Data quality monitoring method and system based on correlation analysis of networking observation values

Abstract

The invention discloses a data quality monitoring method and system based on correlation analysis of networking observation values, and relates to the technical field of meteorological data quality monitoring, and the method comprises the steps: obtaining the position information of each meteorological station in a target region from high-resolution elevation DEM data, and then classifying each meteorological station in the target region, and the regional radius is determined according to the classification condition, so that the correlation of each meteorological station and the correlation among the meteorological stations in the regional radius are analyzed, the reliability of data quality is improved to a great extent, the spatial correlation and regional feature correlation among meteorological elements are deepened, and the accuracy of the data quality is improved. And meanwhile, the final suspicious degree of each meteorological station is obtained through correlation analysis, so that each suspicious meteorological station is obtained, the current monitoring data of each suspicious meteorological station is corrected based on the historical monitoring data of each suspicious meteorological station, and the accuracy of meteorological monitoring is ensured to a great extent.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of meteorological data quality monitoring, and particularly relates to a data quality monitoring method and system based on correlation analysis of networked observation values. BACKGROUND

[0002] Meteorological detection mainly relies on various detection equipment, wherein the ground meteorological observation station is consistent with the space where people are located, and the accuracy of ground meteorological data is of great significance for reference of meteorological application. Since the distribution of meteorological elements has continuity in time and space, there is no sudden change, so the meteorological data collected by the ground meteorological station can represent the meteorological element conditions in a certain space range.

[0003] In recent years, the number of meteorological stations is increasing, and an automatic quality control method is needed to solve the continuous monitoring of thousands of automatic stations. At present, the meteorological system mainly uses the meteorological data business system MDOS for quality control, and the abnormal station judgment algorithm mainly includes the climatological extreme value range, spatial consistency, time consistency and multi-source data consistency of observation data. The quality control method is too simple and lacks flexibility, and therefore the application provides a data quality monitoring method and system based on correlation analysis of networked observation values.

[0004] The prior art such as the invention application patent with the announcement number CN119439083A discloses a meteorological radar data quality control method, system and equipment based on fuzzy logic, which comprises the following steps: constructing a radar reflectivity intensity product in a three-dimensional Cartesian coordinate system; performing OpenGL rendering on the three-dimensional reflectivity intensity product as texture data; obtaining the elevation data of the radar effective scanning range, and performing three-dimensional terrain rendering on the elevation value as the texture data of the rendered three-dimensional terrain; providing a constant height plane position indication map product and a combined reflectivity product to judge the echo type, and providing an arbitrary vertical section product to check whether the echo type is judged accurately; after determining the echo type, providing interactive selection of a target echo area, counting the radar parameter characteristics of the target echo area and visualizing; localizing the membership function of the ground clutter to complete the quality control of the radar reflectivity intensity data.

[0005] The prior art such as the application patent for a method for quality control of hydro-meteorological data using correlation, with publication number CN120086769A, relates to the field of marine meteorology, and uses the correlation between elements to control the quality of hydro-meteorological element data, that is, a correlation quality control method. First, the data is sorted to include information such as time, location, and instrument status, and abnormal values are identified through routine quality control testing. The correlation between elements is analyzed, and is divided into natural correlation and algorithm correlation. Natural correlation determines data abnormalities by testing the reasonableness and contradiction of element values. Algorithm correlation determines the abnormality of related elements after detecting abnormalities through routine quality control. The correlation quality control method covers time, location, salinity, vector, relative humidity, wind-wave-flow, and other elements, and emphasizes judging the correlation based on the source and method of element observation to prevent over-control. The method has a clear process and strong scalability, and can effectively detect suspicious or abnormal data.

[0006] For the above-mentioned scheme, there are the following technical problems: 1. Current artificial intelligence methods are widely used in meteorological data anomaly detection, but the application of artificial intelligence and neural networks makes it difficult for people to understand the specific quality control process, and lacks interpretability. Similarly, a prediction model for each element of a single station can be established by support vector machines, and the problem of the station can be detected by comparing the predicted value with the measured value, but only a single element of each meteorological station is trained, which is a simple fitting of data and a single algorithm. Different meteorological elements need to be set with an over-limit threshold, which is prone to false identification.

