A weather sequence pattern recognition method and system for thunderstorm gale prediction

Abstract

The present application relates to the technical field of weather prediction, and provides a weather sequence pattern recognition method and system for thunderstorm gale prediction, which comprises the following steps: continuously acquiring high-frequency weather data, and determining a current evolution scenario from candidate evolution scenarios according to the space-time evolution characteristics of the high-frequency weather data; loading a scenario recognition strategy corresponding to the current evolution scenario, extracting a precursor index change sequence from the high-frequency weather data, and determining a precursor matching result and an initial warning credibility according to a precursor pattern recognition rule; adjusting the initial warning credibility according to a warning credibility calculation and adjustment rule and the current evolution scenario to obtain a current warning credibility; and generating a warning information when the precursor matching result indicates that the precursor index change sequence is consistent with a precursor pattern under the current evolution scenario, and the current warning credibility reaches a preset warning threshold. The present application has the effect of improving the accuracy and timeliness of thunderstorm gale warning.

Description

Technical Field

[0001] This invention relates to the field of meteorological forecasting technology, specifically to a meteorological sequence pattern recognition method and system for forecasting thunderstorms and strong winds. Background Technology

[0002] In the field of meteorological forecasting, accurately predicting sudden and destructive weather phenomena such as thunderstorms and strong winds is crucial for ensuring social safety and economic operation. Existing meteorological forecasting systems typically analyze long-term meteorological data collected from various observation devices, including sounding radar, ground weather stations, and satellites, to identify specific patterns of change associated with thunderstorm and strong wind events. These systems can provide relatively reliable warnings when facing typical severe convective weather caused by large weather systems (such as fronts and cyclones). However, in practice, forecasters have found that not all thunderstorms and strong winds follow these classic evolutionary paths. Some highly destructive thunderstorms and strong winds do not present obvious warning signals that can be identified by traditional systems before they occur. These events often arrive suddenly, and their triggering mechanisms are more complex and covert.

[0003] In-depth analysis reveals that the precursors to such events often manifest as the convergence and interaction of multiple weak meteorological disturbances in a specific time and space, rather than changes in a single significant meteorological element. For example, a weak upper-level disturbance, coupled with a slight increase in lower-level humidity and an inconspicuous surface airflow convergence line—conditions that, individually, are insufficient to trigger strong convection—may coincidentally combine to generate localized severe thunderstorms and strong winds. This phenomenon, formed by the superposition of multiple weak signals, makes it difficult for systems that traditionally rely on a single or a few strong features to detect, leading to missed warnings.

[0004] Further research indicates that the superposition of these weak signals is not a simple accumulation, but rather a nonlinear triggering process. When these weak disturbances reach a specific configuration in time and space, they collectively disrupt the original atmospheric balance, leading to the rapid release of unstable atmospheric energy. For example, a previously stable inversion layer may be rapidly eroded in a short period due to the delicate coordination of weak descending airflow at high altitudes and the lifting effect of warm, moist air at low altitudes, thus opening a "channel" for the outbreak of convection. The opening of this "channel" is often accompanied by drastic changes in meteorological parameters within a very short time, rather than a slow accumulation. This means that if we only focus on the absolute values ​​of meteorological elements or their average trends over longer time scales, we can easily miss this rapid, nonlinear evolution process.

[0005] To capture these rapidly evolving processes with greater precision, meteorological departments began introducing higher-frequency observational data, such as minute-level or even second-level data from automatic ground stations, encrypted radar scan data (e.g., Doppler radar velocity products and reflectivity factor products), and high-temporal-resolution geostationary meteorological satellite products (e.g., cloud top brightness temperature and water vapor channel information). Through analysis of this high-frequency data, technicians discovered that some very subtle but rapid "change sequences" do indeed exist before these "sudden" thunderstorm and gale events. These sequences manifest as specific, non-linear trajectories of change in multiple meteorological parameters within extremely short time windows (e.g., 10-30 minutes). However, identifying these constantly changing "evolutionary trajectories" presents new challenges. These trajectories are not fixed patterns; they are highly dependent on the prevailing atmospheric conditions. For example, in the summer afternoon, a specific sequence of wind field convergence and rising humidity may foreshadow thunderstorms and gales; but at night, a completely different sequence of temperature drop and air pressure rise may also foreshadow gales in a different context. This means that a "change sequence" that is a strong warning signal in one context may be just normal fluctuation or even completely unrelated in another. If all these different "change sequences" in various contexts are used indiscriminately for identification, the system will struggle to distinguish between genuine warning signals and irrelevant interference. This can cause the system to become overly "sensitive," generating numerous false alarms, or become overly "sluggish" to avoid false alarms, thus continuing to miss warnings.

[0006] The core issue lies in the fact that interpreting a specific "change sequence" requires considering the real-time, continuously changing large-scale and mesoscale atmospheric background conditions to accurately determine its warning significance. Simply matching "change sequences" without this context significantly reduces reliability and may even be misleading. These background conditions are not merely static; they are themselves a continuously changing state. For example, a slight oscillation of an upper-level jet stream axis or the formation and dissipation of a mesoscale vortex can alter the warning effectiveness of a "change sequence." This necessitates that the identification system understand "change within change"—that is, it must not only identify the sequence changes in meteorological parameters but also the continuous changes in the background environment in which these sequence changes occur, and adjust its identification logic accordingly. This ability to perceive and adapt to a constantly changing background is generally lacking in existing systems.

[0007] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0008] This application discloses a meteorological sequence pattern recognition method and system for thunderstorm and gale prediction, aiming to solve the problems that existing meteorological forecasting systems have difficulty capturing nonlinear precursor signals formed by the superposition of multiple weak meteorological disturbances when facing sudden and destructive thunderstorm and gale events, and that the interpretation of specific change sequences is detached from real-time and continuously changing atmospheric background conditions, resulting in missed warnings or false alarms.

[0009] The technical solution of this application is as follows: In a first aspect, this application discloses a meteorological sequence pattern recognition method for thunderstorm and strong wind prediction, comprising the following steps: Acquire initial meteorological data, which includes meteorological observation information and short-term forecast information; Based on initial meteorological data, candidate evolution scenarios of the future atmospheric environment are inferred, and corresponding scenario identification strategies are configured for different candidate evolution scenarios. The scenario identification strategies include precursor pattern recognition rules and warning credibility calculation and adjustment rules. The candidate evolution scenarios represent the evolution path or evolution type of the future atmospheric environment forming thunderstorm and strong wind process. Continuously acquire high-frequency meteorological data, and determine the current evolution scenario from candidate evolution scenarios based on the spatiotemporal evolution characteristics of the high-frequency meteorological data; the high-frequency meteorological data includes at least one of wind speed, wind direction, air pressure, temperature, humidity, radar echo or satellite observation data acquired continuously at preset time intervals; Load the scenario recognition strategy corresponding to the current evolution scenario, extract the change sequence of precursor indicators from high-frequency meteorological data, and determine the precursor matching result and the credibility of the initial warning according to the precursor pattern recognition rules; Based on the rules for calculating and adjusting the credibility of early warnings and the current evolution scenario, the initial credibility of early warnings is adjusted to obtain the current credibility of early warnings. When the current precursor matching result indicates that the sequence of precursor indicator changes is consistent with the precursor pattern under the current evolution scenario, and the current warning confidence reaches the preset warning threshold, a warning message is generated.

[0010] This technical solution effectively addresses the problems of existing thunderstorm and gale prediction systems struggling to capture weak precursor signals and adapt to complex and variable atmospheric backgrounds. Through scenario inference, high-frequency data analysis, and dynamic reliability adjustment, it significantly improves the accuracy and timeliness of thunderstorm and gale warnings, thereby avoiding warning omissions and false alarms, achieving unexpected technical results.

[0011] Furthermore, the step of adjusting the initial warning credibility based on the warning credibility calculation and adjustment rules and the current evolution scenario to obtain the current warning credibility includes: Continuously monitor the scenario consistency between the high-frequency meteorological data and the current evolving scenario to obtain scenario consistency monitoring results; When the scenario consistency monitoring result indicates that the high-frequency meteorological data is continuously inconsistent with the current evolving scenario, the background correction process is initiated. In the background correction process, local triggering factors are extracted from the high-frequency meteorological data. These local triggering factors include local convergence enhancement features, near-surface wind direction change features, humidity increase features, or temperature decrease features. Based on the local triggering factors, the current evolution scenario is modified to obtain the modified current evolution scenario; Based on the revised current evolution scenario, the scenario recognition strategy is adjusted to obtain the adjusted scenario recognition strategy. Based on the adjusted scenario recognition strategy, the initial warning confidence of the precursor indicator change sequence is recalculated to obtain the corrected confidence. Determine the current warning credibility based on the revised credibility. When the scenario consistency monitoring result indicates that the high-frequency meteorological data and the current evolution scenario are not continuously inconsistent, the initial warning credibility is adjusted according to the warning credibility calculation and adjustment rules to obtain the current warning credibility.

