Tropical cyclone forecasting method and tropical cyclone forecasting system using ensemble forecasting data

Abstract

The invention relates to the technical field of intelligent weather forecasting, and discloses a tropical cyclone forecasting method and system using ensemble forecasting data, and the method comprises the steps: obtaining and fusing multi-source data, and forming a cyclone form and path feature set representing a tropical cyclone state; based on the feature set, parameters are dynamically corrected in combination with the sea temperature field and vertical wind shear data, and a comprehensive feature data set is generated; synchronously evaluating the cyclonic form stability and the path deviation trend by using the data set, and generating preliminary risk information; fusing the risk information and forecast data updated in real time, and dividing an initial risk level; performing deviation checking on the path, and dynamically adjusting a parameter correction strategy to optimize the risk level of the path; and finally, based on the optimized risk level, integrating multi-source information, updating morphological parameters, and feeding back to the initial step to form closed-loop data processing and forecast iteration. The tropical cyclone forecasting method improves the accuracy and timeliness of tropical cyclone forecasting.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent weather forecasting, in particular to a tropical cyclone forecasting method and system using ensemble forecasting data. BACKGROUND

[0002] Tropical cyclones are major meteorological disasters affecting coastal areas around the world, and the strong winds, heavy rain and storm surges they bring pose a serious threat to people's lives and property safety and social and economic operation. Therefore, achieving accurate prediction of their path and intensity has always been a core issue in the field of meteorology. For a long time, meteorological workers have relied on satellite remote sensing, numerical prediction and other multi-source data for analysis, but existing prediction methods still have significant limitations when faced with such extreme weather systems.

[0003] Traditional methods usually focus on the output of a single data source or static model, making it difficult to comprehensively utilize multi-dimensional information such as shape, path and environmental field, resulting in insufficient description of the correlation between the internal structure evolution of the cyclone and the external moving track. In addition, even if multi-source data such as ensemble forecasting is introduced, there is often a lack of effective real-time fusion and dynamic correction mechanism, which cannot timely adjust the prediction conclusion when the cyclone rapidly intensifies or the path suddenly changes, resulting in a lag in the timeliness of the warning. More importantly, the existing technical process is mostly open-loop processing, and the prediction result is output and then terminated, without the ability to feedback and optimize its parameters according to the prediction deviation and the latest actual situation, so the stability and scientificity of its risk assessment cannot be guaranteed under the influence of complex and variable environmental fields such as sea surface temperature and wind shear.

[0004] The above-mentioned problems such as rigid data utilization, poor dynamic adaptability and lack of system self-optimization capability jointly restrict the further improvement of the accuracy and reliability of tropical cyclone prediction, and an intelligent prediction new method that can penetrate data fusion, dynamic correction, real-time evaluation and closed-loop feedback is urgently needed. SUMMARY

[0005] To solve the above technical problems, the present application provides a tropical cyclone forecasting method and system using ensemble forecasting data, which is used to improve the accuracy and timeliness of cyclone risk assessment and provide a scientific basis for disaster warning.

[0006] In a first aspect, the present application provides a tropical cyclone forecasting method using ensemble forecasting data, which comprises: Step S1: acquiring and fusing multi-source data to form a cyclone shape and path feature set representing the state of the tropical cyclone; Step S2: based on the cyclone shape and path feature set, combining sea surface temperature field and vertical wind shear data to perform dynamic correction of parameters and generate a comprehensive feature data set; Step S3: Utilize the comprehensive feature dataset to synchronously evaluate the cyclone shape stability and path deviation trend, and generate preliminary risk information; Step S4: Fuse the preliminary risk information and real-time updated forecast data to determine the dynamic change of path deviation, and divide the initial risk level; Step S5: Deviation check is performed on the path corresponding to the initial risk level, and the parameter correction strategy is dynamically adjusted based on the check result to optimize the path risk level; Step S6: According to the optimized path risk level, a new round of multi-source information is integrated to update the cyclone shape parameters, and the updated parameters are fed back to the feature forming step to complete the closed-loop data processing and prediction iteration.

[0007] In a second aspect, the application provides a tropical cyclone prediction system using ensemble prediction data, the system comprising: A data fusion and feature generation module is configured to obtain and fuse multi-source data to form a cyclone shape and path feature set representing the state of a tropical cyclone; A parameter correction and feature optimization module is configured to perform dynamic correction of parameters based on the cyclone shape and path feature set, in combination with sea surface temperature field and vertical wind shear data, to generate a comprehensive feature dataset; A synchronous evaluation and risk preliminary judgment module is configured to utilize the comprehensive feature dataset to synchronously evaluate the cyclone shape stability and path deviation trend, and generate preliminary risk information; A real-time fusion and risk division module is configured to fuse the preliminary risk information and real-time updated forecast data to determine the dynamic change of path deviation, and divide the initial risk level; A deviation feedback and correction optimization module is configured to perform deviation check on the path corresponding to the initial risk level, and dynamically adjust the parameter correction strategy based on the check result to optimize the path risk level; A closed-loop update and iteration driving module is configured to integrate a new round of multi-source information to update the cyclone shape parameters according to the optimized path risk level, and feed back the updated parameters to the feature forming step to complete the closed-loop data processing and prediction iteration.

[0008] Compared with the prior art, the application has at least the following advantages: 1. The technical scheme provided by the application constructs a tropical cyclone prediction system with multi-source fusion, dynamic correction and closed-loop iteration, which significantly improves the prediction accuracy, timeliness and scientificity of risk assessment. First, by fusing multi-source heterogeneous data such as satellite remote sensing, ensemble prediction and global numerical model, and using distributed synchronous technology to form a unified feature set, the consistency and completeness of the data are improved from the source, laying a reliable foundation for subsequent analysis, and directly improving the accuracy of path and intensity prediction.

[0009] 2. By introducing an environmental variable dynamic correction mechanism based on sea surface temperature field and vertical wind shear, and triggering a secondary evaluation by threshold in the evaluation, adjusting the path weight combined with wind shear, the system can respond to the subtle changes of the cyclone structure and environmental field in real time, thereby greatly enhancing the timeliness of the forecast, especially in the scene of rapid development or path mutation of the cyclone, the warning information can be issued faster.

[0010] 3. By synchronously evaluating the shape stability and path deviation trend, and using interaction analysis, deviation checking and other means, the risk judgment which is traditionally dependent on experience is converted into data-driven quantification, so that the risk division result is more scientific and has better interpretability, and the credibility of decision support is improved.

