A method and system for quick prediction response of offshore oil spill based on nearshore sea area

Abstract

The present application relates to the technical field of marine environmental disaster prediction and emergency response, and particularly relates to a method and system for rapid prediction and response of offshore oil spill based on nearshore sea area, which comprises S1, obtaining meteorological driving data; S2, obtaining marine flow field data; S3, configuring a unified timing automation scheduling workflow; S4, determining the initial position of oil spill based on oil spill event information, and forming the background field data required for oil spill prediction; S5, calling a predefined oil spill prediction model, and outputting the oil spill diffusion trajectory information of multiple time scales and the affected sensitive target area in the next prediction period based on the background field data through the oil spill prediction model. The present application realizes the cooperative linkage and business operation of the meteorological-oceanic-oil spill model, improves the timeliness and accuracy of oil spill prediction, solves the problems of response lag and poor adaptability of traditional methods, and can quickly support the emergency disposal decision of oil spill accidents in the national nearshore sea area.

Description

Technical Field

[0001] This invention relates to the field of marine environmental disaster prediction and emergency response technology, and in particular to a rapid prediction and response method and system for oil spills in nearshore waters. Background Technology

[0002] With the continuous growth of economic activities such as shipping, offshore oil and gas development, and aquaculture in my country's nearshore waters, oil spill accidents have become increasingly frequent, causing numerous adverse effects on the stability of marine ecosystems, coastal industrial development, and the safety of the living environment. Emergency response to oil spills typically relies on rapid prediction of the oil slick's spread trajectory and impact range to support decision-making regarding containment and interception, cleanup and recovery, and protection of sensitive areas. For example, predictive response plans for nearshore oil spills primarily depend on numerical simulation technology, using meteorological models, marine hydrodynamic models, and oil spill spread models to extrapolate the oil slick's drift path and impact range. However, in practical emergency applications, this type of nearshore oil spill predictive response technology still faces several bottlenecks: First, the poor coordination between the various models results in low data flow efficiency. For example, in traditional solutions, meteorological models (such as WRF), ocean models, and oil spill models are often built and run independently, lacking a collaborative mechanism. This means that data transfer between models (such as wind field driving the ocean model, and wind and current fields jointly driving the oil spill model) often requires a lot of manual intervention. This not only makes the entire prediction process time-consuming and difficult to meet the "golden" timeliness requirements of oil spill emergency response, but also easily introduces errors due to human error or inconsistent spatiotemporal matching of data, reducing the reliability of the prediction base data. Secondly, existing model configuration strategies are relatively simplistic and cannot adapt to the complex hydrodynamic environment of nearshore waters. Specifically, my country's nearshore waters include not only open shelf seas but also numerous semi-enclosed bays and narrow estuaries. Therefore, it can be said that the dominant hydrodynamic factors differ greatly across different scales of the sea area and cannot be simply lumped together with a fixed landform. For example, large bays are affected by tides, shelf circulation, and thermo-salinity-density currents; while in smaller bays or estuaries, where the water depth is shallower, tidal currents and wind stress often play a dominant role, and the thermo-salinity stratification effect is not obvious. If a "one-size-fits-all" model configuration strategy, commonly used in existing technologies, is adopted, that is, using ocean models of the same complexity and the same driving factors for all sea areas, this configuration strategy not only leads to a waste of computational resources and slow response speed in small-scale sea areas, but also results in insufficient prediction accuracy due to the neglect of key driving factors (such as thermo-salinity), failing to meet the dual requirements of prediction efficiency and accuracy.

[0003] However, oil spill emergency response is a time-sensitive process, requiring predictive systems to start up quickly, operate stably, and continuously output results. Relying solely on existing event-driven technology frameworks that remain at the application level means that emergency responders will still need to spend a significant amount of valuable time debugging models and preparing data when faced with sudden oil spills, severely limiting the speed and efficiency of the emergency response. Summary of the Invention

[0004] Based on this, in order to address the shortcomings of existing technologies, a rapid prediction and response method for oil spills in nearshore waters is proposed.

[0005] To achieve the above design objectives, the technical solution of the present invention is as follows: A rapid prediction and response method for oil spills in nearshore waters includes: S1. Construct a meteorological field model, and conduct numerical forecasting of nearshore waters based on the meteorological field model to obtain meteorological driving data for oil spill diffusion calculation; S2. Construct a hierarchical marine hydrodynamic model system. Combine the meteorological driving data and the spatial scale characteristics of the sea area where the oil spill event is located, match and run the corresponding level of the bay model in the marine hydrodynamic model system to obtain marine flow field data for oil spill diffusion calculation. The marine hydrodynamic model system includes at least a large bay partition model, a medium bay partition model or a small bay partition model. S3. Introduce a business process management tool, and configure a unified timed automated scheduling workflow for the meteorological field model and the ocean hydrodynamic model system based on the business process management tool; S4. Based on the oil spill event information of the current prediction period, determine the initial location of the oil spill, i.e. the spatial location of the sea area where the oil spill is located. Based on the initial location of the oil spill, dynamically match the corresponding meteorological driving data and ocean current field data through the timed automated scheduling workflow to form the background field data required for oil spill prediction. S5. Call the predefined oil spill prediction model. Based on the background field data, output the oil spill diffusion trajectory information at multiple time scales in the next prediction period, as well as the affected sensitive target areas, through the oil spill prediction model.

[0006] Based on the same inventive essence, this application also proposes a system based on the aforementioned rapid prediction and response method for oil spills in nearshore waters, characterized in that it includes: The meteorological field model construction unit is used to perform numerical forecasting of nearshore waters based on the meteorological field model and to obtain meteorological driving data for oil spill diffusion calculation. The hierarchical hydrodynamic coupling unit is used to construct a hierarchical marine hydrodynamic model system. It combines the meteorological driving data and the spatial scale characteristics of the sea area where the oil spill event is located, matches and runs the corresponding level of the bay model in the marine hydrodynamic model system to obtain marine flow field data for oil spill diffusion calculation. The business process management unit is used to introduce business process management tools and configure a unified timed automated scheduling workflow for the meteorological field model and the ocean hydrodynamic model system based on the business process management tools. The first processing unit is used to determine the initial location of the oil spill, i.e. the spatial location of the sea area where the oil spill is located, based on the oil spill event information of the current prediction period. Based on the initial location of the oil spill, the corresponding meteorological driving data and ocean current field data are dynamically matched through the timed automated scheduling workflow to form the background field data required for oil spill prediction. The second processing unit is used to call a predefined oil spill prediction model, and based on the background field data, output the oil spill diffusion trajectory information at multiple time scales in the next prediction period, as well as the affected sensitive target areas.

[0007] Implementing the embodiments of the present invention will have the following beneficial effects: This invention addresses the problems of poor coordination between meteorological and oceanographic models, low operational efficiency, insufficient regional adaptability, limited initial oil slick acquisition, and untimely dynamic correction in existing nearshore oil spill prediction and response technologies. It provides a rapid prediction and response method for nearshore oil spills by constructing an operational multi-model collaborative system, a hierarchical and regional oceanographic model adaptation mechanism, and differentiated initial oil slick acquisition methods. This enables rapid and accurate prediction of oil spill diffusion in nearshore waters nationwide, providing efficient and reliable technical support for emergency response decision-making. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] in: Figure 1 This is a flowchart of the basic steps corresponding to the solution described in this invention; Figure 2 This is a schematic diagram of the ECFLOW-driven model business operation process corresponding to the implementation case of this invention. Detailed Implementation

[0010] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0011] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. It is understood that the terms “first,” “second,” etc., as used herein may be used to describe various elements, but these elements are not limited by these terms. These terms are used only to distinguish one element from another. For example, a first element may be referred to as a second element without departing from the scope of this application, and similarly, a second element may be referred to as a first element. Both the first element and the second element are elements, but they are not the same element.