[0012] Further, the step of generating warning information when the precursor matching result indicates that the precursor indicator change sequence is consistent with the precursor pattern under the current evolution scenario, and the current warning confidence reaches a preset warning threshold, includes: The application scenario categories for obtaining early warning information recipients; Based on the application scenario category, select and load the early warning release strategy, which includes scenario threshold adjustment rules and early warning information content customization rules; According to the scenario threshold adjustment rules, the preset warning threshold is adjusted to obtain the scenario-based warning threshold; Determine whether the current warning credibility reaches the scenario-based warning threshold; When the current warning credibility reaches the scenario-based warning threshold, warning information is generated according to the warning information content customization rules, and the warning information is sent to the corresponding warning information recipient; When the current warning confidence level does not reach the scenario-based warning threshold, no warning information is generated, and new high-frequency meteorological data continues to be acquired to update the current warning confidence level.

[0013] Furthermore, after sending the warning information to the corresponding warning information recipient, the method further includes: Continuously receive new high-frequency meteorological data; According to the preset cycle, the current evolution scenario and precursor indicator change sequence are re-identified based on new high-frequency meteorological data, and the current warning credibility is recalculated. Based on the re-identified current evolution scenario, the sequence of changes in precursor indicators, and the recalculated current warning confidence, new warning elements are generated; wherein, the warning elements include warning intensity, scope of impact, and evolution path; The new warning elements are compared with the warning elements in the already sent warning information to obtain the element comparison results. When the element comparison result indicates that the new warning element differs from the warning element in the sent warning information, the warning information update process is triggered. In the aforementioned early warning information update process, revised early warning information is generated based on the new early warning elements; The revised warning information will be sent again through the original transmission channel; When the element comparison result indicates that the new warning element is no different from the warning element in the previously sent warning information, the previously sent warning information is maintained.

[0014] Further, the step of generating warning information when the precursor matching result indicates that the precursor indicator change sequence is consistent with the precursor pattern under the current evolution scenario, and the current warning confidence reaches a preset warning threshold, includes: The system acquires information on the environmental vulnerability of the area where the recipient of the early warning information is located, the current activity intensity, and the response time limit requirements. The response time limit requirements indicate the time required for the recipient of the early warning information to complete risk avoidance, work stoppage, evacuation, or protective measures. Based on the environmental vulnerability information, the current activity intensity information, and the response time limit requirement information, assess the risk exposure level of the recipient of the early warning information; Based on the risk exposure level, the preset warning threshold is adjusted to obtain the risk exposure warning threshold; Determine whether the current warning credibility reaches the risk exposure warning threshold; When the current warning confidence level reaches the risk exposure warning threshold, a warning message is generated, and the risk exposure level and the current warning confidence level are written into the warning message; When the current warning credibility does not reach the risk exposure warning threshold, no warning information is generated, and the high-frequency meteorological data and risk exposure level of the area where the warning information recipient is located continue to be monitored.

[0015] Furthermore, the step of continuously acquiring high-frequency meteorological data includes: A multi-channel data receiving architecture is established, which includes a primary and backup data link, a data caching module, and a data quality monitoring module. The system receives high-frequency meteorological data to be processed through the primary and backup data links, and switches to the backup link when the primary link fails. The data caching module smooths the data stream when the receiving rate of the high-frequency meteorological data to be processed fluctuates. The data quality monitoring module detects abnormal values ​​in the high-frequency meteorological data to be processed. When no abnormal values ​​are detected in the high-frequency meteorological data to be processed, the received high-frequency meteorological data to be processed will be used as high-frequency meteorological data. When an abnormal value is detected in the high-frequency meteorological data to be processed, a data retransmission request is triggered or data interpolation processing is performed to obtain the retransmitted or interpolated high-frequency meteorological data to be processed, and the retransmitted or interpolated high-frequency meteorological data to be processed is used as high-frequency meteorological data.

[0016] Furthermore, after the step of triggering a data retransmission request or performing data interpolation processing when an outlier is detected in the high-frequency meteorological data to be processed, to obtain the retransmitted or interpolated high-frequency meteorological data to be processed, the method further includes: A consistency check is performed on the high-frequency meteorological data to be processed after retransmission or interpolation. The consistency check includes comparing the changing trends between the high-frequency meteorological data to be processed after retransmission or interpolation and the meteorological data at adjacent time points and adjacent spatial points. Physical constraint checks are performed on the high-frequency meteorological data to be processed after retransmission or interpolation, and the physical constraint checks include checking the physical correlations between meteorological elements. When both the consistency check and the physical constraint check pass, the high-frequency meteorological data to be processed after retransmission or interpolation is retained. If the consistency check or the physical constraint check fails, the high-frequency meteorological data to be processed after retransmission or interpolation is cleared, and the data retransmission request is triggered again or the data interpolation process is re-executed. The consistency check and physical constraint check are then re-executed on the high-frequency meteorological data to be processed after retransmission or re-interpolation.

[0017] Furthermore, the consistency verification steps include: Read the current evolution scenario information corresponding to the current evolution scenario, which includes the typical spatiotemporal evolution scale and local characteristics of thunderstorm and gale events; Based on the current evolution scenario information, the spatiotemporal range and verification tolerance of the verification window are adjusted so that consistency verification can be performed within the adjusted verification window. Within the adjusted verification window, the changing trends of the high-frequency meteorological data to be processed after retransmission or interpolation are compared with those of meteorological data at adjacent time points and adjacent spatial points to obtain the change comparison results. When the change comparison result indicates that the change trend is within the adjusted verification tolerance range, it is determined that the high-frequency meteorological data to be processed after retransmission or interpolation conforms to the atmospheric evolution continuity characteristics, and it is determined that the high-frequency meteorological data to be processed after retransmission or interpolation passes the consistency verification. When the change comparison result indicates that the change trend is not within the adjusted verification tolerance range, it is determined that the high-frequency meteorological data to be processed after retransmission or interpolation has failed the consistency verification.

[0018] Furthermore, the step of adjusting the spatiotemporal range and verification tolerance of the verification window based on the current evolution scenario information, so as to perform consistency verification within the adjusted verification window, includes: Continuously monitor the evolution of the current atmospheric environment; When the current atmospheric environment does not meet the preset scenario switching conditions, the spatiotemporal range and verification tolerance of the verification window are adjusted according to the current evolution scenario information, so as to perform consistency verification on the high-frequency meteorological data to be processed after retransmission or interpolation within the adjusted verification window. When the current atmospheric environment evolution meets the preset scenario switching conditions, the system identifies the change in the evolution scenario of thunderstorm and strong wind events and obtains information on the new evolution scenario after the switch. Based on the new evolution scenario information, the spatiotemporal range and verification tolerance of the verification window are adjusted so as to perform consistency verification on the high-frequency meteorological data to be processed after retransmission or interpolation within the adjusted verification window.

[0019] Secondly, this application also discloses a meteorological sequence pattern recognition system for thunderstorm and strong wind prediction, used to execute the aforementioned meteorological sequence pattern recognition method for thunderstorm and strong wind prediction, including: The initial meteorological acquisition module is used to acquire initial meteorological data, which includes meteorological observation information and short-term forecast information. The evolution scenario inference module is used to infer candidate evolution scenarios of the future atmospheric environment based on initial meteorological data, and to configure corresponding scenario recognition strategies for different candidate evolution scenarios. The scenario recognition strategies include precursor pattern recognition rules and early warning credibility calculation and adjustment rules. The evolution scenario determination module is used to continuously acquire high-frequency meteorological data and determine the current evolution scenario from candidate evolution scenarios based on the spatiotemporal evolution characteristics of the high-frequency meteorological data. The data processing execution module is used to load the scenario recognition strategy corresponding to the current evolution scenario, extract the change sequence of precursor indicators from high-frequency meteorological data, and determine the precursor matching result and the credibility of the initial warning according to the precursor pattern recognition rules. The early warning credibility determination module is used to adjust the initial early warning credibility based on the early warning credibility calculation and adjustment rules and the current evolution scenario to obtain the current early warning credibility. The early warning information generation module is used to generate early warning information when the current early warning matching result indicates that the change sequence of the precursor index is consistent with the precursor pattern under the current evolution scenario, and the current early warning credibility reaches the preset early warning threshold.

[0020] This application provides a meteorological sequence pattern recognition system for thunderstorm and gale prediction. Through its modular design, the system can efficiently execute the above-mentioned methods, achieving accurate prediction and early warning of thunderstorms and gales. It solves the problem of insufficient early warning capability of existing systems under complex meteorological conditions and has significant practical value.

[0021] Beneficial Effects: This application discloses a meteorological sequence pattern recognition method for thunderstorm and strong wind prediction. It acquires initial meteorological data and infers candidate evolution scenarios of the future atmospheric environment. Corresponding scenario recognition strategies are configured for different scenarios, including precursor pattern recognition rules and warning credibility calculation and adjustment rules. Subsequently, high-frequency meteorological data is continuously acquired, and the current evolution scenario is determined based on its spatiotemporal evolution characteristics. On this basis, the scenario recognition strategy corresponding to the current evolution scenario is loaded, precursor indicator change sequences are extracted from the high-frequency meteorological data, and precursor matching results and initial warning credibility are determined according to the precursor pattern recognition rules. Finally, the initial warning credibility is adjusted according to the warning credibility calculation and adjustment rules and the current evolution scenario to obtain the current warning credibility. Warning information is generated when the precursor matching results are consistent and the current warning credibility reaches a preset warning threshold.