[0007] 2. The current technology can also analyze the correlation between time, location, instrument status and other information to control and analyze meteorological data, but the current technology mainly carries out correlation rule mining on the internal correlation between single station meteorological elements, detects stations that do not meet the element correlation rules, and the current technology is based on the time consistency criterion of a single station. However, each station has its own characteristics, and there is a correlation between meteorological stations. The current technology ignores this aspect, which leads to a lack of comprehensiveness and accuracy of the quality control results. SUMMARY

[0008] The purpose of the present application is to provide a data quality monitoring method and system based on network observation value correlation analysis, which solves the problems in the background art.

[0009] To solve the above technical problems, the present application adopts the following technical scheme: the present application provides a data quality monitoring method based on network observation value correlation analysis in the first invention, comprising: step one, obtaining the position information of each meteorological station in the target area according to the high-resolution elevation DEM data, then classifying each meteorological station, and determining the regional radius according to the classification, so as to analyze the density of each meteorological station.

[0010] Step two, obtain each non-suspect meteorological data and each suspect meteorological data of each site according to historical meteorological data analysis of each site, and then analyze the correlation of each meteorological station according to each non-suspect meteorological data, first analyze the suspicious degree of each meteorological station, then analyze the correlation suspicious degree of each type of meteorological station within the radius of the region, and finally comprehensively analyze the final suspicious degree of each meteorological station.

[0011] Step three, obtain each suspect meteorological station according to the final suspicious degree of each meteorological station, and monitor each suspect meteorological station, obtain the data correction value of each suspect meteorological station according to the monitored meteorological data and meteorological data trend analysis, and then correct the meteorological data of each suspect station.

[0012] The second application provides a data quality monitoring system based on network observation value correlation analysis, which includes: a meteorological station classification module: used to obtain the position information of each meteorological station in the target region according to high-resolution elevation DEM data, and then classify each meteorological station, and determine the radius of the region according to the classification, so as to analyze the density of each meteorological station.

[0013] The suspicious degree analysis module is used to obtain each non-suspect meteorological data and each suspect meteorological data of each site according to historical meteorological data analysis of each site, and then analyze the correlation of each meteorological station according to each non-suspect meteorological data, first analyze the suspicious degree of each meteorological station, then analyze the correlation suspicious degree of each type of meteorological station within the radius of the region, and finally comprehensively analyze the final suspicious degree of each meteorological station.

[0014] The suspect station correction module is used to obtain each suspect meteorological station according to the final suspicious degree of each meteorological station, and monitor each suspect meteorological station, obtain the data correction value of each suspect meteorological station according to the monitored meteorological data and meteorological data trend analysis, and then correct the meteorological data of each suspect station.

[0015] The beneficial effects of the application are: 1, the data quality monitoring method and system based on network observation value correlation analysis provided by the application obtains the position information of each meteorological station in the target region from high-resolution elevation DEM data, and then classifies each meteorological station in the target region, and determines the radius of the region according to the classification, so as to analyze the correlation of each meteorological station and the correlation between each meteorological station within the radius of the region, greatly improving the reliability of data quality, deepening the spatial correlation and regional characteristic correlation of meteorological elements, and obtaining the final suspicious degree of each meteorological station through correlation analysis, so as to obtain each suspect meteorological station, and correct the current monitoring data of each suspect meteorological station based on the historical monitoring data of each suspect meteorological station, greatly ensuring the accuracy of meteorological monitoring.

[0016] 2、The application realizes the classification of each meteorological station in the target area through GIS technology, lays a foundation for subsequent correlation analysis of each meteorological station, and not only performs correlation analysis on a single meteorological station, but also performs correlation analysis among each meteorological station, greatly improves the reliability of data quality, and deepens the spatial correlation and regional characteristic correlation of meteorological elements.

[0017] 3、The application realizes meteorological station data constraint rule mining and automatic updating based on correlation analysis in data mining, thereby converting the implicit rules of each meteorological data into explicit rules, improving the accuracy and reliability of data quality, and simultaneously performing dynamic rule updating, which can better adapt to changes in the meteorological environment, reduce labor costs, and provide data basis for subsequent meteorological data correction. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0019] Figure 1 The method embodiment of the present application is a schematic diagram of the process flow.

