A method and system for self-adaptive coupling prediction and hierarchical early warning of waves, currents, temperature and oxygen in a deep water net cage culture area

Abstract

The application discloses a kind of deep water net cage breeding area wave flow temperature oxygen self-adapting coupling prediction and hierarchical early warning method and system, method includes: obtaining target sea area multi-source data, establish the nested calculation domain of regional scale and breeding area fine scale, and the regional calculation result is passed to fine model as boundary constraint;Construct net cage group subgrid resistance source term;Establish multi-physical field coupling prediction model;According to the time-varying degree of environmental state and wave flow coupling deviation, construct comprehensive scheduling index, dynamically adjust wave flow coupling frequency;Based on error covariance constraint, local correction is carried out by fusing field observation;Extract the threshold amplitude and duration of each risk factor, respectively generate structure safety risk and ecological stress risk and determine the comprehensive risk grade, output early warning information.The application considers prediction accuracy and calculation efficiency, and is suitable for net cage point rolling prediction and joint hierarchical early warning.

Description

Technical Field

[0001] This invention relates to the field of marine numerical forecasting and smart aquaculture technology, and in particular to a method and system for adaptive coupled forecasting and graded early warning of waves, currents, temperature and oxygen in deep-sea cage aquaculture areas. Background Technology

[0002] Deep-sea cage aquaculture is an important form of marine aquaculture developing towards large-scale, intensive, and offshore operations. Compared with nearshore ponds and shallow-water raft aquaculture, deep-sea cages are typically deployed in sea areas with higher levels of openness and more complex marine dynamic conditions. Their operational safety and the health of cultured organisms are simultaneously affected by multiple environmental factors such as waves, currents, storm surges, water temperature, and dissolved oxygen. Events such as typhoons, storm surges, disastrous waves, abnormally strong currents, high temperatures, and low dissolved oxygen are characterized by their suddenness, persistence, and multi-factor coupling, which may lead to increased mooring loads on cages, net deformation, decreased water exchange capacity, and enhanced stress responses in cultured organisms. In severe cases, this can cause structural damage to cages, fish escape, or mortality of cultured organisms. Therefore, developing refined environmental forecasts with spatial positioning capabilities, continuous forecasting capabilities, and risk identification capabilities for specific aquaculture areas and cage clusters is a crucial technical foundation for improving the safety and security of deep-sea cage aquaculture.

[0003] Existing marine environmental forecasting products primarily serve regional marine disaster prevention, port shipping, and nearshore engineering activities. Their computational scale, spatial resolution, and output methods are often insufficient to directly correspond to individual fish cages or fish cage groups. Aquaculture enterprises typically utilize a combination of regional weather forecasts, sea state forecasts, on-site observations, and empirical judgment for risk control in actual production. While on-site buoys can provide real-time or near-real-time monitoring information on environmental elements such as current velocity, water temperature, and dissolved oxygen, their observations mainly reflect the environmental state at a limited location and current moment, making it difficult to independently describe the spatial evolution of environmental elements in the aquaculture area over the next few days. Although regional numerical forecasts can provide some lead time, their accuracy in specific aquaculture areas can still be affected by factors such as seafloor topographic resolution, boundary condition errors, local seabed friction, fish cage group resistance, and insufficient observational correction. Therefore, relying solely on regional forecasts or single-point monitoring is usually insufficient to simultaneously meet the needs of fish cage-scale environmental prediction and production risk decision-making.

[0004] Waves, currents, water temperature, and dissolved oxygen correspond to ocean dynamics, thermodynamics, and ecological processes, respectively, and these processes exhibit significant interactions. For example, current fields and water levels alter wave propagation, refraction, and breaking processes, while the radiation stress and enhanced mixing generated by waves affect local current fields. Hydrodynamic conditions determine the transport, diffusion, and exchange processes of heat and dissolved oxygen, and changes in water temperature further influence saturated dissolved oxygen concentration and oxygen consumption levels in aquaculture organisms. Existing forecasting methods treat these processes using independent calculations, unidirectional transmission, or fixed-time interval exchanges. While fixed, longer coupling exchange intervals can reduce computational load, they may be insufficient to reflect rapid feedback between waves and currents during non-stationary processes such as typhoons, strong winds and waves, and rapid water increases. Conversely, fixed, shorter exchange intervals, while improving the accuracy of time-varying processes, can lead to significant computational redundancy under relatively stable sea conditions. Therefore, dynamically adjusting the wave-current coupling frequency based on the degree of sea state change to achieve a reasonable balance between computational efficiency and forecast accuracy is one of the key challenges in short-term rolling forecasts for aquaculture areas.

[0005] Deep-water cages and their netting structures also generate additional resistance to the local hydrodynamic environment. Mesh size, wire diameter, netting density, attached biomass coverage, netting angle against the current, and cage arrangement can all affect velocity distribution and water exchange capacity within the cages and on the back side. Completely ignoring the cage structure's influence in regional numerical models may lead to inaccurate estimates of local flow velocities and the transport capacity of pollutants, heat, and dissolved oxygen in the cage area. Scale-by-scale geometric analysis of the wire and mesh requires computational grids much smaller than the regional forecast scale, resulting in high computational costs and modeling expenses, which is unfavorable for large-scale, multi-scenario, and long-term window forecasts. Therefore, a parametric method is needed to transform cage geometry, netting density, attachment rate, and upstream state into equivalent resistance source terms in the regional computational grid, thereby characterizing the impact of cage groups on the local flow field while maintaining engineering computability.

[0006] Furthermore, existing early warning methods for aquaculture environments typically rely on single environmental factors or instantaneous thresholds as primary criteria, rarely considering the magnitude and duration of risk factor exceedances, or the combined effects of multiple factors. For deep-sea cage aquaculture, short-term increases in wave height and persistent strong currents can have different impacts on cage structures, and the simultaneous occurrence of high temperatures and low dissolved oxygen can significantly enhance ecological stress on cultured organisms. Therefore, early warning based solely on a single moment or variable is insufficient to comprehensively reflect the actual impact of complex marine environmental processes on cage structural safety and the health of cultured organisms.

[0007] In summary, existing technologies still need to address the following issues: establishing an effective connection between regional-scale numerical forecasting and cage-scale production needs; achieving coordinated forecasting of waves, currents, water temperature, and dissolved oxygen at an acceptable computational cost; rationally characterizing the resistance effect of cage groups on the local hydrodynamic environment; dynamically adjusting the coupling calculation frequency according to the degree of environmental change; using on-site observations to locally correct cage location forecasts; and comprehensively considering the intensity, duration, and combined effects of environmental factors to form tiered early warning results that simultaneously address cage structural safety and ecological stress on aquaculture organisms. Summary of the Invention

[0008] This invention aims to address the following problems in existing deep-sea cage aquaculture environmental forecasting: regional-scale forecast results are difficult to directly adapt to the production and management needs at the cage scale; the characterization of the local resistance effect of cage groups is insufficient; fixed coupling time intervals are difficult to balance computational efficiency and simulation accuracy of rapidly changing sea states; the degree of integration between field observation information and numerical forecast results is low; and there is a lack of unified quantification and hierarchical expression of structural safety risks and ecological stress risks. To address these issues, this invention proposes an adaptive coupled forecasting and hierarchical early warning method and system for wave, current, temperature and oxygen in deep-sea cage aquaculture areas.