[0022] This method effectively solves two core problems existing in thunderstorm and gale prediction systems: first, the difficulty in capturing nonlinear precursor signals formed by the superposition of multiple weak meteorological disturbances; and second, the interpretation of specific change sequences in warnings being detached from real-time and continuously changing atmospheric background conditions, leading to missed warnings or false alarms. By introducing "candidate evolution scenarios" and "scenario recognition strategies," this application can configure different precursor pattern recognition rules and warning credibility calculation and adjustment rules for different atmospheric background conditions, thereby overcoming the "one-size-fits-all" recognition mode of traditional systems and improving the ability to capture complex and concealed precursor signals. Simultaneously, by continuously acquiring "high-frequency meteorological data" and dynamically determining the "current evolution scenario," it ensures that the interpretation of precursor indicator change sequences is always combined with real-time and continuously changing atmospheric background conditions, avoiding misjudgments caused by changes in background conditions. This dynamic and contextualized recognition mechanism enables the system to more accurately identify nonlinear triggering processes formed by the superposition of multiple weak signals, effectively solving the problem of missed warnings in existing systems when facing such sudden thunderstorm and gale events, and significantly improving the accuracy and timeliness of thunderstorm and gale warnings. Attached Figure Description

[0023] Figure 1 This is a flowchart of a meteorological sequence pattern recognition method for thunderstorm and strong wind prediction according to one embodiment of the present invention; Figure 2 This is a flowchart illustrating the method for obtaining the current warning confidence level in an embodiment of the present invention. Figure 3 This is a system block diagram of a meteorological sequence pattern recognition system for thunderstorm and strong wind prediction according to an embodiment of the present invention; Explanation of reference numerals in the attached figures: 1. Meteorological sequence pattern recognition system for thunderstorm and strong wind forecasting; 11. Initial meteorological acquisition module; 12. Evolution scenario inference module; 13. Evolution scenario determination module; 14. Data processing execution module; 15. Early warning credibility determination module; 16. Early warning information generation module. Detailed Implementation

[0024] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0025] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0026] This application proposes a meteorological sequence pattern recognition method for thunderstorm and strong wind prediction, combining... Figure 1 As shown, it includes the following steps: S1, Acquire initial meteorological data, which includes meteorological observation information and short-term forecast information; S2, based on initial meteorological data, infers candidate evolution scenarios of the future atmospheric environment, and configures corresponding scenario identification strategies for different candidate evolution scenarios. The scenario identification strategies include precursor pattern recognition rules and warning credibility calculation and adjustment rules. The candidate evolution scenarios represent the evolution path or evolution type of the future atmospheric environment forming thunderstorm and strong wind process. S3, continuously acquire high-frequency meteorological data, and determine the current evolution scenario from candidate evolution scenarios based on the spatiotemporal evolution characteristics of the high-frequency meteorological data; the high-frequency meteorological data includes at least one of wind speed, wind direction, air pressure, temperature, humidity, radar echo or satellite observation data acquired continuously at preset time intervals; S4, load the scenario recognition strategy corresponding to the current evolution scenario, extract the precursor index change sequence from high-frequency meteorological data, and determine the precursor matching result and the initial warning credibility according to the precursor pattern recognition rules; S5. Based on the rules for calculating and adjusting the credibility of the early warning and the current evolution scenario, adjust the initial credibility of the early warning to obtain the current credibility of the early warning. S6. When the current precursor matching result indicates that the precursor indicator change sequence is consistent with the precursor pattern under the current evolution scenario, and the current warning credibility reaches the preset warning threshold, a warning message is generated.

[0027] To better understand the method proposed in this application, some key terms are first explained. "Initial meteorological data" refers to the dataset acquired at the beginning of the thunderstorm and gale forecasting process to establish preliminary atmospheric environmental knowledge. It typically includes meteorological observation information and short-term forecast information. Meteorological observation information can be real-time or near-real-time meteorological element data acquired through observation platforms such as ground-based meteorological stations, radiosondes, radar, and satellites, including data such as temperature, humidity, air pressure, wind speed, wind direction, precipitation, cloud cover, and radar echo intensity. Short-term forecast information can be the forecast results of meteorological elements for the next few hours to days output by numerical weather prediction models, such as temperature field, humidity field, wind field, convective available potential energy, convective suppression energy, and vertical wind shear. The aforementioned initial meteorological data collectively describe the current and short-term future atmospheric environmental state, providing a data foundation for subsequent candidate evolution scenario inference.

[0028] "Candidate evolution scenarios" refer to the evolution paths or types of future atmospheric conditions that may lead to thunderstorms and strong winds, inferred from initial meteorological data. For example, under different atmospheric circulation backgrounds, thermal conditions, and dynamic triggering conditions, frontal-triggered thunderstorm and strong wind scenarios, orographic lifting thunderstorm and strong wind scenarios, localized thermal convection thunderstorm and strong wind scenarios, or squall line thunderstorm and strong wind scenarios can occur. Each candidate evolution scenario corresponds to a type of change process that the atmospheric environment may undergo before the occurrence of thunderstorms and strong winds. "Scenario identification strategy" is a set of identification rules configured for different candidate evolution scenarios, including precursor pattern identification rules and warning credibility calculation and adjustment rules. Precursor pattern identification rules define which meteorological element change sequences can be considered precursor signals of thunderstorms and strong winds under specific evolution scenarios, and how to determine the consistency between real-time change sequences and precursor patterns. Warning credibility calculation and adjustment rules calculate the initial warning credibility based on precursor matching results and dynamically adjust the initial warning credibility based on real-time atmospheric environmental changes.

[0029] "High-frequency meteorological data" refers to meteorological data acquired continuously at relatively short time intervals, which may include at least one of wind speed, wind direction, air pressure, temperature, humidity, radar echo, or satellite observation data. This high-frequency meteorological data has high temporal resolution and can be used to capture rapidly changing atmospheric signals before thunderstorms and strong winds occur. "Precursor index change sequence" refers to the trajectory formed by the changes of multiple meteorological elements over time within a specific time window, extracted from high-frequency meteorological data. For example, before a thunderstorm and strong wind occur, there may be continuous changes such as a rapid increase in wind speed, a sudden change in wind direction, a sharp drop in air pressure, a rapid increase in humidity, a rapid increase in radar echo, or a rapid decrease in cloud top brightness temperature. "Warning credibility" is an indicator used to measure the probability of thunderstorms and strong winds occurring or the reliability of warnings. The higher the value, the more suitable the current atmospheric environment is for thunderstorms and strong winds, or the more reliable the corresponding warning result.

[0030] The meteorological sequence pattern recognition method for thunderstorm and gale prediction proposed in this application is based on inferring candidate evolution scenarios of the future atmospheric environment from initial meteorological data and configuring corresponding scenario recognition strategies for different candidate evolution scenarios. Subsequently, the system continuously acquires high-frequency meteorological data, determines the current evolution scenario based on the spatiotemporal evolution characteristics of the high-frequency meteorological data, and loads the scenario recognition strategy corresponding to the current evolution scenario, extracting precursor indicator change sequences from the high-frequency meteorological data. The system determines the precursor matching result and initial warning confidence based on precursor pattern recognition rules, and then dynamically adjusts the initial warning confidence according to the warning confidence calculation and adjustment rules to obtain the current warning confidence. When the current precursor matching result indicates that the precursor indicator change sequence is consistent with the precursor pattern under the current evolution scenario, and the current warning confidence reaches a preset warning threshold, a thunderstorm and gale warning is generated.

[0031] In practical implementation, the initial meteorological data is acquired first. This initial meteorological data can consist of meteorological observation information and short-term forecast information. Meteorological observation information may include real-time temperature, humidity, air pressure, wind speed, and wind direction data from automatic weather stations; reflectivity factor, radial velocity, and echo top height data from Doppler weather radar; and cloud top brightness temperature, water vapor channel information, and cloud system development characteristics from geostationary meteorological satellites. Short-term forecast information may include forecast products from numerical weather prediction models for the next 6, 12, or 24 hours, such as convective available potential energy, convective suppression energy, vertical wind shear, low-level jet stream, temperature and humidity field, and wind field. The above data can be acquired in batches from meteorological databases via meteorological data interfaces, or continuously received from meteorological observation networks or forecast product platforms through real-time data stream subscription services. By uniformly formatting, temporally aligning, and spatially interpolating the initial meteorological data, an initial atmospheric environment description for inferring candidate evolution scenarios can be formed.

[0032] Secondly, based on initial meteorological data, candidate evolution scenarios for the future atmospheric environment are inferred, and corresponding scenario recognition strategies are configured for different candidate evolution scenarios. Specifically, the system can infer the evolution path of potential future thunderstorms and strong winds based on the large-scale circulation background, mesoscale system configuration, thermal instability conditions, vertical wind shear conditions, and topographic influence factors in the initial meteorological data. For example, when the initial meteorological data shows the presence of a significant frontal system, strong vertical wind shear, and low-level water vapor transport, a front-triggered thunderstorm and strong wind scenario can be inferred; when the near-surface temperature is high, humidity is high, and atmospheric instability energy accumulation is significant in the summer afternoon, a localized thermal convection thunderstorm and strong wind scenario can be inferred; when radar echoes show a linear development trend and there is significant convergence in the surface wind field, a squall line thunderstorm and strong wind scenario can be inferred. Candidate evolution scenarios can be inferred using expert experience rules, decision trees, support vector machines, recurrent neural networks, or other models.

