A method, computer device and program product for predicting oceanic hypoxia

Abstract

The present disclosure provides a method, device, computer equipment and program product for predicting marine hypoxia, wherein the method comprises: obtaining a mode-driven input field of a target sea area at different times; using a plurality of coupled models in a preset marine hypoxia prediction model to respectively identify and predict the mode-driven input field at different times, to obtain a three-dimensional marine physical state field, a biogeochemical state field and a three-dimensional dissolved oxygen concentration state field of the target sea area within a first preset time period in the future; generating marine environment prediction information of the target sea area within the first preset time period in the future according to a target time interval and rolling broadcasting; extracting a target dissolved oxygen concentration state field matched with a preset region of the target sea area from the three-dimensional dissolved oxygen concentration state field, and determining hypoxia indication information in the preset region and generating visual hypoxia prediction information according to the target dissolved oxygen concentration state field and different hypoxia thresholds.

Description

Technical Field

[0001] This disclosure relates to the field of environmental monitoring and prediction technology, and more specifically, to a method, apparatus, computer equipment, and program product for predicting marine hypoxia. Background Technology

[0002] In recent years, the increasing eutrophication in nearshore waters has led to a growing problem of oxygen depletion in the bottom waters, posing a serious threat to nearshore marine ecosystems and fishery resources. Therefore, there is an urgent need for methods to monitor and provide early warnings for dissolved oxygen in nearshore bottom waters in order to proactively address and resolve this issue. However, current monitoring and early warning methods often suffer from insufficient spatiotemporal coverage of monitoring data and untimely warnings, exhibiting significant drawbacks. Summary of the Invention

[0003] This disclosure provides at least one method, apparatus, computer device, and program product for predicting marine hypoxia.

[0004] In a first aspect, embodiments of this disclosure provide a method for predicting marine hypoxia, comprising:

[0005] The model-driven input fields for the target sea area at different times are obtained. These input fields include at least the atmospheric forcing field corresponding to meteorological forecast data within a first preset time period, historical model initial fields, historical boundary fields at a second preset time period, and runoff files. The model initial field characterizes the initial state of marine physical and ecological environmental variables in the target sea area. The boundary field characterizes the average scale boundary of the marine physical and ecological environmental variables. The runoff files indicate the runoff attribute information of the target sea area. Multiple coupled models in a preset marine hypoxia forecasting model are used to identify and predict the model-driven input fields at different times to obtain the target sea area's... The system will generate and broadcast marine environmental forecast information for the target sea area within the first preset time period, based on the three-dimensional marine physical state field, biogeochemical state field, and three-dimensional dissolved oxygen concentration state field, according to the target time interval. It will also extract the target dissolved oxygen concentration state field that matches a preset area of ​​the target sea area from the three-dimensional dissolved oxygen concentration state field, and determine the hypoxia indication information under the preset area based on the target dissolved oxygen concentration state field and different hypoxia thresholds. Finally, it will generate visualized hypoxia forecast information according to the target time interval and the hypoxia indication information.

[0006] Secondly, embodiments of this disclosure also provide a marine hypoxia prediction device, comprising:

[0007] The acquisition module is used to acquire the model-driven input fields of the target sea area at different times. The model-driven input fields at different times include at least the atmospheric forcing field corresponding to the meteorological forecast data within a future first preset time period, the historical model initial field, the boundary field at a historical second preset time period, and the runoff file. The model initial field is used to characterize the initial state of the marine physical and ecological environmental variables of the target sea area. The boundary field is used to characterize the average scale boundary of the marine physical and ecological environmental variables. The runoff file is used to indicate the runoff attribute information of the target sea area.

[0008] The prediction module is used to identify and predict the mode-driven input field at different times using multiple coupled models in the preset ocean hypoxia prediction model, so as to obtain the three-dimensional ocean physical state field, biogeochemical state field and three-dimensional dissolved oxygen concentration state field of the target sea area in the first preset time period in the future.

[0009] The first forecast module is used to generate marine environmental forecast information for the target sea area in the future within the first preset time period according to the target time interval, based on the three-dimensional ocean physical state field, biogeochemical state field and three-dimensional dissolved oxygen concentration state field, and to broadcast it on a rolling basis.

[0010] The second forecast module is used to extract the target dissolved oxygen concentration state field that matches the preset area of ​​the target sea area from the three-dimensional dissolved oxygen concentration state field, and determine the hypoxia indication information under the preset area according to the target dissolved oxygen concentration state field and different hypoxia thresholds, and generate visualized hypoxia forecast information according to the target time interval and the hypoxia indication information.

[0011] Thirdly, an optional implementation of this disclosure also provides a computer device, including: a processor and a memory, wherein the memory stores machine-readable instructions executable by the processor, and the processor is used to execute the machine-readable instructions stored in the memory, wherein the machine-readable instructions, when executed by the processor, perform the steps of the first aspect described above.

[0012] Fourthly, an optional implementation of this disclosure also provides a computer program product, including a computer program that, when run, implements the steps of the first aspect described above.

[0013] The marine hypoxia prediction method, apparatus, computer equipment, and program products provided in this disclosure, compared with existing technologies that use buoy observations and satellite remote sensing data for qualitative assessment of short-term marine hypoxia, suffer from insufficient spatiotemporal coverage of monitoring data and untimely early warning, combine atmospheric forcing fields within a future first preset time period, as well as historical model initial fields and boundary fields and runoff files at a historical second preset time period for prediction. This allows for comprehensive consideration of the influence of model input fields at different times on the three-dimensional ocean physical state field, biogeochemical state field, and three-dimensional dissolved oxygen concentration state field, thereby overcoming the problems of insufficient comprehensiveness and real-time availability of marine data used for prediction and laying a reliable foundation for improving prediction accuracy and timeliness. By setting a preset marine hypoxia prediction model containing multiple coupled models, each coupled model can fully consider various dynamic and ecological processes within the target sea area. Combining real-time and comprehensive model input fields for prediction can improve the accuracy and timeliness of predictions for various state fields of the target sea area within a future first preset time period. By generating and continuously broadcasting marine environmental forecast information for the target sea area within a first preset time period based on various state fields according to target time intervals, the future changing trends of marine ecological processes can be intuitively displayed. By extracting the target dissolved oxygen concentration state field under a preset area of ​​the target sea area and combining it with different hypoxia thresholds to generate visualized hypoxia forecast information, the hypoxia situation in the target sea area can be intuitively displayed. This facilitates relevant personnel in timely prevention and treatment of potential marine disasters based on the hypoxia situation, significantly improving the early warning capability for hypoxia disasters in the target sea area.

[0014] To make the above-mentioned objects, features and advantages of this disclosure more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings used in the embodiments will be briefly described below. These drawings are incorporated in and constitute a part of this specification. They illustrate embodiments conforming to this disclosure and, together with the specification, serve to explain the technical solutions of this disclosure. It should be understood that the following drawings only show some embodiments of this disclosure and should not be considered as limiting the scope. Those skilled in the art can obtain other related drawings based on these drawings without creative effort.

[0016] Figure 1 A flowchart of a method for predicting marine hypoxia provided in an embodiment of this disclosure is shown;

[0017] Figure 2 A schematic diagram of the overall framework of a marine hypoxia prediction method provided by an embodiment of this disclosure is shown;

[0018] Figure 3 This diagram illustrates a visual hypoxia forecast information provided by an embodiment of the present disclosure;

[0019] Figure 4 A schematic diagram of another marine hypoxia prediction device provided by an embodiment of this disclosure is shown;

[0020] Figure 5 A schematic diagram of the structure of a computer device provided in an embodiment of this disclosure is shown. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. The components of the embodiments of this disclosure described and shown herein can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this disclosure is not intended to limit the scope of the claimed disclosure, but merely represents selected embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.

[0022] Furthermore, the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein.

[0023] In this article, "multiple or several" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0024] Studies have found that dissolved oxygen in the ocean is a crucial biogenic parameter of the marine environment and a key factor for the maintenance and development of marine ecosystems. A state of hypoxia is generally defined as a dissolved oxygen (DO) concentration < 2.0 mg / L in marine waters. When the DO concentration is < 2.0 mg / L, most aquatic organisms face a severe survival crisis, and seabed trawling becomes ineffective in catching fish and shrimp populations. In recent years, the increasing eutrophication in nearshore waters has led to a higher frequency, wider range, and longer duration of hypoxia in bottom waters, posing potential harm or changes to the marine ecological environment. Furthermore, hypoxia alters the community structure of marine organisms, reducing the abundance of fish and benthic animals, thus affecting fisheries production and causing direct or indirect economic losses. While some research has been conducted on monitoring and early warning of marine hypoxia in recent years, current monitoring and early warning methods mainly utilize buoy observations and satellite remote sensing data, combined with statistical empirical models, to qualitatively assess short-term marine hypoxia, lacking research on prediction and forecasting of marine hypoxia. Current technologies utilize small-scale data from buoy observations and satellite remote sensing for prediction. However, at the data application level, the high cost and low density of buoy deployment result in limited data representativeness. While satellite remote sensing can monitor large areas of the ocean surface, it struggles to penetrate the water to obtain information on hypoxia at the bottom. Furthermore, interference from typhoons and clouds further reduces the reliability of satellite remote sensing data. At the model mechanism level, existing technologies largely rely on traditional empirical statistical models, which are significantly insufficient in depicting the complex coupling processes of ocean dynamics and biogeochemistry. These models often oversimplify key physical processes such as turbulent mixing and advection transport, leading to large errors in simulating marine ecological processes. They also rely too heavily on historical marine data, making them ill-suited for real-time extreme marine conditions. Moreover, they fail to effectively incorporate core ecological processes such as nutrient cycling, phytoplankton dynamics, and organic matter degradation, thus failing to accurately simulate the entire process of hypoxia formation, development, and dissipation. Therefore, existing technologies have significant drawbacks in terms of the spatiotemporal coverage of monitoring data and the timeliness of early warnings.

