Multi-dimensional fusion inspection and evaluation method and system for deterministic heavy rainfall forecast

Abstract

The invention relates to a multi-dimensional fusion inspection and evaluation method and system for deterministic heavy rainfall forecast. The method comprises the following steps: acquiring mode rainfall forecast data and site live data of a target evaluation area as to-be-evaluated data; inputting the to-be-evaluated data into a pre-established heavy rainfall multi-dimensional fusion inspection and evaluation system to obtain an evaluation result; and carrying out visual display on the evaluation result. According to the method, a mode rainfall forecast process stability evaluation method and a spatial inspection similarity inspection method suitable for heavy rainfall characteristics are added on the basis of a traditional inspection method, and more inspection dimensions are added for mode rainfall forecast. The invention further constructs a multi-dimensional fusion inspection and evaluation method, multi-dimensional inspection indexes can be displayed through graphs, and mode research and development and a user can visually know the forecasting ability of each dimension of the mode; and a formed comprehensive test evaluation index (CEI) reflects a comprehensive evaluation result of the six core rainfall forecasting capacities of the mode.

Description

Technical Field

[0001] The present invention relates to the technical field related to atmospheric science, and in particular to a multidimensional fusion verification and evaluation method and system for deterministic heavy rainfall forecasts. Background Art

[0002] The precipitation forecast verification methods commonly used in the field of atmospheric science can be divided into the following three categories:

[0003] (1) Deterministic prediction test of continuous variables. Specific test methods include: relative error, absolute error, mean square error, correlation coefficient, etc.

[0004] (2) Deterministic prediction test of binary events. Specific test methods include: accuracy, TS score, BIAS score, hit rate, missed rate, and false alarm rate.

[0005] (3) Spatial field prediction verification. Common verification methods include neighborhood method, MODE, CRA, etc.

[0006] The deterministic forecast test of continuous variables reflects the deviation of the forecast value relative to the actual value; the deterministic forecast test of binary events reflects the presence or absence of a forecast for the actual event; and the spatial field forecast test reflects the deviation in spatial characteristics between the actual field and the forecast field.

[0007] An example is used to illustrate the characteristics of the three evaluation methods mentioned above. A weather station A recorded a 24-hour cumulative precipitation of 47 mm, and a weather station B, 2 km away, recorded a cumulative precipitation of 52 mm. The model predicted that the precipitation at both stations A and B was 52 mm. To evaluate the model's ability to predict heavy rain (accumulated precipitation of more than 50 mm in 24 hours) at station A, the first type of test method showed that the error in the precipitation forecast was only 5 mm, which was not a large deviation. However, according to the second type of test method, it was determined that the model failed to accurately predict the heavy rain, but instead falsely reported heavy rain at station A. Combined with the third type of test method, the model accurately predicted the heavy rain area within 5 km of station A, indicating that the model's overall forecast effect for heavy rainfall in the opposite direction is still relatively good. It can be seen that different test methods only express the evaluation significance of different dimensions, and a single test method cannot contain all the information about the model's forecast performance.

[0008] In addition, the evaluation of the model's precipitation forecasting capability also requires the addition of verification methods in the time dimension, such as the forecast stability of the model for multiple consecutive days and the accuracy of hourly precipitation trend forecasts.

[0009] For precipitation grid forecasts based on deterministic models, there is currently a lack of a comprehensive multi-dimensional heavy precipitation verification method to fully evaluate the model capabilities. Summary of the Invention

[0010] The purpose of the present invention is to address at least one of the deficiencies of the prior art and to provide a multi-dimensional fusion verification and evaluation method and system for deterministic heavy rainfall forecasts.

[0011] In order to achieve the above object, the present invention adopts the following technical solutions:

[0012] Specifically, a multi-dimensional fusion verification and evaluation method for deterministic heavy precipitation forecast is proposed, including the following:

[0013] Obtain model precipitation forecast data and site actual data in the target assessment area as data to be assessed;

[0014] Input the data to be evaluated into the pre-established heavy rainfall multi-dimensional fusion inspection and evaluation system to obtain the evaluation results;

[0015] Visually displaying the evaluation results;

[0016] Specifically, the process of establishing a multi-dimensional fusion inspection and evaluation system includes:

[0017] Establish deterministic forecast test modules for continuous variables and binary events,

[0018] Based on the process prediction stability test method of multidimensional test, a model process prediction stability test module is established.

