Enteromorpha prolifera green tide full life cycle wind action coefficient determination method
By combining satellite remote sensing and marine observation data with numerical models, the wind effect coefficient of the entire life cycle of Ulva prolifera green tides is dynamically determined, which solves the problem of inaccurate prediction of Ulva prolifera green tide drift paths caused by fixed wind effect coefficients, and realizes higher accuracy in drift path prediction and environmental driving mechanism analysis.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIHAI FORECASTING CENT OF STATE OCEANIC ADMINISTRATION ((QINGDAO MARINE FORECASTING STATION OF STATE OCEANIC ADMINISTRATION) (QINGDAO MARINE ENVIRONMENT MONITORING CENT OF STATE OCEANIC ADMINISTRATION))
- Filing Date
- 2026-03-10
- Publication Date
- 2026-05-19
AI Technical Summary
In existing predictions of Ulva prolifera green tide drift, the wind effect coefficient is treated as a fixed constant, failing to fully consider the dynamic changes at different development stages, resulting in insufficient accuracy in drift path prediction.
By interpreting satellite remote sensing images and tracking observation data of marine drift, combined with a short-term drift numerical model of Ulva prolifera green tide, the optimal wind action coefficient for different development stages was determined, and a dynamic model of wind action coefficient for the entire life cycle of Ulva prolifera green tide was established. The velocity fitting method and trajectory similarity optimization method were combined for optimization.
It significantly improves the prediction accuracy of the drift path and coverage of Ulva prolifera green tides, providing more reliable decision support for disaster prevention and mitigation, and deepening the understanding of the relative contribution and coupling mechanism of ocean currents and wind fields in the long-distance transport of Ulva prolifera green tides.
Smart Images
Figure CN121835513B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of predicting the drift of Ulva prolifera green tides, and more specifically to a method for determining the wind action coefficient of Ulva prolifera green tides throughout their entire life cycle. Background Technology
[0002] Green tides of *Ulva prolifera* are characterized by high biomass, long-distance migration, and significant impacts, severely affecting marine ecosystems, coastal tourism, marine sports, and aquaculture. Controlling green tides of *Ulva prolifera* has become a major task in the marine field. Existing research has found that the development process of green tides of *Ulva prolifera* mainly consists of two stages: the initial floating stage, where suspended green algae, scattered green algae, and small patches gradually grow and aggregate into strips and large patches; and the large-scale formation stage. The surface of *Ulva prolifera* patches has a certain degree of roughness, and their drift is mainly driven by sea surface winds and ocean currents. The wind effect coefficient refers to the ratio of the drift velocity of floating *Ulva prolifera* patches to the wind speed when only sea surface winds are involved, reflecting the dragging effect of wind on the *Ulva prolifera* patches. The patch area and floating state of *Ulva prolifera* green tides vary significantly at different development stages, resulting in different wind effect coefficients at different stages. However, currently, in predicting the drift of *Ulva prolifera* green tides, the wind effect coefficient is mainly determined through empirical data and sea trials, and is generally a fixed constant. This clearly fails to fully consider the dynamic changes in the wind effect coefficient at different stages of development, which to some extent limits the accuracy of drift path prediction. Summary of the Invention
[0003] Based on the above-mentioned technical problems, this invention proposes a method for determining the wind action coefficient of the entire life cycle of Ulva prolifera green tide.
[0004] The technical solution adopted in this invention is:
[0005] A method for determining the wind action coefficient of Ulva prolifera green tide throughout its entire life cycle includes the following steps:
[0006] S1. Divide the different development stages of the green tide of Ulva prolifera into the early stage, the outbreak stage and the extinction stage;
[0007] Acquire satellite remote sensing images of different development stages of the green tide of Ulva prolifera and interpret the satellite remote sensing images to obtain distribution data of the green tide of Ulva prolifera; and / or acquire marine drift tracking observation data of different development stages of the green tide of Ulva prolifera.
[0008] S2. Construct a numerical model for the short-term drift of the green tide of *Ulva prolifera*;
[0009] S3. Obtain surface current and sea surface wind data for the study area;
[0010] S4. Determine the optimal wind action coefficient;
[0011] Using the distribution data of Ulva prolifera green tide obtained from satellite remote sensing images of different development stages in S1, and / or the marine drift tracking observation data, as well as the surface current and sea surface wind data of the study area obtained in S3, and combined with the constructed numerical model of short-term drift of Ulva prolifera green tide, the optimal wind action coefficient for different development stages of Ulva prolifera green tide was determined.
