Hail early warning method and system based on lightning jump and support vector machine

Abstract

The invention discloses a hail early warning method and system based on lightning jump and a support vector machine, and the method comprises the steps: obtaining a combined reflectivity factor, a vertical integral liquid water content and echo top height data through radar networking data, recognizing a severe convective echo region, and calculating a lightning time change rate and a standard deviation thereof; a potential hail event is screened in combination with a lightning jump increase condition, a vertical integral liquid water content maximum value VILmax, a jump increase GVIL and a density VILD are further extracted as feature parameters, and the feature parameters are input into a support vector machine classification model based on historical sample training for secondary judgment, so that the false alarm rate of a traditional 2 sigma algorithm is effectively reduced, and the hail early warning accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of severe convective weather forecasting and warning services, and particularly relates to a hail warning method and system based on lightning jump and support vector machine. Background Art

[0002] Hail is a common meteorological natural disaster, often accompanied by lightning activities. With the improvement of lightning detection capabilities, many scholars at home and abroad have confirmed that the frequency of lightning occurrences will also suddenly increase before and after hail occurs.

[0003] Currently, the main methods for calculating lightning jumps are the Gatlin algorithm and the σ algorithm. Gatlin proposed an algorithm for identifying and warning severe storms based on lightning jumps, namely the Gatlin algorithm.

[0004] As a main hail warning method based on lightning data, the algorithm has achieved good application results, but there are still problems with a relatively high false alarm rate. However, dual-polarization radars have not yet achieved full business coverage, and areas without dual-polarization radar data cannot use this method to improve the hail warning effect; in addition, the identification results of the hydrometeor phase states of dual-polarization radars highly depend on the data quality of polarization parameters, and the polarization data is prone to large errors under the influence of various factors during the actual operation of the radar and requires careful and complete quality control processing.

[0005] In recent years, with the gradual improvement of the Support Vector Machine (SVM) method, support vector machine has become a new computer learning method for dealing with highly non-linear classification, regression and other problems. It is a novel small-sample learning method with a solid theoretical basis, which realizes efficient "transductive inference" from training samples to prediction samples, greatly simplifies the usual classification and regression problems, and performs well in dealing with problems of high-dimensional data and small sample sizes.

[0006] In summary, based on lightning and radar data, deeply studying the hail warning method that integrates lightning jump and support vector machine to improve the forecasting and warning level of severe convective weather such as hail is of great significance for effectively enhancing the ability of disaster prevention and mitigation and reducing people's lives and property losses. Summary of the Invention

[0007] In order to effectively make up for the deficiencies of the existing hail weather warning methods in weather forecasting operations and solve the problem of too high a false alarm rate of hail warnings in the existing technology, a hail warning method based on lightning jump and support vector machine is provided. The steps include:

[0008] Based on the radar networking mosaic data in the warning area, obtain the two-dimensional grid data of the radar combined reflectivity factor CR, vertically integrated liquid water content VIL, and echo top height TOP;

[0009] Based on the two-dimensional grid data of the radar combined reflectivity factor CR, obtain the grid position information of the severe convective echo connected area;

[0010] Based on the lightning data in the warning area and the grid position information of the severe convective echo connected area, calculate the lightning time rate of change D and the standard deviation of the lightning time rate of change D ;

[0011] Based on the two-dimensional grid data of the vertically integrated liquid water content VIL and echo top height TOP and the grid position information of the severe convective echo connected area, obtain the maximum value VIL of the vertically integrated liquid water content VIL max and the jump increment GVIL and density VILD of the VIL max ;

[0012] Based on the lightning time rate of change D and the standard deviation of the lightning time rate of change D perform lightning jump increment condition judgment;

[0013] For the times that meet the lightning jump increment conditions, based on the maximum value VIL of the vertically integrated liquid water content VIL max and the jump increment GVIL and density VILD of the VIL max perform hail warning condition judgment using a support vector machine to achieve hail warning.

[0014] Preferably, use the neighborhood method to obtain the grid position information of the severe convective echo connected area, and the steps include:

[0015] Perform binary image processing on the two-dimensional grid data of the radar combined reflectivity factor CR;

[0016] For the data after binary image processing, use the neighborhood method to identify the severe convective echo connected area and obtain the grid position information of the severe convective echo connected area.

[0017] Preferably, the steps of using the neighborhood method to identify the severe convective echo connected area include:

[0018] Taking the current severe convective echo grid point as the center, using the eight adjacent grid points above, below, left, right, upper left, upper right, lower left, and lower right as the neighborhood of the current grid point, judge whether any grid point in the neighborhood is a severe convective echo grid point, and the judgment conditions include:

[0019] If the grid point is a severe convective echo grid point, then it belongs to the same connected area as the center grid point;

[0020] If the grid point is not a severe convective echo grid point, then it is a non-connected grid point;

[0021] In this way, the neighborhood grid points of all grid points in the same connected area are judged in a loop until all the neighborhood grid points of the connected area are non-connected grid points, then the judgment and recognition of a connected area are completed.

[0022] Preferably, calculate the lightning time rate of change D and the standard deviation of the lightning time rate of change D The steps include:

[0023] Based on the lightning data in the early warning area and the grid point position information of the severe convective echo connected area, count the lightning frequency of the severe convective echo connected area every minute;

[0024] Based on the lightning frequency of the severe convective echo connected area, calculate the lightning time rate of change D and the standard deviation of the lightning time rate of change D .

[0025] Preferably, the steps of counting the lightning frequency of the severe convective echo connected area include:

[0026] Based on the lightning data in the early warning area and the grid point position information of the severe convective echo connected area, judge whether the lightning occurrence position is in the severe convective echo area. If so, add 1 to the lightning frequency count; wherein, the lightning includes: cloud flash and ground flash;

[0027] Taking minutes as the interval, count the lightning frequency every minute.

