A meteorological drought prediction method and system based on WRF model

Abstract

This invention discloses a meteorological drought prediction method and system based on a WRF model, belonging to the field of meteorological drought prediction technology. The method includes: acquiring driving data; constructing a WRF model for the target region and setting initial physical parameter combinations; running the model and saving precipitation results, evaluating them from the dimensions of precipitation deviation ratio, wet day frequency deviation, and intensity frequency ratio; when preset rules are not met, optimizing parameters through dynamic weight parameter interaction constraints, driving data uncertainty quantification feedback, and physical process consistency verification until the optimal parameter combination is determined. Based on future driving data from global climate models and the optimal parameter model operation, drought assessment is completed by combining the SPI index. This invention achieves a comprehensive improvement in prediction accuracy, stability, physical interpretability, optimization efficiency, and scenario adaptability, providing efficient and reliable technical support for the scientific management and decision-making of drought disasters.

Description

Technical Field

[0001] This invention relates to the field of drought prediction technology, specifically to a meteorological drought prediction method and system based on the WRF model. Background Technology

[0002] Drought refers to a natural phenomenon caused by persistent water shortages in meteorological, climatic, hydrological, and ecosystem systems. It is characterized by high frequency, long duration, strong destructiveness, and insidious nature. Once drought occurs, it has a significant impact on natural ecosystems and human activities. Based on the affected areas, droughts are classified into meteorological droughts, hydrological droughts, agricultural droughts, and socioeconomic droughts. Meteorological drought is the most fundamental type of drought, forming the basis for other types. It is on this basis that meteorological drought further spreads and develops, ultimately leading to other types of drought events. Therefore, accurate early drought prediction and proactive response measures are crucial for reducing disaster losses and improving risk management.

[0003] Despite significant progress in statistical modeling, climate dynamics simulation, and machine learning in recent years, meteorological drought prediction still faces several shortcomings. First, data and model uncertainties are substantial, especially in regions where high-spatial-resolution meteorological data is limited. Model-driven biases and observational errors significantly impact prediction accuracy. Second, traditional statistical models struggle to characterize the nonlinear and spatiotemporal dependencies between meteorological factors, while deep learning models, despite their strong fitting capabilities, generally lack physical interpretability and generalization ability. Third, climate models (GCM / RCM) primarily focus on large-scale studies and suffer from systematic biases. Furthermore, the applicability of bias correction methods is limited, leading to considerable uncertainty in regional-scale predictions. Fourth, while downscaling using global and regional climate models can characterize small- and mesoscale climate phenomena and address the issue of significant systematic biases, these methods typically employ a combination of global climate models and mathematical statistical downscaling, which cannot accurately characterize the spatiotemporal dependencies of drought factors. A method for dynamic downscaling drought prediction using WRF small- and mesoscale models combined with global climate models has not yet been developed, and an effective method for optimizing complex parameters within WRF models has not yet been proposed. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a meteorological drought prediction method and system based on the WRF model, aiming to solve the problems in the background technology.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a meteorological drought prediction method based on a WRF model, comprising the following steps:

[0006] Step S1: Obtain driving data and construct a WRF model for the target area, and determine the time range of the WRF model simulation and the latitude and longitude boundaries of the target area;

[0007] Step S2: Based on the constructed WRF pattern, set the initial physical parameter combination of the WRF model;

[0008] Step S3: Based on the initial combination of physical parameters set for the WRF model, start the WRF model; after the model runs, save the precipitation results.

[0009] Step S4: Evaluate the precipitation results, including precipitation deviation ratio, wet day frequency deviation, and intensity frequency ratio.

[0010] Step S5: If the precipitation evaluation result does not meet the preset evaluation rules, the initial physical parameter combination of the WRF model is optimized until the preset evaluation rules are met; if the evaluation result meets the preset evaluation rules, the current parameters are selected as the best physical parameter combination.

