Marine phenomenon prediction method, device, equipment and storage medium

Abstract

This application provides a method, apparatus, device, and storage medium for forecasting marine phenomena. The method includes: acquiring marine meteorological forecast data for a future period; extracting forecast data for at least one key factor from the marine meteorological forecast data, wherein the key factor is associated with at least one target phenomenon; determining corresponding detection algorithms for the target phenomena associated with the key factor according to a preset mapping relationship; the preset mapping relationship indicating the association rules between the key factor and the detection algorithms for the target phenomena; and processing the marine meteorological forecast data using the detection algorithms determined for each target phenomenon to obtain a forecast result for at least one target phenomenon. The target phenomena include at least one of the following: mesoscale eddies, atmospheric waveguides, oceanic acoustic channels, and temperature strata. Thus, by extracting key factors and dynamically configuring algorithm parameters, forecast accuracy and reliability can be improved.

Description

Technical Field

[0001] This application relates to the field of marine phenomena technology, and more specifically, to a method, apparatus, device, and storage medium for forecasting marine phenomena. Background Technology

[0002] The ocean is an important component of the Earth's environment. Oceanic weather conditions influence people's daily lives and activities. Therefore, the observation and prediction of oceanic phenomena are crucial. Oceanic phenomena such as mesoscale eddies, atmospheric waveguides, oceanic acoustic channels, and thermoclines have significant impacts on maritime navigation, resource development, and environmental monitoring. Existing forecasting technologies are mainly based on physical diagnostic models with fixed algorithms and thresholds. Their parameters are fixed and difficult to adapt to the complex and ever-changing air-sea environment, with performance degrading under abnormal or marginal conditions. How to accurately and efficiently forecast oceanic phenomena is a topic of great concern in this field. Summary of the Invention

[0003] This application aims to provide a method, apparatus, device, and storage medium for forecasting marine phenomena, with the goal of accurately and efficiently forecasting marine phenomena. The following scheme is adopted in this application.

[0004] In a first aspect, embodiments of this application provide a method for forecasting marine phenomena. The method includes: acquiring marine meteorological forecast data for a future period; extracting forecast data of at least one key factor characterizing the state of the marine environment from the marine meteorological forecast data, wherein the key factor is associated with at least one target phenomenon; determining corresponding detection algorithms for each of the at least one target phenomenon associated with the key factor according to a preset mapping relationship; wherein the preset mapping relationship indicates the association rules between the key factor and the detection algorithms for the target phenomenon; and processing the marine meteorological forecast data using the detection algorithms determined for each target phenomenon to obtain a forecast result for at least one target phenomenon. In embodiments of this application, the target phenomena include at least one of the following: mesoscale eddies, atmospheric waveguides, oceanic acoustic channels, and temperature strata.

[0005] In some embodiments, association rules indicate the association between the physical type of a key factor and the type of detection algorithm.

[0006] Secondly, embodiments of this application also provide a marine phenomenon forecasting device, comprising: an acquisition module for acquiring marine meteorological forecast data for future periods; a processing module for extracting forecast data of at least one key factor characterizing the state of the marine environment from the marine meteorological forecast data acquired by the acquisition module, wherein the key factor is associated with at least one target phenomenon, and the target phenomenon includes at least one of the following: mesoscale eddies, atmospheric waveguides, oceanic acoustic channels, and temperature strata; the processing module is further configured to determine corresponding detection algorithms for the at least one target phenomenon associated with the key factor according to a preset mapping relationship; wherein the preset mapping relationship indicates the association rules between the key factor and the detection algorithms of the target phenomenon; the processing module is further configured to process the marine meteorological forecast data using the detection algorithms determined for each target phenomenon to obtain a forecast result for at least one target phenomenon.

[0007] Thirdly, embodiments of this application also provide an electronic device, including: a processor and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method in any possible implementation of the first aspect.

[0008] Fourthly, embodiments of this application also provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the method in any possible implementation of the first aspect described above.

[0009] The marine phenomenon forecasting method, apparatus, equipment, and storage medium provided in this application acquire marine meteorological forecast data for future periods, extract key factors from the data, and perform marine meteorological forecasts based on the adaptive mapping relationship between the key factors and the detection algorithm. This improves forecast accuracy and reliability. Furthermore, this scheme supports parallel processing of multiple phenomena, enabling the analysis and forecasting of various target phenomena such as mesoscale eddies, atmospheric waveguides, ocean acoustic channels, and temperature strata from the same forecast data, significantly improving computational efficiency and product consistency. Attached Figure Description

[0010] Figure 1 This is a schematic diagram illustrating an application scenario of a marine phenomenon forecasting method provided in an embodiment of this application.

[0011] Figure 2 A flowchart of a marine phenomenon forecasting method provided in this application embodiment;

[0012] Figure 3 A schematic diagram of the module structure of a marine phenomenon forecasting device provided in this application embodiment;

[0013] Figure 4This is a hardware structure diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0014] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0015] The exemplary embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the exemplary embodiments described herein are for illustration and explanation only and are not intended to limit this application. Furthermore, the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0016] In some embodiments, this application provides an intelligent forecasting system for marine meteorological phenomena. This system is a comprehensive platform integrating multi-source data fusion, intelligent key factor screening, adaptive algorithm mapping, and collaborative forecasting of multiple phenomena. The system collects key environmental factors in real time through intelligent sensing terminal devices deployed at the ocean-atmosphere interface and in the deep atmosphere. Combined with reanalysis data and multi-source data such as satellite remote sensing, it utilizes a key factor-driven adaptive algorithm library to automatically identify and intelligently forecast various marine meteorological phenomena, including mesoscale eddies, atmospheric waveguides, ocean acoustic channels, and temperature strata, serving fields such as marine engineering, shipping safety, and environmental monitoring.

[0017] The intelligent forecasting system for marine meteorological phenomena provided in this application has the following characteristics: multi-source data fusion: integrating on-site observations, reanalysis data, and satellite remote sensing data to form a three-dimensional monitoring network covering the air-sea interface; multi-phenomenon collaborative forecasting: outputting forecast results for multiple phenomena such as mesoscale eddies, atmospheric waveguides, ocean acoustic channels, and temperature strata in parallel based on the same set of forecast data; adaptive intelligent decision-making: achieving dynamic adaptive forecasting that varies "depending on the environment" through key factor-driven algorithm selection; operationalization: supporting full-process automation from data access to product release, with high timeliness and stability.

[0018] refer to Figure 1 The intelligent forecasting system 10 may include a terminal perception layer 110, a data transmission and storage layer 120, a core algorithm engine layer 130, and an application service layer 140.

[0019] 1. The terminal perception layer 110 may include the following types of intelligent sensing terminal devices.

[0020] Ocean observation array 111: includes profile buoys equipped with temperature, salinity, and depth sensors, current meter arrays, deep-sea moorings, etc., used to acquire real-time vertical profile data such as seawater temperature, salinity, current velocity, and current direction; Atmospheric observation array 112: includes meteorological buoys, shore-based meteorological stations, radiosondes, etc., used to measure atmospheric boundary layer elements such as sea surface pressure, air temperature, humidity, wind speed, and wind direction; Satellite data receiving terminal 113: used to receive remote sensing data such as sea surface height anomalies, sea surface geostrophic current velocity, sea surface temperature, and ocean color; Communication module 114: transmits real-time data to the central server via satellite, 4G / 5G, or underwater acoustic communication networks.

