Parameterized combination evaluation method, weather forecast method, equipment, medium and product

Abstract

The invention relates to the technical field of scheme evaluation, in particular to a parameterized combination evaluation method, a weather forecast method, equipment, a medium and a product. The parameterization combination evaluation method comprises the steps of obtaining a forecasting mode of a target scene, combining multiple parameterization schemes according to multiple physical processes in a forecasting modulus, obtaining index variables and observation data of the index variables, conducting simulation forecasting on a combination constructed by the parameterization schemes through forecasting mode application, and obtaining corresponding simulation data. The index variables and the data errors of the index variables are fitted to a precision observation model, the precision observation model is analyzed to select a target parameterization combination of the index variables, and an observation result corresponding to the target parameterization combination is an observation result conforming to performance expectation. According to the method, the technical problems that the evaluation result through a single index variable is lack of comprehensiveness and the selected parameterized combination is lack of precision are solved, and the comprehensiveness and the accuracy of the evaluation result are improved.

Description

Technical Field

[0001] The present application relates to the technical field of solution evaluation, and in particular to a parameterized combination evaluation method and a weather forecasting method, equipment, medium and product. Background Art

[0002] In the evaluation of parameterized schemes, a single-metric static evaluation strategy is often used. For example, in weather forecasting, the simulation accuracy of specific meteorological elements is verified by presetting a fixed combination of physical processes and a single parameterized scheme, relying on empirical rules to select a limited combination of schemes for simulation verification. This technical approach ignores the systematic impact of the synergistic effects of multiple physical processes on multidimensional meteorological elements. Traditional single-metric evaluation methods have difficulty quantifying the sensitivity of different physical processes to multiple metrics, nor can they assess the critical impact of differences in parameterized schemes on the stability of system simulations. As a result, the results of evaluations based on a single indicator variable lack comprehensiveness and the selected parameterized combinations lack precision. Summary of the Invention

[0003] The present application provides a parameterized combination evaluation method and a weather forecast method, device, medium and product to at least solve the technical problems that the results of evaluation through a single indicator variable lack comprehensiveness and the selected parameterized combination lacks accuracy.

[0004] The present application provides a parameterized combination evaluation method, which includes: obtaining a forecast model for a target scenario; wherein the forecast model includes multiple physical processes; combining multiple parameterization schemes to construct multiple parameterized combinations of the forecast model; obtaining indicator variables and their observation data; using the forecast model to apply the parameterized combination to simulate the forecast to obtain simulated data of the output indicator variables; fitting the indicator variables and their data errors to a precision observation model; wherein the data error represents the offset of the simulated data compared to the observed data; analyzing the precision observation model to select a target parameterized combination of the indicator variables; wherein the observation results corresponding to the target parameterized combination meet performance expectations.

[0005] The present application also provides a weather forecasting method, which includes: obtaining a target variable for weather forecasting; selecting a target parameterized combination of the target variable using any parameterized combination evaluation method; and applying the target parameterized combination to a weather forecasting model to perform weather forecasting and obtain a forecast output.

[0006] The present application also provides an electronic device, the electronic device comprising: memory for storing computer programs; A processor is used to implement the steps of any of the above-mentioned parameterized combination evaluation methods when executing a computer program; or to implement the steps of the above-mentioned weather forecast method.

[0007] The present application also provides a computer-readable storage medium, in which a computer program is stored, wherein when the computer program is executed by a processor, the steps of any one of the above-mentioned parameterized combination evaluation methods are implemented; or the steps of the above-mentioned weather forecasting method are implemented.

[0008] The present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of any of the above-mentioned parameterized combination evaluation methods; or implements the steps of the above-mentioned weather forecasting method.

[0009] This application establishes a scenario adaptation foundation for subsequent parametric evaluation by obtaining a target scenario forecast model that includes multiple physical processes; constructs a full combination of multiple parameterization schemes to form a complete coverage of the parameterization space, eliminating the evaluation blind spots caused by empirical selection, and further generates data errors based on the simulated data of the indicator variable observation data and the parameterized combination, establishes an objective quantitative benchmark, fits the indicator variables and their data errors to the precision observation model, and constructs a multi-dimensional performance evaluation method. Finally, the target parameterization combination that meets the multi-indicator performance expectations is dynamically screened out through the model analysis mechanism.

[0010] Therefore, this method can solve the technical problems of lack of comprehensiveness of the evaluation results through a single indicator variable and lack of accuracy of the selected parameterized combination, thereby achieving the technical problem of improving the comprehensiveness and accuracy of the evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0012] Figure 1 An application environment diagram of a parameterized combination evaluation method provided in an embodiment of the present application; Figure 2 A flow chart of a parameterized combination evaluation method provided in an embodiment of the present application; Figure 3 A flow chart of another parameterized combination evaluation method provided in an embodiment of the present application; Figure 4 A schematic diagram of a parameterized combination evaluation result provided in an embodiment of the present application; Figure 5 A schematic diagram of the structure of a parameterized combination evaluation device provided in an embodiment of the present application; Figure 6 A schematic diagram of a weather forecasting method according to an embodiment of the present invention; Figure 7 A schematic structural diagram of a weather forecast device provided in an embodiment of the present application; Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0013] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0014] It should be noted that, in the description of this application, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. The terms "first," "second," etc., in this application are used to distinguish similar objects, and are not used to describe a particular order or sequence.

