Sea fog forecasting method, medium and system based on ocean station observation data

Abstract

The invention provides a sea fog forecasting method, medium and system based on ocean station observation data, and belongs to the technical field of electrical digital data processing, and the method comprises the steps: firstly, obtaining years of continuous hydro meteorological observation data of an observation station, including air temperature, dew point temperature, seawater temperature, wind direction and wind speed and other data; performing multi-scale decomposition through wavelet transform to obtain variable components of different time scales; then, forecasting factors including temperature difference, stability index and the like are calculated, and layered Fisher discriminant analysis is carried out according to seasons and weather system types. A polynomial discrimination equation set is established by calculating a weight coefficient, and a discrimination critical value is determined by combining a probability density function, so that whether the weather is foggy or not is judged. And finally, verifying by adopting a three-level inspection system, and constructing a complete forecast product release system including the strength grade of fog, the space-time evolution and the like, thereby realizing the accurate forecast of the sea fog. The technical problem that the accuracy of a traditional forecasting method fluctuates greatly under different conditions is solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of electrical digital data processing. Specifically, it relates to a sea fog forecasting method, medium and system based on ocean station observation data. Background Art

[0002] As a common marine meteorological phenomenon, accurate forecasting of sea fog has important strategic significance and practical application value for maritime traffic safety, offshore engineering operations, marine fishery production, and coastal military activities. In recent years, with the rapid development of global maritime trade, the maritime traffic flow has continued to grow, and the trend of ship enlargement is obvious, which puts forward higher requirements for the accuracy and timeliness of sea fog forecasting.

[0003] Currently, three types of technical methods are mainly used for sea fog forecasting: The first type is the method based on numerical forecasting models, which simulates the formation and evolution of sea fog by establishing a numerical model that includes water vapor, heat, and momentum exchange processes. Although this type of method has a strong physical basis, due to the complexity of the sea fog formation mechanism and the high dependence of the model on the initial field and boundary conditions, its forecasting effect is often not ideal. The second type is the statistical forecasting method, which mainly relies on historical data to establish a statistical forecasting model. This type of method is easy to operate, but it is difficult to accurately describe the non-linear characteristics of sea fog formation and has poor adaptability to abnormal weather conditions. The third type is the forecasting method based on expert experience, which mainly relies on the experience judgment of forecasters. Although it can make a forecast in combination with the actual situation, there are problems such as strong subjectivity and difficulty in inheriting experience.

[0004] Traditional forecasting methods often focus on single meteorological elements or simple combinations of elements, ignoring the complex interactions between multi-scales and multi-elements during the formation process of sea fog, resulting in large fluctuations in the forecasting accuracy under different circumstances. Summary of the Invention

[0005] In view of this, the present invention provides a sea fog forecasting method, medium and system based on ocean station observation data, which can solve the technical problem that traditional forecasting methods often focus on single meteorological elements or simple combinations of elements, ignoring the complex interactions between multi-scales and multi-elements during the formation process of sea fog, resulting in large fluctuations in the forecasting accuracy under different circumstances.

[0006] The present invention is implemented as follows:

[0007] The first aspect of the present invention provides a sea fog forecasting method based on ocean station observation data, including: obtaining multi-year continuous hydrometeorological observation data of the station; performing multi-scale decomposition on the hydrometeorological observation data by using the wavelet transform method; calculating forecasting factors based on the multi-scale decomposition data; dividing the forecasting factors into time periods and performing secondary classification to obtain hierarchical Fisher discriminant sample data; calculating the scatter matrix based on the Fisher secondary discriminant principle; establishing a Fisher discriminant function coefficient matrix by using the cross-validation method; constructing and solving a polynomial discriminant equation system to obtain a sea fog forecasting discriminant equation; calculating the discriminant critical value of the sea fog forecasting discriminant equation; verifying the sea fog forecasting discriminant equation by using a three-level test system; and constructing a forecasting product release system for release.

[0008] Among them, the hydrometeorological observation data includes the station air temperature value, the station dew point temperature value, the upstream measuring point dew point temperature value, the station surface seawater temperature value, the station wind direction and wind speed data, the station air pressure data, the station relative humidity data, and the synchronous observation data of adjacent stations. The synchronous observation data of adjacent stations adopts a sampling frequency of once per hour.

[0009] Among them, the multi-scale decomposition obtains an interannual variation component, a seasonal variation component, a daily-scale variation component, and a random perturbation component.

[0010] Among them, the forecasting factors include the difference between the upstream dew point temperature and the local air temperature, the difference between the station air temperature and the dew point temperature, the difference between the station air temperature and the seawater surface temperature, the vertical gradient of the sea-air temperature difference, the stability index of the near-surface atmosphere, the convergence intensity of the water vapor flux, and the characteristic parameters of the inversion layer structure.

[0011] Among them, the time period division is carried out according to winter, spring, summer, autumn, and two transition periods of spring and autumn. The secondary classification is based on the weather system type, and the weather system type includes cold high pressure type, warm high pressure type, and frontal type.

[0012] Among them, the scatter matrix includes a weighted between-class scatter matrix and a within-class scatter matrix. The weighted between-class scatter matrix characterizes the difference degree between the weighted means of the foggy weather and non-foggy weather samples, and the within-class scatter matrix characterizes the dispersion degree of the samples within their respective categories.

[0013] Among them, the cross-validation method randomly groups the forecasting factors multiple times, calculates the within-class scatter cross-product sum of the forecasting factors, and establishes a Fisher discriminant function coefficient matrix including a regularization term.

[0014] Among them, the discrimination critical value of the sea fog forecast discrimination equation selects the optimal classification threshold by analyzing the probability distribution characteristics of foggy samples and non-foggy samples. When the calculation result of the discrimination equation is greater than the discrimination critical value, it is determined as foggy weather, and when the calculation result of the discrimination equation is less than the discrimination critical value, it is determined as non-foggy weather.