[0021] 2. The data transmission and storage layer 120 may include the following devices.

[0022] Edge Gateway 121: Performs preliminary cleaning, outlier removal, time alignment, and format standardization on raw data; Cloud Server Cluster 122: Receives and stores multi-source data, including real-time sensor data, reanalysis grid data (such as AVISO, CORA, ERA5), satellite inversion products, and historical phenomenon tag libraries; Database System 123: Uses a time-series database to store high-frequency sensor data, a spatial database to manage gridded environmental field and phenomenon vector data (such as vortex boundaries and waveguide layers), and a relational database to store mapping rules and configuration parameters.

[0023] 3. The core algorithm engine layer 130 may include the following modules, etc.

[0024] Key Factor Screening Module 131: Based on historical data, this module calculates the spatial distribution of correlation metrics between candidate factors (such as sea surface height anomalies, vertical temperature gradients, and velocity shear) and target phenomena (mesoscale eddies, atmospheric waveguides, ocean acoustic channels, and temperature strata). It then selects key factors from the candidate factors that have a clear indicative role for various phenomena and stores them in the knowledge base. Adaptive Mapping Rule Base 132: This module stores the preset mapping relationships between key factors and detection algorithms, such as an algorithm mapping table based on the physical type of the key factors. Multi-Phenomenon Collaborative Forecasting Engine 133: As the core decision-making and execution unit of the system, this engine receives real-time forecast data, extracts the forecast values ​​of key factors, queries the adaptive mapping rule base 132 to determine the corresponding detection algorithm for each activated key factor associated with the target phenomenon, calls the detection algorithm, processes the forecast data, and generates the detection results for the target phenomenon. Optionally, the forecast results can be compared with subsequently acquired actual observation data to calculate forecast performance indicators. When the performance falls below a threshold, the system automatically triggers fine-tuning and optimization of the parameter configuration rules for the corresponding key factors and writes the updated rules back to the adaptive mapping rule base to update the system.

[0025] 4. The application service layer 140 may include the following modules.

[0026] Integrated Forecast Visualization Platform 141: Provides users with an interactive interface to display various forecast products generated by the core algorithm engine layer, including: mesoscale eddy distribution maps (including eddy core, radius, and boundary), atmospheric waveguide profile maps (including waveguide layer and intensity), ocean acoustic channel axis depth maps, temperature gradient depth distribution maps, and comprehensive situation maps with multiple phenomena overlaid; supports custom regional queries, time-sliding forecasts, and threshold warning settings. API Data Service 142: Provides standardized data interfaces for third-party systems, allowing access to data including forecast result fields of various phenomena, key factor environmental fields, historical frequency statistics, etc., supporting integration with GIS systems, professional models, and mobile applications.

[0027] As an example, the workflow of the aforementioned intelligent forecasting system 10 for marine meteorological phenomena is as follows.

[0028] The first step is data aggregation. Intelligent sensing terminal devices upload observation data at a preset frequency (e.g., every 6 hours), and the system synchronously updates reanalysis data and satellite remote sensing data to form environmental field data covering the target sea area. The second step is key factor identification. Predicted values ​​of key factors (such as sea surface height anomaly gradient areas and atmospheric temperature inversion layers) are extracted from the real-time environmental field. The third step is adaptive decision-making: the core algorithm engine queries the adaptive mapping rule base and determines the detection algorithm for each target phenomenon based on the extracted key factor type and intensity. The fourth step is target phenomenon detection. If there are multiple target phenomena to be detected, multiple pre-configured detection algorithms (each corresponding to a specific target phenomenon) can be called in parallel to process the same environmental field data, outputting preliminary detection results for mesoscale eddies, atmospheric waveguides, ocean acoustic channels, and temperature strata. Optionally, the following steps may also be included. The fifth step is result optimization and fusion: spatial smoothing, quality control, and confidence assessment are performed on the preliminary results to generate multi-phenomenon collaborative forecast products. The fifth step is visualization and release, automatically generating a comprehensive forecast map, marking the location, intensity, and key characteristics of various phenomena, and overlaying warning icons in high-risk areas (such as vortex active areas and strong waveguide areas). The sixth step is warning push, sending customized warning information to subscribed users when the forecast result of any phenomenon in a certain area exceeds the preset risk threshold.

[0029] Optionally, the forecast results can be compared with the actual observations periodically. When the performance of the algorithm associated with a certain type of key factor continues to decline, the parameter rule optimization is automatically triggered and the mapping rule base is updated.

[0030] The marine phenomenon forecasting method provided by exemplary embodiments of this application will be described below with reference to the accompanying drawings and the above application scenarios. It should be noted that the above application scenarios are shown only to facilitate understanding of the spirit and principles of this application, and the embodiments of this application are not limited in any way.

[0031] like Figure 2 As shown in the embodiments of this application, the marine phenomenon forecasting method includes the following steps.

[0032] 210. Obtain marine meteorological forecast data for future periods.

[0033] In this embodiment of the application, the future time period can be a period in which marine weather forecasting is required.

[0034] 220. Extract forecast data of at least one key factor characterizing the state of the marine environment from marine meteorological forecast data. The key factor is associated with at least one target phenomenon, which includes at least one of the following: mesoscale eddies, atmospheric waveguides, ocean acoustic channels, and temperature strata.

[0035] In some embodiments, the key factors associated with different target phenomena may be different. The following sections will explain the basic approach to selecting key factors in light of different target phenomena.

[0036] I. Atmospheric Waveguides; Atmospheric waveguides are a common electromagnetic anomaly in marine and near-Earth atmospheric environments, significantly altering the propagation path of electromagnetic waves. Based on formation conditions and typical characteristics, atmospheric waveguides are generally classified into three types: evaporation waveguides, surface waveguides, and rise waveguides. From a physical mechanism perspective, the formation of atmospheric waveguides is closely related to the anomalous variation of atmospheric refractive index with altitude. The refractive index is mainly determined by the distribution of temperature, humidity, and air pressure. When these factors exhibit an anomalous gradient in the vertical direction, a stable waveguide structure may form. The formation and evolution of atmospheric waveguides are influenced by a variety of atmospheric environmental factors, and their essence is closely related to the spatial distribution of atmospheric refractive index. The refractive index is controlled by thermodynamic conditions such as temperature, humidity, and air pressure, and is also modulated by wind transport and large-scale circulation characteristics. Therefore, when studying the mechanism and analyzing the distribution characteristics of atmospheric waveguides, it is necessary to comprehensively consider relevant factors at the thermodynamic, dynamic, and circulation levels. In this application, six factors were selected as candidate factors: relative humidity, temperature, U-component wind speed, V-component wind speed, total wind speed, and geopotential height. Key factors were then selected from these.

[0037] (1) Relative humidity. Relative humidity is a key indicator reflecting the water vapor content and saturation level in the atmosphere. The formation of evaporation waveguides mainly depends on the vertical humidity gradient caused by sea surface evaporation, so relative humidity has a direct impact on the refractive index profile. When sea surface evaporation is vigorous and the humidity in the lower layer drops sharply, a strong evaporation waveguide layer is easily formed. Studying the changes in relative humidity not only helps to understand the intensity and thickness characteristics of evaporation waveguides, but also reveals their distribution patterns under different seasons and meteorological conditions.