[0028] Preferably, the steps of calculating the lightning time rate of change D and the standard deviation of the lightning time rate of change D The steps include:

[0029] Step S61: Based on the lightning frequency of the severe convective echo connected area, calculate the average value of the lightning frequency per minute in the last two time units;

[0030] Step S62: Repeat step S61 to calculate the average values of the lightning frequency per minute in a period of several time units pushed forward from the current time. Each average value corresponds to a specific time interval;

[0031] Step S63: Based on the average values of the lightning frequency per minute obtained in step S61 and step S62, calculate the difference between adjacent two average values of the lightning frequency per minute in turn to obtain a plurality of lightning time rates of change D;

[0032] Step S64: Based on a set of lightning time rates of change D obtained in step S63, calculate the standard deviation representing the degree of dispersion of these time rates of change .

[0033] Preferably, the method for determining the hail warning conditions of the support vector machine includes: inputting radar characteristic parameters. If the output is 1, a hail warning is issued; if the output is 0, no hail warning is issued. Among them, the support vector machine classification model uses a radial basis kernel function and is trained based on lightning jump increase data samples.

[0034] The present invention also provides a hail warning system based on lightning jump increase and support vector machine. The system is used to implement the above method and includes: a collection module, a position acquisition module, a first calculation module, a second calculation module, a judgment module, and a warning module.

[0035] The collection module is used to obtain two-dimensional grid data of the radar combined reflectivity factor CR, the vertically integrated liquid water content VIL, and the echo top height TOP based on the radar network mosaic data of the warning area.

[0036] The position acquisition module is used to obtain the grid position information of the severe convective echo connected area based on the two-dimensional grid data of the radar combined reflectivity factor CR.

[0037] The first calculation module is used to calculate the lightning time rate of change D and the standard deviation of the lightning time rate of change D based on the lightning data of the warning area and the grid position information of the severe convective echo connected area. ;

[0038] The second calculation module is used to obtain the maximum value VIL of the vertically integrated liquid water content VIL, the jump increment GVIL, and the density VILD of the VIL based on the two-dimensional grid data of the vertically integrated liquid water content VIL and the echo top height TOP and the grid position information of the severe convective echo connected area. max and the VIL max jump increment GVIL and density VILD;

[0039] The judgment module is used to judge the lightning jump increase conditions based on the lightning time rate of change D and the standard deviation of the lightning time rate of change D. Perform lightning jump increase condition judgment;

[0040] The warning module is used to, for the times that meet the lightning jump increase conditions, based on the maximum value VIL of the vertically integrated liquid water content VIL max and the VIL max jump increment GVIL and density VILD, perform the hail warning condition judgment of the support vector machine to achieve hail warning.

[0041] Preferably, the position acquisition module uses the neighborhood method to obtain the grid position information of the severe convective echo connected area, and the process includes:

[0042] Perform binary image processing on the two-dimensional grid data of the radar combined reflectivity factor CR.

[0043] For the data after binarization image processing, the neighborhood method is used to identify the convective echo connected region, and the grid point position information of the convective echo connected region is obtained.

[0044] Preferably, the process of identifying the convective echo connected region by the neighborhood method includes:

[0045] Taking the current convective echo grid point as the center, the upper, lower, left, right, upper left, upper right, lower left, and lower right 8 adjacent grid points are used as the neighborhood of the current grid point, and it is judged whether any grid point in the neighborhood is a convective echo grid point. The judgment conditions include:

[0046] If the grid point is a convective echo grid point, it belongs to the same connected region as the center grid point;

[0047] If the grid point is not a convective echo grid point, it is a non-connected grid point;

[0048] In this way, all grid points in the same connected region are cyclically judged for neighboring grid points until all neighboring grid points in the connected region are non-connected grid points, then the judgment and identification of a connected region are completed.

[0049] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0050] Based on historical hail weather processes, through statistical analysis Under the cases of effective warning and ineffective warning of hail process by the lightning jump method, the distribution differences of radar characteristic parameters of the convective system are used to construct a classification model based on support vector machine, realizing the elimination of empty ineffective hail warnings, which can effectively make up for the deficiencies of existing hail warning methods in weather forecasting services and improve the application level of lightning data in severe convective weather monitoring and warning. The technical solution provided by the present invention improves the hail warning effect of the lightning jump method, effectively reduces the false alarm rate of hail warnings, and can be applied to existing hail warning business services. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the present invention, the drawings required for use in the embodiments are briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0052] Figure 1 It is a flow chart of the hail warning method based on lightning jump and support vector machine taking Province A as an example in the embodiments of the present invention;

[0053] Figure 2Time distribution map of the minute lightning frequency of the convective system in County B, Province A on May 4, 2020, which is an embodiment of the present invention;

[0054] Figure 3 Time distribution map of the strongest echo of the convective system in County B, Province A on May 4, 2020, which is an embodiment of the present invention;

[0055] Figure 4 Time distribution map of the judgment of the support vector machine hail warning conditions for the convective system in County B, Province A on May 4, 2020, which is an embodiment of the present invention. Detailed implementation manners

[0056] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with 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.

[0057] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners.

[0058] Embodiment 1

[0059] This embodiment provides a hail warning method step based on lightning jump increase and support vector machine. The steps include:

[0060] S1. Based on the radar network mosaic data of the warning area, obtain the two-dimensional grid data of the radar combined reflectivity factor CR, the vertically integrated liquid water content VIL, and the echo top height TOP.