[0015] It should be noted that the terms "S1", "S2", etc. are used only for the purpose of describing the steps and do not specifically refer to the order or sequence, nor are they used to limit this application. They are merely for the convenience of describing the method of this application and should not be understood as indicating the order of the steps. In addition, the technical solutions between the various embodiments can be combined with each other, but this must be based on the fact that ordinary technicians in this field can implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0016] In order to solve the technical problems that the results of evaluation through a single indicator variable lack comprehensiveness and the selected parameterization combination lack precision, the present application obtains a forecast model for the target scenario; wherein the forecast model includes multiple physical processes; combines multiple parameterization schemes to construct multiple parameterization combinations of the forecast model; obtains indicator variables and their observation data; uses the forecast model to apply the parameterization combination to simulate the forecast to obtain the simulated data of the output indicator variables; fits the indicator variables and their data errors to a precision observation model; wherein the data error represents the offset of the simulated data compared to the observed data; analyzes the precision observation model to select the target parameterization combination of the indicator variables; wherein the observation results corresponding to the target parameterization combination meet the performance expectations, thereby achieving the technical effect of improving the comprehensiveness and accuracy of the evaluation results.

[0017] In order to make those skilled in the art better understand the present application, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods. The parameterized combination evaluation method provided by the present application can be applied to Figure 1 , Figure 1 This diagram illustrates an application environment for a parameterized combination evaluation method provided in an embodiment of the present application. Terminal 12 communicates with server 14 via a network. Terminal 12 may be, but is not limited to, various personal computers, laptops, smartphones, tablet computers, and portable wearable devices. Server 14 may be implemented as a standalone server or a server cluster consisting of multiple servers.

[0018] In order to enable those skilled in the art to better understand the present application, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0019] The embodiments of the present application provide a parameterized combination evaluation method, and the method is described in detail in conjunction with the execution process of the parameterized combination evaluation method.

[0020] In one embodiment, Figure 2 As shown, Figure 2 A flow chart of a parameterized combination evaluation method provided in an embodiment of the present application.

[0021] S101: Obtain a forecast model of a target scene; wherein the forecast model includes multiple physical processes.

[0022] In this embodiment, the target scenario refers to a specific application scenario for which prediction is required, such as an application environment for a specific weather forecast. A forecast model may refer to a specific model for which prediction is required in the target scenario, such as the WRF (Weather Research and Forecasting) model for simulating short-term weather forecasts, atmospheric processes, and long-term climate simulations. The WRF model is one of the most widely used numerical weather forecast models in the meteorological field. The WRF model can be used for regional weather forecasts, such as forecasting precipitation, temperature, and wind speed, as well as for scientific research, such as climate change and data assimilation.

[0023] Physical processes refer to the parameters that need to be selected in the forecast model for the target scenario. For example, in the WRF model, these are modular components that handle atmospheric phenomena, including microphysical processes, cumulus convection processes, planetary boundary layer processes, land surface processes, longwave radiation processes, and shortwave radiation processes. These physical processes are provided by the forecast model.

[0024] In this embodiment, by establishing a scenario adaptability foundation, we ensure the specific evaluation requirements of the subsequent implementation and avoid the loss of accuracy caused by a general model. At the same time, the modular physical process supports flexible expansion and enhances the versatility of the method.

[0025] S102: Combining multiple parameterization schemes to construct multiple parameterization combinations of forecast models.

[0026] In this embodiment, the parameterization scheme refers to a mathematical implementation algorithm of a physical process.

[0027] Specifically, each type of physical process includes multiple parameterization schemes to choose from, each corresponding to the physical process. After selecting a physical process, multiple parameterization schemes can be selected accordingly. For example, there are nine parameterization schemes available for both long-wave radiation and short-wave radiation microphysics processes, and up to 29 parameterization schemes available for microphysics processes. A parameterization combination refers to the joint configuration of parameterization schemes corresponding to multiple physical processes. For example, a combination may include one microphysics scheme, two short-wave schemes, and three long-wave schemes; or it may include two microphysics schemes, two short-wave schemes, and one long-wave scheme.

[0028] Furthermore, by selecting one or more parameterization schemes corresponding to the physical process, a parameterization combination is generated through full combination to serve as the evaluation space. For example, if three physical processes are selected, three parameterization schemes are selected for each physical process.

[0029] Specifically, taking the WRF mode as an example, for parameterization schemes that specify different physical processes, you can modify the corresponding values of the physical processes in the WRF mode configuration file namelist.input to specify different parameterization schemes. For example, for shortwave radiation processes, specify ra_sw_physics=1, which means using parameterization scheme 1 for shortwave radiation processes; for longwave radiation processes, specify ra_lw_physics=2, which means using parameterization scheme 2 for longwave radiation processes; for microphysical processes, specify mp_physics=3, which means using parameterization scheme 3 for microphysical processes.

[0030] Furthermore, based on the selected parameterization scheme, 27 parameterization combinations are formed, each of which corresponds to a namelist.input file.

[0031] In this embodiment, the objectivity of the evaluation is improved through systematic combination, comprehensive coverage of the evaluation space is achieved, and the foundation is laid for multi-indicator optimization.

[0032] S103: Obtain indicator variables and their observation data.

[0033] In this embodiment, the indicator variable refers to the quantitative objective to be evaluated. For example, in the WRF model, the indicator variable can be precipitation, maximum temperature, minimum temperature, or wind speed. By selecting multiple indicator variables to be evaluated, the limitations of traditional single-indicator evaluation can be overcome.

[0034] Observational data refers to actual monitoring values. Observational data comes from a variety of sources, including but not limited to site observations, satellite observations, radar observations, and reanalysis data products that integrate observations with numerical forecast products.

[0035] In this embodiment, the indicator variables to be evaluated are determined, and objective benchmark data, that is, real observation data of the indicator variables, are obtained.

[0036] S104: Utilizing the forecast model to apply parameterized combination simulation forecast to obtain simulated data of the output indicator variables.

[0037] In this embodiment, simulated forecasting refers to the process of running a forecast model to generate forecast results. For example, in the WRF model, numerical calculations are performed through parameterized combinations. Simulated data refers to the predicted values output by the forecast model, such as simulated precipitation or temperature series.