[0038] (2) Temperature. Temperature is an important controlling factor for atmospheric stratification and stability. The formation of waveguide layers is usually closely related to temperature inversion, that is, the air temperature increases with altitude, resulting in an anomalous distribution of the refractive index gradient. The typical characteristics of surface waveguides and uplift waveguides are caused by this temperature inversion layer. Therefore, the temperature factor plays a core role in revealing the atmospheric thermal structure, determining the conditions for waveguide formation, and assessing waveguide strength.

[0039] (3) The U-component and V-component wind speeds, the horizontal components of the wind field (U for east-west and V for north-south), can characterize the directionality and transport effect of atmospheric motion. Wind fields in different directions not only change the energy and water vapor exchange at the air-sea interface, but may also change the distribution structure of temperature and humidity through horizontal advection, thereby affecting the position and thickness of the waveguide layer. For example, the sea-land wind circulation commonly seen in coastal areas can create significant temperature and humidity differences locally, which is conducive to the formation of evaporation waveguides or surface waveguides. Therefore, using the U-component and V-component wind speeds as factors can better analyze the regulatory effect of atmospheric dynamic processes on waveguides.

[0040] (4) Total wind speed. Total wind speed comprehensively reflects the strength of atmospheric dynamic activity. Stronger wind speeds tend to enhance atmospheric turbulent mixing and disrupt the stability of stratification, thus hindering the maintenance of the waveguide layer. Under weak wind conditions, atmospheric stratification is more likely to remain stable, and temperature inversion layers and humidity anomalies are more likely to form and persist for a longer period. The total wind speed factor can help identify the probability of occurrence and persistence characteristics of atmospheric waveguides under different wind field conditions and is an important supplementary indicator of the dynamic background.

[0041] (5) Potential height: Potential height is one of the fundamental variables characterizing atmospheric circulation and pressure field distribution. It reflects the dynamic background of large-scale atmospheric systems and has an overall regulatory effect on the distribution of temperature, humidity, and wind fields. When studying uplifted waveguides, potential height can effectively reveal the altitude range of the waveguide layer and its changing trend with the evolution of weather systems. Therefore, the potential height factor has irreplaceable value for understanding the distribution law and evolution mechanism of waveguides from the perspective of large-scale circulation background.

[0042] In summary, relative humidity and temperature represent atmospheric thermodynamic conditions, U and V component wind speeds and total wind speed reveal the dynamic characteristics of the atmosphere, while geopotential height reflects the large-scale circulation background. These six factors systematically cover the formation mechanism and evolution conditions of atmospheric waveguides from three levels: thermodynamic, dynamic, and circulation. They can reveal the mechanisms of local small-scale processes (such as sea surface evaporation and local temperature inversions) and reflect the moderating effect of large-scale background fields (such as circulation patterns and seasonal variations). Therefore, selecting these six factors as research objects can comprehensively and systematically characterize the formation environment and variation patterns of atmospheric waveguides, providing a scientific basis for subsequent mechanism analysis and forecast modeling.

[0043] II. Oceanic Mesoscale Eddies; Mesoscale eddies are typical mesoscale circulation structures in the ocean. Mesoscale eddies are generally classified into two types: anticyclonic eddies (ACEs) and cyclonic eddies (CEs). The former corresponds to the high sea surface height (SLA) center and is characterized by subsidence and deepening of the thermocline, while the latter is the opposite. Their formation and evolution are driven by a combination of mechanisms, including ocean thermal state, density stratification structure, wind stress excitation, Coriolis modulation, and atmosphere-ocean interactions. The following factors are selected as the basis for the study of mesoscale eddies in the embodiments of this application.

[0044] (1) Eastward geostrophic velocity anomalies are one of the core variables describing the dynamic characteristics of mesoscale eddies. Eastward geostrophic velocity (zonal velocity of geostrophic flow) is usually used to identify the rotation direction and intensity of eddies, especially in the western boundary flow region, where its anomalous changes are often highly correlated with eddy generation, merging, or dissipation processes. Through comparative analysis of multi-year monthly mean fields, eastward geostrophic velocity anomalies can serve as a bridge between circulation changes and mesoscale structural evolution.

[0045] (2) The meridional velocity of the geostrophic flow (meridional geostrophic velocity) is an indicator of the latitudinal velocity anomaly of the geostrophic flow, which has an indicative role in the horizontal propagation path of the vortex. In regions such as the South China Sea and the East China Sea, the anomalous evolution of the meridional velocity of the geostrophic flow directly reflects the vortex's movement speed and morphological changes, which is of great significance for vortex tracking and life cycle identification.

[0046] (3) Sea surface height anomaly (SLA) is one of the most widely used vortex identification indicators, which can directly reflect the disturbance of the water column mass field. Low SLA regions often represent anticyclonic vortices (warm vortices), while high SLA regions often correspond to cyclonic vortices (cold vortices).

[0047] (4) Sea surface temperature (SST) is a key variable affecting the thermal structure of mesoscale eddies. SST can reveal the distribution of thermal components in eddies, such as cold eddies causing a decrease in sea surface temperature and warm eddies exhibiting a high-temperature core region. Remote sensing SST data has become an important tool for monitoring eddy intensity, heat flux exchange, and eddy-air coupling.

[0048] (5) Surface salinity and temperature together determine the hydrostatic stability of seawater, which is an important physical basis for mesoscale eddy stratification. Recent studies have shown that eddies can influence water buoyancy and local stratification by regulating surface salinity, thereby controlling their vertical development capacity.

[0049] (6) The eastward flow velocity u is a direct dynamic force value that reflects the local vortex rotation and energy transfer. The u component data can be used for vortex trajectory inversion, kinetic energy estimation and non-geostrophic process identification, and is often used in vortex structure identification.

[0050] (7) The northward velocity v is also crucial in assessing the propagation path and kinetic energy transfer of mesoscale eddies. The combination of the northward velocity and the meridional velocity of the geostrophic flow can jointly construct the eddy plane velocity field, which helps to reconstruct the eddy's direction of rotation, intensity, and degree of boundary closure.

[0051] (8) Vertical velocity; There is significant vertical disturbance inside mesoscale eddies, especially during merging or breaking up, where the vertical velocity (w component) can reach several mm / s. Vertical velocity helps to determine the influence of eddies on the transport of nutrients, thermohaline transport and water mass mixing in the upper layer, and is an important supplementary indicator for identifying the intensity of eddy activity.

[0052] (9) Sea surface pressure is an important variable reflecting wind stress, cyclone activity and atmosphere-ocean coupling. During typhoons or strong wind events, pressure changes can induce sea surface rise / fall, affecting the energy input and structural evolution of mesoscale eddies.

[0053] III. Thermoclines. A thermocline is a layer of water in the ocean whose temperature changes drastically with depth, typically located between the surface warm water layer and the deep cold water layer. Based on their formation and duration, thermoclines can be divided into seasonal thermoclines and permanent thermoclines.

