A gridded precipitation correction method and system
The grid precipitation correction method using Bayesian maximum entropy and multi-source data integration addresses the accuracy and timeliness issues in precipitation forecasting, enhancing hydropower dispatching through improved grid-by-grid corrections.
Patent Information
- Application Number
- CN201810446295.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2018-05-11
- Publication Date
- 2024-12-10
- Estimated Expiration
- 2038-05-11
AI Technical Summary
Current precipitation forecasting methods, particularly for hydropower dispatching, suffer from low accuracy and poor timeliness due to the intermittent nature of precipitation, with existing corrections mainly focused on single stations or local areas, failing to provide grid-by-grid spatial and temporal corrections.
A grid precipitation correction method using Bayesian maximum entropy to adjust residual values in precipitation forecast data, combining data from conventional meteorological stations, weather radar, and Fengyun satellites, and integrating a mesoscale meteorological model with an ensemble Kalman filter assimilation method to improve forecast accuracy and timeliness.
Enhances precipitation forecast accuracy and timeliness, providing robust data support for hydropower scheduling by correcting grid-by-grid precipitation forecasts, thereby improving hydrological forecasting.
Smart Images
Figure CN110472762B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of new energy technology, and in particular to a grid precipitation correction method and system. Background Art
[0002] As the scale of new energy continues to expand, the safety of new energy grid connection becomes more prominent. Take hydropower as an example. During the flood season, the rainfall is abundant, and the hydropower generation capacity is far greater than the power consumption and transmission demand, which makes it difficult to absorb. During the dry season, the hydropower generation capacity drops significantly, and it is necessary to closely track the dynamic changes in hydropower generation capacity and arrange the start-up of thermal power and reservoir water levels. The intermittent nature of hydropower generation has a great impact on the power balance and operation control of the entire network.
[0003] In the current dispatching management, hydropower dispatching mainly relies on precipitation forecasting, and precipitation forecasting mainly relies on traditional empirical forecasting models. Although it has certain guiding significance for precipitation forecasting, it has problems such as low prediction accuracy, relatively rough prediction results, and difficulty in forming effective support for power grid dispatching. Hydropower dispatching also relies on the results of hydrological forecasting models, but precipitation needs to be input into hydrological forecasting models. Therefore, improving the accuracy of precipitation forecasting is a key issue for hydrological researchers. With the continuous improvement of observation methods and the continuous expansion of observation space scales, observation methods such as satellites and radars are becoming more and more mature, but the observation of precipitation can only guarantee accuracy within a certain time and space scale, and the forecast timeliness is very short, which does not meet the needs of hydropower dispatching. Although numerical models have greatly improved the forecast accuracy of precipitation, numerical models are mathematical models based on certain assumptions and still have large systematic deviations.
[0004] Improving the accuracy of precipitation forecasts is mainly achieved through corrections to reduce errors. The existing corrections are mainly targeted at a single station or local precipitation, and it is impossible to perform grid-by-grid statistical corrections on the spatial scale for gridded precipitation. Due to the intermittent nature of precipitation, the current corrections to precipitation are mainly for precipitation area, and there are very few corrections to precipitation in the time series, resulting in poor forecast timeliness. Summary of the invention
[0005] In order to solve the above-mentioned deficiencies in the prior art, the present invention provides a grid precipitation correction method and system.
[0006] The technical solution provided by the present invention is: a grid precipitation correction method, comprising:
[0007] Collect current grid precipitation forecast data based on a pre-established weather forecast system;
[0008] Correcting the current grid precipitation forecast data according to the residual value in the pre-generated precipitation observation data set;
[0009] The residual value in the precipitation observation data set is calculated by: the historical grid precipitation measured value and the historical grid precipitation forecast data.
[0010] Preferably, the correcting the current grid precipitation forecast data according to the residual value in the precipitation observation data set includes:
[0011] The residual value is adjusted by using Bayesian maximum entropy to obtain a predicted average value of the residual value of the to-be-corrected point in the grid;
[0012] The current grid precipitation forecast data is corrected based on the predicted average value.
