Meteorological early warning method for highway geological disasters under rainfall stress

Abstract

The invention discloses a rainfall-stress highway geological disaster meteorological early warning method, relates to the technical field of geological disaster meteorological early warning, and provides a rainfall-stress highway geological disaster meteorological early warning method which comprises the following steps: by taking a highway geological disaster which has occurred in a province as a distribution sample, analyzing meteorological factors such as rainfall intensity in 120 hours, 96 hours, 72 hours, 48 hours, 24 hours and the maximum hour before occurrence of the geological disaster; and synthesizing underlying surface factors such as elevation, fault, land utilization and vegetation coverage indexes to obtain key environment variables and threshold values of geological disasters. And in combination with rainfall grid forecast, high-precision geological disaster meteorological risk early warning is obtained. The geological disaster meteorological risk early warning method is simple and convenient in environment variable acquisition, simple in content and simple in operation, has a manual revision function, continuously adjusts the rainfall threshold along with the actual occurrence of subsequent geological disasters, and continuously improves the reliability and practicability of geological disaster meteorological risk early warning. Alarm and management suggestions can be effectively given out, and loss is reduced.

Description

Technical Field

[0001] The invention relates to the technical field of meteorological early warning of geological disasters, and in particular to a meteorological early warning method for highway geological disasters under precipitation stress. Background Art

[0002] Geological disasters are one of the most frequent and dangerous natural disasters. Geological disasters are largely caused by precipitation. With global climate change, extreme weather and climate events are becoming more frequent. Therefore, increasing meteorological monitoring, improving the accuracy of meteorological warnings, and providing refined meteorological services are the "first line of defense" for disaster prevention and mitigation.

[0003] At present, extensive research has been conducted at home and abroad on geological disasters, and a deeper understanding of the geological conditions and inducing factors that affect geological disasters has been achieved. Some meteorological early warning methods for geological disasters have also been proposed.

[0004] Chinese patent document CN 114093133 A discloses a regional geological disaster meteorological early warning method, which includes establishing a regional geological disaster meteorological early warning model based on a digital elevation model DEM, geological disaster prone zoning, geological disaster meteorological risk level, etc. It can automatically realize regional geological disaster meteorological early warning based on DEM segmentation technology and ArcGIS software technology, combined with regional geological disaster risk level, while making full use of existing regional geological disaster prone zoning and daily forecast rainfall and 5-day forecast rainfall.

[0005] Chinese patent document CN 114999114 A discloses a geological disaster early warning method and system based on meteorological and slope multivariate information. The method includes gridding the early warning area to obtain multiple slope body calculation units; obtaining current rainfall monitoring data, slope geometry information and slope rock and soil parameter information corresponding to the slope body calculation unit; calculating the current slope stability coefficient of the slope body calculation unit according to a preset slope wetting front depth model after rainfall infiltration and a slope stability quantitative analysis model; determining the warning level of each slope body calculation unit according to a preset slope stability coefficient-warning level correspondence relationship and the current slope stability coefficient of each slope body calculation unit, and performing geological disaster early warning.

[0006] Chinese patent document CN 115050168 A discloses a waterlogging monitoring and early warning system and early warning method based on cloud-edge-end collaboration. The method belongs to the field of waterlogging monitoring and early warning. The early warning system includes a terminal, an edge end, and a cloud server; the video device is used to monitor water accumulation and precipitation; the edge computing device is used to receive the video data sent by the video device and process and analyze it, and determine whether there is a waterlogging risk based on the training model built into the edge computing device; the cloud server assimilates multivariate data such as rainfall meteorological data and analysis data, simulates and analyzes it through the built-in coupling model, and synchronously returns it to the edge end to correct the early warning information.

[0007] However, these methods have the following shortcomings:

[0008] 1. The consideration of precipitation is not comprehensive, and the design of thresholds is not scientific. Precipitation cannot only take the daily precipitation value, but should be accurate to the hourly precipitation. With global climate change, extreme weather and climate events are frequent. Record-breaking heavy rainstorms have occurred in many parts of the country (such as Zhengzhou and Nanning). Just a few hours or even 1 hour of precipitation is enough to cause disasters. Therefore, in the analysis of historical precipitation data, research on hourly precipitation should be carried out. At the same time, in the consideration of precipitation forecast conclusions, the maximum hourly rainfall intensity can be considered to design a more practical and scientific threshold.

[0009] 2. The combination of environmental variables and precipitation factors needs to be improved. Many current methods take into account altitude, topography, population environment, etc., and these factors are considered relatively comprehensively, but the combination with precipitation factors needs to be improved, and it is not convincing in analyzing the causes of geological disasters caused by precipitation.