[0038] In this embodiment, the forecast model simulates the multiple parameterization combinations determined, and the results of the selected indicator variables are obtained as simulation data based on the simulation results. Specifically, as described in step S102, WRF simulations can be performed using the configuration files corresponding to the 27 parameterization scheme combinations, thereby obtaining the output results of 27 WRF simulations as simulation data.

[0039] Specifically, simulated data are compared with observed data. For example, observed precipitation, maximum temperature, and minimum temperature data are compared with WRF model simulations to assess the ability of different parameterization scheme combinations to simulate indicator variables.

[0040] S105: Fitting the indicator variables and their data errors to the precision observation model; wherein the data error represents the offset of the simulated data compared to the observed data.

[0041] In this embodiment, data error is a quantification of the deviation between simulated data and observed data. The accuracy observation model refers to a visual or mathematical evaluation framework for evaluating a multi-parameter combination in a selected target environment using selected indicator variables and data error.

[0042] S106: The analytical accuracy observation model selects a target parameterized combination of indicator variables; wherein the observation result corresponding to the target parameterized combination meets the performance expectation.

[0043] In this embodiment, the target parameterized combination refers to the optimal solution configuration, that is, the parameterized combination with the best overall performance of each indicator variable. Performance expectation means that each indicator variable under the optimal parameterized combination meets the actual business needs.

[0044] In this embodiment, by obtaining a target scenario prediction model containing multiple physical processes, a scenario adaptation foundation is established for subsequent parametric evaluation; a full combination of multiple parameterization schemes is constructed to form a complete coverage of the parameterization space, eliminating the evaluation blind spots caused by empirical selection, and further generating data errors based on the simulated data of the indicator variable observation data and the parameterized combination, establishing an objective quantitative benchmark, fitting the indicator variables and their data errors to the precision observation model, and finally dynamically screening out the target parameterized combination that meets the multi-index performance expectations through the model analysis mechanism. This embodiment can solve the technical problems of the lack of comprehensiveness of the results of evaluation through a single indicator variable and the lack of precision of the selected parameterized combination, and achieve the technical effect of improving the comprehensiveness and accuracy of the evaluation results.

[0045] In one embodiment, Figure 3 As shown, Figure 3 A flowchart of another parameterized combination evaluation method provided in an embodiment of the present application.

[0046] S201: Acquire each physical process and a plurality of preset parameterization schemes of each physical process to form a plurality of preset parameterization combinations.

[0047] In this embodiment, the physical process in the forecast model of the target scene is first obtained.

[0048] Specifically, the WRF model includes a variety of physical processes, including microphysical processes, cumulus convection processes, planetary boundary layer processes, land surface processes, longwave radiation processes, and shortwave radiation processes. For example, you can select shortwave radiation, longwave radiation, and microphysical processes.

[0049] Furthermore, each physical process includes multiple corresponding parameterization schemes. These parameterization schemes are provided by the forecast model and are pre-set parameterization schemes. After selecting a physical process, the user can further select multiple pre-set parameterization schemes for each selected physical process. For example, if the first three parameterization schemes for shortwave radiation, longwave radiation, and microphysical processes are selected, nine parameterization schemes are currently selected.

[0050] Furthermore, these nine parameterization schemes were fully combined, resulting in 27 parameterization combinations. Specifically, the parameterization schemes applied to the physical processes within each parameterization combination were selected from the pre-set parameterization schemes. In particular, the parameterization scheme applied to at least one physical process in two parameterization combinations differed.

[0051] In this example, by selecting key physical processes and configuring diverse parameterization schemes for them, a fully combinatorial parameterization space was constructed, covering all possible scheme combinations. This systematic combination generation approach eliminates the blind spots of traditional empirical selection and supports subsequent collaborative optimization evaluation of multiple indicator variables.

[0052] S202: Obtain indicator variables and their observation data.

[0053] In this embodiment, there are multiple indicator variables, which are used to evaluate the comprehensive performance of parameterized combinations under multiple indicator variables.

[0054] The indicator variables can be selected according to the actual forecast requirements. For example, three indicator variables can be selected: precipitation, maximum temperature, and minimum temperature.

[0055] Furthermore, the corresponding real observation data are obtained based on the selected indicator variables. The observation data can come from site observation data, satellite observation data, radar observation data, and reanalysis data products that integrate observation and numerical forecast products.

[0056] Specifically, the observation data of precipitation, maximum temperature and minimum temperature can be obtained by the aforementioned method.

[0057] In this embodiment, by selecting multi-dimensional indicator variables and integrating multi-source observation data, an objective benchmark is established for subsequent evaluation, eliminating the limitations and subjective bias of single indicator evaluation, thereby laying a data foundation for systematically evaluating the comprehensive simulation capabilities of parameterized solutions for actual business needs.

[0058] S203: Acquire simulation data of the output indicator variables.

[0059] In this embodiment, a parameterized combination formed by the aforementioned parameterization scheme is selected and applied to a forecast model to form a new target forecast model for simulation forecasting.

[0060] Specific parameterized combinations and forecast models are further run until all parameterized combinations are applied to the forecast model and all simulated data for each indicator variable are output.

[0061] Specifically, the 27 parameterization combinations formed above can be selected and applied to the forecast model to form a target forecast model. The target forecast model is then used to simulate and forecast the selected precipitation, maximum temperature, and minimum temperature. Each parameterization combination corresponds to simulated data for precipitation, maximum temperature, and minimum temperature, resulting in 81 sets of simulation data, i.e., 27 parameterization combinations x 3 indicator variables = 81 sets of simulation data.

[0062] In this embodiment, by systematically executing simulation forecasts of fully parameterized combinations, a standardized simulation data set covering a multi-dimensional evaluation space is generated, thereby establishing a unified analysis benchmark for subsequent precision observation models. At the same time, the blind spots of empirical scheme selection are eliminated, ensuring the objectivity of multi-indicator collaborative optimization, and providing core data support for the comprehensive capabilities of precise quantitative parameterization schemes.

