South China sea surface pollutant diffusion path analysis method based on drifting buoy trajectory data

Abstract

The invention discloses a South China Sea surface pollutant diffusion path analysis method based on drifting buoy trajectory data. The method comprises the following steps: S1, acquiring and screening buoy trajectory data passing through a pollution source and an adjacent area; s2, performing standardization processing on the trajectory and constructing a unified sample library; s3, according to the position of the pollution source, calling the sample library to carry out Lagrange tracing analysis of instantaneous emission, or simulating continuous emission through random sampling with release; and S4, performing gridding probability density calculation and risk assessment based on an analysis or simulation result. The core of the method lies in that an analysis framework with clear logic is constructed, data preparation and business analysis are separated, limited single diffusion observation is converted into simulation of long-term continuous emission through a random resampling technology, and on the basis of real observation data, the analysis framework is used for analyzing the long-term continuous emission of the long-term continuous emission of the long-term continuous emission of the long-term continuous emission of the long-term continuous emission. The accuracy and reliability of South China Sea pollutant diffusion path prediction and risk assessment are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of marine environmental monitoring and pollutant diffusion simulation technology, specifically to a method for analyzing the diffusion paths of surface pollutants in the South China Sea based on drifting buoy trajectory data. Background Technology

[0002] As an important marginal sea of ​​the western Pacific Ocean, the South China Sea is strongly influenced by a variety of dynamic processes, including the Kuroshio Current intrusion, monsoon-driven circulation, and multi-scale turbulent processes. The diffusion paths of pollutants (such as oil spills, radionuclides, and plastic debris) in this sea area are controlled by these complex dynamic processes. Accurately assessing their diffusion direction, speed, and potential affected areas is of great strategic significance for marine ecological protection, maritime risk management, and pollution emergency response.

[0003] Currently, marine pollutant dispersion prediction mainly relies on numerical simulation methods (such as ROMS, HYCOM, and POM models), which simulate circulation fields and track pollutant trajectories by solving fluid dynamic equations. However, due to limited model flow field accuracy, imperfect parameterization schemes, and the accumulation of initial field errors, simulation results often deviate significantly from actual dispersion paths. This is especially true in complex marginal sea environments like the South China Sea, where mesoscale eddies and internal waves are difficult to simulate accurately, leading to significant prediction uncertainties. While remote sensing monitoring methods can provide real-time spatial distribution of pollutants, they cannot directly predict future dispersion paths and are limited in spatiotemporal coverage by cloud cover and sea state. Empirical statistical methods are simple and easy to use but lack physical mechanism support, limiting their applicability and accuracy. The Lagrange tracing method, which uses drifting buoy trajectories to directly reflect real surface current movements, is an important observational tool for dispersion analysis. The Global Drifting Buoy Program (GDP) has accumulated a large amount of surface flow field observation data.

[0004] However, existing technologies for the South China Sea region still face the following challenges: First, data utilization efficiency and methodological systems are incomplete. There is a lack of systematic methods specifically for assessing pollutant diffusion pathways. Key technical aspects such as efficiently screening buoy trajectories related to pollution sources, conducting quality control, and converting discrete trajectories into continuous diffusion information have not yet been standardized. Second, there is insufficient understanding of pollutant diffusion patterns under real ocean current conditions. The marine dynamic information contained in limited buoy data is not fully extracted, making it difficult to quantify the probability distribution of different diffusion directions and assess the risk level of pollutants reaching sensitive areas. Third, the ability to simulate continuous emission scenarios is insufficient. Traditional single-release analysis cannot reflect the spatiotemporal evolution of long-term pollution sources, and there is a lack of spatial visualization and quantitative expression methods for diffusion results. Therefore, there is an urgent need to develop a South China Sea surface pollutant diffusion pathway analysis method based on real observation data that can systematically address the above-mentioned technical bottlenecks, in order to improve pollution prediction and risk assessment capabilities and provide reliable decision-making basis for marine environmental protection and pollution emergency management. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a method for analyzing the diffusion paths of surface pollutants in the South China Sea based on drifting buoy trajectory data. This method systematically solves the core technical problems of traditional methods in data utilization, understanding of real ocean current patterns, and simulation of continuous emissions by constructing an integrated technical framework of "data preparation - sample construction - application analysis - risk assessment".