[0011] Step S6: Obtain driving data for future periods of the global climate model. Based on the driving data for future periods of the global climate model and the optimal combination of physical parameters, start the WRF model and output precipitation values.

[0012] Step S7: Based on the precipitation value and the calculation method of the SPI drought index, forecast the SPI drought index for the future study period and complete the drought status assessment.

[0013] Furthermore, in step S2, the physical parameters in the initial combination of physical parameters of the WRF model include microphysical parameters, longwave radiation parameters, shortwave radiation parameters, planetary boundary layer parameters, cumulus cloud parameters, and land surface scheme parameters.

[0014] Furthermore, in step S5, the specific process of optimizing the initial combination of physical parameters of the WRF model is as follows:

[0015] Step S5.21, Construction of dynamic weight parameter interaction constraints: Let the first... The microphysical parameters are , No. The longwave radiation parameters are: , No. The shortwave radiation parameters are: , No. The boundary layer parameters of the planets are: , No. The parameters of the cumulus cloud are: , No. The parameters for each land surface scheme are: Based on the historical drought spatiotemporal characteristics of the target area, subdivided scenarios are defined, and a parameter interaction weight matrix is ​​generated to constrain the interaction relationship of the initial physical parameters of the WRF model.

[0016] Step S5.22, Uncertainty Quantification and Stability Verification of Driving Data: The uncertainty of the driving data is quantified by ensemble Kalman filtering to generate a disturbance driving dataset. The fluctuation coefficient of the evaluation index of the physical parameter combination under multiple sets of disturbance data is calculated to increase the stability constraint.

[0017] Step S5.23, Physical process consistency verification: Screen key physical variables related to drought, construct physical variable deviation formulas, and verify the rationality of physical processes corresponding to combinations of physical parameters;

[0018] Step S5.24, Iterative Optimization and Termination Judgment: If the combination of physical parameters simultaneously meets the preset evaluation rules, stability verification and physical process consistency verification, it is determined to be the optimal combination of physical parameters; if not, the physical parameters are adjusted based on feedback orientation, and steps S3-S4 and steps S5.21-S5.23 are repeated until they are met.

[0019] Furthermore, the specific process of step S5.21 is as follows:

[0020] Step S5.211: Divide the target area into subdivided scenarios: Based on historical drought data of the target area, divide it into six scenarios; the six scenarios include "drought core area", "peripheral area", "dry season", "normal water season" and "wet season";

[0021] Step S5.212, Calculate the parameter interaction sensitivity coefficient: Define the parameter interaction sensitivity coefficient as follows:

[0022] ;

[0023] In the formula, Indicates the first Type of physical parameters and the first The interaction sensitivity coefficient of a physical parameter; This represents the precipitation results simulated by the WRF model; , They represent the first Type of physical parameters and the first Physical parameters; , They represent the first Type of physical parameters and the first Historical average values ​​corresponding to various physical parameters; This represents the average of the simulated precipitation results; Indicates the sign of the partial derivative;

[0024] Step S5.213: Generate a scenario-based weight matrix: For the first... Scene , , construct the first Parameter interaction weight matrix for similar scenarios The elements of the parameter interaction weight matrix are represented as follows: , express The weighting coefficients; during the optimization process, the corresponding parameter interaction weight matrix is ​​invoked based on the current scenario. The direction is adjusted by constraining the combination of physical parameters.

[0025] Furthermore, the specific process of step S5.22 is as follows:

[0026] Step S5.221: Generate the perturbation-driven dataset: Perturb the driving data using ensemble Kalman filtering to generate... Group perturbation dataset, denoted as , Indicates the generated first Group perturbation dataset;

[0027] Step S5.222: Calculate the fluctuation coefficient of the evaluation index: (The text abruptly ends here, likely due to an incomplete sentence or a formatting error.) Combination of physical parameters Inputting a WRF model, using multiple sets of perturbation datasets as driving data, yields multiple sets of evaluation metrics; these evaluation metrics include: set 1 to set 2. The precipitation bias ratio corresponding to the perturbation datasets, from group 1 to group 2. The wet day frequency deviations corresponding to the perturbation datasets, from group 1 to group 2. The intensity-frequency ratio of the group of perturbation datasets; the fluctuation coefficient of each evaluation index is calculated. ;