[0010] 3. The geological disaster meteorological risk warning method needs to be further improved. Some current methods require too many parameters to be input, many of which are difficult to accurately measure or quantify, and overemphasize the full automation of the warning process, lacking appropriate human intervention. At the same time, the precipitation threshold parameters of most existing geological disaster meteorological warning methods generally remain unchanged for a long time and cannot change in time with the changes in the number and location of subsequent geological disasters.

[0011] 4. The current road traffic meteorological service only has coarse-grained qualitative precipitation weather forecasts, but no refined geological disaster meteorological risk warning products, which makes it difficult to meet actual needs. Summary of the invention

[0012] In view of the above problems, the present invention aims to provide a meteorological early warning method for highway geological disasters under precipitation stress, which has simple environmental variables, simple content and simple operation. With the actual occurrence of subsequent geological disasters, the precipitation threshold will be continuously adjusted to continuously improve the reliability and practicality of geological disaster meteorological risk early warning. It can effectively issue alarms and management suggestions to reduce losses.

[0013] The main idea of ​​the technical solution adopted by the present invention is to use the geological disasters that have occurred along the highways in the province as distribution samples, analyze the meteorological factors such as the maximum hourly rainfall intensity 120h, 96h, 72h, 48h, 24h before the occurrence, and the comprehensive underlying surface factors such as elevation, faults, land use, and vegetation coverage index to obtain the key environmental variables and thresholds of geological disasters. Combined with the precipitation grid forecast, a high-precision meteorological risk warning of geological disasters is obtained. The high precision is reflected in the accuracy of the risk level of geological disasters based on pixel cells, so as to realize the meteorological warning of geological disasters along the highways.

[0014] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0015] A meteorological early warning method for highway geological disasters under precipitation stress, characterized in that it comprises the following steps:

[0016] Step 1: Based on the geological disasters that have occurred along the highway, ArcGIS is used for rasterization processing to obtain a raster map of the existing sample points of geological disasters along the highway;

[0017] Step 2: Using the distribution grid map obtained in step 1 as a sample, obtain multiple environmental variables in the distribution grid map, process the multiple environmental variables through machine learning models and ArcGIS software, and obtain key environmental variables of geological disasters and thresholds of key environmental variables;

[0018] Step 3: Based on the key environmental variables obtained in step 2, obtain the geological disaster related layers, analyze the geological disaster related layers, and obtain the geological disaster susceptibility zoning;

[0019] Step 4: Input the actual precipitation conditions of all meteorological monitoring stations in the geological disaster susceptibility zone 96 hours, 72 hours, 48 ​​hours, and 24 hours before the day obtained in step 3, as well as the forecast conclusions of precipitation in the next 24 hours, 48 ​​hours, 72 hours, and 96 hours on the day, into the machine learning model to obtain weather data, and then input the weather data into ArcGIS software to obtain the raster map of geological disaster meteorological risk forecast for 24 hours, 48 ​​hours, 72 hours, and 96 hours;

[0020] Step 5: Overlay the thresholds of key environmental variables obtained in step 2, the geological disaster susceptibility zones obtained in step 3, and the grid map obtained in step 4 to obtain a meteorological map. Perform spatial analysis and analysis of the meteorological risk level of geological disasters through the meteorological map to determine the geological disaster risk level of each pixel, and issue meteorological warnings based on the obtained geological disaster risk level.

[0021] Furthermore, the multiple environmental variables in step 2 include 120h precipitation, 96h precipitation, 72h precipitation, 48h precipitation, 24h precipitation, maximum hourly precipitation, elevation, fault, land use, vegetation cover index, curvature, slope, and aspect.

[0022] Furthermore, step 2 includes:

[0023] Step 2-1: First, input meteorological factors and underlying surface factors as environmental variables into the machine learning model to obtain preliminary model simulation results and the contribution of each environmental variable;

[0024] Step 2-2: Obtain the correlation coefficient matrix between different environmental variables through band statistics in ArcGIS software;

[0025] Step 2-3: Based on the correlation coefficient of the two environmental variables in the matrix in step 2-2, r≥0.8, combined with the contribution of each variable in the simulation results of all environmental variables in step 2-1, and the environmental conditions of geological disaster data, exclude environmental variables with high correlation, low contribution and low impact, and obtain key environmental variables;

[0026] Step 2-4: After modeling and analyzing the key environmental variables in step 2-3 through a machine learning model, a distribution diagram of the existence probability and each key environmental variable is obtained;

[0027] Step 2-5: Based on the existence probability in step 2-4 and the changing trend of the occurrence probability in the distribution diagram of each key environmental variable and the 95% confidence interval, the threshold value of each key environmental variable is obtained respectively.

[0028] Furthermore, step 3 includes:

[0029] Step 3-1: Input the key environmental variables obtained in step 2 into the receiver operating characteristic curve to obtain the geological disaster related layer;

[0030] Step 3-2: Input the geological disaster related layers in step 3-1 into ArcGIS software and convert them into raster, conduct reclassification analysis, and obtain the geological disaster susceptibility zoning.

