Parameterization analysis method and system based on sea wave spectrum mode

Abstract

The invention relates to the technical field of data analysis, and discloses a parameterization analysis method and system based on a sea wave spectrum mode. The method comprises the following steps: carrying out abnormal value detection and missing data filling on multi-source ocean data by adopting a 3 sigma criterion and Kriging interpolation, and unifying to a 0.1-degree grid to obtain a standardized data set; decomposing a sea wave energy balance equation source item into three modules for numerical calculation; configuring ST2, ST4 and ST6 schemes for the numerical solutions, and carrying out statistical analysis to obtain performance evaluation indexes; based on the evaluation indexes, a genetic algorithm is combined with particle swarm optimization to carry out parameter optimization; dividing the offshore China into four sub-regions for differential adjustment according to the optimization parameters; and carrying out trend checking and parameter increment updating by adopting a sliding window technology based on region configuration. The technical problems that a sea wave spectrum mode parameterization scheme lacks a systematic comparison and optimization mechanism, parameter configuration cannot adapt to regional marine environment differences, and parameters lacks real-time monitoring and dynamic adjustment capabilities are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data analysis, and in particular to a parameterization analysis method and system based on a sea wave spectrum mode. BACKGROUND

[0002] The existing sea wave numerical prediction technology is mainly based on the third generation sea wave spectrum mode, and the generation, propagation and dissipation process of the wave is described by using the sea wave energy balance equation, wherein the source term includes three components of wind energy input term, nonlinear interaction term and energy dissipation term. The traditional sea wave spectrum mode realizes the numerical simulation of the sea wave field through the WAM, SWAN and WaveWatch III models, and these models are widely used in global and regional scale sea wave prediction, providing important technical support for marine engineering, navigation safety and marine resource development. The parameterization schemes in the prior art mainly include different versions such as ST2, ST4 and ST6, and each scheme adopts different mathematical descriptions and parameter settings for physical processes such as wind wave generation, wave-wave interaction and wave breaking.

[0003] However, the existing technology has the problem of significant differences in the applicability of the parameterization scheme, and the prediction accuracy of different schemes under specific sea areas and meteorological conditions differs greatly, lacking a systematic scheme selection and optimization mechanism. The traditional parameter setting often uses empirical values or global unified configuration, which fails to fully consider the differences of regional marine environment, resulting in insufficient adaptability of the parameterization scheme in complex sea areas such as China's coastal waters. In addition, the existing technology lacks the ability to dynamically adjust parameters, and cannot automatically optimize parameter configuration according to the change of prediction performance. When the marine environmental conditions change, it is difficult to maintain stable prediction accuracy with fixed parameter settings.

[0004] Based on the above analysis, the core technical problems faced by the existing technology include: how to establish a standardized processing mechanism for multi-source marine data, how to build a modular algorithm framework for the sea wave spectrum mode, how to realize intelligent selection and performance evaluation of the parameterization scheme, how to design an adaptive parameter optimization algorithm, how to establish a regional parameter configuration strategy, and how to build a real-time monitoring and dynamic updating mechanism for parameters. The solution of these technical problems needs systematic technical innovation from multiple dimensions such as data processing, algorithm optimization, scheme selection, parameter adjustment and real-time monitoring, to form a complete sea wave spectrum mode parameterization analysis technology system. SUMMARY

[0005] The present application provides a parameterization analysis method and system based on a sea wave spectrum mode, which solves the technical problems of lack of systematic selection and optimization mechanism for sea wave spectrum mode parameterization scheme, inability of parameter configuration to adapt to regional marine environment differences, and lack of real-time monitoring and dynamic adjustment capability for parameters.

[0006] Firstly, this application provides a parametric analysis method based on wave spectrum models. This method includes: performing outlier detection and missing data imputation on multi-source ocean data using the 3σ criterion and Kriging interpolation algorithm; unifying the processed data to a 0.1° × 0.1° spatial grid to obtain a standardized ocean environment dataset; decomposing the source terms of the wave energy balance equation into three modules—wind energy input, nonlinear interaction, and energy dissipation—based on the standardized ocean environment dataset for numerical calculation to obtain numerical solutions for the wave spectrum modules; and configuring the numerical solutions of the wave spectrum modules as ST2, ST4, and ST5, respectively. The T6 parameterization scheme is subjected to correlation coefficient and root mean square error statistical analysis to obtain the performance evaluation index of the parameterization scheme. Based on the performance evaluation index of the parameterization scheme, a genetic algorithm combined with particle swarm optimization is used to constrain and optimize the growth rate coefficient of the wind input term to obtain the global optimal parameter combination. According to the global optimal parameter combination, the coastal waters of China are divided into four sub-regions and the bottom friction coefficient and ocean current influence coefficient are adjusted differently to obtain the regional differentiated parameter configuration. Based on the regional differentiated parameter configuration, the sliding window technique is used to perform Mann-Kendall trend test and parameter incremental update on the 72-hour forecast data to obtain the dynamic parameter update mechanism.

[0007] Secondly, this application provides a parameterized analysis system based on wave spectrum models, the parameterized analysis system based on wave spectrum models comprising:

[0008] The imputation module is used to perform outlier detection and missing data imputation on multi-source ocean data using the 3σ criterion and Kriging interpolation algorithm. The processed data is then unified to a 0.1 degree × 0.1 degree spatial grid to obtain a standardized marine environment dataset.

[0009] The decomposition module is used to decompose the source term of the wave energy balance equation into three modules—wind energy input term, nonlinear interaction term, and energy dissipation term—based on the standardized marine environment dataset, and then perform numerical calculations to obtain the numerical solution of the wave spectrum module.

[0010] The analysis module is used to perform correlation coefficient and root mean square error statistical analysis on the numerical solutions of the wave spectrum module configured with ST2, ST4 and ST6 parameterization schemes respectively, and obtain the performance evaluation index of the parameterization scheme;

[0011] The optimization module is used to perform constrained optimization of the wind input growth rate coefficient based on the performance evaluation index of the parameterized scheme using a genetic algorithm combined with particle swarm optimization, so as to obtain a global optimal parameter combination.

[0012] The adjustment module is used to divide the coastal waters of China into four sub-regions based on the global optimization parameter combination and to make differentiated adjustments to the bottom friction coefficient and ocean current influence coefficient to obtain regional differentiated parameter configurations.

[0013] The update module is used to perform Mann-Kendall trend tests and incremental parameter updates on 72-hour forecast data based on the regionally differentiated parameter configuration using the sliding window technique, thereby obtaining a dynamic parameter update mechanism.

[0014] Thirdly, a parametric analysis device based on wave spectrum patterns is provided, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor invokes the instructions in the memory to cause the parametric analysis device based on wave spectrum patterns to execute the aforementioned parametric analysis method based on wave spectrum patterns.

[0015] Fourthly, a computer-readable storage medium is provided, wherein instructions are stored therein, which, when executed on a computer, cause the computer to perform the above-described parametric analysis method based on wave spectrum patterns.