[0020] Figure 2 The system structure connection diagram of the present application. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0022] Referring to Figure 1 The present application provides a data quality monitoring method and system based on network observation value correlation analysis in the first aspect, including the following steps: step one, obtaining the position information of each meteorological station in the target area according to high-resolution elevation DEM data, then classifying each meteorological station, and determining the regional radius according to the classification, thereby analyzing the density of each meteorological station.

[0023] It should be noted that the position information includes the altitude, slope and distance between each meteorological station.

[0024] In one specific example, the position information of each weather station in the target area is obtained according to high-resolution elevation DEM data, and the specific process is as follows: high-resolution elevation DEM data of each weather station in the target area is obtained from a data platform, and then the elevation pixel coordinates in the high-resolution elevation DEM data of a certain weather station in the target area are marked as (x, y), the elevation of the elevation point is marked as h xy , the horizontal resolution of the pixel is marked as hx, the vertical resolution of the pixel is marked as hy, the elevations of the pixel points above, below, left and right are marked as h x,y-1 , h x,y+1 , h x-1,y and h x+1,y , the horizontal slope factor of the pixel point is marked as S x , and the vertical slope factor of the pixel point is marked as S y , wherein According to the calculation formula: The slope S of the weather station is calculated, and the slopes S of each weather station in the target area are calculated accordingly. i , wherein i is the number of each weather station in the target area, and i is a positive integer.

[0025] It should be noted that the data platform includes geographic spatial data cloud and resource environment science and data sharing center, etc.

[0026] At the same time, the distance between each weather station is obtained by calculating the latitude and longitude of the high-resolution elevation DEM data, and the altitude of each weather station is obtained by matching the precision of the high-resolution elevation DEM data.

[0027] It should be noted that the latitude and longitude calculation and precision matching are prior art, and thus will not be described again.

[0028] In one specific example, the weather stations are further classified, and the area radius is determined according to the classification, so as to analyze the density of each weather station, and the specific process is as follows: each weather station with a slope greater than 15° or an altitude greater than 500 meters is recorded as each mountain weather station, and each weather station with a slope less than or equal to 15° and an altitude less than or equal to 500 meters is recorded as each plain weather station, and then each weather station in the target area is classified.

[0029] The distance between each mountain weather station is obtained based on the distance between each weather station, so as to determine the area radius R corresponding to the mountain weather station, and the number of stations in the area radius R corresponding to the mountain weather station is not less than 10, and the area radius R' corresponding to each plain weather station is determined accordingly.

[0030] It should be noted that the mountain weather stations and the plain weather stations will have different weather data at the same time point due to the difference in altitude and slope, and therefore different regional radii are used to determine the density of each mountain weather station and each plain weather station.

[0031] Step two, according to the historical weather data of each station, each non-suspect weather data and each suspect weather data of each station are obtained, and then according to each non-suspect weather data, the correlation analysis of each weather station is carried out, first the suspicious degree of each weather station is analyzed, then the correlation suspicious degree of each type of weather station in the regional radius is obtained, and finally the final suspicious degree of each weather station is obtained by comprehensive analysis.

[0032] In a specific example, the non-suspect weather data and the suspect weather data of each station are obtained according to the historical weather data of each station: the historical weather data of each weather station is obtained from the weather data management center of the target area, and after the mean value processing of each historical weather data, -3 times of standard deviation is taken as the lower limit value of the non-suspect weather data score, and at the same time, after the mean value processing of each historical weather data, +3 times of standard deviation is taken as the upper limit value of the non-suspect weather data score, and thus the non-suspect weather data score interval of each weather station in the target area is obtained.

[0033] The non-suspect weather data of each weather station in the preset time period is obtained by comparing each weather data monitored by each weather station in the preset time period with the non-suspect weather data score interval.

[0034] It should be noted that the length of the preset time period is set by the relevant staff, and is not limited here, wherein the longer the preset time period, the more weather data obtained, and the more accurate the correlation analysis of each weather station.

