A method and system for comprehensively evaluating the regional representativeness of precipitation data of a meteorological observation station based on multiple precipitation characteristic quantities

Abstract

This invention discloses a method and system for comprehensively evaluating the regional representativeness of precipitation data from meteorological observation stations based on multiple precipitation characteristics, relating to the application and evaluation of meteorological observation data. By selecting monthly total precipitation and monthly maximum daily precipitation as key precipitation characteristics, corresponding sample sets are constructed. Relationship models between the observation structure functions of monthly total precipitation and monthly maximum daily precipitation and the distance between stations are established. The random mean square error of observations is estimated based on the zero-distance intercept. The mean square deviation caused by the observation station's precipitation data replacing the target area's data is calculated, and a normalized deviation index is constructed. Representativeness levels and spatial applicability are then classified. This achieves a comprehensive quantitative evaluation of the regional representativeness error and spatial representativeness of precipitation data, improving the objectivity and applicability of regional precipitation data applications. It can provide a basis for selecting reference meteorological stations for precipitation data in areas with few or no stations, regional climate analysis, station network layout optimization, and related engineering applications.

Description

Technical Field

[0001] This invention belongs to the field of meteorological observation data application and evaluation, and involves spatial representativeness assessment and error quantification analysis technology of precipitation observation data. Specifically, it is a method and system for comprehensively assessing the regional representativeness of precipitation data from meteorological observation stations based on multiple precipitation characteristic quantities. Background Technology

[0002] Representativeness assessment of precipitation data is a crucial step in ensuring that meteorological data can effectively support applications such as engineering design, climate analysis, and disaster assessment. The core issue is determining whether the precipitation data from the observation station is consistent with or highly similar to the precipitation characteristics of the target area. Existing representativeness assessment techniques typically fall into two categories: qualitative assessment and quantitative analysis. Qualitative assessment relies primarily on factors such as spatial distance, climate environment, and underlying topography. However, this method lacks quantitative standards and depends heavily on subjective judgment based on expert experience, making it difficult to guarantee the objectivity and reliability of the results. Quantitative analysis involves establishing a dedicated precipitation observation station in the target area for 1-3 years of short-term precipitation observation, followed by correlation and error analysis with a representative meteorological station to determine the appropriateness of precipitation representativeness. However, this method requires at least one year of precipitation observation data for the target area and does not consider potential errors caused by the surrounding environment of newly established stations.

[0003] For example, Chinese patent CN121303674A discloses a representative assessment method for regional wind speed meteorological reference stations based on a correlation function field. It estimates substitution errors by constructing a regression model of the distance between stations and the correlation coefficient. However, this method is based on the assumption of smooth evolution of a continuous field, making it difficult to effectively remove non-stationary random noise caused by local convection from precipitation elements, resulting in insufficient accuracy in characterizing the discrete features of precipitation. CN110515139B discloses a multi-scale topographic representative quantitative analysis method for meteorological and hydrological stations. It focuses on the analysis of station site topographic matching, using Thiessen polygons and topographic complexity difference coefficients as its core, but it lacks methods for separating observation errors, calculating data substitution bias, and establishing distance classification criteria.

[0004] Meanwhile, precipitation itself exhibits significant spatial heterogeneity and local sensitivity. Its statistical distribution is not only controlled by large-scale weather patterns but also influenced by local topography, underlying surface roughness, river and lake distribution, urban-rural thermal differences, and the randomness of convective activity. Especially in severe convective precipitation, short-duration rainstorms, and local extreme precipitation events, even two closely located observation points may show significant differences. Therefore, relying solely on whether the trends of data from two locations are consistent, or solely on the distance between stations, is insufficient to fully explain the magnitude of the substitution error of precipitation data from a particular observation station for a target point. Furthermore, existing assessment methods typically fail to uniformly characterize the inherent errors in the observation data, interference from the surrounding environment, and additional biases caused by regional microclimate differences. They also lack a hierarchical threshold system for the spatial representativeness of precipitation data. For the large amounts of regional precipitation data generated under high-density observation networks such as automatic weather stations, there remains a lack of mature, unified, and widely applicable technical pathways for objectively determining the representativeness of a particular observation station at different distances without requiring new long-term comparative observations, and for further providing error criteria that can be used for engineering design and disaster assessment.

[0005] In summary, existing methods for assessing the representativeness of precipitation data generally suffer from insufficient quantification, long assessment cycles, limited applicability to target areas without stations, and difficulty in simultaneously reflecting spatial differences and the magnitude of data substitution errors. Therefore, establishing an assessment method that can be applied to regional precipitation data scenarios, quantitatively characterize the representativeness error of station precipitation data to target areas, and determine its spatial representativeness is an urgent technical problem to be solved. Summary of the Invention

[0006] (a) Purpose of the invention To address the aforementioned deficiencies and shortcomings of existing technologies, this invention aims to provide a method and system for comprehensively evaluating the regional representativeness of precipitation data from meteorological observation stations based on multiple precipitation characteristics. By simultaneously selecting monthly total precipitation and monthly maximum daily precipitation as key precipitation characteristics for regional representativeness evaluation, corresponding sample sets, observation structure functions, and their quantitative relationships with inter-station distances are constructed. Furthermore, the random mean square error of observations is estimated, and the mean square deviation caused by the substitution of target area data with observation station data is calculated. Under the synergistic constraints of multiple precipitation characteristics, representativeness levels and spatial applicability ranges are defined, thereby achieving a comprehensive quantitative evaluation of the regional representativeness of precipitation data from meteorological observation stations. This provides a basis for screening precipitation data from reference meteorological stations in areas with few or no stations.