[0013] Preferably, the method of adjusting the residual value by using Bayesian maximum entropy to obtain a predicted average value of the residual value of the to-be-corrected point in the grid comprises:
[0014] Calculating the prior probability density function of the points to be corrected in the grid using the maximum entropy principle;
[0015] Obtaining a posterior probability density function based on the residual value, the current grid precipitation forecast data and the prior probability density function;
[0016] The predicted average value of the residual values of the points to be corrected in the grid is calculated according to the posterior probability density function.
[0017] Preferably, the maximum entropy principle is used to calculate the prior probability density function of the points to be corrected in the grid, which is calculated as follows:
[0018]
[0019] Where: f G : Prior probability density function; χ pre : random variable value; μ0: constant term of normalized constraints; g: function vector of available core knowledge base; μ: coefficient vector of relative importance of each function.
[0020] Preferably, the posterior probability density function is obtained based on the residual value, the current grid precipitation forecast data and the prior probability density function, and is calculated as follows:
[0021] f K (x k )=A -1 ψ[χ k ,f G (x pre )]=A -1 ∫f G (x k )dx k
[0022] Where: fK : posterior probability density function; χ k : The predicted value of the point to be corrected; A: Normalization parameter.
[0023] Preferably, the predicted average value of the residual value of the to-be-corrected point in the grid is calculated according to the posterior probability density function, and is calculated as follows:
[0024]
[0025] Where: The predicted mean of the residual values for the points to be corrected.
[0026] Preferably, the current grid precipitation forecast data is corrected based on the predicted average value, and is calculated as follows:
[0027]
[0028] Where: Revised(s,t): revised grid precipitation forecast data; P s,t : Historical grid precipitation forecast data.
[0029] Preferably, after correcting the current gridded forecast data, the method further includes: reorganizing the corrected data into a gridded precipitation data time series.
[0030] Preferably, the pre-generated precipitation observation data set includes:
[0031] Collect observation data from conventional meteorological stations, weather radar data, Fengyun satellite data, and historical grid precipitation forecast data;
[0032] The observation data of the conventional meteorological station, the weather radar data and the Fengyun satellite data are processed into grid data and fused through Bayesian to generate the historical grid precipitation measured values;
[0033] The residual value is calculated based on the historical grid precipitation measured value and the corresponding historical grid precipitation forecast data;
[0034] The historical grid precipitation measured values, the historical grid precipitation forecast data and the residual values generate a precipitation observation data set.
[0035] Preferably, the processing of the observation data of the conventional meteorological station, the weather radar data and the Fengyun satellite data into gridded data and fusing them through Bayesian to generate the historical grid precipitation measured values comprises:
[0036] The observation data of the conventional meteorological station is interpolated into the first gridded data consistent with the preset parameters of the mesoscale meteorological model by using the inverse distance weight;
[0037] The weather radar data and the Fengyun satellite data are converted into the second gridded data consistent with the preset parameters of the mesoscale meteorological model by resampling or interpolation;
[0038] The first gridded data and the second gridded data are fused through Bayesian method to generate the historical gridded precipitation measured values.
[0039] Preferably, the residual value is calculated based on the historical grid precipitation measured value and the corresponding historical grid precipitation forecast data, as shown in the following formula:
[0040] χ hard(s,t) =O s,t -P s,t
[0041] Where: hard(s,t) : residual value; O s,t : Historical grid precipitation measured value; P s,t : Historical grid precipitation forecast data; S: longitude and latitude; t: time.
[0042] Preferably, the pre-established weather forecast system comprises:
[0043] Set the parameterization scheme, nesting scheme and model resolution of the mesoscale meteorological model;
[0044] Based on the observation data collected from conventional meteorological stations, weather radar data, Fengyun satellite data and the ensemble Kalman filter assimilation method, the time dimension is introduced to establish an assimilation module;
[0045] The mesoscale meteorological model is combined with the assimilation module to construct a weather forecast system.