[0063] In this embodiment, the observed data and the simulated data may be further processed for resolution.

[0064] Since the simulated data is a regular grid data, while the observed data is usually discrete points or irregular grid data, the spatial distribution of the observed data and the simulated data may not match. Therefore, interpolation can be used to match the resolution of the observed data and the simulated data.

[0065] Specifically, the grid cell where the observed data is located is determined, and the indicator variable at that location is calculated using the difference. The simulated data after the difference is compared with the observed data to assess their accuracy. Methods such as bilinear interpolation, nearest neighbor interpolation, and bicubic interpolation can be used to match the resolution of the two.

[0066] In this example, matching the resolution of observed and simulated data improves comparability between the two. Interpolating the simulated data to the observed data position aligns the simulated and observed data at the same spatial scale, facilitating direct comparative analysis. This method is also applicable to a variety of observational data, demonstrating its versatility.

[0067] S204: Unify the dimensions of the observed data and / or simulated data.

[0068] In this embodiment, since the data of the selected indicator variables may have different units, in order to evaluate the simulation capabilities of different parameterization combinations for indicator variables of different dimensions, it is necessary to perform feature scaling on the simulated data and the observed simulation to unify the dimensions.

[0069] Specifically, the simulated precipitation data at different times and locations simulated by the forecast model of the target scenario are Ps1, Ps2, ..., Psn, and the corresponding observed precipitation data are Po1, Po2, ..., Pon.

[0070] First, the mean and spread of the observed data and the mean and spread of the simulated data are evaluated.

[0071] Specifically, calculate the mean value μ of the observed data o , and the dispersion value σ oBy calculating the difference between the observed data and the average value of the observed data, as the first factor, the ratio of the first factor and the discrete value of the observed data is used as the observed normalized value Po i '.

[0072] Similarly, calculate the mean μ of the simulated data s , and the dispersion value σ s By calculating the difference between the simulated data and the average value of the simulated data, as the second factor, the ratio of the second factor to the dispersion value of the simulated data is used as the simulated normalized value Ps i '.

[0073] The dispersion value is the standard deviation, which can be calculated using the standard deviation formula commonly used in mathematics.

[0074] You can also use minimum-maximum normalization, that is, obtain the maximum and minimum values of Ps1, Ps2, ..., Psn and Po1, Po2, ..., Pon, and convert the maximum and minimum values into numbers between [0, 1]. Similarly, the above minimum-maximum normalization is also performed on the data of other indicator variables, such as performing minimum-maximum normalization on the maximum temperature and minimum temperature indicator variables selected above, and converting the maximum and minimum values into numbers between [0, 1]. Specifically, for example, the maximum value is P max , the minimum value is P min , the normalization method can refer to formula 1-1 and formula 1-2: Ps i '=(Ps i -P min ) / (P max -P min ) Formula 1-1; Po i '=(Po i -P min ) / (P max -P min ) Formula 1-2; Among them, Ps i ' is the simulated normalized value, Po i ' is the observed normalized value.

[0075] In this embodiment, the first factor and the second factor are constructed by independently calculating the average value and the dispersion value of the observation data and the simulation data, and are compared with the corresponding dispersion values respectively to generate dimensionless normalized values of the observation results and the simulation results, thereby eliminating the original dimensional differences of the multi-index variables and achieving cross-indicator comparability; at the same time, the systematic error and the random error are separated to provide standardized input data for the subsequent input accuracy observation model, ensuring that the evaluation results of the multi-parameter combination under the multi-index variables are mathematically consistent.

[0076] S205: Fit the indicator variables and their data errors to the precision observation model.

[0077] In this embodiment, the observed normalized value and the simulated normalized value are used to evaluate the discreteness values of the observed data and the simulated data, the linear relationship strength value between the two, and the deviation value as the data error.

[0078] The linear relationship strength value is the correlation coefficient, the dispersion value and standard deviation, and the deviation value is the root mean square error. Specifically, the correlation coefficient, standard deviation, and root mean square error can be calculated using a calculation formula commonly used in the field of mathematics.

[0079] Furthermore, the discrete degree value, the linear relationship strength value, and the deviation degree value are used as relevant parameters, and the relevant parameters are mapped to the observation model. The accuracy observation model includes an observation chart. The observation chart can be a Taylor diagram.

[0080] Specifically, mapping the relevant parameters to the observation model includes calculating the ratio of the discrete degree value of the simulation result to the discrete degree value of the observation result to obtain a third factor, and using the third factor as the coordinate value of the first direction of the Taylor diagram, where the first direction is usually the horizontal axis; and using the linear relationship strength value as the coordinate value of the second direction of the observation diagram, where the second direction is usually the vertical axis.

[0081] For example, the aforementioned 27 parameterization scheme combinations and 3 indicator variables can form 27×3=81 points on the Taylor diagram.

[0082] In this embodiment, by performing a first normalization process on the observed data to form an observed normalized value, and performing a second normalization process on the simulated data to form a simulated normalized value, the dimensional unification of the indicator variables of different units is achieved, and the evaluation bias caused by unit differences is eliminated. By using the observed normalized value and the simulated normalized value to evaluate the discrete degree value, linear relationship strength value and deviation degree value of the observed data and the simulated data, and mapping these related parameters to the observation model, a multi-dimensional performance evaluation framework is constructed, which improves the comprehensiveness, accuracy and interpretability of the evaluation results, thereby supporting the objective quantitative analysis of the parameterization scheme and avoiding the limitations of the static evaluation of a single indicator.

[0083] S206: Analyze the target parameterized combination of indicator variables in the precision observation model.

[0084] In this embodiment, the reference points of the observation chart are evaluated.