[0007] (II) Technical Solution To achieve the objective of this invention and solve its technical problems, the present invention adopts the following technical solution: The first objective of this invention is to provide a method for comprehensively evaluating the regional representativeness of precipitation data from meteorological observation stations based on multiple precipitation characteristics. The method, when implemented, includes at least the following steps: SS1. Data Acquisition: Collect geographical environment information and multi-year daily precipitation observation data of meteorological observation stations in the assessment area, conduct completeness statistics and correlation tests on precipitation observation data, and establish a sample database containing valid station information and daily precipitation data; SS2. Construction of precipitation sample set: Based on the sample database, the total monthly precipitation and the maximum daily precipitation of each station are statistically analyzed. Valid precipitation data of the same month in each year are selected to form a precipitation element sample set corresponding to the total monthly precipitation and the maximum daily precipitation of each month. SS3. Integrated Modeling of Multiple Precipitation Characteristics: Based on the precipitation sample set, the inter-station distances between all stations are statistically analyzed. d and the corresponding monthly total precipitation observation structure function b rt ′( d ), monthly maximum daily precipitation observation structure function b rm ′( d ), fitting b rt ′( d )and d The first relational model and b rm ′( d )and d The second relation model, and based on the two relation models in d Intercept at =0 a 0、 b 0, estimate the random mean square error of observations for total monthly precipitation and maximum daily precipitation, respectively. s rt 2 = a 0 / 2、 s rm 2 = b 0 / 2; SS4. Mean Square Deviation Calculation and Normalized Deviation Index Construction: Based on the first and second relational models, the representative distance of precipitation observation data for any meteorological observation station is calculated. d The root mean square bias of total monthly precipitation and maximum daily monthly precipitation resulting from target point precipitation data. E rt ( d )= b rt ′( d )- a 0 / 2 andE rm ( d )= b rm ′( d )- b 0 / 2, and based on this, normalized deviation indices for total monthly precipitation and maximum daily monthly precipitation are constructed respectively. I rt ( d )= E rt ( d ) / s rt 2 and I rm ( d )= E rm ( d ) / s rm 2 ; SS5. Representativeness Level Threshold Classification: Based on the normalized deviation index of total monthly precipitation and maximum monthly daily precipitation. I rt ( d ), I rm ( d The deviation index was normalized, and a normalized deviation index ≤2.0 was determined as the upper limit of the representative excellent level, and >3.0 was determined as the lower limit of the representative poor level. An intermediate level was divided between the excellent level and the poor level threshold standard. SS6. Comprehensive Representativeness Assessment: Substitute the mean square deviation thresholds corresponding to each representativeness level into the first and second relationship models to determine the inter-station distance ranges corresponding to the monthly total precipitation and the monthly maximum daily precipitation, respectively. The inter-station distance ranges that simultaneously meet the same representativeness level requirements are determined as the comprehensive assessment results of the regional representativeness of precipitation data from meteorological observation stations.

[0008] The second objective of this invention is to provide an evaluation system for the regional representativeness of precipitation data from meteorological observation stations, which is used to implement the above-mentioned method of comprehensively evaluating the regional representativeness of precipitation data from meteorological observation stations based on multiple precipitation characteristics.

[0009] (III) Technical Effects Compared with existing technologies, the method and system for comprehensively evaluating the regional representativeness of precipitation data from meteorological observation stations using multiple precipitation characteristics provided by this invention have the following beneficial and significant technical effects: (1) This invention uses the total monthly precipitation and the maximum monthly daily precipitation as the basis for the collaborative evaluation of multiple precipitation characteristics. It establishes a precipitation sample set, observation structure function and quantitative relationship between them and the distance between stations. It also achieves a unified quantitative characterization of representative error through observation random mean square error, mean square deviation and normalized deviation index. Compared with existing methods that rely only on spatial distance, underlying surface conditions or correlation analysis, it can simultaneously reflect the average precipitation characteristics and heavy precipitation characteristics, and improve the objectivity, completeness and quantification of the regional representativeness evaluation of precipitation data.

[0010] (2) Based on multiple precipitation characteristics, the present invention divides the representative level and reverses the corresponding spatial distance range. Then, the two types of precipitation characteristics are simultaneously satisfied at the same level as the comprehensive criterion to form the comprehensive precipitation representative level spatial distance. This can directly give the reference meteorological station data substitution error range and spatial applicable boundary of the target area with no station or few stations. It can provide more reliable basic data support for the selection of precipitation data in engineering design, regional climate analysis, extreme precipitation risk analysis and meteorological disaster impact assessment. Attached Figure Description

[0011] Figure 1 The diagram shows the implementation process of the method for comprehensively evaluating the regional representativeness of precipitation data from meteorological observation stations based on multiple precipitation characteristics, according to the present invention. Figure 2 The diagram shows the relationship between the observation structure function of monthly total precipitation in the Jianghan Plain and the distance between stations, calculated using the method of this invention. Figure 3 The diagram shows the relationship between the observation structure function of the maximum monthly daily precipitation in the Jianghan Plain and the distance between stations, calculated using the method of this invention. Figure 4 The figure shows the relationship between the moving average of the monthly total precipitation observation structure function and the distance between stations, as well as a polynomial fitting diagram, calculated using the method of this invention. Figure 5 The figure shows the relationship between the moving average of the monthly maximum daily precipitation observation structure function and the distance between stations, as well as a polynomial fitting diagram, calculated using the method of this invention. Figure 6 The figure shows the relationship between the mean square deviation of the total monthly precipitation in the Jianghan Plain and the distance between stations, as well as the distance between representative precipitation levels, calculated using the method of this invention. Figure 7 The diagram shows the relationship between the mean square deviation of the maximum daily precipitation in the Jianghan Plain and distance, as well as the distance to the representative precipitation level, calculated using the method of this invention. Detailed Implementation

[0012] This invention aims to provide a method and system for comprehensively evaluating the regional representativeness of precipitation data from meteorological observation stations based on multiple precipitation characteristics. To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions in the embodiments of this invention will be described in more detail below with reference to the accompanying drawings. The following embodiments are only some implementations of this invention and are used for illustrative purposes, not to limit the scope of protection of this invention. Based on the disclosure of this invention, equivalent substitutions or conventional modifications made by those skilled in the art without departing from the technical concept of this invention should fall within the protection scope of this invention.