[0015] Specifically, the method of the present invention includes the following steps:

[0016] S01. Obtain continuous hydrometeorological observation data of the station for many years. The hydrometeorological observation data includes the station air temperature value, the station dew point temperature value, the upstream measuring point dew point temperature value, the station surface seawater temperature value, the station wind direction and wind speed data, the station air pressure data, the station relative humidity data, and the synchronous observation data of adjacent stations. The synchronous observation data of adjacent stations adopts a sampling frequency of once per hour;

[0017] S02. Perform multi-scale decomposition on the hydrometeorological observation data by using the wavelet transform method to obtain the interannual change component, the seasonal change component, the daily-scale change component, and the random perturbation component;

[0018] S03. Calculate the forecast factors based on the multi-scale decomposition data. The forecast factors include the difference between the upstream dew point temperature and the local air temperature, the difference between the local air temperature and the dew point temperature, the difference between the local air temperature and the seawater surface temperature, the vertical gradient of the sea-air temperature difference, the stability index of the near-surface atmosphere, the convergence intensity of the water vapor flux, and the characteristic parameters of the inversion layer structure;

[0019] S04. Divide the forecast factors according to winter, spring, summer, autumn, and two transition periods of spring and autumn, and perform secondary classification according to the weather system type to obtain hierarchical Fisher discrimination sample data;

[0020] S05. Calculate the weighted between-class scatter matrix and the within-class scatter matrix based on the Fisher secondary discrimination principle, and determine the weight coefficient through sample representativeness;

[0021] S06. Use the cross-validation method to randomly group the forecast factors multiple times, calculate the within-class scatter cross-product sum of the forecast factors, and establish a Fisher discrimination function coefficient matrix including a regularization term;

[0022] S07. Based on the Fisher discrimination function coefficient matrix, construct and solve a polynomial discrimination equation system considering the non-linear relationship of the forecast factors to obtain the sea fog forecast discrimination equation for each combination of season and weather system type;

[0023] S08, calculating the critical value of the sea fog forecast discriminant equation based on the probability density function method, and determining that there is foggy weather when the calculation result of the discriminant equation is greater than the critical value, and determining that there is no foggy weather when the calculation result of the discriminant equation is less than the critical value;

[0024] S09, using a three-level verification system to verify the sea fog forecast discriminant equation to obtain the forecast accuracy, false alarm rate, missed alarm rate, threat score and skill score;

[0025] S10. Construct a forecast product release system, wherein the forecast product includes fog or no fog discrimination results, fog intensity level forecast, time evolution forecast and spatial distribution forecast.

[0026] On the basis of the above technical solution, the sea fog forecasting method based on ocean station observation data of the present invention can also be improved as follows:

[0027] Among them, the weather system types include cold high pressure type, warm high pressure type, and frontal type.

[0028] Cold high pressure type: usually brings clear, dry and cold weather. Cold high pressure system is a high pressure system formed by the sinking of dense cold air, which is common in winter.

[0029] Warm high pressure type: usually brings clear, warm and stable weather. Warm high pressure system is a high pressure system formed by warm air rising and then sinking, which is common in summer.

[0030] Frontal type: formed at the junction of cold and warm air masses, according to the relative position and movement direction of the cold and warm air masses, it can be divided into cold front, warm front, stationary front and occluded front, etc. Frontal systems usually bring precipitation and significant temperature changes.

[0031] Furthermore, the weighted inter-class scatter matrix represents the degree of difference between the weighted means of samples in foggy weather and non-fog weather, and the intra-class scatter matrix represents the degree of dispersion of samples within their respective categories.

[0032] Furthermore, the hydrological and meteorological observation data are quality controlled, outliers are eliminated, and the missing data are interpolated using objective analysis methods.

[0033] Furthermore, the optimal forecast lead time of the forecast factor is determined by lagged correlation analysis.

[0034] Furthermore, the critical value of the sea fog forecast discriminant equation selects the optimal classification threshold by analyzing the probability distribution characteristics of foggy samples and non-fog samples.

[0035] Furthermore, the three-level test system includes contingency table test, forecast skill evaluation and forecast stability test.

[0036] Furthermore, the prediction model is dynamically optimized through a prediction correction mechanism.

[0037] The second aspect of the present invention provides a computer-readable storage medium, in which program instructions are stored, and when the program instructions run on a computer, they are used to execute the above-mentioned sea fog prediction method based on ocean station observation data.

[0038] The third aspect of the present invention provides a sea fog prediction system based on ocean station observation data, which includes the above-mentioned computer-readable storage medium.

[0039] Furthermore, in S01, the quality control equation of hydrometeorological observation data is as follows:

[0040]

[0041] In the formula, Q i is the standardized residual; X i is the i-th observed value; is the sample mean; σ is the sample standard deviation. When Q i > 3, it is determined as an outlier.

[0042] For the objective analysis and interpolation of missing data, the inverse distance weighted method is adopted:

[0043]

[0044] In the formula, X p is the estimated value of the point to be interpolated; X i is the observed value of the i-th neighboring station; d i is the distance from the i-th station to the point to be interpolated; n is the number of neighboring stations participating in the interpolation.

[0045] In S02, continuous wavelet transform is used to decompose the time series:

[0046]

[0047] In the formula, W f (a, b) is the wavelet coefficient; f(t) is the original signal; ψ(t) is the wavelet mother function; a is the scale parameter; b is the translation parameter; ψ * represents the conjugate complex number of the wavelet function.

[0048] In S03, the calculation of prediction factors includes:

[0049] The difference between the upstream dew point temperature and the local air temperature:

[0050]

[0051] Where, ΔT ud is the temperature difference; is the upstream dew point temperature; T a is the local air temperature.

[0052] The difference between the local air temperature and the dew point temperature:

[0053] ΔT d = T a - T d ;

[0054] Where, ΔT d is the temperature difference; T d is the local dew point temperature.

