Landslide disaster susceptibility and dangerousness coupling evaluation method

Abstract

The embodiment of the invention provides a landslide disaster susceptibility and dangerousness coupling evaluation method, and belongs to the technical field of data processing, and the method specifically comprises the steps: extracting landslide disaster factor features according to the multi-source data of a target area, and carrying out the preprocessing; performing interval division on the grid feature data and calculating the landslide event frequency of each interval; according to rainfall raster data in the raster feature data, calculating a local Moran index of accumulated rainfall in a preset time interval to identify an abnormal region in rainfall spatial distribution; performing susceptibility-danger coupling evaluation based on Bayesian optimization according to the landslide event frequency and the rainfall abnormal factor; and inputting a real-time static disaster-pregnant feature interval index of each grid in the target area based on the disaster-pregnant feature optimal weight to obtain a susceptibility evaluation result graph, and inputting a real-time feature vector of each grid into a risk evaluation model to obtain a risk evaluation result graph. Through the scheme of the invention, the evaluation accuracy and adaptability are improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of data processing technology, and in particular to a method for coupling evaluation of landslide susceptibility and hazard. Background Art

[0002] The shortcomings of current landslide hazard assessment methods are as follows: (1) The susceptibility assessment uses "whether a landslide occurs" as a label, but landslide occurrence is closely related to external factors such as rainfall, resulting in the label being unable to accurately characterize the inherent susceptibility of the region, thus affecting the final hazard assessment. (2) Areas where no landslides have occurred are usually considered negative samples, but there may be potential high-risk areas that have not yet triggered landslides, and the random sampling of negative samples may lead to unstable modeling results, affecting the accuracy of hazard assessment.

[0003] It can be seen that there is an urgent need for a coupled evaluation method for landslide susceptibility and hazard with higher assessment accuracy and adaptability. Summary of the invention

[0004] In view of this, an embodiment of the present invention provides a method for coupling assessment of landslide susceptibility and hazard, which at least partially solves the problems of poor assessment accuracy and adaptability in the prior art.

[0005] The embodiment of the present invention provides a method for coupling assessment of landslide susceptibility and hazard, including: Step 1, extracting the landslide hazard factor characteristics based on the multi-source data of the target area and preprocessing them to obtain raster feature data; Step 2, divide the grid feature data into intervals and calculate the landslide event frequency in each interval; Step 3, calculating the local Moran index of the accumulated rainfall in a preset time interval according to the rainfall grid data in the grid feature data to identify the abnormal area in the rainfall spatial distribution and obtain the rainfall anomaly factor; Step 4: According to the frequency of landslide events and rainfall anomaly factors, a susceptibility-hazard coupling evaluation based on Bayesian optimization is performed to obtain the optimal weights of disaster-prone characteristics and a hazard evaluation model; Step 5: Based on the optimal weight of the disaster-pregnant feature, the real-time static disaster-pregnant feature interval index of each grid in the target area is input to obtain a susceptibility assessment result map, and the real-time feature vector of each grid is input into the hazard assessment model to obtain a hazard assessment result map.

[0006] According to a specific implementation of the embodiment of the present invention, step 1 specifically includes: Step 1.1, dividing the multi-source data of the target area into static factor features and dynamic factor features according to the data feature update cycle, wherein the static factor features include geological structure elements, topographic features, deformation features, and human activity factors, and the dynamic factor features are rainfall raster data; Step 1.2, unify the spatial reference and grid resolution of the static factor features and the dynamic factor features to obtain grid feature data.

[0007] According to a specific implementation of the embodiment of the present invention, step 2 specifically includes: Step 2.1, perform interval processing according to the value type of each feature in the static factor feature, and obtain the interval set corresponding to each feature; Step 2.2, calculating the landslide event frequency of each interval in the interval set corresponding to each feature according to a preset formula, wherein the preset formula is: ; in, is the total number of pixels in the target area whose values ​​of feature f are in interval k, Indicates the number of pixels where landslides occurred. is the total number of positive samples, is the total number of pixels in the target area.

[0008] According to a specific implementation of the embodiment of the present invention, step 3 specifically includes: Step 3.1, calculating the local Moran index corresponding to the rainfall of each pixel within a preset time interval based on the rainfall grid data; Step 3.2, statistically verify the local Moran index of all pixels according to the preset significance threshold to obtain the significance score; Step 3.3: Identify rainfall anomaly pixels based on the significance score and calculate their corresponding rainfall anomaly factors. ; in, Represents pixel The local Moran index, Represents pixel The significance score of represents the significant threshold, Indicates high-high aggregation areas between pixels and their neighbors. Represents areas of high-low aggregation of pixels and their neighbors.