[0006] To solve the above-mentioned technical problems, the technical solution proposed in this application is as follows: This invention provides a method for analyzing the diffusion paths of surface pollutants in the South China Sea based on drifting buoy trajectory data, comprising the following steps: S1. Data Acquisition and Screening Steps: Acquire buoy trajectory data from the Global Drifting Buoy Program, set the research area, define the source area based on the specified pollution source location, screen buoy trajectories that pass through the pollution source and its adjacent area according to preset screening criteria, and perform quality control processing on the screened buoy trajectories. S2. Trajectory Standardization and Sample Library Construction Steps: The buoy trajectories after screening and quality control are regarded as the tracer paths of pollutant particles. The starting points of all trajectories are standardized to construct a standardized trajectory sample library. S3. Path Analysis and Simulation Steps: Selectively perform one or all of the following analyses based on the type of pollution source: For transient pollution sources, the standardized trajectory sample library constructed in step S2 is used to perform Lagrange tracer analysis of the pollutant diffusion path; For persistent pollution sources, continuous release of pollutants is simulated by random sampling with replacement from the standardized trajectory sample library constructed in step S2. S4. Probability density distribution calculation steps: Divide the study area into spatial grids. Based on the path analysis or simulation results obtained in step S3, count the number of times each grid cell is traversed by the trajectory, calculate the probability of pollutants reaching each grid cell, and perform visualization and risk level classification.

[0007] Furthermore, the quality control process in step S1 includes the following sub-steps: S11. Outlier identification and removal: Calculate the displacement velocity at adjacent observation times. If the velocity exceeds the threshold set based on the typical value of the surface current velocity in the South China Sea, it is determined to be an outlier and removed. S12. Interpolation processing of missing data: For data segments with a missing duration of less than or equal to 48 hours, linear interpolation is used to supplement location information; for data segments with a missing duration of more than 48 hours, the trajectory of that segment is disconnected. S13. Segmentation of multiple buoy deployments: Separate different deployment records of the same buoy into independent trajectory segments and assign a unique identifier to each trajectory segment.

[0008] Furthermore, the trajectory starting point standardization process in step S2 includes the following sub-steps: S21. Spatial standardization: The point at which the buoy trajectory first enters the source region or is closest to the center of the source region is defined as the starting point of the trajectory. S22. Time standardization: The actual time when the buoy arrives at the starting point, or the moment it leaves the source area, is defined as the relative time zero point of the trajectory.

[0009] Furthermore, in step S3, the Lagrange tracer analysis for transient pollution sources includes the following sub-steps: S31. Multi-time node statistics: Statistically analyze the positional distribution of all trajectories at multiple preset relative time nodes; S32. Calculation of diffusion characteristic parameters: Based on the location distribution, calculate the diffusion radius, main diffusion direction and diffusion velocity.

[0010] Furthermore, in sub-step S32, the specific method for calculating the diffusion characteristic parameters includes: The diffusion radius is obtained by calculating the distances of all trajectory positions relative to the center of the source region and taking the 90th percentile. The main diffusion direction is extracted by principal component analysis of the spatial distribution of the trajectory positions. The diffusion velocity is obtained by calculating the movement speed of the centroid of the trajectory group at different time points.

[0011] Furthermore, in step S3, the simulation of persistent pollution sources includes the following sub-steps: S33. Parameter settings: Set the total simulation duration and emission frequency; S34. Random Sampling and Trajectory Generation: Based on the emission frequency and the total simulation duration, random sampling with replacement is performed from the standardized trajectory sample library to simulate the continuous release of pollutants within the total simulation duration, generating a set of trajectories for all virtual pollutants. S35. Trajectory Overlay and Effect Calculation: The trajectory set is spatiotemporally overlaid to form a spatial density distribution reflecting the frequency of trajectory passage, and the cumulative residence time of pollutants in the sea area is statistically analyzed.

[0012] Furthermore, it also includes: S36. Parameter sensitivity analysis: By changing at least one of the parameters, namely emission frequency, total simulation duration and / or number of samplings, run multiple simulations to analyze the impact of different parameter configurations on the final spatial density distribution of pollutants.

[0013] Furthermore, in step S4, the specific method for calculating the probability of pollutant arrival is as follows: P i,j =(n i,j / N)×100%, where P i,j Let n be the probability of pollutant arrival in the i-th row and j-th column. i,j The number of buoy tracks passing through this grid is N, where N is the total number of simulated buoy tracks.

[0014] Furthermore, in step S4, the probability calculation based on the simulation results of persistent pollution sources is further adjusted by weights, and the specific calculation formula is as follows: ,in, Let i be the weighted arrival probability of the grid in the i-th row and j-th column. Let k be the time the k-th buoy stays within this grid. To simulate the total duration.

[0015] On the other hand, this application also claims protection for an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the above-described method for analyzing the diffusion paths of surface pollutants in the South China Sea based on drifting buoy trajectory data.