[0061] S2. Based on the two-dimensional grid data of the radar combined reflectivity factor CR, obtain the grid position information of the severe convective echo connected area.

[0062] S201. Perform binary image processing on the two-dimensional grid data of the radar combined reflectivity factor CR;

[0063] S202. For the data after binary processing, use the neighborhood method to identify the severe convective echo connected area and obtain the grid position information of the severe convective echo connected area.

[0064] Taking the current severe convective echo grid as the center, taking the eight adjacent grids above, below, left, right, upper left, upper right, lower left, and lower right as the neighborhood of the current grid, determine whether any grid in the neighborhood is a severe convective echo grid. The judgment conditions include:

[0065] If the grid is a severe convective echo grid, then it belongs to the same connected area as the center grid;

[0066] If the grid point is not a severe convective echo grid point, then it is a non-connected grid point;

[0067] In this way, loop through all the grid points in the same connected area to perform neighborhood grid point judgment until all the neighborhood grid points in the connected area are non-connected grid points, then the judgment and identification of a connected area are completed.

[0068] S3. Based on the lightning data in the warning area and the grid point position information of the severe convective echo connected area, calculate the lightning time rate of change D and the standard deviation of the lightning time rate of change D .

[0069] S301. Based on the lightning data in the warning area and the grid point position information of the severe convective echo connected area, statistically count the lightning frequency LF in the severe convective echo connected area every minute.

[0070] Based on the lightning data in the warning area and the grid point position information of the severe convective echo connected area, judge whether the lightning occurrence position is within the severe convective echo area. If so, increment the lightning frequency count by 1; among them, lightning includes: cloud flash and ground flash;

[0071] Taking minutes as the interval, statistically obtain the lightning frequency LF every minute.

[0072] S302. Based on the lightning frequency LF in the severe convective echo connected area, calculate the lightning time rate of change D and its standard deviation σ.

[0073] S3021. Based on the lightning frequency LF in the severe convective echo connected area, calculate the average value of the minute lightning frequency in the most recent two time units , that is, take the average value of the minute lightning frequency of the current time unit and the previous time unit;

[0074] Step S3022: Repeat Step S3021 to calculate the average values of the minute lightning frequencies in a period of several time units pushed forward from the current time. Each average value corresponds to a specific time interval;

[0075] Step S3023: Based on the average values of the minute lightning frequencies obtained in Step S3021 and Step S3022, successively calculate the differences between adjacent two average values of the minute lightning frequencies to obtain multiple lightning time rates of change D;

[0076] Step S3024: Based on a set of lightning time rates of change D obtained in Step S3023, calculate the standard deviation representing the degree of dispersion of these time rates of change .

[0077] S4. Based on the two-dimensional grid data of the vertically integrated liquid water content VIL and the echo top height TOP, and the grid position information of the severe convective echo connected area, obtain the maximum value VIL of the vertically integrated liquid water content VIL max and VIL max jump increment GVIL and density VILD of VIL.

[0078] Based on the two-dimensional grid data of the vertically integrated liquid water content VIL and the grid position information of the severe convective echo connected area, judge the magnitude of the VIL value in the severe convective echo area for each grid point, and finally obtain the maximum value VIL of VIL max ;

[0079] Based on the two-dimensional grid data of the echo top height TOP and the grid position information of the severe convective echo connected area, judge the magnitude of the TOP value in the severe convective echo area for each grid point, and finally obtain the maximum value TOP of TOP max ;

[0080] Based on the maximum value VIL max , calculate the jump increment GVIL of VIL, and the formula is as follows:

[0081] GVIL = VIL max (i) - VIL max (i - 1),

[0082] where, VIL max (i), VIL max (i - 1) represent the VIL at two adjacent radar time instances max .

[0083] Based on the maximum value VIL max and TOP max , calculate the density VILD of VIL, and the formula is as follows:

[0084] VILD = VIL max / TOP max .

[0085] S5. Based on the lightning time rate of change D and the standard deviation of the lightning time rate of change D judge the lightning jump increment condition.

[0086] Based on the average value of the lightning frequency per minute , the lightning time rate of change D and the standard deviation of the lightning time rate of change , determine whether the lightning jump increment condition is satisfied, that is: and , where, LF thres is the lightning frequency activation threshold; if the average value of the lightning frequency per minute at the current time instance is greater than the activation threshold, and at the same time the lightning frequency time rate of change D is greater than , it indicates that lightning jump occurs, that is, the lightning jump increase condition is satisfied.

[0087] S6. For the time instances that meet the lightning jump increase condition, based on the maximum value VIL of the vertically integrated liquid water content VIL max and the jump increment GVIL and density VILD of VIL max , perform hail warning condition judgment using a support vector machine to achieve hail warning.

[0088] Based on the maximum value VIL of the vertically integrated liquid water content VIL max and its jump increment GVIL and density VILD, determine whether the hail warning condition is met through a support vector machine classification model, that is: input the radar feature parameters (VIL max , GVIL and VILD), if the output is 1, then issue a hail warning, if the output is 0, then do not issue a hail warning; among them, the support vector machine classification model uses a radial basis kernel function and is trained based on lightning jump increase data samples; the lightning jump increase data samples are composed of the radar feature parameters (VIL max , GVIL and VILD) at the adjacent time instances during the lightning jump increase in historical hail weather processes, including both effective warning samples with hail occurring within 1 hour after the lightning jump increase and invalid warning samples without hail occurring within 1 hour after the lightning jump increase.

[0089] Embodiment 2

[0090] Next, this embodiment will be combined to detail how the present invention solves the technical problems in actual work.

