Method and device for testing forecast mode, electronic equipment and program product

Abstract

The invention provides a forecasting mode forecasting test method and device, electronic equipment and a program product. The method comprises the steps that target parameters input by a user are obtained, the target parameters comprise a target area, meteorological elements and a target time period, multiple pieces of observation data of the target area in the target time period and multiple pieces of forecasting data corresponding to multiple forecasting modes respectively are determined, and according to the meteorological elements, multiple pieces of forecasting data corresponding to multiple forecasting modes are obtained; the plurality of observation data and the plurality of forecast data are checked, a check index value corresponding to each forecast mode is obtained, a degradation factor corresponding to each forecast mode is determined according to the check index value, and the degradation factor is used for representing the driving contribution degree of meteorological elements to forecast skill degradation in the forecast mode. And according to the plurality of test index values and the plurality of degradation factors, visual test results corresponding to the plurality of forecasting modes are generated, so that the accuracy of testing the forecasting modes is improved.

Description

Technical Field

[0001] The present application relates to the field of weather forecasting, and in particular to a method, device, electronic equipment and program product for verifying forecasting models. Background Art

[0002] With the surge in multi-source heterogeneous forecast data (such as global models, regional models, and Grib / NetCDF format data released by different institutions), the demand for verifying numerical weather forecast models has shifted from a single numerical error assessment to a comprehensive consideration of spatial structure consistency, multi-factor coupling relationships, and long-term stability.

[0003] Currently, forecast models can be tested using traditional statistical indicators, evaluating the difference between the forecast and the observed values. For example, a significance test can be performed by calculating the regional mean using data in a specific format for a specific regional model.

[0004] Among the above technologies, current testing methods usually adopt static thresholds and single indicators, resulting in low accuracy of the test prediction model. Summary of the Invention

[0005] The present application provides a method, device, electronic equipment and program product for verifying forecast models, which are used to solve the technical problem of low accuracy in verifying forecast models in the prior art.

[0006] In a first aspect, the present application provides a method for verifying a forecast model forecast, the method comprising:

[0007] Obtaining target parameters input by a user, wherein the target parameters include a target area, meteorological elements, and a target time period;

[0008] Determining a plurality of observation data of the target area within the target period and a plurality of forecast data corresponding to the plurality of forecast modes respectively;

[0009] According to the meteorological elements, the plurality of observation data and the plurality of forecast data are inspected and processed to obtain an inspection index value corresponding to each forecast mode;

[0010] Determining a degradation factor corresponding to each forecast mode according to the inspection index value, wherein the degradation factor is used to represent the driving contribution of the meteorological element to the degradation of forecast skill under the forecast mode;

[0011] According to multiple test index values ​​and multiple degradation factors, visual test results corresponding to multiple forecast modes are generated.

[0012] In this way, different inspection methods are constructed according to different meteorological elements, which improves the adaptability of the inspection, and determines the degradation factors corresponding to different areas of each forecast model, so that the degradation process can be monitored and evaluated, and the accuracy of the inspection forecast model is improved.

[0013] Optionally, the method described above, for any forecast mode, performs verification processing on the plurality of observation data and the plurality of forecast data according to the meteorological elements to obtain a verification index value corresponding to the forecast mode, including:

[0014] determining a target algorithm among a plurality of verification algorithms according to the meteorological factors;

[0015] Performing verification processing on the plurality of observation data and the plurality of forecast data according to the target algorithm to obtain a first indicator value corresponding to the forecast mode;

[0016] The first indicator value is resampled to obtain a test indicator value corresponding to the forecast mode.

[0017] In this way, different inspection methods are constructed according to different meteorological elements, which improves the adaptability of the inspection and the accuracy of the inspection and forecast model.

[0018] Optionally, in the above method, the meteorological elements include visibility, precipitation, temperature, zonal wind, meridional wind, and humidity, the multiple verification algorithms include a dynamic threshold verification algorithm and a continuous variable verification algorithm, and determining a target algorithm from the multiple verification algorithms based on the meteorological elements includes:

[0019] If the meteorological element is visibility or precipitation, determining that the target algorithm is a dynamic threshold detection algorithm;

[0020] If the meteorological elements are temperature, latitudinal wind, longitudinal wind and humidity, it is determined that the target algorithm is a continuous variable test algorithm.

[0021] In this way, different inspection methods are constructed according to different meteorological elements, which improves the adaptability of the inspection and the accuracy of the inspection and forecast model.

[0022] Optionally, in the above method, the target algorithm is a dynamic threshold verification algorithm, and the plurality of observation data and the plurality of forecast data are verified and processed according to the target algorithm to obtain the first indicator value corresponding to the forecast mode, including:

[0023] Get the preset sliding window half width;

[0024] Determining a first time period according to the target time period and the preset sliding window half width;

[0025] Determining a dynamic threshold corresponding to the time point based on a preset quantile level and a plurality of observation data corresponding to the first time period;

[0026] performing classification processing on the plurality of observation data corresponding to the first time period and the plurality of forecast data corresponding to the first time period according to the dynamic threshold value to obtain observation levels of the plurality of observation data and forecast levels of the plurality of forecast data;

[0027] Determining, based on the observation levels of the plurality of observation data and the forecast levels of the plurality of forecast data, inspection data corresponding to the time point, the inspection data including the plurality of inspection levels and the number of hits, the number of missed alarms, the number of false alarms, and the number of correct rejections corresponding to each inspection level;

[0028] A first indicator value corresponding to the forecast mode is determined based on the test data corresponding to a plurality of time points.

[0029] In this way, the dynamic threshold test algorithm is used to determine the dynamic threshold, improve the adaptability to nonlinear variables and regional heterogeneity, and improve the accuracy of the test and forecast model.

[0030] Optionally, the method described above, determining the first indicator value corresponding to the forecast mode based on the test data corresponding to the plurality of time points, includes:

[0031] Determining, in descending order of the plurality of inspection levels, level inspection data corresponding to the plurality of inspection levels in sequence from the inspection data corresponding to the plurality of time points, the level inspection data including the number of hits, the number of missed alarms, the number of false alarms, and the number of correct rejections corresponding to the same inspection level at the plurality of time points;

[0032] For any inspection level, determining the inspection score of the inspection level according to the level inspection data corresponding to the inspection level;

[0033] The first indicator value corresponding to the forecast mode is determined according to the inspection scores corresponding to the multiple inspection levels and the weight values ​​corresponding to the multiple inspection levels.