[0019] Establish the precipitation spatial test similarity of multidimensional test, and build a spatial test module based on it.

[0020] Based on the deterministic forecast verification module, process forecast stability verification module and spatial verification module, the multidimensional fusion inspection and evaluation indicators of heavy rainfall are determined and a multidimensional fusion inspection and evaluation system is established.

[0021] Furthermore, specifically, a deterministic forecast test module for continuous variables and a deterministic forecast test module for binary events are established, including:

[0022] Establish the relative error, absolute error, and mean square error of the cumulative precipitation forecast; establish the cumulative precipitation classification TS score, BIAS score, hit rate, missed rate, and false alarm rate; establish the hourly precipitation forecast accuracy and correlation coefficient, and then establish a deterministic forecast verification module.

[0023] Furthermore, specifically, a process prediction stability test module based on multidimensional test is established, including:

[0024] The forecast standard deviation, range, and process forecast error are used as test indicators for the stability of the model process forecast. A model process forecast stability test module is established based on the test indicators.

[0025] The calculation method of the forecast standard deviation is:

[0026]

[0027] Where, f i is the precipitation value reported for the i-th time in the process, is the average of the n reported precipitation values,

[0028] The range is calculated as follows:

[0029] R=f max -f min ,

[0030] Where, f max With f min are the maximum and minimum precipitation values ​​among the n reported precipitation values,

[0031] The process forecast error is calculated as follows:

[0032]

[0033] Where, is the average of the n reported precipitation values, and o is the actual precipitation value.

[0034] Furthermore, the model process prediction stability test module is also used to calculate the comprehensive index of process stability.

[0035] The process stability comprehensive index is obtained by adding the forecast standard deviation, range, and absolute value of the process forecast error according to preset weights. The smaller the value of the process stability comprehensive index, the better the forecast stability of the model and the process forecast effect.

[0036] Furthermore, specifically, a multidimensional test of precipitation spatial similarity is established, including:

[0037] Target object identification: Process the grid data in the precipitation field into a two-dimensional array of size xDim*yDim, and use Gaussian threshold filtering to identify and merge precipitation objects to obtain forecast and actual objects;

[0038] After obtaining the predicted and live target objects, the live object is searched within a square with the geometric center of the predicted object as the center. The specific method is to obtain the bounding rectangle of the predicted object, and then extend the grid width of the preset value outward from each side of the bounding rectangle to obtain a search rectangle. If the live object falls within the search rectangle, then the live object is added to the score set of the current predicted object. If there is no live object match, the report is empty; if there are multiple live objects that match, they are scored and the object with the highest score is selected to generate an object pair;

[0039] The scoring is performed through the precipitation spatial test similarity of the multidimensional test. Specifically, the precipitation spatial test similarity of the multidimensional test is as follows:

[0040]

[0041] Through evaluation experiments, the value of M is 4.

[0042] F 1,j is the overlapping area ratio of the j-th object pair, F 1,j = [number of crossed grid points / (actual grid points + predicted grid points)]*2

[0043] F 2,j is the area ratio of the j-th object pair:

[0044] When 0<=R<=0.8, F 2,j =R / 0.8,

[0045] When R>0.8, F 2,j =1,

[0046] Wherein, R is the area of ​​the forecast object, i.e., the total number of grid points, divided by the area of ​​the live object, i.e., the total number of grid points, or the area of ​​the live object, i.e., the total number of grid points, divided by the area of ​​the forecast object, i.e., the total number of grid points, where the ratio is less than 1;

[0047] F 3,j is the long axis angle difference of the j-th object pair, i.e., the cosine value of the angle between the longest axis of the predicted object geometry and the longest axis of the actual object geometry,

[0048] F 3,j =|COS(angleDelta)|,

[0049] F 4,j is the geometric center distance of the j-th object pair,

[0050] Specific parameters: distance D is the distance between the geometric center of the actual object and the geometric center of the predicted object, the optimal distance Dmin, the maximum tolerance distance Dmax,

[0051] When D <= Dmin: F 4,j =1,

[0052] When Dmin <D<=Dmax:F 4,j =1-(D-Dmin) / [Dmax-Dmin],

[0053] When D>Dmax:F 4,j =0,

[0054] Here, set Dmin = 30KM, Dmax = 300KM,

[0055] Function weight w: 0.4, 0.3, 0.1, 0.2,

[0056] Confidence c: C1 = 1, C2 = 1, C4 = the ratio of small area to large area in the actual and forecast objects, where r f and r o are the aspect ratios of the forecast and observation objects respectively. When r f or r o When it approaches 1, the value of the credibility function c3 is close to 0.