[0091] Taking the area of Province A as an example, as Figure 1 shown, the specific steps for implementing a hail warning method based on lightning jump increase and support vector machine include:

[0092] Step S1: Obtain the radar network mosaic data generated by the short-term and nowcasting business system (SWAN) for severe weather operation of the meteorological observatory of Province A, and obtain the two-dimensional grid data of the radar composite reflectivity factor CR, vertically integrated liquid water content VIL, and echo top height TOP.

[0093] Step S2: Obtain the lightning data detected by the three-dimensional lightning location network operated by the lightning protection center of Province A and perform quality control.

[0094] Step S3: Perform binary image processing on the CR two-dimensional grid data obtained in Step S1, and use the "8" neighborhood method to identify the strong convective echo connected area to obtain the grid position information of the connected area.

[0095] Step S4: Based on the lightning data obtained in Step S2 and the grid positions of the severe convective echo area obtained in Step S3, statistically count the lightning frequency LF of the severe convective echo area minute by minute;

[0096] Step S5: Based on the VIL and TOP two-dimensional grid data obtained in Step S1 and the grid positions of the severe convective echo area obtained in Step S3, calculate the maximum value VIL of VIL in the severe convective echo area max 、VIL max jump increment GVIL and VIL max density VILD;

[0097] Step S6: Based on the lightning frequency LF per minute obtained in Step S4, calculate the lightning time variability D and its standard deviation ;

[0098] Step S7: Based on D obtained in Step S6 and , conduct lightning jump increment condition judgment;

[0099] Step S8: For the times that meet the lightning jump increment conditions in Step S7, based on VIL max 、GVIL and VILD obtained in Step S5, use the support vector machine classification model to conduct hail warning condition judgment and obtain the hail warning result.

[0100] Furthermore, in Step S3, the specific steps for identifying the severe convective echo connected area are:

[0101] Step S31: Perform binary image processing on the CR two-dimensional grid data obtained in Step S1, that is: set the grid points with CR≥40dBZ to 1; while set the grid points with CR<40 dBZ to 0;

[0102] Step S32: Use the "8"-neighborhood method to identify the severe convective echo connected area for the binary processed CR two-dimensional grid data obtained in Step S31, that is: take the current severe convective echo grid point as the center, and the 8 grid points above, below, left, right, upper left, upper right, lower left, and lower right are the "8"-neighborhood of the current grid point. Judge whether any grid point in the "8"-neighborhood is a severe convective echo grid point (with a value of 1). If so, this grid point and the central grid point belong to the same connected area. If not, it is a non-connected grid point; loop through this way to judge all the neighborhood grid points of the same connected area until all the neighborhood grid points of the connected area are non-connected grid points, then the judgment and identification of a severe convective echo connected area are completed;

[0103] Furthermore, in Step S4, the method for statistically counting the lightning frequency LF of the severe convective echo area minute by minute is:

[0104] Step S41: Based on the quality-controlled three-dimensional lightning location data obtained in Step S2 and the grid positions of the severe convective echo area obtained in Step S3, determine whether the lightning (cloud flash and ground flash) occurrence location (longitude and latitude) is within the severe convective echo area. If so, increment the lightning frequency count by 1;

[0105] Step S42: Execute Step S41 at minute intervals to statistically obtain the lightning frequency LF per minute;

[0106] Furthermore, in Step S5, calculate the VIL, max GVIL, and VILD of the severe convective echo area. The specific steps are as follows:

[0107] Step S51: Based on the VIL two-dimensional grid data obtained in Step S1 and the grid positions of the severe convective echo area obtained in Step S3, judge the magnitude of the VIL value at each grid point within the severe convective echo area, and finally obtain the maximum value VIL of VIL max , that is: VIL max = max(VIL); where: max is to find the maximum value;

[0108] Step S52: Based on the TOP two-dimensional grid data obtained in Step S1 and the grid positions of the severe convective echo area obtained in Step S3, judge the magnitude of the TOP value at each grid point within the severe convective echo area, and finally obtain the maximum value TOP of TOP max , that is: TOP max = max(TOP); where: max is to find the maximum value;

[0109] Step S53: Based on the VIL max obtained in Step S51, calculate the jump increment GVIL of VIL max , that is: GVIL = VIL max (i) - VIL max (i - 1), where VIL max (i), VIL max (i - 1) represent the VIL of two adjacent radar time instances max ;

[0110] Step S54: Based on the VIL max obtained in Step S51 and the TOP max obtained in Step S52, calculate the density VILD of VIL max , that is: VILD = VIL max / TOP max ;

[0111] Furthermore, in Step S6, the specific steps for calculating D and values are as follows:

[0112] Step S61: Based on the minute-by-minute lightning frequency LF obtained in step S4, calculate the average value of the minute-by-minute lightning frequency in the most recent 2 min (the previous 0 - 2 min), i.e., (unit: min -1 ), that is: , where LF(0) and LF(-1) respectively represent the minute-by-minute lightning frequencies in the previous 0 - 1 min and the previous 1 - 2 min;

[0113] Step S62: Repeat step S61 to calculate a total of 6 average values of the minute-by-minute lightning frequency within the time period of the previous 2 - 14 min, that is: 、 、 、 、 、 ;

[0114] Step S63: Based on the 7 average values of the minute-by-minute lightning frequency obtained in step S61 and step S62, successively calculate the difference between two adjacent average values of the minute-by-minute lightning frequency to obtain the lightning frequency time rate of change D (unit: min -2 ), that is: ; Thus, 6 D values can be obtained, namely: D(0), D(-2), D(-4), D(-6), D(-8), D(-10);

[0115] Step S64: Based on a set of lightning frequency time rates of change D obtained in step S63, calculate the standard deviation representing the dispersion degree of the lightning frequency time rate of change (unit: min -2 ), that is: ; where std is to calculate the standard deviation;