[0033] For different candidate evolution scenarios, the system configures corresponding scenario recognition strategies. For localized thermal convection-type thunderstorm and gale scenarios, the precursor pattern recognition rules can focus on changes such as near-surface temperature rise, increased humidity, enhanced local convergence, rapid radar echo generation, and rapid decrease in cloud top brightness temperature; the warning credibility calculation and adjustment rules can focus on the convection development speed, the degree of thermal instability energy accumulation, and whether the triggering conditions persist. For front-triggered thunderstorm and gale scenarios, the precursor pattern recognition rules can focus on signals such as wind direction changes near the front, pressure changes, enhanced temperature and humidity gradients, and organized development of radar echoes; the warning credibility calculation and adjustment rules can focus on the frontal movement speed, vertical wind shear intensity, and echo enhancement trends. The above scenario recognition strategies can be pre-set by meteorological experts based on historical thunderstorm and gale cases, or they can be automatically learned from a large number of historical meteorological sequences and disaster records through machine learning.

[0034] Next, high-frequency meteorological data is continuously acquired, and the current evolution scenario is determined from candidate evolution scenarios based on the spatiotemporal evolution characteristics of the high-frequency meteorological data. High-frequency meteorological data can include wind speed, wind direction, air pressure, temperature, and humidity data collected at minute or second intervals, or continuous observation data such as radar reflectivity factor, radial velocity, and satellite cloud image features. For example, wind speed, wind direction, air pressure, temperature, and humidity data from ground automatic weather stations can be acquired at a frequency of once per minute, and Doppler radar reflectivity factor and radial velocity data can be acquired at a frequency of once every 6 minutes. The system performs spatiotemporal evolution analysis on the high-frequency meteorological data to identify the direction, rate, spatial movement trend, and organizational structure changes of meteorological elements. For example, dynamic time warping, hidden Markov models, long short-term memory networks, or other pattern recognition methods can be used to compare real-time high-frequency meteorological data with typical evolution paths of each candidate evolution scenario. When real-time data exhibits spatiotemporal evolution characteristics highly consistent with a particular candidate evolution scenario, that candidate evolution scenario is determined as the current evolution scenario. For example, if real-time high-frequency data shows a rapid change in near-surface wind direction, and the radar echo significantly strengthens and forms an arc-shaped echo structure in a short period of time, the current evolution scenario can be identified as a squall line thunderstorm wind scenario.

[0035] Then, the system loads the scenario recognition strategy corresponding to the current evolution scenario, extracts the precursor indicator change sequence from high-frequency meteorological data, and determines the precursor matching result and initial warning credibility according to the precursor pattern recognition rules. In specific implementation, once the current evolution scenario is determined to be of a certain type, the system calls the precursor pattern recognition rules corresponding to that type to determine the precursor indicators to be extracted and the corresponding time windows. For example, for a localized thermal convection type thunderstorm and gale scenario, the system can extract the change sequences of near-surface temperature, humidity, wind speed, wind direction, air pressure, and radar echo intensity over the past 15 or 30 minutes; for a front-triggered thunderstorm and gale scenario, the system can extract the air pressure change, wind direction change, temperature and humidity gradient change, and echo organization change sequences along the frontal movement direction. The precursor indicator change sequence can be represented as a numerical sequence of multiple meteorological elements changing over time, or as the trajectory of meteorological element changes at time-space points. The system matches the precursor indicator change sequence with typical precursor patterns according to the precursor pattern recognition rules under the current evolution scenario to obtain the precursor matching result. If the sequence of changes in precursor indicators matches typical precursor patterns in terms of change order, change magnitude, duration, and spatial organization, a consistent precursor matching result can be obtained. At the same time, the initial warning confidence is calculated based on the degree of matching; the higher the degree of matching, the higher the initial warning confidence.

[0036] Next, based on the warning credibility calculation and adjustment rules and the current evolution scenario, the initial warning credibility is adjusted to obtain the current warning credibility. Specifically, the initial warning credibility is mainly based on the degree of matching between the precursor indicator change sequence and the precursor pattern. However, the formation process of thunderstorms and strong winds is continuously affected by real-time thermal conditions, dynamic conditions, triggering conditions, and the state of convective development. Therefore, the system needs to dynamically adjust the initial warning credibility according to the warning credibility calculation and adjustment rules corresponding to the current evolution scenario. For example, when the current evolution scenario shows a continuous increase in convective effective potential energy, enhanced low-level convergence, rapid development of radar echoes, or the emergence of new triggering mechanisms, the warning credibility can be appropriately increased even if the initial precursor matching degree has not yet reached a high level; when the current evolution scenario shows a weakening of convective instability energy, the disappearance of triggering conditions, a weakening of radar echoes, or a weakening of wind field convergence, the warning credibility can be decreased. Adjustments can be achieved through preset weighting factors, fuzzy logic reasoning, Bayesian updates, or machine learning models to obtain a current warning credibility that better reflects the real-time atmospheric evolution state.

[0037] Finally, when the current precursor matching result indicates that the precursor indicator change sequence is consistent with the precursor pattern under the current evolution scenario, and the current warning credibility reaches the preset warning threshold, a warning message is generated. In specific implementation, the system uses both the precursor matching result and the current warning credibility as warning triggering conditions. When there is only a short-term fluctuation of a single meteorological element but no precursor indicator change sequence consistent with the current evolution scenario, a warning message may not be generated temporarily to reduce the risk of false alarms; when multiple precursor indicator change sequences are consistent with the precursor pattern under the current evolution scenario, and the current warning credibility reaches the preset warning threshold, the system generates a thunderstorm gale warning message. The warning message may include the warning level, affected area, expected occurrence time, expected duration, main precursor indicators, current warning credibility, and suggested defense measures. Through the above processing, this application can infer candidate evolution scenarios based on initial meteorological data, and dynamically determine the current evolution scenario, identify precursor indicator change sequences, and adjust warning credibility by combining high-frequency meteorological data, thereby improving the adaptability to different evolution paths and the reliability of warning results in thunderstorm gale forecasting.

[0038] Optional, combined Figure 2 As shown, the steps to adjust the initial warning credibility to obtain the current warning credibility, based on the warning credibility calculation and adjustment rules and the current evolution scenario, include: A1. Continuously monitor the scenario consistency between high-frequency meteorological data and the current evolving scenario to obtain scenario consistency monitoring results; A2, When the scenario consistency monitoring results indicate that the high-frequency meteorological data is consistently inconsistent with the current evolving scenario, the background correction process is initiated; A3. In the background correction process, local triggering factors are extracted from high-frequency meteorological data. Local triggering factors include local convergence enhancement characteristics, near-surface wind direction change characteristics, humidity increase characteristics, or temperature decrease characteristics. A4. Based on local triggering factors, the current evolution scenario is modified to obtain the modified current evolution scenario; A5. Based on the revised current evolution scenario, adjust the scenario recognition strategy to obtain the adjusted scenario recognition strategy; A6. Based on the adjusted scenario recognition strategy, the initial warning confidence of the precursor indicator change sequence is recalculated to obtain the corrected confidence. A7. Determine the current warning confidence level based on the revised confidence level; A8. When the scenario consistency monitoring results indicate that the high-frequency meteorological data and the current evolving scenario are not continuously inconsistent, the initial warning credibility is adjusted according to the warning credibility calculation and adjustment rules to obtain the current warning credibility.

[0039] Specifically, continuously monitoring the consistency between high-frequency meteorological data and the current evolving scenario refers to assessing the degree of matching between high-frequency meteorological data (such as wind speed, wind direction, radar echoes, etc.) and the expected evolution patterns or trends of meteorological elements under the current evolving scenario by comparing them in real time. The scenario consistency monitoring result can be a quantitative indicator, such as correlation coefficient, deviation value, or model matching degree, used to indicate the degree of agreement between the current actual atmospheric conditions and the preset scenario.

[0040] Specifically, when scenario consistency monitoring results indicate a persistent inconsistency between high-frequency meteorological data and the current evolution scenario, it means that the actual atmospheric evolution may have deviated from the initially determined scenario path, requiring the initiation of a background correction process. Persistent inconsistency can be understood as the deviation between high-frequency meteorological data and the current evolution scenario exceeding a preset tolerance threshold over a period of time or at multiple consecutive observation points.

[0041] In the background correction process, extracting local triggering factors from high-frequency meteorological data is a crucial step. Local triggering factors refer to meteorological characteristics that can significantly influence the occurrence and development of thunderstorms and strong winds on a small scale. For example, enhanced local convergence indicates that airflow converges and strengthens in a local area, which is conducive to the formation of upward motion; abrupt changes in near-surface wind direction may indicate the formation of gust fronts or cold pool outflows; a sudden increase in humidity indicates a rapid increase in atmospheric water vapor content, providing sufficient water vapor for convection development; and a decrease in temperature may be related to cold air intrusion or convective descending airflow. The presence of these factors often foreshadows the localization of thunderstorm and strong wind events or changes in their evolution path.

[0042] Furthermore, based on the extracted local triggering factors, the current evolution scenario is revised to obtain the revised current evolution scenario. This revision can be a fine-tuning of the original scenario parameters, or a switch to a sub-scenario that better matches the current local characteristics, or a new evolution scenario. For example, if strong local convergence is detected, the current evolution scenario may be revised to a more explosive convection development scenario.

[0043] Therefore, the scenario identification strategy is adjusted based on the revised current evolution scenario. The scenario identification strategy includes precursor pattern recognition rules and early warning confidence calculation and adjustment rules. When the evolution scenario is revised, the corresponding identification and calculation rules also need to be adaptively adjusted to ensure that subsequent precursor identification and confidence calculation can accurately reflect the revised scenario characteristics. For example, if the revised scenario predicts faster storm development, the threshold for precursor pattern recognition may be adjusted to be more sensitive, or the weight of certain indicators may be increased.