[0031] Furthermore, determining the geological disaster risk level of each pixel in step 5 for meteorological warning includes:

[0032] Step 5-1: Based on the thresholds of each key environmental variable obtained in step 2, the geological disaster susceptibility zoning obtained in step 3, and the grid map obtained in step 4, grid overlay is performed to obtain a meteorological map;

[0033] Step 5-2: The meteorological risk warning threshold risk levels of geological disasters in the meteorological map of step 5-1 are represented by yellow, orange and red in the grid map, respectively, as 0.18<D≤3.27 for moderate risk, 3.27<D≤5.59 for high risk, and 5.59<D: extremely high risk;

[0034] Step 5-3: Carry out geological disaster meteorological warning according to the different geological disaster meteorological risk warning threshold risk levels in step 5-2.

[0035] The beneficial effect of the present invention is that, compared with the prior art, the improvement of the present invention is:

[0036] 1. The method of the present invention establishes a geological disaster meteorological early warning model with comprehensive meteorological factors and underlying surface factors as environmental variables and existing geological disasters as sample points. It is a new method for geological disaster meteorological early warning with simple environmental variables, simple content and concise operation, which can improve the accuracy and work efficiency of geological disaster meteorological early warning.

[0037] 2. The method of the present invention can make full use of the existing highway geological disaster susceptibility zoning and the future 24h, 48h, and 72h precipitation forecasts, based on ArcGIS software technology, precise targeted publishing technology, and combined with the meteorological risk level of highway geological disasters, automatically realize meteorological early warning of highway geological disasters.

[0038] 3. The meteorological factor thresholds of the method of the present invention, whether it is 24h, 48h precipitation, 72h precipitation, maximum hourly rainfall intensity, etc., can be adjusted in time. According to the actual occurrence of subsequent geological disasters, the machine learning model will be input to carry out further simulation and prediction, and the relevant precipitation thresholds will be adjusted, so as to continuously improve the reliability and practicality of geological disaster meteorological risk warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 This is a technical architecture diagram of the geological disaster meteorological risk warning technology of the present invention.

[0040] Figure 2 This is the geological disaster meteorological risk identification standard and forecast process of the present invention.

[0041] Figure 3 This is a grid map of the distribution of existing sample points of geological disasters along the highway of the present invention.

[0042] Figure 4 This is the grid diagram of the 24-hour geological disaster meteorological risk forecast of the present invention.

[0043] Figure 5 This is the grid diagram of the 48-hour geological disaster meteorological risk forecast of the present invention.

[0044] Figure 6This is the grid diagram of the 72-hour geological disaster meteorological risk forecast of the present invention.

[0045] Figure 7 This is the training model score graph obtained by the jackknife method of the present invention.

[0046] Figure 8 It is the response curve and receiver operating characteristic curve of NDVI in the present invention.

[0047] Fig. 9 It is the response curve and receiver operating characteristic curve of DEM in the present invention.

[0048] Fig.10 It is the response curve of the interruption layer and the receiver operating characteristic curve diagram of the present invention.

[0049] Fig.11 It is the response curve and receiver operating characteristic curve diagram of the specificity in the present invention. DETAILED DESCRIPTION

[0050] In order to enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0051] See also Figure 1-Figure 11 The present application discloses a meteorological early warning method for highway geological disasters under precipitation stress, comprising the following steps:

[0052] Step 1: Based on the geological disasters that have occurred along the highway, ArcGIS is used for rasterization processing to obtain a raster map of the existing sample points of geological disasters along the highway; Figure 3 As shown;

[0053] The geological disasters include debris flow, landslide, collapse, ground subsidence, ground subsidence or cracking. ArcArcGIS software is used for raster processing to obtain the distribution raster map of existing sample points of geological disasters along the highway.

[0054] Step 2: Using the distribution grid map obtained in step 1 as a sample, obtain multiple environmental variables in the distribution grid map, process the multiple environmental variables through machine learning models and ArcGIS software, and obtain key environmental variables of geological disasters and thresholds of key environmental variables;

[0055] Among them, the key environmental variables in step 2 are obtained through a machine learning model (MaxEnt); when performing layer analysis and raster calculations, ArcArcGIS software is used, that is, all mapping and output use ArcArcGIS software.

[0056] In step 2, the specific operation of obtaining multiple environmental variables in the distribution grid map is as follows: select all regional meteorological monitoring stations within 2km along the highway, and obtain the precipitation 120h (5 days) before the day when the geological disaster occurred in the historical records based on the historical records of precipitation at such stations and the time and place of geological disasters that have occurred in the historical records. According to meteorological elements such as 120h, 96h, 72h, 48h, 24h, and maximum hourly rainfall intensity, the underlying surface factor is used as the environmental variable, and a machine learning model is used to obtain the key environmental variables of geological disasters.