[0034] In this way, by comprehensively considering and weighted calculating multi-dimensional data of different inspection levels, the performance of the forecast model can be comprehensively evaluated, providing a clear direction for the improvement of the forecast model, thereby improving the accuracy and reliability of the forecast model.

[0035] Optionally, in the above method, the target algorithm is a continuous variable test algorithm, and the plurality of observation data and the plurality of forecast data are tested and processed according to the target algorithm to obtain the first indicator value corresponding to the forecast mode, including:

[0036] Obtaining a latitude weighted value corresponding to the target area;

[0037] Based on the multiple observation data and the multiple forecast data, and the latitude weighted value, the target algorithm is used to determine multiple first indicator values ​​corresponding to the forecast mode, wherein the first indicator values ​​include an anomaly correlation coefficient value, a deviation value, and a root mean square error value.

[0038] In this way, through the continuous variable verification algorithm, the system can evaluate the spatial consistency of the forecast model and the performance of large-scale weather systems, thereby improving the accuracy of the verification forecast model.

[0039] Optionally, the method described above, for any forecast mode, determining the degradation factor corresponding to the forecast mode according to the inspection index value, includes:

[0040] Determine the indicator type corresponding to the inspection indicator value;

[0041] Determine the target formula and reference value based on the indicator type;

[0042] According to the inspection index value and the reference value, a degradation index corresponding to the inspection index value is determined by the target formula; according to the degradation index and the weight values ​​of multiple sites in the target area, a degradation factor corresponding to the forecast model is determined.

[0043] In this way, the degradation factor can be used to dynamically evaluate the degradation trend of forecast skill with the timeliness of forecast, thereby providing directional guidance for model improvement.

[0044] In a second aspect, the present application provides a device for verifying a forecast model prediction, the device comprising:

[0045] An acquisition module is used to acquire target parameters input by a user, wherein the target parameters include a target area, meteorological elements, and a target time period;

[0046] A determination module, configured to determine a plurality of observation data of the target area within the target period and a plurality of forecast data corresponding to the plurality of forecast modes;

[0047] A first verification module is configured to perform verification processing on the plurality of observation data and the plurality of forecast data according to the meteorological elements to obtain a verification index value corresponding to each forecast mode;

[0048] a second inspection module, configured to determine a degradation factor corresponding to each forecast mode according to the inspection index value, wherein the degradation factor is used to represent a driving contribution of the meteorological element to the degradation of forecast skill under the forecast mode;

[0049] The display module is used to generate visual inspection results corresponding to multiple forecast modes according to multiple inspection index values ​​and multiple degradation factors.

[0050] Optionally, in the above device, the first inspection module is specifically configured to:

[0051] determining a target algorithm among a plurality of verification algorithms according to the meteorological factors;

[0052] Performing verification processing on the plurality of observation data and the plurality of forecast data according to the target algorithm to obtain a first indicator value corresponding to the forecast mode;

[0053] The first indicator value is resampled to obtain a test indicator value corresponding to the forecast mode.

[0054] Optionally, in the device described above, the meteorological elements include visibility, precipitation, temperature, zonal wind, meridional wind, and humidity; the multiple verification algorithms include a dynamic threshold verification algorithm and a continuous variable verification algorithm; and the first verification module is specifically configured to:

[0055] If the meteorological element is visibility or precipitation, determining that the target algorithm is a dynamic threshold detection algorithm;

[0056] If the meteorological elements are temperature, latitudinal wind, longitudinal wind and humidity, it is determined that the target algorithm is a continuous variable test algorithm.

[0057] Optionally, in the above device, the target algorithm is a dynamic threshold verification algorithm, and the first verification module is specifically configured to:

[0058] Get the preset sliding window half width;

[0059] Determining a first time period according to the target time period and the preset sliding window half width;

[0060] Determining a dynamic threshold corresponding to the time point based on a preset quantile level and a plurality of observation data corresponding to the first time period;

[0061] performing classification processing on the plurality of observation data corresponding to the first time period and the plurality of forecast data corresponding to the first time period according to the dynamic threshold value to obtain observation levels of the plurality of observation data and forecast levels of the plurality of forecast data;

[0062] Determining, based on the observation levels of the plurality of observation data and the forecast levels of the plurality of forecast data, inspection data corresponding to the time point, the inspection data including the plurality of inspection levels and the number of hits, the number of missed alarms, the number of false alarms, and the number of correct rejections corresponding to each inspection level;

[0063] A first indicator value corresponding to the forecast mode is determined based on the test data corresponding to a plurality of time points.

[0064] Optionally, in the above device, the first inspection module is specifically configured to:

[0065] Determining, in descending order of the plurality of inspection levels, level inspection data corresponding to the plurality of inspection levels in sequence from the inspection data corresponding to the plurality of time points, the level inspection data including the number of hits, the number of missed alarms, the number of false alarms, and the number of correct rejections corresponding to the same inspection level at the plurality of time points;

[0066] For any inspection level, determining the inspection score of the inspection level according to the level inspection data corresponding to the inspection level;

[0067] The first indicator value corresponding to the forecast mode is determined according to the inspection scores corresponding to the multiple inspection levels and the weight values ​​corresponding to the multiple inspection levels.

[0068] Optionally, in the above device, the target algorithm is a continuous variable testing algorithm, and the first testing module is specifically configured to:

[0069] Obtaining a latitude weighted value corresponding to the target area;

[0070] Based on the multiple observation data and the multiple forecast data, and the latitude weighted value, the target algorithm is used to determine multiple first indicator values ​​corresponding to the forecast mode, wherein the first indicator values ​​include an anomaly correlation coefficient value, a deviation value, and a root mean square error value.

[0071] Optionally, in the above device, for any forecast, the second verification module is specifically configured to:

[0072] Determine the indicator type corresponding to the inspection indicator value;

[0073] Determine the target formula and reference value based on the indicator type;

[0074] Determining a degradation index corresponding to the inspection index value by using the target formula according to the inspection index value and the reference value;

[0075] A degradation factor corresponding to the forecast mode is determined according to the degradation index and weight values ​​of multiple sites in the target area.

[0076] In a third aspect, the present application provides an electronic device, comprising: a processor, and a memory communicatively connected to the processor;

[0077] The memory stores computer-executable instructions;

[0078] The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of the first aspects.

[0079] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement the method as described in any one of the first aspects.

[0080] In a fifth aspect, the present application provides a computer program product, comprising a computer program, which, when executed by a computer, implements the method as described in any one of the first aspects.