[0012] S5. Establish a dynamic model of the wind action coefficient throughout the entire life cycle of Ulva prolifera green tide;
[0013] Based on the optimal wind action coefficients for different development stages of the Ulva prolifera green tide determined in S4, a quadratic polynomial fitting was performed to establish a dynamic model of wind action coefficients applicable to the entire life cycle of the Ulva prolifera green tide. The wind action coefficient values for the entire life cycle of the Ulva prolifera green tide were determined based on this dynamic model.
[0014] The beneficial technical effects of the present invention are as follows:
[0015] (1) This invention breaks through the limitation of the traditional wind action coefficient being a constant, and constructs a dynamic model of the wind action coefficient applicable to the entire life cycle of Ulva prolifera tides; that is, by dividing the different development stages of Ulva prolifera tides, based on satellite remote sensing monitoring and marine tracking data, and combined with numerical models of short-term drift of Ulva prolifera tides, a quantitative model of the dynamic change of the wind action coefficient with different development stages of Ulva prolifera tides is creatively established. This invention can significantly improve the prediction accuracy of Ulva prolifera tide drift.
[0016] (2) Traditional operational forecasting models use a fixed wind action coefficient (usually based on limited experience or results from specific phases of experiments), ignoring the significant differences in patch area and floating state of *Ulva prolifera* as it progresses from early sporadic patches and large-scale aggregation during outbreaks to its eventual disappearance. This invention establishes a quantitative functional relationship between the wind action coefficient and the growth and disappearance of *Ulva prolifera*. The resulting dynamically changing quantitative model inherently reflects the physical evolution of wind-driven forces as *Ulva prolifera* grows, aggregates, and disappears, significantly improving the prediction accuracy of *Ulva prolifera* drift paths and coverage areas, and providing more reliable decision support for the three lines of defense in disaster prevention and mitigation: "marine salvage, nearshore interception, and shoreline cleanup."
[0017] (3) The present invention uses a combination of velocity fitting method and trajectory similarity optimization method to reasonably determine the optimal wind action coefficient; that is, through a two-level optimization strategy: the first stage determines the initial value of the parameter based on the dynamic subprocess (velocity matching) to ensure that the solution is located in the physically feasible neighborhood; the second stage uses the overall trajectory similarity as the ultimate index for fine-tuning, effectively suppressing the error accumulation effect, thereby obtaining the optimal wind action coefficient that is consistent with the observation in both dynamic response and motion form.
[0018] (4) This invention provides a new method for quantitatively analyzing the relative contributions and coupling mechanisms of ocean currents and wind fields in the long-distance transport of Ulva prolifera green tides, which is of great scientific significance for elucidating the environmental driving mechanism of Ulva prolifera green tides. Attached Figure Description
[0019] Figure 1 This is a flowchart illustrating the method for determining the wind action coefficient throughout the entire life cycle of *Ulva prolifera* green tide according to the present invention.
[0020] Figure 2 This is a comparison chart of the predicted and measured results of satellite remote sensing monitoring of Ulva prolifera green tide under the optimal wind action coefficient in a specific application example of the present invention;
[0021] Figure 3 This is a comparison chart of the drift trajectory of seaweed green tide at sea and the prediction results under the optimal wind action coefficient in a specific application example of the present invention;
[0022] Figure 4 This is a quadratic polynomial fitting curve of the green tide wind effect coefficient of *Ulva prolifera* in a specific application example of the present invention;
[0023] Figure 5 This is a comparison chart of the drift trajectory of the Ulva prolifera green tide predicted based on the wind action coefficient throughout the entire life cycle of the green tide and the traditional wind action coefficient. Detailed Implementation
[0024] The patch area and floating state of *Ulva prolifera* vary significantly at different developmental stages, resulting in different wind influence coefficients for *Ulva prolifera* green tides at different stages. Current *Ulva prolifera* green tide drift prediction models treat the wind influence coefficient as a constant at different developmental stages, failing to fully consider its dynamic changes, which to some extent limits the accuracy of drift path prediction. Based on this, this invention proposes a method for determining the wind influence coefficient throughout the entire life cycle of *Ulva prolifera* green tides. This method constructs a dynamic wind influence coefficient capable of determining the entire life cycle of *Ulva prolifera* green tides, which not only helps improve the prediction accuracy of *Ulva prolifera* green tide drift paths but also has significant scientific implications for a deeper understanding of the different contributions and coupling mechanisms of ocean currents and wind fields in the long-distance transport of *Ulva prolifera* green tides.
[0025] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.
[0026] like Figure 1 As shown, a method for determining the wind effect coefficient of the entire life cycle of *Ulva prolifera* green tide includes the following steps:
[0027] S1. Acquire satellite remote sensing images and marine drift tracking observation data of different development stages of the green tide of seaweed.