[0009] According to a specific implementation of the embodiment of the present invention, step 4 specifically includes: Step 4.1, based on the frequency of landslide events, the frequencies of different types of features are combined according to the set initial weight vector to obtain their corresponding weight coefficients, and the scores of each feature interval and the weight coefficients of the corresponding features are combined in each pixel to obtain the susceptibility factor that characterizes the fragility of the geological environment; Step 4.2, randomly generate the date of each pixel, then form a negative sample set with the pixels that meet the preset conditions, and combine it with the positive samples composed of all landslide events to form a hazard model evaluation sample, wherein the preset conditions are that the rainfall anomaly factor corresponding to the pixel is 0 and the susceptibility factor is less than the susceptibility threshold; Step 4.3, construct the dangerous feature set based on the dynamic factor characteristics, rainfall anomaly factors and susceptibility factors; Step 4.4, according to the coordinates and date of each sample data in the risk model evaluation sample, index and obtain the corresponding feature value from the risk feature set and divide it into a training set and a test set according to a preset ratio; Step 4.5, use the training set to train the random forest model and calculate the accuracy index of the trained model on the test set; Step 4.6, construct the objective function according to the accuracy and F1 score in the precision index, take minimizing the objective function value as the optimization goal, solve the objective function through Bayesian optimization iteration, and obtain the optimal weight and optimal susceptibility threshold of the disaster-prone features; Step 4.7, take the optimal susceptibility threshold as the new susceptibility threshold and execute steps 4.2 to 4.5 again to obtain the risk assessment model.

[0010] According to a specific implementation of the embodiment of the present invention, the expression of the objective function is: ; in, Indicates accuracy, represents the weight coefficient, represents the susceptibility threshold, Denotes the F1 score.

[0011] The coupled evaluation scheme for the susceptibility and hazard of landslide disasters in the embodiment of the present invention comprises: step 1, extracting the characteristics of landslide hazard factors according to the multi-source data of the target area and performing preprocessing to obtain grid feature data; step 2, dividing the grid feature data into intervals and calculating the frequency of landslide events in each interval; step 3, calculating the local Moran index of the cumulative rainfall in a preset time interval according to the rainfall grid data in the grid feature data to identify the abnormal area in the spatial distribution of rainfall, and obtaining the rainfall anomaly factor; step 4, performing a susceptibility-hazard coupling evaluation based on Bayesian optimization according to the landslide event frequency and the rainfall anomaly factor, and obtaining the optimal weight of the disaster-pregnant feature and the hazard evaluation model; step 5, based on the optimal weight of the disaster-pregnant feature, inputting the real-time static disaster-pregnant feature interval index of each grid in the target area, obtaining the susceptibility evaluation result map, and inputting the real-time feature vector of each grid into the hazard evaluation model to obtain the hazard evaluation result map.

[0012] The beneficial effects of the embodiments of the present invention are as follows: through the scheme of the present invention, firstly, according to the historical landslide point samples, by counting the frequency of landslide events in each characteristic interval and weighting and stacking, the weighted results of all spatial units are summarized to construct a preliminary susceptibility evaluation factor; secondly, the susceptibility is combined with dynamic disaster-inducing factors to establish a hazard assessment model; finally, through Bayesian optimization, the susceptibility factor weights and negative sample constraints are adaptively adjusted to obtain the susceptibility-hazard evaluation result with the best accuracy in the test set. This scheme can not only improve the accuracy and stability of landslide hazard assessment in terms of method, but also provide a more targeted scientific basis for geological disaster risk management and emergency decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0014] Figure 1 A schematic flow chart of a method for coupling landslide susceptibility and hazard assessment provided by an embodiment of the present invention; Figure 2 A schematic diagram of the results of anomaly detection of historical seven-day rainfall characteristics provided by an embodiment of the present invention; Figure 3 A susceptibility-criticality coupling evaluation framework diagram provided by an embodiment of the present invention; Figure 4 A schematic diagram of a susceptibility evaluation result provided by an embodiment of the present invention; Figure 5A schematic diagram of a risk assessment result provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0015] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0016] The following describes the embodiments of the present invention through specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0017] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein can be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present invention, it should be understood by those skilled in the art that an aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects described herein can be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein can be used to implement this device and / or practice this method.