[0011] 4. By using the deviation feedback mechanism to dynamically adjust the environmental variable correction weight, and based on the optimized risk level to drive the backtracking update of the shape parameter, a closed loop of "evaluation-correction-verification-optimization" is formed, so that the whole prediction system has self-learning and continuous optimization adaptive ability, and the stability and prediction performance of long-term operation are continuously enhanced; the present application not only effectively solves the problems of isolated data utilization, lagging warning and strong subjectivity of risk assessment in the prior art, but also provides a complete technical solution with high precision, high timeliness and strong adaptability for tropical cyclone prediction business through the closed-loop intelligent processing architecture. BRIEF DESCRIPTION OF DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0013] Figure 1 A step flow chart of a tropical cyclone prediction method using ensemble prediction data in an embodiment of the present application; Figure 2 An impact analysis diagram of environmental variables on path deviation in an embodiment of the present application; Figure 3 A comparison diagram of prediction accuracy of different methods in an embodiment of the present application; Figure 4 A composition structure diagram of a tropical cyclone prediction system using ensemble prediction data in an embodiment of the present application. DETAILED DESCRIPTION

[0014] The embodiments of the present application provide a tropical cyclone forecasting method and system using ensemble forecasting data. The terms "first", "second", "third", "fourth" and the like (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" or "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to only those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0015] Embodiment one: The existing tropical cyclone forecasting method cannot accurately and timely evaluate the influence of path deviation and shape change when facing rapid changes in cyclone path and shape, resulting in deviations between the forecasting results and the actual situation, especially under variable environmental conditions. The existing technology lacks effective dynamic evaluation means for cyclone shape stability and path deviation trend. The existing meteorological forecasting method usually relies on a single data source, lacks comprehensive utilization and real-time updating of multi-source data, resulting in insufficient accuracy and timeliness of path prediction. In addition, there are problems such as poor consistency of data sources, untimely dynamic updating in the process of meteorological data fusion and parameter correction, making it difficult to achieve accurate fusion of path and shape characteristics, thereby affecting the accuracy and reliability of tropical cyclone forecasting. Therefore, the present application proposes a method of multi-source data fusion, real-time data updating and path risk evaluation optimization, aiming to improve the accuracy and timeliness of tropical cyclone forecasting, especially when the cyclone shape changes rapidly or the path mutates, to provide more accurate and scientific early warning information.

[0016] For ease of understanding, the specific process of the embodiments of the present application is described below. Please refer to Figure 1 The tropical cyclone forecasting method using ensemble forecasting data in the embodiments of the present application comprises: Step S1: Obtain and fuse multi-source data to form a cyclone shape and path feature set representing the state of the tropical cyclone.

[0017] Among them, step S1 further comprises: based on satellite remote sensing images, quantitatively extracting eye wall density, cloud system distribution uniformity, core structure compactness and cloud band symmetry as shape key parameters; obtaining central pressure data and path probability distribution information as path features from a meteorological ensemble forecasting system and a global numerical forecasting system; using distributed node synchronization technology, real-time fusion of shape key parameters, central pressure data and path probability distribution information to generate a cyclone shape and path feature set.

[0018] Specifically, to solve the problem that the existing tropical cyclone prediction method cannot fully combine multi-source data and realize efficient fusion of path and morphological characteristics, the application provides an accurate and efficient tropical cyclone prediction method by using the multi-source information of ensemble prediction data, to enhance the stability and accuracy of path prediction, especially when the cyclone morphological changes rapidly or the path deviates, it can quickly respond and adjust in time, as follows.

[0019] In the embodiment, the real-time morphological information of the cyclone is extracted using satellite remote sensing images, and quantitative analysis is performed on the eye wall density and cloud system distribution uniformity, and key parameters are extracted in combination with the core structure compactness and cloud band symmetry, wherein the key parameters refer to a plurality of quantitative indexes related to the cyclone morphology extracted by satellite remote sensing images, specifically including an eye wall density quantitative value, a cloud system distribution uniformity parameter, a compactness index calculated in combination with the core structure compactness, and symmetry calculation based on the uniformity of cloud band distribution in four quadrants; the satellite image data captures the cloud top brightness temperature and structural details of the cyclone through visible and infrared channels, and in processing, the image is segmented into an eye wall region, the density distribution of pixels is calculated to quantify the density of the eye wall, and the eye wall density value is defined as the proportion of cloud cluster pixels in the eye wall region exceeding a first numerical proportion, thereby quantifying the tightness of the eye wall; the core structure compactness identifies the eye wall boundary through an edge detection algorithm, calculates the compactness index of the cloud system within the boundary, further analyzes the cloud band symmetry, and calculates the uniformity of the cloud band distribution in the four quadrants of the cyclone, and if the difference in cloud coverage area in each quadrant does not exceed a first threshold value, the symmetry is good; the key parameters: eye wall density, core structure compactness and cloud band symmetry are quantitative data obtained based on satellite remote sensing image analysis, and then used for path feature fusion processing, by extracting these parameters, the morphological characteristics of the cyclone can be more accurately described, providing a reliable data basis for subsequent path prediction and risk assessment.

[0020] The central pressure data and path probability distribution information are obtained by interfacing with a meteorological ensemble prediction interface and a global numerical prediction system. Specifically, the minimum pressure data and path probability distribution information for the next 24 hours are collected by interfacing with a meteorological ensemble prediction system in real time through a standard API interface. The path probability distribution information includes multiple possible path trajectories of the cyclone in the future and the probability value of each path, representing the likelihood of different paths and the uncertainty of the path. These data are used as supplementary inputs and are combined with the morphological data to enter the data fusion processing stage. The fusion of data is realized by distributed node synchronization technology, ensuring that the multi-source data of key parameters from satellite images, central pressure data, and path probability distribution information can be efficiently and synchronously processed. In this process, each computing node is responsible for processing a subset of data, ensuring data synchronization consistency through a consensus algorithm, and combining morphological parameters with path probability data using a weighted average technique to generate an initial set of cyclone morphology and path characteristics.

[0021] In the data fusion process, distributed node synchronization technology is used to distribute the key parameters from satellite images, central pressure data, and path probability distribution information to different nodes for processing. Each node aligns the data timestamps according to the time synchronization protocol to ensure the temporal consistency of the data. Then, the cyclone morphology and path data are combined through vector splicing to form a unified feature set. Consistency verification is performed to ensure the accuracy and integrity of the data, and the variance of the data in the set is calculated to ensure that it does not exceed a predetermined threshold, ensuring data consistency. This technical means can effectively avoid the bias caused by different data sources, ensuring the efficiency and accuracy of data processing. Finally, by fusing multiple data sources, an initial feature set is generated, which serves as an important basis for subsequent path bias correction and optimization.

[0022] For example, during the monitoring of a certain cyclone, satellite images show that the eye wall intensity quantization value is 0.85, the cloud system uniformity is 0.72, the core structure compactness parameter is 0.90, and the cloud band symmetry is 0.88. These values are obtained through the above quantitative steps, ensuring the real-time and reliability of the morphological data and providing high-precision data support. In the monitoring of a strong typhoon, the consistency of the cyclone path probability reaches 95% through distributed node synchronization and weighted average technology, effectively improving the stability and prediction accuracy of path monitoring.