[0057] Furthermore, specifically, the heavy rainfall multi-dimensional fusion test evaluation indicators are determined, including:

[0058] Based on the experiment of multiple heavy precipitation case tests and evaluations, the 24-h mean absolute error of model precipitation forecast, TS score for heavy rain or above, rainstorm forecast deviation, hourly clear and rainy forecast accuracy, rainstorm spatial similarity, and comprehensive process stability index are selected from the output results of the deterministic forecast test module, the process forecast stability test module, and the spatial test module for dimensionless fusion to construct a multidimensional fusion test and evaluation index.

[0059] Furthermore, the method also includes performing equal-weighted arithmetic averaging on the dimensionless fused 24-hour mean absolute error, TS score for heavy rain or above, rainstorm forecast deviation, hourly clear and rainy forecast accuracy, rainstorm spatial similarity, and comprehensive process stability index to obtain a comprehensive inspection and evaluation index CEI. The higher the CEI score, the better the model's multi-dimensional fusion forecast capability for heavy rainfall.

[0060] Furthermore, specifically, the way to perform visual display is:

[0061] The evaluation results are formed into a radar chart, and the radar chart and the CEI value are visually displayed.

[0062] The present invention also proposes a multi-dimensional fusion verification and evaluation system for deterministic heavy rainfall forecast, including the following:

[0063] A data acquisition module is used to obtain model precipitation forecast data and site actual data of the target assessment area as data to be assessed;

[0064] Deterministic forecast test module, used to perform deterministic forecast test of continuous variables and deterministic forecast test of binary events on the data to be evaluated;

[0065] The model process prediction stability test module is used to test the model process prediction stability of the evaluation data based on the process prediction stability test method of multidimensional test;

[0066] Establish a spatial verification module to perform precipitation spatial similarity test on the data to be evaluated based on the precipitation spatial similarity of multidimensional test;

[0067] A module for determining the multidimensional fusion test and evaluation index for heavy rainfall is used to determine the multidimensional fusion test and evaluation index for heavy rainfall based on the deterministic forecast test module, the process forecast stability test module, and the spatial test module, and then obtain the evaluation results;

[0068] The visualization display module is used to visualize the evaluation results.

[0069] The beneficial effects of the present invention are:

[0070] The present invention proposes a multi-dimensional fusion inspection and evaluation method and system for deterministic heavy precipitation forecasts, which overcomes the one-sidedness of the existing single precipitation inspection method mentioned in the background art, which can only evaluate the forecasting ability of a single dimension of the model. On the basis of the traditional inspection method, the present invention supplements the model precipitation forecast process stability assessment method and the spatial inspection similarity inspection method applicable to heavy precipitation characteristics, adding more inspection dimensions to the model precipitation forecast. The present invention also further constructs a multi-dimensional fusion inspection and evaluation method, and the multi-dimensional inspection indicators can be displayed graphically, so that model developers and users can intuitively understand the forecasting ability of each dimension of the model; the formed comprehensive inspection and evaluation index (CEI) reflects the comprehensive evaluation results of the model's six core precipitation forecasting capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] The above and other features of the present disclosure will become more apparent through a detailed description of the embodiments shown in conjunction with the accompanying drawings. The same reference numerals in the drawings of the present disclosure represent the same or similar elements. Obviously, the drawings described below are only some embodiments of the present disclosure. It is possible for a person skilled in the art to derive other drawings based on these drawings without inventive effort. In the drawings:

[0072] Figure 1 Shown is a flow chart of the multi-dimensional fusion verification and evaluation method for deterministic heavy rainfall forecast of the present invention;

[0073] Figure 2 FIG2 is an algorithm principle diagram of the multi-dimensional fusion test and evaluation method for deterministic heavy rainfall forecast according to the present invention;

[0074] Figure 3 Shown is a diagram showing evaluation results of the present invention in a specific application. DETAILED DESCRIPTION

[0075] The following will be combined with the embodiments and drawings to clearly and completely describe the concept, specific structure and technical effects of the present invention so as to fully understand the purpose, scheme and effect of the present invention. It should be noted that the embodiments and features in the embodiments of this application can be combined with each other unless there is a conflict. The same reference numerals used throughout the drawings indicate the same or similar parts.