[0028] Step S5.223, Stability Screening: Set a stability constraint threshold; if the... Combination of physical parameters Evaluation index fluctuation coefficient If all values ​​are less than or equal to the stability constraint threshold, then proceed to the physical process consistency check; otherwise, adjust the physical parameters in the physical parameter combination that are sensitive to the driving data and recalculate the fluctuation coefficient.

[0029] Furthermore, the specific process of step S5.23 is as follows:

[0030] Step S5.231: Screening key physical variables: Select physical variables directly related to drought formation, including land surface evapotranspiration. Planetary boundary layer height Vertical gradient of soil moisture Determine reasonable ranges for physical variables directly related to drought formation based on historical observation data of the target area;

[0031] Step S5.232, Calculate the deviation of physical variables: Define the deviation of physical variables. , represented as:

[0032] ;

[0033] In the formula, Indicates the first Combination of physical parameters The WRF model simulates the specific values ​​of physical variables directly related to drought formation. , These represent the upper and lower limits of the reasonable range for the corresponding physical variables, respectively.

[0034] Step S5.233, Consistency Judgment: If the deviation of all physical variables... If all values ​​are equal to 0, the consistency check passes; if any physical variable deviates... If the value is greater than 0, then the physical parameters corresponding to that physical variable should be adjusted first, and the deviation of the physical variable should be re-simulated and calculated. .

[0035] Furthermore, the driving data utilizes the global climate reanalysis dataset.

[0036] A meteorological drought prediction system based on the WRF model includes:

[0037] The WRF pattern construction module acquires driving data and constructs a WRF pattern for the target area, determining the time range of the WRF pattern simulation and the latitude and longitude boundaries of the target area.

[0038] The physical parameter setting module sets the initial physical parameter combination of the WRF model based on the constructed WRF pattern.

[0039] The initial model running module executes the WRF model run based on the set initial physical parameter combination; after the model run is completed, the precipitation results are saved.

[0040] The evaluation module performs an evaluation of precipitation results, with evaluation dimensions including precipitation deviation ratio, wet day frequency deviation, and intensity frequency ratio.

[0041] The parameter optimization module optimizes the initial physical parameter combination of the WRF model until the precipitation evaluation results meet the preset evaluation rules if the precipitation evaluation results do not meet the preset evaluation rules. If the evaluation results meet the preset evaluation rules, the current parameters are selected as the optimal physical parameter combination.

[0042] The model running module is optimized to acquire driving data for future periods of the global climate model. Based on the driving data for future periods of the global climate model and the optimal combination of physical parameters, the WRF model is started and the precipitation value is output.

[0043] The drought prediction module calculates the SPI drought index based on precipitation data and the SPI drought index, and then performs a drought status assessment.

[0044] An electronic device includes a processor, a memory, and a bus, wherein the processor and the memory are connected via the bus, the memory is used to store a set of program code, and the processor is used to call the program code stored in the memory to execute a meteorological drought prediction method based on a WRF model.

[0045] A non-volatile computer storage medium storing computer-executable instructions that execute a meteorological drought prediction method based on a WRF model.

[0046] Compared with existing technologies, the present invention has the following advantages:

[0047] (1) This invention achieves a comprehensive improvement in the accuracy, stability, physical interpretability, optimization efficiency and scenario adaptability of meteorological drought prediction through the collaborative innovation of dynamic weight parameter interaction constraints, driving data uncertainty quantification feedback and physical process consistency verification. It not only accurately depicts the nonlinear correlation of meteorological parameters and the physical mechanism of drought formation under different spatiotemporal scenarios, but also effectively filters out data uncertainty interference, reduces blind iteration, and makes the prediction results accurate and clear. At the same time, it has the ability to adapt to multiple scenarios, providing efficient and reliable technical support for the scientific management and decision-making of drought disasters.