[0016] Compared with the prior art, the present invention achieves the following beneficial technical effects: This method effectively overcomes the errors of traditional numerical simulations by utilizing real drifting buoy trajectory data, significantly improving the accuracy and reliability of pollutant diffusion path prediction in the South China Sea. Its innovative random resampling technology enables efficient simulation of the long-term cumulative diffusion patterns of persistent pollution sources. Through a probabilistic and gridded risk characterization system, it can intuitively quantify the risk levels of pollutants reaching different sea areas and reveal their dynamic evolution patterns across multiple time scales, thus providing a more scientific and reliable basis for marine environmental supervision and pollution emergency response. Attached Figure Description

[0017] 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating a method for analyzing the diffusion paths of surface pollutants in the South China Sea based on drifting buoy trajectory data, provided as an embodiment of the present invention.

[0019] Figure 2The map shows the trajectory and number distribution of drifting buoys in the South China Sea and its adjacent waters. (a) is the buoy trajectory map, and (b) is the trajectory density map. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] The present invention discloses a method for analyzing the diffusion paths of surface pollutants in the South China Sea based on drifting buoy trajectory data. The overall technical process is as follows: Figure 1 As shown. The core of this method lies in constructing a hierarchical and interconnected technical framework. This framework first lays the foundation for the reliability of the analysis through rigorous data preprocessing, then builds a standardized data sample library as a unified source for all analyses, subsequently conducts refined path analysis or stochastic simulation based on the type of pollution source, and finally outputs the risk assessment results through probabilistic and visualization methods. There are close dependencies and synergistic effects between each step.

[0022] S1. Data Acquisition and Filtering Steps: Constructing a High-Quality Input Dataset This step aims to accurately filter out high-quality buoy trajectories related to pollution sources from massive amounts of raw data, and it forms the data foundation for the entire method.

[0023] First, download historical buoy trajectory data for the South China Sea and adjacent waters (10°S~30°N, 100°E~140°E) from the official website of the Global Drift Buoys Programme (GDP). Define the boundaries of the study area and exclude buoys outside this range. For example... Figure 2 As shown, there are more than 1,200 buoy tracks in the South China Sea and its adjacent waters, spanning from February 14, 1979 to the present. The spatial distribution is more concentrated in the northern part of the South China Sea and the Luzon Strait.

[0024] More importantly, this method establishes a systematic set of pollution source correlation buoy screening criteria and quality control procedures. Based on the specified pollution source location (such as the oil spill accident site), the source area is defined (usually a circular area with a radius of 0.5° to 2°), and screening is carried out according to multiple criteria such as spatial proximity, temporal correlation, and directional consistency.

[0025] To ensure data reliability, the quality control process is broken down into three key sub-steps: Outlier identification and removal: By calculating the displacement velocity at adjacent observation times and comparing it with a threshold (such as 5 m / s) set based on typical surface current velocity values ​​in the South China Sea, outlier jump points caused by positioning errors can be effectively identified and removed.

[0026] Data missing interpolation processing: To address the issue of missing data, a differentiated strategy is adopted based on the duration of the missing data: For short-term missing data (≤48 hours), linear interpolation is used to supplement the missing data to ensure spatiotemporal continuity; for long-term missing data (>48 hours), the trajectory is decisively disconnected to avoid introducing excessive errors.

[0027] Multiple buoy deployment segmentation: Identify and separate multiple deployment records of the same buoy into independent trajectory segments, and assign a unique identifier to each segment to ensure the independence and accuracy of each analysis object.

[0028] Through the above-mentioned refined processing, a high-quality trajectory dataset suitable for subsequent analysis is finally output.

[0029] S2. Trajectory Standardization and Sample Library Construction Steps: Creating a Unified Analysis Benchmark This step places all the selected trajectories under a unified spatiotemporal reference system to construct a standardized "trajectory sample library".

[0030] This process includes two essential standardization sub-steps: Spatial standardization: The point at which each buoy's trajectory first enters the source area or is closest to the center of the source area is clearly defined as the starting point of that trajectory. This unifies the diffusion starting point of all trajectories to the vicinity of the pollution source, making different trajectories comparable.

[0031] Time standardization: The actual time when the buoy reaches the aforementioned starting point (or the moment it leaves the source region) is defined as the relative time zero point (t=0) of the trajectory. All subsequent observation points are recorded using relative time, thereby eliminating the influence of different actual observation times and facilitating unified spatiotemporal evolution analysis.

[0032] Based on all standardized trajectories, a standardized trajectory sample library is ultimately constructed for the cost method. This provides a data source for all subsequent path analysis and simulation operations, ensuring the consistency of the analysis process and the repeatability of the results.

[0033] S3. Path Analysis and Simulation Steps: Implementing Multi-Scenario Diffusion Simulation Based on the essential characteristics of pollution sources, this step provides two parallel technical paths to conduct detailed simulations of instantaneous and continuous emissions, respectively.