[0025] Based on the above research, this disclosure provides a method, apparatus, computer equipment, and program product for predicting marine hypoxia. By combining the atmospheric forcing field within a first preset time period, as well as historical model initial fields and boundary fields and runoff files from a second preset time period, prediction can fully consider the influence of model input fields at different times on the three-dimensional marine physical state field, biogeochemical state field, and three-dimensional dissolved oxygen concentration state field. This overcomes the problems of insufficient comprehensiveness and real-time availability of marine data used for prediction, laying a reliable foundation for improving prediction accuracy and timeliness. By setting a preset marine hypoxia prediction model containing multiple coupled models, each coupled model can fully consider various dynamic and ecological processes within the target marine area. Combining real-time and comprehensive model input fields for prediction can improve the accuracy and timeliness of predicting various state fields of the target marine area within the first preset time period. By generating and continuously broadcasting marine environmental forecast information for the target marine area within the first preset time period according to target time intervals and various state fields, the future changing trends of marine ecological processes can be intuitively displayed. By extracting the target dissolved oxygen concentration state field under a preset area of ​​the target sea area and combining it with different hypoxia thresholds to generate visualized hypoxia forecast information, the hypoxia situation of the target sea area can be displayed intuitively. This makes it easier for relevant personnel to prevent and deal with possible marine disasters in a timely manner based on the hypoxia situation, and significantly improves the early warning capability for hypoxia disasters in the target sea area.

[0026] The shortcomings of the above solutions are the result of the inventor's practical experience and careful research. Therefore, the discovery process of the above problems and the solutions proposed in this disclosure below should be considered as the inventor's contribution to this disclosure.

[0027] The technical solutions of this disclosure will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. The components of this disclosure described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this disclosure provided in the accompanying drawings is not intended to limit the scope of the claimed disclosure, but merely to illustrate selected embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.

[0028] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0029] To facilitate understanding of this embodiment, a method for predicting marine hypoxia disclosed in this disclosure will first be described in detail. The execution subject of the marine hypoxia prediction method provided in this disclosure is generally a computer device with certain computing power. The following description uses a terminal device as the execution subject to illustrate the marine hypoxia prediction method provided in this disclosure.

[0030] like Figure 1 The flowchart shown is a method for predicting marine hypoxia provided in an embodiment of this disclosure, which may include the following steps:

[0031] S101: Obtain the model-driven input fields of the target sea area at different times; the model-driven input fields at different times include at least the atmospheric forcing field corresponding to the meteorological forecast data within the first preset time period, the historical model initial field, the historical boundary field at the second preset time period, and the runoff file; the model initial field is used to characterize the initial state of the marine physical and ecological environmental variables of the target sea area; the boundary field is used to characterize the average scale boundary of the marine physical and ecological environmental variables; the runoff file is used to indicate the runoff attribute information of the target sea area.

[0032] Here, the marine hypoxia prediction method provided in this disclosure can be applied to environmental prediction scenarios to provide real-time forecasts of dissolved oxygen in target sea areas. The target sea area can be any sea area requiring dissolved oxygen forecasting, and the model initial field is used to indicate the initial state of the ecophysical, chemical, and biological processes driving the evolution of the target sea area.

[0033] The model-driven input field can include different types of input fields, which can correspond to different acquisition times. Specifically, the model-driven input field includes the atmospheric forcing field within a first preset time period, the historical model initial field, the boundary field and runoff file at a second preset time period. The first preset time period can be the next five days or the next seven days, and the second preset time period can be the past 30 years or the past 20 years, etc. The atmospheric forcing field can be an atmospheric forcing field for the next five days constructed based on atmospheric forcing data for the next five days. For example, the atmospheric forcing field can provide a preset ocean hypoxia forecasting model with high spatiotemporal resolution data on wind stress, heat flux, longwave and shortwave radiation, relative humidity, evaporation and precipitation acting on the sea surface over the next five days. For simplicity, the preset ocean hypoxia forecasting model will be referred to as the forecasting model in the following text.

[0034] The model initial field is used to provide the forecast model with the initial state of marine physical and environmental variables of the target sea area at the start of the forecast. Historical model initial fields can be selected based on the forecast process of the forecast model; the specific selection steps will be explained later.

[0035] The boundary field is used to characterize the average-scale boundary of marine physical and ecological environmental variables of the target sea area within a second preset historical time period. This average boundary scale can be a monthly average boundary scale. Taking sea area A as an example, the boundary field of sea area A consists of the eastern open boundary, the southern open boundary, and the western open boundary. The runoff file is used to indicate the runoff attribute information of the target sea area. The boundary field / runoff file under the second preset historical time period can be a boundary field constructed based on the monthly average / annual average open boundary data of the past N years; the runoff file under the second preset historical time period can be runoff data constructed based on the monthly average / annual average runoff data of the past N years. Here, N is a positive integer, and the specific value of N can be selected according to the prediction needs. This application embodiment does not impose specific limitations.

[0036] S102: Using multiple coupled models in the preset marine hypoxia prediction model, the model-driven input fields at different times are identified and predicted to obtain the three-dimensional marine physical state field, biogeochemical state field and three-dimensional dissolved oxygen concentration state field of the target sea area in the first preset time period in the future.

[0037] Here, the pre-trained marine hypoxia prediction model is used to receive input fields driven by different time periods, predicting the future three-dimensional marine physical state field, biogeochemical state field, and three-dimensional dissolved oxygen concentration state field of the target sea area. The prediction model can include multiple coupled models, including a hydrodynamic calculation model, a biogeochemical model, and a dissolved oxygen calculation model.

[0038] A three-dimensional ocean physical state field characterizes the ocean's physical state, including water flow, heat distribution, salinity structure, and turbulent mixing intensity. A biogeochemical state field quantitatively characterizes the dynamic cycling processes and spatiotemporal distribution of biomass of key components in the marine ecosystem driven by the physical environment. A three-dimensional dissolved oxygen concentration state field characterizes the distribution and evolution of dissolved oxygen concentration in a target sea area within three dimensions.

[0039] For example, the acquired model-driven input fields can be input into the ocean hypoxia prediction model. Based on the target model parameters configured later, the hydrodynamic calculation module in the ocean hypoxia prediction model can also simultaneously input the real-time updated three-dimensional ocean physical state field into the biogeochemical module and the dissolved oxygen calculation model. Using the biogeochemical module, numerical predictions are performed based on the input key ecological processes to obtain the biogeochemical state field of the target sea area within a preset future time period. Using the dissolved oxygen calculation model, based on biogeochemical information related to dissolved oxygen production and consumption, as well as the model input field, the prediction of the three-dimensional dissolved oxygen concentration state field with high spatiotemporal resolution within a preset future time period is achieved by solving the dissolved oxygen budget control equation.

[0040] S103: Based on the target time interval, and according to the three-dimensional ocean physical state field, biogeochemical state field, and three-dimensional dissolved oxygen concentration state field, generate marine environmental forecast information for the target sea area in the first preset time period in the future and broadcast it on a rolling basis.

[0041] Here, the target time interval is used to indicate the output time interval of various predicted state fields, and also to indicate the forecast interval for the hypoxia prediction information visualized later. The target time interval can specifically be 1 hour, 2 hours, 6 hours, etc.

[0042] In practice, the marine hypoxia forecasting model outputs a three-dimensional marine physical state field, a biogeochemical state field, and a three-dimensional dissolved oxygen concentration state field according to target time intervals. Using these output state fields, marine environmental forecast information for the target sea area within the first preset time period is generated and displayed to users on a rolling basis.

[0043] S104: Extract the target dissolved oxygen concentration state field from the three-dimensional dissolved oxygen concentration state field that matches the preset area of ​​the target sea area, and determine the hypoxia indication information in the preset area according to the target dissolved oxygen concentration state field and different hypoxia thresholds, and generate visualized hypoxia forecast information according to the target time interval and hypoxia indication information.

[0044] Here, because oxygen depletion events in the ocean often occur in the bottom region, this embodiment further predicts and outputs the dissolved oxygen concentration state field of the bottom region. The preset region can be the bottom region of the target sea area; for example, the preset region can be the sea area corresponding to the grid below a preset layer in the simulation model mesh. The target dissolved oxygen concentration state field is the dissolved oxygen concentration state field of the bottom region.

[0045] The hypoxia threshold is a critical value of dissolved oxygen concentration defined based on the survival needs of marine organisms and the response of the ecosystem. It is typically defined as a hypoxic state when the dissolved oxygen concentration in the water is <2.0 mg / L. For example, this implementation can be configured with hypoxia thresholds of 1 mg / L, 2 mg / L, and 3 mg / L. Hypoxia indication information is used to indicate the hypoxia status of a preset area under different hypoxia thresholds, such as the hypoxia range, hypoxia probability, and the impact of hypoxia on the marine ecosystem.

[0046] In practical implementation, target dissolved oxygen data corresponding to the bottom layer region of the target sea area can be extracted from the three-dimensional dissolved oxygen concentration state field to obtain the target dissolved oxygen concentration state field of the bottom layer region. Then, it is determined whether the target dissolved oxygen concentration state field has a hypoxia problem under different hypoxia thresholds to obtain the hypoxia results of the target sea area under different hypoxia thresholds. The hypoxia results can include sub-regions in the target sea area with hypoxia problems, and the hypoxia data of the sub-regions. The sub-regions can include one or more, and any sub-region can be smaller than or equal to a preset region. For each sub-region, hypoxia indication information is determined based on the hypoxia results, the total prediction duration of the preset marine hypoxia forecast model, the information complexity of the model input field, and the proportion of extreme information in the model input field. After obtaining the hypoxia indication information, it can be visualized to obtain visualized hypoxia forecast information, which is output and displayed to the user according to the target time interval.