[0048] (2) This invention addresses the core problems of poor spatiotemporal adaptability and lack of physical interpretability of parameters in traditional methods. It generates a parameter interaction weight matrix based on the scenario through dynamic weight parameter interaction constraints, and accurately matches spatiotemporal features. Then, through physical process consistency verification, it ensures that the parameter combination conforms to the physical laws of key variables such as land surface evapotranspiration and planetary boundary layer height, thus fundamentally solving the pain points of insufficient spatiotemporal adaptability and lack of physical interpretation.

[0049] (3) In view of the problems of large uncertainty in data and model and low efficiency of parameter optimization, this invention drives the uncertainty of data by quantizing the Kalman filter and screening anti-interference parameter combinations to improve the prediction consistency under different data conditions; then, the parameters are adjusted in a progressive process to reduce invalid iterations, which solves the stability problem caused by data interference and improves the efficiency of parameter optimization.

[0050] (4) This invention addresses the problems of insufficient regional adaptability of the model and weak practical decision support. By using scenario-based weight matrix and dynamic physical variable verification, the method can be flexibly adapted to different climate zones and drought types. At the same time, the accurate and stable output results provide high-fidelity data support for disaster prevention and mitigation decision-making. Attached Figure Description

[0051] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0052] like Figure 1 As shown, the present invention provides a technical solution: a meteorological drought prediction method based on the WRF model, comprising the following steps:

[0053] Step S1: Obtain driving data (driving data uses the global climate reanalysis dataset (ERA5), specifically the land surface reanalysis dataset (ERA5_Land)), and construct a WRF model for the target area, determining the time range of the WRF model simulation and the latitude and longitude boundaries of the target area (spatial resolution is set to no higher than 3km*3km).

[0054] Step S2: Based on the constructed WRF pattern, set the initial physical parameter combination of the WRF model.

[0055] Step S3: Based on the initial combination of physical parameters set for the WRF model, start the WRF model to run; after the model runs, save the precipitation results (the precipitation results are saved in the wrfout file, and the data is stored in netcdf format, and then the precipitation data is extracted by the Python tool for further processing).

[0056] Step S4: Evaluate the precipitation results. The evaluation dimensions include precipitation deviation ratio, wet day frequency deviation, and intensity frequency ratio.

[0057] Step S5: If the precipitation evaluation result does not meet the preset evaluation rules, the initial physical parameter combination of the WRF model is optimized until the preset evaluation rules are met; if the evaluation result meets the preset evaluation rules, the current parameters are selected as the best physical parameter combination.

[0058] Step S6: Obtain driving data for future periods from global climate models (GCMs). Based on the driving data for future periods from global climate models and the optimal combination of physical parameters, start the WRF model and output precipitation values.

[0059] Step S7: Based on the precipitation value and the calculation method of the SPI drought index, forecast the SPI drought index for the future study period and complete the drought status assessment.

[0060] In step S2, the physical parameters in the initial combination of physical parameters of the WRF model include microphysical parameters, longwave radiation parameters, shortwave radiation parameters, planetary boundary layer parameters, cumulus cloud parameters, and land surface scheme parameters.

[0061] In step S4:

[0062] Precipitation deviation ratio The calculation formula is: , express The precipitation results at any given time. express Observed precipitation at any given time;

[0063] wet day frequency deviation The calculation formula is: , Indicates the number of elements in the set. Indicates the precipitation threshold;

[0064] The formula for calculating the intensity-frequency ratio is: , Indicates the first The intensity frequency ratio of each precipitation percentile, Indicates the first A set of time indices corresponding to each precipitation percentile.