[0023] The embodiment successfully combines information of multiple data sources efficiently by fusing satellite remote sensing images, meteorological ensemble prediction data and global numerical prediction data and a distributed node synchronization technology, and provides more accurate and reliable technical means for real-time monitoring and path prediction of tropical cyclones; not only improves the stability of path prediction, but also provides more scientific and reasonable data support for cyclone prediction, so as to realize accurate evaluation and optimization of path deviation and cyclone morphological change; the comprehensive application of these technical means effectively solves the problems of inaccurate cyclone morphological feature acquisition and path deviation prediction and early warning lag in the prior art, and further promotes the development of tropical cyclone prediction technology.

[0024] Step S2: Based on the cyclone morphological and path feature set, the parameter dynamic correction is performed in combination with the sea surface temperature field and vertical wind shear data to generate a comprehensive feature data set.

[0025] In the step S2, the sea surface temperature field distribution data and the vertical wind shear intensity data related to the current cyclone are obtained according to the cyclone morphological and path feature set; the parameters in the cyclone morphological and path feature set are physically corrected by using the sea surface temperature field distribution data and the vertical wind shear intensity data; the corrected parameters are integrated to generate a comprehensive feature data set containing cyclone structure description and path trend information through real-time data transmission and consistency verification.

[0026] Specifically, to solve the problem of lack of effective parameter correction means in the prior art tropical cyclone prediction method, especially the insufficient accuracy of cyclone path and morphological prediction under variable environmental conditions, the present application uses sea surface temperature field distribution and vertical wind shear intensity data to correct parameters by combining multiple data sources, thereby improving the accuracy and stability of cyclone path prediction, as follows.

[0027] Firstly, the current sea surface temperature field distribution data is extracted according to the morphological parameters in the initial feature set, such as eye wall density and cloud system distribution uniformity; the influence of sea surface temperature field distribution on cyclones mainly lies in the influence of sea surface temperature change on cyclone intensity, so the sea surface temperature gradient is analyzed by satellite remote sensing image to quantify this influence; when the sea surface temperature is higher than the average value, the cyclone intensity related parameters are correspondingly adjusted to correct the intensity of the cyclone; secondly, the path prediction parameters are further adjusted in combination with the vertical wind shear intensity data; the wind shear intensity refers to the difference in wind speed between different height layers, which is obtained through a meteorological ensemble prediction interface and applied to the correction of path probability distribution; through the above method, the corrected parameters can more accurately reflect the change trend of the cyclone under different environmental conditions, thereby improving the reliability of path prediction, wherein the parameters corrected include path deviation trend, eye wall density and cloud system distribution uniformity and other morphological features, which are adjusted in combination with the sea surface temperature field distribution and the vertical wind shear intensity data to ensure the accuracy and stability of cyclone path prediction.

[0028] After the parameter correction is completed, the corrected information is integrated with other environmental variable data such as central pressure data through a data real-time transmission mechanism; this process utilizes distributed node synchronization technology to ensure the real-time and consistency of the data, avoiding prediction errors caused by delays; specifically, through distributed node synchronization, the corrected morphological parameters and meteorological variable data are integrated onto the same platform, and the synchronization between the various data subsets is ensured; through this method, efficient fusion between different data sources can be achieved, ensuring the timeliness of data processing.

[0029] After integration, data consistency verification is performed to ensure the accuracy of the corrected cyclone morphology and path feature data by comparing the matching degree of multi-source data; in this process, if temporal and spatial inconsistencies are found in the sea surface temperature field data and wind shear data, secondary data fusion is triggered to ensure the consistency and integrity of the final data set; after verification, the corrected cyclone structure description and path trend preliminary information generated are used as the basis for the comprehensive feature data set for subsequent path risk assessment, providing reliable support for accurate prediction of cyclone paths and avoiding risk misjudgments caused by data inconsistency.

[0030] For example, in a scenario where a tropical cyclone is approaching land, assuming that the initial feature set shows an eye wall density quantization value of 0.8 and the sea surface temperature gradient is calculated to be 2 degrees Celsius per hundred kilometers through remote sensing data, at this time, the intensity parameter of the cyclone is adjusted upward by 10% through correction, which helps to more accurately predict the strengthening trend of the cyclone and thus improve the accuracy of path prediction; for the correction of vertical wind shear intensity, assuming that the cyclone cloud system distribution uniformity is low, if the wind shear intensity is 5 meters per second, the path probability weight will be adjusted downward, further reducing path deviation and thus ensuring the accuracy and timeliness of warning information.

[0031] In the data consistency verification process, if the matching degree is below a certain value, a secondary fusion step will be triggered; this ensures the reliability of the final generated cyclone structure description and path trend preliminary information, preventing false predictions caused by data inconsistency; through the above technical means, the present application can effectively solve the problems of inaccuracy in cyclone path prediction and data processing lag in the prior art, greatly improving the accuracy and stability of tropical cyclone path prediction and providing strong data support for disaster warning and emergency management.

[0032] Step S3: Utilize the comprehensive feature data set to synchronously evaluate cyclone morphological stability and path deviation trend, generating preliminary risk information.

[0033] The step S3 further comprises: if the eye wall density quantitative value in the comprehensive feature data set exceeds a preset threshold, triggering secondary evaluation of cloud system distribution uniformity and core structure compactness; combining vertical wind shear intensity information, adjusting the weight configuration of the path probability distribution; based on the secondary evaluation result and the adjusted weight, determining the stability level of the cyclone shape and the preliminary risk level of the path deviation, to form the preliminary risk information.

[0034] Specifically, to solve the problem of inaccuracy and untimeliness of cyclone shape stability and path deviation trend evaluation in existing tropical cyclone prediction methods, especially in the case of rapid change of cyclone shape or path deviation, the existing methods cannot efficiently and accurately adjust the predicted path and risk assessment; through the comprehensive feature data set and multi-source data fusion technology, the present application provides a method for effectively evaluating the stability of cyclone shape and path deviation trend, which can greatly improve the accuracy and timeliness of cyclone path prediction, as follows.

[0035] In the present embodiment, the stability of cyclone shape and path deviation trend are evaluated through the comprehensive feature data set to generate preliminary risk information; specifically, if the eye wall density quantitative value in the comprehensive feature data set exceeds a preset threshold, triggering secondary evaluation of cloud system distribution uniformity and core structure compactness; by collecting the eye wall density quantitative value in the satellite remote sensing image and comparing it with the set threshold, when the quantitative value exceeds a certain set threshold, the secondary evaluation process is triggered; using distributed node synchronization technology, satellite image data and meteorological ensemble prediction interface data are integrated, cloud system distribution uniformity is quantitatively calculated, and variance method is used to calculate the variance of cloud band coverage area, the smaller the variance value, the higher the uniformity; based on core structure compactness feature extraction, image processing algorithms such as edge detection method are used to quantify the compactness of eye wall and core area, the edge detection process includes denoising the image by Gaussian filtering, then applying Canny operator to extract the edge contour, and finally calculating the compactness value.