[0047] In one embodiment, S101 described above can be implemented according to the following steps:

[0048] S101-1: Determine whether the preset ocean hypoxia prediction model has an output initial field.

[0049] Here, the output initial field of the model can be the latest output three-dimensional ocean physical state field, biogeochemical state field, and three-dimensional dissolved oxygen concentration state field of the preset ocean hypoxia prediction model.

[0050] In practice, since the initial model field that the forecast model can use for the first prediction of the target sea area is different from that used for subsequent predictions, an intelligent judgment mechanism is introduced to ensure that the prediction task is carried out accurately and in an orderly manner. When the model-driven input field of the target sea area is obtained at different times, the mechanism actively judges whether there is an output initial model field and determines the initial model field that the forecast model needs to use for the current prediction based on the judgment result.

[0051] S101-2: If not, then obtain the model input data for the target sea area; the model input data includes meteorological forecast data for the first preset time period in the future, historical satellite remote sensing data and observation data, and historical open boundary data and runoff data for the second preset time period.

[0052] Before constructing the initial field of the model, the model input data needs to be collected and preprocessed. For collecting weather forecast data, officially authorized 5-day weather forecast data can be obtained and downloaded from the internet. Furthermore, since the officially provided weather forecast data is dynamically updated, to ensure that the forecast model can use the latest weather forecast data in a timely manner, the latest officially authorized 5-day weather forecast data can be periodically obtained and downloaded at preset time intervals and updated to the preset ocean hypoxia forecast model. The preset time interval can be 1 hour, 6 hours, etc.

[0053] Historical satellite remote sensing data refers to data obtained from continuous observations of the target sea area over a past period (such as several years) prior to the forecast start time. Historical observational data refers to sea area observation data acquired over a specific historical period prior to the forecast start time. Because the collected observational data may originate from different acquisition devices and using different acquisition methods, quality control and preprocessing of the observational data are necessary.

[0054] Historical open boundary data at the second pre-set time period can be a dataset acquired at a specific time scale (such as monthly average) that can at least describe the marine environmental state at the open boundary connecting the target sea area predicted by the forecast model to the open sea or a larger sea area. Historical runoff data at the second pre-set time period is a raw dataset provided at a specific time scale (such as monthly average) that reflects the characteristics of land-based runoff.

[0055] S101-3: Using interpolation methods, meteorological forecast data is interpolated into a preset three-dimensional grid to obtain the atmospheric forcing field in the first preset time period in the future; the preset three-dimensional grid is a simulation model grid constructed for the target sea area.

[0056] Here, the pre-set 3D mesh can be a structured model formed by discretizing and simulating the target sea area space in 3D according to the target sea area range and forecast accuracy requirements. For example, the interpolation method can be bilinear interpolation, which is to use the linear combination of the function values ​​of the four data points around the interpolation point as the estimated value of the function value of the interpolation point.

[0057] In practical implementation, for the construction of the atmospheric forcing field, bilinear interpolation can be used. Based on the coordinates of the horizontal grid cells of a pre-defined three-dimensional grid, the meteorological forecast data for the next five days is divided into forecast data that can be directly interpolated and forecast data that requires transformation before interpolation. After determining the directly interpolable data and the transformed data, for each grid cell in the pre-defined three-dimensional network, the corresponding forecast data around each grid cell can be located based on the coordinates of that grid cell and the coordinates of each data point. The meteorological parameter values ​​of each grid cell are estimated through two linear interpolation operations, achieving a match between the meteorological data and the horizontal scale of the grid. Subsequently, considering the core requirement of the atmospheric forcing field for driving the air-sea interface, the horizontally interpolated meteorological parameters (such as air pressure) are allocated to the corresponding grid cells of the pre-defined three-dimensional grid, thereby constructing a continuous atmospheric forcing field that is fully adapted to the pre-defined three-dimensional grid and covers the first pre-defined time period.

[0058] S101-4: Using interpolation methods, satellite remote sensing data and observation data are synchronously interpolated to a preset three-dimensional grid to obtain the initial field of the model.

[0059] In practice, the collected satellite remote sensing data and observation data can be used to interpolate various data into the corresponding grid cells in the preset three-dimensional grid, thereby creating the initial field of marine physical and ecological environmental variables required for model calculation.

[0060] S101-5: Extract target boundary data that matches the boundary of the target sea area from the open boundary data, and use the target boundary data to construct the boundary field at the second preset time in history.

[0061] Here, the target boundary data that matches the boundary of the target sea area can be the boundary data located on the boundary of the target sea area in the open boundary data of the preset area, or it can be the boundary data located around the boundary of the target sea area.

[0062] In practice, based on the boundary range of the target sea area, partial open boundary data that spatially matches the boundary range can be selected from the open boundary data. The target open boundary data is then determined based on this partial open boundary data. Next, bilinear interpolation is used to spatially adapt the target boundary data to the boundary grid cells of a pre-defined 3D grid, ensuring that each boundary cell corresponds to an accurate boundary variable value. Finally, the physical and ecological variable data of all boundary cells are integrated to construct the boundary field at the second pre-defined historical time.

[0063] S101-6: Distribute runoff data to the preset three-dimensional grid runoff source region locations to obtain the runoff file for the second preset historical time.

[0064] Here, the location of the runoff source area is a spatial configuration preset in the three-dimensional grid of the model for injecting runoff data, such as specifying the location of a river or a sea area.

[0065] In practice, the latitude and longitude coordinates of each data point in the runoff data can be converted into index coordinates for grid cells. The runoff files are then assigned to the grid cells corresponding to the index coordinates in the preset 3D model, and the runoff parameters of each grid cell are integrated to form a runoff file that is adapted to the preset 3D grid and covers a second preset historical time scale (such as monthly average).

[0066] In another embodiment, regarding the step of obtaining historical model initial fields, if the preset ocean anoxic forecasting model has an already output model initial field, it indicates that the forecast for the target sea area is not the first prediction. In this case, the latest output model initial field can be used as the historical model initial field currently required by the preset ocean anoxic forecasting model. The latest output model initial field is related to the latest output three-dimensional ocean physical state field, biogeochemical state field, and three-dimensional dissolved oxygen concentration state field of the preset ocean anoxic forecasting model. For example, the model initial field output the previous day can be used as the historical model initial field currently required.

[0067] In one embodiment, this application also proposes a parameterization scheme for different sea areas to focus on the ecosystem heterogeneity and hypoxia sensitivity mechanisms of typical sea areas. Specifically, before performing S102 above, the model parameters required for the forecast model to predict for the target sea area can be determined according to the following steps 1-2:

[0068] Step 1: Obtain the pre-built model parameter library; the model parameter library includes different model parameters and operating rules corresponding to different sea areas.

[0069] Here, the model parameter library systematically stores the optimal values ​​of key parameters determined through sensitivity analysis and intelligent optimization for different sea areas, based on the heterogeneity of ecosystems in different sea areas. Specifically, the model parameters can include various biogeochemical parameters of interest to the forecasting model, including but not limited to: water light attenuation coefficient, phytoplankton specific light attenuation rate, CO2 absorption half-saturation concentration of small phytoplankton, CO2 absorption half-saturation concentration of diatoms, nitrate absorption half-saturation concentration of small phytoplankton, nitrate absorption half-saturation concentration of diatoms, oxidation half-saturation concentration, phosphate absorption half-saturation concentration of small phytoplankton, phosphate absorption half-saturation concentration of diatoms, diatom silicate absorption half-saturation constant, feeding half-saturation constant of microzooplankton, feeding half-saturation constant of medium-sized zooplankton, and the initial slope of the PI curve for small phytoplankton. The model parameters include the initial slope of the diatom PI curve, maximum feeding rate of micro-zooplankton, maximum feeding rate of meso-zooplankton, specific mortality rate of meso-zooplankton, feeding efficiency of micro-zooplankton, feeding efficiency of meso-zooplankton, mortality rate of small phytoplankton, mortality rate of diatoms, detritus decomposition rate, aggregation rate, nitrification rate, maximum chlorophyll-to-carbon ratio of small phytoplankton, maximum chlorophyll-to-carbon ratio of diatoms, maximum specific growth rate of small phytoplankton, maximum specific growth rate of diatoms, half-saturated concentration of ammonium salts absorbed by small phytoplankton, half-saturated concentration of ammonium salts absorbed by diatoms, maximum growth rate of phytoplankton, nutrient half-saturation constant, organic matter remineralization rate, and benthic oxygen consumption rate. Different values ​​of these model parameters can reflect the differences in nutrient gradients, phytoplankton community structure, and hypoxia-dominant mechanisms in different sea areas.

[0070] In practice, before using the forecasting model to forecast the target sea area, a pre-built model parameter library can be obtained using the parameter optimization and input module. The parameter optimization and input module can be used to build the model parameter library and to select and configure target model parameters for different sea areas.

[0071] Step 2: Obtain the target model parameters that match the target sea area from the model parameter library, and configure the target model parameters into the preset marine hypoxia forecast model.

[0072] Here, the target model parameters are a set of optimal biogeochemical control parameters selected from a pre-built model parameter library that match the ecological characteristics of the target sea area.