[0065] In step S5, the preset evaluation rule is:

[0066] The precipitation deviation ratio is within 0.9 ≤ BR ≤ 1.1;

[0067] The frequency deviation for wet days is within 0.8 ≤ WFB ≤ 1.1;

[0068] The intensity-frequency ratio satisfies , , ; Indicates that the precipitation intensity is at Percentile intensity-frequency ratio, Indicates that the precipitation intensity is at Percentile intensity-frequency ratio, Indicates that the precipitation intensity is at The intensity-frequency ratio of the percentile segment.

[0069] In step S5, the specific process of optimizing the initial combination of physical parameters of the WRF model is as follows:

[0070] Step S5.21, Construction of dynamic weight parameter interaction constraints: Let the first... The microphysical parameters are , No. The longwave radiation parameters are: , No. The shortwave radiation parameters are: , No. The boundary layer parameters of the planets are: , No. The parameters of the cumulus cloud are: , No. The parameters for each land surface scheme are: Based on the historical drought spatiotemporal characteristics of the target area, subdivided scenarios are defined, and a parameter interaction weight matrix is ​​generated to constrain the interaction relationship of the initial physical parameters of the WRF model.

[0071] Step S5.22, Uncertainty Quantification and Stability Verification of Driving Data: The uncertainty of the driving data (land surface reanalysis dataset (ERA5_Land)) is quantified using ensemble Kalman filtering (EnKF) to generate a disturbance driving dataset. The fluctuation coefficient of the evaluation index under multiple sets of disturbance data is calculated, and stability constraints are added.

[0072] Step S5.23, Physical process consistency verification: Screen key physical variables related to drought, construct physical variable deviation formulas, and verify the rationality of physical processes corresponding to combinations of physical parameters.

[0073] Step S5.24, Iterative Optimization and Termination Judgment: If the combination of physical parameters simultaneously meets the preset evaluation rules, stability verification and physical process consistency verification, it is determined to be the optimal combination of physical parameters; if not, the physical parameters are adjusted based on feedback orientation, and steps S3-S4 and steps S5.21-S5.23 are repeated until they are met.

[0074] The specific process of step S5.21 is as follows:

[0075] Step S5.211: Divide the target area into subdivided scenarios: Based on the historical drought data of the target area (frequency of occurrence, duration, and scope of impact), divide the area into six scenarios; the six scenarios include "drought core area", "peripheral area", "dry season", "normal water season" and "wet season".

[0076] Step S5.212, Calculate the parameter interaction sensitivity coefficient: Define the parameter interaction sensitivity coefficient as follows:

[0077] ;

[0078] In the formula, Indicates the first Type of physical parameters and the first The interaction sensitivity coefficient of a physical parameter; This represents the precipitation results simulated by the WRF model; , They represent the first Type of physical parameters and the first Physical parameters; , They represent the first Type of physical parameters and the first Historical average values ​​corresponding to various physical parameters; This represents the average of the simulated precipitation results; This indicates the partial derivative sign.

[0079] Step S5.213: Generate a scenario-based weight matrix: For the first... Scene , , construct the first 6×6 Parametric Interaction Weight Matrix for Scenario The elements of the parameter interaction weight matrix are represented as follows: , express The weighting coefficients; during the optimization process, the corresponding parameter interaction weight matrix is ​​invoked based on the current scenario. The direction is adjusted by constraining the combination of physical parameters.

[0080] The specific process of step S5.22 is as follows:

[0081] Step S5.221: Generate the perturbation-driven dataset: Perturb the driving data using ensemble Kalman filtering to generate... Group perturbation dataset, denoted as , Indicates the generated first Group perturbation dataset.

[0082] Step S5.222: Calculate the fluctuation coefficient of the evaluation index: (The text abruptly ends here, likely due to an incomplete sentence or a formatting error.) Combination of physical parameters Inputting a WRF model, using multiple sets of perturbation datasets as driving data, yields multiple sets of evaluation metrics; these evaluation metrics include: set 1 to set 2. The precipitation bias ratio corresponding to the perturbation datasets, from group 1 to group 2. The wet day frequency deviations corresponding to the perturbation datasets, from group 1 to group 2. The intensity-frequency ratio of the group of perturbation datasets; the fluctuation coefficient of each evaluation index is calculated. , represented as:

[0083] ;

[0084] In the formula, Indicates the first Evaluation metric values ​​corresponding to the perturbation dataset; express The mean of the evaluation index values ​​corresponding to the perturbation dataset.