[0016] In the technical solution provided in this application, the present invention performs outlier detection and missing data imputation on multi-source ocean data using the 3σ criterion and Kriging interpolation algorithm. The processed data is then unified into a 0.1 degree × 0.1 degree spatial grid to obtain a standardized ocean environment dataset. This solves the problems of inconsistent data quality and irregular spatial distribution in traditional ocean data processing. The statistical principle of the 3σ criterion can effectively identify abnormal observations caused by equipment failure and extreme weather. The Kriging interpolation algorithm accurately estimates missing data based on spatial correlation theory. The standardized grid processing eliminates the spatiotemporal deviation between different data sources, providing a high-quality input data foundation for subsequent wave spectrum model calculations. Based on a standardized marine environmental dataset, the source term of the wave energy balance equation is decomposed into three modules: wind energy input, nonlinear interaction, and energy dissipation. Numerical calculations are then performed to obtain the numerical solution for the wave spectrum module. This modular design allows each physical process to employ the most suitable numerical algorithm and parameterization scheme. The wind energy input uses Miles theory to accurately describe the wave generation mechanism; the nonlinear interaction uses an improved discrete mutual approximation algorithm to efficiently handle wave-to-wave energy exchange; and the energy dissipation uses a probabilistic-deterministic hybrid model to accurately characterize the breaking process. This modular architecture not only improves computational accuracy but also enhances the scalability and maintainability of the scheme. The numerical solution of the wave spectrum module is configured with ST2, ST4, and ST6 parameterization schemes, and correlation coefficients and root mean square errors are statistically analyzed to obtain performance evaluation indicators for the parameterization schemes. This systematic scheme comparison overcomes the subjectivity and uncertainty of traditional techniques that rely on experience to select parameterization schemes. Quantitative statistical indicators objectively evaluate the applicability of different schemes in specific sea areas, providing a scientific basis for parameter optimization. Based on the performance evaluation index of the parameterized scheme, a genetic algorithm combined with particle swarm optimization is used to constrain and optimize 15 parameters, including the wind input growth rate coefficient, to obtain a globally optimized parameter combination. The application of artificial intelligence algorithms is the core technical contribution of this invention. The combination of the global search capability of the genetic algorithm and the local development capability of the particle swarm optimization algorithm enables efficient optimization in high-dimensional parameter space. The algorithm's contribution to the scheme is reflected in its ability to automatically discover the optimal parameter configuration without extensive manual experimentation. The setting of constraints ensures the physical rationality of the optimization results. The hybrid optimization strategy significantly improves the efficiency and accuracy of parameter tuning. Based on the globally optimized parameter combination, the coastal waters of China are divided into four sub-regions: the Bohai Sea, the Yellow Sea, the East China Sea, and the South China Sea. Differentiated adjustments to the bottom friction coefficient and ocean current influence coefficient are made to obtain regionally differentiated parameter configurations. The regionalization strategy fully considers the differences in water depth, topography, meteorology, and ocean dynamic environment in different sea areas. Targeted parameter adjustments significantly improve the forecast accuracy of the model in complex sea areas and solve the technical problem that traditional uniform parameter configurations cannot adapt to regional environmental differences.Based on regionally differentiated parameter configuration, a sliding window technique is used to perform Mann-Kendall trend tests and incremental parameter updates on 72-hour forecast data to obtain a dynamic parameter update mechanism. Real-time monitoring and adaptive adjustment capabilities enable the wave spectrum model to automatically optimize parameter settings as the marine environment changes. The sliding window technique ensures the continuity and timeliness of performance evaluation, and the application of statistical testing methods guarantees the scientific and reliable nature of parameter adjustment decisions. The dynamic update mechanism overcomes the limitations of traditional fixed parameter configuration and realizes the intelligent operation of the wave spectrum model. The overall scheme forms a complete technical chain from data preprocessing, modular calculation, intelligent optimization, regional configuration to dynamic adjustment, providing an important theoretical foundation and practical tool for the development of numerical wave forecasting technology. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of an embodiment of the parameterized analysis method based on wave spectrum patterns in this application.

[0019] Figure 2 This is a schematic diagram of an embodiment of the parameterized analysis system based on wave spectrum patterns in this application.

[0020] Figure 3 This is a schematic block diagram of the parametric analysis device based on wave spectrum mode in an embodiment of the present invention. Detailed Implementation

[0021] This application provides a parametric analysis method and system based on wave spectrum patterns. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0022] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the parameterized analysis method based on wave spectrum models in this application includes:

[0023] Step S101: The 3-Sigma Rule and Kriging interpolation algorithm are used to detect outliers and fill in missing data in the multi-source marine data. The processed data are then unified into a 0.1-degree × 0.1-degree spatial grid to obtain a standardized marine environment dataset.

[0024] Step S102: Based on the standardized marine environment dataset, the source term of the wave energy balance equation is decomposed into three modules: wind energy input term, nonlinear interaction term, and energy dissipation term, and numerical calculation is performed to obtain the numerical solution of the wave spectrum module.

[0025] Step S103: Configure the numerical solution of the wave spectrum module with parameterization schemes ST2, ST4 and ST6 respectively, and perform statistical analysis on correlation coefficient and root mean square error to obtain the performance evaluation index of the parameterization scheme;

[0026] Step S104: Based on the performance evaluation index of the parameterized scheme, a genetic algorithm combined with particle swarm optimization is used to constrain and optimize the growth rate coefficient of the wind input term to obtain the global optimal parameter combination;

[0027] Step S105: Divide the Chinese coastal waters into four sub-regions according to the global optimization parameter combination and make differentiated adjustments to the bottom friction coefficient and ocean current influence coefficient to obtain the regional differentiated parameter configuration;

[0028] Step S106: Based on the regionally differentiated parameter configuration, the sliding window technique is used to perform Mann-Kendall trend testing and incremental parameter updates on the 72-hour forecast data to obtain a dynamic parameter update mechanism.

[0029] It is understood that the executing entity of this application can be a parameterized analysis system based on wave spectrum models, or it can be a terminal or a server; the specific implementation is not limited here. This application's embodiments use a server as an example for illustration.

[0030] Specifically, the multi-source ocean data acquisition encompasses four data sources: satellite remote sensing data, ocean buoy observation data, shore-based radar observation data, and numerical weather prediction data. A standardized interface is used for continuous 72-hour data acquisition, obtaining raw data including sea surface wind speed, wind direction, significant wave height, spectral peak period, and mean wave direction. The 3σ criterion outlier detection identifies and removes data points exceeding three times the standard deviation by calculating the mean and standard deviation of the data sequence. The Kriging interpolation algorithm establishes a spatial covariance function model and uses the least squares method to solve for the interpolation weight coefficients to perform spatial interpolation calculations on missing data points. Finally, a bilinear interpolation method is used to resample the data to a 0.1° × 0.1° regular grid, forming a standardized ocean environmental dataset.

[0031] Based on a standardized marine environmental dataset, an energy balance equation for ocean waves was established, decomposing the source term into three independent modules according to physical mechanisms. For the wind energy input term, a growth rate calculation model was established using Miles theory. The wind stress transfer coefficient was corrected by calculating atmospheric stability parameters using the Richardson number, and the nonlinear characteristics of the sea surface drag coefficient varying with wave age were incorporated into the calculation process. For the nonlinear interaction term, an improved discrete mutual approximation algorithm was used. The interaction wavenumber combination was identified through four-wave resonance conditions, and high-precision numerical integration of the energy transfer rate was performed using adaptive mesh refinement technology. For the energy dissipation term, a probabilistic-deterministic hybrid model was adopted. Wave breaking conditions were determined by combining statistical characteristics of sea surface slope and local wave steepness parameters. When wave steepness exceeded a critical value, breaking dissipation calculations were performed. The numerical solutions of the three modules were merged to form the numerical solution of the wave spectrum module.

[0032] Numerical solutions for the wave spectrum module were configured according to three parameterization schemes: ST2, ST4, and ST6. Continuous numerical calculations were performed over 30 days with a time step of 300 seconds in the coastal areas of China. The ST2 scheme employed Tolman-Chalikov wind input parameterization and an exponential dissipation model; the ST4 scheme used a saturated breaking dissipation model based on Ardhuin's physical process description; and the ST6 scheme integrated the latest physical understanding, employing observation-supported wind input functions and improved dissipation parameterization. By comparing effective wave height, spectral peak frequency, and wave direction parameters with ocean buoy observation data, the linear correlation between each scheme and the observation data was calculated using the Pearson correlation coefficient method. Furthermore, the root mean square error, mean absolute error, and bias index were calculated to establish a comprehensive scoring system and generate performance evaluation indicators for the parameterization schemes.