[0057] The 120-hour precipitation refers to the sum of the precipitation from the day before the geological disaster to the 5 days before the day when the geological disaster occurs; the 96-hour precipitation refers to the sum of the precipitation from the day before the geological disaster to the 4 days before the day when the geological disaster occurs; the 72-hour precipitation and 48-hour precipitation are similar; the 24-hour precipitation refers to the cumulative precipitation on the day before the geological disaster occurs; the maximum hourly rainfall intensity refers to the maximum precipitation within one hour from 1 day to 3 days before the day when the geological disaster occurs.

[0058] Specifically, the obtained multiple environmental variables were input into the MaxEnt model, 75% of the data were used to train the model, and 25% of the data were used to verify the model. The model was iterated 10,000 times. In order to reduce the variability of the model results caused by randomness and improve the stability and reliability of the model results, the repeated operation was set to 10 times, and the repeated operation type was bootstrap. The jackknife method was selected to calculate the contribution rate and replacement importance of the environmental variable data. The initial model was rerun by excluding each variable one by one to determine the contribution of each environmental variable to the model. The key environmental variables were screened in the jackknife method.

[0059] It should be noted that the multiple environmental variables in step 2 include 120h precipitation, 96h precipitation, 72h precipitation, 48h precipitation, 24h precipitation, maximum hourly precipitation, elevation, fault, land use, vegetation cover index, curvature, slope, and aspect. Among them, 120h precipitation, 96h precipitation, 72h precipitation, 48h precipitation, 24h precipitation, and maximum hourly precipitation are all meteorological factors, and elevation, fault, land use, vegetation cover index, curvature, slope, and aspect are all underlying surface factors.

[0060] Specifically, step 2 includes:

[0061] Step 2-1: First, input meteorological factors and underlying surface factors as environmental variables into the machine learning model to obtain preliminary model simulation results and the contribution of each environmental variable;

[0062] Step 2-2: Obtain the correlation coefficient matrix between different environmental variables through band statistics in ArcGIS software;

[0063] Step 2-3: Based on the correlation coefficient of the two environmental variables in the matrix in step 2-2 || r ≥ 0.8, combined with the contribution of each variable in the simulation results of all environmental variables in step 2-1, and the environmental conditions of geological disaster data, exclude environmental variables with high correlation, low contribution and low impact, and obtain key environmental variables.

[0064] It should be noted that due to the influence of the correlation between factors, the model simulation may be overfitted. Therefore, all environmental variables were first input into the machine learning model to obtain the preliminary model simulation results and the contribution of each environmental variable, and then the correlation coefficient matrix between different environmental variables was obtained through band statistics in ArcGIS software. Then, according to the correlation coefficient of the two environmental variables in the matrix ||r≥0.8, combined with the contribution of each variable in the simulation results based on all environmental variables, and the environmental conditions of the geological disaster data used in this study, environmental variables with high correlation, low contribution and low impact were excluded.

[0065] Step 2-4: After modeling and analyzing the key environmental variables in step 2-3 through a machine learning model, a distribution diagram of the existence probability and each key environmental variable is obtained;

[0066] Step 2-5: Based on the existence probability in step 2-4 and the changing trend of the occurrence probability in the distribution diagram of each key environmental variable and the 95% confidence interval, the threshold value of each key environmental variable is obtained respectively.

[0067] The distribution diagram in steps 2-4 is as follows Figure 7 As shown, Figure 7 In the table, 120 represents the accumulated precipitation in 120 hours, 24 represents the accumulated precipitation in 24 hours, and so on; h represents the maximum hourly precipitation in 48 hours; dc represents the fault; dem represents the elevation; ndvi represents the vegetation cover index; podu represents the slope; poxiang represents the slope direction, and qulv represents the curvature. Figure 7 The changing trend of the occurrence probability and the 95% confidence interval were used to obtain the thresholds of each key environmental variable. Figure 8-Figure 11 The response curves of several factors and the receiver operating characteristic curves are used as examples to illustrate that the role of factor response curves and receiver operating characteristic curves is very important. They can not only determine whether the factors are related to the occurrence of geological disasters, but also show the probability of occurrence of geological disasters, and at what specific value of a certain factor the probability of geological disasters occurring is the highest (and what is the maximum probability), etc.

[0068] For example, if the vegetation cover index is between 0.32 and 0.58, the risk of geological disasters is relatively high (the probability exceeds 50%); if it is between 0.4 and 0.5, the risk of geological disasters is very high (the probability exceeds 60%); for example, if the precipitation in 48 hours is between 180 and 210 mm, the risk of geological disasters is relatively high, and if it exceeds 210 mm, the risk is very high; if the precipitation in 120 hours is above 280 mm, the risk of geological disasters is very high; if the maximum hourly rainfall intensity within 72 hours exceeds 65 mm, the risk of geological disasters is very high.