[0012] To address the urgent need to effectively improve my country's emergency response capabilities for oil spill disasters in its nearshore waters, this invention essentially designs a multi-model collaborative technical solution for oil spill prediction and emergency response in nearshore waters nationwide, integrating meteorological, oceanographic, and oil spill technologies. This solution aims to create a rapid oil spill prediction and response mechanism that incorporates various strategies, including integrated operational procedures, tiered adaptation, differentiated initial information acquisition, and dynamic correction.

[0013] Based on the aforementioned design requirements, the overall architecture of this application is established through the following core steps, including: First, for the specific scenario of "timeliness of oil spill emergency response", the WRF meteorological model, the hierarchically constructed marine hydrodynamic model and the oil spill prediction model are integrated and automatically scheduled to form a multi-model operational collaborative operation mechanism. Through this mechanism, a fully automated, timed workflow from data acquisition to result archiving is completed. Secondly, considering the complex geography and hydrodynamics of nearshore waters, this application integrates and adapts a zoning method of "three-level zoning of large, medium, and small bays, overlapping configuration schemes, and bay coastline alignment" with a differentiated driving mode of "dual-driven tidal current + wind for small bays, and multi-driven tidal current + wind + temperature + salinity for large / medium bays" to form a hierarchically constructed marine hydrodynamic model system. This system, through a site-specific configuration strategy, significantly improves computational speed while maintaining the accuracy of oil spill prediction in small bays, achieving a balance between computational efficiency and simulation accuracy. It resolves the contradiction between the large-scale coverage and localized high accuracy / efficiency resulting from the "one-size-fits-all" configuration of existing technologies, producing a synergistic effect of "high speed, sufficient accuracy, and no blind spots." Finally, for the real-world scenario of "potentially missing oil spill source information," the two initialization modes of "determined source strength" and "uncertain source strength" are unified in the same strategy. For uncertain scenarios, a technical path of "SAR and optical radar collaborative observation + deep learning inversion" is designed to invert information such as the distribution and thickness of the initial oil film, thereby driving model initialization. Through this mechanism, scenarios that were traditionally unpredictable or had extremely low prediction accuracy are transformed into predictable scenarios, greatly enhancing the applicability of the method in actual oil spill accidents.

[0014] Based on the above design framework, this embodiment proposes a rapid prediction and response method for oil spills in nearshore waters, such as... Figure 1 As shown, the method includes the following steps: S1. Construct a meteorological field model, and conduct numerical forecasting of nearshore waters based on the meteorological field model to obtain meteorological driving data for oil spill diffusion calculation; S2. Construct a hierarchical marine hydrodynamic model system. Combine the meteorological driving data and the spatial scale characteristics of the sea area where the oil spill event is located, match and run the corresponding level of the bay model in the marine hydrodynamic model system to obtain marine flow field data for oil spill diffusion calculation. The marine hydrodynamic model system includes at least a large bay partition model, a medium bay partition model or a small bay partition model. S3. Introduce a business process management tool, and configure a unified timed automated scheduling workflow for the meteorological field model and the ocean hydrodynamic model system based on the business process management tool; S4. Based on the oil spill event information of the current prediction period, determine the initial location of the oil spill, i.e. the spatial location of the sea area where the oil spill is located. Based on the initial location of the oil spill, dynamically match the corresponding meteorological driving data and ocean current field data through the timed automated scheduling workflow to form the background field data required for oil spill prediction. S5. Call the predefined oil spill prediction model. Based on the background field data, output the oil spill diffusion trajectory information at multiple time scales in the next prediction period, as well as the affected sensitive target areas, through the oil spill prediction model.

[0015] In some specific embodiments, S1 employs a general large-scale meteorological field model, such as a global or regional numerical weather prediction model, to forecast nearshore sea areas and extract the required meteorological elements, specifically including the following steps: In step S1, the meteorological field model adopts the WRF (Weather Research and Forecasting) model. Numerical forecasting of nearshore waters is performed using the WRF model to obtain meteorological driving data for oil spill dispersion calculations within the current forecast period, thereby generating a high-resolution rasterized wind field dataset covering the entire nearshore waters of China. This provides accurate meteorological driving data for subsequent oil spill predictions. Specifically, this includes the following steps: S11. Before or at the start of the current cycle, configure the model domain: configure the corresponding simulation region parameters based on the sea surface wind field characteristics of the target sea area near the coast; the preferred configuration is to set the latitude and longitude of the simulation region center, such as (115°E, 21°N), the horizontal resolution of the model domain is no greater than 0.125°×0.125°, the number of grids is no less than 400×400, and at least 30 vertical layers are set in the vertical direction to refine the wind field structure near the sea surface; wherein, setting the horizontal resolution to no greater than 0.125°×0.125° and the number of grids to no less than 400×400 is to capture the small and medium scale caused by the complex topography of the nearshore area and the thermal difference between land and sea. Meteorological phenomena, such as sea and land breezes and local circulation, have a significant impact on nearshore wind fields and sea surface heat flux. Vertical stratification with no fewer than 30 layers, especially refining the wind field structure near the sea surface, is crucial for accurately simulating momentum, heat, and water vapor exchange at the air-sea interface. This is because oil spill diffusion is driven by sea surface wind stress, which is closely related to the near-surface wind speed profile. This allows for the precise capture of subtle changes in the nearshore wind field, improving the adaptability of wind field data to the driving force of nearshore oil spill diffusion.

[0016] S12. Selecting Initial and Boundary Field Data for the Model: Based on the model domain, global weather forecast data is selected as the initial and boundary field data for the WRF model to provide large-scale background information for the regional high-resolution meteorological model. The global weather forecast data includes at least core meteorological elements such as temperature, pressure, humidity, wind speed, and wind direction. The initial field defines the three-dimensional atmospheric state at the start of the model's operation, while the boundary field provides external atmospheric conditions at the lateral boundaries of the model domain, ensuring that the regional model remains consistent with the global-scale weather system. This data is typically provided by global numerical weather prediction centers (such as ECMWF, NCEP, etc.) and includes basic meteorological elements such as temperature, pressure, humidity, wind speed, and wind direction. It also includes elements used to drive the ocean hydrodynamic model, such as heat flux, temperature, pressure, evaporation, and precipitation. These elements collectively determine the temperature, salinity, density structure, and ocean current conditions of the surface water, indirectly affecting the physicochemical processes and diffusion of oil spills. This setting ensures that the model's initial state is consistent with global atmospheric circulation characteristics, laying the foundation for accurate wind field forecasts.

[0017] S13. Data Standardization Processing: Based on the actual needs of oil spill prediction, the time step and forecast duration of the WRF model are set, and rasterized meteorological driving data is output. The forecast duration matches the oil spill prediction cycle. Preferably, the model forecast duration is set to 72 hours (i.e., the next 3 days) to accurately match the subsequent oil spill prediction simulation cycle; the time step is set to 60 seconds to balance computational accuracy and operational efficiency. The meteorological field model is run and hourly meteorological driving data is output, such as hourly sea surface 10m wind speed, wind direction, temperature, and air pressure data for nearshore waters across the country. Running the meteorological field model and outputting hourly meteorological driving data provides a high temporal resolution external forcing field. Hourly data can capture the diurnal variation and short-term fluctuations of meteorological elements, which is crucial for accurately simulating wind-induced drift and air-sea interface exchange during oil spill diffusion. Matching the forecast duration of the meteorological field model with the oil spill prediction cycle ensures that continuous and effective meteorological driving data is available throughout the entire oil spill prediction cycle. For example, if the oil spill prediction period is 72 hours, then the meteorological model should also forecast for at least 72 hours. Setting the time step is a crucial parameter in numerical simulation, requiring a balance between computational accuracy and operational efficiency. A smaller time step can improve simulation stability and accuracy but significantly increase computational load; a larger time step may lead to numerical instability or decreased accuracy. Therefore, users can optimize the selection based on model characteristics, computational resources, and required accuracy. The formation of a standardized rasterized wind field dataset refers to performing format conversion, spatial interpolation, and timestamp normalization on the meteorological driving data to create a standardized dataset for use in the marine hydrodynamic model system and the oil spill prediction model.