[0046] Based on the same inventive concept, the present invention also provides a gridded precipitation correction system, comprising:
[0047] A collection module, used to collect current grid precipitation forecast data based on a pre-established weather forecast system;
[0048] A correction module, used for correcting the current grid precipitation forecast data according to the residual value in the pre-generated precipitation observation data set;
[0049] The calculation module is used to calculate the residual value in the precipitation observation data set from the historical grid precipitation measured value and the historical grid precipitation forecast data.
[0050] Preferably, the correction module comprises:
[0051] An acquisition submodule uses Bayesian maximum entropy to adjust the residual value to obtain a predicted average value of the residual value of the to-be-corrected point in the grid;
[0052] The correction submodule is used to correct the current grid precipitation forecast data based on the predicted average value.
[0053] Preferably, the acquisition submodule includes:
[0054] A priori unit, used for calculating the priori probability density function of the points to be corrected in the grid by using the maximum entropy principle;
[0055] A posterior unit, used to obtain a posterior probability density function based on the residual value, the current grid precipitation forecast data and the prior probability density function;
[0056] The predicted average value calculation unit is used to calculate the predicted average value of the residual values of the points to be corrected in the grid according to the posterior probability density function.
[0057] Compared with the closest prior art, the technical solution provided by the present invention has the following beneficial effects:
[0058] The technical solution provided by the present invention collects current grid precipitation forecast data based on a pre-established weather forecast system; and corrects the current grid precipitation forecast data according to the residual value in the pre-generated precipitation observation data set; wherein the residual value in the precipitation observation data set is calculated by: historical grid precipitation measured values and historical grid precipitation forecast data, and the gridded precipitation data is corrected grid by grid on a spatial scale using the Bayesian maximum information entropy method, thereby improving the accuracy of precipitation forecast data, solving the problem of real-time correction of precipitation, and improving the accuracy of hydrological forecasts, thereby providing strong data support and scientific basis for hydropower scheduling decisions. BRIEF DESCRIPTION OF THE DRAWINGS
[0059] Figure 1 A general flow chart of a gridded precipitation correction method provided by the present invention;
[0060] Figure 2 Detailed flowchart of the correction method in an embodiment of the present invention. DETAILED DESCRIPTION
[0061] In order to better understand the present invention, the content of the present invention is further described below in conjunction with the accompanying drawings and examples.
[0062] Example 1
[0063] Figure 1 This is a general flow chart of a gridded precipitation correction method, such as Figure 1 As shown, including:
[0064] Step S101, collecting current grid precipitation forecast data based on a pre-established weather forecast system;
[0065] Step S102, correcting the current grid precipitation forecast data according to the residual value in the pre-generated precipitation observation data set;
[0066] Step S103: the residual value in the precipitation observation data set is calculated by: the historical grid precipitation measured value and the historical grid precipitation forecast data.
[0067] The method also includes: after the gridded forecast data are corrected, the corrected data are reconstructed into a gridded precipitation data time series.
[0068] The weather forecast system pre-established in step S101 specifically includes:
[0069] Set the parameterization scheme, nesting scheme and model resolution of the mesoscale meteorological model;
[0070] Based on multi-source meteorological data and ensemble Kalman filter assimilation method, the time dimension is introduced to establish an assimilation module;
[0071] The mesoscale meteorological model is combined with the assimilation module to construct a weather forecast system.
[0072] Among them, the assimilation module includes: introducing the time dimension into the observation data collected from conventional meteorological stations, weather radar data and Fengyun satellite data using the ensemble Kalman filter assimilation method to correct the initial field and boundary conditions of the numerical weather forecast model.
[0073] Step S102 specifically includes:
[0074] Firstly, the residual value is adjusted by using Bayesian maximum entropy to obtain the predicted average value of the residual value of the to-be-corrected point in the grid;
[0075] ① Use the maximum entropy principle to calculate the prior probability density function of the points to be corrected in the grid, as follows:
[0076]
[0077] Where: f G : Prior probability density function; χ pre : random variable value; μ0: constant term of normalized constraints; g: function vector of available core knowledge base; μ: coefficient vector of relative importance of each function.