[0033] Based on the performance evaluation index of the parameterized scheme, 15 key optimization parameters were determined, including the wind input growth rate coefficient, sea surface roughness parameter, nonlinear interaction coupling strength parameter, and dissipation term breakage threshold. A multi-objective weighted objective function was established. A genetic algorithm was used to initialize the population to 100 individuals, randomly encoding the parameter genes of each individual. Tournament selection, single-point crossover, and Gaussian mutation were employed to perform population evolution calculations. A particle swarm optimization algorithm was used for local fine-grained search, setting cognitive and social coefficients in particle velocity updates. The optimal solution in the parameter space was found through iterative updates of particle position and velocity. Parameter constraints ensured that the growth rate coefficient, sea surface roughness parameter, and breakage threshold were within physically reasonable ranges. Iterative calculations terminated when the improvement of the objective function was less than a set threshold for 50 consecutive generations, generating a globally optimized parameter combination.

[0034] Based on the global optimization parameter combination, China's coastal waters are divided into four sub-regions: the Bohai Sea, the Yellow Sea, the East China Sea, and the South China Sea, according to differences in water depth distribution, topographic features, meteorological conditions, and marine dynamic environment. In the Bohai Sea, a semi-enclosed shallow sea area, deep-water dispersion correlation correction coefficients and bottom friction coefficients are used to adjust shallow-water effect parameters. In the Yellow Sea, ocean current influence coefficients and tidal harmonic analysis are introduced to correct for ocean current-wave interaction. In the East China Sea, the growth rate coefficient is adjusted and high-frequency dissipation parameters are optimized within the typhoon influence radius. In the South China Sea, standard deep-water parameter configurations are used to optimize far-field swell propagation attenuation parameters. The effectiveness of the parameter configurations for each sub-region is verified using 12 consecutive months of fixed-point buoy data and satellite altimeter data. A hierarchical real-time monitoring system is constructed to continuously collect and process hourly data on effective wave height, spectral peak period, average wave direction parameters from buoy observations, and satellite altimeter along-orbit wave height. A sliding window technique is used to continuously evaluate the forecast results for the most recent 72 hours, calculating the time series mean, standard deviation, trend coefficient, and periodic characteristic statistics of the forecast error. The Mann-Kendall trend test is used to detect the monotonic trend of the forecast error time series, the Kolmogorov-Smirnov distribution consistency test is used to verify the stability of the forecast error distribution, and the runs test is used to detect the randomness of the forecast error series. When a systematic bias or significant trend change is detected, a parameter adjustment warning is triggered. A support vector machine classifier identifies the categories of parameters that need adjustment, a random forest regressor predicts the magnitude of parameter adjustments, and an incremental update strategy is used to limit the magnitude of each adjustment and set continuous adjustment intervals, forming a dynamic parameter update mechanism.

[0035] In one specific embodiment, the process of performing step S101 may specifically include the following steps:

[0036] Satellite remote sensing data, ocean buoy observation data, shore-based radar observation data, and numerical weather prediction data were collected continuously for 72 hours using a standardized interface to obtain multi-source ocean data including sea surface wind speed, wind direction, significant wave height, spectral peak period, and average wave direction.

[0037] The mean and standard deviation of the data sequence are calculated using the 3σ criterion combined with the sliding window statistical method for the multi-source ocean data. Data points exceeding 3 times the standard deviation are marked as outliers and removed to obtain quality-controlled ocean observation data.

[0038] Based on the quality-controlled marine observation data, a spatial covariance function model is established using the Kriging interpolation algorithm. The interpolation weight coefficients are solved by the least squares method to perform spatial interpolation calculations on the missing data points, thereby obtaining spatially continuous marine parameter data.

[0039] The spatially continuous ocean parameter data is spatiotemporally synchronized according to a unified time coordinate system and spatial coordinate system, and then resampled to a 0.1 degree × 0.1 degree spatial grid using a bilinear interpolation method to obtain a standardized ocean environment dataset.

[0040] Specifically, the standardized interface data acquisition process obtains marine environmental information from four types of data sources through preset communication protocols and data format specifications. Satellite remote sensing data includes sea surface temperature and sea surface wind field information; ocean buoy observation data provides measured wave height and period values ​​at fixed locations; shore-based radar observation data covers wave propagation information in nearshore waters; and numerical weather prediction data supplements atmospheric driving field parameters. A continuous 72-hour time window acquisition means tracing back 72 hours of historical data from the current moment to ensure sufficient initial and boundary conditions for wave spectrum model calculations. Sea surface wind speed is measured using an anemometer to obtain the vector wind field; wind direction is recorded using a wind vane; significant wave height represents the statistical characteristic value of wave energy; spectral peak period reflects the frequency characteristics of the main energy concentration of the waves; and average wave direction describes the dominant direction angle of wave propagation. The core of the 3σ criterion outlier detection lies in the statistical principle that data points in a normal distribution that are more than three standard deviations from the mean have an extremely low probability of occurrence. The sliding window statistical method uses a fixed-length time window to move across the data sequence, calculating local statistical characteristics point by point. In the specific data processing process, a time series is first established for each observation parameter. Then, the mean and standard deviation of the series are calculated within the preset time window. When the absolute value of the difference between a certain observation value and the mean exceeds 3 times the standard deviation, the data point is marked as an outlier and removed from the original dataset. The sliding window continues to move forward until the entire data series is processed. The quality-controlled marine observation data has eliminated outlier observation values ​​caused by equipment failure, communication interference, and extreme weather conditions.

[0041] The Kriging interpolation algorithm, based on geostatistical principles, assumes a correlation between spatially adjacent observations. A spatial covariance function model describes how this spatial correlation varies with distance. The algorithm first calculates the distances and corresponding covariance values ​​between all pairs of observation points, establishing a semi-variogram to quantify spatial variability. Then, it determines the parameters of the covariance function by fitting a theoretical model. During the least squares method to solve for the interpolation weight coefficients, a system of linear equations is established. The coefficient matrix of this system consists of the covariance values ​​between observation points, and the right-hand vector consists of the covariance values ​​between the point to be interpolated and each observation point. The obtained weight coefficients reflect the contribution of each observation point to the interpolation result. The interpolation value for missing data points is equal to the weighted sum of all observation values ​​and their corresponding weight coefficients. Spatially continuous ocean parameter data fills the gaps in the original observation grid.

[0042] Spatiotemporal synchronization processing unifies data from different data sources, time bases, and spatial coordinate systems into a standard reference frame. The unified time coordinate system uses Coordinated Universal Time (UTC) as the base, converting timestamps from various data sources into a unified format. The unified spatial coordinate system uses the WGS84 geographic coordinate system, converting the location information from each data source into latitude and longitude representations. During the bilinear interpolation resampling process, for each node in the target grid, the smallest rectangular region containing that node in the original data is found. The values ​​of the target node are calculated using the values ​​of the four vertices of the rectangle through a bilinear interpolation formula. The interpolation calculation considers linear variations in both longitude and latitude. A 0.1° × 0.1° spatial grid resolution corresponds to approximately 11 km of spatial resolution in both longitude and latitude directions. The standardized marine environmental dataset forms a regular grid structure, with each grid node containing values ​​for five parameters: sea surface wind speed, wind direction, significant wave height, spectral peak period, and mean wave direction.

[0043] In one specific embodiment, the process of performing step S102 may specifically include the following steps:

[0044] Based on spatial grid data, an ocean wave energy balance equation is established. The source term is decomposed into three independent modules according to the physical mechanism: wind energy input term, nonlinear interaction term, and energy dissipation term, resulting in a modular ocean wave spectrum calculation system.

[0045] A growth rate calculation model is established for the wind energy input term using Miles theory. The wind stress transmission coefficient is corrected by calculating the atmospheric stability parameter using the Richardson number. The nonlinear characteristics of the sea surface drag coefficient as a function of wave age are introduced into the calculation process to obtain the numerical solution of the wind energy input term.

[0046] An improved discrete mutual approximation algorithm is used to apply the nonlinear interaction term. The interaction wavenumber combination is identified by the four-wave resonance condition. The energy transfer rate is calculated with high precision by adaptive mesh refinement technology to obtain the numerical solution of the nonlinear interaction term.