[0018] In some specific embodiments, S2 constructs a hierarchical and zonal multi-level marine hydrodynamic model, i.e., a marine hydrodynamic model system, to generate accurate data on ocean current velocity and direction for different bay zones, based on the differences in hydrodynamic characteristics of bays of different sizes in the nearshore waters of the country. This specifically includes the following steps: S21. Based on the hydrodynamic characteristics of bays of different scales in nearshore waters, a hierarchical marine model configuration strategy is customized to form a hierarchically constructed marine hydrodynamic model system. The bay hydrodynamic characteristics include at least the openness, enclosure, and depth of the water area. This step involves the hierarchical zoning construction process of nearshore waters, which uses a pre-set marine model configuration strategy to hierarchically divide the national nearshore waters. Preferably, the marine model configuration strategy adopts a three-level zoning strategy, that is, based on the openness, enclosure, and depth characteristics of the nearshore waters, the national nearshore waters are divided into three levels: The Great Bay Zone: This refers to areas with open waters and significant influence from the open sea (such as the Northwest Taiping Bay model).

[0019] Zhonghai Bay Zone: This refers to semi-enclosed sea areas with moderate depths (such as the Bohai Sea, Yellow Sea, East China Sea, and South China Sea, which are medium-sized or semi-enclosed sea areas).

[0020] Small bay zoning: for narrow, shallow ports, estuaries and key coastal sections (such as specific bays covering an area of ​​more than 10 km²).

[0021] In this embodiment, the aforementioned zoning strategy aims to configure the most suitable model based on the geographical features and hydrodynamic characteristics of different sea areas. The marine hydrodynamic model system includes at least a large bay zoning model, a medium bay zoning model, or a small bay zoning model. Here, the marine hydrodynamic model refers to a numerical computation system (such as FVCOM, ADCIRC, ROMS, MIKE, etc.) that uses the Navier-Stokes equations for discrete solution based on the finite difference method or the finite element method. This system can reconstruct the flow field, water level, and temperature-salinity structure in three-dimensional space based on the input data; existing mainstream marine models can be used.

[0022] Preferably, step S21 further includes: optimizing the boundary regions between the various partition models within the marine hydrodynamic model system, i.e., based on the principle of cross-overlap: setting cross-overlap regions between adjacent partitions (large-medium, medium-small) between various partition models, and the width of the overlap region is not less than a preset distance (e.g., ≥5km), to ensure that in the oil spill diffusion calculation, oil particles are always within the effective calculation range of at least one bay partition model when crossing the boundary within the specified simulation time, avoiding particle loss or calculation interruption due to boundary discontinuity. The preset distance can be set based on the maximum possible displacement of oil particles within a single time step, or determined based on the model mesh resolution and hydrodynamic characteristics combined with experience, to ensure the smoothness of data transmission and model connection.

[0023] Preferably, step S21 further includes: performing shoreline fitting constraint processing on the small bay partition model within the marine hydrodynamic model system. That is, when constructing the computational grid of the small bay partition model, ensure that the grid boundary can closely follow the actual geographical shoreline contour to avoid water leakage or inaccurate flow field simulation caused by coarse or mismatched grids. This can be achieved by using unstructured grids or locally refined grid technology to strictly ensure that its shoreline basically fits the actual shoreline, guarantee the requirements for fine nearshore simulation, improve the prediction accuracy of small-scale oil spill accidents, and achieve more accurate diffusion processes within harbors and wharves.

[0024] S22. Call the predefined global ocean model to configure the initial state, lateral boundary conditions, and surface boundary conditions for the ocean hydrodynamic model system, so as to obtain the basic data related to the initial field and driving field of each regional model. The lateral boundary conditions refer to the corresponding lateral open boundary constraint values ​​obtained based on the global ocean model data, so as to simulate the propagation and input of large-scale ocean circulation and tidal waves into the computational domain. The surface boundary conditions refer to the corresponding sea surface meteorological field constraint values ​​obtained based on the meteorological driving data. This step mainly completes the configuration of initial state data and boundary conditions for each regional model based on global ocean model data and meteorological driving data. Preferably, the global ocean model data adopts at least one of CMEMS and HYCOM. For example, global ocean model CMEMS data is introduced to extract ocean currents (simulating the influence of large-scale ocean currents such as the Kuroshio Current and coastal currents on bays), circulation influencing factors (such as temperature and salinity), and sea level information (driving tidal waves into the subsequent computational domain) as the lateral boundary conditions and initial state of the regional ocean hydrodynamic model. Simultaneously, meteorological field data, including sea surface wind field, air pressure, temperature, humidity, heat flux, and precipitation, are acquired as key meteorological input data driving ocean hydrodynamic processes and applied as boundary conditions on the sea surface, specifically at the uppermost grid of the regional ocean hydrodynamic model (i.e., the air-sea interface). Wind stress, heat flux, air pressure, and precipitation data output by the meteorological field model are applied. Specifically, wind field data is converted into tangential forces acting on the sea surface using wind stress formulas, driving surface seawater flow. Heat flux and precipitation data control the vertical flux of sea surface temperature and salinity through thermodynamic exchange equations, simulating the energy and material exchange between the ocean and atmosphere, thus achieving the synergistic fusion of meteorological and oceanographic data.

[0025] S23. Hydrodynamic Simulation and Flow Field Generation: Based on the aforementioned basic data, and combined with predefined differentiated initial field and driving field configuration strategies, corresponding initial fields and driving fields are configured for each level of the regional model within the marine hydrodynamic model system; the differentiated initial field and driving field configuration strategies include: A dual-drive model of tidal currents and wind (a simplified simulation process, using only tidal boundary conditions and sea surface wind stress as external driving forces to calculate the velocity and direction of seawater flow in the area) was used to run the corresponding bay zoning model for hydrodynamic simulation calculations. The output of the ocean current velocity and direction data for the bay zoning was used as the core input data for subsequent oil spill diffusion prediction. The tidal current driving force was achieved by applying tidal boundary conditions at the open boundary of the zoning model, while the wind driving force was achieved by inputting wind field data from the meteorological driving data as the sea surface stress boundary condition. Since bays are typically shallow and the water is well-mixed... Temperature and salinity stratification are not obvious, and tides and winds are the main driving forces of its hydrodynamic processes. Therefore, this model does not require additional input of temperature and salinity parameters, and does not consider the influence of thermo-salinity-density current on ocean currents. This simplifies the model calculation process while ensuring the basic requirements of prediction accuracy (for example, it can be a two-dimensional or quasi-three-dimensional model that does not consider or simplifies the consideration of thermo-salinity effects). This ensures that the amount of calculation is significantly reduced in a short time while still effectively capturing the main water flow characteristics in small bays and generating output results, meeting the needs of rapid response to small bay oil spill accidents, and thus improving rapid response capabilities. Furthermore, the tidal driving force in this example, namely the open boundary tidal forcing force, can be understood as the process of sea surface fluctuation caused by the gravitational influence between various celestial bodies. In the driving analysis process, from a global perspective, if simulating a small bay, it is necessary to incorporate the fluctuations or flow of the open sea. This is generally achieved by providing a time-varying water level sequence (tidal water level boundary) at the open boundary of the zoning model to simulate the propagation and deformation of tidal waves from the open sea within the bay, thereby generating reciprocating currents or rotating currents. Essentially, this tidal driving force relies on the open boundary water level forcing of the open sea to introduce tidal waves, generating bay currents under topographic constraints, which is the tidal driving force of the entire bay circulation.