[0036] Further, the weight of the path probability distribution is adjusted in combination with the vertical wind shear intensity information; the vertical wind shear intensity data is obtained through a global numerical prediction interface system and is mapped into the path probability distribution model, and if the wind shear intensity is higher than a second value (meters per second), the probability weight of the path will be reduced by a certain percentage; this adjustment helps to reflect the actual impact of atmospheric dynamics on the cyclone path and reduce the error of path deviation; through these adjustments, the stability level of the cyclone shape and the preliminary risk of path deviation trend are judged according to the evaluation results, ensuring the comprehensiveness and reliability of the evaluation; the cyclone stability level is calculated by combining the cloud system uniformity, compactness value and wind shear adjustment weight through a weighted average method, and if the total score exceeds a third value, it is determined as a high stability level; the risk is quantified according to the path deviation trend, and if the path deviation exceeds a preset threshold, such as 5 degrees, the risk level is increased.

[0037] In the implementation process, the integrity of all input data is verified by using a data consistency checking mechanism to ensure that all data are synchronized and error-free, such as checking that the time synchronization difference between satellite images and forecast data does not exceed 1 hour, to ensure the reliability of the evaluation; through these technical means, the present application can generate stable cyclone shape description and path trend preliminary information, greatly improving the accuracy and timeliness of path prediction, especially in the case of cyclone path deviation and drastic shape change; for example, in tropical cyclone monitoring, when the eye wall density degree quantization value is 0.85, through secondary evaluation, the cloud system distribution uniformity variance is 0.12 and the core structure compactness is 0.92, thereby ensuring accurate identification of cyclone stability and path risk and further improving the scientificity of the early warning system.

[0038] The advantage of the technical solution is that through the fusion and correction of multi-source data, in combination with the influence of vertical wind shear and sea surface temperature field, the cyclone path can be dynamically and accurately evaluated and adjusted, especially in the context of continuous monitoring and multi-period updating, the risk evaluation can be updated in real time, and stable and reliable early warning information can be provided; the above closed-loop processing method can ensure the comprehensiveness and consistency of path prediction and shape analysis, and provide strong data support for meteorological warning and disaster emergency management.

[0039] Among them, generating preliminary risk information further includes: quantitatively grading the cyclone shape stability according to the shape and path parameters in the comprehensive feature data set in combination with the environmental variables; analyzing the contribution weight of vertical wind shear intensity and sea surface temperature field distribution to path deviation trend; cross- verifying the stability grading and path deviation evaluation results through multi-source data fusion technology to generate preliminary risk information.

[0040] Specifically, to solve the problem that the evaluation of cyclone shape stability and path deviation trend is not accurate and timely in existing tropical cyclone prediction methods, especially in the case of rapid cyclone shape change and unclear path deviation trend, the existing technology lacks an effective dynamic evaluation mechanism; through the use of comprehensive feature data sets, combined with multi-source data fusion technology, the cyclone shape stability can be accurately evaluated, the path deviation trend can be optimized, and the reliability and accuracy of the path prediction can be improved.

[0041] In the present embodiment, by comprehensively using the shape parameters and path features in the comprehensive feature data set, combined with environmental variable data, the stability of the cyclone shape and the path deviation trend are evaluated, and preliminary risk information is generated; first, according to the eye wall density quantitative value and the cloud system distribution uniformity parameter in the comprehensive feature data set, as the core indicators of the cyclone shape; the eye wall density quantitative value is extracted from the satellite remote sensing image, and compared with the preset threshold value, for example, when the quantitative value exceeds a certain set threshold value, further secondary evaluation is triggered; the satellite image data is combined with the path probability distribution data provided by the meteorological ensemble prediction interface to calculate the cloud system distribution uniformity, and the uniformity of the cloud system distribution is quantified by using the variance method, the smaller the variance value, the more uniform the cloud system, otherwise it may indicate that the cyclone shape is unstable; the compactness of the eye wall and the core area is extracted by image processing algorithm such as Canny edge detection, to further evaluate the stability of the cyclone shape.

[0042] Then, for the path deviation trend, the influence weight of vertical wind shear intensity and sea surface temperature field distribution is analyzed, the vertical wind shear intensity data is obtained through the global numerical prediction docking system, and the contribution of the data to the path deviation is calculated to adjust the initial weight of the path deviation; for example, if the wind shear intensity is greater than 10 meters per second, the influence weight of wind shear on the path is increased; the sea surface temperature field distribution data is integrated to evaluate its guiding effect on the cyclone path, and the corresponding weight is calculated, based on these weights, the prediction model parameters of the path deviation trend are adjusted to ensure that the path prediction is more consistent with the actual meteorological dynamics.

[0043] The evaluation results are verified by multi-source data fusion technology to generate preliminary risk information, ensuring the scientificity of the evaluation process and the reliability of the results; meteorological ensemble prediction interfaces and data from global numerical prediction systems are accessed as multi-source inputs, and distributed node synchronization technology is used to fuse these data with the aforementioned evaluation results, perform data consistency checking, and eliminate outliers to ensure data quality; verification algorithms are applied to verify the fused data to confirm the accuracy of the cyclone shape stability and path deviation trend; for example, if the eye wall density quantization value is 0.85, and the preliminary risk determination of the path deviation trend is moderate, the verification process ensures the reliability of the evaluation results; in a scenario close to land, if satellite images show that the cloud band symmetry is good and the consistency is high, the path deviation risk is determined to be low, which helps to improve the accuracy of the forecast and the efficiency of the emergency response.

[0044] Through the implementation of the present application, the path deviation trend and shape stability of the cyclone can be evaluated with higher accuracy, thereby providing more accurate risk information for weather warning; the technical solution not only enables real-time adjustment of cyclone path prediction, but also provides effective data support for dynamic risk management, ensuring the scientificity and accuracy of the forecast in multiple cyclone monitoring periods.

[0045] Step S4: fuse the preliminary risk information with the real-time updated prediction data to determine the dynamic change of the path deviation and divide the initial risk level.

[0046] Among them, step S4 further includes: fusing the stability level and path deviation trend in the preliminary risk information with the real-time acquired path probability distribution and environmental variable correction results; performing interactive analysis on the influence of the cloud band symmetry characteristics of the cyclone and the sea temperature field distribution; using distributed node synchronization technology to update the dynamic state of the path deviation trend, and dividing the initial risk level according to the updated trend.