[0035] In one specific example, the first analysis of each weather station's own suspicious degree is as follows: six element values are extracted from the weather data of each weather station: temperature, precipitation, air pressure, sunshine hours, relative humidity and wind speed, respectively denoted as u, v, w, p, q and z, correlation analysis is performed on each element value by Pearson correlation coefficient method, the correlation coefficients between each element value are obtained, and each correlation coefficient is sorted in descending order, and then the correlation coefficient values of the top three element values are obtained, respectively uw, up and uq, wherein uw>up>uq, and the expression 1=A*up+B*uw is constructed, wherein A and B represent the weight factors of the correlation coefficients up and uw respectively, and A+B=1, the real-time weather data of a certain weather station is input into the expression to obtain: sc=A*up+B*uw, wherein SC represents the correlation coefficient of up and uw, and the self-correlation suspicious degree sd of the element value of the weather station is calculated by the formula: sd=1-sc, and the correlation coefficient value sc of the element value of each weather station can be calculated in the same way. i and the self-correlation suspicious degree sd i .

[0036] It should be noted that the Pearson correlation coefficient method is a prior art, and thus will not be described again.

[0037] It should be noted that the specific values of A and B are set by relevant personnel, and are not specifically limited here, for example, the historical weather data of each weather station within the previous 365 days is obtained, and the weight factors of the correlation coefficients up and uw can be calculated by the analytic hierarchy process, wherein the analytic hierarchy process is a prior art, and thus will not be described again.

[0038] In one embodiment, the reanalysis obtains the correlation suspiciousness of each meteorological station in the region radius, and the specific analysis process is as follows: setting the element value sequence of each meteorological station in the region radius in the previous 365 days as Ri, the correlation between x0 and x6, x8 and x9 is calculated by the Pearson correlation coefficient method, and the three correlation coefficient values are x0-6, x0-8 and x0-9, wherein |x0-6|>|x0-9|>|x0-8|, that is, the correlation between x0 and x6 is the strongest, and the correlation between x0 and x8 is the weakest, taking the two correlation coefficients with the strongest correlation, and constructing the expression: 1=C*X0-6+D*X0-9, wherein C and D represent the weight coefficients corresponding to the correlation coefficients x0 and x8 respectively, and C+D=1, inputting the real-time meteorological data of a meteorological station into the expression to obtain rc=C*X0-6+D*X0-9, wherein rc represents the correlation coefficient of the correlation coefficients x0-6 and x0-9, and the self-correlation suspiciousness rd of the meteorological station is calculated by the formula: rd=1-rc, and the correlation coefficient value rc of the element value sequence of the meteorological station in the region radius in the previous 365 days is calculated in the same way. i and the self-correlation suspiciousness rd i .

[0039] It should be noted that the values of C and D are obtained in the same way as A and B, and thus will not be described again.

[0040] In one embodiment, the final comprehensive analysis obtains the final suspiciousness of each meteorological station, and the specific analysis process is as follows: according to the self-correlation suspiciousness sd i of the element value of each meteorological station in the target region and the self-correlation suspiciousness rd i of the element value sequence of each meteorological station in the region radius in the previous 365 days, the final suspiciousness fd i of each meteorological station is calculated by the formula: fd i =E*sd i +F*rd i , wherein E and F represent the weight factors corresponding to sd i and rd i respectively, and E+F=1.

[0041] It should be noted that the specific configuration of E and F is set by the relevant staff according to the distribution of each meteorological station, and is not specifically limited here, for example, if the number of stations within the radius of the region is greater than 15, then E=0.3, F=0.7; if the number of stations within the radius of the region is greater than 10, then E=0.4, F=0.6; if the number of stations within the radius of the region is greater than 5, then E=0.5, F=0.5; if the number of stations within the radius of the region is greater than 3, then E=0.6, F=0.4; if the number of stations within the radius of the region is less than 3, then E=0.7, F=0.3, etc.

[0042] It should be noted that the weight factors A, B, C and D are continuously iteratively updated using historical meteorological data for the previous 365 days.

[0043] Step three, according to the final suspicious degree analysis of each meteorological station, each suspicious meteorological station is obtained, and each suspicious meteorological station is monitored, and the data correction value of each suspicious meteorological station is obtained according to the monitored meteorological data and meteorological data trend analysis, and then the meteorological data of each suspicious station is corrected.

[0044] In a specific example, the final suspicious degree of each meteorological station is normally distributed, and the final suspicious degree of two-thirds of the normal distribution is recorded as the suspicious degree threshold of the meteorological station, and the final suspicious degree greater than the suspicious degree threshold is recorded as each suspicious meteorological station.