[0073] For example, when a forecasting model needs to predict hypoxia in sea area A, it can automatically retrieve matching items from the model parameter library as target model parameters based on the geographic code or ecological type identifier of sea area A. The selected target model parameters are then configured into the corresponding models in the preset marine hypoxia forecasting models, such as the biogeochemical model and the dissolved oxygen model, thereby significantly improving the regional adaptability of hypoxia forecasting for area A. After configuring the target model parameters, the forecasting model can drive the input field based on the configured target model parameters and the input pattern to output various state fields.

[0074] In one embodiment, the above-mentioned model parameter library is further constructed using the following steps S1 to S4:

[0075] S1: Obtain sample patterns from different sea areas to drive the input field.

[0076] Here, the sample pattern-driven input field can serve as the input field for training the forecast model and building the model parameter library. Different sample pattern-driven input fields can be generated for different sea areas.

[0077] In practice, for different sea areas, historical observation data and future prediction data can be collected and organized to form sample pattern-driven input fields. Then, for each sea area, each sample pattern-driven input field can be input into the preset marine hypoxia forecasting model to be trained for identification and processing to obtain initial prediction results. Then, using the corresponding label prediction results of the sample pattern-driven input fields, the prediction loss of the forecasting model is determined, and the model is iteratively trained using the prediction loss until the training cutoff condition is met, resulting in a trained preset marine hypoxia forecasting model.

[0078] S2: For any sea area, the input field is driven by sample patterns from different sea areas, and sensitivity analysis is performed on each initial model parameter in the preset marine hypoxia forecast model to screen out the parameters to be optimized in different sea areas.

[0079] Here, the initial model parameters can be the default parameters set for the preset marine hypoxia forecasting model before performing region-specific optimization. Sensitivity analysis is used to analyze the sensitivity of various initial model parameters in different sea areas. This sensitivity can indirectly characterize the importance of the initial model parameters in different sea areas. For any given sea area, the parameters to be optimized are the key initial model parameters for the forecasting model when making predictions for that sea area.

[0080] For example, for any sea area, the sample pattern driving input field of that sea area can be input into a preset marine hypoxia prediction model to obtain initial prediction results. Then, using sensitivity analysis methods (such as the Morris method or the Sobol index method), the initial model parameters are adjusted while the prediction task is executed, and the correlation strength between parameter adjustments and output results is quantified. Parameters whose sensitivity indices exceed preset thresholds are automatically selected, forming a list of parameters to be optimized specific to the target sea area. For example, the sensitivity analysis of various initial model parameters can be performed using the following formula (1-1) to obtain the sensitivity of various parameters:

[0081] (1-1)

[0082] Among them, S C,X Indicates sensitivity; C X C represents the average annual biomass of phytoplankton on the sea surface. X % This represents the annual average biomass of phytoplankton on the sea surface after parameter changes; an exemplary proportion can be set to 50%. X represents the value of the initial model parameter that needs to be adjusted. X % These are the parameter values ​​after proportional reduction; the specific limit for setting the ratio is not specified here. For example, each initial model parameter that needs to be adjusted is reduced by 50% (i.e., X). X % =0.5X) Rerun the model to obtain C X % Finally, the values ​​are substituted into the formula to calculate the sensitivity. The higher the sensitivity value, the greater the impact of the parameter on biomass, thus identifying it as a key biogeochemical parameter. For example, if the sensitivity of the maximum specific growth rate parameter of small phytoplankton is significantly higher than that of other parameters, it is considered a key biogeochemical parameter (i.e., a parameter to be optimized).

[0083] S3: Use the target genetic algorithm to optimize the parameters to be optimized in different sea areas to obtain the optimized parameters.

[0084] Here, the target genetic algorithm is an intelligent optimization algorithm adapted to the multi-parameter, nonlinear optimization characteristics of marine hypoxia prediction models. In specific implementation, a multi-objective genetic algorithm can be used to set reasonable physical value ranges for each parameter to be optimized and initialize the population. Then, new parameter combinations are generated iteratively through selection, crossover, and mutation operators. Each iteration uses sample patterns to drive the input field to perform model prediction and uses the fit between the prediction results and the actual data as the fitness function for survival of the fitter. Finally, through multiple generations of evolution, the optimized parameters that minimize the regional prediction error are obtained.

[0085] S4: Construct a model parameter library based on the optimized parameters and initial model parameters corresponding to different sea areas.

[0086] In practice, the parameter sets optimized by genetic algorithms for each sea area can be integrated and compared with the corresponding initial model parameters to generate standardized parameter configuration files containing parameter version identifiers, optimization dates, and sea area characteristic descriptions. Then, using structured database technology, the parameter configuration files for different sea areas are associated and stored with corresponding geographic identifiers, applicable conditions, and operating rules, ultimately constructing a model parameter library that supports rapid retrieval by sea area, version tracking, and dynamic loading.

[0087] In one embodiment, the above-described S2 can be implemented using the following steps:

[0088] S2-1: Input the sample pattern-driven input field into the preset ocean hypoxia prediction model using the initial model parameters to obtain the initial prediction results; the initial prediction results include the predicted three-dimensional ocean physical state field, biogeochemical state field, and three-dimensional dissolved oxygen concentration state field.

[0089] S2-2: Adjust at least one candidate parameter in the initial model parameters, and input the sample pattern of the sea area into the preset marine hypoxia prediction model using the adjusted candidate parameters to obtain the target prediction result.

[0090] Here, candidate parameters are specific parameters selected from all biogeochemical parameters in the preset ocean hypoxia prediction model, which are then subjected to sensitivity analysis. The target prediction result is the set of predicted output data obtained by configuring the adjusted candidate parameters into the preset ocean hypoxia prediction model and driving the input field based on the same sample pattern.

[0091] In practice, at least one key biogeochemical parameter can be selected as a candidate parameter from the initial model parameters, such as the maximum phytoplankton growth rate or the organic matter degradation rate. Subsequently, the value of this candidate parameter is adjusted, for example, by increasing or decreasing it by a specific percentage (e.g., 10%) from its baseline value, and the adjusted candidate parameter is then configured into the prediction model. The sample pattern-driven input field is then fed into the prediction model for prediction, and the target prediction result is output.

[0092] S2-3: Based on the degree of deviation between the initial prediction results and the target prediction results, determine whether the candidate parameters belong to the parameters to be optimized.

[0093] Here, the degree of bias can quantify the impact of changes in candidate parameters on the prediction results of the forecast model.

[0094] In practice, for the same sample pattern driving the input field, the degree of deviation between the initial prediction result and the target prediction result can be determined. This deviation can be quantified using absolute or relative deviation. Based on a preset deviation threshold, it is determined whether the candidate parameter belongs to the parameter to be optimized. If the calculated deviation exceeds the threshold (for example, a relative deviation absolute value greater than 50% is considered to have a significant impact), it indicates that the model is very sensitive to changes in the parameter and should be identified as a parameter to be optimized. If the deviation is below the threshold, it is not considered a parameter to be optimized.

[0095] Furthermore, after obtaining the parameters to be optimized, for the above S3, the following step A1 can be implemented:

[0096] A1: Using a target genetic algorithm, the parameters to be optimized in the sea area are optimized according to the degree of deviation and the importance of the parameters to be optimized.

[0097] Here, the importance of the parameters to be optimized can be determined in real time based on the real-time ecological environment characteristics of the marine area.

[0098] For example, the importance weights of the parameters to be optimized can be determined by utilizing the degree of deviation and importance of the parameters to be optimized. Then, a multi-objective genetic algorithm is used to optimize and adjust the parameters according to the importance weights to obtain the optimized parameters. Specifically, when using a multi-objective genetic algorithm for optimization, the following steps can be taken: First, the importance weights of genetic algorithm parameters such as population size, crossover probability, mutation probability, and maximum number of iterations can be determined by utilizing the degree of deviation and importance of the parameters to be optimized. The default values ​​of the genetic algorithm are then weighted according to the importance weights to obtain the target values ​​for various genetic algorithm parameters. For example, the target value for population size is 20, the target value for crossover probability is 0.60, the target value for mutation probability is 0.10, and the target value for the maximum number of iterations is 1000. Then, different combinations of genetic algorithm parameters are encoded using real numbers to form individuals in the genetic algorithm (each individual corresponds to a set of preset ecological parameter settings for a marine hypoxia prediction model). During the optimization iteration process, the cost function, i.e., the F-value, for each individual is calculated using formula (2-1) to characterize the fitness.

[0099] (2-1)

[0100] Where N is the number of observations, and i represents the i-th observation data point. Predict sea surface chlorophyll concentration (unit: ), Observational data (unit: ) from long-term observational data from official websites and satellite datasets ), This represents the standard deviation of the observed data.

[0101] Optionally, different degrees of deviation can correspond to different optimization strategies. After obtaining the degree of deviation and the importance of the parameters to be optimized, the importance weights of various genetic algorithm parameters can be determined using the importance of the parameters to be optimized. The default values ​​of the genetic algorithm parameters are then weighted according to the importance weights to obtain the target values ​​of various genetic algorithm parameters. Finally, the genetic algorithm is used to optimize the parameters to be optimized according to the optimization strategy matched to the degree of deviation, resulting in the optimized parameters.

[0102] In one embodiment, when using multiple coupled models for prediction in S102 above, the prediction process of different coupled models is different. The prediction process of different coupled models will be described below:

[0103] The prediction process for the hydrodynamic calculation model can be implemented according to the following step B1:

[0104] B1: Using a hydrodynamic calculation model, based on the model-driven input field and three-dimensional ocean fluid dynamics control parameters, simulate the physical operation process in the marine ecosystem to obtain the three-dimensional ocean physical state field of the target sea area in the first preset time period in the future; wherein, the physical operation process includes at least the tidal operation process, the three-dimensional temperature change process, the salinity change process, the flow field change, and the sea surface height change process; the three-dimensional ocean physical state field includes at least the three-dimensional velocity field, the sea surface height field, the temperature field, the salinity field, and the turbulent diffusion coefficient.