[0026] Taking into full account the influence of temperature and salinity fields on the complex hydrodynamic processes of large and medium-sized bays, the corresponding large bay zoning model or medium bay zoning model is run using multiple driving modes of tidal currents, wind, temperature and salinity to simulate hydrodynamics and output the corresponding ocean current velocity and direction data as the core input data for subsequent oil spill diffusion prediction. Among them, the tidal current driving force is achieved by setting at least eight major tidal constituents (the amplitude and phase of tidal constituents such as M2, S2, N2, K1, O1, P1, and Q1) at the open boundary of the large bay and medium bay models to fully consider the driving effect of major tides on ocean currents and thus accurately reconstruct the actual tidal propagation process. The wind driving force is achieved by inputting the wind field data from the meteorological driving data. The temperature and salinity driving forces are achieved by extracting temperature field and salinity field data from the global ocean model. The addition of temperature and salinity driving forces is to account for the impact of uneven temperature and salinity distribution when solving the Navier-Stokes equations for the large and medium bays (these bays typically have greater water depths, and may exhibit significant thermohaline stratification; density-driven currents and topographic effects significantly influence hydrodynamic processes). Therefore, based on the aforementioned multi-driving model of tidal currents, wind, temperature, and salinity, the ocean hydrodynamic zoning model can simultaneously use tidal boundary conditions, sea surface wind stress, temperature field, and salinity field as external driving forces to calculate the flow velocity and direction of seawater in the area. This allows for a more accurate simulation of the complex three-dimensional ocean current structure of tidal currents, wind-driven currents, and thermohaline circulation and their interactions, providing high-precision ocean current field data for oil spill dispersion prediction.

[0027] Therefore, this step is one of the key steps in this scheme. It addresses the complex geographical environment of "significant differences in hydrodynamic characteristics of bays of different sizes in the nearshore waters of the country" by forming a hierarchical marine hydrodynamic model system. Based on this, it implements differentiated driving modes according to the bay level, forming a synergistic "hierarchical and zoning + differentiated driving" mechanism. This solves the contradiction between "large-scale coverage" and "local high precision / high efficiency" in the model configuration used by traditional technical solutions (which can be understood as a one-size-fits-all solution), thus producing a synergistic technical effect of "fast calculation speed, high prediction accuracy, and no blind spots in the whole area".

[0028] In some specific embodiments, S3 specifically designs a workflow architecture for oil spill emergency scenarios, defining business process management tools such as the professional numerical weather prediction business process management tool ECFLOW software. Based on this business process management tool, a unified timed automated scheduling workflow is configured for the meteorological field model and ocean hydrodynamic model system. The automated scheduling process includes: periodically downloading basic data such as global meteorological forecast data and global ocean model data; automatically generating initial fields based on the downloaded data; automatically starting model forecasts; post-processing the model output results (including data format conversion and accuracy verification); archiving and uploading qualified result data to a designated database; and cleaning up temporary data generated during operation to release storage resources. This achieves fully automated, unattended operation from data acquisition to result archiving. Through the integrated management of ECFLOW, compared with existing technologies that rely on manual intervention and fragmented data flow, this invention shortens model preparation time from hours to minutes, ensuring that the latest meteorological and ocean background field data can be immediately accessed when an oil spill occurs, thus truly meeting the requirements of oil spill emergency response. Simultaneously, automated operation eliminates human error, improving data consistency and the reliability of prediction results.

[0029] In some specific embodiments, under the aforementioned automated scheduling framework of the multi-model collaborative operation mechanism, the design purpose of S4 is to require it to determine the target bay zone corresponding to the sea area where the oil spill event is located based on the oil spill event information, and to select the corresponding level of marine hydrodynamic model for configuration and operation according to the bay type of the target bay zone. In this way, it can automatically and intelligently identify the bay type (large bay, medium bay or small bay) to which the oil spill event belongs based on the actual location of the oil spill event, and automatically call and configure the marine hydrodynamic model of the corresponding complexity to avoid using the most complex model for all areas, thereby affecting the computational efficiency.

[0030] To achieve the above objectives, step S4 specifically includes the following steps: S41. Obtain oil spill event information for the current prediction period, extract initial morphological data of the oil spill source location from the oil spill event information, and verify the initial morphological data of the oil spill source location. If the verification rules are met, proceed to step S42; otherwise, proceed to step S43. The verification rules are used to perform a difference operation on the certainty of the oil spill source location after the oil slick is discovered. Specifically, the initial morphological data of the oil spill source location contains latitude and longitude coordinates of the oil spill source, or contains regional description information that can calculate the geometric center coordinates of the oil spill source, i.e., contains data sufficient to directly determine the initial location coordinates of the oil spill. This data can be a single latitude and longitude coordinate point, or a clear regional description (such as a polygonal or circular region) that can calculate the geometric center coordinates. For example, when the initial morphological data of the oil spill source location contains clear latitude and longitude coordinates, the system can directly extract and use it as the oil spill location and directly initialize the oil spill prediction model based on this information. In addition, the reason for setting up the difference operation is that when obtaining oil spill event information, it is possible to use manual input, with operators manually inputting various information of the event based on reports or on-site investigation results, which will cause inaccuracies in the information description.

[0031] S42. Based on the verified initial morphological data of the oil spill source location, identify the target bay partition type corresponding to its sea area, and dynamically match and run the meteorological driving data and ocean current field data of the corresponding level through the timed automated scheduling workflow to form the background field data required for oil spill prediction. S43. Based on the failed initial morphological data of the oil spill source location, acquire satellite imagery data of the sea area where the oil spill event is located, and based on the initial morphological correction strategy for the oil spill source location, identify the target bay partition type corresponding to the sea area where it is located. Dynamically match and run the corresponding level of meteorological driving data and ocean current field data through the timed automated scheduling workflow to form the background field data required for oil spill prediction. The initial morphological correction strategy for the oil spill source location includes: inputting the satellite imagery data into a pre-trained deep learning inversion model, outputting initial oil film distribution data, which at least includes the oil film coverage area and oil film boundary coordinates (oil film coverage area, i.e., the actual area occupied by the oil film on the sea surface; oil film boundary coordinates, precisely defining the outer contour of the oil film), and determining the corresponding latitude and longitude coordinates of the oil spill source based on the initial oil film distribution data, or... The system provides regional description information sufficient to calculate the geometric center coordinates of the oil spill source. Preferably, the satellite imagery data (which can penetrate clouds, is not limited by day or night, and directly detects the existence and distribution of oil slicks on the sea surface, providing an objective and real-time basis for determining the location of the oil spill) is obtained through a combined observation method of synthetic aperture radar (SAR) and optical radar. Furthermore, the satellite imagery data undergoes time matching, geometric correction, and registration with the same coordinate system to obtain standardized satellite imagery input data. The deep learning inversion model employs a convolutional neural network inversion model, which is established by learning the complex mapping relationship between a large number of historical satellite imagery images and the actual oil slick distribution. This allows it to invert the physical properties of the oil slick from the characteristics of the satellite imagery signal (such as backscatter intensity and texture), ensuring that the inversion error is controlled within 10%, and providing accurate initial boundary conditions for model initialization. This ensures that when the oil spill event information only contains a vague location description (such as "near a certain sea area"), or the uncertainty radius of the provided location data exceeds a preset threshold (which can be set according to the grid resolution of the bay model), it is determined that this step needs to be initiated to perform the initialization process of satellite image observation and deep learning inversion. That is, based on the initial oil film distribution data, the oil spill location is determined. For example, by using a simple threshold segmentation or boundary detection algorithm, the geometric center or maximum range of the oil film is extracted from the oil film distribution data output by the model as the oil spill location. Preferably, in step S43, determining the latitude and longitude coordinates of the corresponding oil spill source based on the initial oil film distribution data, or determining the regional description information that can calculate the geometric center coordinates of the oil spill source, includes the following steps: S431. Perform connected component extraction on the oil film coverage area to obtain at least one candidate oil film region; S432. Calculate the area of ​​each candidate oil film region, and remove candidate oil film regions that do not meet the requirements based on the preset area threshold and shape constraints. S433. The candidate oil film region with the highest conformity to the preset criteria among the remaining candidate oil film regions is determined as the target oil film region; the geometric center coordinates or the boundary polygon of the target oil film region are used as the latitude and longitude coordinates of the oil spill source at the oil spill location, or the region description information that can calculate the geometric center coordinates of the oil spill source.