[0045] In a specific example, the data correction value of each suspicious meteorological station is obtained according to the monitored meteorological data, and then the meteorological data of each suspicious station is corrected: the historical element values of each suspicious meteorological station for the previous 365 days are obtained, the historical element values are input into the LSTM neural model, and then the time trend analysis of each element value of each suspicious station is performed, the data deviation between each historical element value and each reference element value of each suspicious station is calculated, the data deviation trend of each suspicious station is fitted by nonlinear regression method, and then each dynamic correction coefficient is obtained, each dynamic correction coefficient is input into the regional element space distribution model, if the data of the corrected suspicious station makes the regional space distribution smooth, then each meteorological station is corrected using each dynamic correction value; otherwise, the trend model parameters are adjusted again until the spatial regional distribution is smooth.

[0046] It should be noted that the meteorological station with the lowest final suspicious degree is recorded as the reference meteorological station, and the element values collected by the reference meteorological station are recorded as the reference element values.

[0047] It should be noted that the LSTM neural model and the regional space distribution model are both existing model technologies, and will not be described here.

[0048] The application provides a data quality monitoring system based on network observation value correlation analysis in the second application, comprising: a meteorological station classification module: used for obtaining the position information of each meteorological station in the target area according to high-resolution elevation DEM data, and then classifying each meteorological station, and determining the regional radius according to the classification, so as to analyze the density of each meteorological station.

[0049] Referring to Figure 2 The application provides a data quality monitoring system based on network observation value correlation analysis in the second aspect, comprising the following modules: a meteorological station classification module: used for obtaining the position information of each meteorological station in the target area according to high-resolution elevation DEM data, and then classifying each meteorological station, and determining the regional radius according to the classification, so as to analyze the density of each meteorological station.

[0050] A suspicious degree analysis module: used for analyzing each non-suspected meteorological data and each suspected meteorological data of each station according to the historical meteorological data of each station, and then performing correlation analysis on each meteorological station according to each non-suspected meteorological data, first analyzing the suspicious degree of each meteorological station, then analyzing the correlation suspicious degree of each type of meteorological station within the regional radius, and finally comprehensively analyzing the final suspicious degree of each meteorological station.

[0051] A suspicious station correction module: used for analyzing each suspected meteorological station according to the final suspicious degree of each meteorological station, and performing key monitoring on each suspected meteorological station, analyzing the data correction value of each suspected meteorological station according to the monitored meteorological data and meteorological data change trend, and then correcting the meteorological data of each suspected station.

[0052] The data quality monitoring method and system based on network observation value correlation analysis provided by the application greatly improve the reliability of data quality by obtaining the position information of each meteorological station in the target area from high-resolution elevation DEM data, classifying each meteorological station in the target area, and determining the regional radius according to the classification, so as to analyze the correlation of each meteorological station itself and the correlation between each meteorological station within the regional radius, deepen the spatial correlation and regional characteristic correlation of meteorological elements, and at the same time, through correlation analysis, the final suspicious degree of each meteorological station is obtained, so as to obtain each suspected meteorological station, and based on the historical monitoring data of each suspected meteorological station, the current monitoring data of each suspected meteorological station is corrected, which greatly ensures the accuracy of meteorological monitoring.

[0053] The above merely provides the description and illustration of the concept of the present application, and those skilled in the art can make various modifications or supplements to the described specific embodiments or adopt similar ways to replace, as long as the modifications or supplements do not deviate from the concept of the present application or exceed the scope defined by the present application, which shall belong to the protection scope of the present application.

Claims

1. A method for monitoring data quality based on correlation analysis of networked observations, characterized in that, The application relates to a suspicious meteorological station screening method and device. Step one: obtaining position information of each meteorological station in a target region according to high-resolution elevation DEM data, then classifying each meteorological station, and determining a region radius according to the classification, so as to analyze the density of each meteorological station; Step two: obtaining each non-suspicious meteorological data and each suspicious meteorological data of each station according to historical meteorological data of each station, then performing correlation analysis on each meteorological station according to each non-suspicious meteorological data, first analyzing the suspicious degree of each meteorological station, then analyzing the correlation suspicious degree of each type of meteorological station within the region radius, and finally comprehensively analyzing the final suspicious degree of each meteorological station; Step three: obtaining each suspicious meteorological station according to the final suspicious degree of each meteorological station, and performing key monitoring on each suspicious meteorological station, analyzing the data correction value of each suspicious meteorological station according to the monitored meteorological data and meteorological data change trend, and then correcting the meteorological data of each suspicious station.