[0009] On the one hand, to achieve the above objectives, this invention provides a method for adaptive coupled forecasting and graded early warning of waves, currents, temperature, and oxygen in deep-sea cage aquaculture areas, comprising: Acquire multi-source data of the target sea area, establish a nested relationship between the regional scale calculation domain and the fine-scale calculation domain of the aquaculture area, and transfer the regional scale calculation results to the fine-scale calculation domain of the aquaculture area as boundary constraints. The netting is equivalent to an anisotropic porous resistance medium, and a subgrid anisotropic resistance source term for the net cage group is constructed in the fine-scale computational domain of the aquaculture area. A multiphysics coupled prediction model is established, which includes hydrodynamic processes, wave processes, temperature transport processes, and dissolved oxygen transport processes. The hydrodynamic processes and wave processes exchange information bidirectionally. The hydrodynamic processes drive the transport of temperature and dissolved oxygen, and the temperature participates in determining the dissolved oxygen saturation concentration and biological oxygen consumption rate. Based on the time-varying nature of the marine environment and the deviation of the wave-current coupling effect, a comprehensive scheduling index is constructed. The frequency of coupling and interaction between the wave process and the hydrodynamic process is dynamically adjusted by comparing the comprehensive scheduling index with a preset threshold. Based on the background error covariance and the observation error covariance, the model prediction results are locally constrained and corrected using field observation data to obtain the corrected prediction information of the cage locations. Based on the corrected forecast information, the degree and duration of exceeding the threshold of each risk element are extracted, and they are aggregated to form structural safety risk and ecological stress risk, respectively. A comprehensive risk is determined based on the structural safety risk and the ecological stress risk, and early warning information is output based on the comprehensive risk.

[0010] Preferably, the multi-source data includes static basic data and dynamic environmental data; The static basic data includes at least the coastline, water depth, seabed type, and seabed roughness; the dynamic environmental data includes at least the tide level, current velocity, wind speed, wind direction, air pressure, wave spectrum, surface heat flux, salinity background field, and dissolved oxygen background field. The fine-scale calculation domain of the aquaculture area sets up locally denser grids in the aquaculture area, shoreline transition area, water depth gradient change area, and around the cage group.

[0011] Preferably, constructing the subgrid anisotropic drag source term includes: The net density is calculated based on the mesh size, net wire diameter, and thickness of attached organisms. The normal and tangential velocity components are determined based on the velocity of the water relative to the net. The normal and tangential drag coefficients are determined based on the net Reynolds number, net density, degree of attachment, and net deformation state, respectively. The netting projected area is mapped to the netting area density per unit water volume according to the intersection weight with the computational grid, thus generating an anisotropic drag source term; When multiple cages act on the same computational grid, the resistance source terms corresponding to each cage are superimposed.

[0012] Preferably, the equivalent mesh diameter after the attachment of biological agents is [missing information]. The solidity of the square mesh is ,in, To clean the diameter of the network cable, The average thickness of biological adhesion on the surface of the network cable. m The mesh side length; No. Resistance source terms within a computational grid for: ; In the formula, The density of seawater, The netting area density per unit volume of water. The normal drag coefficient of the mesh is... The tangential drag coefficient of the mesh is... For the normal component of the netting, This represents the tangential component of the mesh.

[0013] Preferably, dynamically adjusting the coupling frequency between the wave process and the hydrodynamic process includes: The time-varying degree of the environmental state is determined by the weighted sum of the relative changes of at least two variables among the significant wave height, flow velocity, water level, and wind speed at adjacent time points; The deviation of the wave-current coupling effect is determined based on the change in wave radiation stress and the difference between the flow velocity at the previous call time and the current hydrodynamic flow velocity. The comprehensive scheduling index is composed of a weighted average of the time-varying degree of the environmental state and the deviation of the wave-current coupling effect; The preset threshold includes a startup threshold and a recovery threshold, wherein the recovery threshold is less than the startup threshold; When the comprehensive scheduling index continuously reaches the start threshold, it switches to high-frequency coupling state; when the comprehensive scheduling index continuously falls below the recovery threshold, it returns to low-frequency coupling state; the coupling interaction frequency is limited to between a preset minimum frequency and a maximum frequency.

[0014] Preferably, in the local constraint correction, the analysis state satisfies: , In the formula, This represents the background state of the model. For the observation vector, For the observation operator, The background error covariance matrix, The observation error covariance matrix; For analysis of the status; The background error covariance matrix is ​​locally corrected according to the spatial distance between the observation point and the model point, the difference in water depth, the coastal barrier, and the direction of the main current. When the observation data is insufficient, the local constraint correction degenerates into a scalar weighted correction that decreases with the increase of spatial distance, time difference, and observation error.

[0015] Preferably, the risk factors include significant wave height, flow velocity, water temperature, and dissolved oxygen; The degree of exceeding the threshold refers to the normalization of the predicted value of the risk element relative to the risk initiation threshold and the high-risk threshold; The duration of exceeding the threshold is the ratio of the continuous or cumulative time exceeding the threshold within the forecast period to the total forecast duration; The structural safety risk is obtained by aggregating the significant wave height risk, flow velocity risk, and wave-current combined load risk. The wave-current combined load risk is determined based on the cage structure parameters and the wave-current combined force. The ecological stress risks are obtained by combining high temperature risk and low dissolved oxygen risk. The overall risk is the larger of the structural safety risk and the ecological stress risk.

[0016] Preferably, multiple forecast members are constructed by using at least one of the following: disturbing wind field, open sea waves, tidal boundary, netting attachment parameters, and oxygen consumption parameters. The proportion of members whose risk thresholds are exceeded is used as the forecast confidence level, and the forecast is output synchronously with the comprehensive risk. The early warning information includes the effective wave height, flow velocity, water temperature, dissolved oxygen, structural safety risk, ecological stress risk, comprehensive risk level, dominant risk factor, expected start time, expected peak time, expected duration, and forecast confidence level for each cage or cage section.

[0017] Preferably, it further includes: Weather forecasts, offshore boundary data, and field observation data are updated on a rolling basis according to a preset cycle. After the forecast ends, the actual observation results are compared with the forecast results, and the background error covariance, the observation error covariance, the preset threshold, and the cage resistance parameters are updated accordingly.

[0018] On the other hand, to achieve the above objectives, the present invention also provides an adaptive coupled forecasting and hierarchical early warning system for wave, current, temperature, and oxygen in deep-sea cage aquaculture areas, comprising: The data acquisition and processing module is used to acquire and process multi-source data from the target sea area; A multi-scale nested module is used to establish a nested relationship between the regional scale calculation domain and the fine-scale calculation domain of the aquaculture area, and to transfer the regional scale calculation results to the fine-scale calculation domain of the aquaculture area. The cage resistance parameterization module is used to treat the netting as an anisotropic porous resistance medium and construct the sub-grid anisotropic resistance source term of the cage group in the fine-scale computational domain of the aquaculture area. The multiphysics coupling prediction module is used to perform coupled calculations of hydrodynamic processes, wave processes, temperature transport processes, and dissolved oxygen transport processes; An adaptive scheduling module is used to construct a comprehensive scheduling index based on the time-varying degree of environmental conditions and the deviation of wave-current coupling effect, and to dynamically adjust the coupling interaction frequency between wave process and hydrodynamic process by comparing the comprehensive scheduling index with a preset threshold. The observation correction module is used to perform local constraint correction on the model prediction results based on the background error covariance and the observation error covariance, using field observation data to obtain the correction prediction information of the cage location. The risk assessment module is used to calculate structural safety risks and ecological stress risks based on the degree and duration of each risk factor exceeding the threshold, and to determine the comprehensive risk level. The early warning output module is used to output early warning information.