[0105] Here, the hydrodynamic calculation model serves as the fundamental physical engine for the prediction process. It can be constructed using a mature Regional Ocean Model System (ROMS) as its core framework, with its core task being the accurate solution of the three-dimensional ocean hydrodynamic governing equations. The solution process meticulously simulates tides, three-dimensional temperature, salinity, and flow fields, ultimately outputting a high spatiotemporal resolution three-dimensional ocean physical state field for a first predetermined future time period. This three-dimensional ocean physical state field can include at least the three-dimensional velocity field (u, v, w), sea surface height field (zeta), temperature field (T), salinity field (S), turbulent diffusion coefficient Kh, and viscosity coefficient Kv. The three-dimensional ocean hydrodynamic governing parameters can specifically be the parameters of each equation in the three-dimensional ocean hydrodynamic governing equations.

[0106] For example, the three-dimensional ocean hydrodynamic governing equations used in the hydrodynamic calculation model include, but are not limited to, the equations shown in Equations 3-1 to 3-6 below:

[0107] (1) The momentum equation describing fluid motion:

[0108] (3-1)

[0109] (3-2)

[0110] in, This represents the partial derivative of u with respect to the time dimension. Represents the ocean current velocity vector. Represents the advection term. represents the Coriolis parameter, and v represents the seawater velocity scalar; , Indicates the source term, Represents the horizontal dissipation term. This represents the partial derivative of u on the z-axis. This represents the average vertical turbulent diffusion term. This represents the partial derivative of the seawater flow velocity over time. This represents the partial derivative of v on the z-axis.

[0111] (2) Vertical static equilibrium equations:

[0112] (3-3)

[0113] in, This represents the gradient of the gravitational potential in the vertical direction. The fluid density is represented by g, and g represents the acceleration due to gravity. This represents the fluid reference density.

[0114] (3) Convection-diffusion equations for temperature and salinity:

[0115] (3-4)

[0116] in, This indicates how temperature and salinity change over time. Indicates the molecular diffusion coefficient. Represents the vertical turbulent diffusion term. Indicates the source term, This represents the horizontal dissipation term.

[0117] (4) Equation of state for seawater:

[0118] (3-5)

[0119] in, The values ​​represent the state of seawater, where T represents temperature, S represents salinity, and P represents pressure.

[0120] (5) Continuity equation:

[0121] (3-6)

[0122] In practical implementation, the acquired initial fields of various modes can be input into the hydrodynamic calculation model. Using the hydrodynamic calculation model, discrete solutions are performed based on the configured set of governing equations and the input information. Within each time step, the following are calculated sequentially: flow field changes driven by wind stress and tidal gravity; three-dimensional temperature transport dominated by sea surface heat flux and energy exchange; salinity diffusion regulated by evaporation precipitation and turbulent mixing; and vertical and horizontal turbulent diffusion parameterized by a turbulent closure scheme. Then, the hydrodynamic calculation model is used to dynamically and interactively calculate the above physical processes by coupling the momentum equation, thermodynamic equation, and salinity transport equation. After completing the calculations for a predetermined first time period (e.g., 120 hours), the hydrodynamic calculation model outputs a gridded dataset as the three-dimensional ocean physical state field.

[0123] The prediction process for biogeochemical models can be achieved through the following step B2:

[0124] B2: Using a biogeochemical model, based on the three-dimensional ocean physical state field, the model-driven input field, and biogeochemical control parameters, numerical simulations are performed on various biogeochemical cycles and species dynamic changes in the marine ecosystem to obtain the biogeochemical state field in the first preset time period in the future.

[0125] Here, the biogeochemical model can operate under the drive of a three-dimensional ocean physical state field, achieving online synchronous coupling calculations with the physical processes. The biogeochemical control parameters can specifically be the parameters of each equation in the biogeochemical governing equation set.

[0126] For example, the biogeochemical governing equations used in the calculation process of the biogeochemical model may include, but are not limited to, the following equation sets 4-1 to 4-4:

[0127] (4-1)

[0128] (4-2)

[0129] (4-3)

[0130] (4-4)

[0131] in, and These represent the dynamic ecological states of phytoplankton functional groups P1 and P2, respectively. and The dynamic ecological states of chlorophyll Chl1 and Chl2 are represented respectively; NPP1 represents new productivity at P1, RPP1 represents reproductive productivity at P1, and G1 represents the amount of food consumed by small zooplankton. This indicates the loss due to the death of phytoplankton P1. This indicates the aggregation of phytoplankton P1 biomass; NPP2 represents the vertical sinking flux at P1; RPP2 represents the new productivity at P2; RPP2 represents the regeneration productivity at P2; and G2 represents the amount of food consumed by medium-sized zooplankton. This indicates the loss due to phytoplankton P2 death. This indicates the aggregation of phytoplankton P2 biomass; This represents the vertical sinking flux of P2. express Primary productivity; These represent the mortality rates for Chl1 and Chl2, respectively. express The amount of food consumed by small zooplankton, 2. Intake by medium-sized zooplankton; , They represent , Biomass aggregation below; , They represent , The vertical sinking flux below.

[0132] The exemplary equations described above integrate new productivity (NP), regenerated productivity (RP), feeding consumption (G-type terms), and mortality loss by simulating the biomass dynamics of different phytoplankton functional groups (P1, P2) and chlorophyll (Chl1, Chl2) in the ocean. Mortality-like items), biomass aggregation ( Aggregate-like terms), vertical sinking flux ( Key processes such as biomass generation, transfer and loss in marine ecosystems are quantitatively characterized, providing a core dynamic description that couples marine ecological processes and biogeochemical cycles.

[0133] In practical implementation, biogeochemical models can be used. Based on the spatiotemporal distribution and patterns of hydrological elements such as temperature, salinity, and ocean currents provided by the three-dimensional ocean physical state field, and external inputs including atmospheric forcing fields and boundary field elements, as well as three major categories of key parameters: nutrients, plankton, and detritus, numerical simulations of the biogeochemical cycles and species dynamics of marine ecosystems can be performed. That is, at each time step, by solving a system of equations, the advection and diffusion of substances (such as nutrients) driven by the three-dimensional physical field, as well as the complex biogeochemical cycles and species dynamic changes in the ocean, can be simulated.

[0134] The prediction process for the dissolved oxygen calculation model can be implemented according to the following step B3:

[0135] B3: Using a dissolved oxygen calculation model, based on the three-dimensional ocean physical field, biogeochemical state field, model-driven input field, and control parameters of dissolved oxygen concentration, a numerical simulation of the dynamic interaction mechanism of dissolved oxygen changes in the marine ecosystem is performed to obtain the three-dimensional dissolved oxygen concentration state field in the first preset time period in the future.

[0136] Here, the dissolved oxygen calculation model, as the core target model for anoxic prediction, is closely coupled with the hydrodynamic calculation model and the biogeochemical model, and calculations are performed simultaneously. The core of the dissolved oxygen calculation model is solving the three-dimensional dissolved oxygen concentration governing equations, thereby obtaining advection transport terms, turbulent diffusion terms, and biochemical source-sink terms. Specifically, the advection transport term indicates the advection transport results of dissolved oxygen in the target sea area; the turbulent diffusion term indicates the turbulent diffusion results of dissolved oxygen in the target sea area; and the biochemical source-sink terms indicate the production and consumption results of dissolved oxygen in the target sea area.

[0137] A three-dimensional dissolved oxygen concentration state field is used to indicate the dissolved oxygen concentration at various locations within a target sea area. It can be characterized using the dissolved oxygen concentration within each grid cell of a pre-defined three-dimensional grid. The control parameters for dissolved oxygen concentration can specifically be the parameters in various dissolved oxygen control equations.

[0138] For example, the governing equations for the three-dimensional dissolved oxygen concentration used when running the calculation process using the dissolved oxygen calculation model include, but are not limited to, the equations shown in 5-1 to 5-2 below:

[0139] (a) Transport-biogeochemical coupling equation:

[0140] (5-1)

[0141] in, The local time change rate of dissolved oxygen (DO) concentration refers to the rate of change of a physical quantity (such as dissolved oxygen concentration, flow velocity, temperature, biomass, etc.) over time at a fixed location in space (a point or small area that does not move with the fluid movement). The horizontal advection transport term (u and v are horizontal velocities) is used to describe the transport process of DO with the horizontal flow field. The term represents the vertical advection transport (where w is the vertical velocity), which describes the transport process of DO with respect to the vertical flow field. This indicates vertical turbulent diffusion and is used to characterize vertical mixing of DO caused by turbulence; represents the vertical turbulent diffusion coefficient; BIO represents the biogeochemical dissolved oxygen source and sink terms. The source term in the source and sink terms can be a dissolved oxygen production term, and the sink term can be a dissolved oxygen consumption term.

[0142] Biogeochemical dissolved oxygen source sink equation:

[0143]

[0144] in, This represents the total oxygen production from phytoplankton photosynthesis. Indicates phytoplankton ( ) Utilizing ammonia nitrogen ( ,Right now The oxygen production rate of photosynthesis at the available concentration. This represents the maximum growth rate of phytoplankton. Let be the illumination constraint function. The stoichiometric coefficient for photosynthetic oxygen production using ammonia nitrogen; Basic ammonia-related oxygen consumption, i.e., oxygen consumption related to ammonia ( The related basal oxygen consumption processes, Related to the concentration or conversion rate of ammonia. For zooplankton oxygen consumption, For the oxygen consumption of zooplankton respiration, This refers to the oxygen consumption during the metabolic or excretory processes of zooplankton after feeding. For the oxygen consumption term of detrital mineralization, Oxygen is consumed for the mineralization of small debris. It is the first preset coefficient. Oxygen is consumed by the mineralization of large debris. It is the second preset coefficient; This indicates air-sea exchange, which is an "extra source" of dissolved oxygen.

