall event;

[0082] S3.2 Based on the historical landslide disaster data of the research area, classify the rainfall events in S3.1 into disaster-causing rainfall events and non-disaster-causing rainfall events;

[0083] S3.3 According to the preset warning probability P st to determine the probability rainfall threshold and obtain the landslide rainfall threshold curve.

[0084] The preset warning probability P st is determined according to the threshold level; the threshold levels include red warning, orange warning, yellow warning, and blue warning, and P st gradually decreases.

[0085] In the present invention, for red warning, orange warning, yellow warning, and blue warning, the preset warning probabilities P st are taken as 16%, 6.7%, 2.3%, and 0.62% respectively.

[0086] The present invention also relates to an application of the above-described method for characterizing the rainfall threshold curve of landslides based on slope units, which is applied to providing spatio-temporal coordinate information and evaluating the risk degree of landslides under given rainfall conditions when predicting landslide disasters.

[0087] In the actual application process of the present invention, since the characteristic information, rainfall information, and landslide information corresponding to each type of slope unit are obtained, during the actual rainfall process, by obtaining rainfall information data, it can be accurately corresponding to whether each type of slope unit will have a landslide. In addition to giving the risk degree of landslide disasters corresponding to the landslide rainfall threshold curve, spatio-temporal coordinate information can also be given.

[0088] The following uses specific examples to illustrate the construction of the rainfall-induced shallow soil landslide I-D threshold curve of the present invention; the research area is a certain third-level basin in a certain province, and the geological disasters occurring in this area are mainly high-level shallow soil landslides. During the plum rain or typhoon events, such landslides cause immeasurable losses of life and property locally, and it is urgent to establish a reasonable regional rainfall threshold to improve the local disaster prevention and mitigation capabilities.

[0089] (1) Obtain the underlying surface geographic information data, historical landslide disaster data, and meteorological observation data of the research area, and preset the value ranges of the positive-negative sample ratio r, the number of repeated samplings N, the number of characteristic variables n, and the number of slope unit categories m.

[0090] In this case, the types of underlying surface data considered include elevation, slope, aspect, plane curvature, profile curvature, terrain humidity index, terrain position index, terrain ruggedness index, flow intensity index, terrain surface texture, lithology, distance to fault, normalized vegetation index, distance to highway, distance to river, and land use. The data sources include publicly available Internet materials and relevant departments. The positive-negative sample ratio is preset to be 1:3, the number of repeated samplings is taken as 1500 times, the value range of the finally selected number of characteristic variables is selected from 2 to 27, and the value range of the number of slope unit categories is selected from 2 to 20.

[0091] (2) Based on the digital elevation model of the research area, divide the mountainous area into several slope units.

[0092] Use the r.slopeunits method to automatically divide the slope units of the research area. A local example is shown in Figure 2 . The main steps include:

[0093] (1) Clear the plain areas in the digital elevation model where the slope is less than 5°.

[0094] (2) Input the predefined parameters of the algorithm, where the initial cumulative flow threshold is taken as 1500000 m 2 , the reduction coefficient is taken as 5, and the cleaning scale threshold is taken as 10000 m 2; The preselected minimum area threshold of slope units includes (5000, 10000, 25000, 50000, 75000, 100000, 125000, 150000, 200000) m 2 , and the minimum circular variance of slope aspect includes (0.1, 0.15, 0.2, 0.25, 0.3, 0.35, 0.4, 0.45, 0.5);

[0095] (3) Obtain the slope unit division results corresponding to different parameter combinations, and calculate the average circular variance of slope aspect within the slope units in the study area and the Moran's index of the average slope aspect of the slope units. To minimize the two indicators as much as possible, the minimum area threshold of slope units is taken as 10000 m 2 , and the minimum circular variance of slope aspect is taken as 0.2 to obtain the final slope unit division result.

[0096] (3) Extract the multi-dimensional characteristic variables of each slope unit based on the underlying surface geographic information data.

[0097] Based on the underlying surface geographic information data obtained in step (1) and the slope unit division results in step (2), calculate the multi-dimensional characteristic variables of each slope unit, including the statistical indicators of each type of underlying surface information, as well as the terrain undulation and shape index of the slope unit, as shown in Table 1 specifically.

