Landslide analysis method based on risk factor sensitivity

Abstract

The landslide analysis method based on the sensitivity of risk factors solves the problem of the existing difficulty in landslide prevention analysis and belongs to the field of disaster prevention technology. The present invention includes: collecting data of landslide disaster points and various landslide risk factors; dividing classification feature intervals for the raster data of each landslide risk factor to obtain a variety of different classification schemes, and each classification scheme includes multiple classification feature intervals; constructing a classification model for each landslide risk factor by using the deviation sum index and the fitting deviation sum index, and selecting the classification scheme when the classification model value is closest to 1; optimizing the classification feature intervals of the selected classification scheme; calculating the sensitivity DC of the landslide risk factor in the optimized classification feature interval; if DC>0, it means that the landslide is likely to occur in the corresponding classification feature interval; if DC<0, it means that the landslide is not likely to occur in the corresponding classification feature interval; if DC≈0, it means that it cannot be determined whether the landslide is likely to occur in the corresponding classification feature interval.

Description

Technical Field

[0001] The present invention relates to a landslide analysis method based on the sensitivity of risk factors, belonging to the technical field of disaster prevention. Background Art

[0002] The system composition of landslide risk factors is an important part of susceptibility evaluation. However, most of the existing studies only simply use some basic data that can be obtained for extraction and analysis, and do not consider much whether the extracted factors are relevant, which largely causes data redundancy, increases the workload of related research, and has great blindness. Therefore, how to reasonably select landslide risk factors for landslide susceptibility evaluation has become an important entry point for current research.

[0003] Scientific selection of landslide risk factors necessarily leads to the assessment method for the sensitivity of landslide risk factors, and the coefficient of determination method is an effective method for quantifying the sensitivity of landslide risk factors. By judging the positive and negative of the coefficient of determination, the sensitivity of risk factors can be intuitively determined.

[0004] The occurrence of a landslide is a complex process and is affected by many factors. Therefore, it is very difficult to accurately analyze its risk factors. Summary of the Invention

[0005] Aiming at the problem of difficult prevention and analysis of landslides in the prior art, the present invention provides a landslide analysis method based on the sensitivity of risk factors.

[0006] A landslide analysis method based on the sensitivity of risk factors according to the present invention, the method includes:

[0007] S1. Collect data of landslide disaster points and various landslide risk factors;

[0008] S2. Discretize the raster data of each landslide risk factor, divide the classification feature intervals, each division method corresponds to a classification scheme, obtain a variety of different classification schemes, and each classification scheme includes multiple classification feature intervals;

[0009] S3. Calculate the deviation sum index SM of each landslide risk factor:

[0010]

[0011] where, x i is the i-th raster value of a certain risk factor, i = 1, 2,..., w, w represents the number of rasters of the certain risk factor, is the mean value of all raster values of this risk factor;

[0012] S4. Calculate the fitting deviation and the index NSM for each classification scheme of each landslide risk factor: NSM = NSM1 + NSM2 + … + NSM m ;

[0013] where NSM c represents the fitting deviation and the index for the c-th classification feature interval, c = 1, 2, …, m, where m represents the number of classification feature intervals, x cg is the g-th grid value of the risk factor within the c-th classification feature interval, g = 1, 2, …, n, where n represents the number of grids in the c-th classification feature interval, is the mean value of all grid values in the c-th classification feature interval;

[0014] S5. Construct a classification model for each landslide risk factor using the deviation sum index and the fitting deviation sum index. The classification model is:

[0015]

[0016] where GF represents the effect of each classification scheme of the corresponding risk factor;

[0017] S6. Select the classification scheme corresponding to when the GF value is closest to 1;

[0018] S7. Re-divide the classification feature intervals of the selected classification scheme: Calculate the probability ratio index PR. Use the endpoint values of each classification feature interval as the abscissa and the PR value as the ordinate to plot a curve, and use the inflection point of the curve as the endpoint value of the classification feature interval, thereby obtaining the optimized classification feature interval:

[0019] where A is the number of landslide disaster points of a certain risk factor in the classification feature interval, A0 is the total number of landslide disaster points, K is the area of a certain risk factor in the corresponding classification feature interval, and K0 is the total area of the risk factor;