[0069] like Figure 8 As shown, the horizontal axis is the vegetation coverage index (NDVI), and the vertical axis is the probability of geological disasters. The red line is the trend of the probability of occurrence, and the blue area is the 95% confidence interval ( Figure 9-10 , similarly).

[0070] Figure 8 It shows that when NDVI is in the range of 0.28 to 0.56, the probability of geological disasters exceeds 50%; when NDVI is between 0.41 and 0.47, the probability of geological disasters exceeds 70%; when NDVI exceeds 0.47, the probability of geological disasters will decrease as NDVI decreases.

[0071] like Fig. 9 As shown, the horizontal axis is the elevation (dem) and the vertical axis is the probability of geological disasters. The above figure shows that the probability of geological disasters increases with the increase of altitude. When the altitude is 300m, the probability of geological disasters is the highest, and then it will decrease with the increase of altitude; when the altitude is 100-600m, the probability of geological disasters exceeds 50%; when the altitude is 120-400m, the probability of geological disasters exceeds 70%.

[0072] like Fig.10 As shown, the horizontal axis is the fault and the vertical axis is the probability of geological disasters. The above figure shows that when the fault is greater than 6000, the probability of geological disasters is less than 50%; when the fault is greater than 18500, the probability of geological disasters increases with the increase of the fault. When the fault is 21000, the probability of geological disasters is the highest, and then it will decrease with the increase of the fault; when the fault is greater than 66000, the probability of geological disasters is stable.

[0073] like Fig.11 As shown in the figure, the AUC of the training data is 0.951, and the AUC of the test data is 0.871, indicating that the MaxEnt model has good simulation accuracy for geological disasters in both training and test data. The model can not only fit the training data well, but also make accurate predictions on unknown data. The prediction results are highly accurate and not random, and the results can be used for geological disaster assessment and prediction.

[0074] Step 3: Based on the key environmental variables obtained in step 2, obtain the geological disaster related layers, analyze the geological disaster related layers, and obtain the geological disaster susceptibility zoning;

[0075] Step 3-1: Input the key environmental variables of geological hazards obtained in step 2 into the receiver operating characteristic curve, and obtain the geological hazard related layer;

[0076] Step 3-2: Input the geological disaster related layers in step 3-1 into ArcArcGIS software and convert them into raster, conduct reclassification analysis, select "natural break point classification method" for classification method, and select "5" for category, i.e. five levels of geological disaster risk: very high, high, medium, low, and low. The geological disaster susceptibility zoning is obtained with a resolution of 30m.

[0077] Specifically, step 3 includes:

[0078] The key environmental variables that have been screened out are tested, and the threshold and related layers (ASCII code) are obtained. The model accuracy is judged by the AUC value of the receiver operating characteristic curve (ROC), and the AUC value is between 0 and 1.0. The larger the AUC value, the higher the model accuracy and the better the credibility. When AUC < 0.6, the prediction result is unqualified; 0.6 ≤ AUC < 0.7, the prediction result is poor; 0.7 ≤ AUC < 0.8, the prediction result is general; 0.8 ≤ AUC < 0.9, the prediction result is good; 0.9 ≤ AUC < 1.0, the prediction result is excellent. AUC = 1, the predicted value is completely consistent with the actual value; Cloglog is selected for the simulation result output, the format is ASCII, and the rest of the parameters use the default values ​​of the MaxEnt model. First, the environmental variables with a contribution rate of < 1% are eliminated from the environmental variables; secondly, SPSS is used to perform Spearman correlation analysis on the environmental variables that have not been eliminated, and the environmental variables with a correlation coefficient of > 0.8 between variables and a small impact on geological disasters are eliminated, in order to screen out key environmental factors that have a greater impact on the occurrence of geological disasters. The present invention uses the default parameter settings of the MaxEnt model and no improvements are made. The performance test method and index of the evaluation model are mainly judged by the AUC value of the receiver operating characteristic curve (ROC), and the AUC value is between 0 and 1, and the larger the better. The training data AUC of this model is 0.95, and the test AUC is 0.87. The prediction results are good, and the estimation and prediction of geological disaster early warning methods can be applied.