[0047] Specifically, to solve the problems in the existing tropical cyclone prediction method that real-time data cannot be fully utilized to update the dynamic change of the path deviation in time, and there is a lack of effective multi-source data fusion to optimize the path risk evaluation; by fusing the stability level, path deviation trend preliminary risk information and real-time path probability distribution data, and combining the interactive analysis of the cloud band symmetry characteristics and the sea temperature field distribution, the present application can effectively improve the accuracy and dynamics of the path prediction, and provide reliable support for cyclone path prediction, especially when the cyclone path changes rapidly, as follows.

[0048] In the present embodiment, firstly, the stability level and the preliminary risk information of the path deviation trend are combined to form a preliminary risk vector, the stability level is quantified based on the eye wall density value and the cloud system distribution uniformity, and the preliminary risk information of the path deviation trend comes from the path probability distribution weight adjusted by the vertical wind shear intensity; these data are fused by the weighted average method to form real-time path probability distribution data, ensuring that the data reflects the dynamic stability of the current cyclone, avoiding the deviation caused by a single data source, and enhancing the reliability of subsequent analysis; wherein the fusion process by the weighted average method is to assign different weights to the stability level, the preliminary risk information of the path deviation trend and the path probability distribution data adjusted by the vertical wind shear intensity according to their respective importance, which is usually based on the degree of influence on the path prediction. Then, these data are weighted and averaged according to the corresponding weights to obtain a comprehensive real-time path probability distribution data, so that the fused data can more accurately reflect the dynamic stability of the cyclone and reduce the deviation that may be caused by a single data source.

[0049] The interaction analysis is carried out for the influence of the cloud band symmetry feature and the sea surface temperature field distribution; for example Figure 2As shown, the influence coefficient of each key environmental variable on the path deviation is calculated in a real-time prediction cycle for a typhoon; analysis shows that the increase of the sea temperature gradient has the most significant traction effect on the path, with an influence coefficient as high as 0.35, while the increase of the vertical wind shear increases the uncertainty of the path, with an influence coefficient of 0.30; based on such quantitative analysis, these influence coefficients are converted into specific parameters for adjusting the weight of the path probability distribution; for example, when the sea temperature gradient influence coefficient is prominent, a higher weight is given to the sea temperature field distribution in the path deviation model, so as to more accurately predict the deflection trend of the cyclone to the warm water side; such quantitative interaction analysis based on real-time data is one of the core links of the risk division result of the present application, which can dynamically reflect the change of the environmental field and further improve the prediction accuracy; the path deviation model is a general analysis model established in the initialization or deployment stage of the present prediction method, the input of which is the dynamic feature data generated after multi-source fusion and interaction analysis, including real-time path probability distribution, environmental variable correction results, and quantified interaction influence coefficients of cloud band symmetry and sea temperature field, etc., and the output is a path deviation trend vector representing the moving direction and intensity of the cyclone in the future period of time, the training of the model is usually based on a large number of historical tropical cyclone path data and environmental field data, and methods such as Kalman filtering, regression analysis or machine learning are used to learn the statistical or dynamic mapping relationship between environmental variables and path changes, in the present application, the specific training process is not the focus of innovation, the core contribution of the present application is to build a closed-loop process that can dynamically provide high-quality input features for the pre-trained model after real-time correction, and to real-time check and weight feedback optimization of the output deviation vector, so as to significantly improve the accuracy and adaptive ability of the model in actual business prediction; specifically, the cloud band edge in the satellite remote sensing image is analyzed in the frequency domain by the Fourier transform method, and the symmetry coefficient is extracted, and the cloud band symmetry is evaluated by the ratio of the low-frequency component to the high-frequency component, wherein the Fourier transform is to convert the cloud band image into a frequency domain signal, the low-frequency component represents the overall symmetric structure, and the high-frequency component represents the local asymmetric details, and the symmetry coefficient is calculated by the ratio of the low-frequency component to the high-frequency component, and the higher the ratio, the better the symmetry; the sea temperature field data is collected, the sea temperature gradient vector is calculated, the gradient value is calculated by the finite difference method, and the vector product operation is performed on the gradient value and the cloud band symmetry coefficient to form an interaction influence matrix representing the influence degree of the sea temperature on the cloud band symmetry; the influence degree of the sea temperature field distribution on the cloud band symmetry is quantified according to the trace value of the interaction influence matrix, if the matrix trace value is greater than a preset threshold, it indicates that the sea temperature field distribution significantly enhances the cloud band symmetry, thereby affecting the stability of the path deviation trend; the interaction analysis result is normalized to map the matrix elements to the interval of 0 to 1 to form a comprehensive interaction feature set, which is further used to update the dynamic change of the path deviation trend.

[0050] The distributed node synchronization technology is used to update the dynamic change of the path deviation trend, and the initial risk classification result of the cyclone path is generated; a plurality of distributed nodes are deployed, each node is responsible for processing local data, for example, one node processes the symmetry characteristics of the cloud band, and another node processes the influence of the sea surface temperature field distribution, and the data consistency between the nodes is ensured through the clock synchronization protocol; when the data difference between the nodes exceeds the preset threshold, the system will recalculate to ensure that the output data of each node is accurate; the synchronized data is input into the path deviation model, and the Kalman filter is used for dynamic updating, the Kalman filter is used to optimize the prediction value of the deviation trend by combining new observation data through prediction and updating steps, and the path prediction accuracy is improved; according to the updated deviation trend vector, the initial risk result is classified, and the vector length is compared with the preset risk threshold to ensure that the final path risk classification accurately reflects the dynamic change of the cyclone.

[0051] For example, in a strong wind shear environment, five distributed nodes are deployed, the node synchronization delay is controlled within 0.5 seconds, the vector length after the Kalman filter is updated is 4.1, which exceeds the high risk threshold 4.0, and the path deviation risk is classified as high risk, thereby ensuring the timeliness and accuracy of the forecast; through this method, the technical scheme can quickly respond to path changes, dynamically adjust risk assessment, optimize cyclone path prediction, and improve the effectiveness of disaster warning; in multiple cyclone monitoring periods, the scheme can realize continuous updating of data and accurate evaluation of path deviation trend, and ensure efficient operation of the warning system in different scenarios; wherein the vector length 4.1 is obtained by calculating the vector length of the path deviation trend, specifically, the vector length is also called the absolute value or size of the vector, which is a measure of the path deviation vector, indicating the total change amplitude of the path deviation trend, if the vector length calculation result of the path deviation is 4.1, it indicates that the change amplitude of the cyclone path is 4.1 units in this environment; in order to calculate this length, the path deviation vector needs to be defined, assuming that the path deviation vector is composed of a series of coordinate points in time, such as longitude and latitude, each coordinate point has a corresponding time stamp, and the path deviation vector can represent the change between these coordinate points, that is, the deviation between each two adjacent coordinate points.