[0078] ② Based on the residual value, the current grid precipitation forecast data and the prior probability density function, the posterior probability density function is obtained and calculated as follows:
[0079] f K (x k )=A -1 ψ[χ k ,f G(x pre )]=A -1 ∫f G (x k )dx k
[0080] Where: f K : posterior probability density function; χ k : The predicted value of the point to be corrected; A: Normalization parameter.
[0081] ③ Calculate the predicted average value of the residual value of the point to be corrected in the grid according to the posterior probability density function, as follows:
[0082]
[0083] Where: The predicted mean of the residual values for the points to be corrected.
[0084] Finally, the grid precipitation forecast data is corrected based on the predicted average value, calculated as follows:
[0085]
[0086] Where: Revised(s,t): revised grid precipitation forecast data; P s,t : Gridded precipitation forecast data.
[0087] Step S103 specifically includes:
[0088] First, the pre-generated precipitation observation dataset includes:
[0089] Collect observation data from conventional meteorological stations, weather radar data, Fengyun satellite data, and historical grid precipitation forecast data;
[0090] The observation data of conventional meteorological stations, weather radar data and Fengyun satellite data are processed into gridded data and fused through Bayesian to generate historical grid precipitation measured values;
[0091] The residual value is calculated based on the historical grid precipitation measured value and the corresponding historical grid precipitation forecast data;
[0092] The precipitation observation dataset is generated by the historical grid precipitation measured values, historical grid precipitation forecast data and residual values.
[0093] Furthermore, the observation data of the conventional meteorological station, the weather radar data and the Fengyun satellite data are processed into grid data and fused through Bayesian to generate the historical grid precipitation measured values, including:
[0094] The observation data of the conventional meteorological station is interpolated into the first gridded data consistent with the preset parameters of the mesoscale meteorological model by using the inverse distance weight;
[0095] The weather radar data and the Fengyun satellite data are converted into the second gridded data consistent with the preset parameters of the mesoscale meteorological model by resampling or interpolation;
[0096] The first gridded data and the second gridded data are fused through Bayesian method to generate the historical gridded precipitation measured values.
[0097] Among them, the residual value between the historical grid precipitation measured value and the corresponding historical grid precipitation forecast data is calculated as follows;
[0098] χ hard(s,t) =O s,t -P s,t
[0099] Where: hard(s,t) : residual value; O s,t : Historical grid precipitation measured value; P s,t : Historical grid precipitation forecast data; S: longitude and latitude; t: time.
[0100] Example 2
[0101] Figure 2 A detailed flowchart of the correction method is shown in Figure 2 As shown, the correction methods include:
[0102] Step 1: Generate multi-source meteorological datasets
[0103] Obtain conventional weather station, weather radar data and Fengyun satellite data through the meteorological department;
[0104] Based on all the above data sources, the data is formatted to generate data in a unified format, and after quality control, a multi-source meteorological data set is generated.
[0105] Step 2: Establish a weather forecast system based on mesoscale numerical meteorological models
[0106] Based on the multi-source meteorological data set established in the first step, the ensemble Kalman filter data assimilation method is used to establish a weather forecast system in combination with a mesoscale numerical meteorological model. The main steps are as follows:
[0107] (1) Set initial values for the parameterization scheme, nesting scheme, and model resolution of the mesoscale meteorological model;
[0108] (2) Based on the ensemble Kalman filter assimilation method, the time dimension is introduced to establish a precipitation data assimilation system and a rapid update circulation system;
[0109] (3) The mesoscale meteorological model (The Weather Research and Forecasting Model, WRF) is combined with the assimilation module to build a weather forecast system, debug the system, and provide real-time forecast grid precipitation forecast data for the test area.
