A method for planning a typhoon-avoiding route for marine shipping in a typhoon cluster event

Abstract

This invention discloses a method for planning typhoon avoidance routes in maritime shipping that integrates multiple typhoon events. By classifying typhoon events into scenarios, extracting core shipping parameters, and constructing scenario-specific hierarchical diagnostic models, extreme sample expansion and parameter optimization are achieved by combining physical mechanism constraints. This yields three types of forecast error correction coefficients: path deviation, wind field expansion, and multi-typhoon interaction. Based on these correction coefficients, the forecast error envelope is scaled, its boundaries expanded, and its direction shifted. Secondary calibration is then performed using parameters such as the ship's wind resistance rating to form a ship-adaptive error range. By real-time monitoring of the overlap between the actual typhoon situation and the error envelope, the correction coefficients are iteratively optimized, and the risk range is dynamically adjusted. This achieves real-time typhoon avoidance route planning that balances safety and efficiency, effectively improving the safety of ship navigation and the rationality of route planning under typhoon conditions.

Description

Technical Field

[0001] This invention relates to the field of maritime meteorological navigation and intelligent route planning technology, specifically to a method for planning maritime typhoon avoidance routes that integrates multiple typhoon events. Background Technology

[0002] The western Pacific Ocean is the sea area with the most frequent typhoon activity in the world. Extreme typhoon cluster events (such as "four typhoons dancing together" or super typhoon clusters) are characterized by high generation density, high intensity, complex paths, and significant multi-system interactions, posing an extremely high threat to the safety of ocean shipping.

[0003] Currently, the establishment of typhoon risk zones for maritime shipping primarily relies on official typhoon forecast data, overlaid with uncertainties and ship navigation characteristics, to generate dynamic, graded risk areas with time windows. Typically, typhoon track forecasts from authoritative institutions such as the JTWC, the China Meteorological Administration, and the Japan Meteorological Agency are used to extract the typhoon's center location, the radius of wind circles of different levels, the maximum wind speed at the center, and the direction and speed of movement. Using the typhoon center as the center and the corresponding wind circle radius as the basis, the core risk zone directly affected by the typhoon is delineated. Considering the errors in typhoon track and intensity forecasts, existing methods delineate a forecast error envelope (probability cone) outside the basic wind circle based on historical forecast deviations. For example, short-term forecasts may extend outwards by 100-200 km, and medium-term forecasts by 300-500 km, thus forming a secondary core zone of indirect impact, encompassing secondary risks such as strong winds, swells, and severe convection. Finally, the risk zones are dynamically adjusted based on the vessel's speed, draft, cargo capacity, and wind resistance rating (e.g., bulk carriers and container ships have different wind resistance capabilities), striking a balance between safety and efficiency. For low-speed or heavily loaded vessels, the risk zones are appropriately expanded to allow more time for maneuvering; for high-speed or empty vessels, the risk zones can be appropriately reduced to improve the economic efficiency of the route.

[0004] However, the existing methods mentioned above are only designed for single typhoon or regular typhoon scenarios. When facing extreme multi-typhoon events, they have the following key shortcomings: First, existing methods usually only perform geometric superposition of multiple typhoon risk areas, simply splicing / merging the basic wind circles and error envelopes of multiple typhoons, without considering the nonlinear risk superposition caused by the interaction of multiple typhoons (Fujiwara effect, wind field distortion, steering flow anomalies) (e.g., in the area where the wind fields of two typhoons meet, the actual wind force will be much greater than the superposition of the wind circles of a single typhoon); Second, existing technologies lack physical mechanism support, relying only on apparent parameters such as path and wind circle, without introducing physical indicators such as GPI, OW index, and subtropical high anomaly, making it impossible to determine... The current technology suffers from several drawbacks. First, it fails to accurately predict the expansion / contraction trend of the risk zone during extreme typhoon clusters (e.g., abnormal westward extension of the subtropical high can cause typhoon wind fields to shift westward, making it easy to miss high-risk areas on the west side of existing risk zones). Second, existing technologies are based on historical deviations of conventional typhoons, and due to the limited sample size of extreme typhoon clusters, the error envelope cannot match the actual deviation, resulting in either excessive avoidance (loss of economy) or insufficient avoidance (posing safety risks). Third, all extreme typhoon clusters use the same outward expansion standard, failing to differentiate the risk zone characteristics of different extreme cluster scenarios such as abnormal subtropical highs during ENSO decay years and active summer monsoon troughs, thus failing to achieve scenario-based and precise correction of the forecast error envelope.

[0005] Among existing technologies for automatic typhoon avoidance route planning, one method is designed only for single typhoon and conventional typhoon scenarios. Even with multiple typhoon parameters, it can only achieve simple geometric superposition of independent wind circles, failing to consider the interaction of multiple typhoons and overlapping high-risk areas in extreme cluster events. Its accuracy in determining overlapping risks is severely insufficient under small sample conditions of extreme events. Another method, based on a critical decision model, relies on conventional typhoon data for its decision threshold. This threshold is prone to failure under extreme cluster typhoon conditions, failing to characterize the formation mechanism and evolution of extreme typhoon cluster events, leading to misjudgments and omissions. Furthermore, it does not incorporate diagnostic indicators related to typhoon physical mechanisms. Yet another method mentions the superposition of multiple meteorological risks, but only performs simple weighted calculations of wind speed and wave height, not based on the physical superposition of typhoon cluster mechanisms. It completely ignores analysis related to extreme typhoon cluster events, and risk zone delineation still relies on traditional wind circle radii, without employing dynamic diagnostic indicators such as vertical wind shear, absolute vorticity, and OW index. Even if some of the existing technologies mentioned above involve calculations of overlapping risk areas of multiple typhoons, they are all simple geometric superposition or intensity weighting. They do not take into account the formation mechanism of extreme typhoon clusters, nor do they introduce physical diagnostic indicators such as GPI, OW index, subtropical high anomaly, and vertical wind shear. In particular, they do not establish a scenario-based correction mechanism for the delineation of the forecast error envelope (probability cone) for extreme cluster scenarios. They cannot solve the technical problems caused by the scarcity of extreme cluster typhoon samples, strong multi-system interactions, and nonlinear amplification of risk superposition, such as inaccurate setting of error envelope, distortion of risk area judgment, and difficulty in balancing the safety and economy of flight route planning. Summary of the Invention

[0006] To address the technical problems of existing technologies, such as inaccurate risk assessment in extreme typhoon cluster scenarios, lack of physical mechanism support for error envelope correction, simple geometric superposition of multiple typhoon risks, and poor adaptability to extreme small samples, this invention proposes a maritime shipping typhoon avoidance route planning method that integrates multiple typhoon cluster events. This method can achieve differentiated, dynamic, and ship-adaptive correction of the forecast error envelope in different scenarios, thereby significantly improving the accuracy of risk area assessment under both regular and extreme typhoon cluster scenarios, and optimizing route economy while ensuring navigation safety.

[0007] To achieve the above-mentioned technical objectives, the technical solution adopted by the present invention is as follows:

[0008] A method for planning typhoon avoidance routes in maritime shipping that integrates multiple typhoon events, the method comprising the following steps:

[0009] S1. Construct a typhoon cluster classification system that includes extreme typhoon cluster events. Divide the events into multi-dimensional scenarios based on the sea area, season, intensity level, and large-scale circulation background to distinguish between regular and extreme typhoon cluster events, and establish a typhoon cluster event scenario classification library. Based on reanalysis data and optimal typhoon path data, use dynamic and thermal diagnostic parameters to extract the formation mechanism of typhoon cluster events in different scenarios, transform it into graded and quantitative diagnostic indicators, form a graded and quantitative diagnostic indicator system for typhoon cluster events, and determine the key control areas corresponding to each scenario.

[0010] S2, combined with the application requirements of marine shipping engineering, selects indicators that are highly correlated with the changes in the forecast error envelope area from the quantitative diagnostic indicators for typhoon cluster occurrence as the core parameters for shipping in each scenario; through physical mechanism constraints and extreme sample feature enhancement processing, establishes a quantitative correlation between the core parameters for shipping and the uncertainty of typhoon forecast, constructs a typhoon cluster occurrence graded diagnostic model for each scenario, and outputs forecast error correction coefficients to characterize the forecast path deviation, wind field expansion and the comprehensive amplification of the risk superposition of multiple typhoons;

[0011] S3 establishes a scenario-specific error envelope adjustment standard based on the physical impact mechanism of shipping core parameters on the forecast error envelope area within key control areas; using the forecast error correction coefficient as a quantitative basis, the forecast error envelope area is dynamically scaled, its boundaries expanded, and its direction shifted, and combined with ship navigation parameters to form a ship-adaptive error envelope area; the overlap is verified through real-time data and the correction coefficient is iteratively optimized, and finally, based on the calibrated risk area, typhoon avoidance routes are planned and dynamically adjusted.

[0012] Step S1 further includes:

[0013] S11: Collect relevant data on historical typhoon cluster events in the research sea area. According to the frequency of occurrence, number of typhoons, and intensity characteristics, typhoon cluster events are divided into regular typhoon cluster events and extreme typhoon cluster events. For extreme typhoon cluster events only, further subdivide the scenarios according to four dimensions: sea area, season, typhoon intensity level, and large-scale circulation background. The extreme typhoon cluster events are divided into several extreme sub-scenarios, including the ENSO decaying year subtropical high anomalous westward extension type, the summer monsoon trough active type, and the multiple typhoon extreme type. Based on typhoon sample data under each scenario, extract key circulation parameters, high-incidence sea area range, and historical forecast deviation characteristics. Statistically analyze the typical intervals of each parameter, clarify the differences in circulation background, spatial distribution, forecast error magnitude, and deviation distribution between regular scenarios and different extreme sub-scenarios, establish the correspondence rules between scenarios and circulation characteristics, spatial distribution, and forecast deviation characteristics, and form a typhoon cluster scenario classification library applicable to the research sea area.

[0014] S12, based on the typhoon cluster occurrence scenario classification library, corresponding key control areas are determined for different cluster occurrence scenarios. The key control areas are the sea areas where typhoon clusters occur in concentrated phases under the corresponding scenario and the historical forecast path deviation exceeds a set deviation threshold. Based on ERA5 reanalysis data, typhoon optimal path data, Nino3.4 index, and satellite observation data, the Typhoon Generation Potential Index (GPI), Okubo-Weiss vorticity deformation index, convective effective potential energy (CAPE), 200–850 hPa vertical wind shear, 850 hPa low-level absolute vorticity, and subtropical high-pressure westward extension ridge point and area anomaly indicators are used in the key control areas corresponding to each scenario to conduct joint diagnosis of the dynamic and thermal fields of different cluster occurrence scenarios. Among them, for conventional typhoon cluster occurrence scenarios, typical dynamic and thermal characteristics related to the forecast error level are extracted; for extreme typhoon cluster occurrence scenarios, multi-system coupling, extreme dynamic conditions, and enhanced air-sea interaction effects are identified, and the intrinsic relationship between the above characteristics and forecast error amplification and error envelope boundary shift is clarified.

[0015] S13. Based on the mechanism diagnosis results, parameters that have significant indicative significance for the development of typhoon clusters and the boundary and range of the forecast error envelope area are selected from the diagnostic parameters of different scenarios. These parameters are then standardized and normalized. The parameter threshold ranges are determined in combination with different cluster scenarios. The parameters are classified according to the warning threshold level, scenario adaptability, and engineering application weight, forming a typhoon cluster classification and quantitative diagnostic index system that includes circulation background indicators, dynamic indicators, thermal indicators, and comprehensive potential indicators.