[0013] Example 1: A method for assessing the regional representativeness of precipitation data from meteorological observation stations like Figure 1 As shown in this embodiment, the method for comprehensively evaluating the regional representativeness of precipitation data from meteorological observation stations based on multiple precipitation characteristics achieves the classification of regional representativeness and the inference of spatial applicability range of precipitation data by jointly sampling the monthly total precipitation and the monthly maximum daily precipitation, performing structure function modeling, error decomposition, classification, and comprehensive distance back-calculation. With the above technical solution, this invention can achieve quantitative and hierarchical evaluation of the representativeness of regional precipitation data, providing unified technical support for the selection of engineering design parameters, meteorological disaster impact analysis, and regional precipitation data screening. Specifically, the method mainly includes the following steps in its implementation: SS1. Data Acquisition: Collect geographical environment information and multi-year daily precipitation observation data from various meteorological observation stations within the assessment area; conduct completeness statistics and correlation tests on the precipitation observation data; and establish a sample database containing valid station information and daily precipitation data. Preferably, step SS1 should include at least the following: S11. Site and Data Collection: Collect geographical environment information and multi-year daily precipitation observation data for each site in the assessment area. The geographical environment information shall include at least the longitude, latitude and altitude of the site. The multi-year daily precipitation observation data shall include at least 6 years or more of continuous observation data. The spatial distribution density of sites in the assessment area shall meet the requirement that the minimum distance between adjacent sites does not exceed 15km. S12. Data completeness screening: Statistically evaluate the completeness of daily precipitation data collected from each station, and remove stations with an annual precipitation data missing rate exceeding 25% or a summer precipitation data missing rate exceeding 10%. S13. Correlation test: Calculate the correlation coefficient between each station and the precipitation data of stations within a 30km radius. Remove stations with a correlation coefficient of less than 0.1 with more than half of the stations within a 30km radius. Based on the effective stations that pass the completeness test and the spatial correlation test, establish a sample database containing station geographical environment information and daily precipitation data.

[0014] SS2. Construction of Precipitation Sample Set: Based on the sample database, the total monthly precipitation and the maximum daily precipitation for each station are statistically analyzed. Valid precipitation data for the same month in each year are selected to form a precipitation element sample set corresponding to the total monthly precipitation and the maximum daily precipitation for each month. Preferably, step SS2 includes at least the following sub-steps: S21. Monthly statistics generation: Based on the daily precipitation sample database established in step SS1, the total precipitation and maximum daily precipitation of each effective station in each month are statistically analyzed to form candidate monthly samples organized by station, year and month. S22. Determination of missing data for a month: If precipitation data for one day is missing in a certain month, then the total monthly precipitation and the maximum daily precipitation for that month will be marked as missing. S23. Annual Sample Validity Screening: Precipitation data for May to September each year must not be missing for more than one month, and precipitation data for other months must not be missing for more than three months. The final number of valid data samples selected from each station must be 36 or more.

[0015] It should be noted that step SS2 selects two types of precipitation characteristics, namely total monthly precipitation and maximum monthly daily precipitation, and applies unified constraints on the validity of samples from months with missing data and years. This ensures that the constructed sample set can both represent the regional average precipitation level and retain information on heavy precipitation events, thus providing a stable and comparable input basis for subsequent structure function modeling, error estimation, and comprehensive discrimination.

[0016] SS3. Integrated modeling of multiple precipitation characteristics: Based on the precipitation sample set, the inter-station distances between all stations were calculated. d and the corresponding monthly total precipitation observation structure function b rt ′( d ), monthly maximum daily precipitation observation structure function b rm ′( d ), fitting b rt ′( d )and d The first relational model and b rm ′( d )and d The second relation model, and based on the two relation models in d Intercept at =0 a 0、 b 0, estimate the random mean square error of observations for total monthly precipitation and maximum daily precipitation, respectively. s rt2 = a 0 / 2、 s rm 2 = b 0 / 2.

[0017] As a preferred option, the integrated modeling of multiple precipitation characteristic observation structure functions should include at least the following: S31. Calculation of inter-station distances: Based on the precipitation sample set formed in step SS2, statistically evaluate the inter-station distances between all valid stations in the evaluation area. d ; S32. Calculation of Observational Structure Function: For both the monthly total precipitation sample and the monthly maximum daily precipitation sample, calculate the observational structure function between all valid stations. b r '( d For any two stations A and B, b r '( A , B The mean square of the time series of the difference between the observed precipitation values ​​at two stations is defined as the average of the differences between the two station observations. S33. Relationship Field Construction: The calculated inter-station distances and corresponding observation structure functions are paired and combined to form a relationship field between inter-station distances and corresponding structure functions. The number of data sets is... N = k ×( k -1) / 2, k To assess the number of valid sites within the region.

[0018] Furthermore, the construction of the first relation model and the second relation model includes at least the following: S34. Moving average processing: for N The group of sites performed a 3km interval moving average on the relational field data, and then processed the data within 1km ≤ d <3km b r '( d The average value is then used as the observation structure function for a distance of 2km. (2km≤) d <4km b r '( d The average value is used as the observation structure function corresponding to a distance of 3km. This process is repeated until the average observation structure function corresponding to the maximum distance in the region is calculated. S35. Polynomial Regression Modeling: Based on the data processed by moving average, fit the monthly total precipitation observation structure function with respect to inter-station distance. d Multinomial regression model b rt′( d )= a 0+ a 1 d + a 2 d 2 +…+ a m d m A first relational model was formed, and the structure function of monthly maximum daily precipitation observations was fitted with respect to distance. d Multinomial regression model b rm ′( d )= b 0+ b 1 d + b 2 d 2 +…+ b n d n This forms a second relational model, in which m , n For the order of the polynomial, a 0、 a 1. a 2. ... a m as well as b 0、 b 1. b 2. ... b n These are the polynomial coefficients; S36. Determining the optimal order: Compare the sum of squared errors between the fitting results of different orders and the actual observed structure function values, and select the order and coefficients corresponding to the smallest sum of squared errors as the optimal fitting result.

[0019] In addition, the accuracy verification of the two constructed polynomial regression models includes at least the following: based on the optimal polynomial regression model of the observed structure function of total monthly precipitation and maximum daily monthly precipitation determined in step S36, calculating the fitted values ​​corresponding to the distances between each station, and comparing them distance-by-distance with the actual observed structure function values ​​processed by the moving average in step S34, and calculating the determination coefficients of the two models respectively. R 2 When the coefficient of determination of two models R 2When both reach a preset goodness-of-fit threshold close to 1.0, it is determined that the first and second relationship models can accurately characterize the quantitative relationship between the observed structure function and the inter-station distance; when the fitting accuracy does not meet the preset requirements, return to steps S35 and S36, readjust the polynomial order and perform regression fitting and optimal order determination until the preset goodness-of-fit threshold requirements are met.