[0085] Specifically, the skill scores of the indicator variables are evaluated separately to obtain the process parameterization combinations corresponding to each indicator variable. The Taylor skill score can be used to evaluate the skill scores of the indicator variables. Based on the Taylor skill score, the points on the Taylor diagram that perform well for different indicator variables are determined. These points correspond to the parameterization combinations that perform well for a particular indicator variable. The center point between these points with the best performance for different indicator variables is then determined and denoted as A, the reference point.

[0086] The Taylor skill score is a numerical summary of the Taylor plot, reflecting a comprehensive measure of forecast skill, where 0 represents the lowest acceptable standard and 1 represents the highest skill level. It is a composite measure of the correlation coefficient, standard deviation, and root mean square error. For any given variance, the Taylor skill score increases monotonically with increasing correlation and, for any given correlation, increases as the simulated variance approaches the observed variance.

[0087] Further, specifically, determining the center point A between the points where different indicator variables perform better includes converting the coordinates of the points where different indicator variables perform better, such as points A1, A2 and A3, from a polar coordinate system to a rectangular coordinate system, calculating the average rectangular coordinate, and then converting from the rectangular coordinate system to a polar coordinate system.

[0088] That is, weight factors are assigned to the linear relationship strength value and the discrete degree value of the process parameterized combination respectively; the linear relationship strength value and the discrete degree value assigned with the weight factors are weighted averaged to obtain the reference linear relationship strength value and the reference discrete degree value; the reference linear relationship strength value is used as the horizontal coordinate value, and the reference discrete degree value is used as the vertical coordinate value to form a reference point.

[0089] Specifically, the weighted average of the correlation coefficients r1, r2, and r3 of the three points A1, A2, and A3 is calculated based on the weights k1, k2, and k3 of different indicator variables, as shown in Formula 2-1: r=(k1*r1+k2*r2+ k3*r3) / (k1+k2+k3) Formula 2-1; Similarly, the weighted average of the root mean square error and standard deviation of points A1, A2, and A3 is calculated. The weights can be the same or different, indicating the corresponding proportion of the indicator variable.

[0090] For example, if the selected indicator variables are precipitation, maximum temperature, and minimum temperature, first calculate the Taylor skill scores of the 27 points corresponding to precipitation, determine point A1 that performs better for the precipitation indicator, and then calculate points A2 and A3 that perform better for the maximum temperature and minimum temperature, respectively. Calculate the center point between these three points (A1, A2, and A3), that is, convert the coordinates of points A1, A2, and A3 from the polar coordinate system to the rectangular coordinate system, calculate the average rectangular coordinate, and then convert from the rectangular coordinate system to the polar coordinate system to obtain the reference point A.

[0091] In this embodiment, the first distances between the coordinate points included in each parameterized combination and the reference point are calculated respectively to obtain one or more first distances of each parameterized combination; wherein the number of the first distances of the parameterized combination is associated with the number of indicator variables.

[0092] For example, calculate the distance between point A and the points corresponding to different parameterized combinations of precipitation, maximum temperature, and minimum temperature, that is, the distance between point A and the 81 points corresponding to the full parameterized combination.

[0093] Furthermore, the first distance average distance of the parameterized combination is evaluated as its selection factor.

[0094] Specifically, for each parameterization combination, the weighted average of the distances between the points corresponding to the precipitation, maximum temperature, and minimum temperature using this parameterization combination is calculated. That is, assuming that for parameterization combination 1, the distances between the points corresponding to precipitation, maximum temperature, and minimum temperature and point A are L1, L2, and L3, respectively, then the weighted average distance between this parameterization combination 1 and the different indicator variables of point A is L1*=k1*L1+k2*L2+ k3*L3) / (k1+k2+k3). Similarly, the weighted average distances L2*, L3*, …, L27* between the points corresponding to the other 26 parameterization combinations and the different indicator variables of point A can be obtained.

[0095] Compare the selection factors corresponding to each parameterized combination, and select the parameterized combination whose selection factor is smaller than the selection factors of the target number of other parameterized combinations as the target parameterized combination.

[0096] For example, the parameterization combination corresponding to the minimum value among L1* to L27* is the target parameterization combination.

[0097] In this example, scoring identifies the points with the best performance for each indicator variable, and a weighted center point is calculated as a comprehensive benchmark for multiple indicators. Polar coordinate distance calculation and weighted averaging are then used to quantify the spatial deviations of all parameterized combinations from the benchmark point. Ultimately, the combination with the minimum weighted distance is selected as the target parameterized combination. This method transforms the evaluation of multidimensional indicators into a unified spatial optimization problem, overcoming the one-sidedness of single-indicator evaluation while also flexibly adapting to business needs through a weighting mechanism, thereby enhancing the scientific nature and business applicability of parameterized combination selection.

[0098] S207: Evaluate the sensitivity of physical processes to indicator variables.

[0099] In this embodiment, the first target distance is used to indicate that the sensitivity of the physical process corresponding to the first target distance to the first target indicator variable is greater than the sensitivity of the physical process corresponding to the second distance to the first target indicator variable.

[0100] First, select the first target indicator variable, that is, select an indicator variable whose sensitivity is to be evaluated from the selected multiple indicator variables as the first target indicator variable. For example, precipitation is selected as the first target indicator variable.

[0101] Furthermore, a process parameterized combination corresponding to the first target indicator variable is obtained, wherein the parameterized combination includes multiple physical processes, and any one physical process among the multiple physical processes is selected as the first selected process.

[0102] For example, a parameter combination among the 27 parameter combinations corresponding to precipitation is selected, and the parameter combination includes a parameterization scheme (a1, b1, c1) corresponding to the longwave radiation process, the shortwave radiation process and the microphysical process respectively.

[0103] Furthermore, the distance between the coordinate point corresponding to any parameterized combination of the first selected process and the coordinate point corresponding to the parameterized combination of other first selected processes is evaluated to obtain a second distance.