2. The method for data quality monitoring based on correlation analysis of networked observations according to claim 1, characterized in that, The specific process of obtaining the position information of each meteorological station in the target region according to the high-resolution elevation DEM data is as follows: Obtaining high-resolution elevation DEM data of each weather station in the target area from the data platform, and then marking the elevation pixel coordinates in the high-resolution elevation DEM data of a certain weather station in the target area as (x, y), the elevation of the elevation point as h xy , the horizontal resolution of the pixel as hx, the vertical resolution of the pixel as hy, the elevations of the upper, lower, left and right of the pixel point as h x,y-1 , h x,y+1 , h x-1,y and h x+1,y , the horizontal slope factor of the pixel point as S x , and the vertical slope factor of the pixel point as S y , wherein According to the calculation formula: , the slope S of the weather station is calculated, and the slope S of each weather station in the target area is calculated accordingly i , wherein i is the number of each weather station in the target area, and i is a positive integer. Meanwhile, the distance between each meteorological station is obtained by latitude and longitude calculation of the high-resolution elevation DEM data, and the altitude of each meteorological station is obtained by precision matching of the high-resolution elevation DEM data.

3. The method for data quality monitoring based on correlation analysis of networked observations according to claim 2, characterized in that, The specific process of classifying each meteorological station and determining the region radius according to the classification, so as to analyze the density of each meteorological station, is as follows: Each meteorological station with a slope greater than 15 degrees or an altitude greater than 500 meters is recorded as each mountain meteorological station, and each meteorological station with a slope less than or equal to 15 degrees and an altitude less than or equal to 500 meters is recorded as each plain meteorological station, so as to classify each meteorological station in the target region; The distance between each mountain meteorological station is obtained based on the distance between each meteorological station, so as to determine the region radius R corresponding to the mountain meteorological station, and the number of stations in the region radius R corresponding to the mountain meteorological station is not less than 10, and the region radius R' corresponding to each plain meteorological station is determined in the same way.

4. The method for data quality monitoring based on correlation analysis of networked observations according to claim 3, characterized in that, The specific process of obtaining each non-suspicious meteorological data and each suspicious meteorological data of each station according to historical meteorological data of each station is as follows: Each historical meteorological data of each meteorological station is obtained from a meteorological data management center of the target region, and after mean value processing of each historical meteorological data, -3 times standard deviation is taken as the lower limit value of the non-suspicious meteorological data score, and after mean value processing of each historical meteorological data, +3 times standard deviation is taken as the upper limit value of the non-suspicious meteorological data score, so as to obtain the non-suspicious meteorological data score interval of each meteorological station in the target region; Each meteorological data monitored by each meteorological station in a preset time period is compared with the non-suspicious meteorological data score interval, so as to obtain each non-suspicious meteorological data of each meteorological station in the preset time period. 5.The method and system for data quality monitoring based on correlation analysis of networked observations, according to claim 4, wherein, The specific analysis process of first analyzing the suspicious degree of each meteorological station is as follows: Six element values of temperature, precipitation, air pressure, sunshine duration, relative humidity and wind speed are extracted from the meteorological data of each weather station, respectively denoted as u, v, w, p, q and z. Correlation analysis is performed on each element value by Pearson correlation coefficient method to obtain the correlation coefficients between each element value. The correlation coefficients are sorted in descending order, and the correlation coefficient values of the top three element values are obtained, respectively denoted as uw, up and uq, wherein uw>up>uq. An expression 1=A*up+B*uw is constructed, wherein A and B represent the weight factors of the correlation coefficients up and uw respectively, and A+B=1. Real-time meteorological data of a weather station is input into the expression to obtain sc=A*up+B*uw, wherein SC represents the correlation coefficient of up and uw. The self-correlation suspiciousness sd of the element value of the weather station is calculated by the formula sd=1-sc. Similarly, the correlation coefficient values sc of the element values of each weather station can be calculated. i and self-correlation suspiciousness sd i .