[0016] Further, in step S13, the circulation background indicators include the longitude of the western extension ridge of the subtropical high, the intensity index, the Nino 3.4 index, and the intensity and location index of the monsoon trough; the dynamic indicators include the 850hPa lower-level absolute vorticity, the magnitude of the 200–850hPa vertical wind shear, and the Okubo–Weiss vorticity deformation index; the thermal indicators include the convective effective potential energy (CAPE), the sea surface temperature anomaly (SSTA), and the mid-level relative humidity; and the comprehensive potential indicators include the typhoon formation potential index (GPI), the distance / intensity of multiple typhoon interactions, and the statistical value of historical track forecast deviations.

[0017] Step S2 further includes:

[0018] S21. In combination with the engineering application requirements of maritime shipping for the stability of typhoon forecasts and the accuracy of risk zone delineation, the correlation and sensitivity analysis of the quantitative diagnostic indicators for typhoon cluster occurrence obtained in step S1 is carried out. Indicators that are correlated with the changes in the boundary and range of the forecast error envelope area under different scenarios and exceed the set threshold are selected and used as the core shipping parameters specific to each scenario.

[0019] S22. Extract the feature values ​​of the corresponding shipping core parameters from the extreme typhoon cluster event samples under each extreme sub-scenario to construct an extreme sample feature library for each extreme sub-scenario. Based on the physical mechanism constraints of typhoon generation and development, interpolate and synthesize the extreme samples to expand the sample quantity of each extreme sub-scenario. At the same time, normalize the shipping core parameters of each extreme sub-scenario to a unified dimension and establish a parameter standardization mapping table for each extreme sub-scenario.

[0020] S23. Based on the core parameters specific to the regular scenarios selected in step S21, the standardized core parameters of each extreme sub-scenario processed in step S22, and the expanded sample data, a multiple linear regression analysis method is used to establish typhoon cluster classification diagnostic models for each scenario, with the optimization objective of maximizing the accuracy of the forecast error envelope correction. The typhoon cluster classification diagnostic model uses the shipping core parameters of the corresponding scenario as input variables, and the typhoon path forecast deviation amplitude, wind field expansion radius, and multi-typhoon interaction effect as intermediate output variables. Finally, the forecast error correction coefficients used to correct the error envelope for the corresponding scenario are fitted. The forecast error correction coefficients are used to characterize the magnitude of the forecast path deviation, the wind field expansion range, and the amplification degree of the risk superposition effect of multiple typhoons.

[0021] S24. By utilizing the extreme dynamic and thermodynamic characteristics of extreme events in various scenarios, the hierarchical diagnostic model for the corresponding extreme sub-scenario is optimized in reverse. The influence weights of different shipping core parameters on the forecast error are adjusted in the scenario, further improving the model's prediction accuracy of the forecast error envelope trend in the extreme sub-scenario. At the same time, the model for the regular scenario is calibrated with regular parameters to ensure the accuracy of forecast error correction in the regular scenario.

[0022] S25: Based on the real-time acquired typhoon location, intensity, and environmental field data, determine the current typhoon cluster scenario, call the corresponding scenario's hierarchical diagnostic model, and calculate the forecast error correction coefficient for that scenario.

[0023] Step S3 further includes:

[0024] S31, based on the different physical mechanisms by which various core shipping parameters affect the forecast error envelope area, it is divided into: circulation indicators that determine the direction of error envelope area deviation, dynamic and thermal indicators that determine the radius expansion of error envelope area, and interactive indicators that determine the superimposed expansion range of multiple typhoon impact areas; based on the above physical mechanisms, a multi-dimensional adjustment logic for the error envelope area within the key control area under different scenarios is established to form an error envelope area expansion and offset standard adapted to each scenario;

[0025] S32, combining the engineering application requirements of ship safety priority and navigation efficiency, performs scenario-based calibration of the expansion and offset standards for each scenario, so that the correction range matches the degree of anomaly of the core shipping parameters and the level of forecast uncertainty.

[0026] S33, typhoon warning data and real-time meteorological parameters are updated in real time, the typhoon cluster outbreak classification diagnosis model in step S2 is called, and the forecast error correction coefficient for the corresponding scenario is dynamically updated; based on the forecast error correction coefficient, the forecast error in the key control area is scaled up as a whole, the wind field boundary is expanded and the path deviation direction is shifted to form the corrected forecast error range.

[0027] S34: Obtain ship navigation parameters including ship wind resistance level, sailing speed, and load, and combine them with the ship's own wind resistance capability to perform a second calibration on the corrected forecast error range to obtain the ship adaptation error envelope area.

[0028] S35 collects the actual movement path and wind field range of the typhoon in real time, and compares the overlap between the ship's adaptive error envelope area and the actual impact area of ​​the typhoon. If the overlap does not meet the preset safety and efficiency thresholds, the forecast error correction coefficient is iteratively updated based on the ship's wind resistance capability and current navigation status. The outer range and offset direction of the forecast error range are dynamically adjusted. The above process is repeated until the route planning is completed or the typhoon warning is lifted.

[0029] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0030] The present invention provides a method for planning typhoon avoidance routes in maritime shipping that integrates multiple typhoon events. By constructing a scenario-based processing system for both regular and extreme typhoon events, it combines large-scale circulation background, dynamic and thermal mechanism diagnosis, and shipping risk correction requirements. This enables precise correction of the forecast error envelope from empirical expansion to physical mechanism-driven, scenario-specific, and vessel-adaptive approaches, effectively improving the scientific rigor and applicability of shipping risk identification under multiple typhoon events. Compared to traditional methods that rely solely on the geometric superposition of wind circles and a fixed outward expansion ratio, this invention can more accurately characterize the interactions of multiple typhoons, the amplification of forecast deviations, and the nonlinear superposition of risks in extreme typhoon cluster events. It still possesses stable risk characterization capabilities even in extreme small-sample scenarios. By coupling the calculation of core shipping parameters and forecast error correction coefficients for different scenarios, the scope and offset direction of the risk area can be dynamically adjusted according to different circulation backgrounds and risk structures. Under the premise of ensuring ship navigation safety, it significantly reduces the voyage loss caused by excessive avoidance, improves the economy and operability of route planning, and provides more reliable theoretical and technical support for the safe and efficient passage of ocean shipping in the Western Pacific under complex typhoon cluster environments. Attached Figure Description

[0031] Figure 1This is a flowchart of the maritime shipping typhoon avoidance route planning method that integrates multiple typhoon events according to the present invention.

[0032] Figure 2 This is a map showing the distribution of 500 hPa geopotential height differences in large-scale circulation during years with multiple typhoon outbreaks and climatological mean conditions; the isolines are spaced 5 gpm apart; blue represents negative values ​​and red represents positive values.

[0033] Figure 3 This is a spatial distribution map showing the difference between the CAPE index and the climate mean convective instability in years with multiple typhoons; the contour lines are spaced 100 J / kg apart, solid lines represent positive values, and dashed lines represent negative values; blue fill represents negative values, and red fill represents positive values.

[0034] Figure 4 This is a spatial distribution map showing the difference between the OW index and the dynamic characteristics of climate mean-state disturbances in years with multiple typhoon clusters; the contour lines are spaced at 1×10⁻⁶ intervals. -8 s -2 Solid lines represent positive values, and dashed lines represent negative values; blue fill represents negative values, and red fill represents positive values.

[0035] Figure 5 This is a spatial distribution map showing the difference between the GPI index for typhoon formation potential in years with multiple typhoon outbreaks and the climate-averaged typhoon; the contour lines are spaced 0.4 apart, solid lines represent positive values ​​and dashed lines represent negative values; blue fills represent negative values ​​and red fills represent positive values. Detailed Implementation

[0036] The embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0037] This invention discloses a method for planning typhoon avoidance routes in maritime shipping that integrates multiple typhoon cluster events. The method includes the following steps:

[0038] S1. Construct a typhoon cluster classification system that includes extreme typhoon cluster events. Divide the events into multi-dimensional scenarios based on the sea area, season, intensity level, and large-scale circulation background to distinguish between regular and extreme typhoon cluster events, and establish a typhoon cluster event scenario classification library. Based on reanalysis data and optimal typhoon path data, use dynamic and thermal diagnostic parameters to extract the formation mechanism of typhoon cluster events in different scenarios, transform it into graded and quantitative diagnostic indicators, form a graded and quantitative diagnostic indicator system for typhoon cluster events, and determine the key control areas corresponding to each scenario.

[0039] S2, combined with the application requirements of marine shipping engineering, selects indicators that are highly correlated with the changes in the forecast error envelope area from the quantitative diagnostic indicators for typhoon cluster occurrence, and uses them as the core parameters for shipping in various scenarios; through physical mechanism constraints and extreme sample feature enhancement processing, establishes a quantitative correlation between the core shipping parameters and the uncertainty of typhoon forecasts, constructs a typhoon cluster occurrence graded diagnostic model for different scenarios, and outputs forecast error correction coefficients to characterize the forecast path deviation, wind field expansion and the comprehensive amplification of the risk superposition of multiple typhoons;

[0040] S3 establishes a scenario-specific error envelope adjustment standard based on the physical impact mechanism of shipping core parameters on the forecast error envelope area within key control areas; using the forecast error correction coefficient as a quantitative basis, the forecast error envelope area is dynamically scaled, its boundaries expanded, and its direction shifted, and combined with ship navigation parameters to form a ship-adaptive error envelope area; the overlap is verified through real-time data and the correction coefficient is iteratively optimized, and finally, based on the calibrated risk area, typhoon avoidance routes are planned and dynamically adjusted.

[0041] (a) Establish a graded and quantitative diagnostic index system for typhoon outbreaks

[0042] (1.1) Typhoon outbreak scenario segmentation and scenario classification library construction

[0043] This step aims to systematically review and standardize the classification of historical typhoon cluster events in the research area. Through multi-dimensional feature identification, it distinguishes between regular and extreme cluster events, and conducts refined scenario division for extreme events, providing a basis for subsequent scenario-based mechanism diagnosis and risk correction.

[0044] First, a basic dataset was constructed by collecting historical best typhoon track datasets, ERA5 reanalysis meteorological field data, sea surface temperature anomaly data, and historical typhoon forecast bias data within the research sea area. Based on statistical characteristics such as event frequency, number of typhoons generated per event, maximum intensity, and duration, typhoon cluster events were classified into two categories: regular typhoon cluster events and extreme typhoon cluster events. Regular typhoon cluster events refer to cluster events with a moderate number of typhoons, predominantly typhoons to strong typhoons, relatively stable track and intensity changes, and weak multi-system interactions. Extreme typhoon cluster events refer to high-impact events with a large number of typhoons, generally high intensity, a high likelihood of multi-typhoon interactions, severe track oscillations, and significantly amplified forecast biases. Regular and extreme typhoon cluster events differ fundamentally in their dynamic structure, evolution patterns, and forecast uncertainties. Conventional events are of moderate intensity, have stable paths, and have small forecast deviations, so the general risk assessment method can meet the shipping safety requirements. However, extreme events are of high intensity, with significant interactions between multiple typhoons and amplified forecast deviations. If conventional standards are used, risks may be missed or excessively avoided. Therefore, this invention distinguishes between the two and implements differentiated treatment.