[0028] Collect at least 5 years of data on the distribution characteristics of *Ulva prolifera* green tides, such as the annual coverage area time series and the time of the first sporadic discovery of *Ulva prolifera* floating. Calculate the growth rate of the *Ulva prolifera* green tide based on the coverage area. Calculate the average start date of sporadic *Ulva prolifera* floating based on the time of the first sporadic discovery of *Ulva prolifera* floating.
[0029] Based on coverage area and growth rate, the period is divided into early stage, outbreak stage, and decline stage. The early stage of the green tide of Ulva prolifera generally refers to the period before the green tide of Ulva prolifera reaches a large scale (covering an area of more than 5 square kilometers); the outbreak stage refers to the period from the large-scale outbreak to a significant decrease in biomass (growth rate less than -10%), generally from May to early July; the decline stage refers to the period from a significant decrease in biomass to the basic disappearance of the green tide of Ulva prolifera, generally after early July.
[0030] Meanwhile, based on the collected time of the first discovery of sporadic floating of Ulva prolifera each year, the average starting time of sporadic floating of Ulva prolifera was calculated.
[0031] High-resolution satellite remote sensing images of different development stages (early stage, outbreak stage, and decline stage) of the green tide of Ulva prolifera under clear weather conditions (such as the sea area near the Subei Shoal, the sea area near 35 degrees, the sea area north of 35 degrees, and the sea area near Yantai and Weihai) were selected and interpreted to obtain the distribution data of the green tide of Ulva prolifera.
[0032] Collect data from tracking devices or ship positioning observations of green tides of seaweed in different sea areas and at different development stages to obtain long-term data on the drift of green tides of seaweed, generally with a time length of not less than 24 hours and each time interval not greater than 1 hour.
[0033] The above-collected data sets (cases) number no less than 100. Considering the limited early monitoring data and the difficulty in tracking the drift of the green tide of Ulva prolifera, it is easier to track the drift of the green tide of Ulva prolifera in other time periods, including no less than 30 outbreak and extinction phases.
[0034] S2. Construct a numerical model for the short-term drift of the green tide of Ulva prolifera.
[0035] Based on the Lagrange particle tracking method, without considering the growth and extinction process of Ulva prolifera, but taking into account the dragging effect of wind and current on Ulva prolifera green tide patches, a numerical model of short-term drift of Ulva prolifera green tide is constructed.
[0036] ;
[0037] In the formula, Let i be the position of the i-th green algae patch at time t; Surface flow velocity, The wind speed at sea surface is 10 m. The flow action coefficient; The wind effect coefficient represents the direct drag effect of wind on the green tide patches of Ulva prolifera; This indicates the effect of wind on altering the direction of movement of the green tide patches of *Ulva prolifera*; the x-axis direction is... The y-axis direction is ,in The angle between the wind and the x-axis coordinate axis, in degrees. The deflection angle due to wind drag is expressed in degrees.
[0038] S3. Obtain surface current and sea surface wind data for the study area.
[0039] Obtain hydrological and meteorological data on surface current velocity and sea surface wind in the research sea area where the data set (case) is located; the main sources are observation data or numerical simulation data, where the time interval of numerical simulation data is generally 1 hour, and the time length is longer than that of observation.
[0040] S4. Obtain the optimal wind effect coefficient for each case (each data group).
[0041] S41. Based on the initial positions of satellite remote sensing feature points using satellite remote sensing data, and using the numerical model of Ulva prolifera green tide drift constructed in S2 and the surface current and sea surface wind data (environmental field) of the study area obtained in S3, a numerical sensitivity experiment on the wind effect coefficient is conducted. In the numerical model of short-term drift of Ulva prolifera green tide, the current effect coefficient R1 is set to 1, and the wind effect coefficient R2 ranges from 0.001 to 0.030. The distance error between the predicted position and the measured position under each experimental scheme is calculated (i.e., the distance error is calculated by continuously changing the value of the wind effect coefficient R2). The optimal wind effect coefficient is determined for each particle or patch of Ulva prolifera green tide during the tracking process, using the minimization of distance error as the evaluation criterion.
[0042] S42. Based on the drift tracking observation data at sea, the drift time series of the *Ulva prolifera* green tide was obtained. Using numerical simulation data of sea surface wind and flow fields, as well as the drift position sequence of the *Ulva prolifera* green tide, the optimal wind influence coefficient was determined using velocity fitting and trajectory similarity optimization methods. The specific methods are as follows:
[0043] S421. Velocity fitting method: Preliminary parameter optimization based on minimizing velocity error;
[0044] This method aims to obtain an initial estimate of the wind action coefficient R² by minimizing the simulation error of the drift velocity. 2,init The specific steps are as follows:
[0045] (a) Construct the objective function for velocity error;
[0046] The objective function for velocity error is the root mean square error of the residual between the simulated velocity and the observed velocity, as shown in the following equation:
[0047] ;
[0048] where: t = 1, 2, … n, representing different moments; is the velocity error objective function; is the simulated velocity; is the observed velocity.