[0098] Table 1: Characteristic variables of slope units considered

[0099]

[0100] a Take the average value as the characteristic variable of the slope unit

[0101] b Take the standard deviation as the characteristic variable of the slope unit

[0102] c Take the mode as the characteristic variable of the slope unit

[0103] (4) Mark the slope units where landslide events have occurred as positive samples, randomly select slope units from the slope units where landslide events have not occurred in the past and mark them as negative samples, and calculate the relative importance of each characteristic variable for landslide occurrence; repeat 1500 times, and calculate the average value of the relative importance of each characteristic variable as the final relative importance and sort them.

[0104] Historical landslide points are obtained through visual interpretation of historical satellite images, such as Figure 3As shown. Based on the collected positive and negative samples, the relative importance of each feature variable in the multi-dimensional feature variables for landslide occurrence is calculated using a random forest model and permutation importance analysis; in the random forest model, the number of trees, the maximum depth of the trees, and the maximum number of features considered at each split are optimized and adjusted, and the most suitable hyperparameter combination is determined based on ten-fold cross-validation and grid search method; the random forest model is trained using positive and negative samples and the original accuracy of the model is calculated; for each feature variable, it is randomly permuted, the random forest model is retrained and the model accuracy is calculated; the difference between the original accuracy and the model accuracy after permutation is compared, and the larger the difference, the more important the corresponding feature; the accuracy differences of all feature variables are normalized so that their sum is 1, which is used as the relative importance. The final relative importance of the feature variables is shown in Figure 4 .

[0105] (V) Select the top n feature variables with the highest importance. After normalization preprocessing, use a clustering algorithm to divide the slope units into m classes.

[0106] Based on the value range of n, for each value of n, select several feature variables with the highest importance. After normalization preprocessing, use a Gaussian mixture density model to classify the slope units; use the elbow method to determine the optimal number of clusters m, and use the clustering algorithm to divide the slope units into m classes; for each class of slope units, extract the meteorological observation data of the surrounding research area, corresponding to rainfall events; based on the historical landslide disaster data of the research area, classify the rainfall events into disaster-causing rainfall events and non-disaster-causing rainfall events; subsequently, for each class of slope units, draw the receiver operating characteristic curve of the rainfall threshold predicting landslide events and calculate the area under the curve AUC; calculate the weighted average AUC of all classes of slope units; finally, draw the relationship curve between n and the weighted average AUC. The relationship curves between the number of feature variables n, the optimal number of clusters m, and the weighted average AUC are shown in Figure 5 .

[0107] When n takes 26, the weighted average AUC reaches the maximum value. At this time, the slope units in the study area are divided into 9 categories, that is, the final clustering result. The distribution of different types of slope units in the study area is shown in Figure 6 .

[0108] (VI) For each class of slope units, extract the relevant disaster-causing rainfall events and non-disaster-causing rainfall events, and determine the probability rainfall threshold according to the preset warning probability P st Determine the probability rainfall threshold.

[0109] For each class of slope units, extract the rainfall events of the ground meteorological observation stations within 5 km around. Based on whether a landslide event occurs, they are classified into disaster-causing rainfall events and non-disaster-causing rainfall events. Construct the probability rainfall thresholds for red warning, orange warning, yellow warning, and blue warning, and the corresponding preset warning probability Pst They are 16%, 6.7%, 2.3%, and 0.62% respectively. The rainfall thresholds for the four types of slope units are as Figure 7 shown.

[0110] The above embodiments are only used to explain the concept of the present invention, rather than limiting the protection scope of the present invention. Any non-substantive modification made to the present invention using this concept shall fall within the protection scope of the present invention.

[0111] To achieve the above, the present invention also relates to a computer-readable storage medium, on which a landslide rainfall threshold curve characterization program based on slope units is stored. When the program is executed by a processor, the above-mentioned landslide rainfall threshold curve characterization method based on slope units is implemented.

[0112] To achieve the above, the present invention also proposes a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the above-mentioned landslide rainfall threshold curve characterization method based on slope units is implemented.

[0113] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0114] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the specified functions in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0115] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured product including an instruction device, and the instruction device implements the functions in Figure 1 one process or multiple processes and / or blocksFigure 1 The functions specified in one or more boxes.

[0116] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are performed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one Figure 1 process or more processes and / or boxes Figure 1 or more boxes.

[0117] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed as including the preferred embodiments and all changes and modifications falling within the scope of the present invention.