[0052] In summary, the technical scheme improves the accuracy and timeliness of path prediction, optimizes the risk assessment process of tropical cyclone prediction, and provides strong data support for meteorological warning and disaster management through multi-source data fusion, interactive analysis of cloud band symmetry and sea surface temperature field distribution, and application of distributed node synchronization technology.

[0053] Step S5: Deviation checking is performed on the path corresponding to the initial risk level, and a parameter correction strategy is dynamically adjusted based on the checking result to optimize the path risk level.

[0054] The step S5 further comprises: comparing the path deviation trend corresponding to the initial risk level with the historical path probability distribution, calculating a path deviation; if the path deviation exceeds a preset threshold, adjusting the weight proportion of the sea surface temperature field and the vertical wind shear data in the parameter correction through a data consistency checking mechanism; and based on the adjusted weight proportion and the checking result, re-determining an optimized path risk level.

[0055] The adjusting the weight proportion of the sea surface temperature field and the vertical wind shear data in the parameter correction through the data consistency checking mechanism comprises: calculating a covariance matrix between the sea surface temperature field distribution data and the vertical wind shear intensity data, and calculating the variance of each environmental variable data; if the sum of all elements on the main diagonal of the covariance matrix exceeds a preset consistency threshold, inversely adjusting the weight proportion of each environmental variable data according to the variance of each environmental variable data, and the environmental variable data with a larger variance is given a lower weight proportion.

[0056] Specifically, to solve the problems of insufficient accuracy, poor timeliness and insufficient multi-source data fusion in the path deviation trend and risk assessment process of the existing tropical cyclone prediction method, the application optimizes the division of the path risk level by analyzing the initial risk division result and checking the path deviation, combining the comparison of the central pressure data and the historical path probability distribution, to ensure the accuracy and reliability of the path prediction, especially when the cyclone path changes rapidly, the risk assessment result can be updated in time.

[0057] In the embodiment, the initial risk division result is derived from the preliminary risk division of the cyclone path, which compares the collected central pressure data such as real-time minimum pressure value with the stored historical path probability distribution database point by point, calculates the similarity index to identify potential deviation patterns; in this way, by comparing and analyzing the historical path data, it can be judged whether the current cyclone path exists abnormal deviation, so as to help early identification of potential risks.

[0058] The deviation of the path deviation trend is checked; in this step, the preliminary risk information of the path deviation trend is based on the dynamic change data in the aforementioned initial risk division result, and the deviation is quantified by calculating the Euclidean distance between the current path vector and the historical average path, the Euclidean distance refers to the straight line distance between the path point coordinates, and the deviation value calculated can reflect the difference between the current path and the historical path, and then quantify the deviation amplitude; the vector data of the current path deviation trend is collected, including longitude and latitude sequence and time stamp, and these vector data are matched with the corresponding sequence in the historical path probability distribution, the coordinate difference value of each time point is calculated, the difference value is summarized and the weighted average method is applied to obtain the overall deviation value, to ensure that the time effectiveness of the path is considered in the calculation process.

[0059] For example, in the tropical cyclone monitoring scenario, assuming that the current path deviation trend of a cyclone shows a 5-degree eastward deviation, while the historical path probability distribution indicates an average deviation of 2 degrees, the deviation value calculated by the Euclidean distance is 3.5, which can help identify the abnormal trend of the path in time and improve the prediction accuracy; according to different cyclone intensity scenarios such as weak cyclones and strong cyclones, the deviation check can also adjust the weight of distance calculation, weak cyclones pay more attention to short-term deviation, while strong cyclones emphasize long-term trend, which can adapt to the variability of cyclone path under different sea surface temperature field distribution.

[0060] When the deviation value exceeds the preset threshold, the data consistency check mechanism is started to adjust the weight proportion of environmental variable correction; this step will detect the data consistency by calculating the variance of sea surface temperature field and vertical wind shear data, and compare with the standard threshold, if inconsistent, adjust the corresponding weight proportion; for example, when the deviation value is 4.2, if the sea surface temperature field and the vertical wind shear data are inconsistent, the check mechanism will adjust the weight of vertical wind shear to 0.3 and the weight of sea surface temperature field to 0.7, thereby optimizing the accuracy of the path prediction result; check whether the deviation value exceeds the threshold, calculate the correlation between environmental variables through the covariance matrix to ensure data synchronization, adjust the weight proportion according to the check result to ensure the reliability of data fusion.

[0061] In a typical cyclone path deviation case, when the path deviation value is 4.2 and the variance of sea surface temperature field and wind shear data is higher than the standard threshold, the system automatically adjusts the weight to make the path prediction more consistent with the actual observation, improving the accuracy of risk level assessment; according to the above check result, the path risk level is optimized to ensure the scientificity and rationality of the assessment result; the optimized deviation value and weight proportion will be mapped to low, medium or high risk level through the predefined risk level table; for example, if the optimized deviation value decreases to a certain value, the path risk is judged as low risk, which helps the timeliness and accuracy of meteorological forecast decision.

[0062] The technical solution can optimize the accuracy and timeliness of cyclone path prediction, and ensure the scientificity and reliability of path risk assessment through the processes of initial risk division, path deviation check, data consistency check and dynamic adjustment of weight; the scheme can effectively meet the path prediction needs under different cyclone intensity and complex environment, and provides strong support for meteorological warning and disaster emergency management.

[0063] Step S6: According to the optimized path risk level, integrate the new round of multi-source information to update the cyclone shape parameters, and feed back the updated parameters to the feature formation step to complete the closed-loop data processing and prediction iteration.

[0064] The step S6 further comprises: integrating the latest multi-source information from the weather ensemble prediction interface and the global numerical prediction system according to the optimized path risk level; using the latest multi-source information and environmental data to update the eye wall density and cloud system distribution uniformity; through real-time data transmission, generating an updated shape parameter set for input in the next prediction cycle, realizing closed-loop iterative updating of the cyclone path and shape prediction.

[0065] Specifically, to solve the problem of insufficient continuous updating and real-time prediction accuracy of cyclone path and shape in the existing tropical cyclone prediction method, especially when the cyclone path deviates and the shape changes rapidly, the existing technology cannot efficiently integrate real-time data, optimize shape parameters and perform closed-loop data updating; the present application provides a precise dynamic updating mechanism by optimizing the path risk level, integrating multi-source information and combining data support of the global numerical prediction system, to improve the accuracy and timeliness of the tropical cyclone path prediction.