[0018] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention. The drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0019] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the aspects described may be practiced without these specific details.

[0020] In the field of landslide risk assessment, the evaluation of susceptibility and hazard is usually a relatively independent process. Generally speaking, landslide susceptibility mainly measures the potential for landslides contained in the geological environment itself, focusing on the impact of internal geological environmental factors such as geological structure, topography, and rock and soil structure on landslide breeding. It describes the vulnerability or potential instability of a region to landslide disasters under relatively static geological conditions. In other words, assuming that the external triggers are the same, areas with higher susceptibility are more likely to experience landslides. On the other hand, hazard assessment is based on susceptibility, and further superimposed or considered dynamic triggering factors, such as rainfall and surface deformation, to more comprehensively characterize the actual possibility of landslide disasters. These dynamic triggers will significantly increase the possibility of landslides in a certain period of time or under specific conditions. In this way, a landslide hazard assessment that is closer to the actual situation can be formed. Therefore, hazard assessment not only includes the inherent vulnerability of the geological environment, that is, susceptibility, but also incorporates the distribution and intensity of external triggering factors in time and space, and ultimately characterizes the probability or possibility of landslide disasters in the current or future period.

[0021] The existing technical solutions for landslide hazard assessment tend to follow the following model: First, conduct a susceptibility assessment: Combine static disaster-pregnant factors such as topography and lithology, and obtain susceptibility results through model learning and evaluation to characterize the vulnerability of the geological environment. The susceptibility assessment is independent of the dynamic induction process of landslide events and focuses on the stability of internal geological conditions. After obtaining the susceptibility results, conduct a comprehensive hazard assessment: Combine the susceptibility assessment results with dynamic disaster-inducing factors, and use the susceptibility results as factors for hazard assessment to form a final assessment of landslide hazard. This process comprehensively considers the vulnerability of the geological environment and the influence of external inducing factors, making the hazard assessment more comprehensive and accurate. However, in practical applications, there are still two outstanding problems. First, there is uncertainty in the sampling of non-landslide samples. When an area where "no landslide has occurred" is used as a negative sample, it is often impossible to determine whether the area is truly in a low-risk state. Some high-risk plots that have not yet triggered disasters are also regarded as negative samples, resulting in deviations in the susceptibility and hazard assessment results. The second is the lack of true value in susceptibility assessment. The fragility of the geological environment is usually calibrated by "whether a landslide actually occurs", but the occurrence of landslides is also affected by external dynamic disaster-inducing factors such as rainfall, and it does not necessarily reflect the potential instability of the area under static conditions. In other words, the "true value" of the susceptibility sample becomes blurred due to external inducements, and it is difficult to accurately distinguish the fragility of the geological environment itself from the triggering effect of external disaster-inducing factors. If the above problems are not solved, the accuracy and applicability of landslide hazard assessment will be significantly affected.

[0022] In summary, the current landslide hazard assessment methods lack an overall solution that can simultaneously deal with the missing true value of susceptibility and the uncertainty of non-landslide sample sampling.

[0023] An embodiment of the present invention provides a method for coupling assessment of landslide susceptibility and hazard, which can be applied to landslide hazard assessment in geological disaster prevention and control scenarios.

[0024] See also Figure 1 , is a flow chart of a method for coupling landslide susceptibility and hazard assessment provided by an embodiment of the present invention. Figure 1 As shown, the method mainly comprises the following steps: Step 1, extracting the landslide hazard factor characteristics based on the multi-source data of the target area and preprocessing them to obtain raster feature data; In specific implementation, the process of extracting the landslide hazard factor characteristics and preprocessing them according to the multi-source data of the target area to obtain the grid feature data is as follows: 1.1 Extraction and processing of landslide hazard factor characteristics In order to evaluate the susceptibility and hazard of landslide disasters, the present invention preprocesses and extracts features from multi-source data in the study area. According to the update cycle of data features, the features are roughly divided into two categories: static features and dynamic features; static features are mainly used to evaluate the susceptibility of landslide disasters, while dynamic features and a few static features, such as the annual average deformation rate, are mainly used for hazard evaluation.