[0116] Furthermore, in step S7, the specific steps for judging the lightning jump condition are as follows:

[0117] Step S71: Based on , D(0) and obtained in step S6, determine whether the lightning jump condition is satisfied, that is: and , where LF thres is the lightning frequency activation threshold (the value is 2 min -1 ); if the average value of the minute-by-minute lightning frequency at the current time is greater than the activation threshold, and at the same time the lightning frequency time rate of change D is greater than , it indicates that a lightning jump occurs, that is, the lightning jump condition is satisfied;

[0118] Furthermore, support vector machine (SVM) is a machine learning algorithm widely used in classification and regression problems, and performs well in dealing with high-dimensional data and small sample size problems. The core idea of ​​SVM is to find a hyperplane that can perfectly separate data of different categories while ensuring that the distance from the data points closest to it to the hyperplane is as far as possible. For linearly inseparable data, support vector machine needs to use kernel techniques to map data to high-dimensional space through kernel functions, thereby converting the original linearly inseparable problem into a linearly separable problem in high-dimensional space. Whether the occurrence of lightning surge can warn hail is also a binary classification problem, namely: effective warning and invalid warning; and compared with other classification problems, hail samples are relatively small. Therefore, based on the radar feature parameters corresponding to effective and invalid lightning surges in historical hail cases, a support vector machine classification model is constructed to assist the lightning surge algorithm in carrying out hail warnings, so as to further improve the problem of high false alarm rate of the lightning surge algorithm.

[0119] Furthermore, in step S8, the specific steps of determining the hail warning conditions are as follows:

[0120] Step S81: Based on the historical hail weather process, the radar characteristic parameters (VIL max , GVIL and VILD) to form a lightning jump sample data set;

[0121] Step S82: Based on the lightning surge sample data set obtained in step S81, each sample is divided into a valid warning sample and an invalid warning sample according to whether hail occurs within 1 hour after the lightning surge occurs; that is, a valid warning sample is one in which hail occurs within 1 hour after the lightning surge occurs, and vice versa, an invalid warning sample is one in which no hail occurs within 1 hour after the lightning surge occurs;

[0122] Step S83: Allocate the lightning surge sample data set obtained in step S81 in a ratio of 8:2 to form a training sample data set and a verification sample data set, and ensure that the ratio of the number of valid warning samples to the number of invalid warning samples in the two sample sets is consistent with the entire lightning surge sample data set;

[0123] Step S84: The support vector machine uses a nonlinear function to map the nonlinear separable problem from the original feature space to a higher-dimensional Hilbert space, thereby converting it into a linear separable problem. At this time, the hyperplane serving as the decision boundary is expressed as follows:

[0124] ,

[0125] In the formula is the mapping function; \(\vec{w}\) represents the normal vector of the hyperplane; \(b\) represents the intercept of the hyperplane; \(T\) represents the transpose operation; \(X\) represents the input sample data.

[0126] Since the mapping function has a complex form and it is difficult to calculate its inner product, the kernel method can be used, that is, define the inner product of the mapping function as the kernel function to avoid the explicit calculation of the inner product:

[0127] ,

[0128] where, \(K(\cdot,\cdot)\) represents the kernel function; \(X_1\) and \(X_2\) both represent the input sample data;

[0129] Based on the training sample data set and the validation sample data set obtained in step S83, a support vector machine with a radial basis kernel function (RBF kernel or Gaussian kernel) is used for training to obtain a support vector machine classification model. Among them, the radial basis kernel function is:

[0130] ,

[0131] where, \(\gamma\) represents a parameter of the radial basis kernel function, which controls the decay speed of the function.

[0132] Step S85: Based on the support vector machine classification model obtained in step S84, perform hail warning condition judgment of the support vector machine (SVM) on the times that meet the lightning jump increase condition in step S7. The judgment function is as follows:

[0133] ,

[0134] That is: if the lightning jump increase condition ( ) is met, then input the radar feature parameters (VIL max , GVIL and VILD) into the support vector machine classification model; if the output is 1, the hail warning condition is met, and if the output is 0, the hail warning condition is not met.

[0135] In this way, the method of the present invention can effectively reduce the problem of more false alarms existing in the application of the traditional lightning jump increase hail warning method.

[0136] In this embodiment, the radar networking mosaic data selected is the composite reflectivity factor (MCR), vertically integrated liquid water content (MVIL), and echo top height (MTOP) generated by the Short-term Nowcasting Business System (SWAN) for severe weather in the business operation of the Meteorological Observatory of Province A. Its grid horizontal resolution is 1 km, and the time resolution is 6 minutes. The radar networking mosaic data is one of the most frequently and importantly used data in the current severe convective weather forecasting and early warning services, and has become an important basis for improving the defense against and accurate release of disaster weather warnings.

[0137] In this embodiment, the lightning data selected is from the Three-dimensional Lightning Location Network of Province A (HBLLS) in the business operation of the Lightning Protection Center of Province A. Its overall detection efficiency is greater than 95%, the horizontal location error is less than 200 m, and the height location error is less than 500 m. Currently, this lightning location network consists of 19 VLF / LF (Very Low Frequency / Low Frequency) lightning locators developed by the Institute of Electrical Engineering, Chinese Academy of Sciences, realizing real-time high-precision networking observations of total lightning (cloud lightning and cloud-to-ground lightning) within the scope of Province A.