[0047] A probabilistic-deterministic hybrid model is adopted for the energy dissipation term. The wave breaking conditions are judged by combining the statistical characteristics of sea surface slope and local wave steepness parameters. When the wave steepness exceeds the critical value, the breaking dissipation calculation is performed to obtain the numerical solution of the energy dissipation term. The numerical solutions of the wind energy input term, the nonlinear interaction term, and the energy dissipation term are combined into the numerical solution of the wave spectrum module.

[0048] Specifically, the wave energy balance equation is established based on wind speed, wave height, and frequency spectrum information from spatial grid data. It abstracts the physical process of wave evolution into mathematical expressions, where source terms represent various physical mechanisms influencing wave energy changes. The wind energy input term describes the process of energy transfer from the atmospheric wind field to the waves, the nonlinear interaction term characterizes the energy exchange mechanism between waves of different frequencies, and the energy dissipation term reflects the energy loss caused by wave breaking, bottom friction, etc. The decomposition of these three independent modules is based on their respective physical mechanisms and spatiotemporal scale characteristics. The modular wave spectrum calculation system allows each source term to employ a dedicated numerical algorithm and parameterization scheme. The growth rate calculation model established by Miles theory describes the resonance mechanism of wind-wave generation. Its core lies in the pressure difference generated by the phase relationship between sea surface fluctuations and the airflow above, which drives wave growth. In the process of calculating atmospheric stability parameters using the Richardson number, wind speed and temperature gradient information from the spatial grid data is used to calculate the ratio of buoyancy frequency to wind shear frequency. When the Richardson number is less than a critical value, the atmosphere is unstable, and enhanced wind turbulence leads to more energy transfer to the waves; conversely, when the atmosphere is stable, energy transfer efficiency decreases. The nonlinear characteristics of the sea surface drag coefficient with wave age reflect the relationship between sea surface roughness and wave development stage. The steeper waves of young sea surfaces increase the drag coefficient, while the gentler waves of mature sea surfaces decrease the drag coefficient. The correction process of the wind stress transfer coefficient takes into account atmospheric stability and wave age effect, and obtains the corrected numerical solution of wind energy input through iterative calculation.

[0049] When improving the discrete mutual approximation algorithm to handle nonlinear interaction terms, the first step is to select wave combinations that satisfy wave number conservation and frequency conservation based on the four-wave resonance condition. Wave number conservation requires that the wave number vectors of the four interacting waves satisfy the closure condition, and frequency conservation requires that the angular frequencies of the four waves satisfy the additive relationship. An adaptive mesh refinement technique dynamically adjusts the computational mesh based on the magnitude of the spectral energy density gradient. Fine meshes are used in regions with large energy gradients to ensure accuracy, while coarse meshes are used in regions with small energy gradients to save computation. During high-precision numerical integration calculations, the Gaussian quadrature formula is used to discretize complex integral expressions. The division of the integration region considers the geometric constraints of the resonance condition and the local characteristics of the spectral shape. The calculated energy transfer rate reflects the direction and intensity of energy flow between different frequency components.

[0050] A probabilistic-deterministic hybrid model combines statistical characteristics of sea surface slope and local wave steepness parameters to determine wave breaking conditions. The statistical characteristics of sea surface slope are obtained by analyzing the wave height differences between adjacent grid points in spatial grid data, while the local wave steepness parameter is calculated from the ratio of wave height to wavelength. The probabilistic nature of wave breaking is reflected in the different breaking probabilities at different locations on the sea surface, while the deterministic nature is reflected in the fact that breaking will inevitably occur when the wave steepness exceeds a physical critical value. The critical value is set based on wave stability theory and observational statistics. The breaking dissipation calculation uses a dissipation rate formula related to wave steepness; the dissipation intensity increases nonlinearly with wave steepness, and the energy dissipation term numerically quantifies the wave energy lost during the breaking process.

[0051] In one specific embodiment, the process of executing step S103 may specifically include the following steps:

[0052] The numerical solution of the wave spectrum module was configured with grid resolution for the coastal area of ​​China according to three parameterization schemes: ST2, ST4 and ST6. 30 days of continuous numerical calculation with a time step of 300 seconds was performed in the range of 105°E (longitude)-130°E and 10°N (latitude)-45°N to obtain the calculation results of the three parameterization schemes.

[0053] The calculation results of the three parameterization schemes were compared and analyzed with the ocean buoy observation data in terms of effective wave height, spectral peak frequency and wave direction parameters. The Pearson correlation coefficient calculation method was used to calculate the linear correlation between each scheme and the observation data, and the correlation coefficient matrix of each parameterization scheme was obtained.

[0054] Based on the correlation coefficient matrix of each parameterized scheme, the root mean square error, mean absolute error, and deviation index of the forecast and observed values ​​are calculated. The forecast accuracy of each scheme is quantitatively evaluated and analyzed by statistical testing methods to obtain multi-dimensional statistical evaluation results.

[0055] Based on the multi-dimensional statistical evaluation results, a comprehensive scoring system is established. The weights of the effective wave height prediction accuracy, the spectral peak frequency prediction deviation, and the wave direction prediction error are set to 0.3 and 0.3 respectively. The performance evaluation index of the parameterized scheme is obtained by weighted calculation.

[0056] Specifically, the grid resolution configuration process for the three parameterization schemes inputs the numerical solutions of the wave spectrum module into the computational frameworks of schemes ST2, ST4, and ST6, respectively. Scheme ST2 uses Tolman-Chalikov wind input parameterization combined with an exponential dissipation model; scheme ST4 uses a saturated breaking dissipation model based on the Ardhuin physical process description; and scheme ST6 integrates observation-supported wind input functions and improved dissipation parameterization. The latitude and longitude range of China's coastal areas covers the main sea areas of the Bohai Sea, Yellow Sea, East China Sea, and South China Sea. The grid resolution configuration divides the computational domain into regular grid cells according to spatial discretization requirements. The 300-second time step ensures that the numerical stability conditions are met. The 30-day continuous numerical calculation covers the complete weather process and wave evolution cycle. The calculation results of the three parameterization schemes include the wave height, period, and wave direction parameters of each grid point at each time step. The comparative analysis process first extracts observation records corresponding to grid points within the computational domain from the ocean buoy observation database. Significant wave height represents the overall level of wave energy, spectral peak frequency reflects the frequency characteristics of the main energy concentration of the waves, and wave direction parameters describe the dominant direction of wave propagation. The Pearson correlation coefficient calculation method quantifies the strength of the linear relationship between forecast and observed values. The calculation process first calculates the mean of the forecast and observation sequences, then calculates the deviation of each data point from the mean. The correlation coefficient is equal to the mean of the product of the deviations of the two sequences divided by the product of their respective standard deviations. The value ranges from -1 to +1, with values ​​close to 1 indicating a strong positive correlation and close to 0 indicating no linear relationship. The correlation coefficient matrix for each parameterized scheme records the correlation performance of each scheme for different parameters.

[0057] The root mean square error (RMSE) calculation process involves squaring the difference between the forecast and observed values, then calculating the mean of all squared differences. Finally, the square root of this mean is taken to obtain a standard measure of error. This indicator is more sensitive to large errors and can highlight differences in forecast performance. The mean absolute error (MAO) is calculated by taking the mean of the absolute values ​​of the differences between the forecast and observed values. This indicator is less affected by extreme errors and is more robust. The deviation indicator calculates the mean of the differences between the forecast and observed values; a positive deviation indicates that the forecast is systematically overestimated, while a negative deviation indicates that the forecast is systematically underestimated. Statistical testing methods include the t-test to verify the significance of the deviation, the F-test to compare the variance differences between different schemes, and the chi-square test to assess the normality of the error distribution. The multi-dimensional statistical evaluation results comprehensively reflect the differences in forecast performance of various parameterized schemes under different statistical indicators.