[0025] (1) Static factor characteristics The static feature set is , these characteristics remain relatively stable over a long time scale and can be used to measure the vulnerability of the regional geological environment. In the present invention, the static characteristics mainly include: ① geological structural elements: distance from the fault, lithology type and soil type, which are formed by vectorization or rasterization to form a data layer that can be superimposed in a unified coordinate system; ② topographic features: including elevation, slope, slope aspect, plane curvature, profile curvature, extracted based on the digital elevation model (DEM); ③ deformation characteristics: based on the InSAR deformation monitoring of historical years, the annual average deformation in the region can be calculated. The present invention regards deformation characteristics as dangerous characteristic factors. Although they have certain time significance, they can also be regarded as relatively static characteristics in dangerousness assessment; ④ other factors such as human activities: according to the road vector data, land use type, optical remote sensing image and other data in the study area, the characteristic factors such as road network density, land use type, NDVI, etc. are extracted.

[0026] (2) Dynamic factor characteristics The dynamic factor characteristics of landslide hazards refer to the external disaster-inducing factors that change significantly with time t, denoted as , which have a more direct short-term impact on the occurrence of landslide disasters. The dynamic factors in the present invention mainly refer to rainfall-related factors, including the cumulative rainfall of the day, the cumulative rainfall of the past seven days, and the maximum rainfall intensity of the day, which are obtained by interpolation of meteorological station observation data products.

[0027] 1.2 Unification of spatial reference and resolution Since geological, remote sensing, meteorological and deformation data often come from different coordinate systems and resolutions, they need to be processed to ensure the consistency and accuracy of subsequent analysis. The first is the unification of spatial reference: all vector and raster data are projected to the same coordinate system EPSG:3857, and at the same time, the coordinate values ​​recorded in the landslide catalog event are converted to XY in the corresponding coordinate system. The second is the unification of raster resolution: for data with inconsistent resolutions, a unified grid size is selected , high-resolution data can be downsampled, and low-resolution data can be upsampled by interpolation to match the target resolution. Finally, invalid and outlier processing: interpolation or moderate filling can be used for missing data areas; extreme outliers can be removed or limited.

[0028] Step 2, divide the grid feature data into intervals and calculate the landslide event frequency in each interval; In specific implementation, the process of calculating the frequency of landslide events within the disaster-prone characteristic interval is as follows: 2.1 Classification of characteristic intervals of landslide hazard factors For the extracted grid feature data in the study area, the interval processing is performed according to the value type. Continuous features such as slope and elevation are classified using the natural breakpoint method. This method obtains relatively reasonable segmentation points by minimizing the variance within the group and maximizing the variance between groups to form several intervals. For discrete features such as land use type and soil type, their possible values ​​are considered as discrete levels, and no continuous division is required. After processing, each feature f has a set of non-overlapping intervals in the study area. ,in is the number of intervals of feature f.

[0029] 2.2 Calculation of landslide event frequency within the characteristic interval After the interval division is completed, in order to measure the susceptibility of landslide events in each interval, the present invention introduces "interval landslide event frequency" as a quantitative indicator, denoted as , defined as shown in formula (1): ; in, Indicates that in the interval The number of pixels where landslides occurred is counted; is the total number of pixels in the whole area whose value of feature f is in interval k; is the total number of positive samples; is the total number of pixels in the entire area. The larger the Within the interval, landslide events are relatively more concentrated and more likely to occur; if the value is small, it indicates that landslide events in the interval are inactive or less.

[0030] Step 3, calculating the local Moran index of the accumulated rainfall in a preset time interval according to the rainfall grid data in the grid feature data to identify the abnormal area in the rainfall spatial distribution and obtain the rainfall anomaly factor; In specific implementation, after extracting the basic rainfall grid data in step 1, the present invention identifies abnormal areas in the spatial distribution of rainfall by calculating the local Moran index of the seven-day cumulative rainfall, and determines the degree of abnormality through statistical tests. The main process is as follows: 3.1 Calculation of local Moran index The local Moran index is used to measure the attribute correlation between unit i and its spatial neighboring units and is often used in clustering and outlier analysis. represents the historical seven-day rainfall of the ith spatial unit in the target period, is the overall mean, is the element of the adjacency weight matrix, which is used to reflect the adjacency relationship between spatial units i and j. Then the local Moran index The definition of is shown in formula (2): ; in, is the normalized term of the global variance, and its calculation method is shown in formula (3), where n represents the total number of spatial units.