[0020] To illustrate the relationship between the precipitation observation structure function and the observation random mean square error, and to derive the method for estimating the observation random mean square error using the zero-distance intercept, this embodiment further explains the construction process of the relationship between inter-station distance and the structure function based on structure function theory and observation error decomposition, as follows: First, based on the daily precipitation data preprocessed in step SS1, and combined with the precipitation samples selected in step SS2 within the assessment area, the distances and structure functions between all valid stations within the assessment area are statistically analyzed to form a matrix field of distances and corresponding structure functions between stations. For any two stations within the area... A and B Let the distance between the two stations be... d Then the theoretical structure function of the precipitation layer between the two stations b r ( A , B ) is defined as: (1) in: (2) (3) In the formula: f ( A )and f ( B ) represent the stations respectively A and sites B The deviation of precipitation values ​​is shown by the overline, which represents the average of the time series. In this embodiment, F ( A )and F ( B (Corresponds to the site) A and B Precipitation sample values ​​under the same month and the same precipitation characteristic caliber. The calculated spatial distances between each pair of stations within the region and their corresponding theoretical structure functions are paired and combined to form a relationship field between the distances between stations and their corresponding structure functions.

[0021] Secondly, for any precipitation observation value within the assessment area, its observation bias typically includes two components: one is the true value of the precipitation element. F deviation f (A The other is observation error. d r ( A ).in, d r ( A ) dimensions and f ( A Consistency, also known as observation random error. According to d r ( A The randomness of ) has A , B The random errors of observations at different points are uncorrelated, and the random error of observation at a certain point is also uncorrelated with the deviations at that point or other points. Therefore: (4) (5) (6) When two points coincide, i.e., A=B, we have: (7) in, d r ( A )and d r ( B (These are the sites) A , B Random observation error; s r 2 ( A ), s r 2 ( B (These are the sites) A , B The mean square value of the observation random error is also known as the observation random mean square error. s r It represents the random standard error of the observation.

[0022] Since actual observation data contains both true values ​​and observation errors, the observation structure function calculated from the observed precipitation data according to equation (1) is: (8) Right now: (9) Expanding equation (9) and combining the results from equations (4) to (7), we can obtain: (10) Substituting equation (1) into equation (10), we get: (11) Because the climate of the assessment area is similar, its structure function satisfies homogeneity and isotropy; the structure function is only a function of distance. It also contains the mean square value of the observational random error at any point. s r 2 They are all equal, when the site A , B The distance is d Then, equation (11) above can be further simplified to obtain the observation structure function. b r ′( d ) and theoretical truth structure function b r ( d The analytical relationship between them: (12) Equation (12) shows that, due to the influence of observational random errors in the observational data, the observation structure function at the same distance... b r ′( d (Greater than the theoretical truth value structure function) b r ( d ).when d When =0, we have b r ( d Substituting )=0 into equation (12) yields: (13) In the formula: b r ′(0) represents the value of the structure function versus distance curve calculated from the observation data extrapolated to zero distance. From this, the random mean square error of the observations can be calculated using the relationship between the structure function and distance. s r 2 .

[0023] It should be noted that the technical essence of step SS3 lies not only in spatial fitting of precipitation data, but also in constructing quantitative relationships between observation structure functions and inter-station distances for both monthly total precipitation and monthly maximum daily precipitation. Furthermore, it extracts the observational random mean square error using the zero-distance intercept, thus separating observational error from regional spatial differences. This transforms the representativeness of the observation station for the target area into a calculable and comparable error metric, while simultaneously considering both average precipitation characteristics and heavy precipitation characteristics. This provides a unified theoretical foundation for subsequent mean square deviation calculation, classification, and comprehensive distance inversion.

[0024] SS4. Mean Square Deviation Calculation and Normalized Deviation Index Construction: Based on the first and second relational models, the representative distance of precipitation observation data for any meteorological observation station is calculated as follows: d The root mean square bias of total monthly precipitation and maximum daily monthly precipitation resulting from target point precipitation data. E rt ( d )= b rt ′( d )- a 0 / 2 and E rm ( d )= b rm ′( d )- b 0 / 2, and based on this, normalized deviation indices for total monthly precipitation and maximum daily monthly precipitation are constructed respectively. I rt ( d )= E rt ( d ) / s rt 2 and I rm ( d )= E rm ( d ) / s rm 2 .

[0025] As a preferred example, steps SS3 and SS4, including the estimation of the random mean square error and the calculation of the mean square deviation, at least include the following sub-steps: S41. Error Decomposition: Decompose the observed precipitation deviation at any station into the sum of the true deviation and the observed error, and establish the observation structure function under the condition that the evaluation area satisfies homogeneity and isotropy. b r ′( d ) and truth structure functions b r ( d Analytical relationships between ) b r ′( d )= b r ( d )+2 s r 2 ; S42. Calculation of random mean square error: in spatial distanced When = 0, use the truth structure function b r ( d Given the condition )=0, we obtain that the observed random mean square error satisfies s r 2 = b r ′(0) / 2, and based on this, extrapolate the values ​​to the zero distance using the first and second relationship models, and calculate the observed random mean square error corresponding to the total monthly precipitation and the maximum monthly daily precipitation. s rt 2 = a 0 / 2、 s rm 2 = b 0 / 2; S43. Mean Square Deviation Calculation and Index Normalization: Mean Square Deviation Based on Station Data Substituting for Target Point Data E r = b r ′( d )- s r 2 The mean square deviations of the total monthly precipitation and the maximum daily precipitation were calculated respectively. E rt ( d )= b rt ′( d )- a 0 / 2 and E rm ( d )= b rm ′( d )- b 0 / 2, and based on this, normalized deviation indices for total monthly precipitation and maximum daily monthly precipitation are constructed respectively. I rt ( d )= E rt ( d ) / s rt 2 and I rm ( d )= E rm ( d ) / s rm 2 .