[0145] In practice, a dissolved oxygen calculation model can be used to couple a three-dimensional ocean physical field, a biogeochemical state field, and a model-driven input field. Based on the control equation of dissolved oxygen concentration, the dynamic interaction mechanism of dissolved oxygen change can be solved through a time integration algorithm, and a three-dimensional dissolved oxygen concentration state field with high spatiotemporal resolution can be output for the first preset time period in the future.

[0146] In one embodiment, B3 described above can be implemented according to the following steps:

[0147] B3-1: Using a dissolved oxygen calculation model, based on the three-dimensional velocity field in the three-dimensional ocean physical field and the control parameters of dissolved oxygen concentration, the advection transport results of dissolved oxygen are determined.

[0148] Here, the advection transport result is the advection transport term. In practice, a dissolved oxygen calculation model can be used to combine the three-dimensional velocity field in the three-dimensional ocean physical field with the advection term control parameters in the dissolved oxygen governing equation. The dissolved oxygen calculation model calculates the spatial gradient of the current three-dimensional dissolved oxygen concentration distribution field at each grid cell and at each time step, and then performs numerical calculations on this gradient and the velocity component at the corresponding location, thereby accurately quantifying the dissolved oxygen flux carried by the large-scale directional movement of seawater, and obtaining the advection transport result of dissolved oxygen concentration in three-dimensional space due to advection.

[0149] B3-2: Using a dissolved oxygen calculation model, the turbulent diffusion result of dissolved oxygen is determined based on the turbulent diffusion coefficient in the three-dimensional ocean physical field and the control parameters of dissolved oxygen concentration.

[0150] Here, the turbulent diffusion result is the turbulent diffusion term, which includes at least the vertical and horizontal diffusion of dissolved oxygen due to turbulent mixing. In practice, a dissolved oxygen calculation model can be used to couple the turbulent diffusion coefficient field in the three-dimensional ocean physical field with the control parameters of the diffusion term in the dissolved oxygen governing equation. Then, the dissolved oxygen calculation model can calculate its spatial second derivative (i.e., concentration gradient divergence) based on the three-dimensional dissolved oxygen concentration field at the current moment, and perform calculations with the diffusion coefficient at the corresponding location to quantify the dissolved oxygen flux generated by turbulent mixing, thus obtaining the turbulent diffusion result of dissolved oxygen due to turbulent diffusion.

[0151] B3-3: Using a dissolved oxygen calculation model, based on the biogeochemical state field, the temperature field in the three-dimensional ocean physical field, and the control parameters of dissolved oxygen concentration, the production and consumption results of dissolved oxygen are determined.

[0152] Here, production and consumption results can quantify the impact of biogeochemical processes within the marine ecosystem on dissolved oxygen (DO) levels. Specifically, these results represent the biochemical source and sink terms of DO. As a decisive factor in predicting the formation and evolution of hypoxic zones, these terms integrate outputs from multiple key processes within the biogeochemical module. The main sources of DO include: phytoplankton photosynthetic oxygen production: directly calculated from the total photosynthetic rate of phytoplankton at each spatiotemporal grid point by the biogeochemical module, converted using the stoichiometry of photosynthesis; and oxygen exchange at the air-sea interface: the oxygen flux between air and seawater at the surface grid of the pre-defined three-dimensional grid. The sink terms of DO rely on the real-time biogeochemical status provided by the biogeochemical module, including oxygen consumption from phytoplankton respiration, organic matter mineralization, nitrification, and sediment.

[0153] In practical implementation, a dissolved oxygen calculation model can be used, invoking parameterized schemes for processes such as oxygen production through photosynthesis, oxygen consumption during organic matter degradation, and oxygen consumption during nitrification from the dissolved oxygen control parameters. According to this scheme, the photosynthetic oxygen production rate is calculated based on the biogeochemical state field, the temperature field in the three-dimensional ocean physical field, phytoplankton biomass, and light conditions. The oxygen consumption rate of microbial respiration and degradation processes is calculated using organic matter concentration and temperature-dependent rate functions. Numerical integration is then used to calculate dissolved oxygen production and consumption within each water layer and grid cell, yielding the results of dissolved oxygen production and consumption generated by ecosystem metabolism.

[0154] B3-4: Based on the results of advection transport, turbulent diffusion, and production and consumption, determine the three-dimensional dissolved oxygen concentration state field within the first preset time period in the future.

[0155] In practice, a dissolved oxygen calculation model can be used to determine the dissolved oxygen sources and sinks in the target sea area based on the production and consumption results. Then, based on the advection transport results and turbulent diffusion results, the three-dimensional dissolved oxygen concentration state field of the target sea area in the first preset time period in the future can be obtained.

[0156] In one embodiment, for B3-3 above, the following steps can also be followed:

[0157] B3-3-1: Using a dissolved oxygen calculation model, the amount of dissolved oxygen generated is determined based on the total photosynthetic rate and the control parameters of dissolved oxygen concentration in the biogeochemical state field.

[0158] In practice, a dissolved oxygen calculation module can be used to read the total photosynthetic rate data for each grid cell and each water layer from the biogeochemical state field. Then, the control parameters of dissolved oxygen concentration are coupled with the photosynthetic rate and converted into oxygen production rate through stoichiometry. Finally, the model numerically integrates this oxygen production rate at each time step to calculate the amount of dissolved oxygen generated by photosynthesis.

[0159] B3-3-2: Using a dissolved oxygen calculation model, determine the amount of oxygen exchange at the air-sea interface based on the temperature field, the dissolved oxygen concentration in the target area of ​​the target sea area, and the control parameters of the dissolved oxygen concentration.

[0160] Here, the dissolved oxygen concentration within the target area is the dissolved oxygen concentration corresponding to the uppermost M layer in the preset grid of the target sea area. M is a positive integer, and its value depends on the definition of the sea surface layer; this application does not impose specific limitations. The target dissolved oxygen concentration state field is the dissolved oxygen concentration state field of the surface region.

[0161] In practice, a dissolved oxygen calculation model can be used to determine the oxygen exchange rate based on the target dissolved oxygen concentration state field of the top N layers of the preset three-dimensional grid, as well as the temperature and salinity field data of the target area, combined with the parameterization scheme of the gas exchange rate in the dissolved oxygen control parameters and the saturation concentration of oxygen in seawater determined by temperature and salinity.

[0162] For example, the oxygen exchange rate at the air-sea interface can be determined using the following formulas 6-1 to 6-3:

[0163] (6-1)

[0164] (6-2)

[0165] (6-3)

[0166] Among them, U 10Indicates (wind speed in 10 meters), T represents temperature. 2, This refers to the oxygen exchange capacity. DO K represents the dissolved oxygen concentration in surface water. r The exchange rate; the saturation concentration of dissolved oxygen in seawater is affected by temperature and salinity; K r Controlled by wind speed and water temperature. The above formula is usually used to calculate the amount of oxygen exchanged per unit time for each surface grid cell; a positive value indicates that oxygen dissolves from the atmosphere into the ocean, and a negative value indicates that oxygen escapes from the ocean into the atmosphere; finally, the net oxygen exchange of the entire air-sea interface can be determined based on the amount of oxygen exchanged in each grid cell.

[0167] B3-3-3: Using a dissolved oxygen calculation model, determine the amount of dissolved oxygen consumed based on oxygen consumption parameters in the biogeochemical state field and control parameters of dissolved oxygen concentration; oxygen consumption parameters include at least the oxygen consumption parameters of planktonic respiration, organic matter mineralization, nitrification, and sediment.

[0168] In practice, a dissolved oxygen calculation model can be used to calculate the following four major oxygen-consuming processes based on key variables related to oxygen consumption processes obtained from the biogeochemical state field, as well as the corresponding oxygen consumption parameterization schemes and rate coefficients in the dissolved oxygen control parameters: 1. Calculating metabolic oxygen consumption based on zooplankton biomass and respiratory oxygen consumption parameters; 2. Calculating microbial degradation oxygen consumption based on organic detritus concentration, temperature, and organic matter mineralization oxygen consumption parameters; 3. Calculating the oxygen consumption of nitrifying bacteria in oxidizing ammonium salts to nitrates based on ammonium salt concentration and nitrification oxygen consumption parameters; 4. Estimating the oxygen flux at the sediment-water interface using sediment oxygen consumption parameters. Finally, based on the calculation results of these four major oxygen-consuming processes, the dissolved oxygen consumption in the target sea area can be determined.

[0169] B3-3-4: Determine the production and consumption results of dissolved oxygen based on the amount generated, the amount of oxygen exchanged, and the amount consumed.

[0170] In practice, a dissolved oxygen calculation model can be used to determine the final dissolved oxygen production result based on the amount generated, and to determine the final dissolved oxygen consumption result based on the amount of oxygen exchanged and consumed.

[0171] In one embodiment, this application also provides a scheduling and control module for automating the prediction process of marine hypoxia. This module can be built on a Linux platform with a centralized scheduling core to achieve automatic triggering of forecast tasks and parallel computation acceleration of three coupled models. It can also integrate breakpoint continuation and anomaly alarm mechanisms. Specifically, in response to the scheduling and control module periodically triggering the forecast task, it can return to the step of obtaining the pattern-driven input field of the target sea area at different times.

[0172] Here, the scheduling and control module is used to centrally manage and globally start, stop, and control the operation sequence covering the entire forecasting process, while continuously monitoring the system's operating status. The forecasting task is the forecasting process described in S101~S104 above.