6. The method for data quality monitoring based on correlation analysis of networked observations according to claim 5, wherein, The specific analysis process of analyzing the correlation suspicious degree of each type of meteorological station within the region radius is as follows: The element value sequence of the area radius corresponding to each weather station in the previous 365 days is set as Ri, the correlation between x0 and x6, x8 and x9 is calculated by the Pearson correlation coefficient method, and the three correlation coefficient values are x0-6, x0-8 and x0-9, wherein |x0-6|>|x0-9|>|x0-8|, that is, the correlation between x0 and x6 is the strongest, and the correlation between x0 and x8 is the weakest, the two correlation coefficients with the strongest correlation are taken, and the expression: 1=C*X0-6+D*X0-9 is constructed, wherein C and D represent the weight coefficients corresponding to the correlation coefficients x0 and x8 respectively, and C+D=1, the real-time weather data of a certain weather station is input into the expression to obtain rc=C*X0-6+D*X0-9, wherein rc represents the correlation coefficient of the correlation coefficient values x0-6 and x0-9, and the self-correlation suspicious degree rd of the weather station is calculated by the formula: rd=1-rc, and the correlation coefficient values rc of the element value sequence of the area radius corresponding to each weather station in the previous 365 days are calculated in the same way i and the self-correlation suspicious degree rd i .

7. The method for data quality monitoring based on correlation analysis of networked observations according to claim 6, wherein, The specific analysis process of comprehensively analyzing the final suspicious degree of each meteorological station is as follows: Based on the degree of suspicion of the intrinsic correlation of the element values ​​of each meteorological station in the target area. i And the autocorrelation suspicion of the element value series within the corresponding regional radius of each meteorological station for the previous 365 days. i Through the calculation formula: fd i =E*sd i +F*rd i The final suspicion level fd of each meteorological station was calculated. i Where E and F represent sd i Corresponding weighting factors and rd i The corresponding weighting factor, and E+F=1. 8.The method and system for data quality monitoring based on correlation analysis of networked observations of claim 7, wherein, The suspicious meteorological stations are obtained according to the final suspicious degree analysis of each meteorological station, and each suspicious meteorological station is monitored, and the specific analysis process is as follows: The final suspicious degree of each meteorological station is normally distributed, and the final suspicious degree of two-thirds of the normal distribution is recorded as the suspicious degree threshold of the meteorological station, and each meteorological station with a final suspicious degree greater than the suspicious degree threshold is recorded as each suspicious meteorological station.

9. The method for data quality monitoring based on correlation analysis of networked observations according to claim 8, wherein, The data correction value of each suspicious meteorological station is obtained according to the monitored meteorological data, and then the meteorological data of each suspicious station is corrected: The historical element values of each suspicious meteorological station in the past 365 days are obtained, the historical element values are input into the LSTM neural model, and then the time trend analysis of each element value of each suspicious station is carried out, the data deviation between each historical element value and each reference element value of each suspicious station is calculated, the data deviation trend of each suspicious station is fitted by nonlinear regression method, and thus each dynamic correction coefficient is obtained, each dynamic correction coefficient is input into the regional element space distribution model, if the data of the corrected suspicious station makes the regional space distribution smooth, then each meteorological station is corrected by using each dynamic correction value; otherwise, the trend model parameters are adjusted until the spatial regional distribution is smooth.

10. A networked observation-based correlation analysis data quality monitoring system for performing the networked observation-based correlation analysis data quality monitoring method according to any one of claims 1 to 9, characterized by It includes: The meteorological station classification module is used for obtaining the position information of each meteorological station in the target region according to the high-resolution elevation DEM data, and then classifying each meteorological station, and determining the regional radius according to the classification, so as to analyze the density of each meteorological station; The suspicious degree analysis module is used for analyzing each non-suspicious meteorological data and each suspicious meteorological data of each station according to the historical meteorological data of each station, and then analyzing the correlation of each meteorological station according to each non-suspicious meteorological data, first analyzing the suspicious degree of each meteorological station, then analyzing the related suspicious degree of each meteorological station in the regional radius, and finally comprehensively analyzing the final suspicious degree of each meteorological station; The suspicious station correction module is used for obtaining each suspicious meteorological station according to the final suspicious degree analysis of each meteorological station, and monitoring each suspicious meteorological station, obtaining the data correction value of each suspicious meteorological station according to the monitored meteorological data and the meteorological data change trend, and then correcting the meteorological data of each suspicious station.

Images