[0045] After completing the two-level classification, further scenario subdivision was conducted only for extreme typhoon cluster events. The subdivision dimensions included the sea area of ​​occurrence, seasonal period, typhoon intensity level, and large-scale circulation background. Ultimately, extreme cluster events were classified into three typical sub-scenarios: ENSO-declining year with a westward-extending subtropical high anomaly, active summer monsoon trough, and extreme number of typhoons. The formation, movement, intensity changes, and forecast error distribution of extreme cluster typhoons are entirely determined by the large-scale circulation background. Different circulation backgrounds lead to significant differences in the direction of typhoon path deviation, wind field expansion patterns, risk superposition areas, and amplified forecast errors. Without subdivision based on circulation background, using the same risk expansion standard for all extreme events would make it impossible to accurately identify high-risk areas and key error areas, hindering targeted correction of the error envelope.

[0046] For the ENSO-declining year with an abnormally westward extension of the subtropical high, it typically occurs in the western part of the Northwest Pacific, concentrated in late summer to autumn, corresponding to the transition from El Niño to La Niña. The subtropical high extends abnormally westward, and the ridge point shifts westward. While the steering flow is stable, it easily leads to westward shifts in typhoon tracks and increased forecast bias. For the active summer monsoon trough type, it mainly occurs in the southern and eastern parts of the Northwest Pacific, concentrated in midsummer. The monsoon trough is deep and extends widely east-west, with abundant moisture transport and vigorous convection, resulting in dense typhoon formation and the frequent coexistence of multiple typhoons. For the extreme number of typhoons type, it is not limited by a single season, but is characterized by a significantly higher number of typhoons generated in a single event. Multiple systems interact strongly, the Fujiwhara effect is pronounced, tracks are complex, and forecast uncertainty is extremely high.

[0047] Based on this, statistical analysis was conducted on samples from various scenarios to extract key circulation parameters, spatial distribution ranges, and historical forecast deviation characteristics. These included the westward extension ridge of the subtropical high, intensity and area anomalies, monsoon trough location and intensity, vertical wind shear range, low-level vorticity, sea surface temperature anomalies, and Nino 3.4 index ranges. Typical value ranges for each indicator under different scenarios were also statistically analyzed. Through comparative analysis, significant differences between conventional scenarios and extreme sub-scenarios in circulation background configuration, spatial distribution patterns, forecast error magnitude, and deviation direction were identified. Correspondence rules between scenarios and circulation characteristics, spatial distribution, and forecast deviation characteristics were established, ultimately forming a typhoon cluster scenario classification library applicable to the research sea area. This provides stable and reliable scenario constraints for subsequent key control area identification, mechanism diagnosis, and indicator construction.

[0048] (1.2) Determination of key control areas and diagnosis of dynamic and thermodynamic mechanisms in different scenarios

[0049] Based on the classification of typhoon cluster events, key control areas were identified for both conventional and extreme sub-scenarios. Furthermore, based on multi-source reanalysis data and observational data, joint dynamic and thermal field mechanism diagnosis was conducted for different scenarios to reveal the formation conditions, evolution patterns, and intrinsic correlations between various cluster events and forecast errors.

[0050] First, based on a typhoon cluster event scenario classification library, key control areas are identified for each scenario. These key control areas refer to the key sea areas where typhoon cluster events occur in concentrated phases, historical path forecast deviations are significant, and the boundaries of the forecast error envelope are prone to abnormal shifts under the corresponding scenario. The location, path oscillation amplitude, and forecast error magnitude of typhoon cluster events are not uniformly distributed across the entire region, but rather exhibit high clustering in specific sea areas. Applying a uniform diagnostic standard and correction rules to the entire sea area would lead to computational redundancy and insufficient accuracy in correcting key areas. Therefore, this invention, by locking onto key control areas, can focus mechanism analysis, indicator extraction, and error correction on high-impact, high-uncertainty regions. This ensures the relevance of physical diagnosis while significantly improving the accuracy of shipping risk identification, enabling targeted correction of key areas.

[0051] After identifying the key control areas, a multi-physical quantity joint diagnosis was conducted within these areas based on ERA5 reanalysis data, optimal typhoon track data, the Nino 3.4 index, and satellite observation data. Indicators such as the Typhoon Formation Potential Index (GPI), Okubo-Weiss vorticity deformation index, convective effective potential energy (CAPE), 200–850 hPa vertical wind shear, 850 hPa low-level absolute vorticity, and the anomalies of the western extension ridge and area of ​​the subtropical high were used to analyze the formation mechanisms of different typhoon cluster scenarios from both dynamic and thermodynamic perspectives. Specifically, for conventional typhoon cluster scenarios, the circulation configuration is relatively stable, and the interaction between multiple systems is weak. The mechanism diagnosis mainly focuses on extracting typical dynamic and thermodynamic backgrounds, identifying stable features related to the baseline level of forecast errors, such as moderate-intensity monsoon troughs, moderate vertical wind shear, and the position of the conventional-intensity subtropical high, providing background constraints for the baseline error envelope area under conventional scenarios. In extreme typhoon cluster scenarios, their formation typically relies on multi-system coupling, extreme dynamic conditions, and strong ocean-atmosphere interaction. Multi-weather system coupling refers to the combined effects of multiple factors, such as subtropical high anomalies, strong monsoon troughs, active equatorial convergence zones, and mid-latitude system disturbances. Extreme dynamic conditions include abnormally high low-level vorticity, vertical wind shear in the optimal range, and strong rotational characteristics of the Okubo-Weiss index, all of which are conducive to typhoon maintenance and orderly wind field expansion. Significantly enhanced ocean-atmosphere interaction is manifested in abnormally high sea surface temperatures, sufficient mid-level humidity, and high convective available potential energy (CAPE), providing ample energy for typhoon cluster formation and intensity maintenance. The above diagnosis reveals the physical connection between extreme dynamic and thermodynamic extremes and forecast uncertainties. When extreme characteristics such as abnormally high low-level vorticity, optimal vertical wind shear, significant deviation of the subtropical high from climatological conditions, and intense interaction among multiple typhoons occur within key control areas, it directly leads to increased typhoon track forecast bias, a shift of the error envelope boundary in a specific direction, and a wind field expansion exceeding conventional expectations. Simultaneously, it intensifies the nonlinear superposition of risks in areas where multiple typhoons converge. Clarifying this underlying physical mechanism allows subsequent indicator selection to move beyond empirical statistics and instead select parameters with strong indicative power for forecast errors and risk distribution based on physical meaning. Furthermore, it provides an interpretable, reproducible, and generalizable mechanistic explanation for adjusting the magnitude, direction, and superposition range of the forecast error envelope for different scenarios.

[0052] Figure 2 This map shows the difference in 500 hPa geopotential height between large-scale circulation patterns in years with multiple typhoon outbreaks and those observed in climatological mean states. It is used to reveal large-scale circulation anomalies corresponding to typhoon outbreaks. Figure 2As shown, the difference in 500 hPa geopotential height between years with multiple typhoon outbreaks and climatological mean-state large-scale circulation reveals a significant positive anomaly (+20 gpm) in the main subtropical high region, with the positive anomaly center in the northeastern South China Sea reaching +28 gpm. The westward extension ridge of the subtropical high shifts westward compared to the climatological state, leading to an increase in the southerly wind component on its western side. Combined with ERA5 reanalysis data, the westward extension anomaly is related to two factors: ENSO decay years and westerly disturbances. A Nino3.4 index above +0.5℃ for four consecutive months followed by a drop to neutral or even negative values ​​within the next seven months is considered an ENSO decay year. The weakening of the equatorial Pacific thermal gradient during El Niño decay years leads to the eastward shift of the western subsidence branch of the Walker circulation. Enhanced subsidence in the western Pacific suppresses convective activity. The enhanced subsidence, through a mass adjustment mechanism, forms positive vorticity advection on the western side of the subtropical high. Positive vorticity advection promotes enhanced low-level convergence, which is conducive to the maintenance of anticyclonic circulation. The westerly disturbances manifest as a positive geopotential height anomaly in the mid-latitudes, with the westerly jet stream axis shifting northward by 3 degrees of latitude. The ascending branch of the secondary circulation in the jet stream inlet region and the descending branch on the north side of the subtropical high form a closed vertical circulation. An anticyclonic anomalous circulation develops south of the westerly jet stream, with its central geopotential height 15 gpm higher than the climatological level. This anticyclone, together with the main body of the subtropical high, forms a "double high-pressure system," enhancing the westward extension of the subtropical high through the principle of potential vortex conservation. Figure 2 It is evident that there is a significant positive geopotential height anomaly in the main region of the subtropical high. The westward extension ridge of the subtropical high extends significantly westward compared to the climatological mean, forming a guiding circulation background that is conducive to the westward movement and westward deviation of the typhoon's path. This is a key circulation factor that leads to increased path forecast bias and westward shift of the error envelope area. This is consistent with the mechanism diagnosis results of the extreme sub-scenario of the subtropical high's anomalous westward extension, and can provide intuitive physical evidence for the westward shift correction of the error envelope area.

[0053] Figure 3 This is a spatial distribution map showing the difference between the CAPE index and the climatological mean convective instability index in years with multiple typhoon outbreaks, used to illustrate the spatial distribution characteristics of atmospheric instability energy during typhoon outbreaks. For example... Figure 3 As shown, the difference between the CAPE index of a year with multiple typhoon outbreaks and the climatological mean convective instability reveals a significant positive anomaly in the western Pacific: the positive CAPE anomaly value over the ocean east of the Philippines exceeds +400 J / kg, with its center located near the Luzon Strait. This region shows a high degree of spatial matching with the location of multiple typhoon formations, confirming the promoting effect of atmospheric instability energy accumulation on typhoon outbreaks. A negative anomaly is observed in the northern South China Sea: a negative anomaly of -300 to -500 J / kg appears in the central and northern South China Sea, possibly related to enhanced subsidence caused by the westward extension of the subtropical high, which inhibits the development of local convection. Figure 3The CAPE in key sea areas of the central and northwestern Pacific Ocean is significantly higher than normal, indicating that the region has vigorous convection development and sufficient thermal conditions, which easily leads to the expansion of typhoon wind field range and the extension of intensity duration. It is an important thermal factor driving the radial expansion of the error envelope area, which is consistent with the adjustment mechanism of dynamic and thermal indicators controlling the wind field expansion range, and provides physical support for the correction of wind field boundary expansion.