[0173] The scheduling and control module can periodically trigger the prediction process according to a preset forecast period. Each triggering of the prediction process means re-executing S101~S104 above. The forecast period can be a periodic forecast at time X every day, where X can be any time from 0:00 to 24:00.

[0174] In practical implementation, the scheduling and control module first automatically schedules various basic data sources, such as open boundary data, runoff data, historical observation data, and model parameter libraries, according to operational needs and S101 above. It simultaneously completes data preprocessing and time-series matching to ensure the input data meets model calculation requirements. Upon triggering the arrival period, a prediction task is initiated. The collected model input data is converted into a model-driven input field and input into the model. The module automatically calls the baseline parameters from the parameter library and the parameters to be optimized after optimization by the target genetic algorithm, and configures them for the prediction model. Then, using the prediction model, based on the model-driven input field and the configured parameters, it predicts the three-dimensional ocean physical state field, biogeochemical state field, and three-dimensional dissolved oxygen concentration state field of the target sea area within the first preset time period, following the steps in S102. Then, by executing S103 and S104, various forecast information is obtained and predictions are made. Simultaneously, the scheduling and control module can monitor the model's operating status in real time. If data is missing or calculation deviations occur, an emergency adjustment mechanism is triggered. After the model prediction is completed, the scheduling and control module can also perform standardized post-processing of the prediction results and complete result storage and visualization according to operational application needs.

[0175] In one embodiment, after the prediction task is triggered, the scheduling control module can also perform real-time status detection and alarm during the prediction process. Specifically, the scheduling control module can be used to detect the execution status and system status information of each prediction task step in the prediction task in real time.

[0176] Here, each prediction task step may include steps S101 to S104 and each sub-step summarized from these steps. System state information is a representation of the hardware resources and software environment on which the prediction task depends. Execution state may encompass the state of the software environment supporting the prediction task's execution.

[0177] In practice, when the preset ocean hypoxia forecasting model executes its prediction task, the scheduling and control module first breaks down the task according to the forecasting process and tracks the execution status of each step in real time. Simultaneously, the scheduling and control module monitors system status information. At the hardware level, it collects real-time data on CPU utilization, memory usage, and storage I / O read / write rates; at the software level, it monitors data transmission links, the running status of the model calculation process, and abnormal records in the operating system logs.

[0178] Furthermore, in response to an execution status indication of an execution anomaly in any task step, a first anomaly alarm message can be generated, and an anomaly handling mechanism matching the cause of the execution anomaly can be used for anomaly handling; and / or, in response to a system status information indicating the existence of a status anomaly, a second anomaly alarm message can be generated.

[0179] Here, the exception handling mechanism can be used to handle detected anomalies in real time. Different exception handling mechanisms can be pre-set for execution anomalies caused by different reasons.

[0180] In practice, the scheduling control module continuously monitors the execution status and system status information of each task step in real time. When the execution status of any task step indicates an anomaly, the scheduling control module will immediately generate a first anomaly alarm message, which includes the identifier of the task step where the anomaly occurred, the anomaly type, and a preliminary determination of the cause. Simultaneously, it will automatically match and activate the corresponding anomaly handling mechanism based on the cause. Furthermore, if the system status information indicates an anomaly, the scheduling control module will generate a second anomaly alarm message, which covers the abnormal system module, key indicator values, and preset thresholds. This is used to prompt maintenance personnel to intervene promptly for resource allocation or environmental remediation, ensuring that the forecasting task can be completed continuously, stably, and efficiently even in complex operating environments.

[0181] like Figure 2As shown in the illustration, this disclosure also provides a schematic diagram of the overall framework of a marine hypoxia prediction method, which may include a model input data collection and preprocessing module, a model-driven input field creation module, a model operation calculation module, a hypoxia early warning product creation module, and a scheduling and control module. The model-driven input field creation module is used to construct model-driven input fields at different times. The model operation calculation module is used for parameter optimization and input module for selecting and configuring target model parameters, and for predicting the three-dimensional ocean physical state field, biogeochemical state field, and three-dimensional dissolved oxygen concentration state field using the hydrodynamic calculation model, biogeochemical model, and dissolved oxygen calculation model in the forecast model, respectively. The hypoxia early warning product creation module is used to standardize the target dissolved oxygen concentration state field of the bottom region and generate hypoxia forecast products. The scheduling and control module framework can be used to centrally manage and globally start, stop, and control the operation sequence covering the entire forecast process, while continuously monitoring the system's operating status. For the specific implementation process of each of the above modules / models, please refer to the descriptions of the above embodiments; they will not be repeated here.

[0182] In one embodiment, S104 described above can be implemented according to the following steps:

[0183] S104-1: Based on the relationship between the target dissolved oxygen concentration state field and different hypoxia thresholds, determine the hypoxia area and hypoxia influence range of the preset region under different hypoxia thresholds.

[0184] Here, hypoxia indication information can specifically include hypoxia area, hypoxia impact range, average dissolved oxygen concentration, and minimum dissolved oxygen concentration. Hypoxia indication information provides a comprehensive and quantitative characterization of the spatial distribution and intensity levels of hypoxia events, offering a multi-dimensional scientific basis for accurately assessing ecological risks and formulating differentiated management strategies.

[0185] In practical implementation, the target dissolved oxygen concentration state field of a predetermined area in the target sea area can be preprocessed. This involves denoising, spatial interpolation to fill in blank grid cells, and matching the predetermined area boundaries to achieve data standardization. For example, it can be converted to a standardized NetCDF format with 6-hour intervals. Then, using the grid cells of the predetermined area as a benchmark, the dissolved oxygen concentration of the target dissolved oxygen concentration state field in each grid cell is compared with different thresholds to identify and mark anoxic grids under different anoxic thresholds. Furthermore, the area of ​​a single grid cell is calculated according to its resolution, and the anoxic area under different thresholds is statistically calculated based on the total number of anoxic grids. Spatial clustering analysis aggregates the anoxic grids into at least one continuous anoxic region. Based on the spatial location of each continuous anoxic region, the anoxic influence range of each continuous anoxic region is obtained. Finally, the anoxic area of ​​each continuous anoxic region is determined based on the grid area corresponding to that region.

[0186] S104-2: Determine the average and minimum dissolved oxygen concentrations of the preset area based on the target dissolved oxygen concentration state field.

[0187] In practice, the minimum dissolved oxygen concentration can be selected from the target dissolved oxygen concentration state field on the corresponding bottom grid cells of each bottom region. Simultaneously, the average dissolved oxygen concentration of the bottom region can be calculated based on the target dissolved oxygen concentration and the total number of bottom grid cells. Optionally, when selecting the minimum dissolved oxygen concentration, the corresponding grid spatial coordinates can be recorded to clarify its specific location. Based on the average and minimum dissolved oxygen concentrations, the overall level and extreme low values ​​of dissolved oxygen in the bottom of the target sea area can be intuitively reflected. Optionally, when determining the scope of hypoxia, if multiple consecutive hypoxic regions are aggregated, the hypoxia area and hypoxia impact range of each consecutive hypoxic region can be output separately. Simultaneously, the average and minimum dissolved oxygen concentrations within each consecutive hypoxic region can be calculated.

[0188] S104-3: Use the hypoxic area and hypoxic influence range, average dissolved oxygen concentration and minimum dissolved oxygen concentration of the preset area under different hypoxic thresholds as hypoxic indication information.

[0189] For example, the hypoxic area and hypoxic influence range, average dissolved oxygen concentration, and minimum dissolved oxygen concentration of a preset area under different hypoxic thresholds can be used as hypoxic indication information under different hypoxic thresholds. Furthermore, after obtaining the hypoxic indication information, it can be visualized. Specifically, basic geographic data such as coastlines and marine functional zones can be overlaid to represent and output the hypoxic indication information under different hypoxic thresholds using visual charts and / or text descriptions. The hypoxic indication information under different hypoxic thresholds can be represented using visual information of different colors. For example, the hypoxic influence range at different hypoxic thresholds can be represented by colors such as red, orange, and yellow. Optionally, when generating visualized hypoxic forecast information, the forecast start time and forecast lead time can also be displayed.

[0190] like Figure 3 The diagram shown is a schematic representation of a visualized hypoxia prediction information provided in an embodiment of this application. Figure 3 Information may include the reporting time, forecast lead time, and the hypoxia area (i.e., the red hypoxia area) corresponding to the minimum hypoxia threshold (e.g., 1 mg / L).

[0191] Those skilled in the art will understand that, in the above-described method of the specific implementation, the order in which each step is written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined by its function and possible internal logic.

[0192] Based on the same inventive concept, this disclosure also provides a marine hypoxia prediction device corresponding to the marine hypoxia prediction method. Since the principle of the device in this disclosure is similar to the marine hypoxia prediction method described above, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0193] like Figure 4 The diagram shown is a schematic representation of a marine hypoxia prediction device provided in an embodiment of this disclosure, comprising:

[0194] The acquisition module 401 is used to acquire the model-driven input field of the target sea area at different times. The model-driven input field at different times includes at least the atmospheric forcing field corresponding to the meteorological forecast data within a future first preset time period, the historical model initial field, the boundary field at a historical second preset time period, and the runoff file. The model initial field is used to characterize the initial state of the marine physical and ecological environmental variables of the target sea area. The boundary field is used to characterize the average scale boundary of the marine physical and ecological environmental variables. The runoff file is used to indicate the runoff attribute information of the target sea area.