[0081] The present application provides a method, device, electronic device, and program product for verifying forecast model predictions. The method obtains target parameters input by the user, including target area, meteorological elements, and target time period, determines multiple observation data of the target area within the target time period, and multiple forecast data corresponding to multiple forecast models. According to the meteorological elements, the multiple observation data and the multiple forecast data are verified and processed to obtain a verification index value corresponding to each forecast model. According to the verification index value, the degradation factor corresponding to each forecast model is determined. The degradation factor is used to represent the driving contribution of the meteorological elements to the degradation of the forecast skill under the forecast model. According to the multiple verification index values ​​and the multiple degradation factors, visual verification results corresponding to the multiple forecast models are generated. In this way, different verification methods are constructed according to different meteorological elements, which improves the adaptability of the verification. The degradation factor corresponding to each forecast model is determined, so that the degradation process can be monitored and evaluated, and the accuracy of the verification forecast model is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0083] Figure 1 A schematic diagram of the structure of an inspection system provided in an embodiment of the present application;

[0084] Figure 2 A flowchart of a method for verifying a forecast model provided in an embodiment of the present application;

[0085] Figure 3 A flow chart of another method for verifying a forecast model provided in an embodiment of the present application;

[0086] Figure 4 A schematic flow chart of another method for verifying a forecast model provided in an embodiment of the present application;

[0087] Figure 5 A schematic diagram of the structure of a prediction mode inspection device provided in an embodiment of the present application;

[0088] Figure 6A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0089] The above drawings illustrate specific embodiments of the present application, which will be described in more detail below. These drawings and the textual description are not intended to limit the scope of the present application in any way, but rather to illustrate the concepts of the present application to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0090] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0091] It should be noted that although the terms "first" and "second" are used to describe various information in the embodiments of this application, such information should not be limited to these terms. These terms are merely used to distinguish information of the same type from each other. Alternatively, without departing from the scope of this application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information.

[0092] It should be understood that the terms "comprise" and "include" indicate the presence of the previously mentioned features, steps, or operations, but do not exclude the presence, occurrence, or addition of one or at least one other feature, step, or operation. The terms "and / or" and the like used in this application may be interpreted as inclusive, or may mean any one or any combination. Alternatively, "A and / or B" means "any of the following: A; B; A and B." In addition, the character " / " in this document generally indicates that the preceding and following objects are in an "or" relationship.

[0093] With the surge in multi-source heterogeneous forecast data (such as global models, regional models, and Grib / NetCDF format data released by different institutions), the demand for verifying numerical weather forecast models has shifted from a single numerical error assessment to a comprehensive consideration of spatial structure consistency, multi-factor coupling relationships, and long-term stability.

[0094] Currently, forecast models can be tested using traditional statistical indicators, evaluating the difference between the forecast and the observed values. For example, a significance test can be performed by calculating the regional mean using data in a specific format for a specific regional model.

[0095] Among the above technologies, current testing methods usually adopt static thresholds and single indicators, resulting in low accuracy of the test prediction model.

[0096] In order to solve the above technical problems, the embodiment of the present application provides a method for verifying a forecast model. By determining multiple observation data of a target area within a target period and multiple forecast data corresponding to multiple forecast models, the multiple observation data and the multiple forecast data are verified and processed according to meteorological factors to obtain a verification index value corresponding to each forecast model. Based on the verification index value, the degradation factor corresponding to each forecast model is determined. Based on the multiple verification index values ​​and the multiple degradation factors, visual verification results corresponding to the multiple forecast models are generated. In this way, different verification methods are constructed according to different meteorological factors, which improves the adaptability of the verification. The degradation factor corresponding to each forecast model is determined, and the degradation process can be monitored and evaluated, thereby improving the accuracy of the verification forecast model.

[0097] Next, combine Figure 1 , an example is given to illustrate the verification system of the forecast model.

[0098] Figure 1 This is a schematic diagram of the structure of an inspection system provided in an embodiment of the present application. Figure 1 , Figure 1 It can include data layer, indicator calculation layer, analysis layer and visualization layer.

[0099] The data layer can be used to automatically download multiple forecast data and observation data from multiple forecast models. It supports multiple data formats, different pressure layers, and multi-dimensional meteorological elements. For example, multiple data formats can include Grib, NetCDF, and CSV. Different pressure layers can include 200hPa, 500hPa, 700hPa, 850hPa, and 925hPa. Multi-dimensional meteorological elements can include precipitation, temperature, zonal wind, meridional wind, and humidity.

[0100] The data layer can also be used to download reanalysis data.

[0101] The data downloaded by the data layer can have different temporal and spatial resolutions.

[0102] The data layer can automatically schedule data downloads through a scheduled task scheduling mechanism to ensure the acquisition of real-time and accurate forecast data and observation data, significantly reducing the complexity of manual operations and greatly improving the real-time nature of data acquisition and system processing efficiency.

[0103] After the data is downloaded, the system can use the preset program script to perform data quality control.

[0104] Data quality control can include deleting abnormal files, checking data anomalies and deleting files containing abnormal data, repairing or correcting abnormal data that can be repaired, etc.

[0105] For example, abnormal files are files that cannot be opened normally, files with data volumes lower than normal, etc.

[0106] After deleting the file, the data layer can resubmit the data download task for downloading. If repeated submission fails three times, the download task will not be submitted again.

[0107] Data quality control can ensure the accuracy and reliability of input data and avoid deviations in test results due to data defects.

[0108] The data layer can also be used to convert the downloaded data into a unified file format, data format, and spatiotemporal resolution, and to align data with a spatial grid structure.

[0109] For example, a single meteorological element of a pressure layer in each forecast mode can be saved as a data file in NetCDF format with an hourly temporal resolution and a spatial resolution of 0.1°x0.1°.

[0110] The data layer standardizes data from multiple forecast models across diverse sources, overcoming the compatibility issues inherent in existing technologies, such as inconsistent data formats. This significantly improves the efficiency and accuracy of data processing. The data layer successfully addresses the technical challenges of diverse data sources and inconsistent formats, enabling unified management and analysis of multiple data sources.

[0111] The indicator calculation layer can be used to quantify the forecast performance of multiple forecast models.

[0112] The indicator calculation layer can include multiple test indicators.

[0113] For example, the multiple test indicators may include root mean square error, standard deviation, correlation coefficient, threat score, and reliability.

[0114] The analysis layer can be used to diagnose and attribute errors in multiple forecast models based on multiple test indicator values.

