A disaster weather-oriented ground observation data anomaly monitoring method and system

Abstract

A method and system for monitoring anomalies in ground observation data during severe weather is disclosed, relating to the field of integrated meteorological data monitoring technology. The method includes obtaining a national list of automatic ground stations based on the Tianyuan interface and obtaining all hourly ground observation data entry records corresponding to the current operational time based on the Tianqing interface. The national list of automatic ground stations is then cleaned and standardized using these hourly ground observation data entry records to obtain a target receiving station information list corresponding to the current operational time. For each target station in the target receiving station information list, if no corresponding hourly ground observation data entry record exists in the database, the target station is determined to be an abnormal station that has not uploaded observation data for the current operational time. Early warnings are issued for all abnormal stations. This application can promptly and accurately identify missing reporting stations during severe weather, thereby improving the support capability for real-time interpolation services.

Description

Technical Field

[0001] This application relates to the field of integrated meteorological data monitoring technology, specifically to a method and system for monitoring anomalies in ground observation data for severe weather. Background Technology

[0002] Against the backdrop of current global climate change, extreme weather disasters are occurring frequently, seriously threatening people's lives and property. Surface meteorological observation data (i.e., surface observation data), as a crucial foundation for weather forecasting, consists of various meteorological elements measured by instruments and equipment at automatic weather stations, primarily including temperature, humidity, air pressure, wind direction, wind speed, and precipitation. Because extreme weather events (such as torrential rains and typhoons) can easily damage the instruments and equipment of automatic weather stations or disrupt communication networks, surface observation data in affected areas is prone to continuous loss or anomalies during meteorological disasters. This reduces the reliability of forecast analysis results under severe weather scenarios, thereby affecting the scientific rigor and timeliness of meteorological disaster prevention and mitigation decision-making services. Therefore, in the face of severe weather, the real-time and accurate detection of abnormal stations is of paramount importance.

[0003] In related technologies, due to limitations in business performance evaluation requirements and the monitoring granularity of existing monitoring systems, provincial maintenance personnel cannot promptly and accurately detect sudden data anomalies or interruptions at several, dozens, or hundreds of stations in disaster-stricken areas. Under normal circumstances, the current reporting rate (i.e., fault tolerance rate) can meet business needs, but in disaster weather emergency scenarios, in order to support refined forecasting in disaster-stricken areas, the observation data of each regional station in the disaster-affected area is indispensable. Its absence may affect the scientific nature of disaster prevention and mitigation decision-making data and the timeliness of emergency response. Moreover, the Tianjing (i.e., meteorological integrated business real-time monitoring system) only monitors the reporting rate of the evaluation stations and does not monitor non-evaluation stations.

[0004] Furthermore, while provincial meteorological bureaus can apply to the national level to initiate a real-time interpolation process when abnormal reporting rates are detected at ground observation data files through the Tianjing full-process monitoring system, in order to promptly fill in missing data using gridded multi-source precipitation real-time products and thus reduce the impact on disaster prevention and mitigation decision-making data, the current emergency real-time interpolation process for severe weather is limited by the reporting rate indicator. The time between the start of the process and the interruption or abnormality of the observation data is usually more than 24 hours, and can be as long as 81 hours. This seriously affects the reliability of forecast analysis results under severe weather scenarios, and consequently affects the scientific nature and timeliness of meteorological disaster prevention and mitigation decision-making services. Summary of the Invention

[0005] This application provides a method and system for monitoring ground observation data anomalies in response to severe weather, which can promptly and accurately identify stations that are missing reports during severe weather, thereby improving the support capability for real-time interpolation services.

[0006] In a first aspect, embodiments of this application provide a method for monitoring anomalies in ground-based observation data for severe weather, the method comprising: The system obtains a list of information on all national automatic ground stations based on the Tianyuan interface, and obtains all hourly ground observation data records corresponding to the current business time based on the Tianqing interface. The information list of national automatic ground stations is cleaned and standardized by recording hourly ground observation data to obtain the target collection station information list corresponding to the current business time. For each target site in the target receivable site information list, if there is no target ground observation hour data entry record corresponding to the target site in all ground observation hour data entry records, then the target site is determined to be an abnormal site that has not uploaded the current business time observation data; Issue warnings for all abnormal sites.

[0007] In conjunction with the first aspect, in one embodiment, the method further includes: If there is a target ground observation hourly data entry record corresponding to the target station in all the ground observation hourly data entry records, then determine whether there is a target meteorological quality control code with a characterization error or missing measurement value in the target ground observation hourly data entry record. If it exists, the target site is determined to be an abnormal site where the feature value is unavailable; If it does not exist, the target site is determined to be a normal site.

[0008] In conjunction with the first aspect, in one implementation, after the step of determining that the target site is an abnormal site that has not uploaded the current service observation data, the method further includes: For each abnormal site, obtain the historical frequency of the same service within a preset time period for that abnormal site; If the historical reporting rate for the same service is zero, the abnormal site is determined to be a non-service application site and is removed from the list.

[0009] In conjunction with the first aspect, in one embodiment, the method further includes: The list of historical severe weather warnings within a preset time period obtained based on the Tianqing interface is parsed to obtain the target information corresponding to each warning message. The target information includes the warning release time, severe weather type code, administrative division code of the issuing unit, and warning message text. For each warning message, the time range and area range of the severe weather impact are determined based on the corresponding target information; Based on the time range of the impact of severe weather, the target early warning information that is effective in the current business time is selected from all early warning information; For each abnormal site, if the spatial location of the abnormal site is within the disaster weather impact area corresponding to the target early warning information, then the abnormal site is determined to be a site to be interpolated. An alarm is issued for the site to be interpolated in order to initiate the emergency real-time interpolation process.

[0010] In conjunction with the first aspect, in one implementation, determining the time range and area range of severe weather impact based on the corresponding target information includes: The warning message text is parsed to obtain the target parsing results; Based on the target analysis results and the severe weather type code, the end time of the severe weather is determined, and the warning issuance time is taken as the start time of the severe weather. Construct the time range of severe weather impact based on the start and end times of severe weather; The scope of the area affected by severe weather was determined based on the target analysis results and the administrative division code of the issuing unit.

[0011] In conjunction with the first aspect, in one implementation, determining the end time of severe weather based on the target parsing results and the severe weather type code includes: If the target analysis result contains specific time keywords, the end time of the severe weather corresponding to the specific time keyword is determined based on the preset mapping relationship between keywords and time points; If the target analysis result does not contain specific time keywords but contains duration keywords, then the end time of the severe weather is determined based on the duration keywords and the warning issuance time. If the target analysis result does not contain specific time keywords and duration keywords but contains date range keywords, then the end time of the severe weather is determined based on the date range keywords and the warning release time. If the target parsing result does not contain specific time keywords, duration keywords, and date range keywords, then the target duration corresponding to the severe weather type code is determined based on the preset mapping relationship between severe weather type and duration, and the end time of severe weather is determined based on the target duration and the warning release time.

[0012] In conjunction with the first aspect, in one implementation method, determining the affected area of ​​severe weather based on the target analysis results and the administrative division code of the issuing unit includes: Determine whether the administrative division code of the issuing unit is a county-level administrative division code; If so, the administrative division code of the issuing unit will be used as the scope of the area affected by severe weather; If not, then if the preset administrative division code table does not contain the target administrative division code corresponding to the keywords of the administrative division unit in the target parsing result, all county-level administrative division codes corresponding to the administrative division code of the publishing unit will be used as the scope of the area affected by the disaster weather. When the preset administrative division code table contains the target administrative division code corresponding to the keywords of the administrative division unit in the target parsing results, the scope of the area affected by the severe weather is determined based on the target administrative division code.

[0013] In conjunction with the first aspect, in one implementation, if the target administrative division code is multiple and divided into target county-level administrative division codes and target city-level administrative division codes, determining the affected area based on the target administrative division codes includes: Determine whether there exists a county-level administrative division code in the target county-level administrative division code that corresponds to the target city-level administrative division code; If they exist, then all target county-level administrative division codes will be considered as the affected area of ​​severe weather; If it does not exist, then all county-level administrative division codes corresponding to the target municipal administrative division code and all target county-level administrative division codes will be considered as the area affected by severe weather.

[0014] In conjunction with the first aspect, in one embodiment, after the step of determining that the abnormal site is a site to be interpolated, the method further includes: If there are multiple interpolation sites located in the same province, the density-based clustering algorithm DBSCAN is used to calculate the density of the multiple interpolation sites located in the same province in order to generate the target cluster. Issue an alarm for the target cluster; The neighborhood radius and minimum number of points of the DBSCAN algorithm are determined according to the density level zone where the province is located. Different density level zones have different preset neighborhood radii and preset minimum number of points.

[0015] Secondly, embodiments of this application provide a ground observation data anomaly monitoring system for severe weather, the ground observation data anomaly monitoring system for severe weather including: The data acquisition layer is used to obtain a list of information on automatic ground stations nationwide based on the Tianyuan interface, and to obtain all hourly ground observation data records corresponding to the current business time based on the Tianqing interface. The processing layer is used to clean and standardize the national list of automatic ground stations by using the ground observation hourly data entry records to obtain the target receivable station information list corresponding to the current business time. For each target station in the target receivable station information list, if there is no target ground observation hourly data entry record corresponding to the target station in all the ground observation hourly data entry records, the target station is determined to be an abnormal station that has not uploaded the observation data for the current business time. The presentation layer is used to issue alerts for all abnormal sites.