[0026] It should be noted that the technical essence of step SS4 lies in: based on the quantitative relationship between the observation structure function and inter-station distance established in step SS3, further expliciting the mean square bias caused by the substitution of target area data with observation station data, and constructing normalized bias indices for monthly total precipitation and monthly maximum daily precipitation respectively. By normalizing the substitution error and the corresponding observation random mean square error, the differences in dimensions and numerical scales of different precipitation characteristics can be eliminated, allowing the two types of precipitation characteristics to be compared horizontally under a unified evaluation benchmark. This provides a consistent, stable, and feasible measurement basis for subsequent representativeness level classification and comprehensive distance determination.

[0027] SS5. Representativeness Level Threshold Classification: Based on the normalized bias index of total monthly precipitation and maximum daily monthly precipitation I rt ( d ), I rm ( d The threshold standards for the regional representativeness of precipitation data from meteorological observation stations were determined respectively. The normalized deviation index ≤2.0 was determined as the upper limit of the representativeness level, and >3.0 was determined as the lower limit of the representativeness level. An intermediate level was divided between the threshold standards for the representativeness level and the poor level.

[0028] As a preferred option, the representative level threshold classification should at least include: a normalized deviation index based on total monthly precipitation and maximum daily monthly precipitation. I rt ( d ), I rm ( d Based on the condition that the deviation caused by distance variation is no greater than the random mean square error of observation, the normalized deviation index ≤2.0 is determined as the threshold standard for the representative excellent level, and >3.0 is determined as the threshold standard for the representative poor level. Between the representative excellent level and the representative poor level, the normalized deviation index >2.0 and ≤2.5 is determined as the good level, and >2.5 and ≤3.0 is the medium level. The four levels are used to determine the representative level of monthly total precipitation and monthly maximum daily precipitation.

[0029] It should be noted that step SS5 uses the normalized deviation index as a unified criterion to establish a quantitative correspondence between the spatial difference deviation caused by the substitution of target area data with observational data and the random mean square error of observation. Specifically, a normalized deviation index of no more than 2.0 is set as the upper limit for the excellent grade, essentially corresponding to the additional deviation caused by distance changes not exceeding the random mean square error of observation; an index greater than 3.0 is set as the lower limit for the poor grade, indicating that the data substitution error has significantly exceeded the acceptable range; and further subdividing the range from 2.0 to 3.0 into good and medium grades at 0.5 intervals improves the resolution of representative grading and engineering applicability while maintaining clear statistical meaning.

[0030] SS6. Representativeness Assessment: Substituting the mean square deviation thresholds corresponding to each representative level into the first and second relationship models, the inter-station distance ranges corresponding to the monthly total precipitation and the monthly maximum daily precipitation are determined respectively. The inter-station distance ranges that simultaneously meet the same representative level requirements are determined as the comprehensive evaluation results of the regional representativeness of precipitation data from meteorological observation stations.

[0031] As a preferred option, the reverse calculation of the upper limit of the applicable spatial range for each representative level should include at least the following: S61. Single Feature Distance Inversion: For any representative level threshold l According to E rt ( d )= l · s rt 2 and E rm ( d )= l · s rm 2 Determine the mean square deviations corresponding to the total monthly precipitation and the maximum daily precipitation, and then respectively based on E rt ( d )= b rt ′( d )- a 0 / 2 and E rm ( d )= b rm ′( d )- b By combining 0 / 2 with the first and second relationship models, we can deduce the spatial distances corresponding to the representative levels of total monthly precipitation and maximum daily monthly precipitation. d rt , drm ; S62. Comprehensive Judgment Based on Multiple Feature Quantities: The spatial distance range that simultaneously meets the same representativeness level requirements for monthly total precipitation and monthly maximum daily precipitation is determined as the comprehensive precipitation representativeness level distance. d r and with d r This serves as the distance basis for selecting alternative precipitation observation station data when there are no target observation points. Specifically, it includes the spatial distance corresponding to the representative level of the total monthly precipitation. d rt Spatial distance corresponding to the representative level of the maximum daily precipitation in a month d rm When they are not equal, d rt and d rm The smaller of the values ​​is determined as the representative level of the comprehensive precipitation at the corresponding representative level. d r This ensures that the comprehensive spatial distance range simultaneously meets the representativeness requirements of both the total monthly precipitation and the maximum monthly daily precipitation at the same representative level.

[0032] It should be noted that step SS6 is not a simple parallel comparison of the representative distances corresponding to the total monthly precipitation and the maximum monthly daily precipitation. Instead, it employs a comprehensive discrimination criterion that simultaneously meets the requirements of the same level to ensure that the observation station data meets the corresponding level requirements in both the average precipitation characteristics and the heavy precipitation characteristics dimensions. When the representative level distances derived from the two types of precipitation characteristics are inconsistent, the smaller one is taken as the comprehensive precipitation representative level distance. The technical basis for this is that the precipitation characteristics corresponding to the smaller distance impose stricter constraints on data substitution errors and can serve as the control boundary for the comprehensive spatial applicability range. If the larger one is taken, it may only meet the level requirements of one type of precipitation characteristic, while causing the substitution error of the other type of precipitation characteristic to exceed the allowable range. Therefore, taking the smaller one ensures that the finally determined comprehensive distance range simultaneously meets the representative requirements of the total monthly precipitation and the maximum monthly daily precipitation at the same level, thereby improving the engineering applicability of the comprehensive evaluation results.

[0033] Example 2: Case Verification Based on the technical solution of Embodiment 1 above, in order to better understand the present invention, the following example, Example 2, further clarifies the content of the present invention by evaluating the representativeness of precipitation data in the Jianghan Plain of Hubei Province (between 29°26′ and 31°37′ north latitude and 111°14′ and 114°36′ east longitude). This example is carried out sequentially according to steps SS1 to SS6 of Embodiment 1, and each step corresponds completely to the method framework of Embodiment 1; the general method principles already described in detail in Embodiment 1 will not be repeated, but the specific data parameters and calculation results for this region will be emphasized. Specifically, when implementing the method of the present invention, it includes: S001. Data Collection and Preprocessing: We collected data from 539 meteorological stations with an altitude below 200m in the Jianghan Plain of Hubei Province, including 34 national meteorological stations and 505 high-density provincial meteorological stations belonging to the same climate zone and possessing daily precipitation observation data. The minimum distance between these stations was within 10km. The collected geographical information included the longitude, latitude, and altitude of the stations, as well as the required daily precipitation observation data for the eight years from 2018 to 2025. The data completeness rate was assessed by excluding 2022 and 2023, years in which half of the provincial stations had an annual precipitation data measurement rate exceeding 25% or a summer precipitation data measurement rate exceeding 10%. Meanwhile, through correlation analysis, stations with a correlation coefficient of less than 0.1 with more than half of the stations within a 30km radius, as well as stations with an altitude exceeding 200m, were removed. After the above processing, a total of 104 provincial stations were removed, resulting in geographical environmental information for 435 stations (including 34 national stations and 401 provincial stations), as well as daily precipitation data for 6 years from 2018 to 2021, 2024, and 2025, forming a precipitation sample database.