[0058] The comprehensive scoring system was established by determining the importance weights of each parameter based on the application requirements of wave forecasting. The weight for significant wave height forecast accuracy was set to 0.4, reflecting the crucial role of wave height forecasting in marine engineering and maritime safety. The weight for peak frequency forecast deviation was set to 0.3, reflecting the importance of frequency forecasting in describing wave spectrum shape. The weight for wave direction forecast error was set to 0.3, reflecting the value of wave direction information in determining wave propagation direction. The weighted calculation process involved multiplying the standardized scores of each parameter by their corresponding weights and then summing them. Standardization eliminated the influence of differences in the dimensions of different parameters. The final performance evaluation index for the parameterized scheme was a dimensionless comprehensive score; a higher value indicated better scheme performance.

[0059] In one specific embodiment, the process of executing step S104 may specifically include the following steps:

[0060] Based on the performance evaluation index of the parameterized scheme, the wind input growth rate coefficient, sea surface roughness parameter, nonlinear interaction coupling strength parameter, and dissipation term breakage threshold optimization parameter are determined. A multi-objective weighted objective function is established to comprehensively evaluate the effective wave height root mean square error, spectral peak period deviation, wave direction correlation coefficient, and extreme event skill score, and the parameter optimization objective function is obtained.

[0061] The parameter optimization objective function is input into the genetic algorithm to initialize the population. The population size is set to 100 individuals. The parameter genes of each individual are randomly encoded. The population evolution calculation is performed using tournament selection, single-point crossover and Gaussian mutation operations to obtain the population evolution result of the genetic algorithm.

[0062] The population evolution results of the genetic algorithm are subjected to a local fine search using a particle swarm optimization algorithm. The cognitive coefficient and social coefficient in the particle velocity update are both set to 2.0. The optimal solution in the parameter space is found by iteratively updating the particle position and velocity, and the search results of the hybrid optimization algorithm are obtained.

[0063] Based on the search results of the hybrid optimization algorithm, parameter constraints are set, the growth rate coefficient is constrained to a reasonable range, the sea surface roughness parameter is constrained to a physically feasible range, and the breakage threshold is constrained to an empirical range. The iterative calculation is terminated when the improvement of the objective function is less than the set threshold for 50 consecutive generations, and the global optimization parameter combination is obtained.

[0064] Specifically, the performance evaluation indicators of the parameterized scheme provide a quantitative basis for determining key optimization parameters. The wind input growth rate coefficient controls the energy transfer efficiency during wave generation. The sea surface roughness parameter reflects the influence of sea surface conditions on wind stress transmission. The nonlinear interaction coupling strength parameter determines the degree of energy exchange between waves of different frequencies. The dissipation term breaking threshold parameter controls the critical conditions for wave breaking. In the process of establishing the multi-objective weighted objective function, the root mean square error of effective wave height measures the overall accuracy of wave height forecast. The weight setting reflects the importance of wave height forecast in marine engineering applications. The spectral peak period deviation assesses the degree of systematic deviation in frequency forecast. The wave direction correlation coefficient quantifies the consistency between directional forecast and observation. The extreme event skill score specifically evaluates the model's ability to forecast extreme sea states such as typhoons. The four evaluation indicators are weighted and summed to form a comprehensive objective function. The magnitude of the parameter optimization objective function directly reflects the quality of the parameter combination.

[0065] The genetic algorithm initializes the population by encoding each optimization parameter into a binary gene string. A population size of 100 individuals ensures a balance between genetic diversity and computational efficiency. Each individual represents a complete set of parameters, and the random encoding process generates initial individuals uniformly within the parameter value range. The tournament selection operation randomly selects several individuals from the population for fitness comparison; individuals with high fitness are selected as parents to participate in reproduction. The selection pressure is controlled by the tournament size. The single-point crossover operation randomly selects a crossover point between the gene strings of two parent individuals, exchanging the gene segments after the crossover point to generate offspring individuals. The crossover probability controls the frequency of the crossover operation. The Gaussian mutation operation randomly perturbs the individual genes, with the mutation intensity following a Gaussian distribution. The mutation probability determines the likelihood of gene mutation. Population evolution calculations improve the average fitness level of the population generation by iteratively executing selection, crossover, and mutation operations. The genetic algorithm's population evolution results include the best individuals of each generation and population statistics.

[0066] The Particle Swarm Optimization (PSO) algorithm uses the optimal solution from the Genetic Algorithm as the initial particle positions. Each particle represents a candidate solution in the parameter space. Particle velocity represents the direction and magnitude of parameter changes. The cognitive coefficient controls the degree to which a particle moves closer to its historical best position, while the social coefficient controls the degree to which a particle moves closer to the global best position. Both coefficients are set to 2.0 to ensure a balance between individual and group experience. During particle position updates, the new position equals the current position plus the velocity vector. Velocity updates consider the combined effects of inertia, cognition, and social terms. The inertia weight controls the tendency of particles to maintain their original direction of motion, the cognitive term drives particles to move towards their individual best solution, and the social term drives particles to move towards the global best solution. During iterative updates, the historical best position and the global best position of each particle are continuously recorded. The local fine-grained search capability allows the algorithm to find a more accurate optimal solution within a limited area. The hybrid optimization algorithm's search results combine the global exploration capability of the Genetic Algorithm and the local development capability of the PSO algorithm.

[0067] The parameter constraints ensure the physical rationality of the optimization results. The reasonable range of the growth rate coefficient is based on the applicability of Miles' theory and observational verification results. The physical feasible range of the sea surface roughness parameter is determined by the actual variation range of the sea surface state. The empirical range of the breakage threshold comes from wave stability theory and a large amount of observational statistics. The constraints are handled using a penalty function method. When the parameter value exceeds the constraint range, a penalty term is added to the objective function to guide the search process to converge to the feasible region. The convergence criterion is set as the improvement of the objective function being less than a set threshold for 50 consecutive generations. This criterion ensures that the algorithm terminates in a timely manner after finding a stable optimal solution, avoiding unnecessary consumption of computational resources. The global optimization parameter combination represents the parameter settings that enable the objective function to reach its optimal value under given constraints.

[0068] In one specific embodiment, the process of executing step S105 may specifically include the following steps:

[0069] Based on the global optimization parameter combination, China's coastal waters are divided into four sub-regions: the Bohai Sea, the Yellow Sea, the East China Sea, and the South China Sea, according to differences in water depth distribution, topographic features, meteorological conditions, and marine dynamic environment, thus obtaining a marine zoning scheme.

[0070] The shallow water effect parameters of the Bohai Sea region were adjusted by using the deep-water dispersion relation correction coefficient and the bottom friction coefficient. The ocean current influence coefficient and tidal harmonic analysis were introduced into the Yellow Sea region to correct the ocean current-wave interaction, and the parameter adjustment results of the shallow sea region were obtained.

[0071] The growth rate coefficient was adjusted and the high-frequency dissipation parameters were optimized for the East China Sea region within the typhoon's influence radius. For the South China Sea region, the far-field surge propagation attenuation parameters were optimized using standard deep-water parameter configuration, resulting in the deep-sea region parameter adjustment results.

[0072] Based on the parameter adjustment results of the shallow sea area and the parameter adjustment results of the deep sea area, an independent verification dataset for each sub-region is established. The effectiveness of the parameter configuration of each sub-region is verified by 12 consecutive months of fixed-point buoy data and satellite altimeter data, and regional differentiated parameter configurations are obtained.