[0016] In conjunction with the second aspect, in one embodiment, the processing layer is further configured to: If there is a target ground observation hourly data entry record corresponding to the target station in all the ground observation hourly data entry records, then determine whether there is a target meteorological quality control code with a characterization error or missing measurement value in the target ground observation hourly data entry record. If it exists, the target site is determined to be an abnormal site where the feature value is unavailable; If it does not exist, the target site is determined to be a normal site.

[0017] In conjunction with the second aspect, in one embodiment, the processing layer is further configured to: For each abnormal site, obtain the historical frequency of the same service within a preset time period for that abnormal site; If the historical reporting rate for the same service is zero, the abnormal site is determined to be a non-service application site and is removed from the list.

[0018] In conjunction with the second aspect, in one embodiment, the processing layer is further configured to: The list of historical severe weather warnings within a preset time period obtained based on the Tianqing interface is parsed to obtain the target information corresponding to each warning message. The target information includes the warning release time, severe weather type code, administrative division code of the issuing unit, and warning message text. For each warning message, the time range and area range of the severe weather impact are determined based on the corresponding target information; Based on the time range of the impact of severe weather, the target early warning information that is effective in the current business time is selected from all early warning information; For each abnormal site, if the spatial location of the abnormal site is within the disaster weather impact area corresponding to the target early warning information, then the abnormal site is determined to be a site to be interpolated. An alarm is issued for the site to be interpolated in order to initiate the emergency real-time interpolation process.

[0019] In conjunction with the second aspect, in one implementation, the processing layer is further configured to: The warning message text is parsed to obtain the target parsing results; Based on the target analysis results and the severe weather type code, the end time of the severe weather is determined, and the warning issuance time is taken as the start time of the severe weather. Construct the time range of severe weather impact based on the start and end times of severe weather; The scope of the area affected by severe weather was determined based on the target analysis results and the administrative division code of the issuing unit.

[0020] In conjunction with the second aspect, in one implementation, the processing layer is further configured to: If the target analysis result contains specific time keywords, the end time of the severe weather corresponding to the specific time keyword is determined based on the preset mapping relationship between keywords and time points; If the target analysis result does not contain specific time keywords but contains duration keywords, then the end time of the severe weather is determined based on the duration keywords and the warning issuance time. If the target analysis result does not contain specific time keywords and duration keywords but contains date range keywords, then the end time of the severe weather is determined based on the date range keywords and the warning release time. If the target parsing result does not contain specific time keywords, duration keywords, and date range keywords, then the target duration corresponding to the severe weather type code is determined based on the preset mapping relationship between severe weather type and duration, and the end time of severe weather is determined based on the target duration and the warning release time.

[0021] In conjunction with the second aspect, in one implementation, the processing layer is further configured to: Determine whether the administrative division code of the issuing unit is a county-level administrative division code; If so, the administrative division code of the issuing unit will be used as the scope of the area affected by severe weather; If not, then if the preset administrative division code table does not contain the target administrative division code corresponding to the keywords of the administrative division unit in the target parsing result, all county-level administrative division codes corresponding to the administrative division code of the publishing unit will be used as the scope of the area affected by the disaster weather. When the preset administrative division code table contains the target administrative division code corresponding to the keywords of the administrative division unit in the target parsing results, the scope of the area affected by the severe weather is determined based on the target administrative division code.

[0022] In conjunction with the second aspect, in one implementation, if the target administrative division code is multiple and divided into target county-level administrative division codes and target city-level administrative division codes, the processing layer is further configured to: Determine whether there exists a county-level administrative division code in the target county-level administrative division code that corresponds to the target city-level administrative division code; If they exist, then all target county-level administrative division codes will be considered as the affected area of ​​severe weather; If it does not exist, then all county-level administrative division codes corresponding to the target municipal administrative division code and all target county-level administrative division codes will be considered as the area affected by severe weather.

[0023] In conjunction with the second aspect, in one embodiment, the processing layer is further configured to: If there are multiple interpolation sites located in the same province, the density-based clustering algorithm DBSCAN is used to calculate the density of the multiple interpolation sites located in the same province in order to generate the target cluster. Issue an alarm for the target cluster; The neighborhood radius and minimum number of points of the DBSCAN algorithm are determined according to the density level zone where the province is located. Different density level zones have different preset neighborhood radii and preset minimum number of points.

[0024] The beneficial effects of the technical solutions provided in this application include: The system obtains the national list of automatic ground stations and all hourly ground observation data records corresponding to the current business session through the Tianyuan and Tianqing interfaces, respectively. Based on the hourly ground observation data records, the national list of automatic ground stations is cleaned and standardized to obtain the target collection site information list corresponding to the current business session. Since the national list of automatic ground stations collected by the Tianyuan system includes both assessment stations and non-assessment stations, the target collection site information list must also include both assessment stations and non-assessment stations. Then, for each target site in the target collection site information list, if there is no corresponding target ground observation hourly data record in all hourly ground observation data records, the target site is determined to be an abnormal site that has not uploaded observation data for the current business session, and an alert is issued for all abnormal sites. As can be seen, this application monitors all automatic weather stations and promptly and accurately identifies abnormal stations during severe weather by checking for the presence or absence of data entry records, regardless of the reporting rate. That is, if there is no record of ground observation hourly data for the current operational period for a certain station, it means that the station has not uploaded the ground observation hourly data for the current operational period, and the station is considered a missing reporting station. This effectively improves the timeliness and accuracy of abnormal station detection, thereby enhancing the support capability for real-time interpolation services, enabling maintenance personnel to promptly initiate emergency real-time interpolation procedures for abnormal stations. Attached Figure Description

[0025] Figure 1This is a flowchart illustrating an embodiment of the ground observation data anomaly monitoring method for severe weather according to this application. Figure 2 This is a schematic diagram of the architecture of a ground observation data anomaly monitoring system for severe weather, as described in the embodiments of this application. Figure 3 This is a flowchart illustrating the data anomaly interruption detection algorithm involved in the embodiments of this application; Figure 4 This is a flowchart illustrating the spatiotemporal range parsing algorithm for disaster weather early warning information involved in the embodiments of this application; Figure 5 This is a schematic diagram of the distance matrix of abnormal sites involved in the embodiments of this application. Detailed Implementation

[0026] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0027] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0028] In a first aspect, embodiments of this application provide a method for monitoring anomalies in ground observation data for severe weather.

[0029] In one embodiment, reference is made to Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the ground-based observation data anomaly monitoring method for severe weather according to this application. Figure 1 As shown, methods for monitoring anomalies in ground-based observation data for severe weather include: Step S10: Obtain the list of information on national automatic ground stations based on the Tianyuan interface, and obtain all ground observation hourly data entry records corresponding to the current business time based on the Tianqing interface.

[0030] As an example, it should be noted that the China Meteorological Administration has now established a comprehensive meteorological observation system covering land, sea, and air. The surface meteorological observation network is responsible for observing and measuring near-surface meteorological elements and weather phenomena, providing crucial foundational data for weather forecasting, climate analysis, scientific research, and meteorological services. The surface meteorological observation network consists of national-level meteorological observation stations (referred to as "national stations") and provincial-level meteorological observation stations (referred to as "regional stations"). National stations are established and planned uniformly by the China Meteorological Administration, providing long-term, continuous, and stable observational data for national and global weather and climate analysis, forecasting, and climate change monitoring. Regional stations, on the other hand, are constructed by each province and serve as an important supplement to the national weather observation network.

[0031] It is worth noting that the operation monitoring, maintenance, equipment management, and quality control of the entire ground meteorological observation station network are handled by the integrated meteorological observation business operation information platform (hereinafter referred to as: Tianyuan). Among them, the ground meteorological observation stations collect meteorological elements in accordance with the observation specifications, encode them in accordance with the data format specifications, and then form files according to the transmission and exchange specifications and upload them to the provincial level.

[0032] Currently, the surface meteorological observation data files compiled and issued by surface meteorological observation stations include hourly observation data from national stations in China, minutely observation data from national stations in China, hourly observation data from unmanned national stations, and hourly observation data from regional stations. The transmission, quality control, decoding, and service of these surface meteorological observation data files are handled by the Meteorological Big Data Cloud Platform (referred to as "Tianqing"), while the arrival rate monitoring for each stage of the entire process is handled by the Meteorological Integrated Business Real-time Monitoring System (referred to as "Tianjing"). Tianqing is a unified national and provincial data processing platform built by the China Meteorological Administration based on a meteorological "private cloud + public cloud" architecture. It consists of an exchange and quality control system, a product processing system, a data mining and analysis system, and a storage and service system. Meteorological observation data files are transmitted from the exchange and quality control system to the national and provincial levels, then decoded and stored in the national and provincial Tianqing storage and service systems, and finally provided to the national and provincial meteorological business systems through standardized interfaces.

[0033] Tianjing is a nationally and provincially unified real-time meteorological business monitoring system covering "all businesses, all processes, and all elements." It consists of a comprehensive monitoring subsystem, a centralized alarm subsystem, a large-screen display subsystem, an operation and maintenance management subsystem, an operation service subsystem, and a monitoring interface subsystem. Based on the log information sent by various systems of Tianqing, Tianjing monitors and alarms the arrival rate of ground meteorological observation data files for each business period at each stage of collection, distribution, storage, and synchronization, ensuring the efficiency and stability of the entire data link. At the same time, Tianjing will determine the list of receiving stations and the arrival rate alarm threshold for different ground meteorological observation data files in accordance with the assessment requirements of the China Meteorological Administration.