[0031] ; when When With surrounding If both are positive or negative, it indicates high-high aggregation (HH) or low-low aggregation (LL), reflecting that pixel i and its neighbors are similar in value distribution and are both high or low; When , it means that pixel i has opposite characteristics to its neighborhood, such as high-low aggregation (HL) or low-high aggregation (LH), indicating potential outliers or mutations; The larger the value, the more significant the degree to which i and its neighbors have the same high value or the same low value in rainfall distribution; The more negative the value, the greater the contrast with the neighborhood.

[0032] 3.2 Statistical significance test After calculating After that, it is necessary to conduct a statistical test on the calculated results. For each unit i, its corresponding Z score can be estimated to express the significance level. The calculation method is shown in formula (4): ; in and are the expectation and variance of the local Moran index under the assumption of random distribution. Exceeding a certain pre-set threshold When , it can be determined that there is statistically significant spatial aggregation or outlier.

[0033] 3.3 Generation of rainfall anomaly feature grid The present invention focuses on the spatial anomaly of rainfall, especially the high-value clustering areas and the high-low clustering areas with strong contrast with the surrounding units are more likely to indicate abnormally concentrated or extreme rainfall. Therefore, the present invention classifies HH and HL types, and The units with significant rainfall patterns are regarded as rainfall anomaly units, and the other types or insignificant units are regarded as rainfall non-anomalous units.

[0034] For each pixel i that has been determined to be rainfall anomaly, its Z score can be directly used as a measure of the degree of anomaly. The larger the value of the anomaly, the higher the degree of anomaly of the unit’s rainfall characteristic value. The corresponding rainfall anomaly index calculation method is shown in formula (5).

[0035] ; Step 4: According to the frequency of landslide events and rainfall anomaly factors, a susceptibility-hazard coupling evaluation based on Bayesian optimization is performed to obtain the optimal weights of disaster-prone characteristics and a hazard evaluation model; In specific implementation, the present invention has completed the characteristic interval division and the calculation of the interval landslide event probability in step 2. In order to further improve the prediction accuracy of the hazard model, this step uses Bayesian optimization to link and adjust the susceptibility weight and the constraint strategy of negative samples, and finally obtains the optimal parameters through iteration, thereby obtaining the optimal susceptibility-hazard coupling evaluation result.

[0036] 4.1 Generation of susceptibility factors After calculating the landslide event frequency corresponding to the disaster-pregnant characteristic value interval in step 2.2, according to the set initial weight vector to combine the frequencies of different features together, where F represents the total number of static disaster-prone features, Represents the weight coefficient of feature f in the overall susceptibility calculation. On the other hand, the scores of each feature interval and the corresponding feature weights are combined. , we can get the vulnerability index that represents the vulnerability of the geological environment , as shown in formula (6): ; in, represents the weight of feature f, Indicate the pixel The partition index of feature f, that is, which value interval or category it belongs to; is the probability of landslide event in this interval.

[0037] 4.2 Hazard Assessment Model Training (1) Dangerous Negative Sample Constrained Sampling In the hazard assessment model, negative samples usually need to be selected from units where landslides have never occurred. If the negative samples are not selected properly, they may include areas with high landslide risks but have not yet triggered disasters, thereby interfering with model learning. To this end, the present invention constrains the sampling of dangerous negative samples by using rainfall anomaly characteristic factors and susceptibility factors, and only includes the pixel in the optional negative sample set when the following conditions are met: ① It is outside the rainfall anomaly range, that is, the rainfall characteristic anomaly calculated according to step 3 is 0; ② The susceptibility index is low, that is, the internal geological environment is relatively stable, and the susceptibility index is less than the set threshold. .

[0038] Before determining the spatial location of the negative sample points, first randomly generate the date , and then screen according to the above conditions to obtain the negative sample candidate set that meets the conditions, as shown in formula (7).

[0039] ; in Indicates on date Randomly select the negative sample candidate set that meets the conditions, is the rainfall anomaly, is the susceptibility index. Finally, the required number of negative samples are randomly selected from the candidate set, denoted as , and spliced ​​with the positive sample points to obtain a set of dangerous sample points. Then, according to the set number of negative sample sampling, a corresponding number of negative samples are randomly sampled from the candidate set. Finally, it is spliced ​​with the positive samples to obtain a set of landslide dangerous sample points.