[0115] For example, the analysis layer can quantify the driving contribution of various meteorological factors to the degradation of forecast skills at different temporal and spatial scales, construct a traceable and interpretable degradation factor map, and provide accurate reference for subsequent meteorological model improvements.

[0116] The visualization layer can be designed based on a lightweight B / S architecture, supporting high-performance interactive operations on the web page, with flexible layer switching, legend scaling and parameter setting capabilities, and adapting to multi-terminal browsing environments.

[0117] The visualization layer can have the "one-click skill report generation" function, automatically summarizing the core assessment results and visualization graphics to form standardized or customized output reports, greatly improving the automation and intelligence level of the business assessment process.

[0118] The following specific embodiments describe in detail the technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0119] The technical solutions shown in this application are described in detail below through specific embodiments. It should be noted that the following embodiments can exist independently or in combination with each other, and the same or similar contents will not be repeated in different embodiments.

[0120] Figure 2 This is a flow chart of a method for verifying a forecast model provided in an embodiment of the present application. The execution subject of the embodiment of the present application may be a processor. The processor may be implemented by software or by a combination of software and hardware. Figure 2 , the method comprising:

[0121] S201: Obtain target parameters input by the user.

[0122] Target parameters include target area, meteorological elements and target time period.

[0123] The target area may be a location area where multiple forecast models are to be tested.

[0124] Meteorological elements may include precipitation, temperature, latitudinal wind, meridional wind and humidity.

[0125] The target period can be a historical period in which multiple forecast models are to be tested.

[0126] Target parameters can be used to describe the parameters that users want to perform data analysis on.

[0127] An interactive interface can be displayed, which may include controls such as text boxes, selection boxes, and sliding bars. In response to the user's interactive operations on the target area selection control, the interactive operations on the meteorological element selection control, the sliding operation of the sliding bar corresponding to the target time period, and the clicking operation of the submit button in the interactive interface, the control values ​​corresponding to the multiple controls are obtained, and the target parameters input by the user are determined based on the multiple control values.

[0128] The target parameters entered by the user can be obtained through the command line interface.

[0129] It should be noted that the target parameters input by the user can be obtained according to any feasible implementation method, and the embodiments of the present application are not limited to this.

[0130] S202: Determine a plurality of observation data of a target area within a target period and a plurality of forecast data corresponding to the plurality of forecast modes.

[0131] Observation data can be collected through ground meteorological stations, automatic weather stations and other equipment, covering physical quantity data such as temperature, humidity, wind speed, etc. of the surface and near-surface atmosphere.

[0132] The forecast data may be data generated by forecasting atmospheric conditions, weather phenomena and related environmental parameters through a meteorological model, statistical algorithm or target algorithm corresponding to the forecast mode.

[0133] According to the target parameters, data can be downloaded through the program script in the data layer to obtain multiple downloaded data corresponding to the meteorological elements of the target area within the target time period. The multiple downloaded data are quality controlled and standardized to obtain multiple observation data of the target area within the target time period and multiple forecast data corresponding to multiple forecast models.

[0134] Among them, quality control processing may include deleting abnormal files, checking data anomalies and deleting files containing abnormal data, repairing or correcting abnormal data that can be repaired, etc.

[0135] Normalization processing can be used to convert the downloaded data into a unified file format, data format, and spatiotemporal resolution, and to align the data in a spatial grid structure.

[0136] It should be noted that, multiple observation data of the target area within the target period and multiple forecast data corresponding to multiple forecast modes can be determined according to any feasible implementation method, and the embodiments of the present application are not limited to this.

[0137] S203. Perform inspection and processing on the plurality of observation data and the plurality of forecast data according to meteorological elements to obtain inspection index values ​​corresponding to each forecast mode.

[0138] Different meteorological elements require different inspection methods and different inspection indicators.

[0139] The inspection index value may be an index value corresponding to the inspection index.

[0140] The test index values ​​can be used to compare the quantitative results calculated from the forecast data and the observation data, and to objectively evaluate the accuracy, reliability and effectiveness of the forecast.

[0141] A first mapping relationship can be obtained, and a target algorithm can be determined based on the meteorological elements and the first mapping relationship. Based on the target algorithm, multiple observation data and multiple forecast data are inspected and processed to obtain inspection index values ​​corresponding to each forecast mode.

[0142] The first mapping relationship may include multiple meteorological elements and an algorithm corresponding to each meteorological element.

[0143] Meteorological factors can be input into the algorithm selection model to obtain the target algorithm. According to the target algorithm, multiple observation data and multiple forecast data are tested and processed to obtain the test index value corresponding to each forecast mode.

[0144] It should be noted that, multiple observation data of the target area within the target period and multiple forecast data corresponding to multiple forecast modes can be determined according to any feasible implementation method, and the embodiments of the present application are not limited to this.

[0145] S204. Determine the degradation factor corresponding to each forecast mode according to the test index value.

[0146] The degradation factor is used to indicate the driving contribution of meteorological factors to the degradation of forecast skill in the forecast model.

[0147] The test index values ​​may include threat score, latitude-weighted anomaly correlation coefficient, root mean square error and bias, etc., which are not limited here.

[0148] The inspection index value may include site index values ​​corresponding to multiple sites.

[0149] Optionally, for any forecast mode, the degradation factor corresponding to the forecast mode can be determined based on the test index value in the following manner: a reference value corresponding to the test index value can be obtained, and the degradation indexes corresponding to the multiple sites can be determined based on the index values ​​and reference values ​​of the multiple sites. The degradation factor corresponding to the forecast mode can be determined based on the degradation indexes corresponding to the multiple sites and the weight values ​​corresponding to the multiple sites.

[0150] For example, for the indicator types of root mean square error and bias, the larger the degradation index, the worse the forecast effect.

[0151] The degradation index corresponding to this indicator type can be determined by the following formula:

[0152]

[0153] Among them, D s,f,t It can represent the degradation index corresponding to site s, meteorological element f, and target period t, M s,f,t It can represent the inspection index value corresponding to the site s, meteorological element f, and target period t, T s,f It can represent the reference value corresponding to the test index value.

[0154] For the indicator type of threat score and latitude-weighted anomaly correlation coefficient, the larger the degradation index, the better the forecast effect.

[0155] The degradation index corresponding to this indicator type can be determined by the following formula:

[0156]

[0157] Among them, D s,f,t It can represent the degradation index corresponding to site s, meteorological element f, and target period t, M s,f,t It can represent the inspection index value corresponding to the site s, meteorological element f, and target period t, T s,f It can represent the reference value corresponding to the test index value.