[0055] The difference between the local air temperature and the sea surface temperature:

[0056] ΔT s = T a - T s ;

[0057] Where, ΔT s is the temperature difference; T s is the sea surface temperature.

[0058] The vertical gradient of the sea-air temperature difference:

[0059]

[0060] Where, γ t is the vertical temperature gradient; z2, z1 are two altitude levels.

[0061] The stability index of the atmospheric boundary layer:

[0062]

[0063] Where, Ri is the Richardson number; g is the acceleration due to gravity; is the average temperature; θ is the potential temperature; u is the wind speed.

[0064] The convergence intensity of the water vapor flux:

[0065]

[0066] Where, Q is the convergence of the water vapor flux; q is the specific humidity; is the wind speed vector.

[0067] The characteristic parameter of the inversion layer structure:

[0068] I t = ΔT max ·H;

[0069] Where, I t is the inversion intensity index; ΔTmax is the maximum temperature difference; H is the thickness of the inversion layer.

[0070] The specific Fisher discriminant analysis in S04 - S07 includes:

[0071] Between-class scatter matrix:

[0072]

[0073] In the formula, S B is the between-class scatter matrix; ω i is the weight of the i-th class; m i is the mean vector of the samples in the i-th class; m is the mean vector of the overall samples.

[0074] Within-class scatter matrix:

[0075]

[0076] In the formula, S W is the within-class scatter matrix; x ij is the j-th sample vector in the i-th class; n i is the number of samples in the i-th class.

[0077] Regularized Fisher discriminant function:

[0078]

[0079] In the formula, J(ω) is the discriminant criterion function; ω is the discriminant vector; λ is the regularization parameter.

[0080] Polynomial discriminant equation:

[0081]

[0082] In the formula, f(x) is the discriminant function; x i , x j are the prediction factors; α ij , β i are the coefficients; c is the constant term; p is the number of prediction factors.

[0083] The determination of the discriminant critical value in S08 includes:

[0084] Probability density function based on kernel density estimation:

[0085]

[0086] In the formula, f h (x) is the probability density function; K(·) is the kernel function; h is the bandwidth parameter; n is the number of samples.

[0087] Optimal discriminant critical value:

[0088] T opt = argmin T {P miss (T) + αP false (T)};

[0089] Where T opt is the optimal critical value; P miss is the miss probability; P false is the false alarm probability; α is the weight coefficient.

[0090] The forecast verification in S09 includes:

[0091] Forecast accuracy rate:

[0092]

[0093] Where ACC is the accuracy rate; H is the number of hits; M is the number of misses; F is the number of false alarms; CN is the number of correct negatives.

[0094] Threat score:

[0095]

[0096] Where TS is the threat score.

[0097] Skill score:

[0098]

[0099] Where SS is the skill score; ACC r is the random forecast accuracy rate.

[0100] The forecast product generation in S10 includes:

[0101] Fog intensity level forecast:

[0102] V = a·exp(-b·RH) + c·ΔT d + d;

[0103] Where V is the visibility; RH is the relative humidity; a, b, c, d are fitting coefficients.

[0104] The spatial distribution forecast uses Kriging interpolation:

[0105]

[0106] Where Z(x0) is the estimated value of the point to be forecast; Z(x i ) is the observed value of the known point; λ i is the weight coefficient.

[0107] The weight coefficient satisfies:

[0108]

[0109] where γ(h) is the variogram; h ij is the distance between point i and point j; μ is the Lagrange multiplier.

[0110] Time evolution prediction:

[0111]

[0112] where f is the prediction element; is the transport velocity; D is the diffusion coefficient; S is the source-sink term.

[0113] Characterization of prediction uncertainty:

[0114]

[0115] where P(f|x) is the posterior probability; P(x|f) is the likelihood function; P(f) is the prior probability; P(x) is the normalization factor.

[0116] Prediction correction:

[0117]

[0118] where f c is the corrected prediction value; f0 is the original prediction value; y is the observed value; is the observation operator; K is the gain matrix.

[0119] Calculation of the gain matrix:

[0120] K = BH T (HBH T + R) -1 ;

[0121] where B is the background error covariance matrix; R is the observation error covariance matrix.

[0122] Compared with the prior art, the beneficial effects of a sea fog prediction method, medium and system based on ocean station observation data provided by the present invention are:

[0123] 1. Make full use of multi-source observation data of ocean stations. It includes not only conventional elements such as air temperature, dew point temperature, wind speed, etc., but also introduces relevant indicators such as upstream dew point temperature, seawater temperature, atmospheric stability, etc., which can more comprehensively reflect the sea-air interaction process. At the same time, wavelet analysis is used to perform multi-scale decomposition on these observation data, effectively capturing the variation laws on different time scales and laying a foundation for accurate prediction.

[0124] 2. A prediction model based on Fisher discriminant analysis was established. By calculating the weighted between-class scatter matrix and within-class scatter matrix, the optimal weights of each prediction factor were determined, improving the discriminant ability of the model. Meanwhile, considering the non-linear relationship between prediction factors, a polynomial discriminant equation system was constructed to further enhance the prediction accuracy.

[0125] 3. The prediction results are presented in a probabilistic description manner. Not only the discriminant results of fog or no fog are given, but also the intensity level, time evolution trend and spatial distribution of fog are quantitatively predicted, and the uncertainty of the prediction is reflected through the Bayesian probability model, enhancing the practicality of the prediction products.

[0126] 4. A complete prediction verification system was established. It includes contingency table test, prediction skill assessment and prediction stability test, comprehensively evaluating the performance of the constructed model. Meanwhile, through the prediction correction mechanism, the model is dynamically optimized to further improve the accuracy and reliability of the prediction.

[0127] In summary, the sea fog prediction method proposed by the present invention fully exploits the information potential of the ocean station observation data, constructs a prediction model based on mathematical statistics, and solves the technical problem that traditional prediction methods often focus on single meteorological elements or simple combinations of elements, ignoring the complex interactions among multiple scales and multiple elements in the sea fog formation process, resulting in large fluctuations in the prediction accuracy under different conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0128] Figure 1 It is a flowchart of the method provided by the present invention.