[0138] In this embodiment, the hail weather process data selected comes from the direct report information of hail disaster situations verified by the county (city, district)-level meteorological departments and reported through the Meteorological Disaster Management System of the China Meteorological Administration, as well as eyewitness reports released through new media means such as disaster situation reports in official media and Weibo. Among them, in order to ensure the authenticity of the hail process and the usability of the early warning method evaluation, only hail records with exact photo records and real-time location information are selected, and hail reports with fuzzy hail sizes, unclear time and location information, and incomplete lightning data are excluded.

[0139] Next, the application effect of a hail early warning method based on lightning jump increase and support vector machine involved in this application will be analyzed in combination with a hail process in County B, Province A on May 4, 2020, and the overall evaluation results based on 116 hail weather processes in Province A from 2015 to 2023.

[0140] In the afternoon of May 4th, severe convective activities occurred in County B of Province A, and hailstones suddenly fell in Village D of Township C in County B. At 16:12, a convective system developed in the southern part of County B. At this time, the maximum echo intensity > 50 dBZ, accompanied by sporadic lightning activities, and there were two other monomers developing synchronously on the left side of the system. After more than 10 minutes of rapid development, at 16:24, the central intensity of the convective system was further enhanced and exceeded 60 dBZ for the first time. At this time, lightning activities began to be active and concentrated near the strong echo center. At 16:48, the intensity of the convective system still remained above 60 dBZ, the maximum echo intensity reached 65 dBZ, and the lightning activities did not significantly weaken. However, the strong echo center rapidly dropped and touched the ground, indicating that hailstones began to fall during this period. Subsequently, the convective system began to weaken during the eastward movement. At 17:18, the maximum echo intensity of the convective system decreased below 60 dBZ, and the lightning activities increased, and the lightning frequency increased significantly.

[0141] Figures 2 - 4 Minute lightning frequency of the convective system in County B of Province A on May 4th, 2020, the strongest echo ( Figure 2 ), Judgment of lightning jump increase conditions ( Figure 3 ), and time distribution map of judgment of hail warning conditions by support vector machine ( Figure 4 ).

[0142] It can be seen from Figure 2 that the echo intensity of the convective system increased rapidly and reached the peak before the hailstones fell, and then remained above 60 dBZ for about 6 minutes. With the occurrence of hailstones, the strongest echo decreased rapidly and reached the lowest 57 dBZ at 17:06. At the same time, the evolution process of the convective system was also accompanied by obvious lightning activities, and its minute lightning frequency basically exceeded the activation threshold of the algorithm, and the maximum minute lightning frequency occurred after the hailstones fell, exceeding 20 minutes -1 ; among them, there were a total of 7 lightning jumps in the convective system ( Figure 3 ). From the actual hail data (the hailstorm time was 16:48), it can be seen that the first 5 lightning jumps (16:04, 16:16, 16:18, 16:20, 16:42) were effective hail warnings, and the last 2 lightning jumps (16:54 and 17:10) were ineffective hail warnings. Based on the earliest moment of lightning jump increase, the warning lead time for this hail process was about 44 minutes.

[0143] It can be seen from Figure 4 that when the first lightning jump of the convective system (16:04) was approaching (16:00), the radar characteristic parameters (VIL max, GVIL, and VILD) input the support vector machine classification model, and the output result is 1, indicating that this lightning jump is an effective hail warning. This embodiment gives a correct hail warning judgment. Similarly, the method of this embodiment also successfully determines that the 2nd to 5th lightning jumps are effective hail warnings. Among them, the 3rd and 4th lightning jumps are adjacent in time and have the same radar time (16:18). In addition, the output results of the 6th and 7th lightning jumps by the support vector machine classification model are 0, that is, they do not meet the hail warning conditions. This embodiment also successfully identifies these two invalid hail warnings, thus effectively reducing two false hail warnings.

[0144] Table 1 shows the hail warning effect evaluation results of the traditional method of 2σ lightning jump and the improved method based on lightning jump and support vector machine of this embodiment for 116 hail weather processes in Province A from 2015 to 2023. Among them, three parameter indicators, namely the hit rate (POD), false alarm rate (FAR), and critical success index (CSI), are used to evaluate the warning effects of the two methods on hail weather in Province A, that is:

[0145] POD = NA / (NA + NC),

[0146] FAR = NB / (NA + NB),

[0147] CSI = NA / (NA + NB + NC),

[0148] In the formula, NA is the number of samples with successful hail warnings, NB is the number of samples with hail warnings but no actual hail, and NC is the number of samples with actual hail but no hail warnings.

[0149] From the hail warning effect evaluation results of the traditional method of 2σ lightning jump and the improved method based on lightning jump and support vector machine of this application embodiment for 116 hail weather processes in Province A from 2015 to 2023 in Table 1, it can be seen that The POD, FAR, and CSI of the lightning jump method (traditional method) are 93.7%, 41.4%, and 56.4% respectively. Its overall warning effect is close to the evaluation results (POD, FAR, and CSI are 80.2%, 41.6%, and 51.1% respectively) of 177 hail events in a certain region. At the same time, compared with the application effect of this method in a certain country, the evaluation results in Province A show the characteristics of a relatively high hit rate and a relatively high false alarm rate.

[0150] Table 1

[0151] .

[0152] Introduce based on radar characteristic parameters (VIL max, after the support vector machine improvement method for hail NA (GVIL and VILD), although the hail NA decreased from 208 to 180, the change in the hit rate POD was not obvious, only decreasing by 2.8%. Instead, the change in NB was particularly obvious, decreasing from 147 to 67, and its false alarm rate FAR decreased from 41.4% to 27.1%, indicating that the improvement method can effectively reduce the false alarm rate of hail while ensuring the correct rate. In addition, the critical success index CSI increased from 56.4% to 67.9%, which also shows that the improvement method has an overall better hail warning effect, that is, based on radar characteristic parameters (VIL max , GVIL and VILD) to construct a support vector machine classification model to assist in improving the hail warning method based on lightning jumps is feasible.