[0110] Step 3: Gridded spatiotemporal statistics correction
[0111] Based on the real-time grid precipitation forecast data collected in the second step, the Bayesian maximum entropy method is used to perform gridded spatiotemporal correction. The main steps are as follows:
[0112] (1) Basic data preparation: Prepare observation data for at least one year, including precipitation observation data from conventional ground monitoring stations, rain gauges, hydrological stations, radar observation data, satellite-derived precipitation data, etc. At the same time, prepare historical grid precipitation forecast data for the required area.
[0113] (2) Preparation of input data for the revised model mainly includes the following steps:
[0114] ①Use the inverse distance weighted interpolation in ArcGIS software to interpolate the meteorological observation data of ground conventional monitoring stations, rainfall stations, and hydrological stations into gridded data consistent with the grid resolution and number of grid points of the WRF model;
[0115] ②Resample or interpolate satellite inversion data and radar quantitative precipitation estimation data into gridded data consistent with the WRF model grid resolution and number of grid points;
[0116] ③ Perform Bayesian fusion between the two to generate historical grid precipitation measured values;
[0117] ④ Use the meteorological interpretation script to calculate the residual value between the actual historical grid precipitation measured value and the corresponding historical grid precipitation forecast data.
[0118] (3) Statistical modeling is performed using Bayesian maximum entropy (BME). BME refers to all collected data and information as the knowledge base (K). K is divided into two categories based on their properties: core knowledge base (G) and specific knowledge (S). G includes spatiotemporal covariance models, physical laws, and scientific theories, and S includes hard data and soft data.
[0119] In this embodiment, hard data refers to the residual value between the historical grid precipitation measured value and the historical grid precipitation forecast data, and the error is considered to be negligible. Soft data refers to the real-time grid precipitation forecast data, which has certain uncertainty and error.
[0120] The calculation process mainly includes the following steps:
[0121] ① Prior stage: Use the maximum entropy principle to find the prior probability density function f that is closest to the actual situation and contains the maximum amount of information G (core knowledge base) G , the calculation formula is as follows:
[0122]
[0123] Where: f G : Prior probability density function; χ pre : random variable value; μ0: constant term of normalized constraints; g: function vector of available core knowledge base; μ: coefficient vector of relative importance of each function.
[0124] Among them, χ pre , calculated as follows:
[0125] χ pre =[χ data ,x k ]
[0126] Where: data : A collection of hard data and soft data; k : The predicted value of a certain point to be corrected;
[0127] χ data =[χ hard ,x soft ]
[0128] Where: hard : the realized value of hard data at the corresponding position; soft : The realized value of soft data at the corresponding position;
[0129]
[0130] Where: m k Hard data; m : m soft data.
[0131] ②Intermediate stage: Express the real-time grid precipitation forecast data in a reasonable form;
[0132] ③ Posterior stage: In this stage, the real-time grid precipitation forecast data and the prior probability density function are combined to obtain the posterior probability density function f based on K under the generalized Bayesian condition. K , as shown below:
[0133] f K (x k )=A -1 ψ[χ k ,fG (x pre )]=A -1 ∫f G (x k )dx k (2)
[0134] Where: f K : posterior probability density function; χ k : The predicted value of the point to be corrected; A: Normalization parameter.
[0135] According to f K The predicted average value of the residual value of the point to be corrected in the grid can be calculated using the following formula:
[0136]
[0137] Where: The predicted mean of the residual values for the points to be corrected.
[0138] ④ Correction of forecast results: real-time correction of grid precipitation forecast data
[0139] Based on the Bayesian maximum entropy method and combined with the WRF mesoscale numerical model results, the correction is performed. The formula is as follows:
[0140]
[0141] Where: Revised(s,t): revised grid precipitation forecast data; P s,t : Historical grid precipitation forecast data.
[0142] Among them, the residual value between the historical grid precipitation measured value and the corresponding historical grid precipitation forecast data is shown in the following formula:
[0143] χ hard(s,t) =O s,t -P s,t
[0144] Where: hard(s,t) : residual value; O s,t : Historical grid precipitation measured value; P s,t : Historical grid precipitation forecast data; S: longitude and latitude; t: time.