[0049] (b) Gradient descent iteration;
[0050] The velocity error objective function is optimized using the gradient descent method. The specific steps include:
[0051] Initialization: Within the range of 0 < R2 < 0.03, set the initial guess value of the wind action coefficient R2(0) = 0.001.
[0052] Gradient calculation: In the k-th iteration, calculate the gradient (first derivative) of the velocity error objective function with respect to the parameter (R2);
[0053] ;
[0054] where: represents the gradient of the velocity error objective function with respect to the parameter; the gradient direction indicates the rate of change and trend of the objective function as R2 increases.
[0055] (c) Parameter update: Adjust the parameter along the opposite direction of the gradient. The update formula is:
[0056] ;
[0057] where: H is the learning rate (step size), and a value comparable to the magnitude of the parameter itself can be set, such as 0.01. If the gradient of the velocity error objective function with respect to the parameter is positive, then reduce R2 to reduce the error.
[0058] (d) Convergence determination: When (gradient threshold, generally set to 10 -4 ), or (parameter change threshold, generally set to 10 -5 ), or when the maximum number of iterations is reached, the iteration terminates. At this time, the obtained R 2,init is the local optimal solution in the sense of minimizing the velocity error, which is also the initial estimate value of the wind action coefficient.
[0059] S422. Fine parameter optimization based on the trajectory similarity criterion;
[0060] After obtaining the initial estimate value R 2,init , further refine the correction of the wind action coefficient based on the overall trajectory matching degree. The specific process is as follows:
[0061] (a) Construct the trajectory similarity objective function;
[0062] A trajectory similarity objective function is constructed, with the root mean square error of spatial position between the simulated trajectory and the observed trajectory as the evaluation index;
[0063] ;
[0064] In the formula: t = 1, 2, ... n, representing different times; To simulate the trajectory; For observation trajectory.
[0065] (b) Gradient descent iteration;
[0066] Initialization: R obtained in the previous step 2,init As the starting point for optimization in this stage, it can effectively improve convergence efficiency and avoid pseudo-local minima.
[0067] Gradient approximation: The commonly used finite difference method is expressed as follows:
[0068] ;
[0069] In the formula, δ represents a small perturbation. Efficient and accurate gradient calculations can also be achieved using adjoint models or automatic differentiation techniques.
[0070] (c) Parameter update: Iterative update along the opposite direction of the gradient:
[0071] ;
[0072] In the formula, β is the learning rate of this stage, which should usually be less than the learning rate (step size) of the first stage, and is set to 0.001.
[0073] (d) Convergence criterion:
[0074] When the trajectory error changes tend to stabilize, such as , To set a threshold (usually set to 10) -4 The iteration terminates when the maximum number of iterations is reached, or when the maximum number of iterations is reached. The final wind action coefficient R is obtained. 2,optimal This is the optimal parameter estimate on the trajectory scale, also known as the optimal wind action coefficient.
[0075] The initial locations of the aforementioned satellite remote sensing feature points, as well as the drift sequence of the Ulva prolifera green tide, all come from the dataset.
[0076] This invention employs a two-stage optimization strategy that combines physical rationality and computational robustness: the first stage determines initial parameter values based on dynamic subprocesses (velocity matching) to ensure that the solution lies in a physically feasible neighborhood; the second stage uses the overall trajectory similarity as the ultimate indicator for fine-tuning, effectively suppressing the cumulative effect of errors, thereby obtaining the optimal wind action coefficient that is consistent with observations in both short-term dynamic response and long-term motion morphology.
[0077] S5. Establish a dynamic model of the wind effect coefficient throughout the entire life cycle of the green tide of Ulva prolifera.
[0078] The variation law of the optimal wind action coefficient determined by S4 with time was analyzed. By performing quadratic polynomial fitting on the optimal wind action coefficients obtained in different time periods, a dynamic model of wind action coefficient applicable to the entire life cycle of Ulva prolifera green tide was established.
[0079] Specifically, the dynamic model of the wind action coefficient is as follows:
[0080] ;
[0081] Where A, B, and C are the values to be fitted; x is the number of days since the sporadic floating of seaweed began. By fitting the data, the specific values of A, B, and C in the dynamic model of wind effect coefficient are determined, and then the wind effect coefficient for different time periods can be calculated.