[0044] Based on this, and using the adjusted scenario identification strategy, the initial warning confidence of the precursor indicator change sequence is recalculated to obtain the corrected confidence. This recalculation process ensures that the warning confidence is derived based on the latest and most accurate evolution scenario and identification strategy. Finally, based on the corrected confidence, the current warning confidence is determined as the final basis for generating warning information.

[0045] Optionally, when the current precursor matching result indicates that the precursor indicator change sequence is consistent with the precursor pattern under the current evolution scenario, and the current warning confidence reaches the preset warning threshold, the steps for generating warning information include: The application scenario categories for obtaining early warning information recipients; Select and load the early warning release strategy according to the application scenario category. The early warning release strategy includes scenario threshold adjustment rules and early warning information content customization rules. Based on the scenario threshold adjustment rules, the preset warning threshold is adjusted to obtain the scenario-based warning threshold; Determine whether the credibility of the current warning has reached the scenario-based warning threshold; When the current warning credibility reaches the scenario-based warning threshold, a warning message is generated according to the warning message content customization rules and sent to the corresponding warning message recipient; When the current warning credibility does not reach the scenario-based warning threshold, no warning information is generated, and new high-frequency meteorological data continues to be acquired to update the current warning credibility.

[0046] Specifically, the application scenario category refers to the industry, field, or response needs of the recipient of the early warning information in response to thunderstorm and strong wind events. For example, it can include agricultural production, transportation, urban management, power supply, construction, and tourist attractions. The purpose is to categorize different early warning information recipients in order to provide more targeted early warning services.

[0047] The early warning release strategy can be understood as a set of specifications for generating and releasing early warning information for specific application scenarios, aiming to guide the customized processing of early warning information. This strategy specifically includes scenario threshold adjustment rules and early warning information content customization rules. Scenario threshold adjustment rules are used to dynamically adjust preset early warning thresholds according to different application scenarios. For example, for transportation scenarios with high sensitivity to thunderstorm and strong wind risks, the early warning threshold can be appropriately lowered to issue warnings earlier; while for agricultural production scenarios with relatively low risk sensitivity, the early warning threshold can be appropriately raised to avoid unnecessary interference. Early warning information content customization rules are used to personalize the specific content, format, and release channels of early warning information according to the characteristics of the application scenario. For example, early warning information for urban management departments may focus on protection suggestions for urban infrastructure, while early warning information for agricultural users may include guidance on windproofing and reinforcing crops.

[0048] In practical applications, when the current warning credibility reaches the scenario-based warning threshold, it means that the probability and impact of a thunderstorm and strong wind event have reached the acceptable risk level for that specific application scenario. At this point, a warning message will be generated and sent according to the customized rules based on the warning information content. Conversely, when the current warning credibility does not reach the scenario-based warning threshold, no warning message will be generated. Instead, new high-frequency meteorological data will continue to be acquired to continuously monitor meteorological evolution and update the current warning credibility, thereby avoiding unnecessary warning issuance.

[0049] Optionally, after sending the warning information to the corresponding warning information recipient, it also includes: Continuously receive new high-frequency meteorological data; According to the preset cycle, the current evolution scenario and precursor indicator change sequence are re-identified based on new high-frequency meteorological data, and the current warning credibility is recalculated. Based on the re-identified current evolution scenario, the sequence of changes in precursor indicators, and the recalculated current warning credibility, new warning elements are generated; among them, the warning elements include warning intensity, scope of impact, and evolution path. The new warning elements are compared with the warning elements in the already sent warning information to obtain the element comparison results. When the element comparison result indicates that the new warning element differs from the warning element in the sent warning information, the warning information update process is triggered. In the early warning information update process, revised early warning information is generated based on the new early warning elements; The revised warning information will be sent again through the original transmission channel; When the element comparison results indicate that the new warning element is no different from the warning element in the already sent warning information, the already sent warning information is maintained.

[0050] Specifically, after the warning information is sent to the recipient, the system will continuously receive new high-frequency meteorological data. High-frequency meteorological data can be understood as at least one of the following: wind speed, wind direction, air pressure, temperature, humidity, radar echo, or satellite observation data, acquired continuously at preset time intervals. Its purpose is to reflect the latest changes in the atmospheric environment in real time.

[0051] Furthermore, the system will, according to a preset cycle, such as every 5, 10, or 15 minutes, re-execute the identification of the current evolution scenario and the extraction of precursor indicator change sequences based on newly received high-frequency meteorological data, and recalculate the current warning confidence level. This process is similar to the initial warning generation process, but uses the latest meteorological data to ensure the real-time nature of the assessment.

[0052] Therefore, based on the re-identified current evolution scenario, the sequence of changes in precursor indicators, and the recalculated current warning confidence, the system generates new warning elements. Warning elements are descriptions of the key characteristics of thunderstorm and gale events, specifically including warning intensity (e.g., gale level, precipitation intensity), impact range (e.g., affected geographical areas), and evolution path (e.g., storm movement direction and speed).

[0053] Subsequently, the newly generated warning elements are compared with the warning elements contained in the previously sent warning messages to obtain the element comparison results. This comparison aims to detect whether there are significant differences between the actual evolution of the thunderstorm and gale event and the initial warning information.

[0054] When the element comparison results indicate that the new warning element differs from the warning element in the previously sent warning information, such as an increase in warning intensity, an expansion of the scope of impact, or a significant deviation in the evolution path, the system will trigger the warning information update process.

[0055] During the early warning information update process, the system will generate revised early warning information based on the new early warning elements. This revised early warning information will contain the latest and more accurate information on thunderstorm and strong wind events.

[0056] As a preferred implementation, the revised warning information will be supplemented and sent to the warning information recipient through the original sending channel to ensure the consistency and timeliness of information transmission. The original sending channel may include SMS, broadcast, APP push, electronic display screen, etc.

[0057] Conversely, when the element comparison results indicate that the new warning elements are no different from the warning elements in the already sent warning information, that is, the evolution of the thunderstorm and gale event is consistent with expectations, the system will maintain the already sent warning information without updating it, thereby avoiding unnecessary duplicate notifications.

[0058] Optionally, when the current precursor matching result indicates that the precursor indicator change sequence is consistent with the precursor pattern under the current evolution scenario, and the current warning confidence reaches the preset warning threshold, the steps for generating warning information include: The system obtains information on the environmental vulnerability of the area where the recipient of the early warning information is located, the current activity intensity, and the response time limit requirements. The response time limit requirements indicate the time required for the recipient of the early warning information to complete risk avoidance, work stoppage, evacuation, or protective measures. Based on environmental vulnerability information, current activity intensity information, and response time limit requirements, assess the risk exposure level of the recipients of the early warning information; Based on the risk exposure level, the preset warning threshold is adjusted to obtain the risk exposure warning threshold; Determine whether the credibility of the current warning has reached the risk exposure warning threshold; When the current warning credibility reaches the risk exposure warning threshold, a warning message is generated, and the risk exposure level and the current warning credibility are written into the warning message; If the current warning credibility does not reach the risk exposure warning threshold, no warning information will be generated, and the high-frequency meteorological data and risk exposure level of the area where the warning information recipient is located will continue to be monitored.

[0059] Specifically, obtaining information on the environmental vulnerability, current activity intensity, and response time limits of the area where the warning information recipient is located aims to comprehensively understand the potential risks to the recipient in the face of thunderstorm and strong wind events. Environmental vulnerability information can be understood as the resilience and recovery capacity of a specific area in the face of natural disasters. This can include, for example, the area's geographical features (such as topography and hydrology), the disaster resistance capacity of infrastructure (such as building structures, power and communication facilities), population density distribution, and historical disaster records, with the aim of quantifying the inherent risks of the area. Current activity intensity information refers to the concentration of people or assets in the area within a specific time period. This can include, for example, real-time population distribution, traffic flow, outdoor activities, or large gatherings, with the aim of reflecting the number of dynamic targets that may be affected when an event occurs. Response time limits indicate the time required for the warning information recipient to complete measures such as risk avoidance, work stoppage, evacuation, or protective measures after receiving the warning, with the aim of determining the urgency of issuing the warning.

[0060] In practical applications, the risk exposure level of the recipient of the early warning information is assessed based on environmental vulnerability information, current activity intensity information, and response time limit requirements. This assessment process can employ a multi-indicator comprehensive evaluation model, such as weighted averaging, fuzzy comprehensive evaluation, or machine learning algorithms, to quantify and integrate the three types of information to obtain a comprehensive risk exposure level. This level can be divided into multiple grades, such as low, medium, high, or more granular levels, with the aim of providing a basis for subsequent adjustments to the early warning threshold.

[0061] Furthermore, the preset warning threshold is adjusted based on the risk exposure level to obtain the risk exposure warning threshold. This means that the warning threshold is no longer fixed, but dynamically adjusted according to the specific risk situation of the recipient. For example, for areas with high risk exposure levels, the preset warning threshold can be appropriately lowered to issue warnings in the early stages of thunderstorm and strong wind events, allowing more response time for high-risk areas; while for areas with low risk exposure levels, the preset warning threshold can remain unchanged or be appropriately increased to avoid unnecessary warnings.