[0073] Specifically, the marine zoning scheme was established based on the analysis results of a global optimization parameter combination. China's coastal waters were stratified according to water depth distribution data. Areas with a depth of less than 50 meters were classified as shallow water, areas with a depth of 50 to 200 meters as medium-depth water, and areas with a depth greater than 200 meters as deep water. Topographic feature analysis was used to divide the regions based on factors such as seafloor slope, coastline shape, and island distribution. Meteorological condition difference analysis covered climatic factors such as monsoon intensity, typhoon path frequency, and temperature gradient. Marine dynamic environment difference analysis included oceanographic factors such as ocean current systems, tidal characteristics, and water mass structure. The Bohai Sea, as a semi-enclosed shallow sea area, has a relatively shallow average water depth and is significantly influenced by land. The Yellow Sea has complex water depth variations and tidal systems. The East China Sea is strongly influenced by the Kuroshio Current and typhoons. The South China Sea, as a deep-water basin, has complete deep-sea environmental characteristics. The division of these four sub-regions was based on multi-factor comprehensive analysis to ensure the relative homogeneity of the marine environment within each region. In the adjustment of shallow water effect parameters in the Bohai Sea, the deep water dispersion relationship correction coefficient is calculated based on the local depth-to-wavelength ratio. When the depth-to-wavelength ratio is less than 0.5, the shallow water effect is significant. The correction coefficient expresses the influence of water depth on wave propagation speed using a hyperbolic tangent function. The bottom friction coefficient adjustment considers seabed sediment type and bottom roughness. Fine sandy bottoms correspond to smaller friction coefficients, while coarse sand and gravel bottoms correspond to larger friction coefficients. The specific values ​​of the friction coefficients are determined through bottom sediment sampling analysis and towing experiments. In the introduction of the ocean current influence coefficient in the Yellow Sea, the velocity distribution data of the Yellow Sea warm current and coastal current are first obtained. Then, the relative relationship between current velocity and wave phase velocity is calculated. When the current direction is consistent with the wave propagation direction, an acceleration effect occurs; when they are opposite, a deceleration effect occurs. Tidal harmonic analysis decomposes complex tidal signals into harmonic components of different frequencies. The main tidal components include semi-diurnal and diurnal tides. The analysis process uses the least squares method to fit the observation data to obtain the amplitude and phase parameters of each tidal component. The parameter adjustment results for shallow sea areas comprehensively consider the combined effects of multiple factors such as water depth, seabed sediment, ocean currents, and tides.

[0074] During the adjustment of typhoon impact parameters in the East China Sea, the typhoon impact radius was determined based on historical typhoon path data and intensity analysis. Typically, waves are significantly affected within a 500-kilometer radius centered on the typhoon center. The growth rate coefficient adjustment considers the nonlinear wave generation mechanism under extreme typhoon wind speeds. When wind speeds exceed 25 meters per second, the traditional linear growth relationship no longer applies, requiring the introduction of a nonlinear correction term. High-frequency dissipation parameter optimization addresses the strong breaking process under typhoon conditions. Wave breaking frequency increases significantly under typhoon influence, and dissipation intensity increases exponentially with wind speed. The deep-water parameter configuration in the South China Sea is based on the physical conditions of a standard deep-sea environment. The far-field swell propagation attenuation parameter considers energy loss during long-distance wave propagation. Attenuation mechanisms include viscous dissipation, air resistance, and nonlinear effects. The attenuation coefficient is related to wave frequency and propagation distance. The deep-sea region parameter adjustment results focus on accurately describing the propagation and evolution characteristics of deep-water waves.

[0075] The independent validation dataset was established by collecting observational data for four sub-regions. Validation data for the Bohai Sea primarily came from nearshore buoy stations and tide gauge stations; data for the Yellow Sea included fixed-point observations from oceanographic research vessels and shore-based radar data; data for the East China Sea focused on collecting extreme value observation records during typhoons; and data for the South China Sea relied on long-term observations from deep-sea buoys and satellite altimeters. A continuous 12-month validation cycle ensured coverage of complete seasonal variations and various sea state conditions. The validation process evaluated the effectiveness of parameter configurations by calculating statistical indicators of simulated and observed values. Fixed-point buoy data included time series of basic parameters such as wave height, period, and wave direction, while satellite altimeter data provided extensive sea surface height observation information. The combination of these two types of data ensured the comprehensiveness and reliability of the validation results.

[0076] In one specific embodiment, the process of executing step S106 may specifically include the following steps:

[0077] A real-time monitoring system with a hierarchical architecture is constructed based on the aforementioned regionally differentiated parameter configuration. It continuously collects and processes data on the effective wave height, spectral peak period, average wave direction parameter, and satellite altimeter along-orbit wave height of buoy observations every hour to obtain a real-time ocean observation data stream.

[0078] The real-time ocean observation data stream is used to continuously evaluate the forecast results of the most recent 72 hours using the sliding window technique. The time series mean, standard deviation, trend coefficient and periodic characteristic statistics of the forecast error are calculated to obtain the time series analysis results of the forecast performance.

[0079] The Mann-Kendall trend test, Kolmogorov-Smirnov distribution consistency test, and Runs test were performed on the time series analysis results of the forecast performance. When a systematic deviation or significant trend change in the forecast error was found, a parameter adjustment warning was triggered, and the statistical significance test results were obtained.

[0080] The statistical significance test results are input into the support vector machine classifier to identify the parameter categories that need to be adjusted. The parameter adjustment magnitude is predicted by the random forest regressor. An incremental update strategy is adopted to limit the single adjustment magnitude to no more than 5% of the current value and the continuous adjustment interval to no less than 24 hours, thus obtaining a dynamic parameter update mechanism.

[0081] Specifically, the construction process of the layered real-time monitoring system is based on regionally differentiated parameter configuration and adopts a three-layer architecture design: a data acquisition layer, a processing layer, and an analysis layer. The data acquisition layer is responsible for acquiring raw observation data from buoy stations and satellite platforms; the processing layer performs data quality control and format conversion; and the analysis layer performs performance evaluation and parameter adjustment decisions. Buoy observation significant wave height data is obtained by measuring sea surface pressure changes using pressure sensors; spectral peak period data identifies dominant frequencies with concentrated energy through spectrum analysis; average wave direction data calculates the average azimuth angle of wave propagation using a multi-directional sensor array; and satellite altimeter along-orbit wave height data measures sea surface height anomalies using a radar altimeter. Continuous hourly data acquisition and processing includes four stages: data reception, decoding, verification, and storage. The data reception module acquires raw data packets from each observation platform according to a preset communication protocol; the decoding module converts binary data into physical quantity values; the verification module checks the integrity and rationality of the data; and the storage module organizes the processed data according to time series and establishes an index, forming a continuous time-series database from the real-time ocean observation data stream. In the sliding window technique for continuous performance evaluation of the most recent 72-hour forecast results, the window length is set to 72 hours to ensure coverage of the complete weather change cycle. The window slides forward in 1-hour increments, and the statistical indicators within the window are recalculated after each move. Forecast error is calculated by subtracting observed values ​​from forecast values ​​to obtain the error series. The time series mean reflects the magnitude and direction of systematic deviation, the standard deviation quantifies the dispersion of the error, the trend coefficient analyzes the trend of error over time using linear regression, and the periodic characteristic statistics identify the periodic components in the error series using Fourier transform. During the calculation of statistics within the sliding window, the mean is calculated by taking the arithmetic mean of all error values ​​within the window; the standard deviation is calculated by first taking the square root of the sum of squares of the deviations between the error values ​​and the mean; the trend coefficient fits the linear relationship between error and time using the least squares method; periodic analysis identifies the main frequency components using fast Fourier transform; and the time series analysis results of forecast performance include the temporal evolution characteristics of the four statistics.

[0082] The Mann-Kendall trend test first calculates the sign difference for each pair of observations in the time series, then counts the number of positive and negative differences, constructs a test statistic, and compares it with the critical value to determine the significance of the trend. The Kolmogorov-Smirnov distribution consistency test assesses whether the error distribution conforms to the expected statistical distribution by comparing the maximum difference between the empirical distribution function and the theoretical distribution function. The runs test analyzes the run characteristics of the error series; a run is defined as a continuous sequence of error values ​​with the same sign, and the statistical characteristics of run length and number reflect the degree of randomness of the series. Systematic bias judgment is based on the results of the Mann-Kendall test; when the test statistic exceeds the critical value, it indicates a significant trend. Significant trend changes are identified by detecting abrupt changes in the trend coefficient. The parameter adjustment early warning mechanism triggers different levels of early warning signals based on the combination patterns of the test results. The statistical significance test results provide a scientific basis for subsequent parameter adjustments.