[0019] Compared with the prior art, the present invention has the following advantages and technical effects: (1) This invention establishes a multi-scale information transmission mechanism between regional marine environmental forecasting and cage-scale fine calculation. By densifying the local grid in aquaculture areas, shoreline transition areas, areas with large water depth gradients and around cage groups, and using regional model results to provide dynamic boundary conditions for aquaculture area fine models, regional marine environmental forecasting results can be further transmitted to specific cages or cage zones, which is conducive to improving the matching degree between forecasting results and aquaculture production management spatial scale.

[0020] (2) This invention treats the netting as a porous resistance medium with directional selectivity. Based on the mesh size, wire diameter, attached organism thickness, netting normal, netting deformation state, and local relative flow velocity, the normal and tangential resistances of the netting are calculated respectively. The projected area of ​​the netting is then mapped to the computational grid according to the intersection area weight, forming an anisotropic resistance field of the net cage group subgrid. Thus, without the need for solid mesh analysis of centimeter-level meshes and wires, the directional influence of the net cage group on local flow velocity, water exchange, and material transport can be characterized in the numerical model of a region or aquaculture area, which is beneficial to reducing the grid size and computational load caused by fine geometric analysis.

[0021] (3) This invention constructs a multi-process forecasting structure that integrates wave-current bidirectional coupling, hydrodynamic-water temperature-dissolved oxygen sequential coupling, and water temperature feedback correction of dissolved oxygen processes. The hydrodynamic module transmits water level and flow velocity to the wave module, while the wave module feeds back radiation stress and wave enhancement mixing parameters to the hydrodynamic module. The hydrodynamic field further drives the transport of heat and dissolved oxygen, and water temperature participates in dissolved oxygen calculation by affecting saturated dissolved oxygen concentration, reoxygenation process, and biological metabolic rate. Compared with the method of calculating each environmental element independently, this technical structure can more completely reflect the physical relationship between dynamic, thermal, and dissolved oxygen processes in the aquaculture area.

[0022] (4) This invention simultaneously utilizes marine environmental state change indicators and wave-current coupling defect indicators to determine the model exchange interval, and achieves adaptive switching between quasi-steady-state low-frequency coupling and unsteady-state high-frequency coupling through start-up thresholds, recovery thresholds, and hysteresis intervals. Under normal sea conditions, the system can use a longer exchange interval to reduce redundant calculations by the wave module; during non-stationary processes such as typhoons, strong winds and waves, rapid water rises, or abnormal strong currents, the exchange interval can be automatically shortened and the wave-current information exchange frequency increased. Therefore, this invention can dynamically allocate computational resources according to the non-stationarity of environmental processes, which is beneficial for coordinating the relationship between rolling forecast lead time, computational load, and simulation accuracy of rapidly changing processes.

[0023] (5) This invention employs a local optimal interpolation method constrained by background error covariance and observation error covariance to fuse buoy or monitoring station observation information with the model background field, and controls the influence range of observation residuals based on spatial distance, water depth differences, coastal obstruction, and mainstream direction. Compared with the correction method that only weights residuals based on distance, this method can distinguish between model background error and observation error, suppress the unreasonable propagation of abnormal observations or long-distance observations to cage location prediction results, and can degenerate into scalar weighted correction when the number of observations is insufficient, thereby improving the stability and applicability of the method of utilizing field observation data.

[0024] (6) This invention separately constructs the structural safety risk of the cage and the ecological stress risk of the aquaculture organisms, comprehensively considering factors such as effective wave height, current velocity, combined wave-current load, water temperature, and dissolved oxygen, and incorporates the threshold exceedance magnitude and the continuous or cumulative threshold exceedance time into the risk calculation. Thus, it can distinguish between short-term, minor exceedance processes and continuous high-risk processes, reducing the distortion of risk levels caused by using only a single moment or a single threshold for judgment, and can identify structural safety risks, ecological stress risks, and their dominant risk factors.

[0025] (7) This invention can construct multiple forecast members by using disturbed wind fields, offshore waves, tidal boundaries, netting attachment rate, and oxygen consumption parameters, and calculate the probability of exceeding the risk threshold. The probability of exceeding the threshold is output as the warning confidence level and risk level simultaneously. This technology can characterize the forecast uncertainty caused by boundary conditions, model parameters, and observation errors in addition to the deterministic risk value, providing supplementary basis for enterprises to judge the reliability of warning information.

[0026] (8) This invention updates meteorological boundaries, offshore boundaries, and field observation data on a rolling basis according to a preset cycle. After the forecasting process is completed, the actual observation results, enterprise handling records, and disaster results are compared with the forecast results to update background error parameters, observation error parameters, adaptive coupling thresholds, and risk thresholds. Through the feedback calibration mechanism, the forecasting model and early warning rules can be continuously corrected with the accumulation of observation data and operational cases, avoiding the model parameters and early warning thresholds from remaining fixed for a long time.

[0027] (9) This invention can output effective wave height, current velocity, water temperature, dissolved oxygen, structural safety risk, ecological stress risk, comprehensive risk, dominant risk factor, expected occurrence time, expected peak time, expected duration, and forecast confidence level for a single net cage or net cage zone. It can also generate differentiated treatment suggestions based on different dominant risks, such as net cage reinforcement, mooring inspection, cessation of offshore operations, reduction of feeding, enhanced aeration, early harvesting, or personnel evacuation. Therefore, this invention can transform marine environmental numerical forecast results into aquaculture production auxiliary decision-making information with spatial positioning, temporal lead time, and risk orientation. Attached Figure Description

[0028] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart of an adaptive coupling forecasting and graded early warning method for wave, current, temperature and oxygen in deep-water cage aquaculture areas, according to an embodiment of the present invention. Figure 2 This is a system architecture and data flow diagram according to an embodiment of the present invention; Figure 3 This is a multi-scale nested model diagram of the regional model and the fine model of the aquaculture area in an embodiment of the present invention; Figure 4 This is a diagram showing the construction of the cage resistance field by mapping the cage structure parameters to the anisotropic resistance source terms of the computational grid, as described in an embodiment of the present invention. Figure 5 This is an adaptive coupling scheduling diagram for adjusting the switching interval based on changes in environmental conditions and wave-current coupling defects, as described in an embodiment of the present invention. Figure 6 This is a local observation correction diagram based on error covariance constraint fusion of field observation and model background field according to an embodiment of the present invention; Figure 7 This is a joint risk warning diagram for the joint calculation and hierarchical output of structural safety risk and ecological stress risk in an embodiment of the present invention. Detailed Implementation

[0029] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0030] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0031] Example 1 This technical solution aims to address the following issues in existing deep-sea cage aquaculture environmental forecasting: regional-scale forecast results are difficult to directly adapt to the production and management needs at the cage scale; the characterization of the local resistance effect of cage groups is insufficient; fixed coupling time intervals are difficult to balance computational efficiency and simulation accuracy of rapidly changing sea states; the degree of integration between field observation information and numerical forecast results is low; and there is a lack of unified quantification and hierarchical expression of structural safety risks and ecological stress risks. The solution provides a method and system for adaptive coupled forecasting and hierarchical early warning of waves, currents, temperature, and oxygen in deep-sea cage aquaculture areas.

[0032] Another objective of this invention is to establish a technical linkage mechanism between regional marine environmental forecasting and cage-scale risk identification. By parameterizing cage geometry, net density, attachment rate, and current-facing state as equivalent resistance source terms in the computational grid, the impact of cage groups on local flow fields and water exchange processes is characterized. By constructing environmental change indicators, the coupling and exchange interval between the wave module and the hydrodynamic module is dynamically adjusted, allowing for a lower computation frequency under normal sea states and a higher coupling update frequency under rapidly changing sea states, thereby coordinating the relationship between forecast accuracy, computational load, and operational timeliness.