[0054] Figure 4 This is a spatial distribution map showing the difference between the Okubo-Weiss vorticity deformation index and the climatological mean-state disturbance dynamic characteristics in years with multiple typhoon outbreaks. This map is used to identify the rotation and deformation characteristics of the flow field during typhoon outbreaks. Figure 4 As shown, the difference between the OW index of years with multiple typhoon outbreaks and the climatological mean-state disturbance dynamic characteristics reveals a significant positive and negative anomaly distribution in the dynamic conditions of the western Pacific. The core of the positive anomaly is located east of the Philippines, with an OW index anomaly value reaching +2.7 × 10⁻⁶. -8 s -2 The area has expanded by 45% compared to climatological levels, and its shape has a high spatial overlap with the initial disturbance centers of multiple typhoons. The southern Japan anomaly zone forms a continuous negative anomaly corridor along the northern edge of the subtropical high, corresponding to the ascending branch of the secondary circulation in the westerly jet stream inlet region. Furthermore, the subtropical high anomaly extends westward, and its southward easterly jet stream forms a horizontal convergence center east of the Philippines, with significantly higher vorticity and advection. This dynamic forcing mechanism leads to the continuous expansion of the positive OW index anomaly area, providing a basic vortex environment for the formation of multiple typhoons. Figure 4 The dense formation area of ​​typhoons corresponds to a significant positive anomaly in the OW index, indicating that the region has strong rotational properties, which is conducive to the maintenance and development of typhoon disturbances. At the same time, the flow field deformation is likely to induce wind field distortion and enhance the interaction between multiple typhoons. This is consistent with the mechanism of risk superposition and expansion controlled by interactive indicators, and can provide a dynamic diagnostic basis for error envelope superposition correction in the scenario of multiple typhoons coexisting.

[0055] Figure 5 This is a spatial distribution map showing the difference between the GPI index for typhoon formation potential in years with multiple typhoon outbreaks and the climatological mean, used to characterize the spatial anomalies of typhoon outbreak potential. The difference between the GPI index for typhoon formation potential in years with multiple typhoon outbreaks and the climatological mean shows that the 200-850 hPa vertical wind shear remains low at 6-8 m / s in the region east of the Philippines (climatological 9-12 m / s), and its negative anomaly contributes to a 25% increase in the dynamic term of the GPI. Combined with ERA5 reanalysis data, the absolute vorticity (850 hPa) in this region reaches 3.2 × 10⁻⁶. -5 s -1The humidity factor increased by 40% compared to the climatological level, significantly enhancing the development of cyclonic vorticity. The humidity factor contributed 28% to the high-potential GPI area in the South China Sea. Many typhoons formed primarily within the high-potential area with a GPI > 0.6, exhibiting a high degree of spatial matching. The high GPI area highly overlaps with the actual typhoon formation location, reflecting the region's superior conditions for typhoon formation and maintenance. This corresponds to the overall increase in forecast uncertainty and the need for overall scaling of the error envelope, matching the mechanism of comprehensive index-based adjustment of the overall scaling factor of the error envelope. This provides a comprehensive potential basis for scenario-based and graded corrections.

[0056] This step, through scenario-based mechanism diagnosis, can clearly identify the dominant influencing factors and key physical processes of different typhoon cluster events, clarify the mechanism differences between normal scenarios and different extreme sub-scenarios, and provide a solid physical basis for the next step of constructing graded and quantitative diagnostic indicators.

[0057] (1.3) Construction of a graded and quantitative diagnostic index system for typhoon outbreaks

[0058] Based on the diagnostic results of the dynamic and thermodynamic mechanisms in different scenarios, key physical parameters that significantly indicate the occurrence, development, intensity evolution, and changes in the boundary and range of the forecast error envelope area were further screened from the diagnostic parameters corresponding to each scenario. The screened parameters were standardized and normalized to unify the dimensions and eliminate differences in numerical magnitude, making them suitable for model calculations and index classification.

[0059] Based on this, and combining the physical characteristics of both conventional typhoon cluster scenarios and various extreme sub-scenarios, threshold ranges for each parameter are determined for each scenario. Since there are significant differences in circulation configuration, dynamic conditions, and thermal background under different scenarios, the critical threshold and risk indication significance of the same parameter differ significantly between conventional and extreme scenarios. Therefore, it is necessary to establish parameter threshold systems applicable to conventional scenarios, the westward extension of the subtropical high during an ENSO-declining year, the active summer monsoon trough, and the extreme scenario with multiple typhoons, respectively, to ensure the relevance and applicability of the indicators to each scenario.

[0060] Next, based on the parameters' indicative ability of typhoon risk, warning threshold level, scenario adaptability, and shipping engineering application weight, a graded value was assigned. Key parameters that significantly reflect path deviation, abrupt intensity changes, and amplified forecast errors were given higher weights, while auxiliary parameters that only characterize background conditions were given standard weights. After completing the threshold determination and graded value assignment, the selected parameters were divided into four categories according to their physical meaning and mechanism of action: circulation background indicators, dynamic indicators, thermal indicators, and comprehensive potential indicators, which together constitute a graded and quantitative diagnostic indicator system for typhoon clusters.

[0061] Preferably, circulation background indices characterize large-scale circulation conditions that control typhoon movement and overall morphology. These are core factors determining the direction of path deviation and the overall shift of the error envelope. Specifically, they include the longitude of the westward-extending ridge of the subtropical high, the subtropical high intensity index, the Nino 3.4 index (ENSO phase), and the intensity and location index of the monsoon trough. Dynamic indices reflect the dynamic forcing conditions of typhoon formation, development, rotation, and wind field expansion, directly affecting typhoon structural stability and the degree of forecast error amplification. These include the 850 hPa lower-level absolute vorticity, the magnitude of the 200–850 hPa vertical wind shear, and the Okubo–Weiss vorticity deformation index. The Okubo–Weiss index can distinguish between rotational and deformation components in the flow field and is a key physical quantity for determining whether the wind field expands in an orderly manner and whether distortion occurs. Thermal indices reflect marine thermal conditions, water vapor supply, and convection development potential, determining the intensity, duration, and size of the typhoon's energy supply and wind field range. These include convective effective potential energy (CAPE), sea surface temperature anomaly (SSTA), and mid-level relative humidity. Comprehensive potential indicators are used to comprehensively reflect the potential for typhoon clusters, the intensity of interactions between multiple typhoons, and the level of uncertainty in historical forecasts. Specifically, they include the Typhoon Formation Potential Index (GPI), the distance and intensity of interactions between multiple typhoons, and the statistical value of historical track forecast deviations.

[0062] The resulting indicator system features scenario adaptability, clear physical meaning, and clear hierarchical quantification, providing a stable and reliable quantitative foundation for subsequent screening of core shipping parameters, training of scenario-specific models, and correction of forecast error envelope.

[0063] (II) Constructing a graded diagnostic model for typhoon outbreaks

[0064] (2.1) Screening of core shipping parameters

[0065] In response to the engineering requirements of maritime shipping for typhoon forecast stability, risk zone delineation accuracy, and route planning reliability, this study conducts correlation and sensitivity analyses on a quantitative diagnostic index system for typhoon cluster occurrence, aiming to improve the accuracy of forecast error envelope correction. By statistically calculating the correlation coefficients between each index and the boundary shift and extent expansion of the forecast error envelope, key indicators with significant impact on the morphological changes of the error envelope are identified, and redundant parameters with weak correlation, lacking prominent physical significance, and low contribution to distinguishing shipping risks are eliminated. Based on this, indicators whose correlation with changes in the boundary and extent of the forecast error envelope exceeds a set threshold under different scenarios are selected and defined as core shipping parameters specific to each scenario.

[0066] The dominant influencing factors and error distribution characteristics differ significantly in the scenarios of regular typhoon clusters, the westward extension of the subtropical high during the ENSO decay year, the active summer monsoon trough, and the extreme number of multiple typhoons. Therefore, the combination of core shipping parameters corresponding to each scenario is significantly different to ensure that the parameters are targeted to the scenario characteristics and that the risk characterization is accurate. For example, in the scenario of a westward extension of the subtropical high during an ENSO-declining year, the longitude of the westward extension ridge point of the subtropical high, vertical wind shear, historical track forecast bias, and the westward shift of the error envelope are highly correlated, with correlation coefficients exceeding a set threshold. Therefore, these parameters were selected as core parameters for shipping in this scenario. In the scenario of an active summer monsoon trough, the intensity of the monsoon trough, low-level absolute vorticity, and CAPE are strong indicators of wind field expansion and southward expansion of the error envelope, thus becoming core parameters for this scenario. In the scenario of an extreme number of typhoons, the interaction distance between multiple typhoons, the Okubo-Weiss vorticity deformation index, GPI and risk nonlinear superposition, and global error amplification are closely related, thus being identified as core parameters. In the scenario of a regular typhoon outbreak, using historical forecast bias statistics, vertical wind shear, and the subtropical high intensity index as core parameters can meet the error envelope correction requirements under basic stable forecasts.

[0067] It should be noted that the screening rules used in this step are not limited to a fixed number and type of scenarios, and have good scalability and universality. When the research area expands, forecasting needs increase, or new extreme event types are added, core shipping parameters suitable for the current scenario can be automatically extracted based on unified correlation and sensitivity criteria, ensuring that the method has stable physical consistency and scenario adaptability under both regular and various extreme typhoon events. The core shipping parameters obtained through this step retain the significance of typhoon dynamics and thermodynamics, and closely align with the engineering requirements of shipping risk assessment, and can be directly used for subsequent scenario-specific model training and forecast error correction coefficient calculation.

[0068] (2.2) Extreme sample feature enhancement and parameter standardization

[0069] To address the issue of scarce sample sizes for extreme typhoon cluster events, which are insufficient to meet model training requirements, this step extracts the core shipping parameter features corresponding to historical extreme typhoon cluster events under each extreme sub-scenario. These features are then categorized and stored according to types such as ENSO decay year subtropical high anomaly westward extension, active summer monsoon trough, and extreme number of typhoons, thus constructing a scenario-specific and structured extreme sample feature library to ensure a one-to-one correspondence between samples and scenario mechanisms.

[0070] Based on this, and constrained by the physical mechanisms of typhoon formation and development, this method interpolates, smooths, and synthesizes existing extreme samples to effectively expand the extreme sample set. Unlike traditional pure mathematical interpolation, which generates samples solely through numerical fitting, this method's sample enhancement process strictly adheres to the physical consistency of the dynamic and thermodynamic fields: interpolation and synthesis must satisfy the inherent constraints between parameters. For example, vertical wind shear must match the typhoon intensity range, low-level vorticity must be coordinated with the monsoon trough / subtropical high configuration, sea surface temperature anomalies must maintain a reasonable correspondence with convective effective potential energy (CAPE), and the interaction distance between multiple typhoons must be physically compatible with the path oscillation amplitude. Parameter combinations that deviate from actual atmospheric evolution patterns must be avoided. For instance, when expanding samples for the westward extension of the subtropical high anomaly during an ENSO decay year, the interpolation process maintains a reasonable configuration between the longitude of the westward extension ridge point of the subtropical high, vertical wind shear, and historical forecast bias. Smoothing interpolation is performed only within the typical threshold ranges of each parameter, preventing the generation of false samples where the ridge point anomaly is too westward but the vertical wind shear is too small, contradicting the physical mechanisms. When synthesizing samples for active summer monsoon trough scenarios, parameters such as monsoon trough intensity, low-level absolute vorticity, and CAPE are ensured to be simultaneously within ranges conducive to dense typhoon formation, maintaining the coordinated changes in dynamic and thermodynamic conditions. In extreme scenarios with multiple typhoons, sample augmentation follows the physical laws governing the interaction of multiple typhoons, ensuring reasonable matching of parameters such as typhoon spacing, vorticity deformation index, and GPI, avoiding physically meaningless extreme combinations. Through sample augmentation constrained by the above mechanisms, it is ensured that the expanded extreme samples still conform to the spatiotemporal evolution characteristics of real typhoon cluster events, without introducing false, anomalous, or non-physical samples, thereby significantly improving the generalization ability of subsequent models under small sample conditions and the reliability of forecast error correction.