[0034] Therefore, national and provincial meteorological operational personnel can detect data interruptions from automatic weather stations by monitoring the arrival rate of ground meteorological observation data files through the "Sky Mirror" full-process monitoring system. Specifically, when extreme weather events such as heavy rain or typhoons occur and the hourly arrival rate of national or regional station observation data is abnormal, provincial meteorological departments will conduct a comprehensive assessment to determine whether to apply to the national level to activate the real-time interpolation process as an emergency measure to ensure the accuracy and timeliness of refined forecasts for disaster-stricken areas. If the real-time interpolation process is activated, the provincial level needs to manually collect the information from the automatic weather stations to be interpolated, mainly including station number, latitude and longitude, altitude, interpolated meteorological elements, and time range, and then submit it to the national meteorological information center. However, due to limitations in the reporting rate, provincial maintenance personnel are unable to promptly and accurately detect sudden anomalies or interruptions in data from several, dozens, or even hundreds of stations in disaster-stricken areas, resulting in severe delays in initiating the emergency real-time interpolation process. Furthermore, the full-process monitoring function of the Tianjing system only includes the assessment stations in the list of stations to be monitored, leaving tens of thousands of non-assessment area stations unmonitored. This seriously affects the reliability of forecast analysis results under severe weather scenarios, thereby impacting the scientific rigor and timeliness of meteorological disaster prevention and mitigation decision-making services.

[0035] It is evident that the Tianjing system's full-process monitoring function, responsible for monitoring and alarming the reporting rate of core business processes such as collection, distribution, decoding, storage, and synchronization of observation data files from all automatic meteorological stations at both the national and provincial levels, meets the needs of normal meteorological operations. However, it cannot meet the monitoring and alarm requirements for abnormal interruptions in automatic meteorological station observation data during meteorological disaster scenarios. Therefore, there is an urgent need for a method for monitoring abnormal interruptions in ground observation data from automatic meteorological stations during severe weather. This method would enable automated real-time monitoring and alarming of abnormal interruptions in ground observation data during severe weather, improve support capabilities for real-time interpolation operations, and enhance the initiative and scientific nature of decision-making in emergency scenarios.

[0036] Understandably, Tianyuan manages all automatic weather stations in the entire ground meteorological observation network. Therefore, its nationwide list of automatic weather station information includes not only assessment stations but also non-assessment stations. Furthermore, all automatic weather station ground meteorological observation data files are transmitted to Tianqing, which decodes and stores the received data to create hourly ground observation data records. Based on this, to achieve comprehensive monitoring of all automatic weather stations, see [link to relevant documentation]. Figure 2 and Figure 3As shown, in this embodiment, the acquisition layer will obtain the national list of automatic ground stations and all ground observation hourly data entry records corresponding to the current operational time through the meteorological automatic station information acquisition interface in the Tianyuan interface and the ground observation hourly data entry record acquisition interface in the Tianqing interface, respectively, providing a data foundation for comprehensive monitoring of abnormal stations. For example, at 10:10 on the hour, the national list of automatic ground stations is obtained from the Tianyuan interface, which mainly includes the station number, station name, station latitude and longitude, station altitude, station level, county name, county-level administrative division code, city name, city-level administrative division code, province name, and provincial-level administrative division code, etc.; and at 30:00 on the hour, all ground observation hourly data entry records for the current operational time are obtained from the Tianqing interface, which mainly include the station number, station level, station latitude and longitude, station altitude, operational time, meteorological element value, meteorological element quality control code, county name, county-level administrative division code, city name, and province name, etc.

[0037] Step S20: Clean and standardize the national list of automatic ground stations by recording hourly ground observation data to obtain the target collection station information list corresponding to the current business time.

[0038] Exemplary, see Figure 3 As shown, this embodiment will clean and standardize the national list of automatic ground stations to ensure that the data sets of receivables and actual receivables are completely consistent in key fields, and finally generate a complete and correct list of receivables at the current business time (i.e., the target list of receivables). Specifically, the following rules can be used to clean and standardize the national list of automatic ground stations: 1) If the administrative division information of a certain station in the national list of automatic ground stations is missing, it shall be supplemented according to the information in the national administrative division code table.

[0039] 2) If the administrative division information of a certain station in the national list of automatic ground stations is missing, the administrative division information of that station will be obtained from the ground observation hourly data entry record to complete the information.

[0040] 3) If the administrative division information of a certain station in the national list of automatic ground stations is missing and cannot be filled in, then the station will be removed.

[0041] 4) If a station number in the hourly ground observation data entry record is not in the national list of automatic ground stations, then the station number and related information shall be added to the national list of automatic ground stations.

[0042] 5) If a station number in the hourly ground observation data entry record is in the national list of automatic ground stations, then the core fields of that station number, such as latitude and longitude, altitude, and administrative division code, will be verified. If they are inconsistent, then the information in the national list of automatic ground stations will be replaced with the information from the hourly ground observation data entry record.

[0043] Step S30: For each target site in the target receivable site information list, if there is no target ground observation hour data entry record corresponding to the target site in all ground observation hour data entry records, then the target site is determined to be an abnormal site that has not uploaded the current business time observation data.

[0044] As an example, it is understandable that for each target site in the target receivable site information list, it is necessary to confirm whether there is a target ground observation hour data entry record corresponding to the target site in all ground observation hour data entry records for the current business period. That is, whether there is a ground observation hour data entry record containing a station number that is the same as the station number of the target site. If it exists, it means that the ground observation hour data of the target site in the current business period has been uploaded to the national-level SkyEngine. If it does not exist, it means that the ground observation hour data of the target site in the current business period has not been uploaded to the national-level SkyEngine. In this case, it needs to be listed as a missing reporting station, that is, it is determined to be an abnormal station that has not uploaded the observation data for the current business period.

[0045] Furthermore, in one embodiment, after determining that the target site is an abnormal site that has not uploaded the current service observation data, the method further includes: For each abnormal site, obtain the historical frequency of the same service within a preset time period for that abnormal site; If the historical reporting rate for the same service is zero, the abnormal site is determined to be a non-service application site and is removed from the list.

[0046] As an example, it should be noted that the specific value of the preset duration can be determined according to actual needs, and there can be one or two preset durations; this is not limited here. For example, the preset duration can be set to 7 days and 24 hours. Based on this, see [link to relevant documentation]. Figure 3 As shown, for stations with missing reports, the reporting rate for the same service time within the past 7 days and the reporting rate for the same service time within the past 24 hours will be statistically analyzed. If both are 0, the station with missing reports will be determined to be a non-service application station such as an emergency station or a test station. That is, although its station information is included in the unified management of Tianyuan, the station's observation data has not yet been officially uploaded to the national and provincial Tianqing systems. Therefore, it will be removed so that these stations will not be included in the subsequent alarm, interpolation station list and density detection.

[0047] Furthermore, in one embodiment, the method further includes: If there is a target ground observation hourly data entry record corresponding to the target station in all the ground observation hourly data entry records, then determine whether there is a target meteorological quality control code with a characterization error or missing measurement value in the target ground observation hourly data entry record. If it exists, the target site is determined to be an abnormal site where the feature value is unavailable; If it does not exist, the target site is determined to be a normal site.

[0048] As an example, it is worth noting that the current Tianjing system's full-process monitoring function is based on log information sent by various systems of the Tianjing system. This means it can only detect stations with missing reports, but cannot detect abnormal meteorological element values ​​caused by malfunctions in automatic weather station observation equipment during severe weather. However, this embodiment will implement the detection of abnormal meteorological element values ​​for all stations that are not missing reports, thereby improving the reliability of forecast analysis results under severe weather scenarios.

[0049] It is understandable that when each automatic weather station uploads its hourly ground observation data to the provincial meteorological database, the provincial database will perform rapid quality control on the received hourly ground observation data and mark it with a quality control code. The numbers and meanings of the quality control codes are shown in Table 1 below: Table 1. Definition of Quality Control Code

[0050] It should be noted that Table 1 above is only a presentation of an embodiment, and the contents of Table 1 can be adapted to meet actual needs, without limitation.

[0051] Based on this, see Figure 3 As shown, if a target ground observation hourly data entry record containing the target station number can be found among all the ground observation hourly data entry records corresponding to the current operational period, it indicates that the target station's ground observation hourly data for the current operational period has been uploaded to the national-level meteorological database. Then, check whether the quality control codes corresponding to the precipitation, temperature, wind speed, wind direction, relative humidity, and air pressure values ​​in the target ground observation hourly data entry record are the target meteorological requirement quality control codes (i.e., 2 and 8) representing incorrect or missing values. If they exist, for example, the quality control code for precipitation is 2 or the quality control code for temperature is 8, it means that although the target station has uploaded the ground observation hourly data for the current operational period to the national-level meteorological database, its element values ​​are unusable and should be listed as an abnormal station, i.e., the target station is determined to be an abnormal station with unusable element values. If they do not exist, it means that the target station has not only uploaded the ground observation hourly data for the current operational period to the national-level meteorological database, but all the uploaded ground observation hourly data for the current operational period is normal, and the target station is determined to be a normal station.

