Sea grass biomass prediction method based on synergistic effect of multiple environmental factors

Abstract

The invention discloses a sea grass biomass prediction method based on a synergistic effect of multiple environmental factors, and belongs to the field of sea grass biomass prediction. The method comprises the following steps: a, the leaf growth rate is influenced by temperature, illumination, water nutritive salt concentration and plant density, the root growth rate is influenced by temperature, soil nutritive salt concentration and plant density, and the leaf respiration rate and the root respiration rate are influenced by temperature; the sea grass biomass comprises leaf biomass and root system biomass, and considering biomass transmission from the root system to the leaf, and constructing a sea grass biomass prediction model; and b, acquiring illumination, seawater temperature, water nutritive salt concentration, soil nutritive salt concentration and plant density data of the target area under different time steps, and predicting the seaweed biomass of the target area in combination with the seaweed biomass prediction model constructed in the step a. According to the method, the sea grass biomass prediction model under the synergistic effect of multiple environmental factors is established, and accurate prediction of the change of the sea grass distribution area is realized.

Description

Technical Field

[0001] The invention relates to the field of seaweed biomass prediction, and in particular to a seaweed biomass prediction method based on the synergistic effect of multiple environmental factors. Background Art

[0002] As an indispensable part of the marine ecosystem, seagrass beds play multiple key ecological functions, such as maintaining biodiversity, purifying seawater quality, reducing wave energy, and resisting coastal erosion. However, under the dual pressures of global climate change and increasingly frequent human activities, seagrass beds are facing a severe crisis of degradation. In this context, accurately predicting the distribution changes of seagrass beds is extremely important for formulating scientific and effective protection strategies and early warning of coastal disasters.

[0003] At present, the research on seagrass beds mainly covers the following aspects: (1) Evaluating the potential impact of seagrass restoration on the ecosystem based on the Ecopath with Ecosim model; (2) Generating simulated satellite images based on the radiation transfer model to evaluate the visual detectability of seagrass beds under different seawater transparency and seagrass coverage; (3) Evaluating the environmental impact of marine aquaculture on seagrass bed ecosystems based on the hydrodynamic-water quality model, light attenuation model and seagrass bed dynamics model; (4) Based on the self-organization theory model of seagrass bed ecosystem, this model mainly focuses on the spatial pattern formation mechanism of seagrass ecosystem.

[0004] In summary, there are currently few studies on seagrass growth models. Although some have been done, they are usually limited to the independent effects of a single or a few environmental factors, and fail to fully reveal the nonlinear response relationship of the multi-factor coupling mechanism to seagrass biomass. This limits the accuracy of predictions when faced with the complex and changeable actual marine environment, making it difficult to make reliable predictions about the growth status of seagrass. Summary of the invention

[0005] In view of the above technical problems, the present invention proposes a seaweed biomass prediction method based on the synergistic effect of multiple environmental factors.

[0006] The technical solution adopted by the present invention is: A method for predicting seagrass biomass based on the synergistic effect of multiple environmental factors comprises the following steps: a. Construct a seagrass biomass prediction model; Seagrass biomass includes leaf biomass and root biomass, and considering the biomass transfer from roots to leaves, a seagrass biomass prediction model is constructed as follows: (1) (2) Among them, B l(t) and B l (t+1) are the leaf biomass at time t and time t+1, B r (t) and B r (t+1) are the root biomass at time t and time t+1, G l (t) and G r (t) are the leaf growth rate and root growth rate at time t, R esl (t) and R esr (t) are the leaf respiration rate and root respiration rate at time t, T rl (t) is the biomass transfer coefficient from roots to leaves at time t, and Δt is the time step; The leaf growth rate is affected by temperature, light, water nutrient concentration and plant density, the root growth rate is affected by temperature, soil nutrient concentration and plant density, and both the leaf respiration rate and root respiration rate are affected by temperature. b. To predict seagrass biomass in the target area; Obtain the light, seawater temperature, water nutrient concentration, soil nutrient concentration and plant density data at different time steps in the target area, and use the seagrass biomass prediction model constructed in step a to predict the seagrass biomass in the target area.