[0034] For transient pollution sources (such as sudden oil spills), this method utilizes the entire sample library for Lagrange tracer analysis. This analysis is further refined into two levels: Dynamic tracking across multiple time scales: At multiple preset time points, such as short-term (e.g., 1-7 days), medium-term (e.g., 1-3 months), and long-term (e.g., 6-12 months), the location distribution of all trajectories is statistically analyzed to capture the full-process characteristics of pollutant diffusion.

[0035] Quantitative extraction of diffusion characteristic parameters: Based on location distribution, a series of quantitative parameters are calculated, including: diffusion radius (characterized by the 90th quantile distance), main diffusion direction (objectively extracted from spatial distribution using principal component analysis), and diffusion velocity (calculated by the movement of the centroid of the trajectory group). Furthermore, by grouping and comparing these parameters by season, the influence of seasonal factors such as monsoons on diffusion paths can be effectively revealed.

[0036] For persistent pollution sources (such as stochastic simulations of oil spills), this method innovatively employs a sample library-based random resampling simulation technique. This simulation is a complete process including parameter setting, cyclic sampling, and effect assessment. Set simulation parameters, including the total simulation duration and emission frequency (e.g., once a month or once a week).

[0037] Random sampling and trajectory generation: Based on the emission frequency, random sampling with replacement is performed from a standardized sample library. Each sample represents the release of a pollutant particle at a specific moment. By repeating this process, a large-scale set of virtual particle trajectories can be generated, thereby simulating long-term continuous emissions.

[0038] Trajectory overlay and cumulative effect calculation: All virtual trajectories are spatiotemporally overlaid to generate a "heat map" reflecting the frequency of pollutant passage, and the cumulative residence time in sensitive areas is calculated to assess the long-term exposure risk.

[0039] To further evaluate the robustness of the simulation results, this method can also introduce parameter sensitivity analysis. By changing parameters such as emission frequency, the changes in the output results can be observed, thereby determining reasonable parameter configurations and understanding the uncertainty of the model.

[0040] S4. Probability density distribution calculation steps: Output a quantitative risk map. This step aims to transform the path information generated by the aforementioned analysis or simulation into an intuitive and quantitative spatial risk distribution.

[0041] First, the study area is divided into regular spatial grids (e.g., 1°×1°). Then, the core task is to apply two probabilistic calculation models with different focuses: Basic arrival probability model: Count the total number of times each grid cell is traversed by all trajectories, and calculate the percentage P of the total number of trajectories. i,j =(n i,j / N)×100%, where P i,j Let n be the probability of pollutant arrival in the i-th row and j-th column. i,j Let N be the number of buoy trajectories passing through the grid, and let N be the total number of simulated emission buoy trajectories. This model intuitively reflects the probability that pollutants will reach any grid after originating from the source region.

[0042] Weighted probability model: For continuous emission simulations, the buoy's dwell time within the grid can be further considered to calculate the weighted probability of arrival. ,in, Let i be the weighted arrival probability of the grid in the i-th row and j-th column. Let k be the time the k-th buoy stays within this grid. (Total simulation duration). Weighted probability of arrival can more accurately reflect the cumulative concentration distribution of pollutants in different sea areas.

[0043] Finally, the calculated probability values ​​are visualized using color codes, and the study area is classified into risk levels based on preset probability thresholds, forming a surface pollutant probability density distribution map that can be directly used for decision support.

[0044] Through the close integration and progressive deepening of the above four steps, this invention achieves a complete transformation from raw buoy data to quantitative risk information, providing a reliable technical tool for pollutant control in the South China Sea.

[0045] In this application, the quality control process in the data acquisition and filtering step (S1) is a refined process that includes multiple verification mechanisms, specifically implemented through the following sub-steps: S11. Outlier Identification and Removal: This sub-step aims to filter out invalid data caused by sensor malfunctions or signal interference. Its core is calculating the displacement velocity between two adjacent observation times (typically 6 hours). To ensure the rationality of the criteria, the velocity threshold is scientifically set based on typical hydrodynamic values ​​of surface current velocities in the South China Sea (e.g., 20 km / h or approximately 5 m / s). Any displacement point exceeding this threshold will be marked as an outlier and automatically removed by the system. To further improve accuracy, a sliding window method or the 3σ criterion can be used for continuous anomaly detection to avoid misjudging individual instantaneous jump points.

[0046] S12. Interpolation Processing of Missing Data: This sub-step is responsible for repairing data interruptions and ensuring the spatiotemporal continuity of the trajectory. The system first identifies observation intervals outside the normal 6-hour interval. For short-term missing data (≤48 hours, i.e., no more than 8 normal observation points), a linear interpolation algorithm is used to supplement the position information of the intermediate time based on the valid observation points at both ends of the missing segment. For long-term missing data (>48 hours), the trajectory is considered to have broken at this point, and this segment of the trajectory is broken at the missing point and treated as two independent trajectory segments to avoid introducing excessive speculation errors.