[0033] This technical solution also aims to improve the applicability of numerical weather prediction results to specific aquaculture areas and cage locations. By calculating the residuals between the measured values ​​from the buoy and the model predictions, and performing spatiotemporal weighted corrections based on the spatial distance between the buoy and the cage location and the observation time difference, a cage location prediction sequence with local observation constraints is formed. This reduces the impact of regional model boundary conditions, local topography, and parameter setting errors on the prediction results for aquaculture areas.

[0034] Based on this, this embodiment comprehensively considers environmental factors such as effective wave height, flow velocity, equivalent hydrodynamic load, water temperature, and dissolved oxygen, and jointly considers the threshold exceedance amplitude, duration, and combined effects of risk factors to construct assessment results for the safety risks of cage structures and the ecological stress risks of aquaculture organisms. It outputs the risk level, dominant risk type, expected occurrence time, and duration, thereby providing a technical basis with spatial targeting and time lead time for production management measures such as cage reinforcement, production scheduling, feeding adjustment, oxygenation, early harvesting, and personnel evacuation.

[0035] A method for adaptive coupled forecasting and graded early warning of waves, currents, temperature, and oxygen in deep-sea cage aquaculture areas includes: Acquire multi-source data of the target sea area, establish a nested relationship between the regional scale calculation domain and the fine-scale calculation domain of the aquaculture area, and transfer the regional scale calculation results to the fine-scale calculation domain of the aquaculture area as boundary constraints. The netting is equivalent to an anisotropic porous resistance medium, and a subgrid anisotropic resistance source term for the net cage group is constructed in the fine-scale computational domain of the aquaculture area. A multiphysics coupled prediction model is established, which includes hydrodynamic processes, wave processes, temperature transport processes, and dissolved oxygen transport processes. The hydrodynamic processes and wave processes exchange information bidirectionally. The hydrodynamic processes drive the transport of temperature and dissolved oxygen, and the temperature participates in determining the dissolved oxygen saturation concentration and biological oxygen consumption rate. Based on the time-varying nature of the marine environment and the deviation of the wave-current coupling effect, a comprehensive scheduling index is constructed. The frequency of coupling and interaction between the wave process and the hydrodynamic process is dynamically adjusted by comparing the comprehensive scheduling index with a preset threshold. Based on the background error covariance and the observation error covariance, the model prediction results are locally constrained and corrected using field observation data to obtain the corrected prediction information of the cage locations. Based on the corrected forecast information, the degree and duration of exceeding the threshold of each risk element are extracted, and they are aggregated to form structural safety risk and ecological stress risk, respectively. A comprehensive risk is determined based on the structural safety risk and the ecological stress risk, and early warning information is output based on the comprehensive risk.

[0036] Furthermore, the multi-source data includes static basic data and dynamic environmental data; The static basic data includes at least the coastline, water depth, seabed type, and seabed roughness; the dynamic environmental data includes at least the tide level, current velocity, wind speed, wind direction, air pressure, wave spectrum, surface heat flux, salinity background field, and dissolved oxygen background field. The fine-scale calculation domain of the aquaculture area sets up locally denser grids in the aquaculture area, shoreline transition area, water depth gradient change area, and around the cage group.

[0037] Specifically, step 1 involves multi-source data acquisition, quality control, and multi-scale computational domain construction: Acquire static basic data such as coastline, water depth, seabed type, and seabed roughness of the target sea area; Acquire dynamic environmental data such as tidal level, current velocity, wind speed, wind direction, air pressure, wave spectrum, surface heat flux, salinity background field, and dissolved oxygen background field; Obtain water level, flow velocity, significant wave height, water temperature, and dissolved oxygen data from buoys or monitoring stations in the aquaculture area; Obtain aquaculture production data such as the center coordinates of the cage, cage diameter, netting depth, mesh size, wire diameter, netting thickness, cage arrangement, aquaculture biomass, and feed amount.

[0038] The data from different sources are sequentially subjected to coordinate system transformation, time reference unification, range verification, rate of change verification, outlier identification, missing measurement marking, and time interpolation.

[0039] For continuous abnormal, out-of-range, or abrupt data, a quality label is set, and the corresponding observation weight is reduced or the observation data is not used in subsequent observation correction processes.

[0040] A regional computational domain covering the open sea boundary to the target aquaculture area is established, with locally refined grids set in the aquaculture area, shoreline transition zones, areas with large water depth gradients, and around the cage clusters. The water level, flow velocity, wave, and temperature information output by the regional model are used to form the boundary conditions of the refined aquaculture model through spatiotemporal interpolation, thereby establishing a nested relationship between regional-scale environmental forecasting and cage-scale refined calculations.

[0041] Furthermore, the subgrid anisotropic drag source term is constructed, including: The net density is calculated based on the mesh size, net wire diameter, and thickness of attached organisms. The normal and tangential velocity components are determined based on the velocity of the water relative to the net. The normal and tangential drag coefficients are determined based on the net Reynolds number, net density, degree of attachment, and net deformation state, respectively. The netting projected area is mapped to the netting area density per unit water volume according to the intersection weight with the computational grid, thus generating an anisotropic drag source term; When multiple cages act on the same computational grid, the resistance source terms corresponding to each cage are superimposed.

[0042] Specifically, step 2, the construction of the sub-mesh anisotropic drag field of the cage group, includes: The netting is considered as a porous resistance medium with directional selectivity, and the netting density is calculated based on the mesh size, netting wire diameter, and the thickness of the attached organisms.

[0043] Considering the equivalent network diameter after attachment of organisms for: (1); In the formula, To clean the diameter of the network cable, The average thickness of organisms attached to the surface of the network cable.

[0044] For square mesh, the mesh density for: (2); In the formula, The side length of the mesh.

[0045] The velocity vector of the water body relative to the netting Decomposed into netting normal components and tangential components : (3); In the formula, Let be the unit normal vector of the mesh surface.

[0046] The normal drag coefficient of the mesh is determined based on the mesh density, mesh Reynolds number, adhesion rate, and mesh deformation degree. and tangential drag coefficient .

[0047] For the A computational grid is used to allocate the projected area of ​​each net surface intersecting the grid according to the intersection area weight, thus obtaining the net area density per unit volume of water. .

[0048] Net cage to the first Resistance source terms generated by each computational grid for: (4); In the formula, This refers to the density of seawater.

[0049] When multiple net cages jointly affect the same computational grid, the drag source terms corresponding to each net cage are superimposed. As the flow direction, velocity, and netting adhesion change, the normal and tangential velocity components of the netting and their corresponding drag source terms are updated in real time.

[0050] Using the above method, without performing solid mesh analysis on centimeter-level meshes and wires, the directional influence of cage groups on local flow velocity, water exchange, and material transport can be characterized.

[0051] Further, step 3, the construction of the multiphysics coupled prediction model, includes: A multiphysics field coupled prediction model was established, which includes hydrodynamic module, wave module, water temperature transport module and dissolved oxygen transport module.

[0052] The hydrodynamic module uses water level and depth-average flow velocity as state variables to calculate the flow field under the combined effects of tides, wind stress, bottom friction, wave radiation stress, and cage resistance. The cage resistance source term is added to the hydrodynamic momentum equation according to equation (4).

[0053] The wave module uses the wave action balance equation to calculate the propagation, refraction, shallowing, breaking, bottom friction dissipation and white cap dissipation of wave energy in the frequency and direction space based on wind field, water depth, water level and flow field, and outputs effective wave height, characteristic period, wave direction and radiation stress.