[0034] S002. Construction of Precipitation Sample Set: Based on precipitation sample data, precipitation data from May to September of each year must be at least one month of data missing, and precipitation data from other months must be at least three months of data missing. If the validity of the annual precipitation sample does not meet the above conditions, the data for that year will not be selected. The number of valid precipitation samples (i.e., non-missing data) selected must be at least 36 sets. After removing 15 provincial stations that did not meet the conditions, a monthly precipitation element sample set of 424 stations (including 34 national stations and 390 provincial stations) was finally obtained, including station code, longitude, latitude, altitude, sample time identifier, total monthly precipitation, and maximum monthly daily precipitation.

[0035] S003. Integrated modeling of multiple precipitation characteristics: Based on the selected monthly precipitation element sample set of 424 valid stations, the pairwise distances and corresponding structure functions of all valid stations were statistically analyzed. Since there are 424 valid stations, two sets of... N=424×423 / 2=89676 sets of data. Scatter plots of the structure function for the total monthly precipitation and the maximum daily monthly precipitation at different distances are shown below. Figure 2 and Figure 3 (Blue dot). Figure 2 This mainly reflects the overall increase in the monthly total precipitation observation structure function as the distance between stations increases. Figure 3 It mainly reflects the discrete distribution characteristics of the monthly maximum daily precipitation observation structure function as the distance between stations increases.

[0036] Based on the relational field data, the method described in step S34 of Example 1 is continued to perform a 3km distance moving average processing to reduce the fluctuation of the structure function caused by sampling factors other than distance. Specifically, the structure function values ​​for distances greater than or equal to 1km and less than 3km are averaged to obtain the structure function value corresponding to a distance of 2km; the structure function values ​​for distances greater than or equal to 2km and less than 4km are averaged to obtain the structure function value corresponding to a distance of 3km; and so on, until the average structure function values ​​corresponding to distances of 2, 3, 4...300km are calculated. The relationship between the structure function values ​​of total monthly precipitation and maximum daily monthly precipitation for different distances is shown in [reference needed]. Figure 3 and Figure 4 The above processing can reduce local fluctuations caused by sampling factors other than distance while preserving the overall trend of change.

[0037] Based on the moving average processing of the precipitation structure function, the precipitation structure function is fitted. b ′( d ) and distance d The fitted polynomial regression model (see) Figure 5 and Figure 6 ),Right now: Total monthly precipitation: ; Maximum daily precipitation in a month: .

[0038] in: b rt ′( d )and b rm ′( d () are distances d The corresponding structure functions for total monthly precipitation and maximum daily monthly precipitation; the goodness of fit (i.e., coefficient of determination R²) of the polynomial regression model are 0.9973 and 0.9787, respectively, both close to 1.0, indicating that the observed structure functions and distance relationships of both types of precipitation characteristics can be well represented by the established polynomial model. Based on the polynomial regression model of precipitation structure function and distance relationship obtained above, when... d When =0, b ′(0)=a Substitute 0 s r 2 = a 0 / 2 can be used to calculate the random mean square error of the observations: the random mean square error of the monthly total precipitation observations. ; Random mean square error of monthly maximum daily precipitation observation .

[0039] S004~S005. Calculation of mean square deviation and classification of representativeness level thresholds: Based on the multinomial regression model of the relationship between the structure function of total monthly precipitation and maximum daily monthly precipitation obtained in step S003 and distance, the random mean square error of the observed total monthly precipitation is obtained. s rt 2 =695.8; Random mean square error of monthly maximum daily precipitation observation s rm 2 =167.5; calculate the corresponding mean square deviation respectively. E rt ( d )= b rt ′( d )−695.8 and E rm ( d )= b rm ′( d -167.5. Furthermore, based on the rules for constructing the normalized deviation index, the normalized deviation index for monthly total precipitation was obtained as follows: The normalized deviation index for the maximum daily precipitation each month is: .

[0040] The above-mentioned total monthly precipitation and maximum daily monthly precipitation were normalized to a standard deviation index. I rt ( d ), I rm ( d The representativeness of precipitation observation station data is divided into four levels according to the following standards: I rt ( d ) / I rm ( d )≤2.0, excellent representativeness (Level I); 2< I rt ( d ) / I rm ( d≤2.5, good representativeness (Level II); <2.5 I rt ( d ) / I rm ( d )≤3.0, representative (Level III); I rt ( d ) / I rm ( d If the score is greater than 3.0, it indicates poor representativeness (Level IV).

[0041] The random mean square error of observations obtained in the Jianghan Plain region s rt 2 =695.8 and s rm 2 =167.5, substitute into I rt ( d ), I rm ( d The normalized deviation index threshold can be converted into the representative quality level threshold standard of monthly total precipitation and monthly maximum daily precipitation data (see Table 1 and Table 2).

[0042] Table 1. Representative Classification of Monthly Total Precipitation Data Table 2. Representative Classification of Monthly Maximum Daily Precipitation Data S006. Achieve comprehensive evaluation of the representativeness level of precipitation data. For the two types of precipitation characteristics, a comprehensive discrimination is performed based on the simultaneous fulfillment of the same level, and the smaller of the corresponding distances is taken as the representative distance of the comprehensive precipitation level. The representative distance of the precipitation observation station calculated based on step S005 is... d Mean square deviation of total monthly precipitation at any point E rt Mean square deviation of maximum daily precipitation in the month E rm (See Figure 6 and Figure 7 Based on the representativeness levels of the spatial distance between the total precipitation and the maximum daily precipitation in the month given in Tables 1 and 2, the mean square deviation of the total monthly precipitation is selected. E rt Mean square deviation of maximum daily precipitation in the month E rmIf all distance thresholds are met, the spatial distance range level of the comprehensive representativeness of precipitation data can be obtained (see Table 3), thus realizing the assessment of the representativeness level of precipitation data in the Jianghan Plain.