[0040] (2) Acquisition of characteristic values ​​of dangerous samples Based on the dynamic features extracted in step 1, the rainfall anomaly factor obtained in step 3, and the susceptibility results obtained in step 4.1, a dangerous feature set is constructed. , mainly including static susceptibility factor and deformation factor, dynamic rainfall factor and abnormal factor, as shown in formula (8): ; in, It is a susceptibility factor; is the deformation factor; is the accumulated rainfall for the day; It is the historical seven-day cumulative rainfall; The maximum rainfall intensity of the day; is the rainfall anomaly factor.

[0041] According to the coordinates in the sample point set and date , index and obtain the corresponding feature value from the feature grid file. This process can be expressed by formula (9): ; (3) Hazard assessment model training and accuracy calculation After constructing the dangerous samples, they are divided according to a fixed ratio of training set to test set. Assume that the sample point set is ,in is the sample label, landslide is a positive sample, and no landslide is a negative sample. The training set and test set are randomly divided, with the training set accounting for 70% and the test set accounting for 30%. The division process can be expressed by formula (10-11): ; The random forest model is selected for training. After the training is completed, the accuracy indicators of the model on the test set are calculated, including accuracy, precision, recall and F1 score, as shown in formula (12-15): ; Among them, TP is the number of true positive examples, TN is the number of true negative examples, FP is the number of false positive examples, and FN is the number of false negative examples.

[0042] 4.3 Bayesian Optimization for Optimal Parameters (1) Optimization objectives and hyperparameter definitions In the present invention, the parameters adjusted by the Bayesian optimization method mainly include two parts: static feature weight parameters in susceptibility evaluation and the susceptibility constraint threshold in dangerous negative sample sampling The optimization objective function is is the sum of the accuracy of the risk assessment model on the test set and the F1 score, as shown in formula (16). as the goal.

[0043] ; (2) Bayesian optimization solution process Initial sampling: In the search space, select several groups of initial candidate points, perform all operations from 4.1 to 4.2 for each group, including generating susceptibility factors, constraining negative samples, training hazard models, calculating ACC and F1, and obtain the corresponding ; Build or update the proxy model: Use the Gaussian process as a proxy model to Fit the sample relationship to obtain the posterior distribution; select new candidate points: determine the next batch of most likely to improve based on the acquisition function of Parameter combination and re-evaluation; iterate until convergence: when the agent model converges or reaches the maximum number of iterations, the optimal solution is output , that is, the parameter configuration that can maximize the ACC+F1 comprehensive index of hazard assessment.

[0044] After completing the Bayesian optimization, we can get the optimal susceptibility factor: The weighted landslide frequency in each characteristic interval is generated ; Optimal Hazard Model: Combining Thresholds After negative sample constraint sampling, the risk assessment model is trained with random forest and achieves the highest accuracy on the test set.

[0045] Step 5: Based on the optimal weight of the disaster-pregnant feature, the real-time static disaster-pregnant feature interval index of each grid in the target area is input to obtain a susceptibility assessment result map, and the real-time feature vector of each grid is input into the hazard assessment model to obtain a hazard assessment result map.

[0046] In specific implementation, based on the optimal weight of the disaster-pregnant feature, the real-time static disaster-pregnant feature interval index of each grid in the target area can be input to obtain the susceptibility assessment result map. In addition, by inputting the feature vectors of all spatial units in the area into the trained hazard assessment model, the hazard assessment results of all spatial units in the area can be obtained.

[0047] The coupled evaluation method for landslide susceptibility and hazard provided in this embodiment performs constrained sampling of hazard negative samples by combining susceptibility and dynamic disaster-inducing factors, thereby avoiding the bias of directly using areas where landslides have not occurred as negative samples in traditional methods, improving the accuracy of negative sample selection, and effectively preventing the omission of potential high-risk areas, thereby significantly improving the stability and accuracy of the hazard evaluation model; the susceptibility-hazard coupling learning method based on Bayesian optimization can adaptively adjust the weights of susceptibility factors and negative sample constraints, optimize the landslide hazard evaluation process, maximize accuracy and F1 scores, and overcome the problem of insufficient fusion of susceptibility factors and dynamic disaster-inducing factors in traditional landslide hazard evaluation methods; the local Moran index is used to detect spatial anomalies of rainfall characteristics, which can effectively identify and quantify rainfall abnormal areas, improve the contribution of rainfall inducements to the hazard evaluation model, and make the hazard assessment results closer to the actual possibility of disasters.