[0054] The formation and evolution of the thermocline are directly controlled by solar radiation and air-sea exchange, and are also influenced by atmospheric dynamic processes and regional thermal conditions. To more comprehensively reveal the physical mechanisms of the thermocline, this application selected six factors as research objects: sea level pressure (MSL), surface temperature (SKT), sea surface temperature (SST), air temperature at 2 meters (T2M), east-west wind speed at 10 meters (U10), and north-south wind speed at 10 meters (V10). The reasons for selecting these factors are as follows.

[0055] (1) Sea level pressure (MSL): Sea level pressure reflects the atmospheric circulation pattern and is an important background factor that determines wind field distribution and air-sea exchange. Changes in the pressure field indirectly affect the thickness of the ocean mixing layer and the depth of the thermocline by regulating wind stress and sea surface convergence and divergence processes.

[0056] (2) Surface temperature (SKT); Surface temperature includes temperature information of land surface and ocean surface, which can characterize the regional thermal balance. The temperature difference between land and ocean drives regional circulation (such as sea breeze), which in turn regulates the heat distribution and mixing intensity in nearshore areas, thus affecting the formation and location of the thermocline.

[0057] (3) Sea surface temperature (SST); SST is a direct factor that determines the thermal state of surface seawater. Higher SST will enhance the temperature difference between the surface and deep layers, thus forming a more significant thermocline; while a decrease in SST may weaken the thermocline.

[0058] (4) 2-meter air temperature (T2M); T2M represents the near-surface air temperature and is an important driving factor for air-sea heat exchange. The temperature difference between T2M and SST determines the strength of the sensible heat flux between the air and sea, thus affecting the rate of warming or cooling of the surface seawater.

[0059] (5) 10-meter east-west wind speed (U10) and north-south wind speed (V10); U10 and V10 together constitute the sea surface wind field, which is the core factor affecting the dynamic processes of the surface ocean. Wind stress affects the thickness of the ocean mixing layer through turbulent mixing, thereby controlling the depth and gradient of the thermocline. In addition, changes in wind direction may lead to surface circulation and upwelling processes, which play an important role in regulating the regional distribution and spatiotemporal variation of the thermocline.

[0060] In summary, MSL, SKT, SST, T2M, U10, and V10, from different perspectives such as large-scale circulation background (MSL), regional thermal conditions (SKT, T2M), ocean surface state (SST), and dynamic processes (U10, V10), together constitute candidate factors influencing the formation and evolution of the thermocline.

[0061] IV. Sound Channel. Ocean sound channel thickness refers to the depth range within the sound velocity profile where an effective waveguide structure forms. Especially in environments with a positive sound velocity gradient, thickness directly affects the sound wave propagation distance and energy attenuation level. The formation and evolution of this structure are regulated by multiple factors, including ocean thermal structure, hydrodynamic processes, and atmospheric coupling mechanisms. To study the characteristics of ocean sound channel thickness variation and key regulatory factors, this application selects the following factors as candidate factors.

[0062] (1) Temperature is one of the main controlling factors affecting the distribution of sound velocity, and its rate of change in the vertical direction directly determines the sound velocity gradient structure. When the surface temperature is significantly higher than the bottom temperature, especially during the summer when the layer is strengthened, the sound velocity increases with depth to form a stable positive gradient structure, thus determining the existence and thickness range of the sound channel. Studies have found that the strength and location of the thermocline are closely related to the height of the top / bottom of the sound channel.

[0063] (2) Although the effect of salinity on sound speed is less than that of temperature, in high latitudes or estuaries, the halocline and temperature structure work together to affect the change in sound speed profile. A strong halocline may enhance the positive gradient structure on the surface or counteract the thermocline effect at the bottom, thus affecting the range and intensity of the sound channel thickness.

[0064] (3) Vertical shearing of meridional velocity may trigger redistribution of thermohaline structure, indirectly altering the stratification of sound velocity field. Especially in mid-to-high latitude sea areas or near fronts, strong north-south shearing is often accompanied by stratification disturbance, which makes the sound channel structure unstable or causes drastic changes in thickness.

[0065] (4) Zonal current velocity (such as eastward ocean current) shear can adjust the thickness of the surface mixed layer and its thermal structure through compression / stretching, thereby affecting the stability of the sound velocity profile.

[0066] (5) Vertical flow velocity reflects the existence of upwelling, downwelling or internal wave processes, which can directly disrupt or strengthen thermo-salt stratification, thereby changing the sound velocity gradient distribution.

[0067] (6) Geostrophic anomalies reflect large-scale circulation, vortex and other dynamic disturbances, which change the thermal structure and baroclinic stability of seawater, thereby regulating the sound velocity distribution and the thickness of the sound channel.

[0068] (7) Abnormal northward geostrophic flow can also cause disturbances in thermohaline structures.

[0069] (8) The atmospheric pressure field indirectly affects the thickness of the surface mixing layer and the structure of the thermocline by regulating the wind field structure, thereby affecting the change of the sound speed profile.

[0070] In some embodiments, for any target phenomenon, key factors can be selected in the following manner: for any target phenomenon, obtain historical marine meteorological data associated with the target phenomenon; based on the historical marine meteorological data, determine multiple candidate factors associated with the target phenomenon; and based on the correlation between each candidate factor and the target phenomenon, select key factors corresponding to the target phenomenon from the multiple candidate factors.

[0071] For example, if the target phenomenon is a mesoscale eddy, the historical marine meteorological data associated with the mesoscale eddy comes from AVISO satellite altimeter products (e.g., including sea level anomalies and geostrophic flow velocity), CORA ocean reanalysis data (e.g., including temperature, salinity, and three-dimensional flow velocity), and ERA5 atmospheric reanalysis data (e.g., including sea level pressure). The candidate factors cover elements such as sea level anomalies, geostrophic flow velocity, temperature, salinity, three-dimensional flow velocity, and sea level pressure.

[0072] If the target phenomenon is atmospheric waveguide, the historical data associated with atmospheric waveguide comes from the multi-layer meteorological element field of ERA5 reanalysis data. Candidate factors include vertical difference of relative humidity, vertical difference of temperature, vertical difference of wind speed, zonal wind speed difference, meridional wind speed difference, and geopotential height difference.

[0073] If the target phenomenon is oceanic acoustic channels, the historical data associated with oceanic acoustic channels are derived from CORA ocean temperature, salinity and current data, AVISO sea surface current field data, and ERA5 meteorological and atmospheric pressure data. In addition, combined with acoustic channel thickness table data, candidate factors include temperature, salinity, three-dimensional velocity, sea surface geostrophic velocity, sea surface height anomaly, and sea level pressure.

[0074] If the target phenomenon is a temperature stratus, the historical data associated with the temperature stratus are derived from CORA ocean reanalysis data, and the candidate factors mainly include vertical temperature profile data. Optionally, sea surface temperature, wind stress, and sea surface height anomalies can also be included as candidate factors.

[0075] In some embodiments, the correlation between candidate factors and target phenomena can be determined as follows: For any candidate factor, historical observation data of the candidate factor at various spatiotemporal locations and historical information on the occurrence status of the target phenomenon at those locations can be determined. This historical information on occurrence status indicates whether the target phenomenon occurred at that spatiotemporal location. Furthermore, based on the historical observation data and the historical information on occurrence status, a spatial distribution of the correlation metric between the candidate factor and the target phenomenon can be generated. This spatial distribution characterizes the spatial variation of the correlation strength index between the candidate factor and the target phenomenon and the statistical significance of the correlation strength index.