[0052] Step S40: Issue an alert for all abnormal sites.

[0053] As an example, in this embodiment, an abnormal station information list is generated based on all abnormal stations (i.e., stations with missing reports and / or stations with abnormal elements). This list may include data type, service time, station number, station name, station latitude and longitude, station altitude, station level, county name, county-level administrative division code, city name, province name, abnormal meteorological element name, etc., and a visual warning is given to all abnormal stations through the display layer.

[0054] As can be seen, this embodiment monitors all automatic weather stations and promptly and accurately identifies abnormal stations during severe weather by checking for the presence or absence of data entry records, regardless of the reporting rate. That is, if there is no record of ground observation hourly data for the current operational period for a certain station, it means that the station has not uploaded the ground observation hourly data for the current operational period, and the station is considered a station with missing reports. This effectively improves the timeliness and accuracy of detecting abnormal stations, thereby enhancing the support capability for real-time interpolation services, enabling maintenance personnel to promptly initiate emergency real-time interpolation procedures for abnormal stations.

[0055] Based on this, this embodiment provides an algorithm for detecting abnormal interruptions in ground observation data from automatic weather stations. Its core lies in constructing a closed-loop monitoring process for interruptions and element anomalies in ground observation data from automatic weather stations through multi-source data fusion, dynamic cleaning and verification, and multi-rule identification. This process can automatically and accurately identify operational application sites where data has not been uploaded (missing reports) and where data is invalid (element anomalies). (1) Dynamic site information cleaning and synchronization rules: A set of multi-level data standardization rules and a closed-loop algorithm design of “acquisition-cleaning-identification-screening” are established. The core is to use real-time observation data and administrative division code table as authoritative sources to dynamically verify and update the core geographical and administrative division information (such as latitude and longitude, altitude, and division code) in the list of ground automatic stations obtained from Tianyuan, and to achieve bidirectional completion of missing information to ensure the accuracy and timeliness of the benchmark list (i.e., the target list of receiving site information).

[0056] (2) Dual anomaly detection mechanism: First, the station number is compared to identify the station that is completely missing; then, for the data that has been reported, the station with abnormal elements is identified based on the national quality control code (especially the code value with incorrect identification or missing test), thus realizing comprehensive monitoring of "whether there is data" and "the quality of data".

[0057] (3) Intelligent elimination of non-business application sites based on historical reporting rate: By analyzing the recent reporting rate of missing reporting sites (simultaneous times in the past 7 days and in the past 24 hours), non-business application sites such as emergency stations and test stations that have no data for a long time are automatically identified, avoiding interference in subsequent alarms and statistics, and improving the targeting of business monitoring.

[0058] This embodiment effectively realizes the automated diagnosis of ground observation data interruptions and element anomalies from automatic weather stations nationwide, significantly improving the efficiency and accuracy of meteorological data business monitoring.

[0059] Furthermore, in one embodiment, the method further includes: The list of historical severe weather warnings within a preset time period obtained based on the Tianqing interface is parsed to obtain the target information corresponding to each warning message. The target information includes the warning release time, severe weather type code, administrative division code of the issuing unit, and warning message text. For each warning message, the time range and area range of the severe weather impact are determined based on the corresponding target information; Based on the time range of the impact of severe weather, the target early warning information that is effective in the current business time is selected from all early warning information; For each abnormal site, if the spatial location of the abnormal site is within the disaster weather impact area corresponding to the target early warning information, then the abnormal site is determined to be a site to be interpolated. An alarm is issued for the site to be interpolated in order to initiate the emergency real-time interpolation process.

[0060] As an example, it should be noted that the specific value of the preset duration can be determined according to actual needs and is not limited here. For example, the preset duration can be set to 24 hours; in this embodiment, see... Figure 4As shown, the system obtains a list of severe weather warnings issued by provinces, cities, and counties in the past 24 hours (i.e., a historical list of severe weather warnings) from the national disaster warning information collection interface in the Tianqing interface. This list is then parsed in HTML (Hypertext Markup Language) format to obtain the warning information text and related descriptive information (such as ID number, warning release time, severe weather type code, warning level, issuing unit name, and issuing unit administrative division code) for each warning in the historical severe weather warning information list. The target information includes the ID number, warning release time, severe weather type code, warning level, issuing unit name, issuing unit administrative division code, and warning information text. For each warning, the corresponding warning information text is parsed to determine the time range and area range of severe weather impact based on the parsing results and the target information. Furthermore, the national administrative division code table can be obtained through the administrative division collection interface for cleaning and standardizing the national automatic weather station information list and determining the area range of severe weather impact.

[0061] Then, based on the time range of severe weather impact, the target warning information that is effective in the current business time is selected from all warning information. That is, the severe weather impact time range corresponding to each warning information is superimposed with the current business time to determine the effective target warning information. For example, if the current business time is 18:00 on January 2, 2026, and the severe weather impact time range of warning information X1 is from 8:00 on January 1, 2026 to 15:00 on January 3, 2026, then warning information X1 is effective in the current business time, and warning information X1 is taken as the target warning information. If the severe weather impact time range of warning information X2 is from 8:00 on January 1, 2026 to 10:00 on January 2, 2026, then warning information X2 is not effective in the current business time, and warning information X2 is not taken as the target warning information.

[0062] For each abnormal station, determine whether the spatial location of the abnormal station is within the disaster weather impact area corresponding to the target warning information. If so, the abnormal station is determined to be a station to be interpolated; otherwise, the abnormal station is determined not to be a station to be interpolated. It should be noted that the spatial location can be represented by the latitude and longitude of the station or by the administrative division code of the station, which is not limited here.

[0063] After identifying all sites to be interpolated, a list of interpolated sites can be generated and displayed through a presentation layer to generate alerts for the sites to be interpolated. This allows maintenance personnel to initiate the emergency real-time interpolation process as soon as possible, effectively improving the efficiency and timeliness of the emergency real-time interpolation process and thus enhancing the reliability of forecast analysis results in severe weather scenarios.

[0064] Furthermore, in one embodiment, determining the time range and area range of severe weather impact based on the corresponding target information includes: The warning message text is parsed to obtain the target parsing results; Based on the target analysis results and the severe weather type code, the end time of the severe weather is determined, and the warning issuance time is taken as the start time of the severe weather. Construct the time range of severe weather impact based on the start and end times of severe weather; The scope of the area affected by severe weather was determined based on the target analysis results and the administrative division code of the issuing unit.

[0065] In this exemplary embodiment, for each warning message, the warning release time obtained from parsing the HTML format is used as the start time of the severe weather. Simultaneously, the warning message text is parsed to obtain the target parsing result. For example, parsing the warning message text corresponding to warning message X1 yields the target parsing result: "Y County Meteorological Observatory issued a yellow thunderstorm warning signal at 13:34 on August 20, 2025: Thunderstorm activity is expected in townships Y1, Y2, and Y3 of our county within the next 6 hours, with the possibility of strong winds, hail, and short-term heavy rainfall in some areas. Please take precautions against lightning." Based on this, the end time of the severe weather can be determined according to the target parsing result and the severe weather type code. Then, the time range of the severe weather's impact can be determined by using the start and end times of the severe weather. Furthermore, the area affected by the severe weather can also be determined by using the target parsing result and the administrative division code of the issuing unit. Based on this, a warning information analysis record can be generated, which preferably includes the severe weather type code, warning level, time range of severe weather impact (i.e., warning release time and end time), and area range of severe weather impact (such as a list of county-level administrative division codes).

[0066] Furthermore, in one embodiment, determining the end time of severe weather based on the target parsing result and the severe weather type code includes: If the target analysis result contains specific time keywords, the end time of the severe weather corresponding to the specific time keyword is determined based on the preset mapping relationship between keywords and time points; If the target analysis result does not contain specific time keywords but contains duration keywords, then the end time of the severe weather is determined based on the duration keywords and the warning issuance time. If the target analysis result does not contain specific time keywords and duration keywords but contains date range keywords, then the end time of the severe weather is determined based on the date range keywords and the warning release time. If the target parsing result does not contain specific time keywords, duration keywords, and date range keywords, then the target duration corresponding to the severe weather type code is determined based on the preset mapping relationship between severe weather type and duration, and the end time of severe weather is determined based on the target duration and the warning release time.

[0067] Exemplary, see Figure 4 As shown, to determine the end time of severe weather, firstly, specific time keywords (such as "morning / noon / afternoon / evening / night / nighttime") are searched in the target parsing results corresponding to the warning information text. If these keywords are present in the target parsing results, the time information is extracted from the target parsing results using a time representation pattern described by a regular expression. Then, the end time of severe weather is generated based on the mapping relationship between keywords and time points. It should be noted that the preferred mapping relationship between keywords and time points is: 'early morning': 2 AM, 'morning': 12 PM, 'noon': 2 PM, 'afternoon': 6 PM, 'evening': 9 PM, 'nighttime': 11 PM, and 'nighttime': 11 PM. The specific time representation patterns include the following four types: Pattern 1: Relative date + keywords Regular expression: '(Today|Tomorrow)(Morning|Noon|Afternoon|Evening|Night|Night)' Matching examples: "this morning", "tomorrow night" For example: The target analysis result is "Z County Meteorological Observatory continued to issue an orange rainstorm warning signal at 21:31 on August 19, 2025: It is expected that Z1 Town, Z2 Town, Z3 Town and Z4 District of our county will be affected by rainstorm weather tonight, with hourly precipitation reaching 70-100 mm and cumulative precipitation of 100-150 mm. Please take precautions." Based on this, it contains the keyword "night", so the time description information extracted is "tonight", and therefore the severe weather end time is generated as "23:00:00 on August 19, 2025".