[0071] After sample augmentation, the core shipping parameters under each extreme sub-scenario are normalized to a unified dimension, mapping parameters of different types and magnitudes to the same numerical range and eliminating the impact of dimensional differences on model training. Based on the distribution characteristics and physical meaning of parameters in each scenario, a parameter standardization mapping table for each extreme sub-scenario is established, achieving a stable conversion from raw parameters to standardized inputs. This provides a standardized, unified, and reusable data foundation for the subsequent construction of scenario-based hierarchical diagnostic models. Furthermore, the sample augmentation and standardization rules used in this step are scalable. When adding new extreme sub-scenarios or expanding the sample set, the same mechanism constraints and standardization process can be used, ensuring consistency and applicability of the method under different scenarios and data conditions.

[0072] (2.3) Construction of a tiered diagnostic model for typhoon outbreaks in different scenarios

[0073] Based on the core shipping parameters specific to the conventional scenario obtained in step S21, the core parameters of each extreme sub-scenario after mechanistic constraint enhancement and standardization in step S22, and the expanded equilibrium sample set, a typhoon cluster classification diagnostic model corresponding to the conventional scenario and each extreme sub-scenario is constructed using multiple linear regression analysis. The model construction process takes maximizing the accuracy of the forecast error envelope correction as the unified optimization objective, ensuring that the output correction coefficients under different scenarios accurately match the actual forecast uncertainty distribution.

[0074] The core objective of this step is to transform the selected core shipping parameters, processed standardized parameters, and sample data into forecast error correction coefficients that can be directly used to correct the forecast error envelope. Through scenario-based modeling, the accuracy of error correction under different scenarios is ensured, adapting to the accuracy requirements of risk zone delineation in maritime shipping typhoon avoidance route planning. The specific implementation process is as follows:

[0075] (2.3.1) Data preparation for modeling

[0076] In the data preparation phase of modeling, modeling datasets for both conventional and extreme sub-scenarios were constructed. For the conventional scenario, selected core shipping parameters, such as the longitude of the western extension ridge of the subtropical high, the 200-850 hPa vertical wind shear, and the 850 hPa low-level absolute vorticity, were used, along with sufficient historical samples to form the basic data. From these, the magnitude of path forecast deviation, the wind field expansion radius, and the interaction effects of multiple typhoons were extracted to form a complete model input-output structure, serving as the basic dataset for modeling. The interaction effects of multiple typhoons can be characterized by the distance between their centers and the intensity superposition coefficient. For each extreme sub-scenarios, the core shipping parameters, after being expanded and standardized by physical mechanism constraints, and the expanded extreme sample sets were used. Similarly, the magnitude of path forecast deviation, the wind field expansion radius, and the interaction effects of multiple typhoons were extracted for each sample set to construct modeling datasets for each extreme sub-scenarios. Each extreme sub-scenarios had at least 30 samples to avoid overfitting.

[0077] (2.3.2) Establishment of Quantization Mapping Relationship

[0078] In constructing the quantitative mapping relationship, a multiple linear regression method is adopted to establish multi-objective quantitative correlations between shipping core parameters and forecast error characteristics for both conventional and various extreme sub-scenarios. The model uses the shipping core parameters corresponding to each scenario as input, and establishes multi-dimensional, multi-component regression mapping relationships for three key uncertainty features: typhoon track forecast deviation magnitude, wind field expansion radius, and multi-typhoon interaction effects. This ultimately forms a set of independently applicable correction coefficients.

[0079] Taking a specific extreme sub-scenario as an example, regression models containing multiple components are constructed for each of the three types of features:

[0080]

[0081]

[0082]

[0083] In the formula, This is the path deviation correction coefficient vector, corresponding to the amplification coefficients for the eastward, westward, and meridional deviations of the path, respectively; This is the vector of wind field expansion correction coefficients, which correspond to the amplification coefficients of the wind field expansion in the four quadrants of the typhoon, respectively. This is a vector of correction coefficients for the interaction of multiple typhoons, corresponding to the amplification coefficients for the superposition of the distance and intensity of multiple typhoons. The standardized shipping core parameter vector includes the standardized subtropical high westward extension ridge point index, the standardized Okubo-Weiss vorticity deformation index, the standardized convective effective potential energy (CAPE), and the standardized multi-typhoon interaction intensity. , and This is the weight matrix for the corresponding scenario. Initial values ​​can be set based on physical mechanism constraints, and subsequent back-optimization is performed in step S24. Constant term. , and It is obtained by fitting historical samples and is used to correct regression bias.

[0084] The aforementioned vector-based regression model can simultaneously output multiple correction components for the same feature type. Each component is dimensionless, typically ranging from 1.0 to 2.5; the larger the value, the greater the correction magnitude for the corresponding dimension's error envelope. In application, different components can be applied separately to the corresponding directions or regions of the error envelope to achieve asymmetric, regionally refined correction. For example, in scenarios where the path prediction deviation is significantly westward, only the western boundary of the error envelope can be amplified to avoid excessive avoidance caused by uniform expansion across the entire region.

[0085] (2.3.3) Construction and Optimization Objectives of Scenario-based Hierarchical Diagnostic Model

[0086] After constructing the regression relationships, they are encapsulated into independent scenario-specific diagnostic models. The conventional scenario model is suitable for clustered events with relatively stable circulation and dynamic conditions, and its output correction coefficients are generally low, corresponding to smaller adjustments to the error envelope. Various extreme sub-scenario models, on the other hand, target events with anomalies in circulation, multi-system coupling, and significant interactions between multiple typhoons, outputting higher correction coefficients to match larger forecast biases and risk expansion. The independence of each scenario model avoids mutual interference between different error characteristics, improving the overall correction effect.

[0087] The input to the standard scenario model is the standard scenario identifier and real-time meteorological parameters. The real-time meteorological parameters are used to extract the real-time values ​​of the core parameters specific to the standard scenario, which are then substituted into the regression formula to calculate the forecast error correction coefficient under the standard scenario. The value is usually 1.0-1.2, which is relatively small and does not require significant correction. The input to the extreme sub-scenario model is the corresponding extreme sub-scenario identifier and real-time meteorological parameters. The real-time meteorological parameters are used to extract the real-time values ​​of the core parameters specific to the extreme sub-scenario, which are then substituted into the corresponding regression formula to calculate the forecast error correction coefficient under the extreme sub-scenario. The value is usually 1.5-2.5, which is relatively large and requires a significant increase in the error envelope area.

[0088] The optimization objective of the scenario-based hierarchical diagnostic model is clearly defined as maximizing the accuracy of the forecast error envelope correction. This is achieved by adjusting the weight coefficients of each core parameter so that the correction coefficients output by the model ensure that the corrected error envelope matches the actual impact range of historical typhoons by more than 90%, thus meeting the safety requirements of maritime shipping typhoon avoidance route planning. The scenario-based hierarchical diagnostic model of this invention adopts a construction method of scenario decoupling, independent modeling, and unified objectives. It does not rely on a fixed number of scenarios and has good scalability. When a new extreme cluster event sub-scenario is added to the research sea area, the sub-model can be independently trained using the corresponding scenario's shipping core parameters and enhanced samples, following the same modeling process. There is no need to reconstruct the overall framework, ensuring the method has stable applicability and physical consistency under different sea areas and event types. Scenario-based modeling effectively avoids sample interference between routine events and extreme events, as well as events with different extreme mechanisms, significantly improving the model's prediction reliability under small sample and high uncertainty conditions.

[0089] (2.3.4) Model Validation

[0090] After model training is complete, independent samples not involved in modeling are used for validation. For example, 10 sets of historical samples not involved in modeling are selected (3 sets for the normal scenario and at least 2 sets for each extreme sub-scenario). The core parameters in the samples are input into the hierarchical diagnostic model of the corresponding scenario to calculate the forecast error correction coefficient. This coefficient is then used to correct the normal forecast error envelope area, and the consistency between the corrected envelope area and the actual impact range of the typhoon is verified. If the consistency does not reach 90%, the process returns to step (2.3.3) to perform reverse optimization of the model parameters until the accuracy requirements are met.

[0091] The typhoon cluster outbreak classification diagnostic model constructed in this invention uses the core shipping parameters corresponding to each scenario as the model input variables, and takes three key uncertainty features, namely the magnitude of typhoon path forecast deviation, wind field expansion radius, and multi-typhoon interaction effect, as intermediate output variables. It is fitted and trained through dual constraints of physical meaning and statistical law, and finally outputs the forecast error correction coefficient applicable to the current scenario, so as to realize the scientific, quantitative and differentiated correction of the error envelope area.

[0092] (2.4) Model inverse parameter optimization and conventional scene calibration

[0093] This step addresses the issues of limited samples in extreme scenarios and the tendency for initial model weights to be biased towards conventional states through reverse optimization under physical constraints. It also calibrates the model for conventional scenarios, ensuring that the error correction accuracy in both scenarios meets shipping safety requirements.

[0094] I. Inverse parameter optimization of extreme sub-scene models

[0095] In step (2.3), the initial weight coefficients of the model are obtained only based on statistical fitting, without fully considering the extreme dynamic and thermodynamic characteristics of extreme events. This can easily lead to the model fitting well under normal conditions, but the prediction deviation is amplified under extreme conditions such as extreme circulation configurations and strong multi-typhoon interactions. Therefore, this step utilizes the extreme dynamic and thermodynamic characteristics of each extreme sub-scenario to perform inverse parameter optimization on the corresponding hierarchical diagnostic model, adjusting the influence weights of different shipping core parameters on the forecast error, so that the prediction accuracy of the forecast error envelope trend under extreme conditions is significantly improved. Specifically, by using the extreme characteristics of circulation, dynamics, and thermodynamics in extreme events as physical constraints for adjusting model weights, the weights of core parameters that are highly indicative of extreme forecast errors are increased through backpropagation or constraint optimization algorithms, while reducing the interference of non-critical parameters, so that the correction coefficients of the model under extreme conditions are more in line with the actual error amplification law.

[0096] For example, in the extreme sub-scenarios of a westward-extending subtropical high during an ENSO-declining year, extreme events commonly exhibit characteristics such as an abnormally westward shift of the subtropical high ridge, weaker vertical wind shear, and significant westward deviation in the path. During optimization, the weighting coefficients of the subtropical high westward-extending ridge index and vertical wind shear on the path deviation correction coefficient are appropriately increased. This allows the model to automatically increase the amplification ratio of the westward error envelope when detecting similar extreme circulation configurations, resulting in a more accurate match with actual forecast deviations. In the extreme sub-scenarios of an active summer monsoon trough, extreme events often exhibit characteristics such as a stronger monsoon trough, higher low-level vorticity, higher convective effective potential energy (CAPE), and significant wind field expansion on the south side of the typhoon. During optimization, by increasing the weighting coefficients of monsoon trough intensity, low-level vorticity, and CAPE on the southward component of the wind field expansion, the model can more reasonably amplify the southward wind field error envelope under strong thermal convection conditions. In the extreme scenario of multiple typhoons, the close interaction distance and significant intensity superposition effects of multiple typhoons amplify both path deviations and wind field distortions. During the optimization process, by increasing the weight of the multi-typhoon interaction intensity and the Okubo-Weiss vortex deformation index on the correction coefficient of multi-typhoon interaction, the model can more accurately reflect the nonlinear superposition effect of risks under extreme conditions of multiple typhoons coexisting. Through the aforementioned reverse optimization under physical constraints, the extreme sub-scenario model can better capture the asymmetric amplification law of the error envelope region in extreme events, avoiding the problem of insufficient model generalization ability due to insufficient sample size.