[0032] Connected component extraction is an image processing technique aimed at identifying interconnected regions with identical pixel values ​​in an image. Here, this operation identifies all interconnected oil film pixels within the initial oil film coverage area as independent "candidate oil film regions." This helps to initially segment potentially scattered or discontinuous oil film information, providing a foundation for subsequent screening and localization. Subsequently, the area of ​​each candidate oil film region is calculated, and regions that do not meet the requirements are eliminated based on preset area thresholds and shape constraints. The initial oil film distribution data inverted by the deep learning model may contain false oil film regions caused by sensor noise, environmental interference, or model errors. By calculating the area of ​​each candidate oil film region and comparing it with a preset area threshold, regions that are too small or too large and do not conform to the actual characteristics of oil spills can be effectively eliminated. For example, a minimum area threshold can be preset; regions smaller than this threshold are considered noise and eliminated. Simultaneously, combining shape constraints—for example, oil films typically exhibit irregular but relatively continuous shapes rather than elongated or overly fragmented shapes—can further eliminate noise regions with abnormal shapes, improving the accuracy of identification.

[0033] Secondly, after screening by area and shape, several candidate oil slick areas may still remain that meet the basic requirements. To accurately determine the oil spill location, the candidate oil slick area that best matches the preset criteria from the remaining candidate oil slick areas needs to be identified as the target oil slick area. Preset criteria can be based on prior knowledge of the oil spill event, historical data, or expert experience; for example, selecting the area with the largest area, the center point closest to the approximate location of the reported oil spill, or the most regular shape. For instance, the largest area among the remaining candidate oil slick areas can be simply selected as the target oil slick area; alternatively, if there is a preliminary reported oil spill location, the candidate oil slick area whose center point is closest to that location can be selected; or, weights can be assigned to multiple factors such as area, shape, and distance from the reported location, a comprehensive score can be calculated, and the area with the highest score can be selected.

[0034] Finally, the geometric center coordinates or the bounding polygon of the target oil film region are used as the oil spill location. Once the target oil film region is determined, it needs to be transformed into a specific oil spill location representation that can be used for subsequent prediction models. The geometric center coordinates provide a single, representative point location, suitable for cases where the oil spill is considered a point source. For example, the average latitude and longitude coordinates of all pixels within the target oil film region can be calculated as its geometric center. The bounding polygon provides more detailed regional information, suitable for cases where the oil spill is considered a surface source or where its initial diffusion range needs to be considered. For example, the boundary pixels of the target oil film region can be extracted, and a minimum bounding rectangle, convex hull, or more complex geometric polygon can be constructed to represent the extent of the region.

[0035] In some specific embodiments, the purpose of S5 is to construct and run an oil spill prediction model: that is, after obtaining initial oil slick distribution data based on the certainty of the oil spill source location, an oil spill prediction model is constructed based on the Lagrange particle tracking method, integrating meteorological and marine data to simulate the oil spill diffusion trajectory information at multiple time scales over a certain period of time, such as 3 days, and simultaneously identifying and outputting potentially affected sensitive areas (such as aquaculture areas, wetlands, nature reserves, etc.) to provide a basis for delineating the scope of emergency response; the reason why the oil spill prediction simulation model uses the Lagrange particle tracking method to characterize the migration and diffusion process of the oil slick is that this method can accurately track the movement trajectory of individual oil particles, thereby reconstructing the entire diffusion evolution process of the oil slick. Its core equation is based on the motion law of oil particles under the combined action of water flow and wind, which can fully reflect the driving effect of tidal currents and wind fields on oil slick diffusion.

[0036] Preferably, in step S5, the oil spill prediction model employs the Lagrange particle tracking method, specifically including: Assumptions: The oil spill is discretized using the Lagrange particle tracing method, representing it as multiple oil particles, each carrying location and quantity attributes. This method treats the oil spill as a large number of independent oil particles, each representing a certain volume or mass of oil, effectively capturing the dispersion, drift, and diffusion behavior of oil in complex flow fields. Furthermore, during model initialization, a large number of oil particles are generated within the spill area based on the spill location and initial leakage amount, and assigned initial attributes such as location, mass, or oil type.

[0037] Simulation calculation: Within each prediction step, the horizontal displacement of oil particles is calculated based on the ocean current field data, and the wind-induced drift displacement is calculated based on the meteorological driving data. A random walk update is then performed on the oil particles using the diffusion coefficient to obtain the temporal position set of the oil particles in the next period. Specifically, for each oil particle, based on its current position, the seawater velocity vector at that position is interpolated from the ocean current field data (e.g., ocean current velocity field and ocean current direction field), and multiplied by the time step to update the oil particle's position. Simultaneously, the wind-induced drift displacement is calculated based on the meteorological driving data. This is typically achieved by obtaining the wind speed vector from the meteorological driving data (e.g., wind speed and direction), calculating it according to a certain wind-induced drift coefficient (e.g., 3% to 4% of the sea surface wind speed) and deflection angle, and then superimposing it into the oil particle's position update. Furthermore, a random walk update is performed on the oil particle diffusion process using the diffusion coefficient. The random walk component is sampled from a probability distribution with zero mean and variance correlated with the diffusion coefficient and time step to simulate random motion caused by turbulence and molecular diffusion. Through the above calculations, the temporal position set of oil particles in the next period can be obtained, recording the position information of all oil particles at different predicted time points.

[0038] Output Results: Based on the time-series location set, the spatiotemporal distribution of the oil film is reconstructed, outputting the oil spill diffusion trajectory, oil film coverage area, and oil film thickness distribution, i.e., multi-timescale oil spill diffusion information for the next cycle. The reconstructed spatiotemporal distribution of the oil film can be achieved by dividing the prediction area into grids and statistically analyzing the number of oil particles or the amount of oil in each grid, or by using methods such as kernel density estimation to smoothly convert discrete particles into a continuous oil film density field. The oil spill diffusion information includes at least the oil spill diffusion trajectory, oil film coverage area, and oil film thickness distribution. The oil spill diffusion trajectory can be represented by connecting the center positions of oil particles at different time points or the geometric center of the oil film; the oil film coverage area can be represented by the coordinate set of the oil film boundary or a rasterized coverage area; and the oil film thickness distribution can reflect the oil pollution concentration in different areas through rasterized oil film thickness values. The multiple timescales include at least 1 hour, 6 hours, 12 hours, 24 hours, and 72 hours to meet the timeliness requirements of prediction at different emergency response stages.

[0039] Preferably, step S5 further includes: overlaying the oil film coverage area with preset target species area spatial data to identify and output the affected sensitive areas; the target species area spatial data is pre-stored in a geographic information system (GIS), including at least one of aquaculture areas, wetlands, and nature reserves. Through spatial overlay operation, the affected target species areas can be identified and output.

[0040] Preferably, for each affected sensitive area, a corresponding impact time window and impact degree index are also output; the impact time window includes the time when the oil film enters the area, the time when it leaves the area, or the duration of its existence; the impact degree index includes the proportion of oil film coverage area, average thickness, or residence time, or the risk level based on a comprehensive assessment of oil type and regional sensitivity, etc.

[0041] In some specific embodiments, to address the technical deficiency of "accumulated errors in long-term oil spill prediction," the method establishes a dynamic correction mechanism and generates standardized prediction reports to support emergency decision-making in order to ensure that the prediction results continuously match the actual oil film diffusion state and improve the stability of prediction accuracy. This forms a closed-loop feedback mechanism of "prediction-observation-correction." This mechanism effectively suppresses error drift in long-term predictions and ensures the continuous accuracy of prediction results. Specifically, it also includes: S6, a step of dynamically correcting the parameters of the oil spill prediction model by setting a correction period and using a sliding window correction strategy, specifically including: Within each correction cycle, acquire the actual distribution data of the oil film observed by synthetic aperture radar or optical radar; The actual oil slick distribution data is compared with the simulation results of the oil spill prediction model during the same period. Based on the deviation analysis results, the model parameters of the oil spill prediction model are dynamically adjusted, and the oil spill diffusion information for the remaining period from the current moment to the end of the next prediction cycle is recalculated using the adjusted model parameters, so as to achieve rolling updates of subsequent prediction results. Obtaining oil slick observation data within a preset correction cycle means setting multiple shorter time intervals as correction cycles throughout the entire oil spill prediction cycle, and at the end of each correction cycle, obtaining the current oil slick distribution information in the sea area through actual observation methods. This observation data can come from various sources, such as using satellite remote sensing technology (e.g., optical satellites, synthetic aperture satellites, SAR satellites) to obtain large-scale oil slick coverage images and thickness information; using UAVs equipped with visible light, infrared, or multispectral sensors for low-altitude aerial photography to obtain high-resolution local oil slick data; or using on-site survey vessels and personnel for on-site sampling and visual observation to record the actual location, shape, and thickness of the oil slick.