[0068] Pattern 2: Specific date + keywords to keywords Regular expression: '({1,2})Month({1,2})Day(AM|Noon|Afternoon|Evening|Night|Night)[To|To](AM|Noon|Afternoon|Evening|Night|Night)' Matching example: "October 5th, morning to afternoon" Pattern 3: Specific date + keywords Regular expression: '({1,2})month({1,2})day(morning|noon|afternoon|evening|night|night)' Matching example: "October 5th afternoon" Pattern 4: Relative Time + Keyword to Keyword Regular expression: 'Today (morning|noon|afternoon|evening|night|night)[to|until](morning|noon|afternoon|evening|night|night)' Matching examples: "this morning to afternoon", "this night to tomorrow morning" If the target parsing result does not contain specific time keywords, then the result is searched for duration keywords (such as "expected / future / continued"). If the target parsing result contains these keywords, the time information is extracted from the target parsing result using a time representation pattern described by a regular expression, and then the end time of the severe weather is generated according to the duration rules. The duration representation pattern has the following two types: Pattern 1: Keyword + Duration Regular expression: '(future|expected|will|continue)(+)(\d+)[small]?[hour]?' Matching example: "the next 3 hours" Pattern 2: Keywords + Duration Range Regular expression: '(future|expected|will|continue)(\d+)(+)[to\-](\d+)[hours]?[hours]' Matching examples: "in the next 3-5 hours", "expected 4 to 6 hours" The duration rules are as follows: Rule 1: Single duration Processing: Directly extract numbers to calculate the end time of severe weather. Example: "Next 3 hours" → End time = Warning issuance time + 3 hours For example: The target analysis result is "Y County Meteorological Observatory issued a yellow thunderstorm warning signal at 13:34 on August 20, 2025: Thunderstorm activity is expected in townships Y1, Y2, and Y3 of our county in the next 6 hours, and localized severe convective weather such as strong winds, hail, and short-term heavy rainfall may occur. Please take precautions against lightning." It contains the keywords "expected" and "future". The extracted time description information is "expected in the next 6 hours". According to Rule 1, the end time of the severe weather is generated as "19:34:00 on August 20, 2025".

[0069] Rule 2: Duration is a time range Processing: Take the maximum value for calculating the end time of the disaster weather Example: "In the next 3 - 5 hours" → End time = Warning release time + 5 hours Rule 3: Multiple durations Processing: Add up the times.

[0070] Example: "Heavy rain in the next 3 hours, which will last for 2 hours" → End time = Warning release time + 3 hours + 2 hours If the target parsing result does not include the duration keyword, extract the time information from the target parsing result using the date range expression pattern described by the regular expression, and then generate the end time of the disaster weather; among which, Regular expression: '(\d+)[~~\-](\d+) days' Matching examples: "1~3 days", "5 - 7 days", "10~12 days" If no time information is extracted from the target parsing result, estimate the end time of the disaster weather according to the mapping relationship between the disaster weather type and the duration; among which, the mapping relationship between the disaster weather type and the duration is shown in Table 2 below.

[0071] Table 2 Relationship table between disaster weather types and durations

[0072] It should be noted that Table 2 above is only a presentation of the embodiments, and the contents in Table 2 can also be adaptively adjusted according to actual needs, which are not limited here.

[0073] It can be understood that each disaster weather type has a unique code corresponding to it (i.e., the disaster weather type code), so the disaster weather type can be determined through the disaster weather type code; based on this, the target disaster weather type can be determined according to the disaster weather type code in the target information, and then the target duration corresponding to the target disaster weather type can be determined according to the relationship in Table 2; then adding the target duration to the warning release time can obtain the end time of the disaster weather.

[0074] Furthermore, in one embodiment, the determining the disaster weather affected area range based on the target parsing result and the administrative division code of the issuing unit includes: Judge whether the administrative division code of the issuing unit is a county-level administrative division code; If so, use the administrative division code of the issuing unit as the disaster weather affected area range; If not, then if the preset administrative division code table does not contain the target administrative division code corresponding to the keywords of the administrative division unit in the target parsing result, all county-level administrative division codes corresponding to the administrative division code of the publishing unit will be used as the scope of the area affected by the disaster weather. When the preset administrative division code table contains the target administrative division code corresponding to the keywords of the administrative division unit in the target parsing results, the scope of the area affected by the severe weather is determined based on the target administrative division code.

[0075] As an example, in this embodiment, the county-level administrative region affected by the severe weather will be determined based on the issuing unit of the early warning information, and, if necessary, in conjunction with the text parsing of the early warning information, will be represented using the county-level administrative division code. For details, see [link to documentation]. Figure 4 As shown, first confirm whether the administrative division code of the issuing unit of the warning information is a county-level administrative division code. If so, use the administrative division code of the issuing unit to represent the area affected by the severe weather. For example, if the target analysis result is "K County Meteorological Observatory issued a yellow warning signal for thunderstorms and strong winds at 21:26 on August 19, 2025: It is expected that all townships in our county will experience thunderstorms and strong winds in the next 6 hours, with gusts reaching level 8-10. Please take precautions.", the issuing unit of the warning information is "K County Meteorological Observatory", and its corresponding administrative division code is "PPPPPP". Therefore, the area affected by the severe weather in this warning information is PPPPPP (K County).

[0076] If the administrative division code of the issuing unit is not a county-level administrative division code, then the administrative division unit keywords (such as "city / county") are searched in the target parsing results to extract the text information corresponding to the administrative division unit keywords (such as "**city / **county" text information). Then, the corresponding administrative division code is searched in the administrative division code table, and it is determined whether the number of successful matches is greater than or equal to 1. If not, it means that the administrative division code table does not contain the target administrative division code corresponding to "**city / **county", and all county-level administrative division codes corresponding to the issuing unit's administrative division code are taken as the scope of the area affected by severe weather. If an administrative code is matched, it means that the administrative code table contains the target administrative division code corresponding to "**city / **county", and then it is necessary to further determine the scope of the area affected by severe weather based on whether the code represented by the matched target administrative division code is a city-level or county-level code.

[0077] Furthermore, in one embodiment, if the target administrative division code is multiple and divided into target county-level administrative division codes and target city-level administrative division codes, determining the affected area based on the target administrative division codes includes: Determine whether there exists a county-level administrative division code in the target county-level administrative division code that corresponds to the target city-level administrative division code; If they exist, then all target county-level administrative division codes will be considered as the affected area of ​​severe weather; If it does not exist, then all county-level administrative division codes corresponding to the target municipal administrative division code and all target county-level administrative division codes will be considered as the area affected by severe weather.

[0078] As an example, in this embodiment, when a target administrative division code corresponding to the keyword of the administrative division unit is matched in the administrative division code table, if there is only one target administrative division code, it is determined whether the target administrative division code is a municipal administrative division code. If not, it means that the target administrative division code is a target county-level administrative division code, and the target administrative division code (i.e., the target county-level administrative division code) is directly used as the scope of the area affected by the severe weather. If so, all county-level administrative division codes corresponding to the target administrative division code (i.e., the target municipal administrative division code) are used as the scope of the area affected by the severe weather.

[0079] If there are multiple target administrative division codes, and one of them contains both county-level administrative division codes (i.e., target county-level administrative division codes) and city-level administrative division codes (i.e., target city-level administrative division codes), then for the target city-level administrative division code, it is determined whether all matched target county-level administrative division codes contain a county-level administrative division code corresponding to that target city-level administrative division code. If so, all target county-level administrative division codes are directly considered as the affected area of ​​severe weather, and the target city-level administrative division code is excluded. If the target analysis result is "V City Meteorological Observatory continued to issue an orange rainstorm warning signal at 21:27 on August 19, 2025: It is expected that the rainfall in the following areas will reach more than 50 mm in the next 3 hours: southeast of V1 area, eastern V2 area, V3 county, V4 county, and V5 county. Please take precautions.", and the issuing unit is "V City Meteorological Observatory", and the target analysis result extracts "V3 county, V4 county, and V5 county", the corresponding county-level administrative division codes are "SSSSSS, RRRRRR, and UUUUUU", then the disaster weather impact area corresponding to this warning information is "SSSSSS, RRRRRR, and UUUUUU".

[0080] If not included, then all county-level administrative division codes corresponding to the target municipal administrative division code will be added, and all the newly added county-level administrative division codes and all the previously matched target county-level administrative division codes will be used as the scope of the area affected by severe weather.

[0081] Based on this, this embodiment proposes a disaster weather spatiotemporal range parsing algorithm for meteorological disaster early warning information. Its core lies in automatically and intelligently extracting standardized, machine-readable disaster spatiotemporal ranges from unstructured early warning information through an integrated "time-space" multi-level, multi-strategy text parsing method. (1) Step-by-step time analysis: A four-level strategy is adopted to accurately extract the disaster end time from the warning information text, namely, extraction based on specific time keywords, calculation based on duration keywords and rules, extraction based on date range patterns, and estimation based on disaster weather type-duration mapping, to ensure coverage of various time expressions.