[0076] Example 1, with reference to Figure 1 as well as Figure 2 The present invention proposes a multi-dimensional fusion verification and evaluation method for deterministic heavy rainfall forecast, including the following:

[0077] Obtain model precipitation forecast data and site actual data in the target assessment area as data to be assessed;

[0078] Input the data to be evaluated into the pre-established heavy rainfall multi-dimensional fusion inspection and evaluation system to obtain the evaluation results;

[0079] Visually displaying the evaluation results;

[0080] Specifically, the process of establishing a multi-dimensional fusion inspection and evaluation system includes:

[0081] Establish deterministic forecast test modules for continuous variables and binary events,

[0082] Based on the process prediction stability test method of multidimensional test, a model process prediction stability test module is established.

[0083] Establish the precipitation spatial test similarity of multidimensional test, and build a spatial test module based on it.

[0084] Based on the deterministic forecast verification module, process forecast stability verification module and spatial verification module, the multidimensional fusion inspection and evaluation indicators of heavy rainfall are determined and a multidimensional fusion inspection and evaluation system is established.

[0085] In this embodiment 1, the one-sidedness of the single-dimensional test method in evaluating the precipitation forecast capability of the model is overcome, and a multi-dimensional fusion precipitation forecast evaluation method is provided, which can comprehensively evaluate the heavy precipitation forecast capability of the deterministic model and intuitively compare the forecast capabilities of various models.

[0086] As a preferred embodiment of the present invention, specifically, a deterministic forecast test module for continuous variables and a deterministic forecast test module for binary events are established, including:

[0087] Establish the relative error, absolute error, and mean square error of cumulative precipitation forecasts; establish the TS score, BIAS score, hit rate, missed rate, and false alarm rate of cumulative precipitation; establish the accuracy rate and correlation coefficient of hourly precipitation forecasts; and then establish a deterministic forecast verification module;

[0088] The correlation coefficient refers to the statistical Where Oi is the station observation value, Gi is the station forecast value, and N is the total number of samples participating in the test (number of stations). In addition, most of the indicators in the deterministic forecast verification module are not used to synthesize multidimensional fusion test evaluation indicators in the following text. These indicators are only used to establish the deterministic forecast verification module to provide data support for subsequent analysis.

[0089] As a preferred embodiment of the present invention, specifically, a process prediction stability test module based on a multi-dimensional test process prediction stability test method is established, including:

[0090] The forecast standard deviation, range, and process forecast error are used as test indicators for the stability of the model process forecast. A model process forecast stability test module is established based on the test indicators.

[0091] The calculation method of the forecast standard deviation is:

[0092]

[0093] Where, f i is the precipitation value reported for the i-th time in the process, is the average of the n reported precipitation values,

[0094] The range is calculated as follows:

[0095] R=f max -f min ,

[0096] Where, f max With f min are the maximum and minimum precipitation values ​​among the n reported precipitation values,

[0097] The process forecast error is calculated as follows:

[0098]

[0099] Where, is the average of the n reported precipitation values, and o is the actual precipitation value.

[0100] Usually, n≥3.

[0101] As a preferred embodiment of the present invention, the model process prediction stability test module is also used to calculate the comprehensive index of process stability.

[0102] The process stability comprehensive index is obtained by adding the forecast standard deviation, range, and absolute value of the process forecast error according to preset weights. The smaller the value of the process stability comprehensive index, the better the forecast stability of the model and the process forecast effect.

[0103] In this preferred embodiment, the absolute values ​​of the above three indicators are added according to the weights (2:1:2) to form a comprehensive index of process stability. The smaller the value, the better the prediction stability of the model and the process prediction effect.

[0104] As a preferred embodiment of the present invention, specifically, a multidimensional test of precipitation spatial similarity is established, including:

[0105] Target object identification: Process the grid data within the precipitation field into a two-dimensional array of size xDim*yDim. Use Gaussian threshold filtering to identify and merge precipitation objects to obtain forecast and actual objects. (The test requires obtaining both the forecast and actual precipitation field data. The objects identified and merged from the forecast grid data are the forecast objects; the objects identified and merged from the actual grid data are the actual objects.)