[0104] For example, to analyze the sensitivity of precipitation to the long-wave radiation process, it is necessary to calculate the distance between the points corresponding to the other two schemes of the long-wave radiation scheme and (a1, b1, c1), that is, to calculate the distance L between (a2, b1, c1) and (a3, b1, c1) respectively. a2a1 and L a3a1 , and calculate L a2a1 and L a3a1 The average value of is used to obtain the second distance, which is recorded as L*a.

[0105] Similarly, any other physical process among the multiple physical processes is selected as the second selected process, and the distance between the coordinate point corresponding to any parameterized combination of the second selected process and the coordinate point corresponding to the parameterized combination of other second selected processes is evaluated to obtain a third distance.

[0106] For example, the shortwave radiation process of precipitation is calculated to obtain the third distance L*b.

[0107] Similarly, the distance L*c of precipitation to microphysical processes can also be calculated.

[0108] Furthermore, a larger value of the multiple distances is selected as a first target distance, where the first target distance indicates that the parameterized combination corresponding to the distance has a greater sensitivity to the selected first target indicator variable.

[0109] Specifically, the physical process corresponding to the maximum value of L*a, L*b, and L*c is the physical process that is most sensitive to precipitation, or the physical process that has the greatest impact on precipitation.

[0110] In this example, the sensitivity of a physical process is objectively quantified by determining the point corresponding to the best-scoring parameterized combination for a specific indicator variable, fixing the other physical process parameterized combinations, and calculating the average distance between other solutions for the same target physical process and this best-scoring point. By comparing the average distance values of all physical processes, the key physical processes with the greatest impact on the indicator variable are directly identified. This process improves the targeted nature of parameterized combination evaluation, helping users quickly identify the physical process modules that require priority optimization and avoiding blind empirical adjustments, thereby enhancing simulation accuracy and operational practicality.

[0111] S208: Evaluate the differences in parameterization schemes in the physical process.

[0112] In this embodiment, the fourth distance represents the difference between each physical process and the second target indicator variable.

[0113] First, a second target indicator variable is selected, and any one of the multiple physical processes is selected as the third selected process. For example, precipitation is selected as the second target indicator variable, and longwave radiation is selected as the third selected process.

[0114] Furthermore, parameterization schemes of other multiple physical processes are combined, and the distance between the coordinate points corresponding to each combination is calculated to obtain a fourth distance.

[0115] Specifically, the average value of the distances between the points when the parameterized combination of shortwave radiation and microphysical processes is the same is calculated, that is, the distances between the three points represented by (*, b1, c1) are calculated, where * represents a1, a2, a3, and b1, c1 represent the parameterized combination of shortwave radiation and microphysical processes respectively. Similarly, the distances between the three points represented by (*, b1, c2) and the distances between the three points represented by (*, b1, c3), (*, b2, c1), (*, b2, c2), (*, b2, c3), (*, b3, c1), (*, b3, c2), (*, b3, c3) are calculated, and a total of 27 distances (L1 b1c1 ,L2 b1c1 ,L3 b1c1 , L1 b2c1 ,L2 b2c1 ,L3 b2c1 ,L1 b3c1 ,L2 b3c1 ,L3 b3c1 , L1 b1c2 ,L2 b1c2 ,L3 b1c2 , L1 b2c2 ,L2 b2c2 ,L3 b2c2 , L1 b3c2 ,L2 b3c2 ,L3 b3c2 ,L1 b1c3 ,L2 b1c3 ,L3 b1c3 , L1b2c3,L2 b2c3 ,L3 b2c3 , L1 b3c2 ,L2 b3c2 ,L3 b3c3 ).

[0116] Averaging these 27 values yields the fourth distance. This distance represents the degree of difference between each physical process and the second target indicator variable. Specifically, for precipitation, this is the degree of difference (L*aa) between different parameterizations of the longwave radiation process.

[0117] Similarly, the differences L*bb and L*cc between different parameterization combinations of shortwave radiation processes and microphysical processes for precipitation can also be calculated. The physical processes corresponding to the maximum values of L*aa, L*bb, and L*cc are the physical processes of the parameterization combination for precipitation that need to be carefully selected, because the selection of the parameterization combination will lead to large differences in the precipitation simulation results.

[0118] In this example, by calculating the average distance between points corresponding to different parameterization combinations of a physical process for a given indicator variable, this method objectively quantifies the degree of dispersion of the parameterization combinations. By comparing the average distance values across all physical processes, physical processes with significantly different parameterization combinations are accurately identified. This process ensures that users prioritize the physical processes that have the greatest impact on the indicator variable, avoiding arbitrary parameterization combination selection.

[0119] Therefore, according to the parameterized combination evaluation method of step S201 to step S208, it can be obtained Figure 4 , Figure 4 A schematic diagram of a parameterized combination evaluation result provided in an embodiment of the present application.

[0120] In one embodiment, Figure 5 As shown, Figure 5 The schematic diagram of the structure of a parameterized combination evaluation device provided in an embodiment of the present application is as follows: The parameterized combination evaluation device may include a first acquisition module 21, a combination module 22, a second acquisition module 23, a simulation module 24, a fitting module 25, and an analysis module 26.

[0121] The first acquisition module 21 is used to acquire a forecast model of a target scene; wherein the forecast model includes multiple physical processes.

[0122] The combination module 22 is used to combine multiple parameterization schemes to construct multiple parameterization combinations of forecast models.

[0123] The second acquisition module 23 is used to obtain indicator variables and their observation data.

[0124] The simulation module 24 is used to use the forecast model to apply parameterized combination simulation forecast to obtain simulated data of the output indicator variables.

[0125] The fitting module 25 is used to fit the indicator variables and their data errors to the precision observation model; wherein the data error represents the offset of the simulated data compared to the observed data.