[0083] In the process of identifying the parameter categories that need adjustment using a Support Vector Machine (SVM) classifier, the input feature vector includes the statistical significance test results and the current parameter configuration information. The classifier learns the correspondence between different error patterns and parameter categories through training samples. The kernel function uses radial basis functions to handle nonlinear classification problems, and the classification result determines the specific parameter type that needs adjustment. When the Random Forest (RFS) regressor predicts the parameter adjustment magnitude, the input variables include the current error level, trend strength, and historical adjustment experience. The regressor establishes a mapping relationship between error and adjustment magnitude through ensemble learning of multiple decision trees, and the prediction result provides a suggested adjustment amount for each parameter. In the process of limiting the single adjustment magnitude using an incremental update strategy, the adjustment magnitude constraint ensures the gradualness of parameter changes. The 5% limit is based on a trade-off between numerical stability and physical rationality, and the continuous adjustment interval limit prevents computational instability caused by frequent parameter changes. The dynamic parameter update mechanism integrates a complete process of early warning triggering, parameter identification, magnitude prediction, and constraint control.

[0084] The parametric analysis method based on wave spectrum models in the embodiments of this application has been described above. The parametric analysis system based on wave spectrum models in the embodiments of this application is described below. Please refer to [link / reference]. Figure 2 One embodiment of the parameterized analysis system based on wave spectrum models in this application includes:

[0085] The imputation module 201 is used to perform outlier detection and missing data imputation on multi-source ocean data using the 3σ criterion and Kriging interpolation algorithm, and to unify the processed data to a 0.1 degree × 0.1 degree spatial grid to obtain a standardized marine environment dataset.

[0086] Decomposition module 202 is used to decompose the source term of the wave energy balance equation into three modules—wind energy input term, nonlinear interaction term, and energy dissipation term—based on the standardized marine environment dataset, and perform numerical calculations to obtain the numerical solution of the wave spectrum module.

[0087] Analysis module 203 is used to perform correlation coefficient and root mean square error statistical analysis on the numerical solutions of the wave spectrum module configured with ST2, ST4 and ST6 parameterization schemes respectively, and obtain the performance evaluation index of the parameterization scheme;

[0088] The optimization module 204 is used to perform constrained optimization of the wind input growth rate coefficient based on the performance evaluation index of the parameterized scheme using a genetic algorithm combined with particle swarm optimization, so as to obtain a global optimal parameter combination.

[0089] The adjustment module 205 is used to divide the coastal waters of China into four sub-regions according to the global optimization parameter combination and to make differentiated adjustments to the bottom friction coefficient and ocean current influence coefficient to obtain regional differentiated parameter configurations.

[0090] The update module 206 is used to perform Mann-Kendall trend tests and incremental parameter updates on 72-hour forecast data using sliding window technology based on the regionally differentiated parameter configuration, thereby obtaining a dynamic parameter update mechanism.

[0091] above Figure 2 The parametric analysis system based on wave spectrum patterns in this embodiment of the invention will be described in detail from the perspective of modular functional entities. The parametric analysis device based on wave spectrum patterns in this embodiment of the invention will be described in detail from the perspective of hardware processing.

[0092] Reference Figure 3 This invention also provides a parametric analysis device based on wave spectrum patterns. This device can be a server, and its internal structure can be as follows: Figure 3 As shown, the wave spectrum-based parametric analysis device includes a processor, memory, display screen, input device, network interface, and database connected via a system bus. The processor provides computational and control capabilities. The memory of the wave spectrum-based parametric analysis device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the wave spectrum-based parametric analysis device stores the data corresponding to this embodiment. The network interface of the wave spectrum-based parametric analysis device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the above-described method.

[0093] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the parametric analysis device based on wave spectrum patterns to which the present invention is applied.

[0094] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the parameterized analysis method based on the wave spectrum pattern.

[0095] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0096] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a parameterized analysis device based on wave spectrum patterns (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0097] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A parametric analysis method based on wave spectrum models, characterized in that, The method includes: Outlier detection and missing data imputation were performed on multi-source ocean data using the 3σ criterion and Kriging interpolation algorithm. The processed data were then unified to a 0.1 degree × 0.1 degree spatial grid to obtain a standardized marine environment dataset. Based on the standardized marine environment dataset, the source term of the wave energy balance equation is decomposed into three modules: wind energy input term, nonlinear interaction term, and energy dissipation term, and numerical calculation is performed to obtain the numerical solution of the wave spectrum module. The numerical solutions of the wave spectrum module were configured with ST2, ST4 and ST6 parameterization schemes respectively, and the correlation coefficient and root mean square error were statistically analyzed to obtain the performance evaluation index of the parameterization scheme. Based on the performance evaluation index of the parameterized scheme, a genetic algorithm combined with particle swarm optimization is used to constrain and optimize the growth rate coefficient of the wind input term to obtain a global optimal parameter combination. Based on the global optimization parameter combination, the coastal waters of China are divided into four sub-regions for differentiated adjustment of bottom friction coefficient and ocean current influence coefficient, resulting in regional differentiated parameter configurations; Based on the regionally differentiated parameter configuration, the sliding window technique is used to perform Mann-Kendall trend tests and incremental parameter updates on the 72-hour forecast data, resulting in a dynamic parameter update mechanism.

2. The parameterized analysis method based on wave spectrum models according to claim 1, characterized in that, The method employs the 3σ criterion and Kriging interpolation algorithm to perform outlier detection and missing data imputation on multi-source ocean data. The processed data is then standardized to a 0.1° × 0.1° spatial grid, resulting in a standardized marine environment dataset, including: Satellite remote sensing data, ocean buoy observation data, shore-based radar observation data, and numerical weather prediction data were collected continuously for 72 hours using a standardized interface to obtain multi-source ocean data including sea surface wind speed, wind direction, significant wave height, spectral peak period, and average wave direction. The mean and standard deviation of the data sequence are calculated using the 3σ criterion combined with the sliding window statistical method for the multi-source ocean data. Data points exceeding 3 times the standard deviation are marked as outliers and removed to obtain quality-controlled ocean observation data. Based on the quality-controlled marine observation data, a spatial covariance function model is established using the Kriging interpolation algorithm. The interpolation weight coefficients are solved by the least squares method to perform spatial interpolation calculations on the missing data points, thereby obtaining spatially continuous marine parameter data. The spatially continuous ocean parameter data is spatiotemporally synchronized according to a unified time coordinate system and spatial coordinate system, and then resampled to a 0.1 degree × 0.1 degree spatial grid using a bilinear interpolation method to obtain a standardized ocean environment dataset.

3. The parameterized analysis method based on wave spectrum models according to claim 1, characterized in that, The source term of the wave energy balance equation is decomposed into three modules—wind energy input, nonlinear interaction, and energy dissipation—based on the standardized marine environment dataset, and numerical calculations are performed to obtain the numerical solution of the wave spectrum module, including: Based on spatial grid data, an ocean wave energy balance equation is established. The source term is decomposed into three independent modules according to the physical mechanism: wind energy input term, nonlinear interaction term, and energy dissipation term, resulting in a modular ocean wave spectrum calculation system. A growth rate calculation model is established for the wind energy input term using Miles theory. The wind stress transmission coefficient is corrected by calculating the atmospheric stability parameter using the Richardson number. The nonlinear characteristics of the sea surface drag coefficient as a function of wave age are introduced into the calculation process to obtain the numerical solution of the wind energy input term. An improved discrete mutual approximation algorithm is used to apply the nonlinear interaction term. The interaction wavenumber combination is identified by the four-wave resonance condition. The energy transfer rate is calculated with high precision by adaptive mesh refinement technology to obtain the numerical solution of the nonlinear interaction term. A probabilistic-deterministic hybrid model is adopted for the energy dissipation term. The wave breaking conditions are judged by combining the statistical characteristics of sea surface slope and local wave steepness parameters. When the wave steepness exceeds the critical value, the breaking dissipation calculation is performed to obtain the numerical solution of the energy dissipation term. The numerical solutions of the wind energy input term, the nonlinear interaction term, and the energy dissipation term are combined into the numerical solution of the wave spectrum module.