[0007] The beneficial technical effects of the present invention are as follows: The present invention optimizes the limitation of traditional models that only focus on single or a few environmental factors by constructing a seagrass biomass prediction model with the synergistic effect of multiple environmental factors. By considering the influence of temperature, light, water nutrient concentration, soil nutrient concentration and plant density on seagrass biomass, a seagrass biomass prediction model with the synergistic effect of multiple environmental factors is established, and accurate prediction of changes in seagrass distribution areas is achieved. In the process of model parameter optimization, the key parameters that have a greater impact on the model results are found through sensitivity analysis, and then these key parameters are optimized in a targeted manner. This modeling method of comprehensive multi-factor analysis and parameter optimization effectively makes up for the prediction bias problem of traditional models caused by one-sided data, greatly improves the accuracy of prediction results, and provides more reliable support for seagrass ecological research and protection.

[0008] The precise prediction results provided by the present invention provide a solid scientific basis for the protection and restoration of seagrass beds. With the help of the prediction data of the model, relevant management departments can gain a deeper understanding of the development trend of seagrass beds under different environmental conditions, and then formulate more targeted and effective management measures. For example, according to the prediction results, the marine protected areas can be rationally planned, and restoration resources can be accurately deployed to avoid the waste of resources caused by blind decision-making, providing a strong guarantee for the sustainable development of the seagrass bed ecosystem. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 This is a schematic diagram of the seagrass biomass change model framework; Figure 2 is the sensitivity of the main parameters of the model; Figure 3 The decrease of the cost function during the assimilation process is shown; Figure 4 This is a comparison chart between the leaf biomass predicted by the model and the measured results; Figure 5 This is a comparison chart between the root biomass predicted by the model and the measured results; Figure 6 The seagrass cover in a certain area in a certain month as observed by shipborne radar; Figure 7 To study the average sea surface temperature and light intensity of the sea area and the simulated average leaf biomass, (a) shows the sea surface temperature, (b) shows the light intensity, and (c) shows the leaf biomass; Figure 8 The figure shows the distribution of leaf biomass in the study area in four seasons, where (a) shows the leaf biomass in spring, (b) shows the leaf biomass in summer, (c) shows the leaf biomass in autumn, and (d) shows the leaf biomass in winter. DETAILED DESCRIPTION

[0010] Traditional modeling methods are usually limited to the independent effects of a single or a few environmental factors, and fail to fully reveal the nonlinear response relationship of the multi-factor coupling mechanism to seagrass biomass. This limits the accuracy of the model's predictions when facing complex and changeable actual marine environments, and makes it difficult to reliably predict the growth of seagrass. In view of this situation, the present invention proposes a seagrass biomass prediction method and application based on the synergistic effects of multiple environmental factors. This method considers the effects of multiple key environmental variables on seagrass biomass, including temperature, light, water nutrient concentration, soil nutrient concentration, and plant density, to establish a seagrass biomass prediction model under the synergistic effects of multiple environmental factors, and further evaluates and optimizes parameter uncertainty, thereby filling the gaps in existing models in terms of comprehensive analysis of multiple factors and accurate prediction, and providing innovative and effective technical means for the study and protection of seagrass ecosystems.

[0011] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0012] A method for predicting seagrass biomass based on the synergistic effect of multiple environmental factors comprises the following steps: a. Construct a seagrass biomass prediction model; Considering the environmental factors affecting seagrass leaf biomass and root biomass, such as temperature, light, water nutrient concentration, soil nutrient concentration and plant density, a seagrass biomass model was constructed. The specific model framework is as follows: Figure 1 shown.