[0097] II. Calibration of Standard Scene Model Parameters

[0098] While optimizing the extreme sub-scenario model, routine parameter calibration was performed on the hierarchical diagnostic model for normal scenarios. Since normal scenarios have sufficient samples and relatively stable forecast error distributions, the calibration process used the statistical values ​​of forecast errors from historical normal typhoon cluster events as a benchmark, fine-tuning the model weights and constant terms to ensure the accuracy of forecast error correction under normal scenarios. The core objective of calibration is to ensure that the correction coefficients output by the model are consistent with the statistical distribution characteristics of forecast errors under normal scenarios, avoiding both excessive amplification leading to increased flight path avoidance costs and insufficient correction resulting in missed risk assessments.

[0099] (2.5) Scene discrimination and prediction error correction coefficient calculation

[0100] This step uses real-time data on typhoon location, intensity, and environmental field to determine the current typhoon cluster scenario, calls the corresponding scenario's hierarchical diagnostic model, and calculates the forecast error correction coefficient for that scenario. This provides accurate and real-time quantitative basis for subsequent fine-grained correction of the error envelope area.

[0101] First, based on real-time acquired typhoon location, intensity, and environmental field data, a real-time feature vector is constructed. The environmental field data focuses on extracting key parameters such as circulation background and dynamic / thermal parameters. These parameters are then compared with the typhoon cluster scenario classification library established in step S11. A scene recognition algorithm is used to determine the current typhoon cluster scenario, classifying it as a typical typhoon cluster scenario or an extreme sub-scenario such as an ENSO-declining subtropical high anomalous westward extension, an active summer monsoon trough, or an extreme multi-typhoon scenario. The scene identification process strictly adheres to the established correspondence rules between scenarios and circulation features and spatial distribution. It combines real-time parameters with the typical feature threshold ranges of each scenario in the classification library to ensure the accuracy and timeliness of scene identification and avoid deviations in the correction coefficient due to misjudgment.

[0102] After scenario identification is completed, the corresponding typhoon cluster outbreak classification and diagnostic model is automatically invoked, and the real-time extracted and standardized shipping core parameters are input into the model. Based on the preset multiple linear regression mapping relationship and the optimized weight parameters, the model calculates the forecast error correction coefficient vector for the scenario, that is, outputs the correction components for each direction of path deviation, the correction components for each quadrant of wind field expansion, and the correction components for each dimension of multi-typhoon interaction. Each component accurately corresponds to the forecast error amplification characteristics in the current real-time scenario.

[0103] For example, if a typhoon cluster event is detected in real time, with its subtropical high ridge point significantly westward, vertical wind shear in an extreme range, and corresponding to the circulation characteristics of an ENSO decay year, and the scenario is identified as an extreme sub-scenario of an abnormal westward extension of the subtropical high during an ENSO decay year, then the hierarchical diagnostic model for that scenario is automatically invoked. The model inputs core parameters such as the real-time standardized subtropical high ridge point index and vertical wind shear. The model outputs a correction coefficient vector including a westward path deviation correction component and a westward wind field extension correction component, providing quantitative support for subsequent targeted correction of the error envelope. If the scenario is identified as a regular typhoon cluster event, then the regular scenario model is invoked, outputting correction coefficients adapted to regular error characteristics, ensuring that the correction magnitude matches the actual error level. The entire process requires no manual intervention, automating scenario identification, model invocation, and coefficient calculation. This ensures the real-time performance of the correction coefficients while leveraging the advantages of scenario-based modeling to guarantee their relevance and accuracy, providing reliable support for subsequent error envelope correction and shipping route planning.

[0104] (III) Dynamic adjustment of error envelope and real-time trajectory planning

[0105] (3.1) Constructing expansion and offset standards for each scenario

[0106] The error characteristics of typhoon paths, intensities, and interaction effects vary significantly under different scenarios. The corresponding error envelope adjustment logic also needs to be designed differently. It is necessary to achieve precise correction of the error envelope in terms of direction, range, and superposition effect in order to both conform to the physical mechanism and meet the actual needs of shipping risk prevention and control.

[0107] This step combines the path, wind field, and interaction multi-component correction logic mentioned earlier. By clarifying the index classification and establishing multi-dimensional adjustment logic, it solves the problem that traditional single correction coefficients cannot adapt to multiple error sources, ensuring that the adjustment of the error envelope region not only conforms to the laws of atmospheric motion but also meets the actual needs of shipping safety.

[0108] First, the core parameter influence mechanisms under various scenarios are decomposed. Combining the hierarchical diagnostic model constructed earlier, the core factors affecting the error envelope area are divided into three categories, each corresponding to different adjustment dimensions: First, circulation indicators (such as the position of the subtropical high and the intensity of the monsoon), which mainly affect the typhoon track forecast deviation correction coefficient and are used to control the overall azimuth shift of the error envelope area; second, dynamic and thermal indicators (such as low-level vorticity and vertical wind shear), which mainly affect the wind field expansion radius correction coefficient and are used to control the radial scaling of the error envelope area; and third, interaction indicators, which mainly affect the multi-typhoon interaction correction coefficient and are used to control the nonlinear superposition and expansion range of the risk area under the joint influence of multiple typhoons, mainly determining the superposition and expansion range of the error envelope area.

[0109] Based on this, and considering the differences in circulation configuration, dynamic and thermal conditions, and the intensity of multi-typhoon interactions under various typhoon cluster scenarios, a multi-dimensional adjustment logic for the error envelope region is established. The quantitative principles for offset direction, expansion magnitude, and overall scaling under various scenarios are clarified, ultimately forming standards for error envelope region expansion and offset suitable for each scenario. This provides a unified basis for subsequent refined corrections based on forecast error correction coefficients. For example, for conventional typhoon cluster scenarios with stable circulation, regular typhoon structures, and no significant multi-system interference, a conservative, mild adjustment strategy is adopted, only making small-scale overall scaling and boundary corrections to the error envelope region. This ensures sufficient risk coverage while avoiding excessive expansion that could cause unnecessary route detours, balancing navigation safety and operational efficiency. For various extreme sub-scenarios, differentiated and asymmetric adjustments are implemented based on their typical physical characteristics: In the scenario of a westward-extending subtropical high anomaly during an ENSO-declining year, the error envelope region is prone to significant westward shift accompanied by abnormal steering airflow. Therefore, the standard explicitly prioritizes strengthening the offset rules of the path deviation direction and correspondingly increases the expansion of the western wind field boundary. In the scenario of an active summer monsoon trough, strong water vapor and convection conditions easily lead to a significant expansion of the typhoon wind field range. The standard focuses on strengthening the wind field boundary expansion rules and increasing the outward expansion ratio of the risk areas on the south and east sides. In the extreme scenario of multiple typhoons, the mutual attraction and circulation superposition effects between typhoons are prominent. The standard focuses on supplementing the superposition expansion rules of the multi-typhoon affected areas, performing more sufficient expansion and scaling on overlapping risk areas, and avoiding risk omissions due to nonlinear superposition. Through the construction of the above-mentioned sub-scenarios standards, the error envelope region can be coordinated and standardized in three dimensions: overall scaling, wind field boundary expansion, and path deviation direction offset. This forms a complete connection with the subsequent quantitative correction process based on the forecast error correction coefficient.

[0110] (3.2) Scenario-based calibration of the expansion and offset standards of each scenario

[0111] The error envelope expansion and offset standards established in step (3.1) are general rules based on the physical mechanisms of typhoons, but they do not fully consider the actual engineering requirements of maritime shipping. Ship navigation needs to balance safety priorities and navigation efficiency; excessive correction will increase route detour costs and extend sailing time, while insufficient correction will lead to typhoon avoidance safety risks. Therefore, this invention quantifies and adapts to shipping needs, adjusting the stringency of the expansion and offset standards for each scenario. This ensures that the correction magnitude of the error envelope is precisely matched with the degree of anomaly in the core shipping parameters and the level of forecast uncertainty, achieving a dual balance between safety and efficiency. Specifically, the calibration logic follows the principle that the higher the uncertainty and the more significant the anomaly in the core parameters, the more stringent the standard after calibration and the larger the correction magnitude; conversely, the lower the uncertainty and the closer the core parameters are to normal levels, the more moderate the standard after calibration and the smaller the correction magnitude. Simultaneously, considering the actual engineering aspects of ship navigation, the calibration focus for different scenarios is differentiated: normal scenarios prioritize efficiency with a safety safety net, while extreme scenarios prioritize safety and comprehensive coverage, ensuring that the calibrated standards conform to atmospheric physical laws and are adapted to the actual needs of typhoon avoidance route planning.

[0112] For example, in normal scenarios, key shipping parameters, including the position of the subtropical high, vertical wind shear, and low-level vorticity, are all within normal threshold ranges, resulting in low forecast uncertainty. Ship navigation efficiency takes precedence over excessive typhoon avoidance. Therefore, the calibration focus is on reducing unnecessary corrections. Only minor adjustments to the established standard for normal scenarios are needed, reducing the wind field boundary expansion by 10%-15% and keeping the path deviation within a minimum reasonable range. This maintains the overall scaling factor between 1.0 and 1.2, ensuring that the error envelope only covers the necessary risk area. This avoids excessively long detour routes and increased fuel consumption due to excessive expansion, balancing navigation safety and operational efficiency.

[0113] For scenarios involving a westward extension of the subtropical high during an ENSO-declining year, the subtropical high ridge index is significantly anomalous, resulting in extremely high uncertainty regarding the westward deviation of the track forecast. This poses a significant risk of track misalignment for vessels, making safety a far greater priority than efficiency. Therefore, the calibration focus should be on strengthening the track misalignment and westward wind field expansion standards. The set westward track misalignment amplitude can be increased by 20%-25%, while simultaneously increasing the outward expansion ratio of the westward wind field boundary. The overall scaling factor should be adjusted to 1.6-1.8 to ensure that the error envelope fully covers the potential risks of westward track misalignment, preventing vessels from unwittingly entering the typhoon's impact zone due to track misalignment.

[0114] In scenarios with active summer monsoon troughs, core dynamic and thermodynamic parameters such as trough intensity and CAPE are abnormally high, significantly increasing the uncertainty of wind field expansion. Typhoon wind field extent is prone to sudden expansion, and the main threat to ship navigation stems from the unpredictability of wind field boundaries. Therefore, the calibration focus is on optimizing wind field expansion standards and performing tiered calibration of the set wind field boundary expansion range: when the core parameter anomaly is low, the wind field expansion range can be increased by 10%-15%; when the core parameter anomaly is high (e.g., CAPE exceeds twice the threshold), the wind field expansion range can be increased by 30%-35%, while appropriately reducing the overall scaling factor to avoid excessive wind field expansion while ensuring that potential risks at the wind field boundary are fully covered.