4. The parameterized analysis method based on wave spectrum models according to claim 1, characterized in that, The numerical solutions of the wave spectrum module are configured with ST2, ST4, and ST6 parameterization schemes, respectively, and subjected to correlation coefficient and root mean square error statistical analysis to obtain performance evaluation indicators of the parameterization schemes, including: The numerical solution of the wave spectrum module was configured with grid resolution for the coastal area of ​​China according to three parameterization schemes: ST2, ST4 and ST6. 30 days of continuous numerical calculation with a time step of 300 seconds was performed in the range of 105°E-130°E and 10°N-45°N to obtain the calculation results of the three parameterization schemes. The calculation results of the three parameterization schemes were compared and analyzed with the ocean buoy observation data in terms of effective wave height, spectral peak frequency and wave direction parameters. The Pearson correlation coefficient calculation method was used to calculate the linear correlation between each scheme and the observation data, and the correlation coefficient matrix of each parameterization scheme was obtained. Based on the correlation coefficient matrix of each parameterized scheme, the root mean square error, mean absolute error, and deviation index of the forecast and observed values ​​are calculated. The forecast accuracy of each scheme is quantitatively evaluated and analyzed by statistical testing methods to obtain multi-dimensional statistical evaluation results. Based on the multi-dimensional statistical evaluation results, a comprehensive scoring system is established. The weights of the effective wave height prediction accuracy, the spectral peak frequency prediction deviation, and the wave direction prediction error are set to 0.3 and 0.3 respectively. The performance evaluation index of the parameterized scheme is obtained by weighted calculation.

5. The parameterized analysis method based on wave spectrum models according to claim 1, characterized in that, The performance evaluation index based on the parameterized scheme employs a genetic algorithm combined with particle swarm optimization to constrain and optimize the growth rate coefficient of the wind input term, resulting in a globally optimized parameter combination, including: Based on the performance evaluation index of the parameterized scheme, the wind input growth rate coefficient, sea surface roughness parameter, nonlinear interaction coupling strength parameter, and dissipation term breakage threshold optimization parameter are determined. A multi-objective weighted objective function is established to comprehensively evaluate the effective wave height root mean square error, spectral peak period deviation, wave direction correlation coefficient, and extreme event skill score, and the parameter optimization objective function is obtained. The parameter optimization objective function is input into the genetic algorithm to initialize the population. The population size is set to 100 individuals. The parameter genes of each individual are randomly encoded. The population evolution calculation is performed using tournament selection, single-point crossover and Gaussian mutation operations to obtain the population evolution result of the genetic algorithm. The population evolution results of the genetic algorithm are subjected to a local fine search using a particle swarm optimization algorithm. The cognitive coefficient and social coefficient in the particle velocity update are both set to 2.

0. The optimal solution in the parameter space is found by iteratively updating the particle position and velocity, and the search results of the hybrid optimization algorithm are obtained. Based on the search results of the hybrid optimization algorithm, parameter constraints are set, the growth rate coefficient is constrained to a reasonable range, the sea surface roughness parameter is constrained to a physically feasible range, and the breakage threshold is constrained to an empirical range. The iterative calculation is terminated when the improvement of the objective function is less than the set threshold for 50 consecutive generations, and the global optimization parameter combination is obtained.

6. The parameterized analysis method based on wave spectrum models according to claim 1, characterized in that, The process involves dividing China's coastal waters into four sub-regions based on the globally optimized parameter combination, and then adjusting the bottom friction coefficient and ocean current influence coefficient accordingly to obtain regionally differentiated parameter configurations, including: Based on the global optimization parameter combination, China's coastal waters are divided into four sub-regions: the Bohai Sea, the Yellow Sea, the East China Sea, and the South China Sea, according to differences in water depth distribution, topographic features, meteorological conditions, and marine dynamic environment, thus obtaining a marine zoning scheme. The shallow water effect parameters of the Bohai Sea region were adjusted by using the deep-water dispersion relation correction coefficient and the bottom friction coefficient. The ocean current influence coefficient and tidal harmonic analysis were introduced into the Yellow Sea region to correct the ocean current-wave interaction, and the parameter adjustment results of the shallow sea region were obtained. The growth rate coefficient was adjusted and the high-frequency dissipation parameters were optimized for the East China Sea region within the typhoon's influence radius. For the South China Sea region, the far-field surge propagation attenuation parameters were optimized using standard deep-water parameter configuration, resulting in the deep-sea region parameter adjustment results. Based on the parameter adjustment results of the shallow sea area and the parameter adjustment results of the deep sea area, an independent verification dataset for each sub-region is established. The effectiveness of the parameter configuration of each sub-region is verified by 12 consecutive months of fixed-point buoy data and satellite altimeter data, and regional differentiated parameter configurations are obtained.

7. The parameterized analysis method based on wave spectrum models according to claim 1, characterized in that, The method based on the regionally differentiated parameter configuration employs a sliding window technique to perform Mann-Kendall trend testing and incremental parameter updates on the 72-hour forecast data, resulting in a dynamic parameter update mechanism, including: A real-time monitoring system with a hierarchical architecture is constructed based on the aforementioned regionally differentiated parameter configuration. It continuously collects and processes data on the effective wave height, spectral peak period, average wave direction parameter, and satellite altimeter along-orbit wave height of buoy observations every hour to obtain a real-time ocean observation data stream. The real-time ocean observation data stream is used to continuously evaluate the forecast results of the most recent 72 hours using the sliding window technique. The time series mean, standard deviation, trend coefficient and periodic characteristic statistics of the forecast error are calculated to obtain the time series analysis results of the forecast performance. The Mann-Kendall trend test, Kolmogorov-Smirnov distribution consistency test, and Runs test were performed on the time series analysis results of the forecast performance. When a systematic deviation or significant trend change in the forecast error was found, a parameter adjustment warning was triggered, and the statistical significance test results were obtained. The statistical significance test results are input into the support vector machine classifier to identify the parameter categories that need to be adjusted. The parameter adjustment magnitude is predicted by the random forest regressor. An incremental update strategy is adopted to limit the single adjustment magnitude to no more than 5% of the current value and the continuous adjustment interval to no less than 24 hours, thus obtaining a dynamic parameter update mechanism.

8. A parameterized analysis system based on wave spectrum models, characterized in that, For implementing the parameterized analysis method based on wave spectrum models as described in any one of claims 1-7, the parameterized analysis system based on wave spectrum models comprises: The imputation module is used to perform outlier detection and missing data imputation on multi-source ocean data using the 3σ criterion and Kriging interpolation algorithm. The processed data is then unified to a 0.1 degree × 0.1 degree spatial grid to obtain a standardized marine environment dataset. The decomposition module is used to decompose the source term of the wave energy balance equation into three modules—wind energy input term, nonlinear interaction term, and energy dissipation term—based on the standardized marine environment dataset, and then perform numerical calculations to obtain the numerical solution of the wave spectrum module. The analysis module is used to perform correlation coefficient and root mean square error statistical analysis on the numerical solutions of the wave spectrum module configured with ST2, ST4 and ST6 parameterization schemes respectively, and obtain the performance evaluation index of the parameterization scheme; The optimization module is used to perform constrained optimization of the wind input growth rate coefficient based on the performance evaluation index of the parameterized scheme using a genetic algorithm combined with particle swarm optimization, so as to obtain a global optimal parameter combination. The adjustment module is used to divide the coastal waters of China into four sub-regions based on the global optimization parameter combination and to make differentiated adjustments to the bottom friction coefficient and ocean current influence coefficient to obtain regional differentiated parameter configurations. The update module is used to perform Mann-Kendall trend tests and incremental parameter updates on 72-hour forecast data based on the regionally differentiated parameter configuration using the sliding window technique, thereby obtaining a dynamic parameter update mechanism.

9. A parametric analysis device based on wave spectrum models, characterized in that, It includes a memory and a processor, the memory storing a computer program that can run on the processor, and the processor executing the computer program to implement the parameterized analysis method based on wave spectrum patterns according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is run by the processor, it causes the processor to perform the parametric analysis method based on wave spectrum patterns as described in any one of claims 1 to 7.

Images