[0066] In the embodiment, the multi-source information obtained by the weather ensemble prediction interface is integrated through the optimized path risk level, the shape parameter set is updated, and closed-loop data processing is completed; specifically, the wind speed distribution and pressure gradient data in the weather ensemble prediction interface are weighted and fused according to the path risk level as a weight factor, forming an integrated multi-source information set, the higher the path risk level, the greater the weight factor, and especially the weight of high uncertainty data also increases; in this way, the dynamic response of the cyclone path change under high risk is ensured; for example, in tropical cyclone monitoring, if the path risk level is high, the weight of the wind speed distribution data in the weather ensemble prediction interface will increase, thereby improving the prediction accuracy; using distributed node synchronization technology, the integrated multi-source information set is preliminarily matched with other environmental variables such as sea surface temperature field distribution and vertical wind shear intensity, to ensure data compatibility and consistency.

[0067] With the data support of the global numerical prediction docking system, the eye wall intensity and cloud system distribution uniformity are updated retrospectively; the sea surface temperature field distribution and vertical wind shear intensity data provided by the global numerical prediction docking system are collected as the basis for retrospective updating. The sea surface temperature field distribution reflects the spatiotemporal variation of the sea surface temperature, which is the source of cyclone energy. The vertical wind shear intensity reflects the stability of the cyclone structure. The current eye wall intensity value is compared with the historical data. If the difference exceeds the threshold, the eye wall intensity value is adjusted according to the wind shear intensity. For example, when the wind shear is strong, the intensity will increase accordingly, reflecting the change in the compactness of the eye wall of the cyclone. The cloud system distribution uniformity is updated retrospectively. The variance of the cloud system distribution is calculated and combined with the sea surface temperature field distribution data to adjust the uniformity index. For example, in high temperature areas, the distribution of cloud systems may be more uniform, resulting in a lower variance value. The results of the above steps are fused to generate the updated eye wall intensity and cloud system distribution uniformity parameter pair, which provides support for the generation of subsequent morphological parameters.

[0068] The morphological parameter set of the next cycle is generated through the real-time data transmission mechanism, and the closed-loop data update of the cyclone path and morphology is completed. The retrospective updated morphological parameter pair is combined with the path risk level to generate the morphological parameter set of the next cycle through the real-time transmission mechanism. These morphological parameter sets include the eye wall intensity, the cloud system distribution uniformity, and the vector representation of the path deviation trend. Consistency check is performed on the generated parameter set. If the check is passed, the cyclone path and morphology data are updated to achieve a closed loop, ensuring that the data of each cycle is based on the output of the previous cycle, forming a continuous processing chain. In a typical cyclone monitoring scenario, for example, if the initial quantization value of the eye wall intensity is 0.8 and the path risk level is high, the weight of the wind speed distribution set increases, and after data fusion, the prediction accuracy is improved. In the retrospective update, assuming that the vertical wind shear intensity is 5 m / s, the eye wall intensity is adjusted from 0.8 to 0.9, and the cloud system distribution uniformity variance is reduced from 0.2 to 0.15, the updated morphological description is more accurate, which helps to reduce the path prediction error.

[0069] In the implementation process, for different cyclone intensity scenarios, the eye wall intensity is updated retrospectively with the support of the sea surface temperature field distribution, which can make the parameters more in line with the actual dynamics and improve the stability of continuous monitoring. After generating the morphological parameter set of the next cycle, closed-loop updating ensures the complete chain of data from initial fusion to the final path trend, which helps to ensure the continuity of real-time prediction and avoid information gaps. In summary, this technical solution can effectively improve the prediction accuracy of cyclone path and morphology by combining multi-source information integration, global numerical prediction support, real-time data transmission mechanism, and closed-loop data updating, ensuring the dynamic and accurate prediction of tropical cyclone path and providing more scientific and reliable decision support for disaster warning and emergency response.

[0070] To verify the actual prediction performance of the method, it is compared with the existing traditional method, as shown in Figure 3 The results show that the error of the method, the "method" curve in the figure, is significantly lower than that of the traditional method in the entire prediction period, and the error range, the "method error range" area in the figure, is significantly narrower and more stable. In particular, in the middle and later stages of the 24-hour to 48-hour prediction, the error of the traditional method increases significantly to about 38 kilometers, while the method benefits from its dynamic correction and closed-loop optimization mechanism, which can continuously integrate real-time information and adjust model parameters, thereby effectively controlling the error to about 16 kilometers. The comparison test fully proves that the application significantly improves the accuracy and stability of tropical cyclone track prediction by constructing an intelligent process of "data fusion-dynamic evaluation-closed-loop update", especially in long-acting prediction. It provides a solid basis for improving the reliability of business prediction.

[0071] Through the cooperation between the above steps, the accuracy and timeliness of tropical cyclone prediction are improved.

[0072] In summary, the application provides a tropical cyclone intelligent prediction method based on multi-source data fusion and closed-loop iteration. The method integrates satellite remote sensing, ensemble prediction, global numerical model and other multi-source heterogeneous data to construct a unified cyclone state feature set, and combines key environmental variables such as sea surface temperature field and vertical wind shear for dynamic parameter correction to form a high-reliability comprehensive feature data set. On this basis, the system synchronously evaluates the cyclone shape stability and path deviation trend, generates preliminary risk information, and realizes path dynamic update and risk level division through real-time data fusion and interactive analysis. Further, by means of deviation checking and weight self-adaptive adjustment mechanism, the risk assessment results are optimized, and finally through the closed-loop design of backtracking update shape parameters and feedback to the prediction starting point, the continuous iteration and self-optimization of the prediction process are realized. The method significantly improves the accuracy, timeliness and scientificity of tropical cyclone path and intensity prediction and risk assessment, providing reliable technical support for disaster prevention and mitigation decision-making.

[0073] Embodiment two: The above describes a tropical cyclone prediction method using ensemble prediction data in an embodiment of the application. The following describes a tropical cyclone prediction system using ensemble prediction data in an embodiment of the application. Please refer to Figure 4 The tropical cyclone prediction system using ensemble prediction data in an embodiment of the application includes: A data fusion and feature generation module is configured to acquire and fuse multi-source data to form a cyclone shape and path feature set representing the state of a tropical cyclone. The parameter correction and feature optimization module is configured to perform dynamic correction of parameters based on the cyclone shape and path feature set, in combination with sea surface temperature field and vertical wind shear data, to generate a comprehensive feature data set. The synchronous evaluation and risk preliminary judgment module is configured to use the comprehensive feature data set to synchronously evaluate cyclone shape stability and path deviation trend, to generate preliminary risk information. The real-time fusion and risk division module is configured to fuse the preliminary risk information and real-time updated prediction data, to determine dynamic changes of path deviation, and to divide initial risk levels. The deviation feedback and correction optimization module is configured to perform deviation checking on paths corresponding to the initial risk levels, and to dynamically adjust parameter correction strategies based on the checking results, to optimize path risk levels. The closed-loop update and iteration driving module is configured to update cyclone shape parameters based on a new round of multi-source information according to the optimized path risk levels, and to feed back the updated parameters to the feature forming step, to complete closed-loop data processing and prediction iteration.