[0013] The leaf growth rate is affected by light, temperature, plant density and water nutrient concentration, while the root growth rate is not affected by light and water nutrient concentration, but is additionally affected by soil nutrient concentration. As for the respiration rate, it is mainly affected by temperature. The growth rate and respiration rate of leaves and roots are expressed as: (1) (2) (3) (4) Among them, G l (t) and G r (t) are the leaf and root growth rates at time t, R esl (t) and R esr (t) are the leaf and root respiration rates at time t, respectively, f I (t), f Tl (t), f Tr (t), f S (t), f Nw (t) and f Ns (t) are the light influence coefficient, leaf temperature influence coefficient, root temperature influence coefficient, plant density influence coefficient, water nutrient concentration influence coefficient and soil nutrient concentration influence coefficient at time t, respectively, expressed as: (5) (6) (7) (8) (9) (10) Where I(t) is the light intensity at time t, which is measured by an on-site light meter; k el is the leaf illumination half-saturation constant; T(t) is the sea temperature at time t, which is calculated by the ocean numerical model; T optl and T optr are the optimum growth temperatures for leaves and roots, respectively; tl and trare the coefficients of dependence of leaf growth and root growth on temperature; n w (t) and n s (t) are the nutrient concentrations of water and soil at time t. The nitrate concentration value is mainly used in the model calculation, which is obtained by measuring and analyzing the seawater and soil samples collected on site; n lmin and n lcrit are the minimum internal nitrogen content of leaves and the critical internal nitrogen content of leaves, respectively; n rmin and n rcrit are the minimum internal nitrogen content of the root system and the critical internal nitrogen content of the root system, respectively; B l (t) is the leaf biomass at time t. First, the initial time, i.e., the leaf biomass at t = 0, is obtained by field collection. l0 Then, the iterative calculation model constructed by formula (1) is used for subsequent calculations to obtain the leaf biomass B at different times. l (t); B lmax is the maximum leaf biomass; k l is the coefficient of leaf growth dependence on space.

[0014] The light intensity I(t,z) at depth z at time t can be calculated from the sea surface light intensity I surface (t) is calculated as: (11) Among them, k z is the light attenuation coefficient.

[0015] The biomass transfer from roots to leaves at time t is represented by the biomass transfer coefficient T rl , expressed as: (12) T rl (t) is the biomass transfer coefficient from roots to leaves at time t, T rlmean is the average biomass transfer coefficient from roots to leaves.

[0016] The model uses dry mass per square meter (g DW m - ²) to represent biomass. The biomass of seagrass consists of leaf biomass and root biomass. The change of biomass is related to growth, respiration and biomass transport. For a given time step Δt, the biomass at time t+1 can be expressed by the biomass at time t: (13) (14) Among them, B l (t) and B l(t+1) are the leaf biomass at time t and time t+1, B r (t) and B r (t+1) are the root biomass at time t and time t+1, G l (t) and G r (t) are the growth rates of leaves and roots at time t, R esl (t) and R esr (t) are the respiration rates of leaves and roots at time t, T rl (t) is the biomass transfer coefficient from roots to leaves at time t.

[0017] In summary, by considering the effects of temperature, light, water nutrient concentration, soil nutrient concentration and plant density on leaf biomass and root biomass, a method for calculating seagrass biomass was given, thus constructing a complete seagrass biomass change model framework.

[0018] b. Optimizing the parameters of the seagrass biomass prediction model constructed in step a; Optimization of model parameters is a key step in improving the accuracy and reliability of the seagrass biomass change model. The present invention uses sensitivity analysis and measured data fitting to collaboratively optimize model parameters.

[0019] b1. Screen key parameters based on sensitivity analysis; In order to study the impact of model parameters on the results, it is necessary to conduct a sensitivity analysis on the main parameters of the model, and explore the extent of its impact on leaf biomass by increasing and decreasing the model parameters by 20%.

[0020] The present invention sets up two sets of sensitivity experiments (SE-1 and SE-2). In these two sets of experiments, relative random errors of +20% and -20% are applied to the main parameters of the model respectively to simulate the errors that may occur in the actual observation process. Each set of experiments is repeated 100 times to ensure the stability and representativeness of the experimental results through multiple repetitions. At the same time, the uncertainty (RU) of the sensitivity experiment results is calculated, and the calculation formula is: (15) Among them, CE is the result of the control experiment, SE is the result of the sensitive experiment, N is the number of simulation days, that is, 365 days, and the calculation of the result uncertainty (RU) can reveal the fluctuation range of the model prediction results under different scenarios, and thus reflect the uncertainty level of the model.