[0129] Figure 2 It is four fog-causing weather situation diagrams in the embodiment, divided into four sub-diagrams. Diagram a: the type of offshore modified high pressure; Diagram b: the type of northward extension of the subtropical high ridge; Diagram c: the type in front of the cyclone; Diagram d: the type in front of the trough behind the ridge; L represents the center of the low-pressure cyclone; H represents the high-pressure center.

[0130] Figure 3 It is a diagram of the wavelet decomposition result of the temperature time series in the embodiment.

[0131] Figure 4 It is a probability density distribution diagram of foggy samples and non-foggy samples in March in the embodiment.

[0132] Figure 5 It is a comparison diagram of the sea fog prediction skill scores under different weather system types in the embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0133] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0134] As shown Figure 1 in the figure, it is a flowchart of a sea fog forecasting method provided by the first aspect of the present invention. This method includes the following steps:

[0135] S01. Obtain continuous hydrometeorological observation data of the station for many years. The hydrometeorological observation data includes the station air temperature value, the station dew point temperature value, the upstream measuring point dew point temperature value, the station surface seawater temperature value, the station wind direction and wind speed data, the station air pressure data, the station relative humidity data, and the synchronous observation data of adjacent stations. The synchronous observation data of adjacent stations adopts a sampling frequency of once per hour;

[0136] S02. Perform multi-scale decomposition on the hydrometeorological observation data by using the wavelet transform method to obtain the interannual variation component, the seasonal variation component, the daily-scale variation component, and the random perturbation component;

[0137] S03. Calculate the forecasting factors based on the multi-scale decomposition data. The forecasting factors include the difference between the upstream dew point temperature and the local air temperature, the difference between the local air temperature and the dew point temperature, the difference between the local air temperature and the seawater surface temperature, the vertical gradient of the sea-air temperature difference, the stability index of the near-surface atmosphere, the convergence intensity of the water vapor flux, and the characteristic parameters of the inversion layer structure;

[0138] S04. Divide the forecasting factors into time periods according to winter, spring, summer, autumn, and two transition periods of spring and autumn, and perform secondary classification according to the weather system type to obtain the hierarchical Fisher discrimination sample data;

[0139] S05. Calculate the weighted between-class scatter matrix and the within-class scatter matrix based on the Fisher secondary discrimination principle, and determine the weight coefficient through sample representativeness;

[0140] S06. Use the cross-validation method to randomly group the forecasting factors multiple times, calculate the within-class deviation cross-product sum of the forecasting factors, and establish a Fisher discrimination function coefficient matrix including a regularization term;

[0141] S07. Construct a polynomial discrimination equation system considering the non-linear relationship of the forecasting factors based on the Fisher discrimination function coefficient matrix and solve it to obtain the sea fog forecasting discrimination equation for each combination of season and weather system type;

[0142] S08. Calculate the discrimination critical value of the sea fog forecasting discrimination equation based on the probability density function method. When the calculation result of the discrimination equation is greater than the discrimination critical value, it is determined as a foggy weather; when the calculation result of the discrimination equation is less than the discrimination critical value, it is determined as a non-foggy weather;

[0143] S09. Use a three-level test system to verify the sea fog forecasting discrimination equation to obtain the forecasting accuracy rate, false alarm rate, missed alarm rate, threat score, and skill score;

[0144] S10. Construct a prediction product release system, where the prediction products include the discrimination results of fog or no fog, the prediction of fog intensity level, the time evolution prediction, and the spatial distribution prediction.

[0145] The following is a detailed description of the specific implementation of the above steps:

[0146] First is step S01. The purpose of this step is to obtain continuous hydrometeorological observation data of a station over the years and perform quality control and missing value imputation on these data.

[0147] Specifically, for the quality control of the observation data, the method of standardized residuals is used to identify outliers. The calculation formula of the standardized residuals is: where Q i represents the standardized residual of the i-th observation value; X i represents the i-th observation value; represents the sample mean; σ represents the sample standard deviation. When Q i > 3, the observation value is determined as an outlier and excluded.

[0148] For the imputation of missing data, the inverse distance weighted method is used for objective analysis. Its calculation formula is: where X p represents the estimated value of the point to be imputed; X i represents the observation value of the i-th neighboring station; d i represents the distance from the i-th station to the point to be imputed; n represents the number of neighboring stations participating in the imputation.

[0149] Through the above data preprocessing, complete and reliable hydrometeorological observation data are obtained, laying a foundation for subsequent analysis.

[0150] Next is step S02. This step uses continuous wavelet transform to perform multi-scale decomposition on time series data. The formula of wavelet transform is: where W f (a, b) represents the wavelet coefficient; f(t) represents the original signal; ψ(t) represents the wavelet mother function; a represents the scale parameter; b represents the translation parameter; ψ * represents the conjugate complex number of the wavelet function.

[0151] Through wavelet transform, the original hydrometeorological observation data can be decomposed into interannual variation components, seasonal variation components, daily scale variation components, and random perturbation components. This multi-scale decomposition helps to better characterize the time variation characteristics of hydrometeorological elements and lays a foundation for the extraction of subsequent prediction factors.

[0152] In step S03, a series of predictors were calculated based on the multi-scale decomposition data obtained in step S02. These predictors include:

[0153] The difference between the dew point temperature upstream and the local air temperature: where, ΔT ud represents the temperature difference; represents the dew point temperature upstream; T a represents the local air temperature.

[0154] The difference between the air temperature at the station and the dew point temperature: ΔT d = T a - T d . where, ΔT d represents the temperature difference; T d represents the dew point temperature at the station.

[0155] The difference between the air temperature at the station and the sea surface temperature: ΔT s = T a - T s . where, ΔT s represents the temperature difference; T s represents the sea surface temperature.