[0047] S13. Segmented Processing of Multiple Buoy Deployments: This sub-step ensures that each independent ocean observation event is analyzed independently. The system separates different deployment records under the same buoy ID based on the deployment count recorded in the data, or by automatically identifying long time intervals. Each deployment is assigned a unique new trajectory identifier (e.g., ID_01, ID_02), and its deployment time, location, and duration are recorded independently. This effectively prevents the erroneous mixing of ocean current observation data from different periods and under different initial conditions, ensuring the clarity of the physical meaning of the analysis results.

[0048] The trajectory standardization and sample library construction step (S2) involves trajectory starting point standardization, which is crucial for ensuring the comparability of all trajectories. Specifically, it includes: S21. Spatial Normalization: This sub-step defines a unified spatial starting point for all trajectories. The system iterates through each trajectory and determines the normalized starting point (lon_0, lat_0) as the location where it first enters the predefined source region polygon, or the observation point that is geometrically closest to the center of the source region throughout the entire trajectory. This point is given a clear physical meaning: representing the initial location where pollutants begin to diffuse from the source region.

[0049] S22. Time Standardization: This sub-step establishes a unified timeline for all trajectories. The system explicitly defines the actual observation time of the buoy reaching the aforementioned spatial starting point (if the starting point is the point of entry into the source region), or the time when the buoy finally leaves the boundary of the source region, as the relative time zero point (t=0) of that trajectory. Thereafter, all observation times on that trajectory are converted into offsets relative to this zero point (e.g., t=6h, t=12h, ...). This allows buoy trajectories deployed in different years and seasons to be compared and analyzed within the same time frame.

[0050] In this application, the Lagrange tracer analysis of transient pollution sources in the path analysis and simulation step (S3) is further refined through the following sub-steps: S31. Multi-Time Node Statistics: This sub-step aims to characterize the dynamic process of pollutant diffusion. The system does not only focus on the final location, but also "freezes" all trajectories at a series of pre-defined representative relative time nodes (e.g., t=1 day, 7 days, 30 days, 90 days, 180 days, 360 days) and statistically analyzes their latitude and longitude positions at each time point. This constitutes the basic dataset for analyzing the spatiotemporal evolution of diffusion.

[0051] S32. Calculation of diffusion characteristic parameters: This sub-step extracts quantitative characteristic indicators from the location set.

[0052] Dispersion radius: At a specific time t, calculate the spherical distance between all trajectory locations and the center of the source region. To eliminate the influence of extreme outliers, instead of using the maximum distance, the statistical 90th percentile distance is adopted as the effective diffusion radius of the pollutant at time t, which better reflects the diffusion range of the main pollutant.

[0053] Main diffusion direction: The latitude and longitude coordinates of all trajectory points at time t are considered as a two-dimensional spatial point set, and principal component analysis is performed on this point set. The extracted first principal component direction, that is, the direction with the largest data variance, is defined as the main diffusion direction of pollutants at that time, objectively revealing the mainstream diffusion path.

[0054] Diffusion velocity: First, calculate the spatial distribution centroid (center of mass) of all trajectory locations at adjacent time points (e.g., t1 and t2). Then, calculate the distance the centroid moves from t1 to t2, and divide it by the time interval (t2-t1) to obtain the average diffusion velocity of the pollutant plume as a whole during this time period.

[0055] In this application, the path analysis and simulation step (S3) for simulating persistent pollution sources is a complete simulation process based on random resampling technology, implemented through the following sub-steps: S33. Parameter Settings: This sub-step defines the simulation scenario. Users need to set core parameters, including the total simulation duration (e.g., 3 years or 1095 days) and the frequency of pollutant emissions (e.g., once a month or once a week). These parameters collectively determine the scale and granularity of the simulation.

[0056] S34. Random Sampling and Trajectory Generation: This sub-step is the core of the simulation. Based on the set emission frequency, at each emission time point (e.g., the first day of each month), the system randomly selects a specified number of trajectories with replacement from the standardized trajectory sample library constructed in S2. Each selected trajectory represents a pollutant being released from the source region at that moment and moving along this real trajectory in the future. This process is repeated until the entire simulation duration is covered, ultimately generating a set containing tens of thousands of virtual particle trajectories, thus achieving statistical simulation of long-term, continuous emissions.

[0057] S35. Trajectory Superposition and Effect Calculation: This sub-step synthesizes and interprets the simulation results.

[0058] Spatial overlay: All virtual particle trajectories generated by S34 (e.g., 3600) are overlaid on the same geographic base map. By calculating the total number of times each geographic location (or grid) is traversed by the trajectory and visualizing it using color depth or density, a "heat map" of pollutant diffusion is formed. High-density areas are high-risk areas.