[0048] The method of the present invention will be further described below in conjunction with a specific embodiment, using the landslide event catalog of Province A from 2022 to 2024, and the characteristic data of disaster factors and rainfall data over multiple periods to illustrate the specific implementation process of the present invention: (1) Data preprocessing A total of 456 landslide events that occurred in the past 2019-2024 were collected. First, the longitude and latitude coordinates recorded in the catalog were converted to coordinates in the EPSG:3857 coordinate system. For the existing data, all features were processed into raster files through data conversion, and then the spatial reference of all raster data was unified to the EPSG:3857 coordinate system, and the grid resolution was uniformly set to 300m×300m. The static feature set includes NDVI, elevation, slope, aspect, soil type, land use type, distance to water system, distance to road, distance to fault, human density, and annual average deformation rate. After extraction, 11 TIF raster files were obtained; after calculation, the dynamic feature set is mainly the daily rainfall, the historical seven-day cumulative rainfall, and the maximum rainfall intensity of the day. An independent raster layer is generated for each period, and the update cycle of dynamic features is one day.

[0049] (2) Calculation of landslide event frequency within the disaster-prone characteristic interval For continuous features such as slope and curvature, each feature is divided into five value intervals using the natural breakpoint method; for discrete features such as soil type and land use type, the original categories are retained. The number of landslide events in different feature intervals within five years is counted, and the frequency of landslide events in different feature intervals is counted according to formula (1).

[0050] (3) Dynamic rainfall feature spatial anomaly detection In the present invention, the feature of historical seven-day cumulative rainfall is used for anomaly detection. The historical seven-day cumulative rainfall grid of a certain day is input, and for each pixel in the area, the local Moran index is calculated according to formula (2), where the neighbor weight The value is taken using the 8-neighborhood mode, that is, when pixel i is within the 8-neighborhood of pixel j, , otherwise it is 0; its global mean and variance are obtained according to the rainfall characteristic data of that day. When calculating statistical significance, the significance threshold is set to 1.96, corresponding to a 95% confidence level. When , it is determined that the pixel has significant spatial aggregation or outliers. Finally, according to formula (5), the anomaly degree of all pixels in the region is assigned to obtain new rainfall anomaly characteristics. The anomaly detection results of the historical seven-day cumulative rainfall on September 13, 2023 are as follows: Figure 2 shown.

[0051] (4) Landslide susceptibility-hazard coupling assessment The landslide susceptibility-hazard coupling learning framework of the present invention is as follows Figure 3 As shown. It is regarded as an adjustable parameter, and its value range can be set to [0, 1]. The susceptibility index under the current weight combination is calculated by the formula When sampling dangerous negative samples, first assign a value to the negative sample date. In the present invention, the ratio of positive to negative samples is 1, so the negative sample date can be simply set to be a copy of the positive sample date. Then, for each date t existing in the negative sample, the candidate set for sampling needs to be filtered first: remove Pixels with normal rainfall are retained; ,like Negative samples are not included. Randomly select the number of points corresponding to date t from the remaining pixels without landslide after double filtering as negative samples. After the negative sample sampling is completed, it is spliced ​​with the positive samples composed of all landslide events to form the hazard model evaluation sample. Then, according to the coordinates of each sample point and date t, obtain the feature value of the corresponding pixel, and the feature set is shown in formula (8). Finally, assign the label column of the sample, with positive samples assigned to 1 and negative samples assigned to 0.

[0052] The samples were divided into training set and test set in a ratio of 7:3. The random forest model was selected, the number of trees was set to 100, the maximum tree depth was set to 4, the training set was input for fitting, the training model was obtained, and the accuracy index of the model on the test set was calculated. The objective function value was calculated according to formula (16), and the minimization of this value was used as the optimization goal. The optimal parameters were finally obtained through continuous iteration of Bayesian optimization, including the optimal weights of disaster-prone features. and the optimal susceptibility constraint threshold Through the best , input the static disaster-prone characteristic interval index of the pixels in the whole province, and obtain the susceptibility evaluation result map, such as Figure 4 shown.

[0053] For each pixel i, input its corresponding feature vector into the trained hazard assessment model, including static feature values ​​and dynamic feature values ​​of a certain period, and the landslide hazard index of the corresponding period can be obtained. When all feature space ranges correspond to the spatial positions of grid cells, the hazard index of all pixels is output to obtain the hazard assessment map of the entire region, such as Figure 5 shown.