[0106] After obtaining the predicted and live target objects, the live object is searched within a square with the geometric center of the predicted object as the center. The specific method is to obtain the bounding rectangle of the predicted object, and then extend the grid width of the preset value outward from each side of the bounding rectangle to obtain a search rectangle. If the live object has a point falling within the search rectangle, then the live object is added to the score set of the current predicted object. If there is no live object match, the report is empty; if there are multiple live objects that match, they are scored and the object with the highest score is selected to form an object pair; (where the points in the live object and the predicted object are grid points)

[0107] The scoring is performed through the precipitation spatial test similarity of the multidimensional test. Specifically, the precipitation spatial test similarity of the multidimensional test is as follows:

[0108]

[0109] Through evaluation experiments, the value of M is 4.

[0110] F 1,j is the overlapping area ratio of the j-th object pair, F 1,j = [number of crossed grid points / (actual grid points + predicted grid points)]*2

[0111] F 2,j is the area ratio of the j-th object pair:

[0112] When 0<=R<=0.8, F 2,j =R / 0.8,

[0113] When R>0.8, F 2,j =1,

[0114] Wherein, R is the area of ​​the forecast object, i.e., the total number of grid points, divided by the area of ​​the live object, i.e., the total number of grid points, or the area of ​​the live object, i.e., the total number of grid points, divided by the area of ​​the forecast object, i.e., the total number of grid points, where the ratio is less than 1;

[0115] F 3,j is the long axis angle difference of the j-th object pair, i.e., the cosine value of the angle between the longest axis of the predicted object geometry and the longest axis of the actual object geometry,

[0116] F 3,j =|COS(angleDelta)|,

[0117] F 4,j is the geometric center distance of the j-th object pair,

[0118] Specific parameters: distance D is the distance between the geometric center of the actual object and the geometric center of the predicted object, the optimal distance Dmin, the maximum tolerance distance Dmax,

[0119] When D <= Dmin: F 4,j =1,

[0120] When Dmin <D<=Dmax:F 4,j =1-(D-Dmin) / [Dmax-Dmin],

[0121] When D>Dmax:F 4,j =0,

[0122] Here, set Dmin = 30KM, Dmax = 300KM,

[0123] Function weight w: 0.4, 0.3, 0.1, 0.2,

[0124] Confidence c: C1 = 1, C2 = 1, C4 = the ratio of small area to large area in the actual and forecast objects, where r f and r o are the aspect ratios of the forecast and observation objects respectively. When r f or r o When it approaches 1, the value of the credibility function c3 is close to 0.

[0125] In this preferred embodiment, the space inspection mainly adopts the MODE method, and the algorithm is mainly divided into three basic steps: target object recognition, target object matching, and target object inspection.

[0126] Target object identification: Grid data within the precipitation area is processed into a two-dimensional array of size xDim*yDim, using a 5km*5km resolution. Gaussian threshold filtering is used to identify and merge precipitation objects. Based on the characteristics of local rainstorms, valid targets must exist only if the merged area has more than 200 grid points (approximately 70km*70km).

[0127] After obtaining the forecast and live target objects, we search for live precipitation objects within a square centered around the forecast object's geometric center. Specifically, we obtain the bounding rectangle of the forecast object and then extend it outward by 10 grid points (set as the search_radius parameter in the program) from each side to create a search rectangle. If any live object falls within the search rectangle, it is added to the score set for the current forecast object. If no object matches, the report is empty; if multiple objects match, they are scored, and the highest-scoring object is selected to form an object pair.

[0128] As a preferred embodiment of the present invention, specifically, determining the heavy rainfall multi-dimensional fusion test evaluation index includes:

[0129] Based on the experiment of multiple heavy precipitation case tests and evaluations, the 24-h mean absolute error of model precipitation forecast, TS score for heavy rain or above, rainstorm forecast deviation, hourly clear and rainy forecast accuracy, rainstorm spatial similarity, and comprehensive process stability index are selected from the output results of the deterministic forecast test module, the process forecast stability test module, and the spatial test module for dimensionless fusion to construct a multidimensional fusion test and evaluation index.

[0130] The six indicators here are derived from the basic inspection indicators mentioned above. Basic indicators correspond to different precipitation levels, and can be expressed as "TS for heavy rain or above" or "TS for rainstorm or above." These six indicators were selected based on the demand for rainstorm assessment in weather forecast inspection operations and are the priority evaluation dimensions of the current project.