[0158] The reference value may be set based on the sliding mean of historical data, the sliding window quantile method, etc., and is not limited here.

[0159] Optionally, the degradation factor corresponding to the forecast model may be determined according to the degradation indices corresponding to the multiple stations in the target area and the weight values ​​corresponding to the multiple stations in the following manner:

[0160]

[0161] Among them, w i It can represent the weight value of the i-th site, D i,f,t It can represent the degradation index corresponding to the meteorological element f and target period t at the i-th station, D f It can represent the degradation factor corresponding to the forecast model.

[0162] The weight value of the i-th site can be calculated by the following formula:

[0163] Since the distribution of station observations is not uniform, in order to avoid weight stacking in dense areas, we follow the principle of "giving less weight to dense stations and more weight to sparse stations" and introduce a normalization coefficient for regional balance:

[0164]

[0165] ρ i : The number of neighboring sites around the site (e.g. within 50km).

[0166] Optionally, for any forecast mode, the degradation factor corresponding to the forecast mode can be determined based on the test index value in the following manner: the indicator type corresponding to the test index value can be determined; based on the indicator type, the target formula and reference value can be determined; based on the test index value and the reference value, the degradation index corresponding to the test index value can be determined through the target formula; based on the degradation index and the weight values ​​of multiple sites in the target area, the degradation factor corresponding to the forecast mode can be determined.

[0167] In this way, based on the degradation index, the main degradation factors in different regions of the forecast model can be determined, providing direction for model improvement.

[0168] It should be noted that the degradation factor corresponding to the forecast mode can be determined according to any feasible implementation method, and the embodiments of the present application are not limited to this.

[0169] Optionally, after determining the degradation factor corresponding to the forecast mode, the method further includes: establishing a skill degradation contribution matrix and a contribution matrix of the degradation factor and the spatial region according to the degradation factor corresponding to the forecast mode.

[0170] The skill degradation contribution matrix can be used to represent the degradation intensity distribution of the degradation factor corresponding to multiple sites in the target area.

[0171] Each element in the contribution matrix of the degradation factor and the spatial region can represent the normalized intensity of skill degradation caused by the degradation factor in the corresponding region of the site.

[0172] The contribution matrix of degradation factors and spatial regions can be used to identify “which degradation factors lead to skill degradation in which regions”.

[0173] S205 : Generate visual inspection results corresponding to a plurality of forecast modes according to the plurality of inspection index values ​​and the plurality of degradation factors.

[0174] The visual test results can be used for visual comparison and testing of spatial, temporal and multiple forecast models.

[0175] According to multiple test index values ​​and multiple degradation factors, multiple preset scripts can be used to generate visual test results corresponding to multiple forecast modes.

[0176] Among them, the visual inspection results can include spatial skill maps, puzzle skill radar maps, time series skill trend maps and uncertainty display maps, etc.

[0177] The spatial skill map can integrate the spatial distribution characteristics of the test index values ​​based on the map background, automatically match the grading standards and generate a color gradient map, realizing the perception of the evaluation pattern of "seeing the whole picture from one map".

[0178] The spatial skill map can also be used to integrate the main degradation factors of each site / region (different degradation factors can be represented by different colors), which can intuitively determine where degradation occurs and why, and assist in regional correction and model optimization.

[0179] Jigsaw puzzle techniques: Radar charts can show the advantages and disadvantages of different forecast models in key meteorological indicators through comprehensive comparison of multiple modes, multiple factors, and multiple levels in the form of radar chart splicing matrices, thereby strengthening the interpretability analysis between models.

[0180] The time series skill trend chart can accurately display the temporal trend of the test indicator values ​​corresponding to different forecast modes, and assist in identifying stability and mutation points.

[0181] The uncertainty display diagram can innovatively output skill indicator distribution histograms and confidence interval box plots in parallel, improving the robustness and reliability of decision-making references.

[0182] The forecast model verification method provided in this embodiment obtains target parameters input by the user, which include a target area, meteorological elements, and a target time period. It then determines multiple observation data for the target area within the target time period and multiple forecast data corresponding to multiple forecast models. Based on the meteorological elements, the multiple observation data and multiple forecast data are verified and processed to obtain a verification index value corresponding to each forecast model. Based on the verification index value, a degradation factor corresponding to each forecast model is determined. The degradation factor is used to represent the contribution of the meteorological element to the degradation of forecast skill under the forecast model. Based on the multiple verification index values ​​and the multiple degradation factors, visual verification results corresponding to the multiple forecast models are generated. In this way, different verification methods are constructed based on different meteorological elements, improving the adaptability of the verification. Furthermore, the degradation factor corresponding to each forecast model is determined, making the degradation process monitorable and evaluable, thereby improving the accuracy of the verified forecast model.

[0183] Next, combine Figure 3 , for any forecast model, the process (S203) of verifying and processing multiple observation data and multiple forecast data according to meteorological elements to obtain the verification index value corresponding to the forecast model is explained.

[0184] Figure 3 This is a flow chart of another method for verifying the forecast mode provided in the embodiment of the present application. Figure 3 , the method comprising:

[0185] S301. Determine a target algorithm from a plurality of verification algorithms based on meteorological factors.

[0186] Meteorological elements include visibility, precipitation, temperature, zonal wind, meridional wind and humidity.

[0187] Zonal winds are winds that blow along the latitudes (east-west).

[0188] Zonal winds can include both decameter and hectometer winds, depending on the altitude at which they are measured.

[0189] Meridional winds are winds that blow along the meridians (north-south).

[0190] Meridional winds can also include decameter and hectometer winds, depending on the altitude at which they are measured.

[0191] The multiple testing algorithms include a dynamic threshold testing algorithm and a continuous variable testing algorithm.

[0192] A first mapping relationship may be obtained, and a target algorithm may be determined among a plurality of verification algorithms according to meteorological elements.

[0193] Optionally, the target algorithm can be determined among multiple test algorithms based on meteorological elements in the following manner: if the meteorological element is visibility or precipitation, the target algorithm is determined to be a dynamic threshold test algorithm; if the meteorological element is temperature, latitudinal wind, longitudinal wind and humidity, the target algorithm is determined to be a continuous variable test algorithm.

[0194] S302: According to the target algorithm, multiple observation data and multiple forecast data are inspected and processed to obtain a first indicator value corresponding to the forecast mode.

[0195] The first indicator value may be an indicator value calculated according to a target algorithm.