[0042] Preferably, the correction period is 12 hours.

[0043] Preferably, the deviation analysis includes: calculating the error value between the oil film observation data and the simulation results, wherein the error value includes at least one of the following: oil film coverage error, oil film thickness error, or oil film boundary position error.

[0044] Preferably, step S6, which dynamically adjusts the model parameters based on the error analysis results to update the oil spill diffusion information at multiple time scales in the next cycle, is the core step in achieving adaptive model correction. Specifically, this includes: adjusting at least one of the diffusion coefficient, wind-induced drift coefficient, evaporation rate, or emulsification rate of the oil spill prediction model using data assimilation techniques or optimization algorithms. The data assimilation techniques include Kalman filtering or ensemble Kalman filtering. For example, based on Kalman filtering or ensemble Kalman filtering, real-time oil film observation data is incorporated into the model's state variables, thereby updating state information such as the position of oil particles and the thickness of the oil film. The optimization algorithms include Bayesian optimization, genetic algorithms, or particle swarm optimization. For example, based on Bayesian optimization, physical parameters in the model, such as the oil film diffusion coefficient, wind-induced drift coefficient, evaporation rate, and emulsification rate, are iteratively adjusted according to the principle of minimizing error, so that the simulation results are closer to the observed data. In some cases, the model parameters can also be manually fine-tuned based on the experience of domain experts. The adjusted model parameters will be used for calculations in subsequent prediction cycles, thereby generating more accurate oil spill diffusion information.

[0045] Preferably, the next prediction period is 72 hours, and the preset correction period is 6 hours or 12 hours, so that each prediction period includes multiple correction periods, clearly defining the iterative relationship between prediction and correction. For example, a complete oil spill prediction period may be set to 72 hours, while the correction period can be set to 6 hours or 12 hours. This means that within the 72-hour prediction period, every 6 or 12 hours, the system will perform oil film observation data acquisition, error analysis, and model parameter adjustment, and then, based on the corrected model state and parameters, re-predict the oil spill diffusion for a future period. This multi-correction-period setting allows the oil spill prediction model to respond promptly to changes in the actual environment, effectively avoiding error accumulation and ensuring the continuous accuracy of the prediction results.

[0046] Preferably, the method further includes step S7: generating a standardized oil spill prediction report based on the output of the oil spill prediction model; the oil spill prediction report includes at least one or more of the following information: a summary of the oil spill event information, a description of the oil spill location and its determination method, the time and spatial range of the meteorological driving data and ocean current field data used, the model configuration parameters and time step of the oil spill prediction model, oil spill diffusion trajectory maps at multiple time scales, oil film coverage maps and oil film thickness distribution maps (intuitively presenting the oil film evolution process), a list of affected target types and areas obtained by overlaying with sensitive areas, error analysis results (indicating prediction accuracy) and confidence level assessment (clarifying the reliability of the results), so as to intuitively present the dynamic evolution process of the oil spill and provide accurate and efficient technical support for emergency response decisions (such as the deployment of oil booms, oil spill recovery, and delineation of personnel evacuation areas). By generating a standardized oil spill prediction report, this application can systematically integrate and clearly present key information such as complex oil spill prediction results, model operation status, and uncertainty assessment. This significantly improves the transparency, readability, and operability of forecast information, enabling emergency responders and decision-makers to quickly and accurately understand the dynamic development of oil spill events, their potential impact range, and the reliability of forecast results. The report includes an oil spill event summary, methods for determining the spill location, details of meteorological and ocean current field data, model configuration parameters, multi-timescale diffusion maps, a list of affected target types and areas, and error analysis results, collectively forming a comprehensive and easily understandable information system. This not only effectively solves the problem of the difficulty in effectively transmitting and interpreting forecast information but also significantly improves the scientific rigor, timeliness, and accuracy of oil spill emergency response decisions, thereby minimizing the damage caused by oil spill events to the nearshore marine ecological environment and economic activities.

[0047] Meanwhile, the results of corresponding simulation experiments also demonstrate the correctness of this method; Example 1: Taking a ship oil spill accident near Qingdao as an example, the rapid prediction and response method for oil spills in coastal waters nationwide based on the present invention is applied. The specific implementation plan is divided into pre-accident preparation and accident impact, and the steps are as follows: Step 1: Construct a national marine meteorological field model: A wind field model covering the nearshore waters of the whole country is constructed using the WRF model. The model domain is set to the central latitude and longitude (115°E, 21°N), with a horizontal resolution of 0.125°×0.125°, a grid number of 400×400, and 30 layers in the vertical direction, with a focus on refining the sea surface to a height resolution of 500m. Global meteorological forecast data is used as the initial and boundary fields, and elements such as temperature, air pressure, humidity, wind speed, and wind direction are input. The forecast duration is set to 72h, the time step is 60s, and the model is run to output hourly wind field data at 10m above the sea surface for the nearshore waters of the whole country. The hourly data for the Pearl River Estuary region with a wind speed of 3.5m / s and a wind direction of southeast are output, forming a rasterized wind field dataset.

[0048] Step 2: Construct a national multi-level regional ocean model system: Zoning: The sea area is divided into a large model of the western Pacific Ocean, a medium model (Yellow and Bohai Sea model, South China Sea model, East China Sea model), and a small model (53 small bays and 10 estuaries across the country), and a refined grid is constructed.

[0049] Basic data acquisition: Extract water level and current velocity data of the boundary field of the Pearl River Estuary from the HYCOM global ocean model; obtain meteorological data such as wind speed, wind direction, and heat flux from the WRF model in step 1.

[0050] Hydrodynamic model configuration: A mainstream ocean model is adopted. Since it is a small bay, it is configured with a dual-drive mode of tidal current + wind, and there is no need to input temperature and salinity parameters.

[0051] Small-scale simulation: Based on tidal and wind-driven field data, the model was run to obtain ocean current data with an average current velocity of 0.8 m / s and a northeastward flow direction in the Pearl River Estuary.

[0052] Step 3: Model Operation The ECFLOW tool was configured to automatically run the WRF model and the Pearl River Estuary small bay ocean model at 00:00 and 12:00 daily. It automatically downloaded global weather forecast data and HYCOM ocean data, generated initial fields, completed forecast calculations, post-processed the output wind field and ocean current data, archived and uploaded them to the emergency response database, and cleaned up temporary data. In this oil spill incident, which occurred at 10:00 on the same day, the latest dataset generated by ECFLOW at 00:00 that day was directly used.

[0053] Step 4: Determining the location of the oil spill and constructing and running the oil spill prediction simulation model. Oil spill location identified: The location of the oil spill source in this accident is clear (121°05′E, 35°15′N), the leakage volume is 50t, and the water level information is known. The oil spill prediction model is directly initialized based on this information.

[0054] Basic model construction: The oil spill model is constructed using the Lagrange particle tracking method, and the core equations are set based on the motion law of oil particles under the combined action of tidal current and southeast wind.