[0059] Cumulative effect calculation: For specific sea areas of concern (such as fishing grounds, marine protected areas, and coastal zones), the system tracks the entry and exit times of each virtual trajectory within that sea area and calculates its residence time. By summing the residence times of all trajectories, the cumulative exposure time of pollutants in that sea area can be obtained, providing a key indicator for assessing long-term ecological risks.

[0060] Furthermore, the path analysis and simulation step (S3) may also include a parameter sensitivity analysis sub-step (S36): this sub-step is used to evaluate the robustness of the model output. By systematically changing the key parameters in S33 (e.g., adjusting the monthly emission frequency from 10 to 5 or 15), and rerunning S34 and S35, the differences in the final spatial density distribution map and cumulative exposure time are observed and compared. This helps to understand the dependence of the model conclusions on the input parameters, determine reasonable parameter value ranges, and enhance the reliability of decision recommendations.

[0061] The core of the probability density distribution calculation step (S4) lies in transforming path information into a probability field, and its specific calculation method includes: Basic arrival probability calculation: This method focuses on assessing the risk of a contaminant "arriving" at a given area. For each spatial grid (i, j), the number of times n that all used trajectories (all trajectories in the sample library for transient analysis; all sampled virtual trajectories for continuous simulation) traverse that grid is counted in step S3. i,j Then, calculate n. i,j The ratio of this ratio to the total number of trajectory attempts N, multiplied by 100%, gives the arrival probability P of that grid. i,j The formula is: P i,j = (n) i,j / N) ×100%, where P i,j Let n be the probability of pollutant arrival in the i-th row and j-th column. i,j Let N be the number of buoys passing through the grid, and N be the total number of simulated buoys released. This probability intuitively reflects the relative likelihood of each point in space being affected by pollution.

[0062] Weighted probability calculation: This method is applicable to persistent pollution source simulation and aims to more accurately assess the risk of "pollution level". It introduces "dwell time" as a weight on top of the basic method. For each grid (i, j), instead of simply counting, the dwell time of all trajectories within that grid is summed. ), then divide by the sum of the total durations of all trajectories ( Then multiply by 100%. The formula is: ,in, Let i be the weighted arrival probability of the grid in the i-th row and j-th column. Let k be the time the k-th buoy stays within this grid. This is to simulate the total duration. This weighted probability better reflects the potential cumulative concentration distribution of pollutants in space, because the longer the residence time, the higher the probability of pollutant accumulation in the area.

[0063] The following is a further description with reference to the embodiments: Example 1 This embodiment uses a hypothetical oil spill site (an instantaneous pollution source) in the South China Sea as an example to illustrate the implementation process of this method in detail.

[0064] S1. Data Acquisition and Filtering Download historical buoy trajectory data for the South China Sea and adjacent waters (10°S~30°N, 100°E~140°E) from the official website of the Global Drifter Program (GDP). Set the boundary of the study area and eliminate trajectories outside the range. Assume that the oil spill occurred in the central part of the South China Sea (113°E, 12°N). Define a circular source area with a radius of 1° centered on this point. Traverse the buoy trajectories and select those that meet the following conditions: (1) Spatial condition: the trajectory has at least one observation point located within the source area or less than 0.2° from the source area boundary; (2) Temporal condition: the trajectory stays near the source area for at least 6 hours; (3) Directional condition: the trajectory moves at least 0.5° after leaving the source area.

[0065] Subsequently, quality control was performed on the selected trajectories: S11. Outlier Identification and Removal: Calculate the displacement velocity at adjacent observation times (6-hour interval), set a velocity threshold of 5 m / s, and mark observation points exceeding this threshold as outliers and remove them.

[0066] S12. Interpolation processing of missing data: For cases where the missing data duration is ≤48 hours, linear interpolation is used to supplement the location information with a 6-hour resolution; for cases where the missing data duration is >48 hours, the trajectory segment is disconnected.

[0067] S13. Segmentation processing of multiple buoy deployments: Based on the deployment records in the data, different deployment segments with the same buoy ID are separated into independent trajectories and new trajectory identifiers are assigned.

[0068] S2, Trajectory Standardization and Sample Library Construction All buoy trajectories after screening and quality control are considered as tracer paths for contaminant particles.

[0069] S21. Spatial Normalization: The position where each buoy trajectory first enters the source area (or the point closest to the center of the source area) is defined as the starting point (lon_0, lat_0) of the trajectory.

[0070] S22. Time standardization: The time when the buoy reaches the starting point is defined as the relative time zero point (t=0) of the trajectory, and all subsequent observation times are converted to relative time relative to this zero point.

[0071] Based on all standardized trajectories, a standardized trajectory sample library is constructed for this analysis.

[0072] S3, Path Analysis and Simulation This example involves a transient pollution source, so all trajectories from the standardized trajectory sample library constructed by S2 are used for Lagrange tracer analysis.