[0020] S8. Calculate the sensitivity DC of the landslide risk factor in the optimized classification feature interval:

[0021]

[0022] where NN a is the ratio of the number of landslide disaster points to the area of a certain risk factor in the classification feature interval, and NN s is the ratio of the total number of landslide disaster points to the total area;

[0023] S9. Make a discrimination based on DC:

[0024] If DC > 0, it means that a landslide is likely to occur in the corresponding classification feature interval;

[0025] If DC < 0, it indicates that landslides are not likely to occur in the corresponding classification feature interval;

[0026] If DC ≈ 0, it indicates that it is impossible to determine whether landslides are likely to occur in the corresponding classification feature interval.

[0027] Preferably, the step S1 further includes:

[0028] Calculating the correlation index of each landslide risk factor collected

[0029]

[0030] is the covariance between the x-th risk factor and the y-th risk factor, where x i is the i-th grid value of the x-th risk factor, and y j is the j-th grid value of the y-th risk factor; is the overall standard deviation of the x-th risk factor, is the overall mean of the y-th risk factor; if it is determined that the correlation is strong. If a certain risk factor has a strong correlation with more than two risk factors at the same time, this risk factor is removed.

[0031] Preferably, w represents the total number of grids.

[0032] Preferably,

[0033] Preferably, the landslide risk factors include factors in terms of topography and geomorphology, geological structure, human activities, and meteorology and hydrology.

[0034] Advantages of the present invention: The present invention constructs an analysis model for determining the sensitivity of landslide risk factors, which is of great significance for predicting landslide susceptibility. The present invention quantifies each landslide risk factor, and makes a reasonable judgment on the sensitivity of the landslide risk factors in this area through the positive and negative of the DC value; at the same time, the present invention comprehensively considers the mutual influence between the risk factors, and uses a more practical theory to guide the geological disaster planning, and has the advantages of being simple and easy to implement, and can be directly used for the related research on geological disaster susceptibility. Brief Description of the Drawings

[0035] Figure 1 is the flow chart of the present invention;

[0036] Figure 2 is the distribution histogram of the disaster points in the implementation area in each classification feature interval of elevation. Detailed Embodiments

[0037] 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 only a part of the embodiments of the present invention, rather than all the embodiments. 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.

[0038] It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments may be combined with each other.

[0039] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but it is not limited to the present invention.

[0040] The landslide analysis method based on the sensitivity of risk factors in this embodiment includes:

[0041] Step 1: Collect data of landslide disaster points and various landslide risk factors;

[0042] Step 2: The types of risk factor data collected include discrete type and continuous type. To unify the data form, discretize the continuous raster data of each landslide risk factor, divide the classification feature intervals, and each division method corresponds to a classification scheme, obtaining multiple different classification schemes. Each classification scheme includes multiple classification feature intervals; for example: the raster data of a certain landslide risk factor is 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, and the classification feature intervals are divided as follows: 1, 2, 3 represent one classification feature interval, 4, 5, 6 represent one classification feature interval, 7, 8, 9, 10 represent one classification feature interval; taking the classification feature interval of 1, 2, 3 as an example, 1 is the first value in this classification feature interval, and 2 is the mean value of all the values in this classification feature interval. The above is just one classification scheme, and similarly, it may also be the classification scheme of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10;

[0043] Step 3: Calculate the deviation sum index SM of each landslide risk factor:

[0044]

[0045] where x i is the i-th raster value of a certain risk factor, i = 1, 2,..., w, and w represents the number of rasters of the certain risk factor, is the mean value of all the raster values of this risk factor;

[0046] Step 4: To facilitate the construction of a classification model to verify the classification effect later, calculate the fitting deviation sum index NSM of each classification scheme of each landslide risk factor:

[0047] NSM = NSM1 + NSM2 + … + NSM m ;

[0048] Among them, NSM c represents the fitting deviation sum and exponent of the c-th classification feature interval, where c = 1, 2, …, m, and m represents the number of classification feature intervals. x cg is the g-th grid value of the risk factor within the c-th classification feature interval, where g = 1, 2, …, n, and n represents the number of grids in the c-th classification feature interval. is the mean value of all grid values in the c-th classification feature interval;

[0049] Step Five: Construct a classification model for each landslide risk factor using the deviation sum exponent and fitting deviation sum exponent, and comprehensively evaluate the classification effect of each classification scheme. The classification model is:

[0050]

[0051] Among them, GF represents the effect of each classification scheme of the corresponding risk factor; the GF value ranges from 0 to 1, 1 indicates an excellent classification effect, 0 indicates a very poor classification effect, and the closer the GF value is to 1, the better the classification effect.