[0021] Figure 2 is the sensitivity of the main parameters of the model determined by the above method in the present invention. Figure 2It can be seen that the model results are more sensitive to parameters related to nutrients, and the sensitivity to root-related parameters is higher than that to leaf-related parameters. This is because the root system plays an important role in the absorption and storage of seaweed nutrients. The root system absorbs nutrients from the pore water of the sediment, and its absorption rate is regulated by relevant parameters. When these parameters change, the nutritional status of the seaweed will change, which will in turn affect the overall growth and biomass of the seaweed. The leaves have a relatively single way of obtaining nutrients, so they are less sensitive to parameter changes. Therefore, the parameter G that has a greater impact on the model results is selected. lmax , G rmax , n lmin , n lcrit , R eslmax , R esrmax ,s tl ,s tr As key parameters, and further optimize and correct the key parameters.

[0022] b2. Optimization of key parameters; The adjoint assimilation model is an efficient parameter optimization method. By combining forward simulation and adjoint simulation, it can quickly find the parameter value that makes the simulation results closest to the observed data. The main steps of the adjoint assimilation model include: forward simulation, adjoint simulation and parameter optimization.

[0023] (1) Forward simulation: Forward simulation is to run the seagrass biomass prediction model according to the current parameter values ​​to generate simulation results. The steps of forward simulation are as follows: According to the key parameters (G lmax , G rmax , n lmin , n lcrit , R eslmax , R esrmax ,s tl ,s tr ) is initialized. The seagrass biomass prediction model is run according to formulas (1)-(12) to generate the leaf biomass B l (t) and root biomass B r (t) time series.

[0024] Establish the difference function between simulation and observation, denoted as cost function MSE, expressed as: (16) Where C (t) = [B l (t), B r (t)] is the model state variable at time t, namely leaf biomass and root biomass, C * (t) = [B l* (t), B r * (t)] is the observed data corresponding to the model state variables at time t, namely the observed data of leaf biomass and root biomass, and N is the number of simulation time steps.

[0025] (2) Adjoint simulation: Adjoint simulation refers to calculating the gradient of the cost function to the model parameters through the adjoint equation, thereby determining the direction of parameter optimization. First, the adjoint equation is constructed based on the kinetic equation of the seagrass biomass prediction model. The adjoint equation describes the cost function to the model state variable (i.e., leaf biomass B l (t) and root biomass B r (t)), and then the sensitivity of the cost function to the model parameters is derived through the chain rule. The specific form of the adjoint equation is as follows: (17) (18) Among them, λ l (t) and λ r (t) are the concomitant variables of leaf and root biomass at time t, respectively.

[0026] Then the adjoint equation is solved. The solution of the adjoint equation starts from the final time point t=N, denoted as t N , and integrate backward to the initial time point t=0, recorded as t0. First, we need to determine the dependent variable at the final time point t N Initial value of: (19) (20) The time interval [t0, t N ] is discretized into N time points, with a time step of Δt, and any time step t=i in the middle is denoted as t i , from t N Start by reversing the calculation of the adjoint variable value at each time point. i , use the Euler method to solve the adjoint equation, then the update formula of the adjoint variable is: (twenty one) (twenty two)

[0027] During the reverse integration process, the value of the adjoint variable λ is stored at each time point l (t) and λ r (t).

[0028] (3) Parameter optimization After solving the inverse of the adjoint equation, the gradient of the cost function with respect to the model parameters is calculated using the adjoint variable. lmax , G rmax , n lmin , n lcrit , R eslmax , R esrmax ,s tl ,s tr ), the gradient calculation formula is: (twenty three) According to the calculated parameter gradient, the gradient descent method is used to update the model parameters to minimize the cost function MSE. The parameter optimization formula is: (twenty four) Among them, θ(i+1) and θ(i) are the values ​​of the parameter θ of the i+1th time iteration step and the ith iteration step respectively, and η represents the learning rate used to determine the size of the gradient descent step. The size of the learning rate η will be adjusted in the specific numerical simulation. The present invention optimizes each parameter in turn using the method of adjoint assimilation. The error between the observed data and the simulation results drives the continuous operation of the model. The parameters are continuously optimized in the adjoint assimilation model. While optimizing the parameters, the optimal learning rate η of each parameter is also obtained.

[0029] Figure 3 The distribution of the inverted parameters and the decrease in the cost function are shown. The decrease in the cost function confirms the feasibility and effectiveness of the model in simulating parameters. The cost functions corresponding to the eight parameters screened by sensitivity analysis all decreased by about 50%, indicating that effective parameter inversion has been achieved. The specific optimized parameter values ​​are shown in Table 1.