[0076] For example, for mesoscale eddies, firstly, AVISO, CORA, and ERA5 multi-source data from the same day are uniformly interpolated onto the AVISO grid. Then, using the contour coordinates from the daily mesoscale eddy identification results, a binary field of mesoscale eddy occurrence is generated as the dependent variable through polygon rasterization. Next, within the study area, based on the AVISO grid, for each marine grid point, a KNN combined with an adaptive radius strategy is used to collect environmental factor values ​​and the binary field of mesoscale eddy occurrence from surrounding grid points. The Pearson correlation coefficient r and significance level p are calculated, ultimately generating a spatial distribution map of the correlation coefficient between each candidate factor and the probability of mesoscale eddy occurrence, which is the spatial distribution of the correlation metric in this embodiment.

[0077] For atmospheric waveguides, multi-layer meteorological element data are first read from ERA5 reanalysis data. A spatiotemporal matching algorithm is used to precisely align the meteorological data with waveguide presence labels, and vertical profile data of meteorological elements at corresponding locations are extracted to construct a labeled dataset. Then, the element differences between adjacent layers are calculated as candidate features among the seven standard pressure layers. Kernel density estimation is used to calculate the probability distribution of positive samples (waveguides present) and negative samples (waveguides absent) at each vertical difference. The distribution differences are quantified using statistical indicators such as KL divergence, t-test, and effect size, generating a correlation metric distribution between each vertical difference feature and waveguide presence.

[0078] For the ocean acoustic duct, CORA ocean data, AVISO sea surface current data, and ERA5 meteorological data are first used to construct a spatial index for efficient spatial point matching. For each acoustic duct point (latitude and longitude), the nearest neighbor grid point is found in the CORA and ERA5 grids, and the corresponding point is located in the AVISO grid using the minimum distance method. Then, linear regression analysis is performed on the preprocessed data of each element and the acoustic duct thickness. The goodness of fit is evaluated by solving the regression coefficient and the coefficient of determination R², generating a linear correlation measure between each element and the acoustic duct thickness. Alternatively, the correlation coefficient field and significance p-value between each element and the acoustic duct thickness can be calculated using the Pearson correlation coefficient, generating the spatial distribution of the correlation measure.

[0079] For the thermocline, the time series of meteorological elements are first spatiotemporally matched with the time series of thermocline depth. Then, the two input multidimensional arrays are flattened into one-dimensional arrays, invalid samples are eliminated by masking, and the Pearson correlation coefficient calculation function is called to calculate the correlation coefficient r and significance level p between meteorological elements and thermocline depth, with a valid sample size of no less than 2. This generates a spatial distribution map of the correlation between each meteorological element and thermocline depth.

[0080] Subsequently, candidate factors that meet preset conditions can be identified as key factors. For example, the preset conditions include: there is a spatially continuous region in the spatial distribution of the correlation metric corresponding to the candidate factor; within the continuous region, the correlation strength index reaches a preset strength threshold; and the proportion of grid points with statistical significance of the correlation strength index exceeds a preset proportion threshold.

[0081] For example, for mesoscale eddies, based on the calculated correlation coefficient r-map, IDW interpolation is performed on the missing values ​​in the sea area, and Gaussian smoothing can be used to improve spatial continuity. Then, it is visualized as a regional correlation distribution map and marked with significant point markers. Candidate factors that exist in continuous strongly correlated regions (e.g., |r|≥0.3) within the study sea area and whose proportion of grid points passing the significance test (p≤0.05) in the region exceeds a preset threshold (e.g., 50%) are selected as key factors.

[0082] For atmospheric waveguides, the probability distribution patterns of positive samples (with waveguides present) and negative samples (without waveguides) at various vertical differences are visualized and compared to identify vertical difference features with significant distribution differences. A triple screening criterion is used: KL divergence greater than 0.5, mean difference exceeding 2 standard deviations, and a clear physical interpretation mechanism. Vertical difference features meeting these criteria are identified as key factors, and their corresponding meteorological elements, vertical levels, and statistical significance indicators are recorded and sorted by influence intensity.

[0083] For marine acoustic tracts, the R² value of the determination coefficient obtained from linear regression analysis is used to evaluate the explanatory power of each candidate factor for the variation in acoustic tract thickness. Factors with high R² values ​​and passing the significance test are selected as key factors. At the same time, the correlation coefficient field (r-map) calculated by Pearson correlation coefficient and the p value estimate are combined to select candidate factors that have continuous significant correlation (p≤0.05) in the study area and whose absolute value of the correlation coefficient reaches the preset threshold (e.g., |r|≥0.3) as key factors.

[0084] For the temperature gradient, the spatial distribution map of the correlation calculated based on the Pearson correlation coefficient is used to visually display the positive and negative correlations using the RdBu_r color scale. Candidate factors that have a continuous strong correlation region (e.g., |r|≥0.3) in the study area and whose proportion of grid points that pass the significance test (p≤0.05) exceeds the preset threshold are selected as key factors.

[0085] 230. Based on the preset mapping relationship, determine the corresponding detection algorithm for at least one target phenomenon associated with the key factor.

[0086] In this embodiment of the application, the preset mapping relationship indicates the association rules between key factors and the detection algorithm of the target phenomenon.

[0087] In some embodiments, the association rule indicates the association between the physical type of a key factor and the type of detection algorithm.

[0088] In some embodiments, the preset mapping relationship indicates at least one of the following association rules.

[0089] For example, key factors include ocean dynamic key factors, such as sea surface height field or current vorticity field; correspondingly, the detection algorithm for this ocean dynamic key factor includes a mesoscale eddy detection algorithm based on the geometric identification of closed contour lines of sea surface height anomalies.

[0090] In some embodiments, the velocity gradient tensor is first calculated using the daily sea surface zonal current field (U / V) output by the ocean numerical model, and then the Okubo-Weiss (OW) parameter field is solved. According to the definition of the OW parameter, when it is less than a certain negative threshold, it indicates that the region is a vortex-dominated region; by combining empirical threshold screening and regional connectivity analysis, the spatial distribution information of mesoscale vortices can be extracted.

[0091] Building upon this foundation, to achieve probabilistic forecasting of vortices, daily forecast fields can be used as input to perform sliding window calculations on the Open Water (OW) parameters and establish a time-series raster dataset. By analyzing the correspondence between changes in the OW index in historical samples and observed / reconstructed vortex events, a mapping model between the occurrence probability of various vortices (cold vortices, warm vortices, weak vortices) and the OW value distribution is statistically established. Ultimately, a daily or hourly "vortex probability field" can be output, reflecting the likelihood of vortices occurring in a specific area within the future forecast lead time.

[0092] In some embodiments, for mesoscale eddy phenomena, the process of determining the detection algorithm based on a preset mapping relationship is as follows: When the selected key factors are key factors of ocean dynamics (such as sea surface height anomalies, geostrophic velocity, and flow field vorticity), the preset mapping relationship maps them to a mesoscale eddy detection algorithm based on the geometric identification of closed contour lines of sea surface height anomalies. The core process of this detection algorithm includes: reading AVISO gridded sea surface height anomaly data, using a regular grid and contour line generation method to generate all candidate closed contours within a given contour line range and step size, and then performing circle fitting on each contour to extract the center position, radius, and fitting error.