[0131] The 24-hour average absolute error, TS score for heavy rain or above, heavy rain forecast deviation, hourly clear and rainy forecast accuracy, and heavy rain spatial similarity correspond to the 24-hour cumulative precipitation average absolute error, 24-hour TS score for heavy rain or above, 24-hour heavy rain forecast deviation, 24-hour hourly clear and rainy forecast accuracy, and 24-hour heavy rain magnitude spatial similarity, respectively;

[0132] As a preferred embodiment of the present invention, the method also includes performing equal-weighted arithmetic averaging on the dimensionless fused 24-hour mean absolute error, TS score for heavy rain or above, rainstorm forecast deviation, hourly clear and rainy forecast accuracy, rainstorm spatial similarity, and comprehensive process stability index to obtain a comprehensive inspection and evaluation index CEI. The higher the CEI score, the better the model's multi-dimensional fusion forecast capability for heavy rainfall.

[0133] The scores of the above indicators are all dimensionlessly fused, with scores ranging from 0 to 100. The closer to 100, the better the performance of the model in this evaluation dimension. The specific fusion methods of each indicator are shown in Table 1:

[0134]

[0135] Table 1 In addition, an application example of the present invention in a specific application is listed as follows: A multi-dimensional fusion test and comparative evaluation of the 2022 Guangdong Province heavy rain forecast results of GIFT, CMA-GFS, CMA-TRAMS, CMA-GD and ECMWF was conducted. Figure 1 Check the process steps.

[0136] Step A: Obtain the 00-96h precipitation grid forecast values ​​reported by the above subjective and objective models starting at 12UTC in 2022.

[0137] Step B performs a multi-dimensional fusion test of heavy rainfall.

[0138] Step C obtains the precipitation multi-dimensional fusion inspection and evaluation results, including radar maps and comprehensive inspection and evaluation index (CEI).

[0139] Step B is specifically as follows:

[0140] Step B1: Conduct deterministic forecast tests for continuous variables, deterministic forecast tests for binary events, process forecast stability tests, and precipitation spatial similarity tests, and obtain the following test results: 24-hour mean absolute error of precipitation, 24-hour TS of heavy rain or above, 24-hour rainstorm forecast deviation, hourly clear and rainy forecast accuracy, rainstorm spatial similarity, comprehensive index of process stability, etc.

[0141] Step B2: Dimensionlessly fuse the 24-hour average absolute error of precipitation, 24-hour TS of heavy rain or above, rainstorm forecast deviation, hourly clear and rainy forecast accuracy, rainstorm spatial similarity, and comprehensive process stability indicators.

[0142] Step B3: Based on the 24-hour average absolute error of precipitation, 24-hour TS of heavy rain or above, rainstorm forecast deviation, hourly clear and rainy forecast accuracy, rainstorm spatial similarity, and comprehensive process stability indicators, a multidimensional inspection and evaluation chart is drawn, and a comprehensive inspection and evaluation index (CEI) is generated.

[0143] Technical effect: A comprehensive assessment was made of the forecast capabilities of various subjective and objective forecast products for different types of heavy rainfall in 2022. By comparing the models, the forecast characteristics of each model in different dimensions can be discovered horizontally.

[0144] The multidimensional test evaluation chart finally obtained in this embodiment is as follows Figure 3 As shown; the comprehensive inspection and evaluation index (CEI) is generated as shown in Table 2 below (the comprehensive inspection and evaluation index of each subjective and objective mode of the 2022 rainstorm classification).

[0145] GIFT ECMWF CMA-GFS CMA-TRAMS CMA-GD Frontal rainstorm 72.4 66.6 46.2 70.5 75.1 Warm-zone rainstorm 56.8 66.1 43.3 68.2 66.3 Tropical cyclone-type heavy rain 66.9 60.8 39.5 73.0 66.5

[0146] Table 2

[0147] As a preferred embodiment of the present invention, specifically, the way of visual display is:

[0148] The evaluation results are formed into a radar chart, and the radar chart and the CEI value are visually displayed.

[0149] The present invention also proposes a multi-dimensional fusion verification and evaluation system for deterministic heavy rainfall forecast, including the following:

[0150] A data acquisition module is used to obtain model precipitation forecast data and site actual data of the target assessment area as data to be assessed;

[0151] Deterministic forecast test module, used to perform deterministic forecast test of continuous variables and deterministic forecast test of binary events on the data to be evaluated;

[0152] The model process prediction stability test module is used to test the model process prediction stability of the evaluation data based on the process prediction stability test method of multidimensional test;

[0153] Establish a spatial verification module to perform precipitation spatial similarity test on the data to be evaluated based on the precipitation spatial similarity of multidimensional test;

[0154] A module for determining the multidimensional fusion test and evaluation index for heavy rainfall is used to determine the multidimensional fusion test and evaluation index for heavy rainfall based on the deterministic forecast test module, the process forecast stability test module, and the spatial test module, and then obtain the evaluation results;

[0155] The visualization display module is used to visualize the evaluation results.