[0079] Step 4: Input the actual precipitation conditions of 96h, 72h, 48h, and 24h at all meteorological monitoring stations in the geological disaster susceptibility zone obtained in step 3, as well as the forecast conclusions of the precipitation forecast for the next 24h, 48h, 72h, and 96h on the same day, into the machine learning model to obtain weather data. Input the weather data into ArcGIS software to obtain the raster maps of the geological disaster meteorological risk forecast for 24h, 48h, 72h, and 96h; Figure 4-6 As shown;

[0080] Among them, the actual precipitation conditions for 96h, 72h, 48h and 24h are the actual conditions before a certain day that needs to be forecasted;

[0081] The forecast starts on the day of the month. If the forecast for the next 24 hours is made in the morning, it will be from 08:00 on the day of the month to 08:00 on the next day; if the forecast is made in the afternoon of the day of the month, it will be from 20:00 on the day of the month to 20:00 on the next day.

[0082] Weather data is the sum of past data and future data;

[0083] The grid map of the geological disaster meteorological risk forecast for 24h, 48h, 72h and 96h is the future grid map;

[0084] Step 5: Overlay the thresholds of key environmental variables obtained in step 2, the geological disaster susceptibility zones obtained in step 3, and the grid map obtained in step 4 to obtain a meteorological map. Perform spatial analysis and analysis of the meteorological risk level of geological disasters through the meteorological map to determine the geological disaster risk level of each pixel, and issue meteorological warnings based on the obtained geological disaster risk level.

[0085] Among them, raster overlay is to superimpose layers such as rainstorm areas, highway maps, and geological disaster distribution points on a map to carry out analysis and services.

[0086] It should also be noted that the forecast is a forecast of geological disaster meteorological risk for the next 24h, 48h, and 72h, and the warning is a warning of geological disaster meteorological risk for the next 1h, 2h, 3h, 4h, 5h, and 6h.

[0087] The geological hazard risk level of each pixel determined in step 5 includes:

[0088] Step 5-1: Based on the thresholds of each key environmental variable obtained in step 2, the geological disaster susceptibility zoning obtained in step 3, and the grid map obtained in step 4, grid overlay is performed to obtain a meteorological map;

[0089] Step 5-2: The meteorological risk warning threshold risk levels of geological disasters in the meteorological map of step 5-1 are represented by yellow, orange and red in the grid map, respectively, as 0.18<D≤3.27 for moderate risk, 3.27<D≤5.59 for high risk, and 5.59<D: extremely high risk;

[0090] Step 5-3: Carry out geological disaster meteorological warning according to the different geological disaster meteorological risk warning threshold risk levels in step 5-2.

[0091] Specifically, Figure 2 The geological disaster meteorological risk identification standard and forecast process are used as an example to illustrate

[0092] All precipitation units are in millimeters and the starting time is 08:00.

[0093] Example: ① Calculate the meteorological risk level of geological disasters in the next 24 hours. If the meteorological risk level of geological disasters on October 5 is predicted, 08:00 on the 5th is the reporting time. In addition to considering the precipitation in the next 24 hours (i.e., the precipitation from 08:00 on October 5 to 08:00 on October 6), it is also necessary to consider the precipitation in 48 hours (i.e., 08:00 on October 4 to 08:00 on October 6), 72 hours (i.e., 08:00 on October 3 to 08:00 on October 6), 96 hours, 120 hours precipitation and so on, as well as the maximum hourly rainfall intensity in 72 hours (i.e., 08:00 on October 3 to 08:00 on October 6). After comprehensively judging the forecast value and the actual value, the meteorological risk level of geological disasters in the next 24 hours is obtained.

[0094] It should be noted that the method of obtaining the meteorological risk level of geological disasters in the next 24 hours by comprehensively judging the forecast value and the actual value is used as an example to illustrate:

[0095] Refer to the industry standard "Meteorological Risk Warning Level for Geological Disasters Induced by Heavy Rain" QX / T487-2019.

[0096] The calculation formula for the provincial geological disaster meteorological risk warning threshold D is:

[0097]

[0098] Note: All precipitation units are in millimeters and the starting time is 08:00.

[0099] 0.136, 0.450, 0.224, and 0.190 are the value-added contributions and total average contributions of each precipitation period (maximum hourly rainfall intensity within 24h, 48h, 72h, and 120h) to geological disasters, and their proportions are used as weight coefficients for constructing meteorological risks of geological disasters;

[0100] The denominators 72.2, 107.2, 119.1, and 36.6 are the average cumulative precipitation in 24 hours, the average cumulative precipitation in 48 hours, the average cumulative precipitation in 72 hours, and the average maximum hourly rainfall intensity within 120 hours when the geological disaster occurred, respectively.

[0101] The geological disaster meteorological risk warning threshold D risk level is: 0.01≤D≤0.18: mild risk; 0.18<D≤3.27: moderate risk; 3.27<D≤5.59: high risk; 5.59<D: extremely high risk. Among the above risk levels, the geological disaster meteorological risk is mild risk and below the level of warning will not be issued, but will be displayed in different colors in the grid map, moderate risk is displayed in yellow, high risk is displayed in orange, and extremely high risk is displayed in red.