[0043] Table 3. Representativeness Level and Assessment of Precipitation Observation Station Data The representativeness assessment of precipitation data in the Jianghan Plain is as follows: Precipitation observation stations located 18 km or less from the target point are considered excellent, and their data can be directly used. Precipitation observation stations located greater than 18 km but less than or equal to 27 km from the target point are considered good; however, comparison with underlying surface characteristics is necessary. If the underlying surface characteristics are similar, the data can be directly used; otherwise, further verification or correction is required. Precipitation observation stations located greater than 27 km but less than or equal to 37 km from the target point are considered moderately representative; comparative analysis of regional microclimate characteristics is required. If the correlation is high, the data can be directly used; otherwise, further verification or correction is required. Precipitation observation stations located greater than 37 km from the target point are considered poorly representative and their use is not recommended.

[0044] The objectives of this invention have been fully and effectively achieved through the above embodiments. Those skilled in the art will understand that this invention includes, but is not limited to, the contents described in the accompanying drawings and the specific embodiments described above. Although the invention has been described with reference to what is currently considered the most practical and preferred embodiments, it should be understood that the invention is not limited to the disclosed embodiments, and any modifications that do not depart from the functional and structural principles of the invention will be included within the scope of the claims.

Claims

1. A method for comprehensively evaluating the regional representativeness of precipitation data from meteorological observation stations based on multiple precipitation characteristics, characterized in that, It should include at least the following steps: SS1. Collect and process geographical environmental information and multi-year daily precipitation observation data of various meteorological observation stations in the assessment area, and establish a sample database containing effective station information and daily precipitation data; SS2. Statistically analyze the total monthly precipitation and the maximum daily precipitation for each station, and select the effective precipitation data for the same month in each year to form a precipitation element sample set with corresponding total monthly precipitation and maximum daily precipitation; SS3. Calculate the distance between each pair of stations for all stations. d and the corresponding monthly total precipitation observation structure function b rt ′( d ), monthly maximum daily precipitation observation structure function b rm ′( d ), fitting b rt ′( d )and d The first relational model and b rm ′( d )and d The second relation model, and based on the two relation models in d Intercept at =0 a 0、 b 0, estimate the observational random mean square error of the two precipitation elements respectively. σ rt 2 = a 0 / 2、 σ rm 2 = b 0 / 2; SS4. Based on the first and second relational models, calculate the representative distance of precipitation observation data for any meteorological observation station. d The root mean square bias of total monthly precipitation and maximum daily monthly precipitation resulting from target point precipitation data. E rt ( d )= b rt ′( d )- a 0 / 2 and E rm ( d )= b rm ′( d )- b Based on 0 / 2, normalized deviation indices for total monthly precipitation and maximum daily monthly precipitation are constructed respectively. I rt ( d )= E rt ( d ) / σ rt 2 and I rm ( d )= E rm ( d ) / σ rm 2 ; SS5. Based on I rt ( d ), I rm ( d The threshold standards for the representativeness of precipitation data at each station area were determined. The normalized deviation index ≤2.0 was determined as the upper limit of the representativeness of the excellent level, and >3.0 was determined as the lower limit of the poor level. An intermediate level was also defined between the excellent level and the poor level threshold standards. SS6. Substitute the mean square deviation thresholds corresponding to each representative level into the first and second relationship models to determine the spatial distance ranges corresponding to the total monthly precipitation and the maximum monthly daily precipitation, and determine the spatial distance ranges that simultaneously meet the requirements of the same representative level as the comprehensive evaluation result.

2. The method according to claim 1, characterized in that, Step SS1 includes the following when implemented: S11. Site and Data Collection: Collect geographical environment information and multi-year daily precipitation observation data for each site in the assessment area. The geographical environment information shall include at least the longitude, latitude and altitude of the site. The multi-year daily precipitation observation data shall include at least 6 years or more of continuous observation data. The spatial distribution density of sites in the assessment area shall meet the requirement that the minimum distance between adjacent sites does not exceed 15km. S12. Data completeness screening: Statistically evaluate the completeness of daily precipitation data collected from each station, and remove stations with an annual precipitation data missing rate exceeding 25% or a summer precipitation data missing rate exceeding 10%. S13. Correlation test: Calculate the correlation coefficient between each station and the precipitation data of stations within a 30km radius. Remove stations with a correlation coefficient of less than 0.1 with more than half of the stations within a 30km radius. Based on the effective stations that pass the completeness test and the spatial correlation test, establish a sample database containing station geographical environment information and daily precipitation data.

3. The method according to claim 1, characterized in that, Step SS2 includes the following when implemented: S21. Based on the daily precipitation sample database, the total precipitation and the maximum daily precipitation for each month at each valid station are statistically analyzed to form candidate monthly samples organized by station, year, and month. S22. If precipitation data for one day in a month is missing, then the total monthly precipitation and the maximum daily precipitation for that month shall be marked as missing. S23. Precipitation data for May to September each year must not be missing for more than one month, and precipitation data for other months must not be missing for more than three months. The number of valid data samples with no missing data selected for each station must be ≥36 groups.

4. The method according to claim 1, characterized in that, In step SS3, the integrated modeling of the structure function of multiple precipitation characteristic measurements includes at least the following sub-steps: S31. Calculation of inter-station distances: Based on the precipitation sample set formed in step SS2, statistically evaluate the inter-station distances between all valid stations in the evaluation area. d ; S32. Calculation of Observational Structure Function: For both the monthly total precipitation sample and the monthly maximum daily precipitation sample, calculate the observational structure function between all valid stations. b r '( d For any two stations A and B, b r '( A , B The mean square of the time series of the difference between the observed precipitation values ​​at two stations is defined as the average of the differences between the two station observations. S33. Relationship Field Construction: The calculated inter-station distances and corresponding observation structure functions are paired and combined to form a relationship field between inter-station distances and corresponding structure functions. The number of data sets is... N = k ×( k -1) / 2, k To assess the number of valid sites within the region.