[0145] ⑤ Reorganize the corrected data into a gridded precipitation data time series.
[0146] ⑥ Write automatic operation scripts based on scripting language and meteorological data interpretation language to realize the automatic operation of ①-⑤.
[0147] Based on the same inventive concept, this embodiment also provides a grid precipitation correction system, including:
[0148] A collection module, used to collect current grid precipitation forecast data based on a pre-established weather forecast system;
[0149] A correction module, used for correcting the current grid precipitation forecast data according to the residual value in the pre-generated precipitation observation data set;
[0150] The calculation module is used to calculate the residual value in the precipitation observation data set from the historical grid precipitation measured value and the historical grid precipitation forecast data.
[0151] In the embodiment, the correction module comprises:
[0152] An acquisition submodule uses Bayesian maximum entropy to adjust the residual value to obtain a predicted average value of the residual value of the to-be-corrected point in the grid;
[0153] The correction submodule is used to correct the current grid precipitation forecast data based on the predicted average value.
[0154] In an embodiment, the acquisition submodule includes:
[0155] A priori unit, used for calculating the priori probability density function of the points to be corrected in the grid by using the maximum entropy principle;
[0156] A posterior unit, used to obtain a posterior probability density function based on the residual value, the current grid precipitation forecast data and the prior probability density function;
[0157] The predicted average value calculation unit is used to calculate the predicted average value of the residual values of the points to be corrected in the grid according to the posterior probability density function.
[0158] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.
[0159] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0160] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0161] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0162] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention are included in the scope of the claims of the present invention to be approved.
Claims
1. A gridded precipitation correction method, characterized in that: include: Collect current grid precipitation forecast data based on a pre-established weather forecast system; Correcting the current grid precipitation forecast data according to the residual value in the pre-generated precipitation observation data set; The residual value in the precipitation observation data set is calculated by: the historical grid precipitation measured value and the historical grid precipitation forecast data; The correcting the current grid precipitation forecast data according to the residual value in the precipitation observation data set includes: The residual value is adjusted by using Bayesian maximum entropy to obtain a predicted average value of the residual value of the to-be-corrected point in the grid; Correcting the current grid precipitation forecast data based on the predicted average value; The method of adjusting the residual value by using Bayesian maximum entropy to obtain a predicted average value of the residual value of the to-be-corrected point in the grid includes: Calculating the prior probability density function of the points to be corrected in the grid using the maximum entropy principle; Obtaining a posterior probability density function based on the residual value, the current grid precipitation forecast data and the prior probability density function; Calculate the predicted average value of the residual value of the to-be-corrected point in the grid according to the posterior probability density function; The maximum entropy principle is used to calculate the prior probability density function of the points to be corrected in the grid, and the calculation is as follows: Where: f G : Prior probability density function; χ pre : random variable value; μ0: constant term of normalized constraints; g: function vector of available core knowledge base; μ: coefficient vector of relative importance of each function; Based on the residual value, the current grid precipitation forecast data and the prior probability density function, the posterior probability density function is obtained, which is calculated as follows: f K (x k )=A -1 ψ[x k ,f G (x pre )]=A -1 ∫f G (x k )dx k Where: f K : posterior probability density function; χ k : predicted value of the point to be corrected; A: normalized parameter; The predicted average value of the residual value of the to-be-corrected point in the grid is calculated according to the posterior probability density function, and is calculated as follows: Where: The predicted average value of the residual value of the point to be corrected; The current grid precipitation forecast data is corrected based on the predicted average value, and is calculated as follows: Where: Revised(s,t): revised grid precipitation forecast data; P s,t : Historical grid precipitation forecast data.
2. The correction method according to claim 1, characterized in that: After the current gridded forecast data is corrected, the method further includes: reorganizing the corrected data into a gridded precipitation data time series.