[0153] In this embodiment, for the actual detection data of the convective system in County B, Province A on May 4, 2020, first read the radar network mosaic data to obtain the two-dimensional grid data of the composite reflectivity factor CR, vertically integrated liquid water content VIL, and echo top height TOP, and perform binary processing on the CR data. Then, use the "8" neighborhood method to identify the strong convective echo connected area, and calculate the maximum value VIL of the VIL in the strong convective echo connected area max and its jump increment GVIL and density VILD. Next, read the lightning data after quality control, and count the lightning frequency per minute in the strong convective echo connected area, and based on the algorithm calculates the lightning time variability D and its standard deviation . Then, judge the lightning jump condition by whether the lightning frequency per minute is greater than the threshold and whether D is greater than . Finally, for the times that meet the lightning jump condition, judge the hail warning condition of the support vector machine based on VIL max , GVIL and VILD to carry out hail warning services. The following conclusions are initially obtained:

[0154] (1) The traditional lightning jump method mainly realizes hail warning based on the frequent jump phenomenon of lightning frequency before hail occurs, and has achieved good warning effects in practical applications. However, non-hail strong convective weather will also show lightning jump phenomena, so there are many false warnings in this method.

[0155] (2) A hail warning method based on lightning jumps and support vector machines involved in this embodiment is based on the lightning jump method, and introduces VIL max, multiple radar characteristic parameters reflecting hail indicators such as GVIL and VILD, and a support vector machine classification model is trained based on the statistical results of a large number of historical hail processes to improve the hail warning effect. From the perspective of the warning results of actual hail processes, the improved method of this application can effectively reduce the false alarm rate FAR while maintaining its probability of detection POD compared with the traditional method, further improving the critical success index CSI, thereby enhancing the hail warning effect. The improved method of this application can provide strong technical support for the hail forecast and warning services using radar and lightning data in meteorological operations.

[0156] From the above conclusions, it can be seen that the method of this embodiment can extract the composite reflectivity factor, vertically integrated liquid water content, and echo top height from the radar network mosaic data, perform binarization processing on the composite reflectivity factor, and then identify the severe convective echo connected area using the "8" neighborhood method in image recognition technology; then, perform quality control on the lightning data and count the minute lightning frequency of the severe convective echo connected area; next, calculate radar characteristic parameters such as the maximum value, jump increment, and density of the vertically integrated liquid water content based on the identified severe convective connected area; finally, based on the judgment result of the lightning jump increment condition, use the support vector machine classification model trained based on the statistical data of historical hail processes to conduct judgment to determine whether it is an effective hail warning, so as to realize the hail warning service based on radar and lightning data.

[0157] Embodiment 3

[0158] This embodiment also provides a hail warning system based on lightning jump increment and support vector machine, including: a collection module, a position acquisition module, a first calculation module, a second calculation module, a judgment module, and a warning module; the collection module is used to obtain the two-dimensional grid data of the radar composite reflectivity factor CR, vertically integrated liquid water content VIL, and echo top height TOP based on the radar network mosaic data of the warning area; the position acquisition module is used to obtain the grid position information of the severe convective echo connected area based on the two-dimensional grid data of the radar composite reflectivity factor CR; the first calculation module is used to calculate the lightning time rate of change D and the standard deviation of the lightning time rate of change D based on the lightning data of the warning area and the grid position information of the severe convective echo connected area ; the second calculation module is used to obtain the maximum value VIL of the vertically integrated liquid water content VIL based on the two-dimensional grid data of the vertically integrated liquid water content VIL and the echo top height TOP and the grid position information of the severe convective echo connected area max and the jump increment GVIL and density VILD of VIL max ; the judgment module is used to judge the lightning jump increment condition based on the lightning time rate of change D and the standard deviation of the lightning time rate of change D ; the warning module is used for the times that meet the lightning jump increment condition, based on the maximum value VIL of the vertically integrated liquid water content VILmax and VIL max The jump increment GVIL and density VILD of max are used to judge the hail warning conditions of the support vector machine, and the hail warning is realized.

[0159] The embodiments described above are only descriptions of the preferred embodiments of the present invention, and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A hail warning method based on lightning jump increase and support vector machine, characterized in that the steps Including: Based on the radar networking mosaic data of the early warning area, obtaining two-dimensional grid data of the combined radar reflectivity factor CR, vertically integrated liquid water content VIL, and echo top height TOP; Based on the two-dimensional grid data of the combined radar reflectivity factor CR, obtaining the grid position information of the severe convective echo connected area; Based on the lightning data of the early warning area and the grid position information of the severe convective echo connected area, calculating the lightning time rate of change D and the standard deviation σ of the lightning time rate of change D; Based on the two-dimensional grid data of the vertically integrated liquid water content VIL and the echo top height TOP and the grid position information of the strong convection echo connected area, the maximum value VIL of the vertically integrated liquid water content VIL is obtained. max And the VIL max The jump amount GVIL and density VILD; Judging the lightning jump increase condition based on the lightning time rate of change D and the standard deviation σ of the lightning time rate of change D; For the time instances that meet the lightning jump increase condition, based on the maximum value VIL of the vertically integrated liquid water content VIL max and the jump increment GVIL and density VILD of the VIL max judge the hail warning conditions of the support vector machine to achieve hail warning.