[0085] Step S5.223, Stability Screening: Let... As the stability constraint threshold, if the th Combination of physical parameters Evaluation index fluctuation coefficient If all values ​​are ≤5%, proceed to physical process consistency verification; otherwise, adjust the physical parameters in the physical parameter combination that are sensitive to the driving data (prioritize adjusting planetary boundary layer parameters and land surface scheme parameters) and recalculate the fluctuation coefficient.

[0086] The specific process of step S5.23:

[0087] Step S5.231: Screening key physical variables: Select physical variables directly related to drought formation, including land surface evapotranspiration. Planetary boundary layer height Vertical gradient of soil moisture Based on historical observation data of the target area, determine the reasonable range of physical variables directly related to drought formation.

[0088] Step S5.232, Calculate the deviation of physical variables: Define the deviation of physical variables. , represented as:

[0089] ;

[0090] In the formula, Indicates the first Combination of physical parameters The WRF model simulates the specific values ​​of physical variables directly related to drought formation. , These represent the upper and lower limits of the reasonable range for the corresponding physical variables, respectively.

[0091] Step S5.233, Consistency Judgment: If the deviation of all physical variables... If all values ​​are equal to 0, the consistency check passes; if any physical variable deviates... If the value is greater than 0, then the physical parameters corresponding to that physical variable (such as land surface evapotranspiration) should be adjusted first. When deviations exceed limits, adjust the land surface scheme parameters, resimulate and calculate the deviations of physical variables. .

[0092] In step S7, the formula for calculating the SPI drought index is: , The probability density function representing the SPI drought index is... Indicates precipitation. This represents shape parameters used to fit the morphological characteristics of precipitation distribution. This represents a scale parameter used to adjust the scale range of precipitation distribution. Represents the gamma function. It represents the base of the natural logarithm.

[0093] A meteorological drought prediction system based on the WRF model includes:

[0094] The WRF pattern construction module acquires driving data and constructs a WRF pattern for the target area, determining the time range of the WRF pattern simulation and the latitude and longitude boundaries of the target area.

[0095] The physical parameter setting module sets the initial physical parameter combination of the WRF model based on the constructed WRF pattern.

[0096] The initial model running module executes the WRF model run based on the set initial physical parameter combination; after the model run is completed, the precipitation results are saved.

[0097] The evaluation module performs an evaluation of precipitation results, with evaluation dimensions including precipitation deviation ratio, wet day frequency deviation, and intensity frequency ratio.

[0098] The parameter optimization module optimizes the initial physical parameter combination of the WRF model until the precipitation evaluation results meet the preset evaluation rules if the precipitation evaluation results do not meet the preset evaluation rules. If the evaluation results meet the preset evaluation rules, the current parameters are selected as the optimal physical parameter combination.

[0099] The model running module is optimized to acquire driving data for future periods of the global climate model. Based on the driving data for future periods of the global climate model and the optimal combination of physical parameters, the WRF model is started and the precipitation value is output.

[0100] The drought prediction module calculates the SPI drought index based on precipitation data and the SPI drought index, and then performs a drought status assessment.

[0101] An electronic device includes a processor, a memory, and a bus, wherein the processor and the memory are connected via the bus, the memory is used to store a set of program code, and the processor is used to call the program code stored in the memory to execute a meteorological drought prediction method based on a WRF model.

[0102] A non-volatile computer storage medium storing computer-executable instructions that execute a meteorological drought prediction method based on a WRF model.