[0102] The method for calculating the disaster meteorological risk level of the address in the next 48 hours and 72 hours is the same as above and will not be elaborated here.

[0103] ② Calculate the meteorological risk level of geological disasters in the next 48 hours. If the meteorological risk level of geological disasters on October 5 and 6 is predicted, 08:00 on May is the starting time, and the precipitation in the next 24 hours (i.e., the precipitation from 08:00 on October 5 to 08:00 on October 6) and 48 hours (i.e., the precipitation from 08:00 on October 5 to 08:00 on October 7) need to be considered respectively; at the same time, the actual value of 48 hours precipitation (i.e., 08:00 on October 4 to 08:00 on October 6) and 72 hours precipitation (i.e., 08:00 on October 3 to 08:00 on October 6) should also be considered, and the 96 hours and 120 hours precipitation should be deduced by analogy, as well as the maximum hourly rainfall intensity of 72 hours (i.e., 08:00 on October 3 to 08:00 on October 6). After comprehensively judging the forecast value and the actual value, the meteorological risk level of geological disasters in 24 hours and 48 hours is obtained.

[0104] ③ Calculate the geological disaster meteorological risk level in the next 72 hours. The method is the same as above. Under normal circumstances, the 24-hour geological disaster meteorological risk forecast is more accurate. It has a strong guiding significance in daily work.

[0105] When the precipitation forecast value takes the maximum hourly precipitation value, the critical precipitation value corresponds to the precipitation threshold of the maximum hourly rainfall intensity, and a 1h to 6h warning is issued.

[0106] When the precipitation forecast value is the 24h, 48h, or 72h precipitation value, the critical precipitation value corresponds to the 24h, 48h, 72h, 96h, or 120h precipitation threshold value.

[0107] Utilizing the latest results of current geological disaster investigations, according to the geological disaster susceptibility zone to which each pixel belongs, the critical precipitation forecast for the next 24h, 48h, and 72h is combined with the actual precipitation conditions (96h, 72h, 48h, 24h) to calculate and conduct geological disaster meteorological risk forecasts 24h, 48h, and 72h in advance; and geological disaster meteorological risk warnings are carried out 6h in advance.

[0108] After a new geological disaster occurs, modeling analysis will be carried out again based on the geological disaster warning situation and the actual occurrence of geological disasters, and the thresholds will be further adjusted to update the geological disaster environmental indicators and thresholds, continuously improving the geological disaster meteorological risk warning capabilities.

[0109] Example

[0110] Taking the expressways in Hunan Province as an example, Figure 1 The following is the technical framework of meteorological early warning for geological disasters on highways in Hunan Province. For geological disasters on highways in the province (including debris flow, landslide, collapse, ground subsidence, ground subsidence or cracking), early warning for geological disasters is carried out according to the following process. The main early warning period is the annual flood season, that is, from April 1 to September 30 each year. A forecast is made every morning and early warning is issued in a timely manner. When yellow, orange, and red warnings appear in the early warning results, meteorological early warning information for geological disasters is promptly released to all sectors of society through special topics, special reports, express reports, and other forms, as well as television programs, broadcasts, and text messages.

[0111] Step 1: Organize basic information such as road network, susceptibility zones, and meteorological monitoring stations. Organize the vector map of geological disaster susceptibility zones of all expressways in the province and all regional meteorological monitoring stations within 2km along the expressways in the province. Use ArcGIS software to rasterize the geological disaster susceptibility zones of all expressways in the province with a resolution of 30m×30m. Each pixel corresponds to a meteorological warning unit for geological disasters in each place, and finally form a raster map of geological disaster susceptibility zones of all expressways in the province.

[0112] Step 2: Process the actual precipitation and forecast values. The actual value is the measured rainfall with the maximum hourly rainfall intensity within 96h, 72h, 48h, 24h and 72h on the forecast day, and the forecast value is the precipitation forecast for the next 24h, 48h and 72h. The precipitation forecast for the next 24h, 48h and 72h is combined with the measured maximum hourly rainfall intensity within 96h, 72h, 48h, 24h and 72h on the forecast day to form the meteorological risk forecast for geological disasters in the next 24h, 48h and 72h; the 24h, 48h and 72h meteorological risk forecast for geological disasters is rasterized in the same way through ArcGIS software, and finally the 24h, 48h and 72h meteorological risk forecast for geological disasters and the forecast raster map are formed.

[0113] Step 3: Implement meteorological early warning for geological disasters on highways. Utilize the raster map of 24h, 48h, and 72h geological disaster meteorological risk forecast formed in the second step, and use ArcGIS software raster overlay spatial analysis method and highway geological disaster risk level determination method to calculate the meteorological early warning for geological disasters on highways. Figure 2 The process is used to carry out geological disaster meteorological risk warning.