[0054] The hydrodynamic module transmits water level and flow velocity to the wave module, while the wave module feeds back radiation stress and wave enhancement mixing parameters to the hydrodynamic module, thus forming a two-way coupling relationship between waves and tidal currents.

[0055] The water temperature transport module utilizes the water level and flow velocity output from the hydrodynamic module for heat transport, and its depth-averaged control equation is: (5); In the formula, For the total water depth, Water temperature This is the depth-averaged flow velocity vector. Where is the thermal diffusivity, Net heat flux per unit horizontal area The specific heat capacity of seawater at constant pressure. For time, This refers to the density of seawater.

[0056] The dissolved oxygen transport module utilizes hydrodynamic and temperature fields for transport and reaction calculations. Its depth-averaged governing equation is: (6); In the formula, Dissolved oxygen concentration, The dissolved oxygen diffusion coefficient, The reoxygenation coefficient is the water vapor re-oxygenation coefficient. Due to water temperature and salinity A determined saturated dissolved oxygen concentration, For oxygen consumption during respiration in cultured organisms, The oxygen consumption is due to the decomposition of uneaten feed and feces. This refers to the oxygen consumption item in the sediment.

[0057] The oxygen consumption of aquaculture organisms is determined based on the aquaculture biomass, water temperature, and growth stage, while the oxygen consumption of uneaten feed and feces decomposition is determined based on the amount of feed, feed utilization rate, and organic matter degradation rate.

[0058] Water temperature is coupled with dissolved oxygen processes by influencing saturated dissolved oxygen concentration, reoxygenation processes, and biological metabolic rates.

[0059] Therefore, this embodiment adopts a multi-process coupling structure of "wave-tidal current bidirectional coupling, hydrodynamic-water temperature-dissolved oxygen sequential coupling, and water temperature-dissolved oxygen feedback correction", rather than calculating each physical process independently.

[0060] Furthermore, dynamically adjusting the coupling frequency between the wave process and the hydrodynamic process includes: The time-varying degree of the environmental state is determined by the weighted sum of the relative changes of at least two variables among the significant wave height, flow velocity, water level, and wind speed at adjacent time points; The deviation of the wave-current coupling effect is determined based on the change in wave radiation stress and the difference between the flow velocity at the previous call time and the current hydrodynamic flow velocity. The comprehensive scheduling index is composed of a weighted average of the time-varying degree of the environmental state and the deviation of the wave-current coupling effect; The preset threshold includes a startup threshold and a recovery threshold, wherein the recovery threshold is less than the startup threshold; When the comprehensive scheduling index continuously reaches the start threshold, it switches to high-frequency coupling state; when the comprehensive scheduling index continuously falls below the recovery threshold, it returns to low-frequency coupling state; the coupling interaction frequency is limited to between a preset minimum frequency and a maximum frequency.

[0061] Specifically, step 4 involves adaptive wave-current coupling scheduling based on state changes and coupling defects, such as... Figure 5 ,include: To balance computational efficiency under normal sea states with computational accuracy under rapidly changing sea states, marine environmental state change indicators are calculated at each candidate coupling time point. (i.e., the time-varying degree of environmental conditions) and wave-current coupling defect index (i.e., the deviation of wave-current coupling effect).

[0062] Marine environmental status change indicators for: (7); In the formula, At least significant wave height, flow velocity, water level, and wind speed should be included. For the corresponding weights, and , To prevent positive numbers with a denominator of zero.

[0063] Wave-current coupling defect index for: (8); In the formula, Let be the wave radiation stress tensor. Given the current hydrodynamic velocity, The flow rate used during the last call to the wave module is represented by the weighting coefficient.

[0064] according to and Constructing comprehensive scheduling indicators I : (9); In the formula, Weights for state change indicators.

[0065] Next coupling switching interval Determined according to the following formula: (10); In the formula, For the target change indicator, To adjust the index, This means limiting the calculation results to the minimum exchange interval. and maximum exchange interval between.

[0066] When the comprehensive scheduling index When the activation threshold is continuously reached, it enters an unsteady high-frequency coupling state, and the wave module and hydrodynamic module repeatedly exchange water level, flow velocity and radiation stress at minimum or small exchange intervals. When the comprehensive scheduling index When the frequency continuously falls below the recovery threshold, and the recovery threshold is less than the start threshold, the quasi-steady-state low-frequency coupling state is restored. By setting different start and recovery thresholds, a hysteresis interval is formed to avoid frequent switching of the coupling mode near the critical state.

[0067] In the quasi-steady state, the wave module recalculates at relatively long intervals, and the wave parameters between adjacent exchange times are passed by either holding or interpolation. In the non-steady state, the wave module updates according to the shortened exchange interval to improve the response capability to typhoons, strong winds and waves, rapid water rise and abnormal strong current processes.

[0068] Further, step 5, local observation correction based on error covariance constraints, includes: At each observation update time, the model-predicted values ​​of water level, flow velocity, significant wave height, water temperature, and dissolved oxygen are interpolated to the buoy or monitoring station location to form the background conditions. The quality-controlled field observations are used to construct the observation vector. Establish the background error covariance matrix. and observation error covariance matrix .

[0069] Calculate and analyze the state using the local optimal interpolation method : (11); In the formula, This represents the background state of the model. For the observation vector, For the observation operator, The background error covariance matrix, Let be the observation error covariance matrix.

[0070] The background error covariance matrix is ​​locally corrected according to the spatial distance between the observation point and the model point, the difference in water depth, the coastal barrier, and the direction of the main current. When the observation data is insufficient, the local constraint correction degenerates into a scalar weighted correction that decreases with the increase of spatial distance, time difference, and observation error.

[0071] Specifically, the background error covariance decreases with the spatial distance between the cage location and the observation station, and can be locally corrected based on the obstruction of the coastline, differences in water depth, and the direction of the main current.

[0072] Its spatial correlation function can be expressed as: (12); In the formula, and The standard deviation of the background error for the corresponding state variable. The distance between the two positions. For spatially relevant scales, Due to the difference in water depth, For water depth related scales, To account for coastal barriers and the localization coefficient of the mainstream direction, For position i With position j The background error covariance between the corresponding state variables.

[0073] When only a single buoy is set or the number of observations is insufficient to construct a complete matrix, equation (11) is degenerated into a scalar weighted correction based on spatial distance, time difference, and observation error. For missing observations, the time decay value of the nearest effective residual is used, and for abnormal observations, the observation error variance is increased or not included in the correction.

[0074] The corrected analysis state is rewritten into the initial field of the fine model or used to correct the cage location forecast sequence, thereby reducing the impact of regional boundary errors, local topographic errors and model parameter errors on the forecast results of specific aquaculture areas.

[0075] Furthermore, the risk factors include significant wave height, flow velocity, water temperature, and dissolved oxygen; The degree of exceeding the threshold refers to the normalization of the predicted value of the risk element relative to the risk initiation threshold and the high-risk threshold; The duration of exceeding the threshold is the ratio of the continuous or cumulative time exceeding the threshold within the forecast period to the total forecast duration; The structural safety risk is obtained by aggregating the significant wave height risk, flow velocity risk, and wave-current combined load risk. The wave-current combined load risk is determined based on the cage structure parameters and the wave-current combined force. The ecological stress risks are obtained by combining high temperature risk and low dissolved oxygen risk. The overall risk is the larger of the structural safety risk and the ecological stress risk.

[0076] Specifically, step 6, the joint calculation of structural safety risk and ecological stress risk, includes: The corrected effective wave height, flow velocity, water temperature and dissolved oxygen time series are extracted at each cage location, and the combined wave and current load is calculated in combination with the cage structural parameters.