[0196] Optionally, the target algorithm is a dynamic threshold verification algorithm, and multiple observation data and multiple forecast data can be verified and processed according to the target algorithm in the following manner to obtain a first indicator value corresponding to the forecast mode: determine multiple time points corresponding to the target time period; obtain a preset sliding window half-width; determine the first time period based on the target time period and the preset sliding window half-width; determine the dynamic threshold corresponding to the time point based on the preset quantile level and the multiple observation data corresponding to the first time period; classify the multiple observation data corresponding to the first time period and the multiple forecast data corresponding to the first time period based on the dynamic threshold to obtain the observation level of the multiple observation data and the forecast level of the multiple forecast data; determine the verification data corresponding to the time point based on the observation level of the multiple observation data and the forecast level of the multiple forecast data; determine the first indicator value corresponding to the forecast mode based on the verification data corresponding to the multiple time points.

[0197] The inspection data includes multiple inspection levels, and the number of hits, missed reports, false alarms, and correct rejections corresponding to each inspection level.

[0198] The dynamic threshold can be determined by the following formula:

[0199] θ t =Quantile q (X t-N:t+N )

[0200] t can be a time point, N can be a preset sliding window half width, X t-N:t+N is the first period, θt is the dynamic threshold at time point t, q is the quantile level, q Can represent quantile calculation functions.

[0201] Optionally, the first index value corresponding to the prediction mode can be determined according to the test data corresponding to the plurality of time points in the following manner: in the test data corresponding to the plurality of time points, the level test data corresponding to the plurality of test levels is determined in turn according to the order from small to large of the plurality of test levels; for any one test level, the test score of the test level is determined according to the level test data corresponding to the test level; and the first index value corresponding to the prediction mode is determined according to the test scores corresponding to the plurality of test levels and the weight values corresponding to the plurality of test levels.

[0202] The level test data includes the number of hits, the number of false alarms, the number of false alarms, and the number of correct refusals corresponding to the same test level in the plurality of time points.

[0203] Optionally, the test score of the test level can include a threat score (TS) and a equitable threat score (ETS) in the following manner according to the level test data corresponding to the test level:

[0204]

[0205] wherein,

[0206] wherein, k can represent the test level, TS k may represent the threat score corresponding to the test level k, ETS k may represent the threat score corresponding to the test level k, H k may represent the number of hits, M k may represent the number of false alarms, F k may represent the number of false alarms, C k may represent the number of correct refusals.

[0207] Optionally, the first index value corresponding to the prediction mode can be determined according to the test scores corresponding to the plurality of test levels and the weight values corresponding to the plurality of test levels in the following manner: determining the product factor of the test score corresponding to each test level and the corresponding weight value, adding the product factors corresponding to the plurality of test levels to determine the first index value corresponding to the prediction mode.

[0208] Optionally, the target algorithm is a continuous variable test algorithm, and the first index value corresponding to the prediction mode can be obtained by performing test processing on the plurality of observation data and the plurality of prediction data according to the target algorithm in the following manner: obtaining the latitude weight value corresponding to the target area; and determining the plurality of first index values corresponding to the prediction mode by the target algorithm according to the plurality of observation data and the plurality of prediction data and the latitude weight value.

[0209] The first indicator value includes the deviation correlation coefficient value, the deviation value and the root mean square error value.

[0210] The latitude weight value corresponding to the target area is preset.

[0211] The target algorithm may include a correlation algorithm for calculating an anomaly correlation coefficient value, a correlation algorithm for a deviation value, and a correlation algorithm for a root mean square error value.

[0212] It should be noted that the first indicator value corresponding to the forecast mode can be determined according to any feasible implementation method, and the embodiments of the present application are not limited to this.

[0213] S303: Resample the first index value to obtain a test index value corresponding to the forecast mode.

[0214] The quantity corresponding to the first index value may be determined, and N resampling operations may be repeated for the first index value to obtain N sample sequences, where the test index value is the N sample sequences.

[0215] The resampling operation is to select an index value from the first index value.

[0216] The implementation content of each step in the embodiment of the present application can refer to the description of the corresponding steps or operations in the above method embodiment, and repeated content will not be repeated.

[0217] The forecast model verification method provided in this embodiment determines a target algorithm from multiple verification algorithms based on meteorological factors; verifies multiple observation data and multiple forecast data using the target algorithm to obtain a first indicator value corresponding to the forecast model; and resamples the first indicator value to obtain a verification indicator value corresponding to the forecast model. In this way, different verification methods are constructed based on different meteorological factors, improving the adaptability of the verification and the accuracy of the verification forecast model.

[0218] Next, combine Figure 4 , the technical solutions shown in this application are explained through specific examples.

[0219] Figure 4 A flow chart of another method for verifying a forecast model provided in an embodiment of the present application. Figure 4 , the method comprising:

[0220] S401: Obtain target parameters input by the user.

[0221] S402: Determine a plurality of observation data of a target area within a target period and a plurality of forecast data corresponding to the plurality of forecast modes.

[0222] S403. Determine a target algorithm from a plurality of verification algorithms based on meteorological factors.

[0223] S404: According to the target algorithm, the plurality of observation data and the plurality of forecast data are tested and processed to obtain a first indicator value corresponding to each forecast mode.

[0224] S405 . Resample the multiple first indicator values ​​to obtain a test indicator value corresponding to each forecast mode.

[0225] S406: For any forecast mode, determine the indicator type corresponding to the test indicator value.

[0226] S407. Determine the target formula and reference value according to the indicator type.

[0227] S408. Determine the degradation index corresponding to the inspection index value through a target formula according to the inspection index value and the reference value.

[0228] S409: Determine a degradation factor corresponding to the forecast model according to the degradation index.

[0229] S410 , generating visual inspection results corresponding to a plurality of forecast modes respectively according to a plurality of inspection index values ​​and a plurality of degradation factors.

[0230] The implementation content of each step in the embodiment of the present application can refer to the description of the corresponding steps or operations in the above method embodiment, and repeated content will not be repeated.

[0231] In this way, different inspection methods are constructed according to different meteorological elements, which improves the adaptability of the inspection, and determines the degradation factor corresponding to each forecast model, so that the degradation process can be monitored and evaluated, and the accuracy of the inspection forecast model is improved.