[0082] The dynamic model of wind action coefficient is combined with the numerical model of short-term drift of Ulva prolifera green tide to predict the drift of Ulva prolifera green tide. That is, the optimal wind action coefficient for different time periods can be obtained in real time based on the dynamic model of wind action coefficient, and then substituted into the numerical model of short-term drift of Ulva prolifera green tide to obtain the prediction result of Ulva prolifera green tide drift.
[0083] In summary, the method of this invention first uses historical data to determine the values to be fitted to the dynamic model of wind action coefficient. During prediction, the starting date of sporadic seaweed drift in the current year is determined, and then the wind action coefficient corresponding to the number of days from that starting date can be predicted. These values are then substituted into the numerical model of short-term drift of seaweed green tides, combined with surface current and sea surface wind data, to obtain the prediction results.
[0084] The invention will be further explained below with reference to specific application examples.
[0085] 1. Collect data on the characteristics of green tides of Ulva prolifera, and satellite remote sensing and marine tracking test data at different development stages (early stage, outbreak stage, and decline stage).
[0086] Data on the distribution characteristics of *Ulva prolifera* green tides in the Yellow Sea were collected and analyzed from 2021 to 2025, including the full-cycle time series of the total annual coverage area and the time of the first sporadic discovery of *Ulva prolifera*. Considering the difficulty in obtaining the coverage area before a large-scale outbreak, the daily growth rate of the *Ulva prolifera* green tide was calculated based on the coverage area time series. The annual *Ulva prolifera* green tide was divided into three stages: early stage, outbreak stage, and decline stage, based on a combination of coverage area and growth rate data. The early stage of the *Ulva prolifera* green tide generally refers to the period from the first sporadic detection of *Ulva prolifera* by ships to the point before a large-scale outbreak (coverage area reaching 5 square kilometers); the outbreak stage refers to the period from a large-scale outbreak to a significant decrease in biomass (growth rate less than -10%), generally from May to early July; the decline stage refers to the period from a significant decrease in biomass to the near disappearance of the *Ulva prolifera* green tide, generally after early July. The average start date of sporadic *Ulva prolifera* floating was calculated based on the time of the first sporadic discovery of *Ulva prolifera*.
[0087] Taking 2024 as an example, on April 15th, ships first discovered sporadic floating green algae in the nearshore waters north of Sheyang; by May 13th, the coverage area reached 5 square kilometers; the growth rate of the green algae tide was less than -10% starting on July 3rd, and the green algae tide basically disappeared by August 14th. Therefore, the early stage was from April 15th to May 12th, the outbreak period was from May 13th to July 2nd, and the decline period was from July 3rd to August 14th. From 2021 to 2025, the earliest time that ships monitored sporadic floating algae was on average April 15th, therefore, the average starting date of sporadic floating algae was April 15th.
[0088] (1) Select high-resolution satellite remote sensing images of different sea areas at different development stages (early stage, outbreak stage and extinction stage) of the green tide of Ulva prolifera under clear weather conditions from 2021 to 2025. Based on the distribution and morphological characteristics of Ulva prolifera, select feature points and construct a data group with more than 2 monitoring data.
[0089] (2) Collect data on the tracking and ship positioning of the green tide of seaweed in different sea areas and at different development stages in the Yellow Sea (such as the Subei Shoal, the sea area near 35 degrees, the sea area north of 35 degrees, Yantai and Weihai, etc.) to obtain long-term data of the drift of the green tide of seaweed. The time length is generally more than 24 hours, and each time interval is no more than 1 hour.
[0090] There are a total of 115 satellite remote sensing and tracking observation data sets: 7 data sets in the early stage, 64 data sets during the outbreak period, and 44 data sets during the decline period.
[0091] The 115 datasets were divided into a training set and a validation set, with 100 datasets in the training set and 15 in the validation set. Both sets included different developmental stages of the green tide of Ulva prolifera.
[0092] 2. Construct a short-term drift prediction model for Ulva prolifera green tides.
[0093] Based on the Lagrange particle tracking method, without considering the growth and extinction process of Ulva prolifera, but taking into account the dragging effect of wind and current on Ulva prolifera green tide patches, a numerical model of short-term drift of Ulva prolifera green tide is constructed.
[0094] ;
[0095] In the formula, Let i be the position of the i-th green algae patch at time t; Surface flow velocity, The wind speed at sea surface is 10 m. The flow action coefficient; The wind effect coefficient represents the direct drag effect of wind on the green tide patches of Ulva prolifera; This indicates the effect of wind on altering the direction of movement of the green tide patches of *Ulva prolifera*; the x-axis direction is... The y-axis direction is ,in The angle between the wind and the x-axis coordinate axis, in degrees. The deflection angle due to wind drag is expressed in degrees.