[0082] (2) Adaptive spatial resolution: A two-level strategy is adopted to determine the precise county-level administrative region scope affected by severe weather. First, the code of the issuing unit is judged; if the county-level condition is not met, the city and county names are extracted from the text and matched with the administrative division code. A backup logic is designed to automatically associate the codes of all subordinate county-level codes when a city-level code is matched, so as to ensure the integrity and accuracy of the regional scope.

[0083] Furthermore, in one embodiment, after the step of determining that the abnormal site is a site to be interpolated, the method further includes: If there are multiple interpolation sites located in the same province, the density-based clustering algorithm DBSCAN is used to calculate the density of the multiple interpolation sites located in the same province in order to generate the target cluster. Issue an alarm for the target cluster; The neighborhood radius and minimum number of points of the DBSCAN algorithm are determined according to the density level zone where the province is located. Different density level zones have different preset neighborhood radii and preset minimum number of points.

[0084] As an example, this embodiment addresses the issue of severe weather causing site or communication base station failures or damage in a certain area, resulting in interruptions or anomalies in observation data from a large number of sites. It employs the density-based spatial clustering algorithm DBSCAN (Density-Based Spatial Clustering of Applications with Noise) to detect the density of missing and abnormal sites within the spatiotemporal range of severe weather. The basic logic of the DBSCAN algorithm is as follows: 1) Core point: Within a specified neighborhood radius eps, if a point's neighborhood contains at least min_samples (i.e., the minimum number of points) points (including itself), then that point is marked as a core point.

[0085] 2) Density propagation: Starting from any core point, all points that are density-reachable from it (including other core points and points in their neighborhood) are grouped into the same cluster.

[0086] 3) Noise points: Points that cannot be propagated to by any core point density are marked as noise.

[0087] Based on this, this embodiment uses the DBSACN algorithm to detect the density of missing reporting stations and abnormal element stations within the spatiotemporal range of severe weather. The algorithm is as follows: (1) Overlay the time range of the impact of severe weather in the early warning information analysis record with the current business time, and extract the early warning information analysis record that is effective in the current business time.

[0088] (2) Overlay the disaster weather impact area in the effective early warning information parsing record with the abnormal stations, and extract the abnormal stations within the disaster weather impact area (i.e., the stations to be interpolated mentioned in the previous embodiment).

[0089] (3) Detecting abnormal sites within the same province: Based on the density level zone of the province to which the abnormal site belongs, different eps and min_sample are used for density detection; for example, the density level zone is divided into high-density zone, medium-density zone and low-density zone, and the site density of the high-density zone is greater than or equal to 150 sites / 10,000 square kilometers, the site density of the medium-density zone is greater than 50 sites / 10,000 square kilometers and less than 150 sites / 10,000 square kilometers, and the site density of the low-density zone is less than or equal to 50 sites / 10,000 square kilometers. Among them, the eps corresponding to the high-density zone is 25km and the min_sample is 4, the eps corresponding to the medium-density zone is 60km and the min_sample is 3, and the eps corresponding to the low-density zone is 120km and the min_sample is 2. At this time, assuming that the site density of province W is greater than 150 sites / 10,000 square kilometers, province W is classified as a high-density zone, and 25km is selected as eps and 2 is selected as min_sample for density detection within province W.

[0090] (4) Record the clusters of abnormal sites identified by the DBSACN algorithm to provide data support for alarms and visualization.

[0091] This embodiment detected 8 anomalous stations in **W Province** (a high-density area, eps=25km, min_sample=4), and the distance matrix (unit: km, approximate value) corresponding to the 8 anomalous stations is as follows: Figure 5 The following is an explanation of the density detection process using the example shown.

[0092] Among them, the neighborhood analysis is based on eps=25km. The points contained in the 25km neighborhood of each point are: A's neighborhood contains A(0km), B(8km), C(12km), D(15km), and E(18km). The total number of points corresponding to this is 5≥min_sample=4, so A is a core point; similarly, B is a core point, C is a core point, D is a core point, E is a core point, F is not a core point, G is not a core point, and H is not a core point.

[0093] The execution process of the DBSCAN algorithm is described as follows: Initialization is performed with eps = 25km and min_sample = 4, meaning all points are marked as unvisited, so the cluster number = 0, and the visit status of each point is A: not visited, B: not visited, C: not visited, D: not visited, E: not visited, F: not visited, G: not visited, H: not visited. Then, starting from point A: since point A is unvisited, its neighborhood is calculated as {A, B, C, D, E} (5 points). Since 5 ≥ 4, A is determined to be a core point, and a new cluster C1 is created. A is marked as visited and added to C1. Simultaneously, all neighboring points {B, C, D, E} of A are added to C1 and marked as visited. Therefore, the current state is cluster C1 = {A, B, C, D, E}. E} and the access status: A: Already, B: Already, C: Already, D: Already, E: Already, F: Not Already, G: Not Already, H: Not Already; then expand cluster C1, that is, traverse the neighborhood points {B,C,D,E} of A: (1) Check B (already visited, in the cluster): calculate the neighborhood of B = {B, A, C, D, E} (5 points), since 5≥4, it is determined that B is the core point, traverse the neighborhood points {A, C, D, E}: A is already in the cluster, C is already in the cluster, D is already in the cluster, E is already in the cluster, and no new points are added; similarly, C, D, and E are checked. Among them, only E is checked and a new point F is found. Then, the cluster C1 is expanded (i.e., the newly added point F is processed): the newly added point F is checked using the same processing principle as above until the expansion of cluster C1 is completed. Unvisited points are processed. When all points have been visited, the algorithm ends and the final target cluster information is output so that staff can accurately distinguish between the impact of wide-area disasters and sporadic equipment failures based on the target cluster information.

[0094] Based on this, this embodiment proposes an algorithm for detecting abnormal station density under severe weather conditions. Its core lies in deeply integrating density clustering algorithms with meteorological operational scenarios. Through dynamic parameter matching and spatiotemporal range overlay, it automatically identifies regional abnormal station clusters caused by large-scale disasters (such as rainstorms and typhoons), effectively distinguishing between the impact of widespread disasters and sporadic equipment failures. It employs an adaptive parameter strategy based on geographical density, innovatively pre-setting multiple sets of DBSCAN algorithm parameters (i.e., neighborhood radius eps and minimum number of points min_samples) according to the significant differences in station distribution density across Chinese provinces. For example, a smaller radius (e.g., 25km) is used for high-density areas, while a larger radius (e.g., 120km) is used for low-density areas, enabling the clustering algorithm to adapt to the station distribution characteristics of different geographical environments, ensuring the accuracy and fairness of detection. Furthermore, it applies the DBSCAN clustering algorithm to a specific operational method for diagnosing the health of ground meteorological observation networks under severe weather conditions, and establishes a rule system for dynamically adapting DBSCAN parameters based on the station density of provincial administrative divisions. In summary, this embodiment effectively improves the ability to perceive the failure of observation data caused by large-scale meteorological disasters, and realizes an intelligent leap from "discovering anomalies at a single station" to "identifying regional faults".

[0095] In summary, this embodiment provides a method for monitoring abnormal interruptions in ground observation data from automatic meteorological stations during severe weather. It employs a combination of "data interruption anomaly perception + disaster location": First, it rapidly identifies stations with missing reports and abnormal elements by combining Tianyuan station information with Tianqing ground observation data entry records, while simultaneously eliminating non-operational application stations. Second, it accurately pinpoints the spatiotemporal range of the disaster through disaster early warning information text parsing. Then, it overlays the spatiotemporal range of the severe weather with abnormal station information and uses a spatial clustering algorithm to label areas of missing or abnormal stations. Finally, it generates a list of ground observation data anomaly interruption alarms and real-time interpolation stations for the disaster area, and visualizes the missing and abnormal stations. This provides a visual and operable decision-making basis for emergency response to severe weather, comprehensively improving the proactive detection, precise location, and intelligent handling capabilities of data interruptions.

[0096] Secondly, embodiments of this application also provide a ground observation data anomaly monitoring system for severe weather.

[0097] In one embodiment, reference is made to Figure 2 , Figure 2 This is a schematic diagram of the functional modules of an embodiment of the ground-based observation data anomaly monitoring system for severe weather, as described in this application. Figure 2 As shown, a ground-based observation data anomaly monitoring system for severe weather includes: The data acquisition layer is used to obtain a list of information on automatic ground stations nationwide based on the Tianyuan interface, and to obtain all hourly ground observation data records corresponding to the current business time based on the Tianqing interface. The processing layer is used to clean and standardize the national list of automatic ground stations by using the ground observation hourly data entry records to obtain the target receivable station information list corresponding to the current business time. For each target station in the target receivable station information list, if there is no target ground observation hourly data entry record corresponding to the target station in all the ground observation hourly data entry records, the target station is determined to be an abnormal station that has not uploaded the observation data for the current business time. The presentation layer is used to issue alerts for all abnormal sites.

[0098] Furthermore, in one embodiment, the processing layer is also used for: If there is a target ground observation hourly data entry record corresponding to the target station in all the ground observation hourly data entry records, then determine whether there is a target meteorological quality control code with a characterization error or missing measurement value in the target ground observation hourly data entry record. If it exists, the target site is determined to be an abnormal site where the feature value is unavailable; If it does not exist, the target site is determined to be a normal site.