3. The correction method according to claim 1, characterized in that: The pre-generated precipitation observation dataset includes: Collect observation data from conventional meteorological stations, weather radar data, Fengyun satellite data, and historical grid precipitation forecast data; The observation data of the conventional meteorological station, the weather radar data and the Fengyun satellite data are processed into grid data and fused through Bayesian to generate the historical grid precipitation measured values; The residual value is calculated based on the historical grid precipitation measured value and the corresponding historical grid precipitation forecast data; The historical grid precipitation measured values, the historical grid precipitation forecast data and the residual values generate a precipitation observation data set.
4. The correction method according to claim 3, characterized in that: The process of processing the observation data of the conventional meteorological station, the weather radar data and the Fengyun satellite data into grid data and fusing them through Bayesian to generate the historical grid precipitation measured values includes: The observation data of the conventional meteorological station is interpolated into the first gridded data consistent with the preset parameters of the mesoscale meteorological model by using the inverse distance weight; The weather radar data and the Fengyun satellite data are converted into the second gridded data consistent with the preset parameters of the mesoscale meteorological model by resampling or interpolation; The first gridded data and the second gridded data are fused through Bayesian method to generate the historical gridded precipitation measured values.
5. The correction method according to claim 3, characterized in that: The residual value is calculated based on the historical grid precipitation measured value and the corresponding historical grid precipitation forecast data, as shown in the following formula: x hard(s,t) =O s,t -P s,t Where: hard(s,t) : residual value; O s,t : Historical grid precipitation measured value; P s,t : Historical grid precipitation forecast data; S: longitude and latitude; t: time.
6. The correction method according to claim 3, characterized in that: The pre-established weather forecast system comprises: Set the parameterization scheme, nesting scheme and model resolution of the mesoscale meteorological model; Based on the observation data collected from conventional meteorological stations, weather radar data, Fengyun satellite data and the ensemble Kalman filter assimilation method, the time dimension is introduced to establish an assimilation module; The mesoscale meteorological model is combined with the assimilation module to construct a weather forecast system.
7. A gridded precipitation correction system, characterized in that: include: A collection module, used to collect current grid precipitation forecast data based on a pre-established weather forecast system; A correction module, used for correcting the current grid precipitation forecast data according to the residual value in the pre-generated precipitation observation data set; A calculation module, used for calculating the residual value in the precipitation observation data set from the historical grid precipitation measured value and the historical grid precipitation forecast data; The correction module comprises: An acquisition submodule uses Bayesian maximum entropy to adjust the residual value to obtain a predicted average value of the residual value of the to-be-corrected point in the grid; A correction submodule, used for correcting the current grid precipitation forecast data based on the predicted average value; The acquisition submodule includes: A priori unit, used for calculating the priori probability density function of the points to be corrected in the grid by using the maximum entropy principle; A posterior unit, used to obtain a posterior probability density function based on the residual value, the current grid precipitation forecast data and the prior probability density function; A prediction average value calculation unit is used to calculate the prediction average value of the residual value of the to-be-corrected point in the grid according to the posterior probability density function; The maximum entropy principle is used to calculate the prior probability density function of the points to be corrected in the grid, and the calculation is as follows: Where: f G : Prior probability density function; χ pre : random variable value; μ0: constant term of normalized constraints; g: function vector of available core knowledge base; μ: coefficient vector of relative importance of each function; Based on the residual value, the current grid precipitation forecast data and the prior probability density function, the posterior probability density function is obtained, which is calculated as follows: f K (x k )=A -1 ψ[x k ,f G (x pre )]=A -1 ∫f G (x k )dx k Where: f K : posterior probability density function; χ k : predicted value of the point to be corrected; A: normalized parameter; The predicted average value of the residual value of the to-be-corrected point in the grid is calculated according to the posterior probability density function, and is calculated as follows: Where: The predicted average value of the residual value of the point to be corrected; The current grid precipitation forecast data is corrected based on the predicted average value, and is calculated as follows: Where: Revised(s,t): revised grid precipitation forecast data; P s,t : Historical grid precipitation forecast data.