[0103] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for predicting meteorological drought based on a WRF model, characterized in that, Comprising the following steps: Step S1: Obtain driving data and construct a WRF model for a target area, determine the time range simulated by the WRF model, and the latitude and longitude boundaries of the target area; Step S2: On the basis of the constructed WRF model, set the initial physical parameter combination of the WRF model; Step S3: Start the WRF model based on the set initial physical parameter combination of the WRF model; after the model runs, save the precipitation results; Step S4: Evaluate the precipitation results, and the evaluation dimensions include precipitation bias ratio, wet day frequency bias, and intensity frequency ratio; Step S5: If the evaluation results of the precipitation do not meet the preset evaluation rules, optimize the initial physical parameter combination of the WRF model until the preset evaluation rules are met; if the evaluation results meet the preset evaluation rules, select the current parameters as the best physical parameter combination; Step S6: Obtain driving data for the future period of the global climate model, start the WRF model based on the driving data for the future period of the global climate model and the best physical parameter combination, and output the precipitation value; Step S7: According to the precipitation value, combine the calculation method of the SPI drought index, and predict the SPI drought index of the future research period to complete the drought condition evaluation; In step S2, the physical parameters in the initial physical parameter combination of the WRF model include microphysical parameters, long-wave radiation parameters, short-wave radiation parameters, planetary boundary layer parameters, cumulus parameters, and land surface scheme parameters; In step S5, the specific process of optimizing the initial physical parameter combination of the WRF model is as follows: Step S5.21, dynamic weight parameter interaction constraint construction: set the first microphysical parameter as , the first longwave radiation parameter as , the first shortwave radiation parameter as , the first planetary boundary layer parameter as , the first cumulus parameter as , and the first land surface scheme parameter as ; Based on the historical spatiotemporal characteristics of the target area, the parameter interaction weight matrix is generated to constrain the interaction relationship of the initial physical parameters of the WRF model; Step S5.22, uncertainty quantification and stability verification of driving data: the ensemble Kalman filter is used to quantize the uncertainty of the driving data, generate a set of perturbed driving data, calculate the evaluation index fluctuation coefficient of the physical parameter combination under multiple sets of perturbed data, and increase the stability constraint; Step S5.23, consistency verification of physical process: screening the key physical variables related to drought, constructing a physical variable deviation formula, and verifying the rationality of the physical process corresponding to the physical parameter combination; Step S5.24, iterative optimization and termination judgment: if the physical parameter combination meets the preset evaluation rules, stability verification, and physical process consistency verification at the same time, it is determined as the best physical parameter combination; if not, the physical parameters are adjusted based on feedback direction, and steps S3-S4 and steps S5.21-S5.23 are repeated until the conditions are met. 2.The weather and drought prediction method based on the WRF model of claim 1, wherein: The specific process of step S5.21 is as follows: Step S5.211, dividing the target area into sub-scenes: based on the historical drought data of the target area, five types of scenes are divided; the five types of scenes include "drought core area", "marginal area", "dry period", "normal period", and "wet period"; Step S5.212, calculating the parameter interaction sensitivity coefficient: the parameter interaction sensitivity coefficient is defined and expressed as: ； In the formula, represents the interaction sensitivity coefficient of the first physical parameter and the second physical parameter; represents the precipitation result simulated by the WRF model; respectively represent the first physical parameter and the second physical parameter; respectively represent the historical mean value corresponding to the first physical parameter and the second physical parameter; represents the mean value of the simulated precipitation result; represents the partial derivative symbol;​​​​​​​​ Step S5.213, generating a scenarioed weight matrix: for the i-th scenario Class scenario , , constructing a parameter interaction weight matrix of the i-th class scenario Class scenario , the parameter interaction weight matrix element is expressed as: , The weight coefficient of the i-th scenario is represented as ; the corresponding parameter interaction weight matrix is called based on the current scenario attribution during the optimization process , and the constraint physical parameter combination adjustment direction. 3.The weather and drought prediction method based on the WRF model of claim 2, wherein: The specific process of step S5.22 is as follows: Step S5.221, generating perturbed driving dataset: perturbing the driving dataset using ensemble Kalman filter to generate the set of perturbed dataset, denoted as , the generated first set of perturbed dataset; Step S5.222: Calculate the fluctuation coefficient of the evaluation index: (The text abruptly ends here, likely due to an incomplete sentence or a formatting error.) Combination of physical parameters Inputting a WRF model, using multiple sets of perturbation datasets as driving data, yields multiple sets of evaluation metrics; these evaluation metrics include: set 1 to set 2. The precipitation bias ratio corresponding to the perturbation datasets, from group 1 to group 2. The wet day frequency deviations corresponding to the perturbation datasets, from group 1 to group 2. The intensity-frequency ratio of the group of perturbation datasets; the fluctuation coefficient of each evaluation index is calculated. ; Step S5.223, Stability Screening: Set a stability constraint threshold; if the... Combination of physical parameters Evaluation index fluctuation coefficient If all values ​​are less than or equal to the stability constraint threshold, then proceed to the physical process consistency check; otherwise, adjust the physical parameters in the physical parameter combination that are sensitive to the driving data and recalculate the fluctuation coefficient. 4.The weather and drought prediction method based on the WRF model of claim 3, wherein: The specific process of step S5.23 is as follows: Step S5.231, screening key physical variables: selecting physical variables directly related to drought formation, including land surface evapotranspiration , planetary boundary layer height , vertical gradient of soil moisture ; determining reasonable intervals of physical variables directly related to drought formation based on historical observation data of the target region; Step S5.232, Compute physical variable deviation: define the physical variable deviation is expressed as: ； wherein represents the th physical parameter combination the specific values of the physical variables directly related to drought formation simulated in the WRF model; , respectively represent the upper and lower limits of the reasonable interval of the corresponding physical variable. Step S5.233, Consistency Check: If the deviation of all physical variables is equal to 0, the consistency check is passed. Step S5.233, Consistency Check: If the deviation of all physical variables is equal to 0, the consistency check is passed. If the deviation of any physical variable If the value is greater than 0, then the physical parameters corresponding to that physical variable should be adjusted first, and the deviation of the physical variable should be re-simulated and calculated. .