[0062] When the current warning credibility is determined to reach the risk exposure warning threshold, a warning message is generated, and the risk exposure level and the current warning credibility are written into the warning message. Writing the risk exposure level and the current warning credibility together with the warning message aims to provide the warning message recipient with a more comprehensive basis for decision-making, enabling them to take more precise and effective countermeasures based on their specific risk situation and the credibility of the warning. When the current warning credibility does not reach the risk exposure warning threshold, a warning message is not generated, and high-frequency meteorological data and risk exposure levels in the area where the warning message recipient is located continue to be monitored. This measure aims to avoid issuing unnecessary warnings when the risk has not reached a specific level, reduce information overload, and maintain continuous attention to potential risks.

[0063] Optionally, the steps for continuously acquiring high-frequency meteorological data include: Establish a multi-channel data receiving architecture, which includes a primary and backup data link, a data caching module, and a data quality monitoring module. High-frequency meteorological data to be processed is received through the primary and backup data links, and the system switches to the backup link when the primary link fails. The data caching module smooths the data stream when the receiving rate of the high-frequency meteorological data to be processed fluctuates. The data quality monitoring module detects outliers in the high-frequency meteorological data to be processed. When no outliers are detected in the high-frequency meteorological data to be processed, the received high-frequency meteorological data to be processed will be used as high-frequency meteorological data. When an outlier is detected in the high-frequency meteorological data to be processed, a data retransmission request is triggered or data interpolation is performed to obtain the retransmitted or interpolated high-frequency meteorological data to be processed, and the retransmitted or interpolated high-frequency meteorological data to be processed is used as the high-frequency meteorological data.

[0064] Specifically, the multi-channel data reception architecture aims to build a robust data acquisition environment to cope with various complex data transmission and quality issues. The core components of this architecture are primary and backup data links, a data caching module, and a data quality monitoring module. The primary and backup data links provide redundancy for data transmission, ensuring that even if the primary link fails, data reception can seamlessly switch to the backup link, thus maintaining data continuity. For example, the primary link can be a dedicated data channel connected via a fiber optic network, while the backup link can be a satellite communication or cellular network connection to provide an alternative when the primary link is interrupted.

[0065] The data caching module is configured to handle fluctuations in the receiving rate of high-frequency meteorological data to be processed. In actual data transmission, due to factors such as network congestion, changes in server load, or uneven sensor transmission intervals, the data stream may experience instantaneous peaks or troughs. The data caching module smooths the data stream by temporarily storing the received data, ensuring that downstream processing modules can acquire data at a stable rate and avoiding data loss or processing delays caused by data rate mismatch.

[0066] The data quality monitoring module is responsible for real-time detection of the received high-frequency meteorological data to identify potential outliers. Outliers may include erroneous readings caused by sensor malfunctions, data corruption due to transmission errors, or data exceeding physically reasonable limits. This module can use statistical analysis methods (e.g., threshold judgment based on historical data distribution), physical constraint checks (e.g., the rate of temperature change cannot exceed a certain physical limit), or machine learning models to identify anomalous data points.

[0067] When the data quality monitoring module does not detect any outliers in the high-frequency meteorological data to be processed, it indicates that the data quality is good. In this case, the received high-frequency meteorological data to be processed can be directly used as high-frequency meteorological data for subsequent scenario determination and pattern recognition.

[0068] However, when outliers are detected in the high-frequency meteorological data to be processed, the system will trigger a data retransmission request or perform data interpolation. A data retransmission request requests the data source to resend data for a specific time period or from a specific sensor, suitable for situations where data is lost or severely corrupted. Data interpolation, on the other hand, is used when retransmission is impossible or takes too long. Algorithms (e.g., linear interpolation, spline interpolation, predictive interpolation based on historical patterns) estimate and fill in missing or outlier data points based on data from adjacent time or spatial points. Through these two methods, the retransmitted or interpolated high-frequency meteorological data to be processed can be obtained and used as valid high-frequency meteorological data, ensuring the integrity and availability of the data sequence.

[0069] Optionally, after the step of triggering a data retransmission request or performing data interpolation processing when an outlier is detected in the high-frequency meteorological data to be processed, and obtaining the retransmitted or interpolated high-frequency meteorological data to be processed, the method further includes: Consistency verification is performed on the high-frequency meteorological data to be processed after retransmission or interpolation. Consistency verification includes comparing the changing trends between the high-frequency meteorological data to be processed after retransmission or interpolation and the meteorological data at adjacent time points and adjacent spatial points. Physical constraint checks are performed on the high-frequency meteorological data to be processed after retransmission or interpolation. The physical constraint checks include checking the physical correlations between meteorological elements. When both the consistency check and the physical constraint check pass, the high-frequency meteorological data to be processed after retransmission or interpolation is retained. If the consistency check or physical constraint check fails, the high-frequency meteorological data to be processed after retransmission or interpolation is cleared, and the data retransmission request is triggered again or the data interpolation process is re-executed. The consistency check and physical constraint check are then re-executed on the high-frequency meteorological data to be processed after retransmission or re-interpolation.

[0070] Consistency verification aims to assess the reasonableness of the repaired data in terms of time series and spatial distribution. Specifically, by comparing the retransmitted or interpolated high-frequency meteorological data to be processed with meteorological data of the same region at adjacent time points or adjacent regions at the same time point, it can be determined whether its trend conforms to normal physical evolution laws. For example, meteorological elements such as wind speed, air pressure, and temperature usually do not undergo drastic and irregular changes in a short period of time or within a small area. Therefore, if the repaired data shows significant and unreasonable changes compared with adjacent data, it may indicate that there are still problems with the data.

[0071] Furthermore, physical constraint checks are used to verify whether the repaired data satisfies basic physical laws and meteorological principles. Specifically, there are inherent physical relationships between meteorological elements, such as the relationship between temperature, humidity, and air pressure, or a reasonable combination of wind speed and direction. By examining these physical relationships, data that may appear statistically reasonable but actually violate physical laws can be identified. For example, at a specific temperature and air pressure, humidity cannot exceed its saturation value; or the combination of wind speed and direction should not lead to a contradiction in the law of conservation of energy.

[0072] When both the consistency check and the physical constraint check pass, it indicates that the high-frequency meteorological data to be processed after retransmission or interpolation has high reliability and accuracy, can be regarded as valid high-frequency meteorological data, and can be retained for subsequent pattern recognition and prediction.

[0073] However, if either the consistency check or the physical constraint check fails, it indicates that the repaired data still has quality issues. In this case, to avoid forecast bias caused by using erroneous data, the data will be wiped, and the system will trigger a data retransmission request or re-execute the data interpolation process. Subsequently, for the high-frequency meteorological data to be processed after retransmission or re-interpolation, consistency checks and physical constraint checks will be performed again until the data quality is confirmed.

[0074] Optionally, the consistency verification steps include: Read the current evolution scenario information corresponding to the current evolution scenario. The current evolution scenario information includes the typical spatiotemporal evolution scale and local characteristics of thunderstorm and strong wind events. Based on the current evolution scenario information, adjust the spatiotemporal range and verification tolerance of the verification window to perform consistency verification within the adjusted verification window; Within the adjusted verification window, the changing trends of the high-frequency meteorological data to be processed after retransmission or interpolation are compared with those of meteorological data at adjacent time points and adjacent spatial points to obtain the change comparison results. When the change comparison results indicate that the change trend is within the adjusted verification tolerance range, it is determined that the high-frequency meteorological data to be processed after retransmission or interpolation conforms to the characteristics of atmospheric evolution continuity, and it is determined that the high-frequency meteorological data to be processed after retransmission or interpolation passes the consistency verification. When the comparison results indicate that the trend of change is not within the adjusted verification tolerance range, it is determined that the high-frequency meteorological data to be processed after retransmission or interpolation has failed the consistency verification.

[0075] Specifically, current evolution scenario information refers to a predefined or real-time inferred description of thunderstorm and gale event characteristics that matches the current state of atmospheric environmental evolution. This information can include the typical duration and evolution rate of thunderstorm and gale events in the time dimension, as well as the typical impact range, movement path, and scale characteristics in the spatial dimension. It can also include local characteristics related to specific geographical regions or topographic conditions, such as the impact of topography on wind fields, urban heat island effect, and water body effect. This information can be stored in a pre-defined database and loaded when the current evolution scenario is determined.

[0076] The spatiotemporal range of the verification window refers to the length of the time series and the size of the spatial neighborhood considered when performing data consistency comparisons. For example, for fast-moving squall lines, whose spatiotemporal evolution scale is small, the spatiotemporal range of the verification window will be adjusted to be smaller to capture their rapid changes; while for large-scale thunderstorms and strong winds, the spatiotemporal range of the verification window will be appropriately expanded. Verification tolerance refers to the maximum allowable deviation range when comparing data change trends. For example, in severe convective weather, elements such as wind speed and air pressure may change drastically in a short period of time. In this case, the verification tolerance will be appropriately relaxed to avoid misjudging normal drastic changes as abnormal data; while under stable weather conditions, the verification tolerance will be tightened.