[0052] Step Six: Select the classification scheme corresponding to when the GF value is closest to 1;

[0053] Step Seven: To further optimize the above classification scheme and make it more in line with the characteristics of the data itself, further merge the above classification scheme in combination with the probability ratio exponent, and re-divide the classification feature intervals of the selected classification scheme: Calculate the probability ratio exponent PR, use the endpoint values of each classification feature interval as the abscissa and the PR value as the ordinate to draw a curve, and use the inflection point of the curve as the endpoint value of the classification feature interval, so as to obtain the optimized classification feature interval;

[0054]

[0055] Among them, A is the number of landslide disaster points of a certain risk factor in the classification feature interval, A0 is the total number of landslide disaster points, K is the area of a certain risk factor in the corresponding classification feature interval, and K0 is the total area of the risk factor;

[0056] Step Eight: Calculate the sensitivity DC of the landslide risk factor in the optimized classification feature interval:

[0057]

[0058] Among them, NN a is the ratio of the number of landslide disaster points of a certain risk factor in the classification feature interval to the area of this classification feature interval, NN sIt is the ratio of the total number of landslide disaster points to the total area;

[0059] Step Nine: Make a judgment according to DC:

[0060] If DC > 0, it means that landslides are likely to occur in the corresponding classification feature interval;

[0061] If DC < 0, it means that landslides are not likely to occur in the corresponding classification feature interval;

[0062] If DC ≈ 0, it means that it cannot be determined whether landslides are likely to occur in the corresponding classification feature interval.

[0063] In Step One of this embodiment, it further includes:

[0064] To avoid redundancy among risk factors, corresponding preprocessing is performed on the risk factors, and the factor correlation index is used to judge the correlation between factors. The factor correlation index is:

[0065] Calculate the correlation index of each landslide risk factor collected

[0066]

[0067] is the covariance between the x-th risk factor and the y-th risk factor, x i is the i-th grid value of the x-th risk factor, y j is the j-th grid value of the y-th risk factor; is the overall standard deviation of the x-th risk factor, is the overall mean of the y-th risk factor; If is determined to have a strong correlation when, if a certain risk factor has a strong correlation with two or more risk factors at the same time, this risk factor is removed.

[0068] w represents the total number of grids.

[0069]

[0070] The landslide risk factors in this embodiment include factors in terms of topography and geomorphology, geological structure, human activities, and meteorology and hydrology. Specific embodiment:

[0072] To achieve the above prediction and planning goals, the present invention provides a method for calculating the sensitivity of landslide risk factors. By considering various influencing factors simultaneously and using the determination coefficient method to judge the sensitivity of landslide risk factors. The following is the specific calculation scheme adopted by the present invention, which will be further elaborated in the following steps:

[0073] 1. Data acquisition:

[0074] Select a certain county in a certain city in a certain province as the example area, and adopt Example Area 1: 650,000 geological hazard distribution map, 1:10,000 county-level second survey data, GF-1 remote sensing image, 30m resolution digital elevation data of GDEM DEM from the Geospatial Data Cloud, as well as the structural outline map of the example area and rainfall station data of the Hubei Meteorological Bureau.

[0075] 2. To reduce data redundancy, select the risk factor data input formula

[0076] Calculate the factor correlation and eliminate the factors with larger correlations.

[0077] If the correlation between two risk factors is strong, then choosing one of the factors can represent such influence characteristics. The specific calculation results of the factor correlations of each risk factor in this example are shown in Table 1. Based on the comprehensive selection rules, a total of 9 landslide risk factors, namely elevation, slope direction, distance to the river, distance to the road, profile curvature, NDVI, slope structure, land category, and rock and soil body category, are finally selected in this example.