5. The weather-based drought prediction method based on the WRF model according to claim 4, characterized in that: The driving data uses a global climate reanalysis data set.

6. A weather drought prediction system based on a WRF model, applied to the weather drought prediction method based on the WRF model in any one of claims 1-5, characterized in that, Comprising: The WRF model construction module performs obtaining driving data and constructing a WRF model for a target area, and determines a time range simulated by the WRF model, and a longitude and latitude boundary of the target area; The physical parameter setting module performs setting an initial physical parameter combination of the WRF model based on the constructed WRF model; The initial model running module performs starting the WRF model running based on the set initial physical parameter combination of the WRF model, and saving a precipitation result after the model running ends; The evaluation module performs evaluating the precipitation result, and the evaluation dimensions include a precipitation deviation ratio, a wet day frequency deviation, and an intensity frequency ratio; The parameter optimization module performs optimizing the initial physical parameter combination of the WRF model if the precipitation evaluation result does not satisfy a preset evaluation rule, until the preset evaluation rule is satisfied; and selecting a current parameter as a best physical parameter combination if the evaluation result satisfies the preset evaluation rule; The optimized model running module performs obtaining driving data of a future period of a global climate model, and starting the WRF model running based on the driving data of the future period of the global climate model and the best physical parameter combination, and outputting a precipitation value; The drought prediction module performs predicting an SPI drought index of a future research period according to the precipitation value and a calculation method of the SPI drought index, and completing drought condition evaluation.

7. An electronic device, comprising: The computer executable instructions perform the meteorological drought prediction method based on the WRF model according to any one of claims 1-5.

8. A non-transitory computer storage medium storing computer-executable instructions, the computer-executable instructions comprising instructions for: The computer executable instructions perform the meteorological drought prediction method based on the WRF model according to any one of claims 1-5.

Images