[0093] For example, key factors include atmospheric thermal key factors, such as the vertical gradient of atmospheric temperature or humidity; correspondingly, the detection algorithm for the atmospheric thermal key factor includes a waveguide detection algorithm based on a modified refractive index vertical gradient threshold.

[0094] In some embodiments, atmospheric factors such as temperature, humidity, and air pressure collectively determine the vertical distribution of the refractive index, and the anomalous variation of the refractive index with altitude is the direct physical cause of atmospheric waveguide formation. To more accurately reflect the propagation characteristics of electromagnetic waves on a curved Earth, studies typically use a modified refractive index instead of the ordinary refractive index. By calculating a large number of samples of the modified refractive index profile, the presence of waveguide structures in the atmosphere can be determined probabilistically.

[0095] In the diagnosis of atmospheric waveguides, the trend of the modified refractive index with altitude is the core basis for waveguide identification. When the modified refractive index shows a decreasing trend, it can be determined that there is a possibility of trapped refraction, and the probability of the trapping layer can be calculated accordingly. The higher the probability of the trapping layer, the greater the possibility that electromagnetic waves are confined and guided in that altitude range, thus producing a more significant waveguide effect.

[0096] In some embodiments, for atmospheric waveguide phenomena, the process of determining the detection algorithm based on the preset mapping relationship is as follows: When the key factors selected are atmospheric thermodynamic key factors (such as temperature vertical gradient, humidity vertical gradient, and wind speed vertical shear), the preset mapping relationship establishes the correspondence between factor type and physical calculation algorithm based on meteorological physics principles - humidity key factors are mapped to the refractive index algorithm for water vapor pressure enhancement calculation and humidity correction, temperature key factors are mapped to the temperature gradient correction and inversion detection algorithm, and wind speed key factors are mapped to the wind shear influence correction and dynamic waveguide detection algorithm.

[0097] In some embodiments, in addition to determining the probability of the existence of the waveguide, its characteristics also need to be probabilistically described. Common indicators include: (1) Waveguide strength: the probability distribution reflects the magnitude of the trapping layer's ability to confine electromagnetic waves. A high probability value area means a stronger possibility of waveguide effect; (2) Waveguide thickness: determined by the probability distribution of the height difference between the upper and lower boundaries of the trapping layer, which can reveal the possibility of different thickness ranges; (3) Waveguide height: represented by the probability distribution of the height of the top of the waveguide, which can quantify the probability of beyond-line-of-sight propagation, blind zone location and radar detection range.

[0098] For example, key factors include ocean stratification key factors, such as the vertical gradient of seawater temperature or salinity. Accordingly, the detection algorithms corresponding to these ocean stratification key factors include stratification detection algorithms based on the identification of the maximum value of the vertical temperature gradient.

[0099] In some embodiments, the presence and properties of thermoclines can be probabilistically identified by calculating the temperature gradient with depth. For example, gradient calculation is first performed on profile data, i.e., comparing the temperature differences between adjacent depth layers. When the rate of temperature decrease with depth exceeds a set threshold, the possibility of a thermocline existing at that depth layer can be determined. For example, when the absolute value of the temperature gradient is greater than 0.1℃ / m, the water body can be considered to have a significant stratified structure, thus meeting the conditions for the formation of a thermocline.

[0100] In some embodiments, the process of determining the detection algorithm based on a preset mapping relationship for temperature stratification phenomena is as follows: When the selected key factors are key factors of ocean stratification (such as seawater temperature vertical gradient, salinity vertical gradient, sea surface temperature, and sea surface height anomaly), the preset mapping relationship maps them to a temperature stratification detection algorithm based on vertical gradient analysis. The core process of this detection algorithm includes: reading temperature vertical profile data, calculating the temperature gradient of each depth layer, identifying the depth where the maximum gradient value is located as the temperature stratification depth, and calculating the stratification intensity.

[0101] In some embodiments, key characteristic parameters of the thermocline can also be quantified by probability distribution function (PDF) or frequency of occurrence.

[0102] (1) Depth of thermocline: The probability distribution corresponding to the depth where the maximum gradient is located can reflect the possibility of the main layer of thermocline appearing; (2) Thickness of thermocline: The probability distribution of the vertical range where the gradient continuously exceeds the threshold can characterize the uncertainty of the spatial scale of thermocline.

[0103] To improve the robustness of forecast results, it is usually necessary to combine statistical analysis with temperature profiles from multiple times and regions. In mid- and low-latitude summers, the thermocline is shallower and more likely to appear; while in high latitudes or regions with strong winds and waves, the thermocline is deeper or less likely to appear. By probabilistically diagnosing these characteristics, the spatiotemporal distribution patterns and uncertainties of the thermocline can be revealed.

[0104] For example, key factors include key factors in marine acoustics, such as temperature and salinity parameters or their gradients that affect the vertical profile of sound velocity; correspondingly, detection algorithms for key factors in marine acoustics include marine acoustic duct detection algorithms based on sound velocity profile analysis.

[0105] In some embodiments, one-dimensional vertical gradient convolution is used to extract the rate of change of sound velocity with depth, and a positive gradient threshold is set to identify depth intervals that continuously satisfy the positive gradient condition, thereby defining the upper and lower boundaries of the sound channel and calculating the sound channel thickness. Based on this, grid statistics and spatial analysis are further performed on the sound channel thickness to output the spatial distribution and intensity map of future time series, forming a sound channel thickness prediction product.

[0106] Since the formation of acoustic duct thickness is closely related to the positive sound velocity gradient structure, and the positive sound velocity gradient is mainly controlled by the vertical structure of temperature and salinity, the sound velocity gradient distribution reconstructed from the forecast hydrological field can effectively reflect the changing trend and spatial probability of acoustic duct thickness in the short term. This method is essentially a fusion of deterministic acoustic gradient diagnosis and probabilistic spatial forecasting: based on multi-time-scale hydrological forecast data as input, it can output the spatial distribution evolution trend of acoustic duct thickness at different time scales; through long-term series modeling and return assessment, it can further construct a spatial statistical map of acoustic duct formation probability, achieving a quasi-probabilistic regional acoustic environment assessment.

[0107] In some embodiments, for ocean acoustic duct phenomena, the process of determining the detection algorithm based on a preset mapping relationship is as follows: When the selected key factors are key ocean acoustic factors (such as temperature gradient, salinity gradient, and temperature-salinity combination parameters affecting the vertical profile of sound velocity), the preset mapping relationship maps them to an ocean acoustic duct detection algorithm based on sound velocity profile analysis. The core process of this detection algorithm includes: calculating the sound velocity profile based on temperature and salinity vertical profile data, identifying the depth layer where the minimum sound velocity is located as the sound duct axis, and determining the upper and lower boundaries of the sound duct and the sound duct thickness.

[0108] 240. Using the detection algorithms determined for each target phenomenon, process marine meteorological forecast data to obtain forecast results for at least one target phenomenon.

[0109] In some embodiments, the forecast results include occurrence status information of the target phenomenon in a future period and characteristic values ​​of the target phenomenon when the occurrence status information indicates that the target phenomenon will occur in that future period.