[0099] Furthermore, in one embodiment, the processing layer is also used for: For each abnormal site, obtain the historical frequency of the same service within a preset time period for that abnormal site; If the historical reporting rate for the same service is zero, the abnormal site is determined to be a non-service application site and is removed from the list.

[0100] Furthermore, in one embodiment, the processing layer is also used for: The list of historical severe weather warnings within a preset time period obtained based on the Tianqing interface is parsed to obtain the target information corresponding to each warning message. The target information includes the warning release time, severe weather type code, administrative division code of the issuing unit, and warning message text. For each warning message, the time range and area range of the severe weather impact are determined based on the corresponding target information; Based on the time range of the impact of severe weather, the target early warning information that is effective in the current business time is selected from all early warning information; For each abnormal site, if the spatial location of the abnormal site is within the disaster weather impact area corresponding to the target early warning information, then the abnormal site is determined to be a site to be interpolated. An alarm is issued for the site to be interpolated in order to initiate the emergency real-time interpolation process.

[0101] Furthermore, in one embodiment, the processing layer is specifically used for: The warning message text is parsed to obtain the target parsing results; Based on the target analysis results and the severe weather type code, the end time of the severe weather is determined, and the warning issuance time is taken as the start time of the severe weather. Construct the time range of severe weather impact based on the start and end times of severe weather; The scope of the area affected by severe weather was determined based on the target analysis results and the administrative division code of the issuing unit.

[0102] Furthermore, in one embodiment, the processing layer is specifically used for: If the target analysis result contains specific time keywords, the end time of the severe weather corresponding to the specific time keyword is determined based on the preset mapping relationship between keywords and time points; If the target analysis result does not contain specific time keywords but contains duration keywords, then the end time of the severe weather is determined based on the duration keywords and the warning issuance time. If the target analysis result does not contain specific time keywords and duration keywords but contains date range keywords, then the end time of the severe weather is determined based on the date range keywords and the warning release time. If the target parsing result does not contain specific time keywords, duration keywords, and date range keywords, then the target duration corresponding to the severe weather type code is determined based on the preset mapping relationship between severe weather type and duration, and the end time of severe weather is determined based on the target duration and the warning release time.

[0103] Furthermore, in one embodiment, the processing layer is specifically used for: Determine whether the administrative division code of the issuing unit is a county-level administrative division code; If so, the administrative division code of the issuing unit will be used as the scope of the area affected by severe weather; If not, then if the preset administrative division code table does not contain the target administrative division code corresponding to the keywords of the administrative division unit in the target parsing result, all county-level administrative division codes corresponding to the administrative division code of the publishing unit will be used as the scope of the area affected by the disaster weather. When the preset administrative division code table contains the target administrative division code corresponding to the keywords of the administrative division unit in the target parsing results, the scope of the area affected by the severe weather is determined based on the target administrative division code.

[0104] Furthermore, in one embodiment, if the target administrative division code is multiple and divided into target county-level administrative division code and target city-level administrative division code, the processing layer is further used for: Determine whether there exists a county-level administrative division code in the target county-level administrative division code that corresponds to the target city-level administrative division code; If they exist, then all target county-level administrative division codes will be considered as the affected area of ​​severe weather; If it does not exist, then all county-level administrative division codes corresponding to the target municipal administrative division code and all target county-level administrative division codes will be considered as the area affected by severe weather.

[0105] Furthermore, in one embodiment, the processing layer is also used for: If there are multiple interpolation sites located in the same province, the density-based clustering algorithm DBSCAN is used to calculate the density of the multiple interpolation sites located in the same province in order to generate the target cluster. Issue an alarm for the target cluster; The neighborhood radius and minimum number of points of the DBSCAN algorithm are determined according to the density level zone where the province is located. Different density level zones have different preset neighborhood radii and preset minimum number of points.

[0106] Understandably, see Figure 2 As shown, the ground observation data anomaly monitoring system for severe weather consists of four parts: acquisition layer, storage layer, processing layer, and display layer. The acquisition layer obtains national disaster early warning information, hourly ground observation data records, administrative division codes, and automatic weather station information by calling the Tianqing and Tianyuan interfaces, and stores the acquired raw information in the Tianjing database as messages. The storage layer employs different storage models based on the different characteristics of the monitored data: log data is stored in an index-based ElasticSearch database, alarms and configuration parameters are stored in a MongoDB database, and hotspot data is stored in a Redis in-memory database.

[0107] The processing layer is the core layer of the entire architecture. It realizes the detection of abnormal interruptions in ground observation data from automatic meteorological stations in disaster-stricken areas and the detection of abnormal station density, and generates alarms and abnormal station lists. This helps national and provincial meteorological operation and maintenance personnel to promptly detect abnormal ground observation data, initiate real-time interpolation, and ensure the quality of refined weather forecasts under severe weather conditions. It parses disaster warning information text to extract the spatiotemporal scope of the disaster, confirms whether the warning information is effective at the current time, and cleans and standardizes the automatic meteorological station station information and station information in the observation data records in conjunction with administrative division information to form a list of stations that should receive ground meteorological observation data for each operational period. Then, based on the database records and observation requirement quality control codes, it quickly identifies missing reporting stations and stations with abnormal elements, while simultaneously eliminating... Unless it is a business application site; then the spatiotemporal range of the disaster weather in the effective warning information list is overlaid with the abnormal site information to obtain the abnormal site information of the disaster area; and the spatial density of abnormal sites is analyzed to confirm whether the abnormal sites are in a cluster. If they are in a cluster, it means that the disaster in the area is serious and multiple sites or communication base stations are damaged. Finally, an abnormal interruption alarm for ground observation data in the disaster area is generated (business time, disaster type, level, number of abnormal counties and districts, total number of abnormal sites, number of dense areas, longest interruption time, etc.) and abnormal site details (station number, station name, province, city and county, administrative division code, latitude and longitude, altitude, total number of missing business time, list of missing business time, continuous missing reporting period, abnormal meteorological elements, etc.).

[0108] The display layer uses a GIS map to show the distribution of abnormal sites, dense areas, and the scope of disaster impact, and displays the latest abnormal site details, alarm information, and early warning information, while also providing access to historical information.

[0109] It is worth noting that the ground observation data anomaly monitoring system for severe weather provided in this embodiment went online in July 2025, generating real-time alarm and real-time interpolation station lists for each province, providing support for provincial severe weather emergency response.

[0110] During the trial operation in January and July: (1) Due to heavy rain, Province L applied to the national level for real-time supplementation at 07:00 on July 21, with 5 supplementation stations. The system issued an alarm at 05:00 on July 21, detecting 5 abnormal stations, which were completely consistent with the supplementation stations proposed by Province L, 2 hours ahead of schedule; (2) Due to heavy rain, Province J applied to the national level for real-time supplementation at 12:00 on July 22, with 26 supplementation stations. The system issued an alarm at 11:00 on July 21, detecting 34 abnormal stations, which not only completely covered the supplementation stations proposed by Province J, but also found 8 additional stations missed by the provincial level, 24 hours ahead of schedule; Due to heavy rain, City T applied to the national level for real-time supplementation at 10:00 on July 28, with 3 supplementation stations. The system issued an alarm at 03:00 on July 28. The system issued an alarm at 00:00 on July 25th, detecting 3 abnormal stations, which were completely consistent with the stations proposed by City T for supplementary reporting, 7 hours ahead of schedule. Province H1 applied to the national level for real-time reporting supplementary reporting at 09:00 on July 26th due to heavy rain, requesting 163 stations for supplementary reporting. The system issued an alarm at 00:00 on July 25th, detecting 195 abnormal stations, which not only completely covered the stations proposed by Province H1, but also discovered an additional 32 stations missed by the provincial level, 33 hours ahead of schedule. Province N applied to the national level for real-time reporting supplementary reporting at 11:00 on July 26th due to heavy rain, requesting 10 stations for supplementary reporting. The system issued an alarm at 02:00 on July 23rd, detecting 23 abnormal stations, which not only completely covered the stations proposed by Province N, but also discovered an additional 13 stations missed by the provincial level, 81 hours ahead of schedule.

[0111] During the official operation period: Based on the alarm and real-time interpolation site list generated by this system, H2 Province applied for real-time interpolation of 110 sites during Typhoon "JIAN" on August 25 and 16 sites during Typhoon "HUA" on September 24; Based on the alarm and real-time interpolation site list generated by this system, G Province applied for real-time interpolation of 35 sites during Typhoon "HUA" on September 24.

[0112] Currently, this system is deployed and officially operational at the national level, generating alarm and interpolation site lists in real time and pushing them to meteorological operation and maintenance personnel in each province. After receiving the alarm, the provincial meteorological operation and maintenance personnel will conduct a comprehensive analysis with the provincial meteorological observatory to determine whether to apply to the national level for real-time interpolation. For example, if Typhoon "HUA" makes landfall in Province G at 07:00 on September 24, 2025, this system will issue an alarm for abnormal interruption of observation data to Province G at 18:00 on the 23rd, and Province G will apply for real-time interpolation on the 24th, which greatly improves the provincial emergency support capabilities and meets the needs of emergency decision-making services.