[0115] In extreme scenarios involving multiple typhoons, core interaction parameters such as the intensity of typhoon interactions and vortex deformation index become abnormal, resulting in significant risk superposition effects and peak forecast uncertainty. Ships face multi-dimensional risks, encompassing path deviation, wind field superposition, and range expansion. Therefore, calibration focuses on strengthening the superposition expansion and overall scaling standards. The expansion range of the set overlapping area of ​​multiple typhoons can be increased by 25%-30%, and the overall scaling factor adjusted to 1.8-2.0. Simultaneously, the path deviation direction can be fine-tuned to ensure that the error envelope covers the path deviation risk and wind field superposition risk caused by the mutual attraction of multiple typhoons, maximizing ship navigation safety. In this case, some navigation efficiency is sacrificed to ensure a safety net.

[0116] Through the aforementioned scenario-based calibration, the expansion and offset standards for each scenario are no longer simply general rules based on physical mechanisms, but rather quantitative standards that fully integrate the needs of shipping engineering. The calibrated standards can accurately match the degree of anomaly and forecast uncertainty of core shipping parameters under different scenarios, avoiding efficiency losses caused by over-correction in conventional scenarios and solving safety risks caused by insufficient correction in extreme scenarios, thus achieving a deep fit between physical mechanisms and shipping needs.

[0117] (3.3) Forecast error range correction

[0118] The development of typhoon cluster events is characterized by dynamic evolution. The location, intensity, and environmental parameters of typhoons continuously change over time, and the corresponding forecast uncertainties and correction needs are also dynamically adjusted. Therefore, it is necessary to ensure the timeliness of forecast error correction coefficients by updating data in real time. Then, using the correction coefficients as quantitative support, and following calibrated scenario-based standards, corrections are performed in three dimensions: overall scaling, wind field boundary expansion, and path deviation direction offset. Ultimately, this results in a forecast error range that accurately covers actual risks and adapts to shipping needs.

[0119] First, typhoon warning data and real-time meteorological parameters are updated in real time. The typhoon warning data includes the typhoon's real-time location, intensity, speed of movement, and path forecast. The real-time meteorological parameters focus on extracting core shipping parameters specific to each scenario to ensure the real-time nature and accuracy of the data, providing reliable input for the dynamic updating of correction coefficients.

[0120] Next, based on the updated real-time parameters, the dynamic evolution of the current typhoon cluster scenario is automatically determined. If the scenario remains unchanged, the original scenario identifier is maintained. If the environmental field parameters undergo significant changes, the scenario determination process is re-executed, the scenario-based hierarchical diagnostic model optimized by reverse engineering is called, and the real-time standardized shipping core parameters are substituted in to dynamically update the forecast error correction coefficient vector for the corresponding scenario. That is, the values ​​of each component of the path deviation correction coefficient, wind field expansion correction coefficient, and multi-typhoon interaction correction coefficient are updated simultaneously to ensure that the correction coefficients accurately match the real-time status of the current typhoon cluster and the level of forecast uncertainty.

[0121] Subsequently, using the dynamically updated forecast error correction coefficient as the core quantitative basis, and following the scenario-based calibration standards for expansion and offset, multi-dimensional fine-grained corrections were performed on the initial forecast error range within key control areas. Simultaneously, three operations were executed: overall scaling, wind field boundary expansion, and path deviation direction offset. Specifically, overall scaling, based on the correction coefficient components corresponding to comprehensive indicators, was used to balance and adjust the error envelope region according to the calibrated scaling ratio, adapting to the overall level of current forecast uncertainty. Wind field boundary expansion, based on the correction coefficient components corresponding to dynamic and thermal indicators, was used to asymmetrically expand the wind field influence boundary according to the calibrated expansion magnitude, targeting the wind field expansion characteristics of different quadrants of the typhoon, focusing on covering the potential risks of sudden wind field expansion. Path deviation direction offset, based on the correction coefficient components corresponding to circulation indicators, was used to directionally offset the error envelope region according to the calibrated offset direction and magnitude, accurately adapting to the actual trend of path forecast deviation. In the case of multiple typhoon scenarios, the correction coefficient components corresponding to interactive indicators were combined to simultaneously expand the overlapping risk areas of multiple typhoons, ensuring that the risk superposition effect was fully covered. Through the above multi-dimensional collaborative correction, the final corrected forecast error range is formed. This range not only fits the real-time evolution characteristics of typhoon clusters, but also fully integrates physical mechanism constraints and shipping engineering requirements. It can accurately cover various risks brought about by typhoon path deviation, wind field expansion and multiple typhoon superposition, while avoiding the increase in route detour costs caused by over-correction. It provides accurate and real-time risk boundary support for the dynamic planning of subsequent ship typhoon avoidance routes.

[0122] (3.4) Secondary calibration

[0123] The same error envelope poses varying degrees of safety threat to different vessels. Therefore, considering the differences in risk tolerance among vessels, this invention transforms a general error range into a vessel-adaptive error envelope through secondary calibration. By combining the vessel's own wind resistance capabilities, the boundary range and risk threshold of the error envelope are adjusted. This ensures that the calibrated envelope meets the safety requirements of different vessels while avoiding excessive avoidance or missed risk assessments due to uniform standards, achieving a precise match between typhoon risk and vessel capabilities. Specifically, firstly, the core navigation parameters of the currently navigating vessel are acquired in real time, including the vessel's wind resistance level, actual speed, and load. Simultaneously, combined with vessel design specifications and navigation safety standards, the vessel's own wind resistance capabilities and risk tolerance thresholds are quantitatively assessed, generating a vessel wind resistance capability assessment report, providing a quantitative basis for secondary calibration. Then, combined with the corrected forecast error range, and using the vessel's wind resistance capability as the core constraint, a secondary fine-tuning calibration is performed on the wind field boundary, path offset range, and overall scaling amplitude of the error envelope, ultimately obtaining the vessel-adaptive error envelope.

[0124] For example, large ocean-going container ships with a wind resistance rating of 10, a speed of 18 knots, and no cargo load have strong wind resistance, good maneuverability, and high tolerance to typhoon wind fields and path deviations. Therefore, the focus of secondary calibration is to appropriately shrink the error envelope area. Based on the correction results, the wind field boundary expansion can be reduced by 15%-20%, the path deviation range can be reduced by 10%-15%, and the overall scaling factor can be lowered by 0.1-0.2 to avoid excessively long route detours due to over-correction, thus balancing safety and navigation efficiency. For small and medium-sized coastal cargo ships with a wind resistance rating of 8, a speed of 12 knots, and full load, these ships have weaker wind resistance, reduced maneuverability under full load conditions, and lower tolerance to sudden changes in typhoon wind fields and path deviations. Therefore, the focus of secondary calibration is to appropriately expand the error envelope area. Based on the correction results, the wind field boundary expansion range can be increased by 10%-15%, the path offset range by 10%, and the overall scaling factor by 0.1-0.2. This emphasizes strengthening the risk coverage of the wind field boundary to avoid safety accidents caused by insufficient wind resistance of vessels. For small fishing vessels with a wind resistance rating of level 6 and a sailing speed of 8 knots, these vessels have weak wind resistance and poor maneuverability, resulting in extremely low tolerance for typhoon risks. Therefore, secondary calibration requires a significant expansion of the error envelope area. Based on the correction results, the wind field boundary expansion range can be increased by 25%-30%, the path offset range by 20%, and the overall scaling factor by 0.3-0.4. Simultaneously, the coverage of overlapping typhoon risk areas should be expanded to minimize the probability of vessels accidentally entering risk areas and prioritize ensuring navigational safety. For special transport vessels such as LNG carriers, the requirements for navigational safety are extremely high, and the characteristics of their cargo dictate that they must strictly avoid typhoon risks. Therefore, the secondary calibration adopts a conservative adjustment strategy. Based on the correction results, the wind field boundary and path offset range remain unchanged, and the focus is on optimizing the boundary accuracy of the error envelope area to ensure that the envelope area can accurately cover potential risks, while avoiding excessive expansion that would increase the difficulty of route planning.

[0125] The secondary calibration uses ship navigation parameters as a quantitative basis, which not only does not deviate from the physical mechanism of typhoons and the correction logic mentioned above, but also fully considers the actual needs of shipping engineering. This allows the final ship-adaptive error envelope to provide a personalized risk boundary for subsequent typhoon avoidance route planning. Different types and states of ships can formulate more reasonable and efficient typhoon avoidance routes based on their own adaptive error envelope, which not only ensures navigation safety, but also minimizes detour costs, achieving a dual optimization of safety and efficiency, and perfectly connecting with the subsequent route planning stage.

[0126] (3.5) Real-time route planning

[0127] Typhoon paths and wind field ranges exhibit significant dynamic evolution characteristics, and ship navigation status also adjusts in real time according to sea conditions and navigation needs. A single error envelope correction and route planning cannot adapt to the dynamic changes of typhoons and ships. Therefore, this invention continuously compares the overlap between the ship-adaptive error envelope and the actual typhoon impact area to determine whether the current error envelope meets the dual thresholds of safety and efficiency for ship navigation. If not, it iteratively updates the correction coefficient and adjusts the error range to ensure that the error envelope always accurately covers the actual typhoon risk while adapting to the real-time navigation status of ships. Ultimately, it formulates a safe, efficient, and dynamically adjustable typhoon avoidance route until route planning is completed or the typhoon warning is lifted.

[0128] This step relies on the marine meteorological monitoring system and the ship navigation monitoring system to collect real-time monitoring data such as the actual movement path and actual wind field range of the typhoon, including wind field intensity, radius of influence, and quadrant distribution. Simultaneously, it acquires the ship's current navigation status, including real-time position, navigation direction, remaining range, and fuel reserves. The real-time collected actual typhoon impact area is spatially overlaid with the ship's adaptive error envelope area for analysis. The overlap between the two is quantitatively calculated to characterize the coverage of the error envelope area with the actual typhoon risk. A dual threshold of safety and efficiency is preset. For example, the safety threshold is set to 90%, and the efficiency threshold is set to 30%. That is, the proportion of the error envelope area covering the actual typhoon risk is not less than 90%, and the proportion of route detours caused by excessive coverage does not exceed 30%. This serves as the core basis for judging whether iterative adjustments are needed. If the overlap rate meets the preset safety and efficiency thresholds, it means that the current ship adaptation error envelope can accurately cover the actual risks of the typhoon without causing a serious decrease in navigation efficiency due to over-coverage. Based on this error envelope, and considering the ship's wind resistance, current navigation status, and destination route requirements, a real-time typhoon avoidance route is formulated, specifying key parameters such as the detour path, turning point, and navigation speed to ensure the route stays away from the typhoon risk area while minimizing the detour distance. If the overlap rate does not meet the preset thresholds, for example, less than 90%, it indicates insufficient risk coverage and a potential safety hazard of the ship accidentally entering the typhoon's impact area; if the overlap rate is greater than 120%, it indicates over-coverage, resulting in an excessively long detour distance and affecting navigation efficiency. In this case, the iterative adjustment process is immediately initiated.

[0129] Combining the ship's current wind resistance capability and real-time navigation status, the data is fed back to the graded diagnostic model. The real-time typhoon monitoring data and the ship's real-time parameters are re-introduced, and the forecast error correction coefficient vector for the corresponding scenario is iteratively updated. Based on the updated correction coefficients, the expansion range, offset direction, and overall scaling ratio of the forecast error range are dynamically adjusted, and a new ship-adaptive error envelope is regenerated.

[0130] The aforementioned process is repeated until the overlap meets the preset threshold, real-time route planning is completed, or the typhoon warning is lifted and ships no longer need to detour, ultimately achieving real-time, accurate, and efficient typhoon avoidance route planning under the dynamic evolution of the typhoon.