[0055] Model Initialization and Execution: A background field initialization model was generated by matching the small bay zones of Jiaozhou Bay (using the ocean current dataset and WRF wind field dataset for these zones). The model was run to simulate the oil spill diffusion process over multiple timescales over the next 72 hours. The output results show that the oil slick generally moved in a northeasterly direction. At 8:00 AM on May 3rd, the oil slick had drifted to a position approximately 19.4 km northeast of its initial location; at 2:00 PM on May 3rd, it had drifted to a position approximately 22.1 km northeast of its initial location. The oil slick initially moved in a southwesterly direction and then northeasterly. At 8:00 AM on May 25th, the oil slick had drifted to a position approximately 5.8 km southwest of its initial location; at 2:00 PM on May 25th, it had drifted to a position approximately 4.0 km northeast of its initial location.

[0056] Step 5: Dynamic Correction and Forecast Report Output Dynamic correction: A 12-hour sliding window correction strategy was adopted. The actual distribution data of the oil spill in the Pearl River Estuary was obtained by SAR radar at 12h, 24h, 36h, 48h and 60h after the accident. The data were compared with the simulation results at the same time. The calculation error values ​​were 4%, 5%, 6%, 5% and 7% respectively, all less than 10%. The model diffusion coefficient was dynamically adjusted to correct the subsequent prediction results.

[0057] Report generation: Generates a standardized oil spill prediction simulation report, including model configuration parameters (current + wind dual drive, time step 60s, etc.), oil spill diffusion graphs at multiple time scales from 1h to 72h, error analysis and confidence assessment (confidence level 93%), clearly indicating that the oil spill will approach the tidal flat wetland after 72h, and recommends deploying an oil boom 5km northeast of the oil spill source. The report is promptly sent to the local emergency response department to provide technical support for oil spill recovery work.

[0058] Based on the same inventive concept, this invention also proposes a rapid prediction and response system for oil spills in nearshore waters, comprising: The meteorological field model construction unit is used to perform numerical forecasting of nearshore waters based on the meteorological field model and to obtain meteorological driving data for oil spill diffusion calculation. The hierarchical hydrodynamic coupling unit is used to construct a hierarchical marine hydrodynamic model system. It combines the meteorological driving data and the spatial scale characteristics of the sea area where the oil spill event is located, matches and runs the corresponding level of the bay model in the marine hydrodynamic model system to obtain marine flow field data for oil spill diffusion calculation. The business process management unit is used to introduce business process management tools and configure a unified timed automated scheduling workflow for the meteorological field model and the ocean hydrodynamic model system based on the business process management tools. The first processing unit is used to determine the initial location of the oil spill, i.e. the spatial location of the sea area where the oil spill is located, based on the oil spill event information of the current prediction period. Based on the initial location of the oil spill, the corresponding meteorological driving data and ocean current field data are dynamically matched through the timed automated scheduling workflow to form the background field data required for oil spill prediction. The second processing unit is used to call a predefined oil spill prediction model, and based on the background field data, output the oil spill diffusion trajectory information at multiple time scales in the next prediction period, as well as the affected sensitive target areas.

[0059] Based on the same inventive concept, the present invention also proposes a computer-readable storage medium including computer instructions that, when executed on a computer, cause the computer to perform the method described thereon.

[0060] Based on the same inventive concept, the present invention also proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the steps of the method when executing the program.

[0061] In summary, the present invention achieves the following technical effects, specifically including: (1) Achieve multi-model collaboration and operationalization to improve response efficiency: This invention integrates and manages the WRF meteorological model and multi-level ocean model through the ECFLOW tool, realizing the fully automated operation from data download to result archiving, which greatly reduces the time spent on manual intervention; at the same time, the data of the meteorological and ocean models are accurately matched in time and space, providing efficient and reliable basic data support for oil spill prediction, and solving the problem of delayed response of traditional methods.

[0062] (2) Strong regional adaptability and high prediction accuracy: The three-level zoning strategy of large, medium and small bays and differentiated model configuration is adopted. The driving mode of small bays is simplified to improve the response speed, and the temperature and salinity driving of large and medium bays is added to ensure the simulation accuracy. The cross-overlapping zoning design avoids prediction blind spots, and the fine-grained coastline of small bays improves the nearshore simulation accuracy, so as to achieve accurate adaptation of oil spill accidents of different scales in bays.

[0063] (3) Differentiated initial oil film acquisition to adapt to diverse scenarios: In response to the differences in the certainty of the location of the oil spill source, a differentiated acquisition method of "direct initialization-deep learning inversion" is adopted. When the location is clear, the process is simplified. When the location is unknown, the accuracy of the initial data is ensured through dual radar collaboration and deep learning. This broadens the applicable scenarios of the method and improves the prediction reliability in complex scenarios.

[0064] (4) Timely dynamic correction and excellent result stability: The 12-hour sliding window dynamic correction strategy is adopted to integrate radar observation data in real time to calibrate model parameters, effectively suppress the accumulation of prediction bias, and ensure the continuous accuracy of prediction results at multiple time scales in the next 3 days, providing a scientific basis for the dynamic adjustment of emergency response decisions.

[0065] For ease of understanding, the following explains some key terms in this embodiment: Oil spill incident information refers to various raw data related to oil spill accidents, which describes the background, preliminary characteristics, and potential impact range of the event. This information forms the basis for subsequent oil spill location determination and response prediction.

[0066] Oil spill location refers to the origin of the oil leak or the area where the initial oil film is distributed when an oil spill occurs. It can be point, line, or area geographic coordinate information and serves as the starting condition for oil spill diffusion simulation.

[0067] The deep learning inversion model is a computational model built on deep learning technology. By learning the mapping relationship between a large amount of satellite image observation data and the actual oil film distribution, it can infer information such as the coverage range and coverage edge of the initial oil film from the satellite image observation data.

[0068] A meteorological field model is a numerical model used to simulate and predict atmospheric motion and changes in meteorological elements. Through mathematical description and calculation of atmospheric physical processes, it provides meteorological driving data such as wind speed, wind direction, temperature, and air pressure, providing atmospheric environmental background for oil spill dispersion prediction.

[0069] A marine hydrodynamic model system refers to a set of numerical models used to simulate the hydrodynamic characteristics of sea areas at different scales. This system can select appropriate large bay models, medium bay models, or small bay models based on the geographical features and water depth conditions of the bay, in order to simulate ocean currents, tides, and other ocean flow field data in a refined manner.

[0070] Oil spill prediction models are numerical models specifically designed to simulate and predict the diffusion, drift, and weathering processes of oil spills in the marine environment. These models comprehensively consider various environmental factors, including meteorological and hydrodynamic factors, to output the future spatiotemporal distribution of the oil slick.

[0071] Oil spill diffusion information refers to detailed data on the future distribution of the oil slick output by oil spill prediction models. It typically includes the oil slick's diffusion trajectory, coverage area, and thickness distribution, and is used to assess the impact of the oil spill.

[0072] Target areas refer to sensitive areas in nearshore waters that have special ecological, economic, or social value. Examples include aquaculture areas, wetlands, and nature reserves. These areas require special attention and protection in the event of an oil spill.

[0073] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A rapid prediction and response method for oil spills in nearshore waters, characterized in that, include: S1. Construct a meteorological field model, and conduct numerical forecasting of nearshore waters based on the meteorological field model to obtain meteorological driving data for oil spill diffusion calculation; S2. Construct a hierarchical marine hydrodynamic model system. Combine the meteorological driving data and the spatial scale characteristics of the sea area where the oil spill event is located, match and run the corresponding level of the bay model in the marine hydrodynamic model system to obtain marine flow field data for oil spill diffusion calculation. The marine hydrodynamic model system includes at least a large bay partition model, a medium bay partition model or a small bay partition model. S3. Introduce a business process management tool, and configure a unified timed automated scheduling workflow for the meteorological field model and the ocean hydrodynamic model system based on the business process management tool; S4. Based on the oil spill event information of the current prediction period, determine the initial location of the oil spill, i.e. the spatial location of the sea area where the oil spill is located. Based on the initial location of the oil spill, dynamically match the corresponding meteorological driving data and ocean current field data through the timed automated scheduling workflow to form the background field data required for oil spill prediction. S5. Call the predefined oil spill prediction model. Based on the background field data, output the oil spill diffusion trajectory information at multiple time scales in the next prediction period, as well as the affected sensitive target areas, through the oil spill prediction model.