[0074] Through the cooperation of the above components, the accuracy and timeliness of the tropical cyclone prediction are further improved.

[0075] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described system and units can refer to the corresponding processes in the foregoing method embodiments, which will not be described herein.

[0076] The integrated units, if realized in the form of software function units and sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts that make contributions to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The foregoing storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0077] The above-described embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A tropical cyclone forecasting method using ensemble forecast data, characterized by, The method comprises: Step S1: acquiring and fusing multi-source data to form a cyclone shape and path feature set representing the state of a tropical cyclone; Step S2: based on the cyclone shape and path feature set, combining sea surface temperature field and vertical wind shear data to perform parameter dynamic correction, and generating a comprehensive feature data set; Step S3: using the comprehensive feature data set, synchronously evaluating cyclone shape stability and path deviation trend, and generating preliminary risk information; Step S4: fusing the preliminary risk information and real-time updated forecast data to determine the dynamic change of path deviation, and dividing the initial risk level; Step S5: bias checking the path corresponding to the initial risk level, and dynamically adjusting the parameter correction strategy based on the checking result to optimize the path risk level; Step S6: according to the optimized path risk level, updating the cyclone shape parameters in a new round of multi-source information integration, and feeding back the updated parameters to the feature forming step to complete the closed-loop data processing and prediction iteration.

2. The tropical cyclone prediction method using ensemble prediction data according to claim 1, characterized in that, The step S1 further comprises: Based on satellite remote sensing images, quantitatively extracting eye wall density, cloud system distribution uniformity, core structure compactness and cloud band symmetry as key morphological parameters; From the meteorological ensemble prediction system and the global numerical prediction system, center pressure data and path probability distribution information are obtained as path features; Using distributed node synchronization technology, the key morphological parameters, the center pressure data and the path probability distribution information are fused in real time to generate the cyclone shape and path feature set.

3. The tropical cyclone prediction method using ensemble prediction data according to claim 1, wherein, The step S2 further comprises: According to the cyclone shape and path feature set, the sea surface temperature field distribution data and the vertical wind shear intensity data related to the current cyclone are obtained; Using the sea surface temperature field distribution data and the vertical wind shear intensity data, the parameters in the cyclone shape and path feature set are physically corrected; Through real-time data transmission and consistency verification, the corrected parameters are integrated to generate a comprehensive feature data set containing cyclone structure description and path trend information.

4. The tropical cyclone prediction method using ensemble prediction data according to claim 1, wherein, The step S3 further comprises: If the eye wall density quantitative value in the comprehensive feature data set exceeds a preset threshold, secondary evaluation of cloud system distribution uniformity and core structure compactness is triggered; Combined with the vertical wind shear intensity information, the weight configuration of the path probability distribution is adjusted; Based on the secondary evaluation results and the adjusted weights, the stability level of the cyclone shape and the preliminary risk level of the path deviation are determined to form the preliminary risk information.

5. The tropical cyclone prediction method using ensemble prediction data according to claim 4, wherein, Generating the preliminary risk information further comprises: According to the shape and path parameters in the comprehensive feature data set, the stability of the cyclone shape is quantitatively graded in combination with environmental variables; The contribution weights of vertical wind shear intensity and sea surface temperature field distribution to path deviation trend are analyzed; Through multi-source data fusion technology, the stability grading and path deviation evaluation results are cross-verified to generate preliminary risk information.

6. The tropical cyclone prediction method using ensemble prediction data according to claim 1, wherein, The step S4 further comprises: The stability level and path deviation trend in the preliminary risk information are fused with the real-time acquired path probability distribution and environmental variable correction results; The influence of cloud band symmetry characteristics of the cyclone and sea surface temperature field distribution on each other is analyzed; The distributed node synchronization technology is adopted to update the dynamic state of the path deviation trend, and the initial risk level is divided according to the updated trend.

7. The tropical cyclone prediction method using ensemble prediction data according to claim 3, wherein, The step S5 further comprises: The path deviation trend corresponding to the initial risk level is compared with the historical path probability distribution, and the path deviation is calculated; If the path deviation exceeds a preset threshold, the weight proportion of the sea surface temperature field and the vertical wind shear data in the parameter correction is adjusted through a data consistency checking mechanism; Based on the adjusted weight proportion and the checking result, the optimized path risk level is re-determined.

8. The tropical cyclone prediction method using ensemble prediction data according to claim 7, wherein, The weight proportion of the sea surface temperature field and the vertical wind shear data in the parameter correction is adjusted through a data consistency checking mechanism, comprising: The covariance matrix between the sea surface temperature field distribution data and the vertical wind shear intensity data is calculated, and the variance of each environmental variable data is calculated; If the sum of all elements on the main diagonal of the covariance matrix exceeds a preset consistency threshold, the weight proportion of each environmental variable data is inversely adjusted according to the variance size of each environmental variable data, and the environmental variable data with larger variance is given lower weight proportion.

9. The tropical cyclone prediction method using ensemble prediction data according to claim 1, wherein, The step S6 further comprises: According to the optimized path risk level, the latest multi-source information from the meteorological ensemble prediction interface and the global numerical prediction system is integrated; The latest multi-source information and environmental data are used to update the eye wall density and the uniformity of cloud system distribution; Through real-time data transmission, an updated shape parameter set for the next prediction cycle input is generated, realizing closed-loop iterative update of the cyclone path and shape prediction.

10. A tropical cyclone forecasting system using ensemble forecast data for implementing a method of tropical cyclone forecasting using ensemble forecast data according to any one of claims 1 to 9, characterized in that, The system comprises: A data fusion and feature generation module for acquiring and fusing multi-source data to form a cyclone shape and path feature set representing the state of a tropical cyclone; A parameter correction and feature optimization module for dynamically correcting parameters based on the cyclone shape and path feature set, combining sea surface temperature field and vertical wind shear data to generate a comprehensive feature data set; A synchronous evaluation and risk preliminary judgment module for using the comprehensive feature data set to synchronously evaluate cyclone shape stability and path deviation trend to generate preliminary risk information; A real-time fusion and risk division module for fusing the preliminary risk information and real-time updated prediction data to determine the dynamic change of path deviation and divide the initial risk level; A deviation feedback and correction optimization module for deviation checking of the path corresponding to the initial risk level and dynamically adjusting the parameter correction strategy based on the checking result to optimize the path risk level; A closed-loop update and iterative driving module for integrating new round of multi-source information to update cyclone shape parameters according to the optimized path risk level, and feeding back the updated parameters to the feature forming step to complete closed-loop data processing and prediction iteration.

Images