[0131] For example, a fully loaded, medium-sized coastal cargo ship with a wind resistance rating of level 8 was navigating when real-time monitoring showed that the actual path of the typhoon was 15 km east of the predicted path. The actual wind field's eastern expansion was greater than the coverage area of ​​the error envelope, with an overlap of only 82%, indicating a potential risk of underestimation of the eastern wind field. In this case, an iterative adjustment was immediately initiated: considering the ship's relatively weak wind resistance under full load, feedback was fed into the scenario-specific diagnostic model, increasing the eastern component of the wind field expansion correction coefficient by 0.2 and the eastward component of the path deviation correction coefficient by 0.15. Based on the updated correction coefficients, the eastern wind field boundary of the error envelope was expanded by 15% and shifted eastward by 10 km, regenerating a ship-adaptive error envelope. The overlap was compared again; if it reached 92%, the route was adjusted based on this envelope, moderately detouring eastward to ensure the ship stayed away from the strong wind area on the eastern side of the typhoon. If this was still insufficient, the iteration continued until the overlap met the target.

[0132] This invention enables dynamic optimization of typhoon avoidance routes for ships, maximizing navigation safety and preventing accidents caused by sudden changes in typhoon paths and wind field expansion, while minimizing navigation costs by reducing unnecessary detour distances and fuel consumption. It perfectly meets the engineering requirements of prioritizing safety and efficiency in maritime shipping, providing personalized and real-time typhoon avoidance route solutions for different types and conditions of ships, thereby improving the safety and operational efficiency of maritime shipping.

[0133] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0134] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for planning typhoon avoidance routes in maritime shipping that integrates multiple typhoon cluster events, characterized in that, The method includes the following steps: S1. Construct a typhoon cluster classification system that includes extreme typhoon cluster events. Divide the events into multi-dimensional scenarios based on the sea area, season, intensity level, and large-scale circulation background to distinguish between regular and extreme typhoon cluster events, and establish a typhoon cluster event scenario classification library. Based on reanalysis data and optimal typhoon path data, use dynamic and thermal diagnostic parameters to extract the formation mechanism of typhoon cluster events in different scenarios, transform it into graded and quantitative diagnostic indicators, form a graded and quantitative diagnostic indicator system for typhoon cluster events, and determine the key control areas corresponding to each scenario. S2, combined with the application requirements of marine shipping engineering, selects indicators that are highly correlated with the changes in the forecast error envelope area from the quantitative diagnostic indicators for typhoon cluster occurrence as the core parameters for shipping in each scenario; through physical mechanism constraints and extreme sample feature enhancement processing, establishes a quantitative correlation between the core parameters for shipping and the uncertainty of typhoon forecast, constructs a typhoon cluster occurrence graded diagnostic model for each scenario, and outputs forecast error correction coefficients to characterize the forecast path deviation, wind field expansion and the comprehensive amplification of the risk superposition of multiple typhoons; S3 establishes a scenario-specific error envelope adjustment standard based on the physical impact mechanism of shipping core parameters on the forecast error envelope area within key control areas; using the forecast error correction coefficient as a quantitative basis, the forecast error envelope area is dynamically scaled, its boundaries expanded, and its direction shifted, and combined with ship navigation parameters to form a ship-adaptive error envelope area; the overlap is verified through real-time data and the correction coefficient is iteratively optimized, and finally, based on the calibrated risk area, typhoon avoidance routes are planned and dynamically adjusted.

2. The maritime shipping route planning method for typhoon avoidance based on multiple typhoon events as described in claim 1, characterized in that, Step S1 further includes: S11: Collect relevant data on historical typhoon cluster events in the research sea area. According to the frequency of occurrence, number of typhoons, and intensity characteristics, typhoon cluster events are divided into regular typhoon cluster events and extreme typhoon cluster events. For extreme typhoon cluster events only, further subdivide the scenarios according to four dimensions: sea area, season, typhoon intensity level, and large-scale circulation background. The extreme typhoon cluster events are divided into several extreme sub-scenarios, including the ENSO decaying year subtropical high anomalous westward extension type, the summer monsoon trough active type, and the multiple typhoon extreme type. Based on typhoon sample data under each scenario, extract key circulation parameters, high-incidence sea area range, and historical forecast deviation characteristics. Statistically analyze the typical intervals of each parameter, clarify the differences in circulation background, spatial distribution, forecast error magnitude, and deviation distribution between regular scenarios and different extreme sub-scenarios, establish the correspondence rules between scenarios and circulation characteristics, spatial distribution, and forecast deviation characteristics, and form a typhoon cluster scenario classification library applicable to the research sea area. S12, based on the typhoon cluster occurrence scenario classification library, corresponding key control areas are determined for different cluster occurrence scenarios. The key control areas are the sea areas where typhoon clusters occur in concentrated phases under the corresponding scenario and the historical forecast path deviation exceeds a set deviation threshold. Based on ERA5 reanalysis data, typhoon optimal path data, Nino3.4 index, and satellite observation data, the Typhoon Generation Potential Index (GPI), Okubo-Weiss vorticity deformation index, convective effective potential energy (CAPE), 200–850 hPa vertical wind shear, 850 hPa low-level absolute vorticity, and subtropical high-pressure westward extension ridge point and area anomaly indicators are used in the key control areas corresponding to each scenario to conduct joint diagnosis of the dynamic and thermal fields of different cluster occurrence scenarios. Among them, for conventional typhoon cluster occurrence scenarios, typical dynamic and thermal characteristics related to the forecast error level are extracted; for extreme typhoon cluster occurrence scenarios, multi-system coupling, extreme dynamic conditions, and enhanced air-sea interaction effects are identified, and the intrinsic relationship between the above characteristics and forecast error amplification and error envelope boundary shift is clarified. S13. Based on the mechanism diagnosis results, parameters that have significant indicative significance for the development of typhoon clusters and the boundary and range of the forecast error envelope area are selected from the diagnostic parameters of different scenarios. These parameters are then standardized and normalized. The parameter threshold ranges are determined in combination with different cluster scenarios. The parameters are classified according to the warning threshold level, scenario adaptability, and engineering application weight, forming a typhoon cluster classification and quantitative diagnostic index system that includes circulation background indicators, dynamic indicators, thermal indicators, and comprehensive potential indicators.

3. The maritime shipping route planning method for typhoon avoidance based on multiple typhoon events as described in claim 2, is characterized in that... In step S13, the circulation background indicators include the longitude of the western extension ridge of the subtropical high, the intensity index, the Nino 3.4 index, and the intensity and location index of the monsoon trough; the dynamic indicators include the 850 hPa lower-level absolute vorticity, the magnitude of the 200–850 hPa vertical wind shear, and the Okubo–Weiss vorticity deformation index; the thermal indicators include the convective effective potential energy (CAPE), the sea surface temperature anomaly (SSTA), and the mid-level relative humidity; and the comprehensive potential indicators include the typhoon formation potential index (GPI), the distance / intensity of multiple typhoon interactions, and the statistical value of historical track forecast deviations.

4. The maritime shipping route planning method for typhoon avoidance based on multiple typhoon events as described in claim 1, characterized in that, Step S2 further includes: S21. In combination with the engineering application requirements of maritime shipping for the stability of typhoon forecasts and the accuracy of risk zone delineation, the correlation and sensitivity analysis of the quantitative diagnostic indicators for typhoon cluster occurrence obtained in step S1 is carried out. Indicators that are correlated with the changes in the boundary and range of the forecast error envelope area under different scenarios and exceed the set threshold are selected and used as the core shipping parameters specific to each scenario. S22, extract the feature values ​​of the corresponding shipping core parameters from the extreme typhoon cluster event samples under each extreme sub-scenario, and construct an extreme sample feature library for each extreme sub-scenario; interpolate and synthesize the extreme samples based on the physical mechanism constraints of typhoon generation and development to expand the sample quantity of each extreme sub-scenario, and at the same time perform unified dimension normalization processing on the shipping core parameters of each extreme sub-scenario to establish a parameter standardization mapping table for each extreme sub-scenario. S23. Based on the core parameters specific to the regular scenarios selected in step S21, the standardized core parameters of each extreme sub-scenario processed in step S22, and the expanded sample data, a multiple linear regression analysis method is used to establish typhoon cluster classification diagnostic models for each scenario, with the optimization objective of maximizing the accuracy of the forecast error envelope correction. The typhoon cluster classification diagnostic model uses the shipping core parameters of the corresponding scenario as input variables, and the typhoon path forecast deviation amplitude, wind field expansion radius, and multi-typhoon interaction effect as intermediate output variables. Finally, the forecast error correction coefficients used to correct the error envelope for the corresponding scenario are fitted. The forecast error correction coefficients are used to characterize the magnitude of the forecast path deviation, the wind field expansion range, and the amplification degree of the risk superposition effect of multiple typhoons. S24. By utilizing the extreme dynamic and thermodynamic characteristics of extreme events in various scenarios, the hierarchical diagnostic model for the corresponding extreme sub-scenario is optimized in reverse. The influence weights of different shipping core parameters on the forecast error are adjusted in the scenario, further improving the model's prediction accuracy of the forecast error envelope trend in the extreme sub-scenario. At the same time, the model for the regular scenario is calibrated with regular parameters to ensure the accuracy of forecast error correction in the regular scenario. S25: Based on the real-time acquired typhoon location, intensity, and environmental field data, determine the current typhoon cluster scenario, call the corresponding scenario's hierarchical diagnostic model, and calculate the forecast error correction coefficient for that scenario.

5. The maritime shipping route planning method for typhoon avoidance based on multiple typhoon events as described in claim 1, characterized in that, Step S3 further includes: S31, based on the different physical mechanisms by which various core shipping parameters affect the forecast error envelope area, it is divided into: circulation indicators that determine the direction of error envelope area deviation, dynamic and thermal indicators that determine the radius expansion of error envelope area, and interactive indicators that determine the superimposed expansion range of multiple typhoon impact areas; based on the above physical mechanisms, a multi-dimensional adjustment logic for the error envelope area within the key control area under different scenarios is established to form an error envelope area expansion and offset standard adapted to each scenario; S32, combining the engineering application requirements of ship safety priority and navigation efficiency, performs scenario-based calibration of the expansion and offset standards for each scenario, so that the correction range matches the degree of anomaly of the core shipping parameters and the level of forecast uncertainty. S33, typhoon warning data and real-time meteorological parameters are updated in real time, the hierarchical diagnostic model in step S2 is called, and the forecast error correction coefficient for the corresponding scenario is dynamically updated; based on the forecast error correction coefficient, the forecast error in the key control area is scaled up as a whole, the wind field boundary is expanded and the path deviation direction is shifted to form the corrected forecast error range. S34: Obtain ship navigation parameters including ship wind resistance level, sailing speed, and load, and combine them with the ship's own wind resistance capability to perform a second calibration on the corrected forecast error range to obtain the ship adaptation error envelope area. S35 collects the actual movement path and wind field range of the typhoon in real time, and compares the overlap between the ship's adaptive error envelope area and the actual impact area of ​​the typhoon. If the overlap does not meet the preset safety and efficiency thresholds, the forecast error correction coefficient is iteratively updated based on the ship's wind resistance capability and current navigation status. The outer range and offset direction of the forecast error range are dynamically adjusted. The above process is repeated until the route planning is completed or the typhoon warning is lifted.

Images