[0126] The parsing module 26 is used to analyze the target parameterized combination of indicator variables selected by the accuracy observation model; wherein the observation results corresponding to the target parameterized combination meet the performance expectations.

[0127] Each module in the aforementioned parameterized combination evaluation device may be implemented in whole or in part through software, hardware, or a combination thereof. Each module may be embedded in or independent of a processor within an electronic device in hardware form, or may be stored in a memory within the electronic device in software form, so that the processor can call and execute the corresponding operations of each module.

[0128] In one embodiment, Figure 6As shown, Figure 6 A schematic flow chart of a weather forecasting method provided in an embodiment of the present application.

[0129] S301: Acquire weather data and use the weather data as data to be predicted.

[0130] S302: Perform parameterized combination evaluation according to the parameterized combination evaluation method to obtain the current combination of weather data.

[0131] S303: Use the current combination to perform weather forecast and obtain weather forecast data.

[0132] In this embodiment, the weather forecast method is not only applicable to the evaluation of multiple meteorological forecast variables, but also allows for flexible selection and combination of these indicator variables and their quantities, such as evaluating wind speed, temperature, and precipitation, or cloud cover, temperature, precipitation, and surface temperature. Furthermore, the evaluation period and evaluation area can also be flexibly selected. It is also applicable to the simulation of extreme weather events, such as heavy rain and heat waves.

[0133] In one embodiment, Figure 7 As shown, Figure 7 The schematic diagram of the structure of a weather forecast device provided in an embodiment of the present application is as follows: The parameterized combination evaluation device may include an acquisition module 31 , a selection module 32 and a weather forecast module 33 .

[0134] The acquisition module 31 is used to acquire weather data and use the weather data as the data to be predicted.

[0135] The selection module 32 is used to perform parameterized combination evaluation according to the parameterized combination evaluation method to obtain the current combination of weather data.

[0136] The weather forecast module 33 is used to perform weather forecast using the current combination to obtain weather forecast data.

[0137] An embodiment of the present application further provides an electronic device, including a memory for storing a computer program; and a processor for executing the computer program to perform at least the following steps: Obtain a forecast model for a target scenario; wherein the forecast model includes multiple physical processes; combine multiple parameterization schemes to construct multiple parameterization combinations of the forecast model; obtain indicator variables and their observation data; use the forecast model to apply the parameterization combination to simulate the forecast to obtain simulated data of the output indicator variables; fit the indicator variables and their data errors to a precision observation model; wherein the data error represents the offset of the simulated data compared to the observed data; analyze the precision observation model to select a target parameterization combination of the indicator variables; wherein the observation results corresponding to the target parameterization combination meet the performance expectations.

[0138] In one embodiment, the electronic device may be a server, and its internal structure diagram may be as shown in FIG. Figure 8 As shown. The electronic device includes a processor, memory, network interface and database connected via a system bus. The processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the electronic device is used to store software management data. The network interface of the electronic device is used to communicate with an external terminal via a network connection.

[0139] Those skilled in the art will understand that Figure 8 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the electronic device to which the solution of the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0140] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the computer program can at least perform the following steps: Obtain a forecast model for a target scenario; wherein the forecast model includes multiple physical processes; combine multiple parameterization schemes to construct multiple parameterization combinations of the forecast model; obtain indicator variables and their observation data; use the forecast model to apply the parameterization combination to simulate the forecast to obtain simulated data of the output indicator variables; fit the indicator variables and their data errors to a precision observation model; wherein the data error represents the offset of the simulated data compared to the observed data; analyze the precision observation model to select a target parameterization combination of the indicator variables; wherein the observation results corresponding to the target parameterization combination meet the performance expectations.

[0141] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, 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.

[0142] An embodiment of the present application further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it can perform at least the following steps: Obtain a forecast model for a target scenario; wherein the forecast model includes multiple physical processes; combine multiple parameterization schemes to construct multiple parameterization combinations of the forecast model; obtain indicator variables and their observation data; use the forecast model to apply the parameterization combination to simulate the forecast to obtain simulated data of the output indicator variables; fit the indicator variables and their data errors to a precision observation model; wherein the data error represents the offset of the simulated data compared to the observed data; analyze the precision observation model to select a target parameterization combination of the indicator variables; wherein the observation results corresponding to the target parameterization combination meet the performance expectations.

[0143] Professionals will understand that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the above-described method embodiments. Any reference to memory, storage, database, or other media used in the various embodiments provided herein may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAMbus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM).

[0144] It can be further appreciated that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0145] The above is a detailed introduction to a parameterized combination evaluation method, system, device and storage medium provided by the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method and core ideas of the present application. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the scope of protection of the present application.

Claims

1. A parameterized combination evaluation method, characterized in that: The parameterized combination evaluation method includes: Obtaining a forecast model for a target scenario; wherein the forecast model includes multiple physical processes; Combining multiple parameterization schemes to construct multiple parameterization combinations of the forecast model; Obtain indicator variables and their observation data; Applying the parameterized combination simulation forecast using the forecast model to obtain output simulated data of the indicator variable; Fitting the indicator variable and its data error to a precision observation model; wherein the data error represents the offset of the simulated data compared to the observed data; Analyze the accuracy observation model to select a target parameterized combination of the indicator variables; wherein the observation results corresponding to the target parameterized combination meet performance expectations.

2. The parameterized combination evaluation method according to claim 1, characterized in that: The combining of multiple parameterization schemes to construct multiple parameterization combinations of the forecast model includes: Obtaining a plurality of preset parameterization schemes for each of the physical processes; Constructing a plurality of parameterization combinations, and selecting a parameterization scheme applied to the physical process in each parameterization combination from the preset parameterization schemes; wherein the parameterization scheme applied to at least one physical process in two parameterization combinations is different.