[0096] 3. Obtain surface current and sea surface wind data for the study area.
[0097] Obtain hydrometeorological data on surface current velocity and sea surface wind in the vicinity of the case study (data set); the primary source is numerical simulation data, where the time interval for numerical simulation data is generally 1 hour, longer than the observation time. In the example, sea surface wind mainly comes from the WRF operational meteorological forecasting system; surface current velocity... The data primarily comes from the ROMS 3D ocean current forecasting system, with a data interval of 1 hour. The data spatial range covers the Yellow Sea, which is larger than the study area.
[0098] 4. Obtain the optimal wind effect coefficient for each case (each data group).
[0099] (1) Conduct numerical sensitivity experiments on wind action coefficients based on satellite remote sensing data. Taking May 13, 2024 as an example, based on five satellite remote sensing feature points in the sea area near the Subei Shoal ( Figure 2 The green dot represents the initial moment of the simulation. Numerical sensitivity experiments were conducted, with the flow action coefficient R1 set to 1, the wind drag deflection angle set to 20°, and the wind action coefficient R2 ranging from 0.001 to 0.030, with intervals of 0.005. The simulated positions under each experimental scheme were calculated. Figure 2 (red dot) and measured position ( Figure 2 The minimum absolute error of the distance between the five feature points (blue dots) was used to obtain the optimal wind action coefficient for each of the five feature points.
[0100] (2)For the data of the sea drift tracking experiment, the time series of the drift positions of the Enteromorpha prolifera green tide was obtained. The flow action coefficient R1 was set to 1, and the wind drag deflection angle was set to 20°. Based on the numerical simulation data of the sea surface wind field and flow field, as well as the drift position sequence of the Enteromorpha prolifera green tide, the velocity fitting method and the trajectory similarity optimization method were used to determine the optimal wind action coefficient. Taking June 13, 2025 as an example, with the initial point of the tracking trajectory as the initial simulation time, based on the numerical simulation data of the sea surface wind field and flow field, using a hypothetical wind action coefficient of 0.01, the short-term drift numerical model of the Enteromorpha prolifera green tide was used to calculate and obtain the simulated drift velocity and drift trajectory ( Figure 3 ).
[0101] Then, an objective function for the drift velocity error was established, and the sum of the squares of the difference between the predicted drift velocity and the true drift velocity was used as the evaluation index; setting 0 < R2 < 0.03, the gradient descent method was used to solve the wind action coefficient that minimized the value of the drift velocity error objective function. The wind action coefficient calculated in the previous step was used as the initial value of the trajectory similarity method to construct a trajectory similarity evaluation function; by calculating the root mean square value of the deviation between the predicted drift trajectory and the observation, the gradient descent method was also used to determine the optimal wind action coefficient value.
[0102] Finally, 100 optimal wind action coefficients at different development stages (early stage, outbreak stage, and extinction stage) were obtained.
[0103] 5. Analyze the distribution characteristics of the wind action coefficient and construct a wind action coefficient model for the entire life cycle of the Enteromorpha prolifera green tide.
[0104] Analyze the distribution characteristics of the wind action coefficient over time (different development stages). At different development stages, the wind action coefficient of the Enteromorpha prolifera green tide shows obvious dynamic change characteristics, and is relatively consistent with the change law of the Enteromorpha prolifera life cycle. In the early stage, when the Enteromorpha prolifera is in a suspended state, the wind action coefficient is 0; after that, when the Enteromorpha prolifera is mainly in a floating state, the wind action coefficient is relatively small, generally below 0.01. After entering the outbreak stage, the wind action coefficient gradually increases, mainly distributed between 0.01 and 0.02, and reaches the peak around June 15. When entering the extinction stage, the wind action coefficient decreases significantly and finally approaches to 0 ( Figure 4 ).
[0105] By performing quadratic polynomial fitting on the 100 optimal wind action coefficients in different time periods (early stage, outbreak stage, and extinction stage), and on this basis, a dynamic model of the wind action coefficient applicable to the entire life cycle of the Enteromorpha prolifera green tide was constructed.
[0106] ;
[0107] where x is the number of days from the starting date of the sporadic floating of the Enteromorpha prolifera.
[0108] 6. Model verification;
[0109] Fifteen validation sets that were not involved in the fitting were used, namely satellite data on the characteristic points or drift tracking of Ulva prolifera green tides at different development stages (early stage, outbreak stage, and decline stage). Combined with environmental field data, the drift trajectory of Ulva prolifera green tides was predicted based on the drift numerical model, and the distance error was calculated to verify the prediction effect of the model after using the parameters.