[0156] The vertical gradient of the sea-air temperature difference: where, γ t represents the vertical temperature gradient; z2, z1 represent two height levels.

[0157] The stability index of the surface layer atmosphere: where, Ri represents the Richardson number; g represents the acceleration due to gravity; represents the average temperature; θ represents the potential temperature; u represents the wind speed.

[0158] The convergence intensity of the water vapor flux: where, Q represents the convergence of the water vapor flux; q represents the specific humidity; represents the wind speed vector.

[0159] The characteristic parameter of the inversion layer structure: I t = ΔT max · H. where, I t represents the inversion intensity index; ΔT max represents the maximum temperature difference; H represents the thickness of the inversion layer.

[0160] These predictors comprehensively reflect the impacts of processes such as air-sea interaction, atmospheric stability, and water vapor transport on the formation of sea fog, providing rich input variables for the construction of subsequent prediction models.

[0161] In step S04, first, the extracted forecasting factors are divided into time periods according to seasons, including winter, spring, summer, autumn, and two transitional periods of spring and autumn. Then, the data is further classified at the secondary level according to the types of weather systems, mainly including three types: cold high-pressure type, warm high-pressure type, and frontal type. Through this hierarchical processing, the differential characteristics of sea fog formation under different seasons and weather systems can be better reflected.

[0162] Next, based on the principle of Fisher discriminant analysis, the weighted between-class scatter matrix S of foggy weather and non-foggy weather samples under each classification combination is calculated. B and the within-class scatter matrix S W . Among them, the calculation formula of the between-class scatter matrix is: where ω i represents the weight of the i-th class; m i represents the mean vector of the i-th class of samples; m represents the mean vector of the overall samples. The calculation formula of the within-class scatter matrix is: where x ij represents the j-th sample vector of the i-th class; n i represents the number of samples in the i-th class.

[0163] These statistics lay the foundation for the construction of the subsequent Fisher discriminant function.

[0164] In step S05, according to the S B and S W calculated in step S04, the weight coefficients of each forecasting factor are determined through sample representativeness. Specifically, the Fisher discriminant criterion function is calculated: where J(ω) represents the discriminant criterion function; ω represents the discriminant vector; λ represents the regularization parameter. By optimizing and solving J(ω), the optimal weight coefficients of each forecasting factor are obtained.

[0165] The purpose of this step is to assign different weight coefficients according to the discrimination ability of each forecasting factor for foggy and non-foggy samples. In this way, the relative importance of each forecasting factor can be better reflected in the subsequent discriminant model.

[0166] In step S06, the extracted forecasting factors are randomly grouped multiple times using the cross-validation method. First, the samples are randomly divided into a training set and a validation set. Then, the within-class deviation cross-product sum is calculated based on the training set, and a Fisher discriminant function coefficient matrix containing a regularization term is established. Finally, the established discriminant function is tested using the validation set.

[0167] Through multiple random grouping cross-validations, the discriminant ability of each predictor can be evaluated more objectively, and a robust Fisher discriminant function coefficient matrix can be obtained. This lays a foundation for the construction of subsequent discriminant equations.

[0168] In step S07, based on the Fisher discriminant function coefficient matrix obtained in step S06, a polynomial discriminant equation system considering the non-linear relationship of predictors was constructed: Among them, f(x) represents the discriminant function; x i , x j represents the predictors; α ij , β i represents the coefficients; c represents the constant term; p represents the number of predictors.

[0169] By using the stepwise regression method, the optimal number of polynomial terms was selected, and the model complexity was reduced as much as possible while ensuring the discriminant accuracy. The finally obtained polynomial discriminant equation system can give respective sea fog prediction discriminant equations for different combinations of seasons and weather system types.

[0170] In step S08, first, based on the probability density functions of foggy samples and non-foggy samples, the optimal discriminant critical value of the sea fog prediction discriminant equation was calculated. Here, the kernel density estimation method was used to determine the probability density function: Among them, f h (x) represents the probability density function; K(·) represents the kernel function; h represents the bandwidth parameter; n represents the number of samples.

[0171] Then, the optimal discriminant critical value was determined by minimizing the weighted sum of the false negative probability and the false positive probability: T opt = argmin T {P miss (T) + αP false (T)}. Among them, T opt represents the optimal critical value; P miss represents the false negative probability; P false represents the false positive probability; α represents the weight coefficient.

[0172] Finally, when the calculation result of the discriminant equation is greater than this discriminant critical value, it is determined as foggy weather; when the calculation result is less than the discriminant critical value, it is determined as non-foggy weather. The purpose of this step is to determine the discriminant boundary between foggy and non-foggy.

[0173] In step S09, a three-level test system was used to verify the constructed sea fog prediction discriminant equation.

[0174] First, the accuracy of the prediction results was evaluated by using the contingency table test method. The calculation formula for the prediction accuracy rate is: Among them, H represents the number of hits; M represents the number of missed alarms; F represents the number of false alarms; CN represents the number of correct negatives.

[0175] Secondly, threat score and skill score are calculated to comprehensively evaluate the prediction performance. The calculation formula for the threat score is: The calculation formula for the skill score is: Among them, ACC r represents the accuracy of the random forecast.

[0176] Finally, through the prediction stability test, the consistency of the prediction results at different time periods is evaluated.

[0177] Through the above three-level tests, the performance and reliability of the constructed sea fog prediction discrimination model can be comprehensively evaluated, providing a basis for the release of subsequent prediction products.

[0178] In step S10, based on the sea fog prediction discrimination equation constructed in the foregoing steps, a comprehensive prediction product including the discrimination results of fog or no fog, the prediction of the fog intensity level, the time evolution prediction, and the spatial distribution prediction is compiled.

[0179] Among them, the fog intensity level prediction adopts an exponential function model: V = a·exp(-b·RH)+c·ΔT d +d. Among them, V represents visibility; RH represents relative humidity; a, b, c, d represent fitting coefficients.