2. The rapid prediction and response method for oil spills in nearshore waters according to claim 1, characterized in that, In step S1, the meteorological field model adopts the WRF model. Numerical forecasting of nearshore waters is performed using the WRF model to obtain meteorological driving data for oil spill dispersion calculations within the current forecast period. This specifically includes the following steps: S11. Configure the model domain: Configure the corresponding simulation area parameters based on the sea surface wind field characteristics of the target sea area in the nearshore sea area. S12. Select the initial field and boundary field data of the model: Based on the model domain, select the initial field and boundary field data of the WRF model; S13. Data standardization processing: Set the time step and forecast duration of the WRF model, and output rasterized meteorological driving data. The forecast duration is matched with the oil spill prediction cycle.

3. The rapid prediction and response method for oil spills in nearshore waters according to claim 1, characterized in that, S2 specifically includes the following steps: S21. A graded and partitioned marine model configuration strategy is customized based on the hydrodynamic characteristics of bays of different sizes in nearshore waters to form a graded marine hydrodynamic model system, wherein the hydrodynamic characteristics of the bays include at least the openness, enclosure and depth of the water area. S22. Call the predefined global ocean model to configure basic data for the ocean hydrodynamic model system, namely, configure the initial state, lateral boundary conditions and surface boundary conditions. The lateral boundary conditions refer to the corresponding lateral open boundary constraint values ​​obtained based on the global ocean model data, and the surface boundary conditions refer to the corresponding sea surface meteorological field constraint values ​​obtained based on the meteorological driving data. S23. Based on the aforementioned basic data and combined with predefined differentiated initial field and driving field configuration strategies, configure corresponding initial fields and driving fields for the partition models at each level within the marine hydrodynamic model system.

4. The rapid prediction and response method for oil spills in nearshore waters according to claim 3, characterized in that, The differentiated initial field and driving field configuration strategies include: The corresponding small bay partition model is run in a dual-drive mode of tidal current and wind to perform hydrodynamic simulation calculations and output the ocean current velocity and direction data of the small bay partition. The tidal current driving force is achieved by applying tidal boundary conditions at the open boundary of the partition model, and the wind driving force is achieved by inputting the wind field data in the meteorological driving data as the sea surface stress boundary condition. The corresponding large bay zoning model or medium bay zoning model is run using multiple driving modes of tidal current, wind, temperature and salinity to perform hydrodynamic simulation and output the corresponding ocean current velocity and direction data. Among them, the tidal driving force is achieved by setting at least eight major tidal constituents at the open boundary, the wind driving force is achieved by inputting the wind field data in the meteorological driving data, and the temperature and salinity driving forces are achieved by extracting temperature field and salinity field data from the global ocean model.

5. The rapid prediction and response method for oil spills in nearshore waters according to claim 1, characterized in that, S4 specifically includes the following steps: S41. Obtain oil spill event information for the current prediction period, extract initial morphological data of the oil spill source location from the oil spill event information, and verify the initial morphological data of the oil spill source location. If the verification rules are met, proceed to step S42; otherwise, proceed to step S43. The verification rules refer to the fact that the initial morphological data of the oil spill source location contains the latitude and longitude coordinates of the oil spill source, or contains regional description information that can calculate the geometric center coordinates of the oil spill source. S42. Based on the verified initial morphological data of the oil spill source location, identify the target bay partition type corresponding to its sea area, and dynamically match and run the meteorological driving data and ocean current field data of the corresponding level through the timed automated scheduling workflow to form the background field data required for oil spill prediction. S43. Based on the initial morphological data of the oil spill source location, acquire satellite imagery observation data of the sea area where the oil spill event is located, and based on the initial morphological correction strategy for the oil spill source location, identify the target bay partition type corresponding to the sea area where it is located. Dynamically match and run the meteorological driving data and ocean current field data at the corresponding level through the timed automated scheduling workflow to form the background field data required for oil spill prediction. The initial morphological correction strategy for the oil spill source location includes: inputting the satellite imagery observation data into a pre-trained deep learning inversion model, outputting initial oil film distribution data, which includes at least the oil film coverage area, oil film thickness, and distribution coordinates, and determining the latitude and longitude coordinates of the corresponding oil spill source or the regional description information that can calculate the geometric center coordinates of the oil spill source based on the initial oil film distribution data.

6. The rapid prediction and response method for oil spills in nearshore waters according to claim 5, characterized in that, In step S43, determining the latitude and longitude coordinates of the corresponding oil spill source based on the initial oil film distribution data, or determining the regional description information that allows for the calculation of the geometric center coordinates of the oil spill source, includes the following steps: S431. Perform connected component extraction on the oil film coverage area to obtain at least one candidate oil film region; S432. Calculate the area of ​​each candidate oil film region, and remove candidate oil film regions that do not meet the requirements based on the preset area threshold and shape constraints. S433. The candidate oil film region with the highest degree of conformity to the preset criteria among the remaining candidate oil film regions is determined as the target oil film region. The geometric center coordinates or the bounding polygon of the target oil film area are used as the latitude and longitude coordinates of the oil spill source at the oil spill location, or the area description information that can calculate the geometric center coordinates of the oil spill source is used.

7. The rapid prediction and response method for oil spills in nearshore waters according to claim 1, characterized in that, In step S5, the oil spill prediction model employs the Lagrange particle tracking method, specifically including: The oil spill is discretely characterized as multiple oil particles, each carrying location and oil quantity attributes; Within each prediction step, the horizontal displacement of oil particles is calculated based on the ocean current field data, the wind-induced drift displacement is calculated based on the meteorological driving data, and the oil particles are updated by random walk in combination with the diffusion coefficient to obtain the temporal position set of oil particles in the next period. Based on the set of time-series locations, the spatiotemporal distribution of the oil film is reconstructed, and the oil spill diffusion trajectory, oil film coverage area, and oil film thickness distribution are output.

8. The rapid prediction and response method for oil spills in nearshore waters according to claim 1, characterized in that, The method further includes: S6, a step of dynamically correcting the parameters of the oil spill prediction model by setting a correction period and using a sliding window correction strategy, specifically including: Within each correction cycle, acquire the actual distribution data of the oil film observed by synthetic aperture radar or optical radar; The actual distribution data of the oil film is compared with the simulation results of the oil spill prediction model. Based on the results of the deviation analysis, the model parameters of the oil spill prediction model are dynamically adjusted. The adjusted model parameters are then used to recalculate the oil spill diffusion information for the remaining period from the current moment to the end of the next prediction cycle, so as to achieve rolling updates of subsequent prediction results.

9. A system based on the rapid prediction and response method for oil spills in nearshore waters as described in any one of claims 1-8, characterized in that, include: The meteorological field model construction unit is used to perform numerical forecasting of nearshore waters based on the meteorological field model and to obtain meteorological driving data for oil spill diffusion calculation. The hierarchical hydrodynamic coupling unit is used to construct a hierarchical marine hydrodynamic model system. It combines the meteorological driving data and the spatial scale characteristics of the sea area where the oil spill event is located, matches and runs the corresponding level of the bay model in the marine hydrodynamic model system to obtain marine flow field data for oil spill diffusion calculation. The business process management unit is used to introduce business process management tools and configure a unified timed automated scheduling workflow for the meteorological field model and the ocean hydrodynamic model system based on the business process management tools. The first processing unit is used to determine the initial location of the oil spill, i.e. the spatial location of the sea area where the oil spill is located, based on the oil spill event information of the current prediction period. Based on the initial location of the oil spill, the corresponding meteorological driving data and ocean current field data are dynamically matched through the timed automated scheduling workflow to form the background field data required for oil spill prediction. The second processing unit is used to call a predefined oil spill prediction model, and based on the background field data, output the oil spill diffusion trajectory information at multiple time scales in the next prediction period, as well as the affected sensitive target areas.

Images