[0077] For the Risk factors are defined by normalized over-threshold magnitude. : (13); In the formula, Forecast values ​​for risk factors, As the risk trigger threshold, The threshold is considered high-risk. Significant wave height, flow velocity, and water temperature are upper-limit risk variables, while dissolved oxygen is a lower-limit risk variable.

[0078] After considering the duration of exceeding the threshold, the first Adjusted risk value of risk-like factors for: (14); In the formula, The duration of consecutive or cumulative exceedances of the threshold within the forecast period. To predict the total duration, This is a duration correction factor.

[0079] Structural safety risks Based on effective wave height risk Flow rate risk Risk of combined wave and current load calculate: (15); Ecological stress risk Based on high temperature risk and low dissolved oxygen risk calculate: (16); Comprehensive Risks The conservative principle was adopted to determine that: (17); Based on comprehensive risk Different levels of early warning are issued for each section, while structural safety risks are also retained. and ecological stress risks The risk levels are determined by classifying the risk components and identifying the elements that contribute the most to the overall risk as the dominant risk factors.

[0080] Preferably, multiple forecast members can be constructed using disturbance wind fields, offshore waves, tidal boundaries, netting attachment rate, and oxygen consumption parameters, and the probability of exceeding each risk threshold can be calculated. The probability of exceeding the threshold is used as the warning confidence level and output simultaneously with the risk level to reduce the impact of a single deterministic forecast result on risk assessment.

[0081] The early warning information includes the effective wave height, flow velocity, water temperature, dissolved oxygen, structural safety risk, ecological stress risk, comprehensive risk level, dominant risk factor, expected start time, expected peak time, expected duration, and forecast confidence level for each cage or cage section.

[0082] Furthermore, the method also includes step 7: rolling forecasting, early warning issuance, and feedback calibration. Weather forecasts, offshore boundary data, and field observation data are updated on a rolling basis according to a preset cycle. After the forecast ends, the actual observation results are compared with the forecast results, and the background error covariance, the observation error covariance, the preset threshold, and the cage resistance parameters are updated accordingly.

[0083] Specifically, according to the preset cycle, weather forecasts, offshore boundary and buoy data are reacquired, the initial state and boundary conditions of the model are updated, and steps 1 to 6 are repeated to form a rolling forecast for seven to fourteen days.

[0084] For each cage or cage section, output the effective wave height, flow velocity, water temperature, dissolved oxygen, structural safety risk, ecological stress risk, overall risk, dominant risk factor, expected start time, expected peak time, expected duration, and forecast confidence level.

[0085] The early warning results are released through the enterprise's smart aquaculture platform, web terminal, application programming interface, or messaging terminal. Based on different dominant risks, recommended measures are provided, such as reinforcing net cages, inspecting moorings, ceasing offshore operations, reducing feeding, increasing oxygenation, early harvesting, or evacuating personnel.

[0086] After the forecasting process is completed, the actual observation results, enterprise response records, and disaster results are compared with the forecast results to update the background error covariance, observation error parameters, adaptive coupling threshold, and risk threshold, thereby achieving continuous calibration of model parameters and early warning rules.

[0087] This embodiment also provides an adaptive coupled forecasting and hierarchical early warning system for wave, current, temperature, and oxygen in deep-water cage aquaculture areas, including: The data acquisition and processing module is used to acquire and process multi-source data from the target sea area; A multi-scale nested module is used to establish a nested relationship between the regional scale calculation domain and the fine-scale calculation domain of the aquaculture area, and to transfer the regional scale calculation results to the fine-scale calculation domain of the aquaculture area. The cage resistance parameterization module is used to treat the netting as an anisotropic porous resistance medium and construct the sub-grid anisotropic resistance source term of the cage group in the fine-scale computational domain of the aquaculture area. The multiphysics coupling prediction module is used to perform coupled calculations of hydrodynamic processes, wave processes, temperature transport processes, and dissolved oxygen transport processes; An adaptive scheduling module is used to construct a comprehensive scheduling index based on the time-varying degree of environmental conditions and the deviation of wave-current coupling effect, and to dynamically adjust the coupling interaction frequency between wave process and hydrodynamic process by comparing the comprehensive scheduling index with a preset threshold. The observation correction module is used to perform local constraint correction on the model prediction results based on the background error covariance and the observation error covariance, using field observation data to obtain the correction prediction information of the cage location. The risk assessment module is used to calculate structural safety risks and ecological stress risks based on the degree and duration of each risk factor exceeding the threshold, and to determine the comprehensive risk level. The early warning output module is used to output early warning information.

[0088] Example 2 The following is combined Figures 1 to 7 The following specific embodiments further illustrate this technical solution. The parameters, time intervals, grid scales, and risk thresholds described below are examples provided for the purpose of illustrating this technical solution and can be adjusted according to the aquaculture area, cage structure, aquaculture species, observation conditions, and computing resources. Equivalent substitutions made to the model type, data interface, parameter expressions, and early warning classification methods without departing from the technical concept of this invention are all within the scope of the technical concept of this invention.

[0089] like Figure 1 As shown, this embodiment operates in the following order: "data acquisition and quality control—multi-scale computational domain construction—cage resistance parameterization—wave-current-temperature-oxygen coupling forecast—adaptive scheduling—local observation correction—joint risk assessment—rolling release and feedback calibration." First, static data such as coastline, water depth, seabed sediment, and substrate roughness are read. Environmental data such as tide level, current velocity, wind field, open sea waves, heat flux, salinity, and dissolved oxygen are also read. Simultaneously, buoy observations, cage structural parameters, and aquaculture production data are integrated.

[0090] like Figure 2 As shown, the data acquisition and processing module performs coordinate transformation, time benchmark unification, range verification, rate of change verification, outlier identification, and missing measurement marking on the acquired data. The processed data sequentially enters the cage resistance parameterization module, the multiphysics coupling prediction module, the adaptive scheduling module, the observation correction module, and the risk assessment module. The early warning output module generates cage location forecasts, spatial risk maps, risk levels, and disposal recommendations. The feedback calibration module updates the model error parameters, coupling thresholds, and risk thresholds based on the measured results and disposal records.

[0091] This embodiment sets the forecast window to 7–14 days, and updates meteorological boundaries, offshore boundaries, and on-site observations on a rolling basis according to a preset cycle. The result output interval can be set to 1 hour or shorter. Virtual result points are deployed at the center, edge, and periphery of each cage, outputting significant wave height, current velocity, water temperature, dissolved oxygen, and risk sequences. The system can be deployed on a local server, scientific computing platform, or cloud node, and the results are published to aquaculture enterprises through web terminals, application programming interfaces, and messaging terminals.

[0092] Construction of the equivalent resistance field of the cage: like Figure 3As shown, a regional model covering the open sea boundary, main coastline, and target aquaculture area is first established. The regional model uses unstructured or structured grids, with a relatively coarse grid in the open sea area and progressively finer grids in areas of coastline transition, large water depth gradient, and the periphery of the aquaculture area. Subsequently, a fine model is established covering the aquaculture area and its adjacent exchange waters, and further fined around the cage clusters, so that the grid can characterize local velocity gradients, wave deformation, and differences in temperature and oxygen transport.

[0093] The water level, velocity components, significant wave height, characteristic period, wave direction, water temperature, and dissolved oxygen output from the regional model are transferred to the refined model through spatiotemporal interpolation. Scalars are interpolated using distance-weighted or conservation interpolation, vectors are interpolated separately in a unified coordinate system, and boundary time series are interpolated using linear or higher-order interpolation. Continuity checks are performed on water level, flow rate, and variable concentrations on both sides of the boundary, thus forming a dynamic boundary transfer relationship from regional-scale forecasts to refined calculations in the aquaculture area.