[0232] Figure 5 This is a schematic diagram of the structure of a test device for a forecast mode provided in an embodiment of the present application. Figure 5 The prediction mode verification device 500 includes an acquisition module 501, a determination module 502, a first verification module 503, a second verification module 504 and a display module 505, wherein,

[0233] An acquisition module 501 is configured to acquire target parameters input by a user, wherein the target parameters include a target area, meteorological elements, and a target time period;

[0234] A determination module 502 is configured to determine a plurality of observation data of the target area within the target period and a plurality of forecast data corresponding to the plurality of forecast modes;

[0235] A first verification module 503 is configured to perform verification processing on the plurality of observation data and the plurality of forecast data according to the meteorological elements to obtain a verification index value corresponding to each forecast mode;

[0236] The second verification module 504 is configured to determine a degradation factor corresponding to each prediction mode according to the verification index value, and the degradation factor is used to represent a driving contribution of the meteorological element to the prediction skill degradation under the prediction mode.

[0237] The display module 505 is configured to generate a visual verification result corresponding to each prediction mode according to the plurality of verification index values and the plurality of degradation factors.

[0238] Optionally, the apparatus as described above, the first verification module 503 is specifically configured to:

[0239] determine a target algorithm from the plurality of verification algorithms according to the meteorological element;

[0240] perform verification processing on the plurality of observation data and the plurality of prediction data according to the target algorithm to obtain a first index value corresponding to the prediction mode;

[0241] perform resampling processing on the first index value to obtain a verification index value corresponding to the prediction mode.

[0242] Optionally, the apparatus as described above, the meteorological element includes visibility, precipitation, temperature, zonal wind, meridional wind and humidity, and the plurality of verification algorithms includes a dynamic threshold verification algorithm and a continuous variable verification algorithm, and the first verification module 503 is specifically configured to:

[0243] if the meteorological element is visibility or precipitation, the target algorithm is determined as the dynamic threshold verification algorithm;

[0244] if the meteorological element is temperature, zonal wind, meridional wind and humidity, the target algorithm is determined as the continuous variable verification algorithm.

[0245] Optionally, the apparatus as described above, the target algorithm is the dynamic threshold verification algorithm, and the first verification module 503 is specifically configured to:

[0246] determine a plurality of time points corresponding to the target time period;

[0247] obtain a preset sliding window half-width;

[0248] determine a first time period according to the target time period and the preset sliding window half-width, determine a dynamic threshold value corresponding to the time point according to a preset quantile level and a plurality of observation data corresponding to the first time period;

[0249] perform classification processing on a plurality of observation data corresponding to the first time period and a plurality of prediction data corresponding to the first time period according to the dynamic threshold value to obtain an observation level of the plurality of observation data and a prediction level of the plurality of prediction data;

[0250] Determining, based on the observation levels of the plurality of observation data and the forecast levels of the plurality of forecast data, inspection data corresponding to the time point, the inspection data including the plurality of inspection levels and the number of hits, the number of missed alarms, the number of false alarms, and the number of correct rejections corresponding to each inspection level;

[0251] A first indicator value corresponding to the forecast mode is determined based on the test data corresponding to a plurality of time points.

[0252] Optionally, in the above-mentioned apparatus, the first inspection module 503 is specifically configured to:

[0253] Determining, in descending order of the plurality of inspection levels, level inspection data corresponding to the plurality of inspection levels in sequence from the inspection data corresponding to the plurality of time points, the level inspection data including the number of hits, the number of missed alarms, the number of false alarms, and the number of correct rejections corresponding to the same inspection level at the plurality of time points;

[0254] For any inspection level, determining the inspection score of the inspection level according to the level inspection data corresponding to the inspection level;

[0255] The first indicator value corresponding to the forecast mode is determined according to the inspection scores corresponding to the multiple inspection levels and the weight values ​​corresponding to the multiple inspection levels.

[0256] Optionally, in the above-mentioned device, the target algorithm is a continuous variable testing algorithm, and the first testing module 503 is specifically configured to:

[0257] Obtaining a latitude weighted value corresponding to the target area;

[0258] Based on the multiple observation data and the multiple forecast data, and the latitude weighted value, the target algorithm is used to determine multiple first indicator values ​​corresponding to the forecast mode, wherein the first indicator values ​​include an anomaly correlation coefficient value, a deviation value, and a root mean square error value.

[0259] Optionally, in the above apparatus, the second inspection module 504 is specifically configured to:

[0260] Determine the indicator type corresponding to the inspection indicator value;

[0261] Determine the target formula and reference value based on the indicator type;

[0262] Determining a degradation index corresponding to the inspection index value by using the target formula according to the inspection index value and the reference value;

[0263] A degradation factor corresponding to the forecast mode is determined according to the degradation index.

[0264] Figure 6This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 6 , the electronic device 600 may include: a memory 601 , a processor 602 , and a transceiver 603 .

[0265] The memory 601 is used to store program instructions;

[0266] The processor 602 is configured to execute the program instructions stored in the memory, so as to enable the electronic device 600 to perform the above method.

[0267] The transceiver 603 may include a transmitter and / or a receiver. The transmitter may also be referred to as a transmitter, a transmitter, a transmission port, a transmission interface, or similar descriptions, and the receiver may also be referred to as a receiver, a reception port, a reception interface, or similar descriptions. For example, the memory 601, the processor 602, and the transceiver 603 are interconnected via a bus 604.

[0268] An embodiment of the present application further provides a computer program product, which can be executed by a processor. When the computer program product is executed, the above method can be implemented.

[0269] The forecast mode inspection device, electronic device, computer-readable storage medium and computer program product of the embodiments of the present application can execute the technical solutions shown in the above-mentioned forecast mode inspection method embodiments. Their implementation principles and beneficial effects are similar and will not be repeated here.

[0270] It should be noted that for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all optional embodiments, and the actions and modules involved are not necessarily required by this application.

[0271] It should be further noted that, although the various steps in the flowchart are shown in sequence as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps may be performed in other orders. Moreover, at least a portion of the steps in the flowchart may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily performed at the same time, but may be performed at different times. The execution order of these sub-steps or stages is not necessarily to be performed in sequence, but may be performed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0272] It should be understood that the above-described device embodiments are merely illustrative, and the device of the present application may also be implemented in other ways. For example, the division of units / modules in the above-described embodiments is merely a logical functional division, and actual implementations may employ other division methods. For example, multiple units, modules, or components may be combined or integrated into another system, or some features may be omitted or not implemented.

[0273] In addition, unless otherwise specified, the functional units / modules in the various embodiments of the present application may be integrated into a single unit / module, each unit / module may exist physically separately, or two or more units / modules may be integrated together. The aforementioned integrated units / modules may be implemented in the form of hardware or software program modules.