[0054] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof.

[0055] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for coupling assessment of landslide susceptibility and hazard, characterized in that: include: Step 1, extracting the landslide hazard factor characteristics based on the multi-source data of the target area and preprocessing them to obtain raster feature data; Step 2, divide the grid feature data into intervals and calculate the landslide event frequency in each interval; Step 3, calculating the local Moran index of the accumulated rainfall in a preset time interval according to the rainfall grid data in the grid feature data to identify the abnormal area in the rainfall spatial distribution and obtain the rainfall anomaly factor; Step 4: According to the frequency of landslide events and rainfall anomaly factors, a susceptibility-hazard coupling evaluation based on Bayesian optimization is performed to obtain the optimal weights of disaster-prone characteristics and a hazard evaluation model; Step 5: Based on the optimal weight of the disaster-pregnant feature, the real-time static disaster-pregnant feature interval index of each grid in the target area is input to obtain a susceptibility assessment result map, and the real-time feature vector of each grid is input into the hazard assessment model to obtain a hazard assessment result map.

2. The method according to claim 1, characterized in that The step 1 specifically includes: Step 1.1, dividing the multi-source data of the target area into static factor features and dynamic factor features according to the data feature update cycle, wherein the static factor features include geological structure elements, topographic features, deformation features, and human activity factors, and the dynamic factor features are rainfall raster data; Step 1.2, unify the spatial reference and grid resolution of the static factor features and the dynamic factor features to obtain grid feature data.

3. The method according to claim 2, characterized in that The step 2 specifically includes: Step 2.1, perform interval processing according to the value type of each feature in the static factor feature, and obtain the interval set corresponding to each feature; Step 2.2, calculating the landslide event frequency of each interval in the interval set corresponding to each feature according to a preset formula, wherein the preset formula is: ； in, is the total number of pixels in the target area whose values ​​of feature f are in interval k, Indicates the number of pixels where landslides occurred. is the total number of positive samples, is the total number of pixels in the target area.

4. The method according to claim 3, characterized in that The step 3 specifically includes: Step 3.1, calculating the local Moran index corresponding to the rainfall of each pixel within a preset time interval based on the rainfall grid data; Step 3.2, statistically verify the local Moran index of all pixels according to the preset significance threshold to obtain the significance score; Step 3.3: Identify rainfall anomaly pixels based on the significance score and calculate their corresponding rainfall anomaly factors. ； in, Represents pixel The local Moran index, Represents pixel The significance score of represents the significant threshold, Indicates high-high aggregation areas between pixels and their neighbors. Represents areas of high-low aggregation of pixels and their neighbors.

5. The method according to claim 4, characterized in that The step 4 specifically includes: Step 4.1, based on the frequency of landslide events, the frequencies of different types of features are combined according to the set initial weight vector to obtain their corresponding weight coefficients, and the scores of each feature interval and the weight coefficients of the corresponding features are combined in each pixel to obtain the susceptibility factor that characterizes the fragility of the geological environment; Step 4.2, randomly generate the date of each pixel, then form a negative sample set with the pixels that meet the preset conditions, and combine it with the positive samples composed of all landslide events to form a hazard model evaluation sample, wherein the preset conditions are that the rainfall anomaly factor corresponding to the pixel is 0 and the susceptibility factor is less than the susceptibility threshold; Step 4.3, construct the dangerous feature set based on the dynamic factor characteristics, rainfall anomaly factors and susceptibility factors; Step 4.4, according to the coordinates and date of each sample data in the risk model evaluation sample, index and obtain the corresponding feature value from the risk feature set and divide it into a training set and a test set according to a preset ratio; Step 4.5, use the training set to train the random forest model and calculate the accuracy index of the trained model on the test set; Step 4.6, construct the objective function according to the accuracy and F1 score in the precision index, take minimizing the objective function value as the optimization target, solve the objective function through Bayesian optimization iteration, and obtain the optimal weight and optimal susceptibility threshold of the disaster-prone features; Step 4.7, take the optimal susceptibility threshold as the new susceptibility threshold and execute steps 4.2 to 4.5 again to obtain the risk assessment model.

6. The method according to claim 5, characterized in that The expression of the objective function is ； in, Indicates accuracy, represents the weight coefficient, represents the susceptibility threshold, Denotes the F1 score.

Images