2. The hail warning method based on lightning jump increase and support vector machine according to claim 1, characterized in that, Using the neighborhood method to obtain the grid position information of the severe convective echo connected area, the steps include: Performing binary image processing on the two-dimensional grid data of the combined radar reflectivity factor CR; For the data after binary image processing, using the neighborhood method to identify the severe convective echo connected area and obtaining the grid position information of the severe convective echo connected area.

3. The hail warning method based on lightning jump increase and support vector machine according to claim 2, wherein The steps of using the neighborhood method to identify the severe convective echo connected area include: Taking the current severe convective echo grid point as the center, taking the 8 adjacent grid points above, below, left, right, upper left, upper right, lower left, and lower right as the neighborhood of the current grid point, and judging whether any grid point in the neighborhood is a severe convective echo grid point. The judgment conditions include: If the grid point is a severe convective echo grid point, it belongs to the same connected area as the center grid point; If the grid point is not a severe convective echo grid point, it is a non-connected grid point; In this way, loop through all grid points in the same connected area to perform neighborhood grid point judgment until all neighborhood grid points in the connected area are non-connected grid points, then the judgment and identification of a connected area are completed.

4. The hail warning method based on lightning jump increase and support vector machine according to claim 1, wherein The steps of calculating the lightning time rate of change D and the standard deviation σ of the lightning time rate of change D include: Based on the lightning data of the early warning area and the grid position information of the severe convective echo connected area, statistically counting the lightning frequency of the severe convective echo connected area every minute; Based on the lightning frequency of the severe convective echo connected area, calculating the lightning time rate of change D and the standard deviation σ of the lightning time rate of change D.

5. The hail warning method based on lightning jump increase and support vector machine according to claim 4, characterized in that, The steps of statistically counting the lightning frequency of the severe convective echo connected area include: Based on the lightning data of the early warning area and the grid position information of the severe convective echo connected area, judging whether the lightning occurrence position is within the severe convective echo area. If so, the lightning frequency count is incremented by 1; where the lightning includes cloud flash and ground flash; Taking minutes as the interval, statistically obtaining the lightning frequency every minute.

6. The hail warning method based on lightning jump increase and support vector machine according to claim 4, characterized in that, The steps of calculating the lightning time rate of change D and the standard deviation σ of the lightning time rate of change D include: Step S61: Based on the lightning frequency of the severe convective echo connected area, calculating the average lightning frequency per minute in the most recent two time units; Step S62: Repeating step S61, calculating the average lightning frequency per minute in a period of several time units pushed forward from the current time. Each average corresponds to a specific time interval; Step S63: Based on the multiple average lightning frequencies per minute obtained in step S61 and step S62, successively calculating the difference between adjacent two average lightning frequencies per minute to obtain multiple lightning time rates of change D; Step S64: Based on a set of lightning time rates of change D obtained in Step S63, calculate the standard deviation σ representing the degree of dispersion of these time rates of change.

7. The hail warning method based on lightning jump increase and support vector machine according to claim 1, characterized in that The method for determining the hail warning conditions using a support vector machine includes: inputting radar characteristic parameters. If the output is 1, a hail warning is issued; if the output is 0, no hail warning is issued. Among them, the support vector machine classification model uses a radial basis kernel function and is trained based on lightning jump data samples.

8. A hail warning system based on lightning jump increase and support vector machine, the system is used to implement the method described in any one of claims 1-7, characterized in that, It includes: a collection module, a position acquisition module, a first calculation module, a second calculation module, a judgment module, and a warning module; The collection module is used to obtain two-dimensional grid data of the radar combined reflectivity factor CR, vertically integrated liquid water content VIL, and echo top height TOP based on the radar network mosaic data in the warning area; The position acquisition module is used to obtain the grid position information of the severe convective echo connected area based on the two-dimensional grid data of the radar combined reflectivity factor CR; The first calculation module is used to calculate the lightning time rate of change D and the standard deviation σ of the lightning time rate of change D based on the lightning data in the warning area and the grid position information of the severe convective echo connected area; The second calculation module is configured to obtain the maximum value VIL of the vertically integrated liquid water content VIL based on the two-dimensional grid data of the vertically integrated liquid water content VIL and the echo top height TOP and the grid position information of the severe convective echo connected region max and the VIL max jump increment GVIL and density VILD thereof; The judgment module is used to judge the lightning jump condition based on the lightning time rate of change D and the standard deviation σ of the lightning time rate of change D; The warning module is used to, for the time instances that meet the lightning jump increase condition, based on the maximum value VIL of the vertically integrated liquid water content VIL max and the VIL max jump increment GVIL and density VILD, perform hail warning condition judgment by means of a support vector machine to achieve hail warning.

9. The hail warning system based on lightning jump increase and support vector machine according to claim 8, characterized in that, The position acquisition module uses the neighborhood method to obtain the grid position information of the severe convective echo connected area. The process includes: Perform binary image processing on the two-dimensional grid data of the radar combined reflectivity factor CR; After binary image processing, use the neighborhood method to identify the severe convective echo connected area and obtain the grid position information of the severe convective echo connected area.

10. The hail warning method based on lightning jump increase and support vector machine according to claim 7, characterized in that, The process of using the neighborhood method to identify the severe convective echo connected area includes: Taking the current severe convective echo grid as the center, taking the eight adjacent grids above, below, left, right, upper left, upper right, lower left, and lower right as the neighborhood of the current grid, and judging whether any grid in the neighborhood is a severe convective echo grid. The judgment conditions include: If the grid is a severe convective echo grid, it belongs to the same connected area as the center grid; If the grid is not a severe convective echo grid, it is a non-connected grid; In this way, loop through all the grid points in the same connected area to perform neighborhood grid point judgment until all the neighborhood grid points in the connected area are non-connected grid points, then the judgment and identification of a connected area are completed.

Images