[0030] Table 1 Optimized values ​​of model parameters

[0031] c. To predict seagrass biomass in the target area; The light, seawater temperature, water nutrient concentration, soil nutrient concentration and plant density data of the target area at different time steps are obtained, and the seagrass biomass prediction model constructed in step a and optimized in step b is used to predict the seagrass biomass in the target area.

[0032] In order to determine whether the model after parameter optimization can truly reflect the growth law of seaweed, the present invention has carried out further verification in the simulation of leaf and root biomass. In the present invention, the initial leaf biomass is set according to the approximate conditions of Calvi Bay in 1993 (data from Gobert et al., 1995). The temperature and light data according to Gobert et al. (1995) are used as input. The actual results of leaf biomass are shown in Figure 4 As shown in the dots, in summer, the maximum accumulation of leaf biomass in the areas with depths of 10 m, 20 m, and 30 m showed obvious differences, which were 700, 550, and 200 g·DW·m -2 At 10 m, 20 m and 30 m, leaf growth usually started in April, peaked in June and began to decline from early July.

[0033] By comparing the biomass data at different depths, it can be found that compared with the depth of 10 meters, the biomass at 20 meters and 30 meters has decreased. Among them, the peak time of leaf biomass at a depth of 30 meters is about 1 month and 1.5 months later than that at a depth of 10 meters, respectively. This phenomenon fully demonstrates that the relative importance of biomass will not only change over time, but also vary with the depth of the community. Further analysis shows that the phenomenon of biomass decreasing with increasing depth is closely related to the low light compensation point. In deep water environments, the ratio of leaf biomass to root biomass may change, and leaf biomass may be more or more advantageous, which also affects the distribution of biomass at different depths to a certain extent.

[0034] The model prediction results are as follows Figure 4 The solid line shows that the comparison between the model prediction results and the field measurement data shows that the size and occurrence time of the leaf biomass peak in the model are highly close to the field measurement results. The dynamic change curves obtained by simulating the leaf and epiphyte biomass have a high degree of match with the actual measurement data at the three depths of 10 meters, 20 meters and 30 meters, which fully verifies that the model has good accuracy and reliability in simulating the changes in seagrass biomass.

[0035] In the process of studying seagrass biomass, there is relatively little measurement of root biomass compared to leaf biomass. Due to the lack of observation data, the simulation depth of root biomass is limited to 10 meters in this paper. Figure 5 ) found that compared with leaf biomass, the seasonal variation of root biomass was very small, and its annual average biomass was about 3000 g·DW·m -2This result shows that in seagrass ecosystems, root biomass is relatively stable in time series, in sharp contrast to the obvious seasonal fluctuations of leaf biomass. This stability is of great significance for a deeper understanding of the structure and function of seagrass ecosystems and for assessing the changing trends of seagrass beds under different environmental conditions.

[0036] The present invention will be further described below in conjunction with specific application examples.

[0037] 1. Determine the research area; In order to verify the practicability of the method of the present invention in complex real-life scenarios, the present invention is based on the survey data of the North Sea Disaster Forecast and Mitigation Center of the Ministry of Natural Resources in a certain area in a certain month ( Figure 6 ), explore the performance of the present invention in actual situations.

[0038] 2. Model input; According to the seagrass bed numerical simulation area, the environmental information of the study area was collected. The light, temperature and water nutrient data input to the model were provided by the Copernicus Marine Environment Monitoring Service (CMEMS). By calling the CMEMS data interface, the sea surface light intensity at different times, seawater temperature at different locations and depths, and water nutrient data information can be obtained. The soil nutrient concentration and plant density information input to the model were obtained through multiple field surveys.

[0039] 3. Changes in seagrass biomass; The data obtained above were input into the seagrass biomass prediction model, and the seagrass bed biomass of that month was used as the baseline data to simulate the seasonal changes in the seagrass bed biomass in the area. The simulation results are as follows: Figure 7 shown.