[0110] In some embodiments, if the target phenomenon is a mesoscale eddy, the characteristic values ​​of the mesoscale eddy include some or all of the following: eddy center location (latitude and longitude coordinates), eddy radius, eddy rotation direction (cyclone / anticyclone), and complete boundary polygon coordinate string. Specifically, after performing circle fitting on each closed contour, the algorithm extracts the center location and radius; during result output and storage, it saves the center latitude and longitude and the complete boundary polygon coordinate string for each contour; during eddy center extraction and boundary reconstruction, it extracts the eddy center latitude and longitude and eddy type (cyclone, anticyclone, or unknown) for the day, and organizes them into a set of closed boundary points for different storage formats.

[0111] In some embodiments, if the target phenomenon includes an atmospheric waveguide, the characteristic values ​​of the atmospheric waveguide accordingly include some or all of the following: waveguide layer initiation height, waveguide layer termination height, waveguide layer thickness, waveguide intensity, waveguide depth, waveguide morphology index, and waveguide type. Specifically, multiple characteristic parameters are calculated for the identified waveguide layer: waveguide intensity is defined as the difference between the improved refractive index M values ​​at the top and bottom of the waveguide layer; waveguide depth is defined as the height at which the gradient reaches its minimum; the waveguide morphology index describes the concavity and convexity of the gradient profile; and waveguide stability is assessed based on the smoothness of the gradient change. Furthermore, waveguides are classified based on the dominant type of key factors, and the output waveguide types include humidity-dominated (evaporation waveguide or humidity waveguide), temperature-dominated (surface waveguide or lifting waveguide), wind shear-dominated (dynamic waveguide), and hybrid (multiple factors acting together).

[0112] In some embodiments, if the target phenomenon includes a marine acoustic channel, the characteristic values ​​of the marine acoustic channel accordingly include some or all of the following: channel axial depth, channel thickness, minimum sound velocity, mean temperature / salinity within the channel, temperature / salinity gradient within the channel, and channel type. Specifically, the basic acoustic characteristics include channel axial depth, thickness, and minimum sound velocity; factor-derived characteristics include mean temperature / salinity within the channel, gradient, outliers, velocity matching characteristics (coupling relationship between the channel and the flow field), and correlation of sea surface height anomalies; interactive characteristics include temperature-salinity co-variance patterns and temperature-salinity-velocity coupling characteristics; and a comprehensive stability index integrates the stability measures of multiple factors. Furthermore, based on the dominant factor, the acoustic channel is classified into thermocline channels (temperature-dominated), halocline channels (salinity-dominated), flow-induced channels (velocity shear-dominated), and mixed-layer channels (meteorological factor-dominated).

[0113] In some embodiments, if the target phenomenon includes a temperature gradient, the characteristic values ​​of the temperature gradient include some or all of the following: temperature gradient depth, temperature gradient intensity, temperature gradient thickness, gradient morphology features, and confidence level. Specifically, the temperature gradient depth is calculated from the temperature data; the gradient intensity is defined as the maximum or average value of the temperature gradient within the gradient; and the gradient thickness is defined as the vertical distance between the upper and lower boundaries of the gradient. In some embodiments, the morphological changes of the gradient in the anomalous region can also be considered to extract more characteristic parameters such as intensity and thickness. In some embodiments, the output includes not only the gradient depth but also the confidence level (high, medium, low) for each grid point, improving the practicality of the product.

[0114] The method provided in this application can be executed individually for a single target phenomenon or in parallel for multiple target phenomena.

[0115] The marine phenomenon forecasting method provided in this application acquires marine meteorological forecast data for future periods, extracts key factors from it, and performs marine meteorological forecasts based on the adaptive mapping relationship between key factors and detection algorithms. In this way, by extracting key factors from future forecast data and determining the corresponding detection algorithms, forecast accuracy and reliability can be improved. Furthermore, this scheme supports parallel processing of multiple phenomena, enabling the analysis and forecasting of various target phenomena such as mesoscale eddies, atmospheric waveguides, oceanic acoustic channels, and temperature strata from the same forecast data, significantly improving computational efficiency and product consistency.

[0116] Based on the same inventive concept, this application also provides a marine phenomenon forecasting device. This marine phenomenon forecasting device can, for example, be located in... Figure 1 The intelligent forecasting system for marine meteorological phenomena shown is 10. (For example...) Figure 3The diagram shows a schematic of a marine phenomenon forecasting device 300, which may include: an acquisition module 310 for acquiring marine meteorological forecast data for future periods; a processing module 320 for extracting forecast data of at least one key factor characterizing the marine environmental state from the marine meteorological forecast data acquired by the acquisition module 310, wherein the key factor is associated with at least one target phenomenon, including at least one of the following: mesoscale eddies, atmospheric waveguides, oceanic acoustic channels, and thermoclines; the processing module 320 is further configured to determine corresponding detection algorithms for each of the at least one target phenomenon associated with the key factor according to a preset mapping relationship; wherein the preset mapping relationship indicates the association rules between the key factor and the detection algorithms of the target phenomenon; and the processing module 320 is further configured to process the marine meteorological forecast data using the detection algorithms determined for each target phenomenon to obtain a forecast result for at least one target phenomenon.

[0117] In some embodiments, the acquisition module 310 is further configured to acquire historical marine meteorological data associated with any target phenomenon; the processing module 320 is further configured to determine multiple candidate factors associated with the target phenomenon based on the historical marine meteorological data; and select key factors corresponding to the target phenomenon from the multiple candidate factors according to the correlation between each candidate factor and the target phenomenon.

[0118] In some embodiments, the processing module 320 is further configured to, for any candidate factor, determine the historical observation data of the candidate factor at various spatiotemporal locations and the historical information on the occurrence status of the target phenomenon at those spatiotemporal locations; generate a spatial distribution of the correlation metric between the candidate factor and the target phenomenon based on the historical observation data and the historical information on the occurrence status, wherein the spatial distribution of the correlation metric characterizes the spatial variation of the correlation strength index and the statistical significance of the correlation strength index between the candidate factor and the target phenomenon; and determine the candidate factors that meet preset conditions as key factors, wherein the preset conditions include: there is a spatially continuous region in the spatial distribution of the correlation metric corresponding to the candidate factor; within the continuous region, the correlation strength index reaches a preset strength threshold; and the proportion of grid points with statistical significance of the correlation strength index exceeds a preset proportion threshold.

[0119] In some embodiments, association rules indicate the association between the physical type of a key factor and the type of detection algorithm.

[0120] In some embodiments, the preset mapping relationship indicates at least one of the following association rules: key factors include marine dynamic key factors, and the detection algorithm corresponding to the marine dynamic key factors includes a mesoscale eddy detection algorithm based on the geometric identification of closed contour lines of sea surface height anomalies; key factors include atmospheric thermal key factors, and the detection algorithm corresponding to the atmospheric thermal key factors includes an atmospheric waveguide detection algorithm based on a modified vertical refractive index gradient threshold; key factors include marine stratification key factors, and the detection algorithm corresponding to the marine stratification key factors includes a stratification detection algorithm based on the identification of the maximum value of the vertical temperature gradient; key factors include marine acoustic key factors, and the detection algorithm corresponding to the marine acoustic key factors includes a marine acoustic channel detection algorithm based on sound velocity profile analysis.