5. The method according to claim 4, characterized in that, In step SS3, the construction of the first relation model and the second relation model includes at least the following sub-steps: S34. Moving average processing: for N The group of sites performed a 3km interval moving average on the relational field data, and then processed the data within 1km ≤ d <3km b r '( d The average value is then used as the observation structure function for a distance of 2km. (2km≤) d <4km b r '( d The average value is used as the observation structure function corresponding to a distance of 3km. This process is repeated until the average observation structure function corresponding to the maximum distance in the region is calculated. S35. Polynomial Regression Modeling: Based on the data processed by moving average, fit the monthly total precipitation observation structure function with respect to inter-station distance. d Multinomial regression model b rt ′( d )= a 0+ a 1 d + a 2 d 2 +…+ a m d m A first relational model was formed, and the structure function of monthly maximum daily precipitation observations was fitted with respect to distance. d Multinomial regression model b rm ′( d )= b 0+ b 1 d + b 2 d 2 +…+ b n d n This forms a second relational model, in which m , n For the order of the polynomial, a 0、 a 1. a 2. ... a m as well as b 0、 b 1. b 2. ... b n These are the polynomial coefficients; S36. Determining the optimal order: Compare the sum of squared errors between the fitting results of different orders and the actual observed structure function values, and select the order and coefficients corresponding to the smallest sum of squared errors as the optimal fitting result.

6. The method according to claim 5, characterized in that, In step SS3, the accuracy verification of the two polynomial regression models includes at least the following: based on the optimal polynomial regression model of the observed structure function of total monthly precipitation and maximum daily monthly precipitation determined in step S36, calculating the fitted values ​​corresponding to the distances between each station, and comparing them distance-by-distance with the actual observed structure function values ​​processed by the moving average in step S34, and calculating the coefficients of determination for each model. R 2 When the coefficient of determination of two models R 2 When both reach a preset goodness-of-fit threshold close to 1.0, it is determined that the first and second relationship models can accurately characterize the quantitative relationship between the observed structure function and the inter-station distance; when the fitting accuracy does not meet the preset requirements, return to steps S35 and S36, readjust the polynomial order and perform regression fitting and optimal order determination until the preset goodness-of-fit threshold requirements are met.

7. The method according to claim 5 or 6, characterized in that, In steps SS3 and SS4, the estimation of the random mean square error and the calculation of the mean square deviation of the observations include at least the following sub-steps: S41. Error Decomposition: Decompose the observed precipitation deviation at any station into the sum of the true deviation and the observed error, and establish the observation structure function under the condition that the evaluation area satisfies homogeneity and isotropy. b r ′( d ) and truth structure functions b r ( d Analytical relationships between ) b r ′( d )= b r ( d )+2 σ r 2 ; S42. Calculation of random mean square error: in spatial distance d When = 0, use the truth structure function b r ( d Given the condition )=0, we obtain that the observed random mean square error satisfies σ r 2 = b r ′(0) / 2, and based on this, extrapolate the values ​​to the zero distance using the first and second relationship models, and calculate the observed random mean square error corresponding to the total monthly precipitation and the maximum monthly daily precipitation. σ rt 2 = a 0 / 2、 σ rm 2 = b 0 / 2; S43. Mean Square Deviation Calculation and Index Normalization: Mean Square Deviation Based on Station Data Substituting for Target Point Data E r = b r ′( d )- σ r 2 The mean square deviations of the total monthly precipitation and the maximum daily precipitation were calculated respectively. E rt ( d )= b rt ′( d )- a 0 / 2 and E rm ( d )= b rm ′( d )- b 0 / 2, and based on this, normalized deviation indices for total monthly precipitation and maximum daily monthly precipitation are constructed respectively. I rt ( d )= E rt ( d ) / σ rt 2 and I rm ( d )= E rm ( d ) / σ rm 2 .

8. The method according to claim 1, characterized in that, In step SS5, the representativeness level threshold classification includes at least the following: a normalized deviation index based on total monthly precipitation and maximum daily monthly precipitation. I rt ( d ), I rm ( d Based on the condition that the deviation caused by distance variation is no greater than the random mean square error of observation, the normalized deviation index ≤2.0 is determined as the threshold standard for the representative excellent level, and >3.0 is determined as the threshold standard for the representative poor level. Between the representative excellent level and the representative poor level, the normalized deviation index >2.0 and ≤2.5 is determined as the good level, and >2.5 and ≤3.0 is the medium level. The four levels are used to determine the representative level of monthly total precipitation and monthly maximum daily precipitation.

9. The method according to claim 1, characterized in that, In step SS6, the reverse calculation of the upper limit of the applicable spatial range for each representative level includes at least the following sub-steps: S61. Single Feature Distance Inversion: For any representative level threshold λ According to respectively E rt ( d )= λ · σ rt 2 and E rm ( d )= λ · σ rm 2 Determine the mean square deviations corresponding to the total monthly precipitation and the maximum daily precipitation, and then respectively based on E rt ( d )= b rt ′( d )- a 0 / 2 and E rm ( d )= b rm ′( d )- b By combining 0 / 2 with the first and second relationship models, we can deduce the spatial distances corresponding to the representative levels of total monthly precipitation and maximum daily monthly precipitation. d rt , d rm ; S62. Comprehensive Judgment Based on Multiple Feature Quantities: The spatial distance range that simultaneously meets the same representativeness level requirements for monthly total precipitation and monthly maximum daily precipitation is determined as the comprehensive precipitation representativeness level distance. d r and with d r As a distance basis for selecting alternative precipitation observation station data when there are no observation target points.

10. The method according to claim 9, characterized in that, In step S62, the spatial distance corresponding to the representative level of the total monthly precipitation. d rt Spatial distance corresponding to the representative level of the maximum daily precipitation in a month d rm When they are not equal, d rt and d rm The smaller of the values ​​is determined as the representative level of the comprehensive precipitation at the corresponding representative level. d r This ensures that the comprehensive spatial distance range simultaneously meets the representativeness requirements of both the total monthly precipitation and the maximum monthly daily precipitation at the same representative level.

11. A system for evaluating the regional representativeness of precipitation data from a meteorological observation station, characterized in that, This method is used to implement the regional representativeness of precipitation data from meteorological observation stations based on a comprehensive assessment of multiple precipitation characteristics, as described in any one of claims 1 to 10.

Images