[0180] The spatial distribution prediction adopts the method of Kriging interpolation, and its basic formula is: Among them, Z(x0) represents the estimated value of the point to be predicted; Z(x i ) represents the observed value of the known point; λ i represents the weight coefficient. The calculation of the weight coefficient needs to satisfy the following matrix equation: Among them, γ(h) represents the variogram; h ij represents the distance between point i and point j; μ represents the Lagrange multiplier.

[0181] The time evolution prediction adopts the form of partial differential equations: Among them, f represents the prediction element; represents the transport velocity; D represents the diffusion coefficient; S represents the source-sink term.

[0182] In addition, in order to reflect the uncertainty of the prediction results, the prediction products are also presented in the form of probability description. The specific formula is: Among them, P(f|x) represents the posterior probability; P(x|f) represents the likelihood function; P(f) represents the prior probability; P(x) represents the normalization factor.

[0183] Finally, the prediction model is dynamically optimized through a prediction correction mechanism. The correction formula is: f c = f0 + K(y - Hf0). Among them, f c represents the predicted value after correction; f0 represents the original predicted value; y represents the observed value; H represents the observation operator; K represents the gain matrix. The calculation formula of the gain matrix is: K = BH T (HBH T + R) -1 . Among them, B represents the background error covariance matrix; R represents the observation error covariance matrix.

[0184] The second aspect of the present invention provides a computer-readable storage medium, wherein program instructions are stored in the computer-readable storage medium, and when the program instructions run on a computer, they are used to execute the above-mentioned sea fog prediction method based on ocean station observation data.

[0185] The third aspect of the present invention provides a sea fog prediction system based on ocean station observation data, which includes the above-mentioned computer-readable storage medium.

[0186] Specifically, the principle of the present invention is as follows:

[0187] 1. Data preprocessing. First, obtain the continuous hydrometeorological observation data of the station for many years, including elements such as air temperature, dew point temperature, sea water temperature, wind speed and direction, air pressure, relative humidity, etc. Perform quality control and missing value supplementation on these observation data to ensure the integrity and reliability of the data, and lay a foundation for subsequent analysis.

[0188] 2. Multi-scale feature extraction. Use wavelet transform to perform multi-scale decomposition on time series data to obtain components at different scales such as interannual variation, seasonal variation, daily scale variation, and random perturbation. This multi-scale decomposition helps to better characterize the time variation characteristics of hydrometeorological elements and provides a basis for the screening of prediction factors.

[0189] 3. Prediction factor calculation. According to the multi-scale decomposition results, a series of prediction factors reflecting processes such as air-sea interaction, atmospheric stability, and water vapor transport are calculated. These factors include: the difference between the upstream dew point temperature and the local air temperature, the difference between the local air temperature and the dew point temperature, the difference between the local air temperature and the sea water temperature, the vertical gradient of the sea-air temperature difference, the stability index of the near-surface atmosphere, the convergence intensity of the water vapor flux, and the characteristic parameters of the inversion layer structure, etc.

[0190] 4. Construction of Classification Forecast Model. First, the samples are hierarchically divided according to seasons and weather system types (such as cold high-pressure type, warm high-pressure type, frontal type, etc.) to reflect the differential characteristics of sea fog formation under different environmental conditions. Then, based on the principle of Fisher discriminant analysis, the weighted between-class scatter matrix and within-class scatter matrix are calculated to determine the optimal weight coefficients of each forecast factor. Finally, a discriminant equation set considering non-linear relationships is constructed in polynomial form, and a sea fog forecast discriminant model is given for different seasons and weather types.

[0191] 5. Determination of Forecast Critical Value. By analyzing the probability density distributions of foggy samples and non-foggy samples, the optimal discriminant critical value is determined by minimizing the probability of missed alarms and false alarms. When the calculation result of the forecast model is higher than this critical value, it is determined as foggy weather; otherwise, it is determined as non-foggy weather.

[0192] 6. Verification and Revision of Forecast Results. A three-level forecast verification system including contingency table test, forecast skill assessment, and forecast stability test is established to comprehensively evaluate the performance of the constructed model. At the same time, through the forecast revision mechanism, the model is dynamically optimized in combination with real-time observation data to further improve the accuracy and reliability of the forecast.

[0193] 7. Generation of Forecast Products. Based on the above forecast model, comprehensive forecast products including foggy / non-foggy discrimination results, fog intensity forecast, time evolution forecast, and spatial distribution forecast are generated. Among them, the fog intensity forecast uses an exponential function model based on relative humidity and temperature difference; the time evolution forecast uses partial differential equations to describe the dynamic changes of fog; the spatial distribution forecast uses Kriging interpolation method to achieve regional forecast.

[0194] The following provides an embodiment of a specific application scenario of the present invention: A marine station in a coastal city has been facing the problem of low accuracy in sea fog forecasting for many years. To solve this problem, the station decides to adopt the sea fog forecasting method based on marine station observation data proposed by the present invention.

[0195] First, the station collected continuous observation data for the past 10 years, including elements such as air temperature, dew point temperature, sea water temperature, wind speed and direction, air pressure, relative humidity, etc. Table 1 lists some of the observation records on March 10, 2024.

[0196] Table 1 Observation Data of the Marine Meteorological Station on March 10, 2024

[0197] Time Wind direction (°) Wind speed (m / s) Water temperature (℃) Air temperature (℃) Humidity (%) Dew point temperature Atmospheric pressure (hPa) 0:00 249 0.8 4.6 0.9 84 -1.4937 1025.1 1:00 217 1.1 4.6 0.8 89 -0.8039 1025 2:00 253 0.8 4.6 -0.1 92 - 1024.8 3:00 233 1.4 4.5 0.9 95 0.19077 1024.5 4:00 212 1.9 4.6 0.9 92 -0.2508 1024.2 5:00 211 1.7 4.6 0.9 92 -0.2508 1024.3

[0198] For these observation data, quality control was first carried out. Taking the air temperature data as an example, the standardized residual Q of each time step was calculated i , and it was found that the data at 02:00 was abnormal (Q i(>3), so it was excluded. For the missing dew point temperature data, the inverse distance weighted method was used for interpolation. After this series of preprocessing, complete and reliable hydrometeorological observation data were obtained.