[0110] like Figure 5 As shown, taking the marine drift tracking experiment conducted using a seaweed tracker in June 2025 as an example, based on the initial position of the seaweed tracker (red dot) at 11:00 on June 12th, the wind action coefficient obtained using the full-cycle wind action coefficient model and the traditional wind action coefficient (generally 0.01) were used to forecast the drift of the seaweed green tide using a numerical model. Two drift trajectories of the seaweed green tide patches were obtained and compared. The results show that the drift distance error of the drift trajectory obtained based on the full-cycle wind action coefficient model is significantly smaller at each moment than that obtained using the traditional wind action coefficient. Comparison with other monitoring data and forecast results shows that the full-cycle wind action coefficient model can better reflect the dynamic periodic characteristics of wind-induced seaweed green tide patch drift, and its forecast distance error is smaller than that obtained using the traditional wind action coefficient.
[0111] In summary, the method of the present invention, by constructing a dynamic wind action coefficient that can determine the entire life cycle of the Ulva prolifera green tide, can help improve the prediction accuracy of the Ulva prolifera green tide drift path.
[0112] For any parts not mentioned above, existing technologies can be adopted or referenced.
[0113] Of course, the above description is only a preferred embodiment of the present invention. The present invention is not limited to the above-described embodiments. It should be noted that any equivalent substitutions or obvious modifications made by those skilled in the art under the guidance of this specification fall within the scope of this specification and should be protected by the present invention.
Claims
1. A method for determining the wind action coefficient of the entire life cycle of *Ulva prolifera* green tide, characterized in that... It includes the following steps: S1. Divide different development stages of the Enteromorpha prolifera green tide, and classify it into the early stage, the outbreak stage and the decline stage; Obtain satellite remote sensing images of different development stages of the Enteromorpha prolifera green tide, and interpret the satellite remote sensing images to obtain the distribution data of the Enteromorpha prolifera green tide; and / or obtain the marine drift tracking observation data of different development stages of the Enteromorpha prolifera green tide; S2. Construct a short-term drift numerical model of the Enteromorpha prolifera green tide; S3. Obtain the surface current and sea surface wind data of the research sea area; S4. Determine the optimal wind action coefficient; Adopt the distribution data of the Enteromorpha prolifera green tide obtained by interpreting the satellite remote sensing images of different development stages of the Enteromorpha prolifera green tide in S1, and / or the marine drift tracking observation data, and the surface current and sea surface wind data of the research sea area obtained in S3, and combine with the constructed short-term drift numerical model of the Enteromorpha prolifera green tide to determine the optimal wind action coefficient of different development stages of the Enteromorpha prolifera green tide; S5. Establish a dynamic model of the wind action coefficient for the entire life cycle of the Enteromorpha prolifera green tide; According to the optimal wind action coefficients of different development stages of the Enteromorpha prolifera green tide determined in S4, perform quadratic polynomial fitting to establish a dynamic model of the wind action coefficient applicable to the entire life cycle of the Enteromorpha prolifera green tide, and determine the wind action coefficient values for the entire life cycle of the Enteromorpha prolifera green tide according to this dynamic model of the wind action coefficient; In S4: For the marine drift tracking observation data, obtain the drift position sequence of the Enteromorpha prolifera green tide; based on the surface current and sea surface wind data of the research sea area and the drift position sequence of the Enteromorpha prolifera green tide, adopt the velocity fitting method and the trajectory similarity optimization method to determine the optimal wind action coefficient; The specific steps of the velocity fitting method are as follows: a. Construct a velocity error objective function; The velocity error objective function is the root mean square error of the residual between the simulated velocity and the observed velocity, as shown in the following formula: ； In the formula: t = 1, 2, ... n, representing different times; Let the velocity error be the objective function; To simulate speed; For observation speed; b. Use the gradient descent method to optimize the velocity error objective function, including: Initialization: Set the initial guess value of the wind action coefficient R2(0)=0.001 within the range of 0<R2<0.03; Gradient calculation: In the kth iteration, calculate the gradient of the velocity error objective function with respect to the parameter; ； In the formula: This represents the gradient of the velocity error objective function with respect to the parameters; the gradient direction indicates the rate of change and trend of the objective function as R² increases. c. Adjust the parameter along the opposite direction of the gradient, and the update formula is: ； In the formula: H is the learning rate; if the gradient of the velocity error objective function with respect to the parameter is positive, then reduce R2 to reduce the error; d. Convergence criterion: When the following conditions are met... , Gradient threshold; or , The parameter variation threshold is used; or the iteration terminates when the maximum number of iterations is reached; at this point, the obtained R... 2,init That is, the local optimum solution in the sense of minimizing velocity error; In obtaining R 2,init Subsequently, the wind effect coefficient was further refined using the trajectory similarity optimization method. The specific steps are as follows: e. Construct a trajectory similarity objective function, and use the root mean square error of the spatial position between the simulated trajectory and the observed trajectory as the evaluation index; ； In the formula: t = 1, 2, ... n, representing different times; To simulate the trajectory; For observation trajectory; f. Perform gradient descent iteration, including: Initialization: R obtained in the previous step 2,init This serves as the starting point for optimization in this phase; The finite difference formula is as follows: ； In the formula: δ is a small perturbation; g. Perform iterative update along the opposite direction of the gradient: ； In the formula, β is the learning rate of this stage; h. Convergence determination: The iteration terminates when the trajectory error stabilizes or the maximum number of iterations is reached; the final wind action coefficient R is obtained. 2,optimal This is the optimal wind action coefficient for the corresponding data set; In S5, the dynamic model of the wind action coefficient is as follows: ； In the formula, A, B, and C are the numerical values to be fitted; x is the number of days from the starting date of the sporadic floating of Enteromorpha prolifera.