[0195] Prediction module 402 is used to identify and predict the mode-driven input field at different times using multiple coupled models in the preset ocean hypoxia prediction model, so as to obtain the three-dimensional ocean physical state field, biogeochemical state field and three-dimensional dissolved oxygen concentration state field of the target sea area in the first preset time period in the future.

[0196] The first forecast module 403 is used to generate marine environmental forecast information for the target sea area in the future within the first preset time period according to the target time interval, based on the three-dimensional ocean physical state field, biogeochemical state field and three-dimensional dissolved oxygen concentration state field, and to broadcast it on a rolling basis.

[0197] The second forecast module 404 is used to extract the target dissolved oxygen concentration state field that matches the preset area of ​​the target sea area from the three-dimensional dissolved oxygen concentration state field, and determine the hypoxia indication information under the preset area according to the target dissolved oxygen concentration state field and different hypoxia thresholds, and generate visualized hypoxia forecast information according to the target time interval and the hypoxia indication information.

[0198] The processing flow of each module in the device and the interaction flow between each module can be referred to the relevant descriptions in the above method embodiments, and will not be detailed here.

[0199] Based on the same technical concept, embodiments of this application also provide a computer device. (Refer to...) Figure 5The diagram shown is a structural schematic of a computer device provided in an embodiment of this application, comprising:

[0200] The system comprises a processor 501, a memory 502, and a bus 503. The memory 502 stores machine-readable instructions executable by the processor 501. The processor 501 executes these machine-readable instructions, and when executed, it performs steps S101-S104 as described above. The memory 502 includes a main memory 5021 and an external memory 5022. The main memory 5021, also called internal memory, is used to temporarily store computational data in the processor 501 and data exchanged with external storage devices such as hard disks 5022. The processor 501 exchanges data with the external storage devices 5022 through the main memory 5021. When the computer is running, the processor 501 communicates with the memory 502 via the bus 503, enabling the processor 501 to execute the instructions mentioned in the above method embodiments.

[0201] This disclosure also provides a computer-readable storage medium storing a computer program that, when executed by a processor, performs the steps of the ocean hypoxia prediction method described in the above-described method embodiments. The storage medium may be a volatile or non-volatile computer-readable storage medium.

[0202] This disclosure also provides a computer program product carrying program code. The program code includes instructions that can be used to execute the steps of the ocean hypoxia prediction method described in the above method embodiments. For details, please refer to the above method embodiments, which will not be repeated here.

[0203] The computer program product can be implemented specifically through hardware, software, or a combination thereof. In one alternative embodiment, the computer program product is specifically embodied in a computer storage medium; in another alternative embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0204] Finally, it should be noted that the above-described embodiments are merely specific implementations of this disclosure, used to illustrate the technical solutions of this disclosure, and not to limit it. The protection scope of this disclosure is not limited thereto. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this disclosure; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure, and should all be covered within the protection scope of this disclosure. Therefore, the protection scope of this disclosure should be determined by the protection scope of the claims.

Claims

1. A method for predicting marine hypoxia, characterized in that, include: The model-driven input fields for the target sea area at different times are obtained. These input fields include at least the atmospheric forcing field corresponding to meteorological forecast data within a first preset time period, historical model initial fields, historical boundary fields at a second preset time period, and runoff files. The model initial field characterizes the initial state of the marine physical and ecological environmental variables of the target sea area. The boundary field characterizes the average scale boundary of the marine physical and ecological environmental variables. The runoff files indicate the runoff attribute information of the target sea area. Using multiple coupled models in the preset ocean hypoxia prediction model, the mode-driven input fields at different times are identified and predicted to obtain the three-dimensional ocean physical state field, biogeochemical state field and three-dimensional dissolved oxygen concentration state field of the target sea area in the first preset time period in the future. According to the target time interval, based on the three-dimensional ocean physical state field, biogeochemical state field and three-dimensional dissolved oxygen concentration state field, marine environmental forecast information for the target sea area in the future within the first preset time period is generated and broadcast on a rolling basis. Extract the target dissolved oxygen concentration state field from the three-dimensional dissolved oxygen concentration state field that matches the preset area of ​​the target sea area, and determine the hypoxia indication information under the preset area according to the target dissolved oxygen concentration state field and different hypoxia thresholds, and generate visualized hypoxia forecast information according to the target time interval and the hypoxia indication information.

2. The prediction method according to claim 1, characterized in that, The acquisition of the pattern-driven input field of the target sea area at different times includes: Determine whether the preset ocean hypoxia prediction model has an output initial field; If not, then obtain the model input data for the target sea area; the model input data includes meteorological forecast data for the first preset time period in the future, historical satellite remote sensing data and observation data, and historical open boundary data and runoff data for the second preset time period in the past; The meteorological forecast data is interpolated into a preset three-dimensional grid using an interpolation method to obtain the atmospheric forcing field for the first preset time period in the future; the preset three-dimensional grid is a simulation model grid constructed for the target sea area; Using an interpolation method, the satellite remote sensing data and observation data are synchronously interpolated to the preset three-dimensional grid to obtain the initial field of the model; From the open boundary data, target boundary data matching the boundary of the target sea area is extracted, and the target boundary data is used to construct the boundary field at the second preset historical time. The runoff data is assigned to the preset three-dimensional grid runoff source region location to obtain the runoff file for the second preset historical time.

3. The prediction method according to claim 1, characterized in that, Before using multiple coupled models in a preset ocean hypoxia prediction model to identify and predict the mode-driven input fields at different times to obtain the three-dimensional ocean physical state field, biogeochemical state field, and three-dimensional dissolved oxygen concentration state field of the target sea area in the future first preset time period, the method further includes: Obtain a pre-built model parameter library; the model parameter library includes different model parameters and operating rules corresponding to different sea areas; Obtain target model parameters that match the target sea area from the model parameter library, and configure the target model parameters into the preset marine hypoxia forecast model.

4. The prediction method according to claim 1, characterized in that, The preset marine hypoxia prediction model includes at least a hydrodynamic calculation model, a biogeochemical model, and a dissolved oxygen calculation model, comprising: The hydrodynamic calculation model is used to predict the three-dimensional ocean physical state field within the first preset time period in the future; the biogeochemical model is used to predict the biogeochemical state field within the first preset time period in the future; and the dissolved oxygen calculation model is used to predict the three-dimensional dissolved oxygen concentration state field within the first preset time period in the future.

5. The prediction method according to claim 4, characterized in that, Using the aforementioned hydrodynamic calculation model, the prediction of the three-dimensional ocean physical state field includes: Using a hydrodynamic calculation model, based on the mode-driven input field and three-dimensional ocean fluid dynamics control parameters, the physical operation process in the marine ecosystem is simulated to obtain the three-dimensional ocean physical state field of the target sea area in the first preset time period in the future. The physical operation process includes at least the tidal operation process, the three-dimensional temperature change process, the salinity change process, the flow field change process, and the sea surface height change process; the three-dimensional ocean physical state field includes at least the three-dimensional velocity field, the sea surface height field, the temperature field, the salinity field, and the turbulent diffusion coefficient.

6. The prediction method according to claim 4, characterized in that, Using the biogeochemical model, predicting the biogeochemical state field includes: Using a biogeochemical model, based on the three-dimensional ocean physical state field, the model-driven input field, and biogeochemical control parameters, various biogeochemical cycles and species dynamic changes in the marine ecosystem are numerically simulated to obtain the biogeochemical state field in the first preset time period in the future.

7. The prediction method according to claim 4, characterized in that, Using the dissolved oxygen calculation model, the prediction of the three-dimensional dissolved oxygen concentration state field includes: Using a dissolved oxygen calculation model, based on the three-dimensional ocean physical field, the biogeochemical state field, the mode-driven input field, and the control parameters of dissolved oxygen concentration, the dynamic interaction mechanism of dissolved oxygen changes in the marine ecosystem is numerically simulated to obtain the three-dimensional dissolved oxygen concentration state field in the first preset time period in the future.

8. A device for predicting marine hypoxia, characterized in that, include: The acquisition module is used to acquire the model-driven input fields of the target sea area at different times. The model-driven input fields at different times include at least the atmospheric forcing field corresponding to the meteorological forecast data within a future first preset time period, the historical model initial field, the boundary field at a historical second preset time period, and the runoff file. The model initial field is used to characterize the initial state of the marine physical and ecological environmental variables of the target sea area. The boundary field is used to characterize the average scale boundary of the marine physical and ecological environmental variables. The runoff file is used to indicate the runoff attribute information of the target sea area. The prediction module is used to identify and predict the mode-driven input field at different times using multiple coupled models in the preset ocean hypoxia prediction model, so as to obtain the three-dimensional ocean physical state field, biogeochemical state field and three-dimensional dissolved oxygen concentration state field of the target sea area in the first preset time period in the future. The first forecast module is used to generate marine environmental forecast information for the target sea area in the future within the first preset time period according to the target time interval, based on the three-dimensional ocean physical state field, biogeochemical state field and three-dimensional dissolved oxygen concentration state field, and to broadcast it on a rolling basis. The second forecast module is used to extract the target dissolved oxygen concentration state field that matches the preset area of ​​the target sea area from the three-dimensional dissolved oxygen concentration state field, and determine the hypoxia indication information under the preset area according to the target dissolved oxygen concentration state field and different hypoxia thresholds, and generate visualized hypoxia forecast information according to the target time interval and the hypoxia indication information.

9. A computer device, characterized in that, include: The processor and the memory, wherein the memory stores machine-readable instructions executable by the processor, the processor is configured to execute the machine-readable instructions stored in the memory, and when the machine-readable instructions are executed by the processor, the processor performs the steps of the ocean hypoxia prediction method as described in any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is run by a computer device, the computer device performs the steps of the ocean hypoxia prediction method as described in any one of claims 1 to 7.

Images