3. The parameterized combination evaluation method according to claim 1 or 2, characterized in that: There are multiple indicator variables; The step of using the forecast model to apply the parameterized combination simulation forecast to obtain output simulation data of the indicator variable includes: Selecting one of the parameterized combinations and applying it to the forecast model to form a target forecast model; Using the target forecast model to simulate and forecast each of the indicator variables to obtain simulated data of the indicator variables; A new parameterization combination is selected and applied to the forecast model to form a new target forecast model for simulation forecasting; until all the parameterization combinations are applied to the forecast model and the simulation data of each indicator variable are output.

4. The parameterized combination evaluation method according to claim 1, characterized in that: Fitting the indicator variables and their data errors to the precision observation model includes: Performing a first normalization process on the observation data to form an observation normalization value; performing a second normalization process on the simulation data to form a simulation normalized value; Using the observed normalized value and the simulated normalized value, the discrete degree values of the observed data and the simulated data, the linear relationship strength value between the two, and the deviation degree value as the data error are evaluated; The discrete degree value, the linear relationship strength value, and the deviation degree value are used as relevant parameters, and the relevant parameters are mapped to the observation model.

5. The parameterized combination evaluation method according to claim 4, characterized in that: The performing a first normalization process on the observation data to form an observation normalization value includes: Evaluate the mean and dispersion values of the observed data; Calculate the difference between the observed data and the average value of the observed data as the first factor; The ratio of the first factor to the observed data dispersion value is used as the observed normalized value; And / or, performing a second normalization process on the simulation data to form a simulation normalized value includes: Evaluate the mean and dispersion values of the simulated data; taking the difference between the simulated data and the average value of the simulated data as the second factor; The ratio of the second factor to the dispersion value of the simulated data is evaluated as the simulated normalized value.

6. The parameterized combination evaluation method according to claim 4, characterized in that: The accuracy observation model includes an observation chart; and mapping the relevant parameters to the observation model includes: Calculating a ratio of the discrete degree value of the simulation data to the discrete degree value of the observation data to obtain a third factor; using the third factor as a coordinate value of a first direction of the observation chart; The linear relationship strength value is used as a coordinate value of a second direction of the observation graph.

7. The parameterized combination evaluation method according to claim 6, characterized in that: The target parameterized combination of the indicator variables selected by analyzing the precision observation model includes: Evaluate the reference points of said observation chart; respectively calculating first distances between coordinate points included in each parameterized combination and the reference point to obtain one or more first distances of each parameterized combination; wherein the number of the first distances of the parameterized combination is associated with the number of the indicator variables; evaluating a first distance average distance of said parameterized combination as a selection factor thereof; The selection factors corresponding to the parameterized combinations are compared, and a parameterized combination whose selection factor is smaller than the selection factors of the target number of other parameterized combinations is selected as the target parameterized combination.

8. The parameterized combination evaluation method according to claim 7, characterized in that: The reference points for evaluating the observation chart include: Evaluating the skill scores of the indicator variables to obtain a process parameterization combination corresponding to each indicator variable; Assigning weight factors to the linear relationship strength value and the discrete degree value of the process parameter combination respectively; The linear relationship strength value and the dispersion degree value after the weight factor is assigned are weighted averaged to obtain the reference linear relationship strength value and the reference dispersion degree value; The reference linear relationship strength value is used as a horizontal coordinate value, and the reference discrete degree value is used as a vertical coordinate value to form the reference point.

9. The parameterized combination evaluation method according to claim 8, characterized in that: The parameterized combination evaluation method further includes: Select the first target indicator variable; Acquire the process parameterization combination corresponding to the first target indicator variable, wherein the parameterization combination includes a plurality of the physical processes; selecting any physical process from the plurality of physical processes as a first selected process, and evaluating a distance between a coordinate point corresponding to any parameterized combination of the first selected process and a coordinate point corresponding to another parameterized combination of the first selected process to obtain a second distance; The second distance is used to identify a sensitivity of the first target indicator variable to the first selected process.

10. The parameterized combination evaluation method according to claim 9, characterized in that: After obtaining the second distance, the method further includes: selecting any other physical process from the plurality of physical processes as a second selected process, and evaluating a distance between a coordinate point corresponding to any parameterized combination of the second selected process and a coordinate point corresponding to another parameterized combination of the second selected process to obtain a third distance; The larger value of the second distance and the third distance is selected as the first target distance, and the first target distance is used to indicate that the sensitivity of the physical process corresponding to the first target distance to the first target indicator variable is greater than the sensitivity of the physical process corresponding to the second distance to the first target indicator variable.

11. The parameterized combination evaluation method according to claim 8, characterized in that: The parameterized combination evaluation method further includes: selecting a second target indicator variable, and selecting any one physical process from the plurality of physical processes as a third selected process; combining parameterization schemes of the other multiple physical processes; Calculating the distance between the coordinate points corresponding to each combination to obtain a fourth distance; The fourth distance is used to represent the difference between each of the physical processes and the second target indicator variable.

12. A weather forecast method, characterized in that: The weather forecast method comprises: Get the target variable of weather forecast; Selecting a target parameterized combination of the target variables using the parameterized combination evaluation method according to any one of claims 1 to 11; The target parameterized combination is applied to a weather forecast model to perform weather forecasting to obtain a forecast output.

13. An electronic device, characterized in that: include: memory for storing computer programs; A processor, configured to implement the steps of the parameterized combination evaluation method according to any one of claims 1 to 11 when executing the computer program; or implement the steps of the weather forecast method according to claim 12.

14. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the computer program implements the steps of the parameterized combination evaluation method according to any one of claims 1 to 11; or implements the steps of the weather forecast method according to claim 12.

15. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the parameterized combination evaluation method according to any one of claims 1 to 11 are implemented; or the steps of the weather forecast method according to claim 12 are implemented.

Images