[0094] The refined model and the regional model use the same coordinate reference, vertical reference, physical quantity units, and forecast start time. The refined model sets result extraction points at the center of the cages, the upstream side, the downstream side, and the edge of the cage group, and performs quality checks on the water level phase, flow velocity direction, wave energy, and temperature and oxygen background values ​​before and after the transfer. When the boundary abrupt change exceeds the preset range, the boundary sequence is smoothed or rolled back to the nearest effective regional forecast result.

[0095] Adaptive switching of wave-current coupling interval: This embodiment uses a circular deep-water net cage as an example, with a cage diameter of 8m and a mesh size of 0.04m. The net depth, cleaned net wire diameter, average thickness of attached organisms, net deformation state, and net cage arrangement are read. The equivalent net wire diameter after attachment is calculated according to formula (1), and the net thickness is calculated according to formula (2). The attachment thickness can be updated from net cleaning records, underwater image recognition results, or periodic manual measurement results.

[0096] The side and bottom nets are discretized into several net surfaces. The relative velocity is determined based on the local flow velocity and the net movement velocity, and decomposed into normal and tangential components according to equation (3). The normal drag coefficient and tangential drag coefficient are determined based on the net Reynolds number, net solidity, adhesion degree and deformation state. Then, the projected area density of each net surface in the intersecting calculation grid is calculated, and an anisotropic drag source term with normal drag as the dominant component and tangential drag as the minor component is constructed according to equation (4). Figure 4 The paper illustrates the equivalent parameters of the mesh, the mapping of the projected area, and the formation process of the drag source term.

[0097] During hydrodynamic calculations, the resistance source term is updated based on flow direction, velocity, netting morphology, and attachment status. When multiple net cages act on the same computational grid, the projected area density and resistance contribution of each net cage are superimposed. The resistance coefficient can be calibrated using on-site velocity observations on the upstream side, inside the cage, and downstream side of the net cage, ensuring that the simulated velocity attenuation rate matches the actual hydrodynamic response of the net cage group, while avoiding solid mesh analysis for centimeter-level meshes and netting lines.

[0098] Water temperature and dissolved oxygen calculations and buoy residual correction: The hydrodynamic module calculates the water level and flow velocity under the combined effects of tide level, tidal current, wind stress, bottom friction, wave radiation stress, and cage resistance; the wave module calculates wave generation, propagation, refraction, shallowing, breaking, bottom friction, and whitecap dissipation, and feeds back radiation stress and wave enhancement mixed parameters to the hydrodynamic module. The water temperature transport module calculates heat transport according to equation (5), and the dissolved oxygen transport module calculates reoxygenation, aquaculture organism respiration, decomposition of uneaten feed and feces, and bottom sediment oxygen consumption according to equation (6), thus forming a two-way coupling of wave and current and a sequential coupling of hydrodynamics, water temperature, and dissolved oxygen.

[0099] At each candidate coupling moment, the environmental state change index is calculated according to Equation (7), the wave-current coupling defect index is calculated according to Equation (8), and the comprehensive scheduling index is formed according to Equation (9). A start threshold and a recovery threshold are set, with the recovery threshold < the start threshold, and a minimum dwell time is set. When the comprehensive scheduling index continuously reaches the start threshold, non-steady-state high-frequency coupling is entered. When the comprehensive scheduling index continuously falls below the recovery threshold and meets the minimum dwell time, quasi-steady-state low-frequency coupling is restored to avoid frequent mode switching in the critical interval.

[0100] Risk classification and early warning output: like Figure 6 As shown, at each observation update time, the model results of water level, flow velocity, significant wave height, water temperature, and dissolved oxygen are interpolated to the buoy or monitoring station location to form the background state. The observation vector will be composed of field observations obtained through quality control. And establish observation operators Background error covariance matrix and observation error covariance matrix The state is calculated using equation (11). Using Equation (12), the background error correlation is localized according to spatial distance, water depth difference, coastal obstruction and mainstream direction.

[0101] When multiple observation stations are deployed and The correction is determined based on historical residual statistics, instrument accuracy, and quality indicators. When there is only a single buoy or insufficient observations, the matrix correction is degraded to a scalar-weighted correction based on distance, time difference, and observation error. For abnormal observations, the corresponding observation error variance is increased or excluded from correction. For periods with missing data, the time decay value of the nearest effective residual is used to prevent local anomalies from being unreasonably propagated to all cage locations.

[0102] The corrected analysis state can be written back to the refined model as the initial field for the next rolling cycle, or it can be directly used to correct the cage location forecast sequence. For each variable, the system limits the maximum influence radius and maximum effective time of the observation residuals, and checks the continuity of water level, consistency of flow velocity direction, reasonable range of temperature, and non-negative dissolved oxygen constraint after correction, thus forming a cage location forecast sequence with local observation constraints.

[0103] System Deployment: like Figure 7 As shown, the system extracts the corrected effective wave height, flow velocity, water temperature, and dissolved oxygen sequence at each cage location, and calculates the combined wave and current load by combining the cage diameter, netting area, drag coefficient, and relative flow velocity. The structural safety threshold is determined by the cage design documents, mooring capacity, and enterprise safety regulations, while the ecological threshold is determined by the aquaculture species, growth stage, and historical production records. The normalized exceedance amplitude is calculated according to equation (13), the continuous or cumulative exceedance duration is introduced according to equation (14), and the structural safety risk is obtained according to equations (15) to (17). Ecological stress risks and comprehensive risks .

[0104] The overall risk is classified into four levels: blue, yellow, orange, and red, and the system simultaneously outputs sub-risks, dominant risk factors, estimated start time, estimated peak time, estimated duration, and forecast confidence level. Based on the dominant risk, recommendations for handling the situation are generated, including cage reinforcement, mooring inspection, cessation of offshore operations, reduction of feeding, enhanced oxygenation, early harvesting, or personnel evacuation. After the forecast period ends, the actual observations, handling records, and disaster results are compared with the forecast results to update the forecast. , Adaptive scheduling thresholds, resistance parameters, and risk thresholds form a closed loop of rolling forecasts, early warning issuance, and feedback calibration.

[0105] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for adaptive coupled forecasting and graded early warning of waves, currents, temperature, and oxygen in deep-water cage aquaculture areas, characterized in that... include: Acquire multi-source data of the target sea area, establish a nested relationship between the regional scale calculation domain and the fine-scale calculation domain of the aquaculture area, and transfer the regional scale calculation results to the fine-scale calculation domain of the aquaculture area as boundary constraints. The netting is equivalent to an anisotropic porous resistance medium, and a subgrid anisotropic resistance source term for the net cage group is constructed in the fine-scale computational domain of the aquaculture area. A multiphysics coupled prediction model is established, which includes hydrodynamic processes, wave processes, temperature transport processes, and dissolved oxygen transport processes. The hydrodynamic processes and wave processes exchange information bidirectionally. The hydrodynamic processes drive the transport of temperature and dissolved oxygen, and the temperature participates in determining the dissolved oxygen saturation concentration and biological oxygen consumption rate. Based on the time-varying nature of the marine environment and the deviation of the wave-current coupling effect, a comprehensive scheduling index is constructed. The frequency of coupling and interaction between the wave process and the hydrodynamic process is dynamically adjusted by comparing the comprehensive scheduling index with a preset threshold. Based on the background error covariance and the observation error covariance, the model prediction results are locally constrained and corrected using field observation data to obtain the corrected prediction information of the cage locations. Based on the corrected forecast information, the degree and duration of exceeding the threshold of each risk element are extracted, and they are aggregated to form structural safety risk and ecological stress risk, respectively. A comprehensive risk is determined based on the structural safety risk and the ecological stress risk, and early warning information is output based on the comprehensive risk.