[0073] S31. Multi-time point statistics: Statistically analyze the location distribution of all trajectories at multiple time points such as t=7 days (short-term), t=30 days (medium-term), and t=90 days (long-term).

[0074] S32. Calculation of diffusion characteristic parameters: Diffusion radius: Calculate the distance of all buoys from the center of the source region at time t, and take the 90th percentile as the effective diffusion radius at that time.

[0075] Main diffusion direction: Perform principal component analysis on the spatial coordinates of all buoy positions at time t, and the direction of the first principal component is the main diffusion direction.

[0076] Diffusion velocity: The average diffusion velocity is obtained by dividing the distance the centroid of the trajectory group moves between adjacent time nodes (e.g., from t=0 to t=30 days) by the time.

[0077] In addition, the trajectories were divided into a winter group (November-March) and a summer group (May-September) according to the deployment season, and S31 and S32 were repeated respectively to compare the differences in diffusion characteristics between the two seasons.

[0078] S4. Calculation of probability density distribution A 1°×1° latitude and longitude grid was established for the study area (10°S~30°N, 100°E~140°E).

[0079] Based on all the trajectories obtained from the S3 analysis, count the number of times n that each grid is traversed i,j . The total number of trajectories N is the total number of trajectories used for analysis in the sample library

[0080] Calculate the arrival probability P of each grid i,j = (n i,j / N) × 100%

[0081] Visualize the probability data as a spatial distribution map, using a color scale to represent the probability magnitude. Geographical elements such as coastlines and marine protected areas can be superimposed. Divide the risk levels according to the probability values. For example: P > 10% is the high-risk area, 5% < P ≤ 10% is the medium-risk area, 1% < P ≤ 5% is the low-risk area, and P ≤ 1% is the extremely low-risk area

[0082] Example 2 In this example, taking the random simulation (persistent pollution source) of an oil spill accident as an example, the implementation process of this method is illustrated. The S1 and S2 steps are the same as those in Example 1, and a high-quality standardized trajectory sample library is constructed

[0083] S3. Path Analysis and Simulation In this example of a persistent pollution source, therefore, from the standardized trajectory sample library constructed in S2, simulate the continuous emission process of pollutants through random sampling with replacement (Bootstrap method)

[0084] S33. Parameter Setting: Set the total simulation duration to 3 years (36 months), and the emission frequency to 100 pollutant particles per month

[0085] S34. Random Sampling and Trajectory Generation: Randomly select 30 trajectories (allowing duplicates) from the sample library each month to simulate the movement paths of 30 pollutants emitted in that month. Repeat this process for 36 months to generate a total of 1080 virtual pollutant particle trajectories

[0086] S35. Trajectory Superposition and Effect Calculation: Superimpose all 1080 trajectories in the geographical space to form a trajectory density map (heat map), and the color depth reflects the passing frequency. Statistically calculate the cumulative residence time of pollutants in specific sensitive sea areas (such as coastal fishing grounds and marine parks) <L

[0087] In addition, perform parameter sensitivity analysis: Change the emission frequency (such as 10 times and 20 times per month) and re-simulate, observe and compare the changes in the final spatial density distribution patterns

[0088] S4. Probability Density Distribution Calculation Establish a 1°×1° latitude-longitude grid for the study area

[0089] Traverse all 1080 virtual particle trajectories and count the number of times n that each grid is traversed i,j The total number of trajectories N is the total number of random samplings, which is 1080.

[0090] Calculate the arrival probability P of each grid i,j = (n i,j / N) × 100%.

[0091] Calculate the weighted arrival probability of each grid , where is the weighted arrival probability of the grid in the i-th row and j-th column,[[]] is the residence time of the k-th buoy in this grid,[[]] is the total simulation time.

[0092] Visualize the probability data as a spatial distribution map, using a color scale to represent the probability magnitude. Geographical elements such as the coastline and marine protected areas can be superimposed. Divide the risk levels according to the probability values. For example: P > 10% is a high-risk area, 5% < P ≤ 10% is a medium-risk area, 1% < P ≤ 5% is a low-risk area, and P ≤ 1% is a very low-risk area.

[0093] Use the Bootstrap method to quantify prediction uncertainty: Randomly draw 1000 sub-sample sets with replacement from the original sample library. Each sub-sample set contains 1080 trajectories. Run the complete S3 and S4 processes for each sub-sample set to obtain 1000 sets of probability distributions. Then calculate the mean and standard deviation of the arrival probability of each grid cell to evaluate the reliability of the prediction.