2. The method for determining the wind action coefficient of the entire life cycle of *Ulva prolifera* green tide according to claim 1, characterized in that, In S1: Collect the distribution characteristic data of the Enteromorpha prolifera green tide, including the time series of the coverage area every year, calculate the growth rate of the Enteromorpha prolifera green tide based on the time series of the coverage area; divide different development stages of the Enteromorpha prolifera green tide into the early stage, the outbreak stage and the decline stage according to the coverage area and the growth rate of the Enteromorpha prolifera green tide; at the same time, determine the starting date of the sporadic floating of Enteromorpha prolifera based on the collected distribution characteristic data of the Enteromorpha prolifera green tide.
3. The method for determining the wind action coefficient of the entire life cycle of *Ulva prolifera* green tide according to claim 2, characterized in that, In S1: Select high-resolution satellite remote sensing images of different development stages of the green tide of Ulva prolifera under clear weather conditions, and select at least two monitoring data points based on the distribution and morphological characteristics of the green tide of Ulva prolifera to form a data set; Collect data from tracking observations or ship positioning observations of green tides of Ulva prolifera at different development stages to obtain long-term series data of green tide drift of Ulva prolifera, with a time length of not less than 24 hours and each time interval not greater than 1 hour, and form a data set. Set up no fewer than 100 data groups, with no fewer than 30 data groups covering both the outbreak and decline phases.
4. The method for determining the wind action coefficient of the entire life cycle of *Ulva prolifera* green tide according to claim 3, characterized in that, In S2: Based on the Lagrange particle tracking method, considering the dragging effect of wind and current on the green tide patches of Ulva prolifera, a numerical model of short-term drift of Ulva prolifera green tide is constructed. ； In the formula, Let i be the position of the i-th green algae patch at time t; Surface flow velocity, The wind speed at sea surface is 10 m. The flow action coefficient; The wind effect coefficient represents the direct drag effect of wind on the green tide patches of Ulva prolifera; This indicates the effect of wind on altering the direction of movement of the green tide patches of *Ulva prolifera*; the x-axis direction is... The y-axis direction is ,in The angle between the wind and the x-axis coordinate axis, in degrees. The deflection angle due to wind drag is expressed in degrees.
5. The method for determining the wind action coefficient of the entire life cycle of *Ulva prolifera* green tide according to claim 4, characterized in that, In S3: Obtain hydrological and meteorological data on surface current velocity and sea surface wind in the sea area where the data set is located; the source of this hydrological and meteorological data is observation data or numerical simulation data.
6. The method for determining the wind action coefficient of the entire life cycle of *Ulva prolifera* green tide according to claim 5, characterized in that, In S4: For the distribution data of Ulva prolifera green tide obtained from satellite remote sensing image interpretation, based on the constructed numerical model of short-term drift of Ulva prolifera green tide and surface current and sea surface wind data of the study sea area, a numerical sensitivity experiment of wind action coefficient is carried out. The distance error between the predicted position and the actual monitoring position under each experimental scheme is calculated. The optimal wind action coefficient for each data group is determined by minimizing the distance error as the evaluation criterion.
7. The method for determining the wind action coefficient of the entire life cycle of *Ulva prolifera* green tide according to claim 6, characterized in that: The dynamic model of wind action coefficient is combined with the numerical model of short-term drift of Ulva prolifera green tide to predict the drift of Ulva prolifera green tide.