[0199] Next, wavelet transform was used to perform multi-scale decomposition on the observed data. Figure 3 The multi-scale decomposition results of the air temperature time series are shown. There are a total of 5 subgraphs, from top to bottom are: the original air temperature series, the seasonal and trend components, the synoptic scale component, the diurnal variation component, and the random noise component. Through wavelet transform, the original complex air temperature time series is decomposed into change characteristics on different time scales, revealing the multi-scale characteristics contained in the air temperature change. Among them, the seasonal and trend components reflect the long-term change trend of the air temperature; the synoptic scale component reflects the change of the synoptic system with a 3-7 day cycle; the diurnal variation component reflects the periodic change of the air temperature within a day; the random noise component represents the random fluctuations that are difficult to explain by deterministic laws. This multi-scale decomposition helps to extract key information from the complex changes of meteorological elements and provides important characteristic parameters for sea fog forecasting. Based on the multi-scale decomposition data, a series of forecast factors reflecting air-sea interaction were calculated at this station, as shown in Table 2.

[0200] Table 2 Calculation results of forecast factors

[0201]

[0202] With these forecast factors, the station then classified the observed data. First, it was divided into winter (December - February), spring (March - May), summer (June - August), autumn (September - November), and the spring-autumn transition period according to seasons. Then, it was further divided into cold high pressure type, warm high pressure type, and frontal type according to the synoptic system type. For the case on March 10th, it belongs to the spring and cold high pressure type weather.

[0203] Based on Fisher discriminant analysis, the station calculated the weighted between-class scatter matrix S of each forecast factor under the conditions of spring and cold high pressure type B and the within-class scatter matrix S W , and the results are shown in Table 3.

[0204] Table 3 Fisher discriminant analysis statistics

[0205]

[0206] Next, the station constructed a polynomial discriminant equation considering the non-linear relationship of the forecast factors:

[0207]

[0208] Through stepwise regression screening, the optimal number of polynomial terms was determined. Meanwhile, using the method of cross-validation, a robust discriminant function coefficient matrix was obtained. For the case on March 10th, the calculation result of the discriminant equation was 4.92.

[0209] To determine the discriminant critical value for foggy and non-foggy conditions, the station calculated the probability density functions of foggy samples and non-foggy samples respectively using the method of kernel density estimation based on historical observational data. Figure 4 It shows the probability density distribution of foggy samples and non-foggy samples in terms of discriminant function values. In the figure, the red curve represents the probability density distribution of foggy samples, while the blue curve represents the probability density distribution of non-foggy samples. The cross-region of the two curves indicates the situation where the forecast may be misjudged. The vertical black dashed line in the figure marks the optimal discriminant critical value of 4.75, which is determined by minimizing the weighted sum of the miss rate and false alarm rate. The discriminant value of 4.92 for the case on March 10th is also marked in the figure, slightly higher than the critical value, so it is judged as foggy weather. In addition, the false alarm region and miss region are clearly marked in the figure, corresponding to the probability regions where non-foggy is misjudged as foggy and foggy is misjudged as non-foggy respectively. This figure intuitively shows the performance of the discriminant model and the basis for its application in actual forecasting.

[0210] Since the calculation result (4.92) of the discriminant equation on March 10th is greater than the critical value (4.75), the station judged that day as foggy weather. To further evaluate the reliability of the forecast results, the station established a three-level inspection system:

[0211] 1. Contingency table test. According to the forecast results and actual observations from March 1st to March 20th, a contingency table as shown in Table 4 was established. From this, the forecast accuracy rate was calculated to be 85%, the threat score was 0.72, and the skill score was 0.68.

[0212] Table 4 Contingency table of forecast results (March 1st - March 20th)

[0213] Observed fog Observed no fog Forecast fog 42 7 Forecast no fog 5 36

[0214] 2. Forecast skill evaluation. In addition to the above overall evaluation indicators, the station also calculated the forecast accuracy rate, miss rate, and false alarm rate for different weather system types respectively, and the results are shown in Table 5. It can be seen that in the cold high-pressure type weather, the forecast performance is better, but there are certain false alarm problems in the frontal type weather.

[0215] Table 5 Forecast performance under different weather system types

[0216] Weather system type Forecast accuracy rate Omission rate False alarm rate Cold high pressure type 90％ 6％ 10％ Warm high pressure type 85％ 12％ 15％ Frontal type 80％ 15％ 20％

[0217] 3. Forecast stability test. The station conducted a time series analysis on the forecast results from March 1st to March 20th and found that the various indicators fluctuated slightly during different periods, indicating good stability of the forecast results.

[0218] Figure 5 It shows the comparison of the forecast skills of sea fog forecasts under three different weather system types (cold high pressure type, warm high pressure type, frontal type). Five evaluation indicators are presented in the form of bar charts in the figure: forecast accuracy rate, threat score, skill score, miss rate, and false alarm rate. It can be seen from the figure that the forecast performance is the best under the cold high pressure type weather, with a forecast accuracy rate of 0.90, threat scores and skill scores of 0.75 and 0.70 respectively, and miss rates and false alarm rates of only 0.06 and 0.10; while the forecast performance is relatively poor under the frontal type weather, with a forecast accuracy rate of 0.80, threat scores and skill scores of 0.60 and 0.58 respectively, and the miss rate and false alarm rate increase to 0.15 and 0.20. This result indicates that the sea fog forecasting method has different applicability to different weather system types, and the forecasting model needs to be further optimized under the frontal type weather system.