[0078] Table 1 Correlation coefficients of landslide risk factors in the example area

[0079]

[0080] In Table 1, X1: annual average rainfall; X2: terrain humidity index; X3: slope structure; X4: distance to the road; X5: distance to the fault; X6: elevation; X7: distance to the river; X8: NDVI; X9: plane curvature; X 10 : slope; X 11 : profile curvature; X 12 : slope direction; X 13 : land category; X 14 : rock and soil body category;

[0081] 3. The collected risk factor data includes discrete type and continuous type. Discretize the continuous type data. According to the clustering and iteration principle, list a classification scheme with a certain preset number of groups for the continuous type data for preliminary classification, and input the corresponding data of various risk factors in the example area into the deviation sum of squares formula and the fitting deviation sum of squares formula;

[0082] 4. Substitute the deviation sum of squares and fitting deviation sum of squares of each calculated risk factor into the classification model, calculate the classification model values corresponding to each classification scheme of each risk factor, and evaluate the classification effect to determine the selected classification scheme.

[0083] GF represents a verification model for the classification effect after the grid values of the corresponding risk factors are classified by each division combination. The GF value ranges from 0 to 1. A value of 1 indicates an excellent classification effect, and a value of 0 indicates a very poor classification effect. The closer the GF value is to 1, the better the classification effect. Here, taking the elevation of the example area as an example of a risk factor for calculation, by comparing the calculation results of the classification models of each classification scheme, the division combination with the calculation value of the classification model closest to 1 is selected as the final classification result of this risk factor. The specific classification results are as Figure 2 shown. The number of disaster points is the largest in the elevation range of 100 - 120, and the smallest in the range of 1300 - 1400, which conforms to the basic characteristics of disaster occurrence in this area.

[0084] 5. To further optimize the above classification scheme and make it more in line with the characteristics of the data itself, it is combined with the probability ratio index to obtain an optimized classification feature interval. Substitute the landslide disaster point data in each classification feature interval determined by the optimal classification scheme for the continuous data of each risk factor into the probability ratio method formula, draw a curve based on the probability ratio calculation results, and reasonably combine the intervals of the risk factors according to the inflection points and overall trends of the curve, so as to obtain an optimized classification feature interval, realize the interval division of continuous factors, and achieve a noise reduction effect. The finally obtained classification feature intervals are shown in Table 2.

[0085] Table 2 Discretization results of factor intervals in the example area

[0086]

[0087] In Table 2, land category: 1. Cultivated land 2. Orchard land 3. Urban, rural and industrial and mining land 3. Forest land 4. Water area and water conservancy facilities land 5. Grassland 6. Other land 7. Transportation land; Rock and soil body category: 1. Loose and soft rock and soil type 2. Clastic rock type 3. Carbonate rock type 4. Magmatic rock and metamorphic rock type; Slope structure: 1. Horizontal slope 2. Downslope 3. Oblique downslope 4. Cross slope 5. Oblique upslope 6. Upslope;

[0088] 6. According to the characteristic intervals obtained from the above calculations, re-divide and count the distribution of landslide disaster points in each optimized classification feature interval and the area of each optimized classification feature interval, and substitute the statistical results into the decisive coefficient DC formula to calculate the sensitivity values of various risk factors.

[0089] By calculation, the sensitivity values of each risk factor in each interval are obtained, and the results are shown in Table 3: Landslides are likely to occur within the elevation range of 0 - 400m, slope direction range of 90 - 180°, NDVI range of -0.3 - 0.2, profile curvature range of -2 - 2, distance to road range of 0 - 100m, distance to river range of 4800 - 6800m, land categories including cultivated land, garden land, urban villages, industrial and mining land, and transportation land, and geotechnical categories including loose and soft rock types and clastic rock types, as well as in the areas of downslope and cross-slope.