[0077] In practical applications, comparing the changing trends of the high-frequency meteorological data to be processed after retransmission or interpolation with meteorological data at adjacent time points and spatial points can be achieved by calculating the gradient of the data sequence, differencing, or using statistical methods such as moving averages. The comparison results are used to quantify the degree of difference between the current data and surrounding data in terms of changing trends. When this degree of difference falls within the adjusted verification tolerance range, it indicates that the data conforms to the continuous characteristics of atmospheric evolution under the current evolution scenario, that is, its changes are in line with physical laws and scenario expectations, and thus it is judged to have passed the consistency verification. Conversely, if it exceeds the tolerance range, the data is considered to be abnormal and fails the consistency verification.

[0078] Optionally, the steps of adjusting the spatiotemporal range and verification tolerance of the verification window based on the current evolution scenario information, so as to perform consistency verification within the adjusted verification window, include: Continuously monitor the evolution of the current atmospheric environment; When the current atmospheric environment does not meet the preset scenario switching conditions, the spatiotemporal range and verification tolerance of the verification window are adjusted according to the current evolution scenario information, so as to perform consistency verification on the high-frequency meteorological data to be processed after retransmission or interpolation within the adjusted verification window. When the current atmospheric environment evolution meets the preset scenario switching conditions, the system identifies the change in the evolution scenario of thunderstorm and strong wind events and obtains information on the new evolution scenario after the switch. Based on the new evolution scenario information, the spatiotemporal range and verification tolerance of the verification window are adjusted so as to perform consistency verification on the high-frequency meteorological data to be processed after retransmission or interpolation within the adjusted verification window.

[0079] Specifically, "continuously monitoring the evolution of the current atmospheric environment" refers to the system acquiring and analyzing the latest meteorological data in real time, such as at least one of wind speed, wind direction, air pressure, temperature, humidity, radar echo, or satellite observation data, to assess the dynamic trends and characteristics of the current atmospheric environment. This monitoring can be conducted using machine learning models, statistical analysis methods, or physical model simulations. The "preset scenario switching conditions" can be understood as a series of predefined meteorological indicator thresholds or patterns. When the evolution of the current atmospheric environment meets these conditions, it indicates that the evolution scenario of the thunderstorm and strong wind event may have changed significantly. For example, a rapid increase in radar echo intensity accompanied by wind field convergence characteristics, or a sudden and sustained near-surface wind speed reaching a certain critical value, may be considered a trigger condition for scenario switching. In practical applications, "identifying the shift in the evolution scenario of the thunderstorm and strong wind event" means that the system, based on the comparison between the monitored evolution state and the preset scenario switching conditions, determines whether the evolution path or type of the current thunderstorm and strong wind event has changed from one scenario to another. This may involve pattern recognition or classification of new meteorological data. "New evolution scenario information" refers to the information acquired by the system after a scenario switch is detected, such as the typical spatiotemporal evolution scale and local characteristics corresponding to the new evolution scenario. This information will be used to guide the subsequent adjustment of the verification window and tolerance.

[0080] This application also proposes a meteorological sequence pattern recognition system for thunderstorm and gale prediction, which is used to perform the above-mentioned meteorological sequence pattern recognition method for thunderstorm and gale prediction.

[0081] Please refer to Figure 3 The meteorological sequence pattern recognition system 1 for thunderstorm and strong wind forecasting includes: The initial meteorological acquisition module 11 is used to acquire initial meteorological data, which includes meteorological observation information and short-term forecast information. The evolution scenario inference module 12 is used to infer candidate evolution scenarios of the future atmospheric environment based on initial meteorological data, and to configure corresponding scenario recognition strategies for different candidate evolution scenarios. The scenario recognition strategies include precursor pattern recognition rules and early warning credibility calculation and adjustment rules. The evolution scenario determination module 13 is used to continuously acquire high-frequency meteorological data and determine the current evolution scenario from candidate evolution scenarios based on the spatiotemporal evolution characteristics of the high-frequency meteorological data. The data processing execution module 14 is used to load the scenario recognition strategy corresponding to the current evolution scenario, extract the precursor index change sequence from high-frequency meteorological data, and determine the precursor matching result and the initial warning credibility according to the precursor pattern recognition rules. The early warning credibility determination module 15 is used to adjust the initial early warning credibility according to the early warning credibility calculation and adjustment rules and the current evolution scenario to obtain the current early warning credibility. The early warning information generation module 16 is used to generate early warning information when the current early warning matching result indicates that the change sequence of the precursor index is consistent with the precursor pattern under the current evolution scenario, and the current early warning credibility reaches the preset early warning threshold.

[0082] The initial meteorological acquisition module 11 is configured to acquire initial meteorological data, which may include meteorological observation information from ground observation stations, upper-air sounding, satellite remote sensing, and other channels, as well as short-term forecast information generated by numerical weather prediction models. This data provides the foundation for subsequent inferences about atmospheric environmental evolution scenarios.

[0083] The evolution scenario inference module 12 is configured to use initial meteorological data and machine learning models or expert systems to infer multiple candidate evolution scenarios for potential future thunderstorm and strong wind processes in the atmospheric environment. Simultaneously, this module will configure a specific scenario recognition strategy for each candidate evolution scenario. This strategy includes precursor pattern recognition rules for identifying precursor patterns and warning confidence calculation and adjustment rules for calculating and adjusting warning confidence.

[0084] The evolution scenario determination module 13 is configured to continuously receive high-frequency meteorological data, such as at least one of wind speed, wind direction, air pressure, temperature, humidity, radar echo, or satellite observation data acquired continuously at minute or second intervals. This module analyzes the spatiotemporal evolution characteristics of these high-frequency meteorological data and compares them with preset candidate evolution scenarios to dynamically determine the current evolution scenario of the atmospheric environment.

[0085] The data processing execution module 14 is configured to load a corresponding scenario recognition strategy after determining the current evolution scenario. This module extracts the index change sequences related to thunderstorm and gale precursors from continuously acquired high-frequency meteorological data, and determines whether these sequences are consistent with the precursor patterns under the current evolution scenario based on precursor pattern recognition rules, thereby determining the precursor matching results and the credibility of the initial warning.

[0086] The early warning credibility determination module 15 is configured to dynamically adjust the initial early warning credibility obtained by the data processing execution module 14 according to the early warning credibility calculation and adjustment rules and the real-time status of the current evolution scenario, so as to obtain a more accurate and reliable current early warning credibility.

[0087] The early warning information generation module 16 is configured to generate early warning information when specific conditions are met. Specifically, when the precursor matching result determined by the data processing execution module 14 indicates that the precursor indicator change sequence is consistent with the precursor pattern under the current evolution scenario, and the current early warning credibility obtained by the early warning credibility determination module 15 reaches the preset early warning threshold, this module will trigger the generation of early warning information.

[0088] Through the above technical solution, this application provides a meteorological sequence pattern recognition system for thunderstorm and gale forecasting. This system can implement complex meteorological sequence pattern recognition methods in a structured and modular manner. This systematic design helps improve the automation level and operational efficiency of the forecasting process, ensuring that key steps such as data processing, scenario recognition, precursor matching, and early warning credibility assessment can be completed quickly and accurately when faced with massive amounts of real-time meteorological data. Furthermore, the clear division of responsibilities among modules makes system maintenance and upgrades more convenient, thus providing solid hardware and software support for accurate forecasting and timely early warning of thunderstorms and gales, effectively enhancing meteorological disaster prevention capabilities.

[0089] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A meteorological sequence pattern recognition method for thunderstorm and strong wind forecasting, characterized in that, include: Acquire initial meteorological data, which includes meteorological observation information and short-term forecast information; Based on the initial meteorological data, candidate evolution scenarios of the future atmospheric environment are inferred, and corresponding scenario identification strategies are configured for different candidate evolution scenarios. The scenario identification strategies include precursor pattern recognition rules and early warning credibility calculation and adjustment rules. The candidate evolution scenarios represent the evolution path or evolution type of the future atmospheric environment forming thunderstorm and strong wind processes. Continuously acquire high-frequency meteorological data, and determine the current evolution scenario from the candidate evolution scenarios based on the spatiotemporal evolution characteristics of the high-frequency meteorological data; the high-frequency meteorological data includes at least one of wind speed, wind direction, air pressure, temperature, humidity, radar echo or satellite observation data acquired continuously at preset time intervals; Load the scenario recognition strategy corresponding to the current evolution scenario, extract the precursor index change sequence from the high-frequency meteorological data, and determine the precursor matching result and the initial warning credibility according to the precursor pattern recognition rule; Based on the aforementioned early warning credibility calculation and adjustment rules and the current evolution scenario, the initial early warning credibility is adjusted to obtain the current early warning credibility. When the precursor matching result indicates that the precursor indicator change sequence is consistent with the precursor pattern under the current evolution scenario, and the current warning credibility reaches the preset warning threshold, a warning message is generated.

2. The meteorological sequence pattern recognition method for thunderstorm and strong wind forecasting according to claim 1, characterized in that, The step of adjusting the initial warning credibility according to the warning credibility calculation and adjustment rules and the current evolution scenario to obtain the current warning credibility includes: Continuously monitor the scenario consistency between the high-frequency meteorological data and the current evolving scenario to obtain scenario consistency monitoring results; When the scenario consistency monitoring result indicates that the high-frequency meteorological data is continuously inconsistent with the current evolving scenario, the background correction process is initiated. In the background correction process, local triggering factors are extracted from the high-frequency meteorological data. These local triggering factors include local convergence enhancement features, near-surface wind direction change features, humidity increase features, or temperature decrease features. Based on the local triggering factors, the current evolution scenario is modified to obtain the modified current evolution scenario; Based on the revised current evolution scenario, the scenario recognition strategy is adjusted to obtain the adjusted scenario recognition strategy. Based on the adjusted scenario recognition strategy, the initial warning confidence of the precursor indicator change sequence is recalculated to obtain the corrected confidence. Determine the current warning credibility based on the revised credibility. When the scenario consistency monitoring result indicates that the high-frequency meteorological data and the current evolution scenario are not continuously inconsistent, the initial warning credibility is adjusted according to the warning credibility calculation and adjustment rules to obtain the current warning credibility.