[0114] Geological disaster meteorological risk classification standards:

[0115] Extremely high-risk area: The probability of geological disasters is greater than 70%

[0116] High-risk areas: Probability of geological disasters 45-69%

[0117] Medium-prone areas: probability of geological disasters 20-44%

[0118] Low-risk areas: The probability of geological disasters is 8-19%

[0119] Extremely low-prone area: The probability of geological disasters occurring is less than 8%.

[0120] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A meteorological early warning method for highway geological disasters under precipitation stress, characterized in that: The following steps are involved: Step 1: Based on the geological disasters that have occurred along the highway, ArcGIS is used for rasterization processing to obtain a raster map of the existing sample points of geological disasters along the highway; Step 2: Using the distribution grid map obtained in step 1 as a sample, obtain multiple environmental variables in the distribution grid map, process the multiple environmental variables through machine learning models and ArcGIS software, and obtain key environmental variables of geological disasters and thresholds of key environmental variables; Step 3: Based on the key environmental variables obtained in step 2, obtain the geological disaster related layers, analyze the geological disaster related layers, and obtain the geological disaster susceptibility zoning; Step 4: Input the actual precipitation conditions of all meteorological monitoring stations in the geological disaster susceptibility zone 96 hours, 72 hours, 48 ​​hours, and 24 hours before the day obtained in step 3, as well as the forecast conclusions of precipitation in the next 24 hours, 48 ​​hours, 72 hours, and 96 hours on the day, into the machine learning model to obtain weather data, and then input the weather data into ArcGIS software to obtain the raster map of geological disaster meteorological risk forecast for 24 hours, 48 ​​hours, 72 hours, and 96 hours; Step 5: Overlay the thresholds of key environmental variables obtained in step 2, the geological disaster susceptibility zones obtained in step 3, and the grid map obtained in step 4 to obtain a meteorological map. Perform spatial analysis and analysis of the meteorological risk level of geological disasters through the meteorological map to determine the geological disaster risk level of each pixel, and issue meteorological warnings based on the obtained geological disaster risk level.

2. The meteorological early warning method for highway geological disasters under precipitation stress according to claim 1 is characterized by: The multiple environmental variables in step 2 include 120h precipitation, 96h precipitation, 72h precipitation, 48h precipitation, 24h precipitation, maximum hourly precipitation, elevation, fault, land use, vegetation cover index, curvature, slope, and aspect.

3. The meteorological early warning method for highway geological disasters under precipitation stress according to claim 1 is characterized in that: Step 2 includes: Step 2-1: First, input meteorological factors and underlying surface factors as environmental variables into the machine learning model to obtain preliminary model simulation results and the contribution of each environmental variable; Step 2-2: Obtain the correlation coefficient matrix between different environmental variables through band statistics in ArcGIS software; Step 2-3: Based on the correlation coefficient of the two environmental variables in the matrix in step 2-2, r≥0.8, combined with the contribution of each variable in the simulation results of all environmental variables in step 2-1, and the environmental conditions of geological disaster data, exclude environmental variables with high correlation, low contribution and low impact, and obtain key environmental variables; Step 2-4: After modeling and analyzing the key environmental variables in step 2-3 through a machine learning model, a distribution diagram of the existence probability and each key environmental variable is obtained; Step 2-5: Based on the existence probability in step 2-4 and the changing trend of the occurrence probability in the distribution diagram of each key environmental variable and the 95% confidence interval, the threshold value of each key environmental variable is obtained respectively.

4. The meteorological early warning method for highway geological disasters under precipitation stress according to claim 1 is characterized in that: Step 3 includes: Step 3-1: Input the key environmental variables obtained in step 2 into the receiver operating characteristic curve to obtain the geological disaster related layer; Step 3-2: Input the geological disaster related layers in step 3-1 into ArcGIS software and convert them into raster, conduct reclassification analysis, and obtain the geological disaster susceptibility zoning.

5. The meteorological early warning method for highway geological disasters under precipitation stress according to claim 1 is characterized in that: Determining the geological disaster risk level of each pixel for meteorological warning in step 5 includes: Step 5-1: Based on the thresholds of each key environmental variable obtained in step 2, the geological disaster susceptibility zoning obtained in step 3, and the grid map obtained in step 4, grid overlay is performed to obtain a meteorological map; Step 5-2: The meteorological risk warning threshold risk levels of geological disasters in the meteorological map of step 5-1 are represented by yellow, orange and red in the grid map, respectively, as 0.18<D≤3.27 for moderate risk, 3.27<D≤5.59 for high risk, and 5.59<D: extremely high risk; Step 5-3: Carry out geological disaster meteorological warning according to the different geological disaster meteorological risk warning threshold risk levels in step 5-2.

Images