[0118] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A method for characterizing the rainfall threshold curve of landslides based on slope units, characterized in that: The method includes the following steps: S1 Define the study area, obtain the underlying surface data and associated data of the study area, and divide the mountainous areas in the study area to obtain a number of slope units; S2 Combine the underlying surface data to obtain the multi-dimensional characteristic variables of each slope unit, divide the weight ratio of each characteristic variable and sort them; classify the slope units based on the sorted characteristic variables; S3 For each type of slope unit, extract rainfall information and landslide information, and determine the probability rainfall threshold according to the preset warning probability P st to obtain the landslide rainfall threshold curve.

2. A method for characterizing the rainfall threshold curve of landslides based on slope units according to claim 1, characterized in that: In S1, the underlying surface data of the study area includes elevation, slope, aspect, plane curvature, profile curvature, lithology, normalized difference vegetation index, land use; The associated data includes the historical landslide disaster data of the study area and the meteorological observation data of the study area.

3. A method for characterizing the rainfall threshold curve of landslides based on slope units according to claim 2, characterized in that: In S1, dividing the mountainous area into a number of slope units based on the digital elevation model of the study area includes the following steps: S1.1 Remove the areas with a slope less than the preset value in the digital elevation model; S1.2 Use the r.slopeunits method to divide the slope units, input the predefined parameters of the algorithm, including the initial cumulative flow threshold, the minimum area threshold of the slope unit, the minimum circular variance of the aspect, the reduction coefficient, and the cleaning scale threshold, to obtain the corresponding slope unit division result.

4. A method for characterizing the rainfall threshold curve of landslides based on slope units according to claim 2, characterized in that: S2 includes the following steps: S2.1 Extract the multi-dimensional characteristic variables of each slope unit based on the underlying surface data; S2.2 Based on the historical landslide disaster data of the study area and the spatial coordinate relationship of the slope units, mark the slope units where landslide events have occurred as positive samples; S2.3 Using the positive and negative sample ratio r , randomly select slope units from the slope units without landslide events corresponding to the historical landslide disaster data in the study area and label them as negative samples; Calculate the relative importance of each characteristic variable in the multi-dimensional characteristic variables for the occurrence of landslides; S2.4 Repeat S2.3 N times, calculate the average value of the relative importance of each feature variable, and sort the feature variables from largest to smallest; S2.5 Select the top n feature variables with the highest importance. After normalization preprocessing, use the clustering algorithm to divide the slope units into m categories.

5. A method for characterizing the rainfall threshold curve of landslides based on slope units according to claim 4, characterized in that: In S2.1, the multi-dimensional characteristic variables are the statistical indicators of the data of each type of underlying surface of each slope unit, and the multi-dimensional characteristic variables also include the perimeter or area of each slope unit.

6. A method for characterizing the rainfall threshold curve of landslides based on slope units according to claim 4, characterized in that: In S2.3, use the random forest model and permutation importance analysis to calculate the relative importance of each characteristic variable in the multi-dimensional characteristic variables for the occurrence of landslides; In the random forest model, optimize and adjust the number of trees, the maximum depth of the trees, and the maximum number of features considered at each split, and determine the most suitable combination of hyperparameters based on ten-fold cross-validation and grid search method; Use the positive and negative samples to train the random forest model and calculate the original accuracy of the model. For each characteristic variable, randomly permute it, retrain the random forest model and calculate the model accuracy; compare the difference between the original accuracy and the accuracy of the permuted model, and normalize the accuracy differences of all characteristic variables as the relative importance.

7. A method for characterizing the rainfall threshold curve of landslides based on slope units according to claim 4, characterized in that: r is 1:2 to 5; N is 1000 to 2000.

8. A method for characterizing the rainfall threshold curve of landslides based on slope units according to claim 2, characterized in that: S3 includes the following steps: S3.1 For each type of slope unit, extract the meteorological observation data of the surrounding study area corresponding to the rainfall events; S3.2 Based on the historical landslide disaster data of the study area, distinguish the rainfall events in S3.1 into disaster-causing rainfall events and non-disaster-causing rainfall events; S3.3 Determine the probability rainfall threshold according to the preset warning probability P st Determine the probability rainfall threshold and obtain the landslide rainfall threshold curve.

9. A method for characterizing a landslide rainfall threshold curve based on a slope unit according to claim 8, characterized in that: The preset warning probability P st is determined according to the threshold level; the threshold level includes red warning, orange warning, yellow warning, and blue warning, P st and gradually decreases.

10. Application of the method for characterizing the rainfall threshold curve of landslides based on slope units according to any one of claims 1 to 9, characterized in that: It is applied to provide spatio-temporal coordinate information and evaluate the landslide disaster risk level under given rainfall conditions when forecasting landslide disasters.

Images