[0156] In addition, the functional modules in various embodiments of the present invention may be integrated into a single processing module, or each module may exist physically separately, or two or more modules may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or software functional modules.

[0157] If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or system that can carry the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0158] Although the present invention has been described in considerable detail and with particularity with respect to several described embodiments, it is not intended to be limited to any of these details or embodiments or any particular embodiment, but rather should be construed as providing a broad possible interpretation of these claims in view of the prior art by reference to the appended claims, thereby effectively encompassing the intended scope of the invention. In addition, the invention has been described above in terms of embodiments foreseen by the inventors for the purpose of providing a useful description, and those insubstantial modifications of the invention that are not currently foreseen may still represent equivalent modifications of the invention.

[0159] The above description is merely a preferred embodiment of the present invention. The present invention is not limited to the above-described embodiments. As long as the technical effects of the present invention are achieved by the same means, they shall fall within the scope of protection of the present invention. Within the scope of protection of the present invention, various modifications and variations of the technical solutions and / or implementation methods may be made.

Claims

1. A multi-dimensional fusion test and evaluation method for deterministic heavy precipitation forecast, characterized by: These include: Obtain model precipitation forecast data and site actual data in the target assessment area as data to be assessed; Input the data to be evaluated into the pre-established heavy rainfall multi-dimensional fusion inspection and evaluation system to obtain the evaluation results; Visually displaying the evaluation results; Specifically, the process of establishing a multi-dimensional fusion inspection and evaluation system includes: Establish deterministic forecast test modules for continuous variables and binary events, Based on the process prediction stability test method of multidimensional test, a model process prediction stability test module is established. Establish the precipitation spatial test similarity of multidimensional test, and build a spatial test module based on it. Based on the deterministic forecast verification module, process forecast stability verification module and spatial verification module, the multidimensional fusion inspection and evaluation indicators of heavy rainfall are determined and a multidimensional fusion inspection and evaluation system is established.

2. The multidimensional fusion verification and evaluation method for deterministic heavy rainfall forecast according to claim 1 is characterized in that: Specifically, a deterministic forecast test module for continuous variables and a deterministic forecast test module for binary events are established, including: Establish the relative error, absolute error, and mean square error of the cumulative precipitation forecast; establish the cumulative precipitation classification TS score, BIAS score, hit rate, missed rate, and false alarm rate; establish the hourly precipitation forecast accuracy and correlation coefficient, and then establish a deterministic forecast verification module.

3. The multidimensional fusion verification and evaluation method for deterministic heavy rainfall forecast according to claim 1 is characterized in that: Specifically, a process prediction stability test module based on multidimensional test is established, including: The forecast standard deviation, range, and process forecast error are used as test indicators for the stability of the model process forecast. A model process forecast stability test module is established based on the test indicators. The calculation method of the forecast standard deviation is: Where, f i is the precipitation value reported for the i-th time in the process, is the average of the n reported precipitation values, The range is calculated as follows: R=f max -f min , Where, f max With f min are the maximum and minimum precipitation values ​​among the n reported precipitation values, The process forecast error is calculated as follows: Where, is the average of the n reported precipitation values, and o is the actual precipitation value.

4. The multidimensional fusion verification and evaluation method for deterministic heavy rainfall forecast according to claim 3 is characterized in that: The model process prediction stability test module is also used to calculate the comprehensive index of process stability. The process stability comprehensive index is obtained by adding the forecast standard deviation, range, and absolute value of the process forecast error according to preset weights. The smaller the value of the process stability comprehensive index, the better the forecast stability of the model and the process forecast effect.