[0094] Finally, it should be noted that: The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it; Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: They can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; And these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for analyzing diffusion path of surface pollutants in the South China Sea based on drift buoy trajectory data, characterized in that, Includes the following steps: S1. Data Acquisition and Screening Steps: Acquire buoy trajectory data from the Global Drifting Buoy Program, set the research area, define the source area based on the specified pollution source location, screen buoy trajectories that pass through the pollution source and its adjacent area according to preset screening criteria, and perform quality control processing on the screened buoy trajectories. S2. Trajectory Standardization and Sample Library Construction Steps: The buoy trajectories after screening and quality control are regarded as the tracer paths of pollutant particles. The starting points of all trajectories are standardized to construct a standardized trajectory sample library. S3. Path Analysis and Simulation Steps: Selectively perform one or all of the following analyses based on the type of pollution source: For transient pollution sources, the standardized trajectory sample library constructed in step S2 is used to perform Lagrange tracer analysis of the pollutant diffusion path; For persistent pollution sources, continuous release of pollutants is simulated by random sampling with replacement from the standardized trajectory sample library constructed in step S2. S4. Probability density distribution calculation steps: Divide the study area into spatial grids. Based on the path analysis or simulation results obtained in step S3, count the number of times each grid cell is traversed by the trajectory, calculate the probability of pollutants reaching each grid cell, and perform visualization and risk level classification.

2. The method of claim 1, wherein, The quality control process in step S1 includes the following sub-steps: S11. Outlier identification and removal: Calculate the displacement velocity at adjacent observation times. If the velocity exceeds the threshold set based on the typical value of the surface current velocity in the South China Sea, it is determined to be an outlier and removed. S12. Interpolation processing of missing data: For data segments with a missing duration of less than or equal to 48 hours, linear interpolation is used to supplement location information; for data segments with a missing duration of more than 48 hours, the trajectory of that segment is disconnected. S13. Segmentation of multiple buoy deployments: Separate different deployment records of the same buoy into independent trajectory segments and assign a unique identifier to each trajectory segment.

3. The method according to claim 1, characterized in that, The trajectory starting point standardization process in step S2 includes the following sub-steps: S21. Spatial standardization: The point at which the buoy trajectory first enters the source region or is closest to the center of the source region is defined as the starting point of the trajectory. S22. Time standardization: The actual time when the buoy arrives at the starting point, or the moment it leaves the source area, is defined as the relative time zero point of the trajectory.

4. The method according to claim 1, characterized in that, In step S3, the Lagrange tracer analysis for transient pollution sources includes the following sub-steps: S31. Multi-time node statistics: Statistically analyze the positional distribution of all trajectories at multiple preset relative time nodes; S32. Calculation of diffusion characteristic parameters: Based on the location distribution, calculate the diffusion radius, main diffusion direction and diffusion velocity.

5. The method according to claim 4, characterized in that, In the S32 sub-step, the specific method for calculating the diffusion characteristic parameters includes: The diffusion radius is obtained by calculating the distance of all trajectory positions relative to the center of the source region and taking the 90th percentile. The main diffusion direction is extracted by principal component analysis of the spatial distribution of the trajectory positions. The diffusion rate is obtained by calculating the movement speed of the centroid of the trajectory group at different time points.

6. The method according to claim 1, characterized in that, In step S3, the simulation of persistent pollution sources includes the following sub-steps: S33. Parameter settings: Set the emission frequency and total simulation duration and / or number of samplings; S34. Random Sampling and Trajectory Generation: Based on the emission frequency and the total simulation duration, random sampling with replacement is performed from the standardized trajectory sample library to simulate the continuous release of pollutants within the total simulation duration, generating a set of trajectories for all virtual pollutants. S35. Trajectory overlay and effect calculation: The trajectory set is spatiotemporally overlaid to form a spatial density distribution reflecting the frequency of trajectory passage, and the cumulative residence time of pollutants in the sea area is statistically analyzed.

7. The method according to claim 6, characterized in that, Also includes: S36. Parameter sensitivity analysis: By changing at least one of the parameters, namely emission frequency, total simulation duration and / or number of samplings, run multiple simulations to analyze the impact of different parameter configurations on the final spatial density distribution pattern of pollutants.

8. The method according to claim 1, characterized in that, The specific method for calculating the pollutant arrival probability in the S4 step is: P i,j = (n i,j / N) x 100%, wherein P i,j is the pollutant arrival probability of the grid at the ith row and the jth column, n i,j is the number of buoy trajectories passing through the grid, and N is the total number of buoy trajectories of the simulated emission.

9. The method according to claim 1, characterized in that, In step S4, the probability calculation based on the simulation results of persistent pollution sources is further adjusted by weights. The specific calculation formula is as follows: ,in, Let i be the weighted arrival probability of the grid in the i-th row and j-th column. Let k be the dwell time of the k-th buoy within this grid. To simulate the total duration.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the South China Sea surface pollutant diffusion path analysis method based on drifting buoy trajectory data as described in any one of claims 1 to 9.

Images