[0113] In summary, this embodiment constructs a monitoring tool for abnormal interruptions in ground observation data from automatic meteorological stations during severe weather. It transforms the complex business scenario of "abnormal interruptions in station observation data during severe weather" into a quantifiable and real-time calculable indicator. First, it acquires the list of automatic meteorological stations and the database records of meteorological observation files based on Tianyuan and Tianqing, respectively, to detect missing reporting stations and stations with abnormal elements in real time. It further analyzes the historical reporting rate time-series characteristics of missing reporting stations to exclude non-business application stations such as emergency stations and experimental stations. Then, it determines the type, level, and spatiotemporal range of severe weather by parsing the disaster warning information text. Next, it overlays the abnormal stations with the spatiotemporal range of severe weather to identify stations with interrupted or abnormal observation data within the disaster's impact area. The DBSCAN algorithm is used to mark the geographical density of missing or abnormal stations. Finally, it generates alarms and sends them to the Tianjing alarm station to notify frontline maintenance personnel. Simultaneously, it generates a structured interpolation station list and dynamically renders station numbers, dense areas, and overlaid real-time warning signals on a GIS map on a visualization page, enhancing decision support capabilities in national and provincial severe weather emergency scenarios.

[0114] The functions of each module in the above-mentioned ground observation data anomaly monitoring system for severe weather correspond to the steps in the above-mentioned ground observation data anomaly monitoring method embodiment for severe weather. Their functions and implementation processes will not be described in detail here.

[0115] It should be noted that the sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0116] The terms "comprising" and "having," and any variations thereof, in the specification, claims, and accompanying drawings of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus. The terms "first," "second," and "third," etc., are used to distinguish different objects, etc., and do not indicate a sequence, nor do they limit "first," "second," and "third" to different types.

[0117] In the description of the embodiments of this application, terms such as "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a concrete manner.

[0118] In the description of the embodiments of this application, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in the text is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of this application, "multiple" means two or more.

[0119] In some processes described in the embodiments of this application, multiple operations or steps are included in a specific order. However, it should be understood that these operations or steps may not be executed in the order they appear in the embodiments of this application, or they may be executed in parallel. The sequence number of the operation is only used to distinguish the different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed sequentially or in parallel, and these operations or steps may be combined.

[0120] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device to execute the methods described in the various embodiments of this application.

[0121] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A disaster weather-oriented ground observation data anomaly monitoring method characterized by comprising: The method for monitoring anomalies in ground-based observation data for severe weather includes: The system obtains a list of information on all automatic ground stations nationwide based on the Tianyuan interface, and retrieves all hourly ground observation data records corresponding to the current business time based on the Tianqing interface. The information list of national automatic ground stations is cleaned and standardized by recording hourly ground observation data to obtain the target collection station information list corresponding to the current business time. For each target site in the target receivable site information list, if there is no target ground observation hour data entry record corresponding to the target site in all ground observation hour data entry records, then the target site is determined to be an abnormal site that has not uploaded the current business time observation data; Issue warnings for all abnormal sites; The list of historical severe weather warnings within a preset time period obtained based on the Tianqing interface is parsed to obtain the target information corresponding to each warning message. The target information includes the warning release time, severe weather type code, administrative division code of the issuing unit, and warning message text. For each warning message, the time range and area range of the severe weather impact are determined based on the corresponding target information; Based on the time range of the impact of severe weather, the target early warning information that is effective in the current business time is selected from all early warning information; For each abnormal site, if the spatial location of the abnormal site is within the disaster weather impact area corresponding to the target early warning information, then the abnormal site is determined to be a site to be interpolated. An alarm is issued for the site to be interpolated in order to initiate the emergency real-time interpolation process; The step of determining that the abnormal site is a site to be interpolated further includes: If there are multiple interpolation sites located in the same province, the density-based clustering algorithm DBSCAN is used to calculate the density of the multiple interpolation sites located in the same province in order to generate the target cluster. Issue an alarm for the target cluster; The neighborhood radius and minimum number of points of the DBSCAN algorithm are determined according to the density level zone where the province is located. Different density level zones have different preset neighborhood radii and preset minimum number of points.

2. The disaster weather-oriented ground observation data anomaly monitoring method according to claim 1, characterized by, The method further includes: If there is a target ground observation hourly data entry record corresponding to the target station in all the ground observation hourly data entry records, then determine whether there is a target meteorological quality control code with a characterization error or missing measurement value in the target ground observation hourly data entry record. If it exists, the target site is determined to be an abnormal site where the feature value is unavailable; If it does not exist, the target site is determined to be a normal site.

3. The disaster weather-oriented ground observation data anomaly monitoring method according to claim 1, characterized by, After the step of determining that the target site is an abnormal site that has not uploaded the current service observation data, the method further includes: For each abnormal site, obtain the historical frequency of the same service within a preset time period for that abnormal site; If the historical reporting rate for the same service is zero, the abnormal site is determined to be a non-service application site and is removed from the list.

4. The disaster weather-oriented ground observation data anomaly monitoring method according to claim 1, characterized by, The determination of the time range and area range of severe weather impact based on the corresponding target information includes: The warning message text is parsed to obtain the target parsing results; Based on the target analysis results and the severe weather type code, the end time of the severe weather is determined, and the warning issuance time is taken as the start time of the severe weather. Construct the time range of severe weather impact based on the start and end times of severe weather; The scope of the area affected by severe weather was determined based on the target analysis results and the administrative division code of the issuing unit.

5. The disaster weather-oriented ground observation data anomaly monitoring method according to claim 4, characterized by, The determination of the end time of severe weather based on the target parsing results and severe weather type code includes: If the target analysis result contains specific time keywords, the end time of the severe weather corresponding to the specific time keyword is determined based on the preset mapping relationship between keywords and time points; If the target analysis result does not contain specific time keywords but contains duration keywords, then the end time of the severe weather is determined based on the duration keywords and the warning issuance time. If the target analysis result does not contain specific time keywords and duration keywords but contains date range keywords, then the end time of the severe weather is determined based on the date range keywords and the warning release time. If the target parsing result does not contain specific time keywords, duration keywords, and date range keywords, then the target duration corresponding to the severe weather type code is determined based on the preset mapping relationship between severe weather type and duration, and the end time of severe weather is determined based on the target duration and the warning release time.

6. The disaster weather-oriented ground observation data anomaly monitoring method according to claim 4, characterized by, The area affected by severe weather, determined based on the target analysis results and the administrative division code of the issuing unit, includes: Determine whether the administrative division code of the issuing unit is a county-level administrative division code; If so, the administrative division code of the issuing unit will be used as the scope of the area affected by severe weather; If not, then if the preset administrative division code table does not contain the target administrative division code corresponding to the keywords of the administrative division unit in the target parsing result, all county-level administrative division codes corresponding to the administrative division code of the publishing unit will be used as the scope of the area affected by the disaster weather. When the preset administrative division code table contains the target administrative division code corresponding to the keywords of the administrative division unit in the target parsing results, the scope of the area affected by the severe weather is determined based on the target administrative division code.

7. The disaster weather-oriented ground observation data anomaly monitoring method according to claim 6, characterized by, If the target administrative division code is multiple and divided into target county-level administrative division codes and target city-level administrative division codes, the determination of the affected area based on the target administrative division code includes: Determine whether there exists a county-level administrative division code in the target county-level administrative division code that corresponds to the target city-level administrative division code; If they exist, then all target county-level administrative division codes will be considered as the affected area of ​​severe weather; If it does not exist, then all county-level administrative division codes corresponding to the target municipal administrative division code and all target county-level administrative division codes will be considered as the area affected by severe weather.

8. A disaster weather-oriented ground observation data anomaly monitoring system characterized by, The ground-based observation data anomaly monitoring system for severe weather includes: The data acquisition layer is used to obtain a list of information on automatic ground stations nationwide based on the Tianyuan interface, and to obtain all hourly ground observation data records corresponding to the current business time based on the Tianqing interface. The processing layer is used to clean and standardize the national list of automatic ground stations by using the ground observation hourly data entry records to obtain the target receivable station information list corresponding to the current business time. For each target station in the target receivable station information list, if there is no target ground observation hourly data entry record corresponding to the target station in all the ground observation hourly data entry records, the target station is determined to be an abnormal station that has not uploaded the observation data for the current business time. The presentation layer is used to issue early warnings for all abnormal sites; The processing layer is also used to parse the list of historical severe weather warnings within a preset time period obtained based on the Tianqing interface, in order to obtain the target information corresponding to each warning message. The target information includes the warning release time, severe weather type code, administrative division code of the issuing unit, and warning message text. For each warning message, the time range and area range of severe weather impact are determined based on its corresponding target information. Based on the time range of severe weather impact, the target warning messages that are effective in the current business time are filtered from all warning messages. For each abnormal site, if the spatial location of the abnormal site is located in If the target warning information falls within the disaster weather impact area, the abnormal station is identified as a station to be interpolated; an alarm is issued for the station to be interpolated to initiate the emergency real-time interpolation process; if there are multiple stations to be interpolated in the same province, the density-based clustering algorithm DBSCAN is used to calculate the density of the multiple stations to be interpolated in the same province to generate a target cluster; an alarm is issued for the target cluster; the neighborhood radius and minimum number of points of the DBSCAN algorithm are determined according to the density level zone where the province is located, and different density level zones have different preset neighborhood radii and preset minimum number of points.

Images