[0090] Table 3

[0091]

[0092] In Table 3, land categories: 1. Cultivated land 2. Garden land 3. Urban villages and industrial and mining land 3. Forest land 4. Water areas and water conservancy facilities land 5. Grassland 6. Other land 7. Transportation land; geotechnical categories: 1. Loose and soft rock types 2. Clastic rock types 3. Carbonate rock types 4. Magmatic and metamorphic rock types; slope structures: 1. Horizontal slope 2. Downslope 3. Downslope and cross-slope 4. Cross-slope 5. Upslope and cross-slope 6. Upslope.

[0093] After investigation and comparison, the calculation results of this embodiment are relatively consistent with the distribution characteristics of historical landslide disaster points obtained statistically in each interval. The DC values of each factor in the high - susceptibility interval of landslides are also greater than 0, indicating that it is reasonable and reliable for us to use the DC method to conduct sensitivity analysis on the attribute intervals of each factor.

Claims

1. A landslide analysis method based on the sensitivity of risk factors, characterized in that The method includes: S1. Collect data of landslide disaster points and various landslide risk factors; S2. Discretize the raster data of each landslide risk factor, divide the classification feature intervals, and each division method corresponds to a classification scheme, obtaining multiple different classification schemes, and each classification scheme includes multiple classification feature intervals; S3. Calculate the deviation sum index SM of each landslide risk factor; where x i is the i-th grid value of a certain risk factor, i = 1, 2, …, w, and w represents the number of grids of the certain risk factor, is the mean value of all grid values of the risk factor; S4. Calculate the fitting deviation sum index NSM of each landslide risk factor in each classification scheme; NSM = NSM1 + NSM2 + … + NSM m ; Among them, NSM c represents the fitting deviation and exponent of the c-th classification feature interval, where c = 1, 2, …, m, and m represents the number of classification feature intervals. x cg is the g-th grid value of the risk factor within the c-th classification feature interval, where g = 1, 2, …, n, and n represents the number of grids in the c-th classification feature interval. is the mean of all grid values in the c-th classification feature interval. S5. Construct a classification model for each landslide risk factor by using the deviation sum index and the fitting deviation sum index. The classification model is: where GF represents the effect of each classification scheme of the corresponding risk factor; S6. Select the classification scheme corresponding to the case when the GF value is closest to 1; S7. Re-divide the classification feature intervals of the selected classification scheme: Calculate the probability ratio index PR, draw a curve with the endpoint values of each classification feature interval as the abscissa and the PR value as the ordinate, and use the inflection point of the curve as the endpoint value of the classification feature interval, so as to obtain the optimized classification feature interval: where A is the number of landslide disaster points of a certain risk factor in the classification feature interval, A0 is the total number of landslide disaster points, K is the area of a certain risk factor in the corresponding classification feature interval, and K0 is the total area of the risk factor; S8. Calculate the sensitivity DC of the landslide risk factor in the optimized classification feature interval; Among them, NN a is the ratio of the number of landslide disaster points of a certain risk factor in the classification feature interval to the area of this classification feature interval, NN s is the ratio of the total number of landslide disaster points to the total area; S9. Make a judgment according to DC: If DC>0, it means that landslides are likely to occur in the corresponding classification feature interval; If DC<0, it means that landslides are not likely to occur in the corresponding classification feature interval; If DC = 0, it means that it cannot be determined whether landslides are likely to occur in the corresponding classification feature interval.

2. The landslide analysis method based on risk factor sensitivity according to claim 1, characterized in that The S1 further includes: Calculate the correlation index of each landslide risk factor collected is the covariance between the x-th risk factor and the y-th risk factor, where x i is the i-th grid value of the x-th risk factor, and y j is the j-th grid value of the y-th risk factor; is the overall standard deviation of the x-th risk factor, is the overall mean of the y-th risk factor; if it is defined as having a strong correlation. If a certain risk factor has a strong correlation with more than two risk factors at the same time, this risk factor is removed.

3. The landslide analysis method based on risk factor sensitivity according to claim 2, wherein w represents the total number of grids.

4. The analysis method for calculating the sensitivity of landslide risk factors according to claim 3, wherein 5. The landslide analysis method based on risk factor sensitivity according to claim 4, wherein The landslide risk factors include factors in terms of topography and geomorphology, geological structure, human activities, and meteorology and hydrology.

Images