2. The adaptive coupled forecasting and graded early warning method for wave, current, temperature, and oxygen in deep-sea cage aquaculture areas according to claim 1, characterized in that, The multi-source data includes static basic data and dynamic environmental data; The static basic data includes at least the coastline, water depth, seabed type, and seabed roughness; the dynamic environmental data includes at least the tide level, current velocity, wind speed, wind direction, air pressure, wave spectrum, surface heat flux, salinity background field, and dissolved oxygen background field. The fine-scale calculation domain of the aquaculture area sets up locally denser grids in the aquaculture area, shoreline transition area, water depth gradient change area, and around the cage group.

3. The adaptive coupled forecasting and graded early warning method for wave, current, temperature, and oxygen in deep-sea cage aquaculture areas according to claim 1, characterized in that, Constructing the subgrid anisotropic drag source term includes: The net density is calculated based on the mesh size, net wire diameter, and thickness of attached organisms. The normal and tangential velocity components are determined based on the velocity of the water relative to the net. The normal and tangential drag coefficients are determined based on the net Reynolds number, net density, degree of attachment, and net deformation state, respectively. The netting projected area is mapped to the netting area density per unit water volume according to the intersection weight with the computational grid, thus generating an anisotropic drag source term; When multiple cages act on the same computational grid, the resistance source terms corresponding to each cage are superimposed.

4. The adaptive coupled forecasting and graded early warning method for wave, current, temperature, and oxygen in deep-water cage aquaculture areas according to claim 3, characterized in that, The equivalent network diameter after attachment by organisms is The solidity of the square mesh is ,in, To clean the diameter of the network cable, The average thickness of biological adhesion on the surface of the network cable. m The mesh side length; No. Resistance source terms within a computational grid for: ； In the formula, The density of seawater, The netting area density per unit volume of water. The normal drag coefficient of the mesh is... The tangential drag coefficient of the mesh is... For the normal component of the netting, This represents the tangential component of the mesh.

5. The adaptive coupled forecasting and graded early warning method for wave, current, temperature, and oxygen in deep-sea cage aquaculture areas according to claim 1, characterized in that, Dynamically adjusting the coupling frequency between the wave process and the hydrodynamic process includes: The time-varying degree of the environmental state is determined by the weighted sum of the relative changes of at least two variables among the significant wave height, flow velocity, water level, and wind speed at adjacent time points; The deviation of the wave-current coupling effect is determined based on the change in wave radiation stress and the difference between the flow velocity at the previous call time and the current hydrodynamic flow velocity. The comprehensive scheduling index is composed of a weighted average of the time-varying degree of the environmental state and the deviation of the wave-current coupling effect; The preset threshold includes a startup threshold and a recovery threshold, wherein the recovery threshold is less than the startup threshold; When the comprehensive scheduling index continuously reaches the start threshold, it switches to high-frequency coupling state; when the comprehensive scheduling index continuously falls below the recovery threshold, it returns to low-frequency coupling state; the coupling interaction frequency is limited to between a preset minimum frequency and a maximum frequency.

6. The adaptive coupled forecasting and graded early warning method for wave, current, temperature, and oxygen in deep-sea cage aquaculture areas according to claim 1, characterized in that, In the local constraint correction, the analysis state satisfies: ， In the formula, This represents the background state of the model. For the observation vector, For the observation operator, The background error covariance matrix, The observation error covariance matrix; For analysis of the status; The background error covariance matrix is ​​locally corrected according to the spatial distance between the observation point and the model point, the difference in water depth, the coastal barrier, and the direction of the main current. When the observation data is insufficient, the local constraint correction degenerates into a scalar weighted correction that decreases with the increase of spatial distance, time difference, and observation error.

7. The adaptive coupled forecasting and graded early warning method for wave, current, temperature, and oxygen in deep-sea cage aquaculture areas according to claim 1, characterized in that, The risk factors include significant wave height, flow velocity, water temperature, and dissolved oxygen. The degree of exceeding the threshold refers to the normalization of the predicted value of the risk element relative to the risk initiation threshold and the high-risk threshold; The duration of exceeding the threshold is the ratio of the continuous or cumulative time exceeding the threshold within the forecast period to the total forecast duration; The structural safety risk is obtained by aggregating the significant wave height risk, flow velocity risk, and wave-current combined load risk. The wave-current combined load risk is determined based on the cage structure parameters and the wave-current combined force. The ecological stress risks are obtained by combining high temperature risk and low dissolved oxygen risk. The overall risk is the larger of the structural safety risk and the ecological stress risk.

8. The adaptive coupled forecasting and graded early warning method for wave, current, temperature, and oxygen in deep-water cage aquaculture areas according to claim 7, characterized in that, Multiple forecast members are constructed by taking at least one of the following: perturbation wind field, offshore waves, tidal boundary, netting attachment parameters and oxygen consumption parameters. The proportion of members whose risk thresholds are exceeded is used as the forecast confidence level, and the forecast is output synchronously with the comprehensive risk. The early warning information includes the effective wave height, flow velocity, water temperature, dissolved oxygen, structural safety risk, ecological stress risk, comprehensive risk level, dominant risk factor, expected start time, expected peak time, expected duration, and forecast confidence level for each cage or cage section.

9. The adaptive coupled forecasting and graded early warning method for wave, current, temperature, and oxygen in deep-sea cage aquaculture areas according to claim 1, characterized in that, Also includes: Weather forecasts, offshore boundary data, and on-site observation data are updated on a rolling basis according to a preset cycle; After the forecast is completed, the actual observation results are compared with the forecast results, and the background error covariance, the observation error covariance, the preset threshold, and the cage resistance parameters are updated.

10. A wave-current-temperature-oxygen adaptive coupling forecasting and graded early warning system for deep-sea cage aquaculture areas, used to implement the wave-current-temperature-oxygen adaptive coupling forecasting and graded early warning method for deep-sea cage aquaculture areas as described in any one of claims 1-9, characterized in that, include: The data acquisition and processing module is used to acquire and process multi-source data from the target sea area; A multi-scale nested module is used to establish a nested relationship between the regional scale calculation domain and the fine-scale calculation domain of the aquaculture area, and to transfer the regional scale calculation results to the fine-scale calculation domain of the aquaculture area. The cage resistance parameterization module is used to treat the netting as an anisotropic porous resistance medium and construct the sub-grid anisotropic resistance source term of the cage group in the fine-scale computational domain of the aquaculture area. The multiphysics coupling prediction module is used to perform coupled calculations of hydrodynamic processes, wave processes, temperature transport processes, and dissolved oxygen transport processes; An adaptive scheduling module is used to construct a comprehensive scheduling index based on the time-varying degree of environmental conditions and the deviation of wave-current coupling effect, and to dynamically adjust the coupling interaction frequency between wave process and hydrodynamic process by comparing the comprehensive scheduling index with a preset threshold. The observation correction module is used to perform local constraint correction on the model prediction results based on the background error covariance and the observation error covariance, using field observation data to obtain the correction prediction information of the cage location. The risk assessment module is used to calculate structural safety risks and ecological stress risks based on the degree and duration of each risk factor exceeding the threshold, and to determine the comprehensive risk level. The early warning output module is used to output early warning information.

Images