[0121] In some embodiments, key marine dynamic factors include sea surface height field or flow vorticity field; key atmospheric thermal factors include vertical gradient of atmospheric temperature or vertical gradient of humidity; key marine stratification factors include vertical gradient of seawater temperature or salinity; and key marine acoustic factors include temperature and salinity parameters or their gradients that affect the vertical profile of sound speed.

[0122] In some embodiments, the forecast results include occurrence status information of the target phenomenon in the future time period and characteristic values ​​of the target phenomenon when the occurrence status information indicates that the target phenomenon will occur in the future time period.

[0123] The marine phenomenon forecasting device provided in this application acquires marine meteorological forecast data for future periods, extracts key factors from it, and performs marine meteorological forecasts based on the adaptive mapping relationship between the key factors and the detection algorithm. In this way, by extracting key factors from future forecast data and dynamically configuring the algorithm, forecast accuracy and reliability can be improved. Furthermore, this scheme supports parallel processing of multiple phenomena, enabling the analysis and forecasting of various target phenomena such as mesoscale eddies, atmospheric waveguides, ocean acoustic channels, and temperature strata from the same forecast data, significantly improving computational efficiency and product consistency.

[0124] Based on the same inventive concept, embodiments of this application also provide an electronic device. This electronic device, for example, can be located in... Figure 1 The intelligent forecasting system 10 shown. Figure 4 The electronic device shown also includes a communication interface 403 and a communication bus 404, wherein the processor 401, the memory 402 and the communication interface 403 communicate with each other through the communication bus 404.

[0125] The memory 402 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. The communication bus 404 can be an ISA bus, PCI bus, or EISA bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, only a single bidirectional arrow is used in the diagram, but this does not imply that there is only one bus or one type of communication bus.

[0126] The communication interface 403 is used to connect to at least one user terminal and other network units through the network interface, and to send the encapsulated message to the user terminal through the network interface.

[0127] Processor 401 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of processor 401 or by instructions in software form. Processor 401 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this disclosure. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this disclosure can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 402. Processor 401 reads the information in memory 402 and, in conjunction with its hardware, completes the steps of the method described in the foregoing embodiment.

[0128] This application also provides a computer storage medium storing computer-executable instructions. When executed by a processor, these instructions are used to implement the marine phenomenon forecasting method described in any of the preceding embodiments; therefore, they will not be repeated here. Furthermore, the beneficial effects of using the same method will also not be repeated. For technical details not disclosed in the computer storage medium embodiments of this invention, please refer to the description of the method embodiments of this invention.

[0129] This application also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it is used to implement the marine phenomenon forecasting method described in any of the preceding embodiments. Therefore, it will not be described again here.

[0130] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0131] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for forecasting marine phenomena, characterized in that, include: Obtain marine meteorological forecast data for future periods; From the marine meteorological forecast data, forecast data of at least one key factor characterizing the state of the marine environment is extracted. The key factor is associated with at least one target phenomenon, which includes at least one of the following: mesoscale eddies, atmospheric waveguides, ocean acoustic channels, and temperature strata. Based on a preset mapping relationship, a corresponding detection algorithm is determined for each of the at least one target phenomenon associated with the key factor; wherein, the preset mapping relationship indicates the detection algorithm of the key factor and the target phenomenon, as well as the association rules between the detection algorithms; The marine meteorological forecast data is processed using the detection algorithm determined for each of the target phenomena to obtain the forecast result for the at least one target phenomenon.

2. The method according to claim 1, characterized in that, The method further includes: For any target phenomenon, acquire historical marine meteorological data associated with the target phenomenon; Based on the aforementioned historical marine meteorological data, several candidate factors associated with the target phenomenon were identified; Based on the correlation between each candidate factor and the target phenomenon, the key factor corresponding to the target phenomenon is selected from the plurality of candidate factors.

3. The method according to claim 2, characterized in that, The step of selecting the key factor corresponding to the target phenomenon from the multiple candidate factors based on the correlation between each candidate factor and the target phenomenon includes: for any candidate factor, determining the historical observation data of the candidate factor at each spatiotemporal location and the historical information of the occurrence state of the target phenomenon at the spatiotemporal location; Based on the historical observation data and the historical information of the occurrence status, a spatial distribution of the correlation metric between the candidate factor and the target phenomenon is generated. The spatial distribution of the correlation metric characterizes the spatial variation of the correlation strength index between the candidate factor and the target phenomenon and the statistical significance of the correlation strength index. Candidate factors that meet preset conditions are identified as key factors. The preset conditions include: there is a spatially continuous region in the correlation metric spatial distribution corresponding to the candidate factor; the correlation strength index reaches a preset strength threshold in the continuous region; and the proportion of grid points with statistical significance of the correlation strength index exceeds a preset proportion threshold.

4. The method according to claim 1, characterized in that, The association rule indicates the relationship between the physical type of the key factor and the type of the detection algorithm.

5. The method according to claim 4, characterized in that, The preset mapping relationship indicates at least one of the following association rules; The key factors include ocean dynamic key factors, and the detection algorithms corresponding to the ocean dynamic key factors include mesoscale eddy detection algorithms based on geometric identification of closed contour lines of sea surface height anomalies. The key factors include atmospheric thermal key factors, and the detection algorithms corresponding to the atmospheric thermal key factors include atmospheric waveguide detection algorithms based on modified refractive index vertical gradient thresholds. The key factors include ocean stratification key factors, and the detection algorithms corresponding to the ocean stratification key factors include a layer-skipping detection algorithm based on the maximum value of the vertical temperature gradient. The key factors include marine acoustic key factors, and the detection algorithms corresponding to the marine acoustic key factors include marine acoustic duct detection algorithms based on sound velocity profile analysis.

6. The method according to claim 5, characterized in that, The key marine dynamic factors include sea surface height field or current vorticity field; the key atmospheric thermal factors include atmospheric temperature vertical gradient or humidity vertical gradient; the key marine stratification factors include seawater temperature or salinity vertical gradient; and the key marine acoustic factors include temperature and salinity parameters or their gradients that affect the vertical profile of sound speed.

7. The method according to claim 1, characterized in that, The forecast results include the occurrence status information of the target phenomenon in the future time period and the characteristic values ​​of the target phenomenon when the occurrence status information indicates that the target phenomenon will occur in the future time period.

8. A marine phenomenon forecasting device, characterized in that, include: The acquisition module is used to acquire marine meteorological forecast data for future periods; The processing module is used to extract forecast data of at least one key factor characterizing the state of the marine environment from the marine meteorological forecast data acquired by the acquisition module. The key factor is associated with at least one target phenomenon, which includes at least one of the following: mesoscale eddies, atmospheric waveguides, ocean acoustic channels, and temperature strata. The processing module is further configured to determine corresponding detection algorithms for the at least one target phenomenon associated with the key factor according to a preset mapping relationship; wherein the preset mapping relationship indicates the association rules between the key factor and the detection algorithm of the target phenomenon; The processing module is further configured to process the marine meteorological forecast data using the detection algorithm determined for each of the target phenomena, and obtain the forecast result of the at least one target phenomenon.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1 to 7.

Images