[0274] If the integrated unit / module is implemented in hardware, the hardware may be digital circuits, analog circuits, etc. The physical implementation of the hardware structure includes, but is not limited to, transistors, memristors, etc. Unless otherwise specified, the processor may be any appropriate hardware processor, such as a CPU, GPU, FPGA, DSP, and ASIC. Unless otherwise specified, the storage unit may be any appropriate magnetic storage medium or magneto-optical storage medium, such as resistive random access memory (RRAM), dynamic random access memory (DRAM), static random access memory (SRAM), enhanced dynamic random access memory (EDRAM), high-bandwidth memory (HBM), hybrid memory cube (HMC), etc.

[0275] If the integrated unit / module is implemented in the form of a software program module and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a memory and includes a number of instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned memory includes various media that can store program codes, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.

[0276] In the above embodiments, the description of each embodiment has its own focus. For parts not described in detail in a certain embodiment, please refer to the relevant description of other embodiments. The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0277] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present application that follow the general principles of the present application and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0278] It should be understood that the present application is not limited to the exact structure described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A method for testing a forecast model, characterized in that: The method comprises: Obtaining target parameters input by a user, wherein the target parameters include a target area, meteorological elements, and a target time period; Determining a plurality of observation data of the target area within the target period and a plurality of forecast data corresponding to the plurality of forecast modes respectively; According to the meteorological elements, the plurality of observation data and the plurality of forecast data are inspected and processed to obtain an inspection index value corresponding to each forecast mode; Determining a degradation factor corresponding to each forecast mode according to the inspection index value, wherein the degradation factor is used to represent the driving contribution of the meteorological element to the degradation of forecast skill under the forecast mode; According to multiple test index values ​​and multiple degradation factors, visual test results corresponding to multiple forecast modes are generated.

2. The method according to claim 1, characterized in that For any forecast mode, the plurality of observation data and the plurality of forecast data are inspected and processed according to the meteorological elements to obtain an inspection index value corresponding to the forecast mode, including: determining a target algorithm among a plurality of verification algorithms according to the meteorological factors; Performing verification processing on the plurality of observation data and the plurality of forecast data according to the target algorithm to obtain a first indicator value corresponding to the forecast mode; The first indicator value is resampled to obtain a test indicator value corresponding to the forecast mode.

3. The method according to claim 2, characterized in that The meteorological elements include visibility, precipitation, temperature, zonal wind, meridional wind, and humidity. The multiple verification algorithms include a dynamic threshold verification algorithm and a continuous variable verification algorithm. Determining a target algorithm from the multiple verification algorithms based on the meteorological elements includes: If the meteorological element is visibility or precipitation, determining that the target algorithm is a dynamic threshold detection algorithm; If the meteorological elements are temperature, latitudinal wind, longitudinal wind and humidity, it is determined that the target algorithm is a continuous variable test algorithm.

4. The method according to claim 2 or 3, characterized in that The target algorithm is a dynamic threshold verification algorithm. According to the target algorithm, the plurality of observation data and the plurality of forecast data are verified and processed to obtain a first indicator value corresponding to the forecast mode, including: Get the preset sliding window half width; Determining a first time period according to the target time period and the preset sliding window half width; Determining a dynamic threshold corresponding to the time point based on a preset quantile level and a plurality of observation data corresponding to the first time period; performing classification processing on the plurality of observation data corresponding to the first time period and the plurality of forecast data corresponding to the first time period according to the dynamic threshold value to obtain observation levels of the plurality of observation data and forecast levels of the plurality of forecast data; Determining, based on the observation levels of the plurality of observation data and the forecast levels of the plurality of forecast data, inspection data corresponding to the time point, the inspection data including the plurality of inspection levels and the number of hits, the number of missed alarms, the number of false alarms, and the number of correct rejections corresponding to each inspection level; A first indicator value corresponding to the forecast mode is determined based on the test data corresponding to a plurality of time points.

5. The method according to claim 4, characterized in that Determining a first indicator value corresponding to the forecast mode according to test data corresponding to a plurality of time points includes: Determining, in descending order of the plurality of inspection levels, level inspection data corresponding to the plurality of inspection levels in sequence from the inspection data corresponding to the plurality of time points, the level inspection data including the number of hits, the number of missed alarms, the number of false alarms, and the number of correct rejections corresponding to the same inspection level at the plurality of time points; For any inspection level, determining the inspection score of the inspection level according to the level inspection data corresponding to the inspection level; The first indicator value corresponding to the forecast mode is determined according to the inspection scores corresponding to the multiple inspection levels and the weight values ​​corresponding to the multiple inspection levels.

6. The method according to claim 2 or 3, characterized in that The target algorithm is a continuous variable testing algorithm. According to the target algorithm, the plurality of observation data and the plurality of forecast data are tested and processed to obtain a first indicator value corresponding to the forecast mode, including: Obtaining a latitude weighted value corresponding to the target area; Based on the multiple observation data and the multiple forecast data, and the latitude weighted value, the target algorithm is used to determine multiple first indicator values ​​corresponding to the forecast mode, wherein the first indicator values ​​include an anomaly correlation coefficient value, a deviation value, and a root mean square error value.

7. The method according to claims 1-6, characterized in that For any forecast mode, determining the degradation factor corresponding to the forecast mode according to the inspection index value includes: Determine the indicator type corresponding to the inspection indicator value; Determine the target formula and reference value based on the indicator type; Determining a degradation index corresponding to the inspection index value by using the target formula according to the inspection index value and the reference value; A degradation factor corresponding to the forecast mode is determined according to the degradation index and weight values ​​of multiple sites in the target area.

8. A device for verifying forecasts of a forecast model, characterized in that: The device comprises: An acquisition module is used to acquire target parameters input by a user, wherein the target parameters include a target area, meteorological elements, and a target time period; A determination module, configured to determine a plurality of observation data of the target area within the target period and a plurality of forecast data corresponding to the plurality of forecast modes; A first verification module is configured to perform verification processing on the plurality of observation data and the plurality of forecast data according to the meteorological elements to obtain a verification index value corresponding to each forecast mode; a second inspection module, configured to determine a degradation factor corresponding to each forecast mode according to the inspection index value, wherein the degradation factor is used to represent a driving contribution of the meteorological element to the degradation of forecast skill under the forecast mode; The display module is used to generate visual inspection results corresponding to multiple forecast modes according to multiple inspection index values ​​and multiple degradation factors.

9. An electronic device, characterized in that: include: a processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that The invention comprises a computer program, which implements the method according to any one of claims 1 to 7 when executed by a processor.

Images