[0219] Based on the above three-level test results, the station is quite confident in the sea fog forecast conclusion on March 10th.

[0220] Therefore, the station issued the following forecast products:

[0221] 1. Fog forecast: There will be fog on March 10th, with a forecast accuracy rate of 85%.

[0222] 2. Fog intensity forecast: According to the exponential function model of relative humidity and temperature difference, it is estimated that the visibility will be between 2 - 3 km.

[0223] 3. Time evolution forecast: Based on the simulation of partial differential equations, it is estimated that the fog will gradually dissipate around 10:00 the next day.

[0224] 4. Spatial distribution forecast: Using Kriging interpolation method, it is estimated that sea fog with low visibility will occur in most coastal areas.

[0225] 5. Forecast uncertainty: Estimated through the Bayesian probability model, the posterior probability of the forecast result is 75%.

[0226] Furthermore, the station also conducted further analysis based on the specific fog-forming weather situation to improve the forecast accuracy rate, such as Figure 2 shown, including the following four fog-forming weather situations:

[0227] Offshore transformed high pressure type: When the high pressure system moves from land to the ocean, its airflow characteristics change, creating conditions suitable for fog formation. The cold air on land flows towards the warmer water surface, forming evaporation fog or advection fog.

[0228] Northward extension type of the subtropical high ridge: When the ridge line of the subtropical high extends northward, the airflow characteristics and temperature and humidity conditions at its edge are conducive to the formation of fog, especially in certain geographical locations.

[0229] Front part of cyclone type: There is often warm and humid airflow in the front part of the low-pressure system (cyclone), which meets the cold air on the ground and creates conditions suitable for the formation of fog.

[0230] Behind the ridge and in front of the trough type: In the area behind the high-pressure ridge and in front of the low-pressure trough, there are often airflow convergence and changes in temperature and humidity conditions, which are conducive to the formation of fog.

[0231] In subsequent monitoring, the station found that the forecast results were in good agreement with the actual observations, demonstrating good forecasting skills. Through dynamic correction, the station continuously optimized the forecasting model, further improving the accuracy and reliability of sea fog forecasting.

[0232] It should be noted that the explanations of relevant variables are shown in Table 6 below:

[0233] Table 6 Explanation of variables

[0234]

[0235]

[0236] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention.

Claims

1. A sea fog forecasting method based on ocean station observation data, characterized in that, Including: Obtaining multi - year continuous hydrometeorological observation data of the station; Performing multi - scale decomposition on the hydrometeorological observation data by using the wavelet transform method; Calculating prediction factors based on the multi - scale decomposition data; Dividing the time periods and performing secondary classification on the prediction factors to obtain hierarchical Fisher discriminant sample data; Calculating the scatter matrix based on the Fisher secondary discriminant principle; Establishing a Fisher discriminant function coefficient matrix by using the cross - validation method; constructing and solving a polynomial discriminant equation system to obtain a sea fog prediction discriminant equation; Calculating the discriminant critical value of the sea fog prediction discriminant equation; Verifying the sea fog prediction discriminant equation by using a three - level test system; Constructing a prediction product release system for release.

2. The sea fog forecasting method based on ocean station observation data according to claim 1, wherein, The hydrometeorological observation data includes the station air temperature value, the station dew - point temperature value, the upstream measuring point dew - point temperature value, the station surface seawater temperature value, the station wind direction and speed data, the station air pressure data, the station relative humidity data, and the synchronous observation data of adjacent stations. The sampling frequency of the synchronous observation data of adjacent stations is once per hour.

3. The sea fog forecasting method based on ocean station observation data according to claim 2, characterized in that, The multi - scale decomposition obtains an inter - annual variation component, a seasonal variation component, a daily - scale variation component, and a random perturbation component.

4. The sea fog forecasting method based on ocean station observation data according to claim 3, characterized in that The prediction factors include the difference between the upstream dew - point temperature and the local air temperature, the difference between the local air temperature and the dew - point temperature, the difference between the local air temperature and the seawater surface temperature, the vertical gradient of the sea - air temperature difference, the stability index of the near - surface atmosphere, the convergence intensity of water vapor flux, and the characteristic parameters of the inversion layer structure.

5. The sea fog forecasting method based on ocean station observation data according to claim 4, wherein The time - period division is carried out according to winter, spring, summer, autumn, and two transitional periods of spring and autumn. The secondary classification is based on the type of weather system, and the types of weather systems include cold - high - pressure type, warm - high - pressure type, and frontal type.

6. The sea fog forecasting method based on ocean station observation data according to claim 5, wherein The scatter matrix includes a weighted between - class scatter matrix and a within - class scatter matrix. The weighted between - class scatter matrix characterizes the degree of difference between the weighted mean values of samples in foggy weather and non - foggy weather. The within - class scatter matrix characterizes the degree of dispersion of samples within their respective categories.

7. The sea fog forecasting method based on the ocean station observation data according to claim 6, wherein The cross - validation method randomly groups the prediction factors multiple times, calculates the within - class scatter cross - product sum of the prediction factors, and establishes a Fisher discriminant function coefficient matrix including a regularization term.

8. The sea fog forecasting method based on ocean station observation data according to claim 7, wherein, The discriminant critical value of the sea fog prediction discriminant equation selects the optimal classification threshold by analyzing the probability distribution characteristics of foggy samples and non - foggy samples. When the calculation result of the discriminant equation is greater than the discriminant critical value, it is determined as foggy weather. When the calculation result of the discriminant equation is less than the discriminant critical value, it is determined as non - foggy weather.

9. A computer-readable storage medium, characterized in that, The computer - readable storage medium stores program instructions. When the program instructions run on a computer, they are used to execute a sea fog prediction method according to any one of claims 1 - 8 based on ocean station observation data.

10. A sea fog forecasting system based on ocean station observation data, characterized in that, Including the computer - readable storage medium according to claim 9.

Images