5. The multidimensional fusion verification and evaluation method for deterministic heavy rainfall forecast according to claim 1 is characterized in that: Specifically, a multidimensional test of precipitation spatial similarity is established, including: Target object identification: Process the grid data in the precipitation field into a two-dimensional array of size xDim*yDim, and use Gaussian threshold filtering to identify and merge precipitation objects to obtain forecast and actual objects; After obtaining the predicted and live target objects, the live object is searched within a square with the geometric center of the predicted object as the center. The specific method is to obtain the bounding rectangle of the predicted object, and then extend the grid width of the preset value outward from each side of the bounding rectangle to obtain a search rectangle. If the live object falls within the search rectangle, then the live object is added to the score set of the current predicted object. If there is no live object match, the report is empty; if there are multiple live objects that match, they are scored and the object with the highest score is selected to generate an object pair; The scoring is performed through the precipitation spatial test similarity of the multidimensional test. Specifically, the precipitation spatial test similarity of the multidimensional test is as follows: Through evaluation experiments, the value of M is 4. F 1,j is the overlapping area ratio of the j-th object pair, F 1,j = [number of crossed grid points / (actual grid points + predicted grid points)]*2 F 2,j is the area ratio of the j-th object pair: When 0<=R<=0.8, F 2,j =R / 0.8, When R>0.8, F 2,j =1, Wherein, R is the area of ​​the forecast object, i.e., the total number of grid points, divided by the area of ​​the live object, i.e., the total number of grid points, or the area of ​​the live object, i.e., the total number of grid points, divided by the area of ​​the forecast object, i.e., the total number of grid points, where the ratio is less than 1; F 3,j The angle difference of the long axis of the j-th object pair is the cosine value of the angle between the longest axis of the predicted object geometry and the longest axis of the actual object geometry. F 3,j =|COS(angleDelta)| F 4,j is the geometric center distance of the j-th object pair, Specific parameters: distance D is the distance between the geometric center of the actual object and the geometric center of the predicted object, the optimal distance Dmin, the maximum tolerance distance Dmax, When D <= Dmin: F 4,j =1, When Dmin < D <= Dmax: F 4,j = 1 - (D - Dmin) / [Dmax - Dmin], When D>Dmax:F 4,j =0, Function weight w: 0.4, 0.3, 0.1, 0.2, Confidence c: C1 = 1, C2 = 1, C4 = the ratio of small area to large area in the actual and forecast objects, where r f and r o are the aspect ratios of the forecast and observation objects respectively. When r f or r o When it approaches 1, the value of the credibility function c3 is close to 0.

6. The multidimensional fusion verification and evaluation method for deterministic heavy rainfall forecast according to claim 1 is characterized in that: Specifically, determine the heavy rainfall multi-dimensional fusion test evaluation indicators, including: Based on the experiment of multiple heavy precipitation case tests and evaluations, the 24-h mean absolute error of model precipitation forecast, TS score for heavy rain or above, rainstorm forecast deviation, hourly clear and rainy forecast accuracy, rainstorm spatial similarity, and comprehensive process stability index are selected from the output results of the deterministic forecast test module, the process forecast stability test module, and the spatial test module for dimensionless fusion to construct a multidimensional fusion test and evaluation index.

7. The multidimensional fusion verification and evaluation method for deterministic heavy rainfall forecast according to claim 6 is characterized in that: The method also includes performing equal-weighted arithmetic averaging on the dimensionless fused 24-hour mean absolute error, the TS score for heavy rain or above, the rainstorm forecast deviation, the hourly clear and rainy forecast accuracy, the rainstorm spatial similarity, and the comprehensive index of process stability to obtain a comprehensive test evaluation index CEI. The higher the CEI score, the better the model's multi-dimensional fusion forecast capability for heavy rainfall.

8. The multi-dimensional fusion verification and evaluation method for deterministic heavy rainfall forecast according to claim 7 is characterized in that: Specifically, the visual display method is: The evaluation results are formed into a radar chart, and the radar chart and the CEI value are visually displayed.

9. A multi-dimensional fusion verification and evaluation system for deterministic heavy rainfall forecasts, characterized by: These include: A data acquisition module is used to obtain model precipitation forecast data and site actual data of the target assessment area as data to be assessed; Deterministic forecast test module, used to perform deterministic forecast test of continuous variables and deterministic forecast test of binary events on the data to be evaluated; The model process prediction stability test module is used to test the model process prediction stability of the evaluation data based on the process prediction stability test method of multidimensional test; Establish a spatial verification module to perform precipitation spatial similarity test on the data to be evaluated based on the precipitation spatial similarity of multidimensional test; A module for determining the multidimensional fusion test and evaluation index for heavy rainfall is used to determine the multidimensional fusion test and evaluation index for heavy rainfall based on the deterministic forecast test module, the process forecast stability test module, and the spatial test module, and then obtain the evaluation results; The visualization display module is used to visualize the evaluation results.

Images