3. The meteorological sequence pattern recognition method for thunderstorm and strong wind forecasting according to claim 1, characterized in that, The step of generating warning information when the precursor matching result indicates that the precursor indicator change sequence is consistent with the precursor pattern under the current evolution scenario, and the current warning confidence reaches a preset warning threshold, includes: The application scenario categories for obtaining early warning information recipients; Based on the application scenario category, select and load the early warning release strategy, which includes scenario threshold adjustment rules and early warning information content customization rules; According to the scenario threshold adjustment rules, the preset warning threshold is adjusted to obtain the scenario-based warning threshold; Determine whether the current warning credibility reaches the scenario-based warning threshold; When the current warning credibility reaches the scenario-based warning threshold, warning information is generated according to the warning information content customization rules, and the warning information is sent to the corresponding warning information recipient; When the current warning confidence level does not reach the scenario-based warning threshold, no warning information is generated, and new high-frequency meteorological data continues to be acquired to update the current warning confidence level.

4. The meteorological sequence pattern recognition method for thunderstorm and strong wind forecasting according to claim 3, characterized in that, After sending the warning information to the corresponding warning information recipient, the process also includes: Continuously receive new high-frequency meteorological data; According to the preset cycle, the current evolution scenario and precursor indicator change sequence are re-identified based on new high-frequency meteorological data, and the current warning credibility is recalculated. Based on the re-identified current evolution scenario, the sequence of changes in precursor indicators, and the recalculated current warning confidence, new warning elements are generated; wherein, the warning elements include warning intensity, scope of impact, and evolution path; The new warning elements are compared with the warning elements in the already sent warning information to obtain the element comparison results. When the element comparison result indicates that the new warning element differs from the warning element in the sent warning information, the warning information update process is triggered. In the aforementioned early warning information update process, revised early warning information is generated based on the new early warning elements; The revised warning information will be sent again through the original transmission channel; When the element comparison result indicates that the new warning element is no different from the warning element in the previously sent warning information, the previously sent warning information is maintained.

5. The meteorological sequence pattern recognition method for thunderstorm and strong wind prediction according to claim 1, characterized in that, The step of generating warning information when the precursor matching result indicates that the precursor indicator change sequence is consistent with the precursor pattern under the current evolution scenario, and the current warning confidence reaches a preset warning threshold, includes: The system acquires information on the environmental vulnerability of the area where the recipient of the early warning information is located, the current activity intensity, and the response time limit requirements. The response time limit requirements indicate the time required for the recipient of the early warning information to complete risk avoidance, work stoppage, evacuation, or protective measures. Based on the environmental vulnerability information, the current activity intensity information, and the response time limit requirement information, assess the risk exposure level of the recipient of the early warning information; Based on the risk exposure level, the preset warning threshold is adjusted to obtain the risk exposure warning threshold; Determine whether the current warning credibility reaches the risk exposure warning threshold; When the current warning confidence level reaches the risk exposure warning threshold, a warning message is generated, and the risk exposure level and the current warning confidence level are written into the warning message; When the current warning credibility does not reach the risk exposure warning threshold, no warning information is generated, and the high-frequency meteorological data and risk exposure level of the area where the warning information recipient is located continue to be monitored.

6. The meteorological sequence pattern recognition method for thunderstorm and strong wind prediction according to claim 1, characterized in that, The steps for continuously acquiring high-frequency meteorological data include: A multi-channel data receiving architecture is established, which includes a primary and backup data link, a data caching module, and a data quality monitoring module. The system receives high-frequency meteorological data to be processed through the primary and backup data links, and switches to the backup link when the primary link fails. The data caching module smooths the data stream when the receiving rate of the high-frequency meteorological data to be processed fluctuates. The data quality monitoring module detects abnormal values ​​in the high-frequency meteorological data to be processed. When no abnormal values ​​are detected in the high-frequency meteorological data to be processed, the received high-frequency meteorological data to be processed will be used as high-frequency meteorological data. When an abnormal value is detected in the high-frequency meteorological data to be processed, a data retransmission request is triggered or data interpolation processing is performed to obtain the retransmitted or interpolated high-frequency meteorological data to be processed, and the retransmitted or interpolated high-frequency meteorological data to be processed is used as high-frequency meteorological data.

7. A meteorological sequence pattern recognition method for thunderstorm and strong wind prediction according to claim 6, characterized in that, After the step of triggering a data retransmission request or performing data interpolation processing when an abnormal value is detected in the high-frequency meteorological data to be processed, to obtain the retransmitted or interpolated high-frequency meteorological data to be processed, the method further includes: A consistency check is performed on the high-frequency meteorological data to be processed after retransmission or interpolation. The consistency check includes comparing the changing trends between the high-frequency meteorological data to be processed after retransmission or interpolation and the meteorological data at adjacent time points and adjacent spatial points. Physical constraint checks are performed on the high-frequency meteorological data to be processed after retransmission or interpolation, and the physical constraint checks include checking the physical correlations between meteorological elements. When both the consistency check and the physical constraint check pass, the high-frequency meteorological data to be processed after retransmission or interpolation is retained. If the consistency check or the physical constraint check fails, the high-frequency meteorological data to be processed after retransmission or interpolation is cleared, and the data retransmission request is triggered again or the data interpolation process is re-executed. The consistency check and physical constraint check are then re-executed on the high-frequency meteorological data to be processed after retransmission or re-interpolation.

8. A meteorological sequence pattern recognition method for thunderstorm and strong wind prediction according to claim 7, characterized in that, The consistency verification steps include: Read the current evolution scenario information corresponding to the current evolution scenario, which includes the typical spatiotemporal evolution scale and local characteristics of thunderstorm and gale events; Based on the current evolution scenario information, the spatiotemporal range and verification tolerance of the verification window are adjusted so that consistency verification can be performed within the adjusted verification window. Within the adjusted verification window, the changing trends of the high-frequency meteorological data to be processed after retransmission or interpolation are compared with those of meteorological data at adjacent time points and adjacent spatial points to obtain the change comparison results. When the change comparison result indicates that the change trend is within the adjusted verification tolerance range, it is determined that the high-frequency meteorological data to be processed after retransmission or interpolation conforms to the atmospheric evolution continuity characteristics, and it is determined that the high-frequency meteorological data to be processed after retransmission or interpolation passes the consistency verification. When the change comparison result indicates that the change trend is not within the adjusted verification tolerance range, it is determined that the high-frequency meteorological data to be processed after retransmission or interpolation has failed the consistency verification.

9. A meteorological sequence pattern recognition method for thunderstorm and strong wind prediction according to claim 8, characterized in that, The step of adjusting the spatiotemporal range and verification tolerance of the verification window based on the current evolution scenario information, so as to perform consistency verification within the adjusted verification window, includes: Continuously monitor the evolution of the current atmospheric environment; When the current atmospheric environment does not meet the preset scenario switching conditions, the spatiotemporal range and verification tolerance of the verification window are adjusted according to the current evolution scenario information, so as to perform consistency verification on the high-frequency meteorological data to be processed after retransmission or interpolation within the adjusted verification window. When the current atmospheric environment evolution meets the preset scenario switching conditions, the system identifies the change in the evolution scenario of thunderstorm and strong wind events and obtains information on the new evolution scenario after the switch. Based on the new evolution scenario information, the spatiotemporal range and verification tolerance of the verification window are adjusted so as to perform consistency verification on the high-frequency meteorological data to be processed after retransmission or interpolation within the adjusted verification window.

10. A meteorological sequence pattern recognition system for thunderstorm and gale prediction, used to execute a meteorological sequence pattern recognition method for thunderstorm and gale prediction as described in any one of claims 1-9, characterized in that, include: The initial meteorological acquisition module is used to acquire initial meteorological data, which includes meteorological observation information and short-term forecast information; The evolution scenario inference module is used to infer candidate evolution scenarios of the future atmospheric environment based on the initial meteorological data, and to configure corresponding scenario recognition strategies for different candidate evolution scenarios. The scenario recognition strategies include precursor pattern recognition rules and early warning credibility calculation and adjustment rules. The evolution scenario determination module is used to continuously acquire high-frequency meteorological data and determine the current evolution scenario from the candidate evolution scenarios based on the spatiotemporal evolution characteristics of the high-frequency meteorological data. The data processing execution module is used to load the scenario recognition strategy corresponding to the current evolution scenario, extract the precursor index change sequence from the high-frequency meteorological data, and determine the precursor matching result and the initial warning credibility according to the precursor pattern recognition rule. The early warning credibility determination module is used to adjust the initial early warning credibility according to the early warning credibility calculation and adjustment rules and the current evolution scenario to obtain the current early warning credibility. The early warning information generation module is used to generate early warning information when the early warning matching result indicates that the sequence of changes in the early warning indicators is consistent with the early warning pattern under the current evolution scenario, and the credibility of the current early warning reaches a preset early warning threshold.

Images