[0040] Depend on Figure 7 It can be seen that the model has good predictive ability for the seasonal changes in leaf biomass. Specifically, the biomass reaches a peak from July to early August, then gradually decreases and reaches a trough from late January to early February. Figure 8 The overall spatial distribution of seagrass, its growth status and the impact of environmental factors on it are presented. Hydrodynamics plays a direct regulatory role in the growth of seagrass. In mudflats near major tidal channels, due to the influence of shear stress, the probability of seagrass leaves being torn and washed away increases significantly, resulting in a slowdown in the growth of seagrass in this area.

[0041] In summary, one of the key technologies of the present invention is to comprehensively consider multiple environmental factors such as temperature, light, water nutrient concentration, soil nutrient concentration and plant density when constructing a seaweed biomass change model. By deeply analyzing the interactions between the various factors, the growth conditions of seaweed under different environmental combinations are accurately simulated. In the model framework, the leaf growth rate is jointly affected by light, temperature, plant density and water nutrient concentration, and the root growth rate is also related to multiple factors. This multi-factor coupled modeling method breaks through the limitations of traditional models that only consider single or a few factors, and fully reflects the complexity and dynamic changes of the seaweed growth environment.

[0042] Furthermore, the model parameters were optimized by combining sensitivity analysis with measured data fitting. First, through sensitivity analysis, the uncertainty of the results was calculated, and the key parameters that had a greater impact on the model results were screened out. Then, the gradient descent algorithm and adjoint assimilation method were used to construct the cost function based on the difference between simulation and observation, and the screened parameters were optimized, which significantly improved the accuracy and reliability of the model.

[0043] The present invention protects a method for predicting changes in seagrass distribution areas based on a seagrass biomass change model, including ideas for constructing a model framework, a comprehensive consideration of multiple environmental factors, a specific method for parameter optimization, and the operation process of the entire prediction method, to prevent others from using, copying, or improving the complete technical solution without authorization.

[0044] In addition, the present invention also protects the application rights of this method in the fields of marine ecology such as scientific research, monitoring and protection, and sustainable development of seagrass bed ecosystems. It ensures that in these fields, without authorization, others cannot use this technology to predict changes in seagrass distribution areas, formulate seagrass bed protection strategies, and other related activities, providing forward-looking and practical key technical support for the field of marine ecology, and ensuring the exclusive application value of the technology of the present invention in related fields.

Claims

1. A method for predicting seagrass biomass based on the synergistic effect of multiple environmental factors, characterized in that The following steps are involved: a. Construct a seagrass biomass prediction model; Seagrass biomass includes leaf biomass and root biomass, and considering the biomass transfer from roots to leaves, a seagrass biomass prediction model is constructed as follows: （1） （2） Among them, B l (t) and B l (t+1) are the leaf biomass at time t and time t+1, B r (t) and B r (t+1) are the root biomass at time t and time t+1, G l (t) and G r (t) are the leaf growth rate and root growth rate at time t, R esl (t) and R esr (t) are the leaf respiration rate and root respiration rate at time t, T rl (t) is the biomass transfer coefficient from roots to leaves at time t, and Δt is the time step; The leaf growth rate is affected by temperature, light, water nutrient concentration and plant density, the root growth rate is affected by temperature, soil nutrient concentration and plant density, and both the leaf respiration rate and root respiration rate are affected by temperature. b. To predict seagrass biomass in the target area; Obtain the light, seawater temperature, water nutrient concentration, soil nutrient concentration and plant density data at different time steps in the target area, and use the seagrass biomass prediction model constructed in step a to predict the seagrass biomass in the target area.

2. The method for predicting seaweed biomass based on the synergistic effect of multiple environmental factors according to claim 1, characterized in that: The leaf growth rate, root growth rate, leaf respiration rate and root respiration rate were calculated using the following formulas: （3） （4） （5） （6） Among them, f I (t), f Tl (t), f Tr (t), f S (t), f Nw (t) and f Ns (t) are the light influence coefficient, leaf temperature influence coefficient, root temperature influence coefficient, plant density influence coefficient, water nutrient concentration influence coefficient and soil nutrient concentration influence coefficient at time t; G lmax is the maximum growth rate of leaves; G rmax is the maximum root growth rate; R eslmax is the maximum leaf respiration rate coefficient; R esrmax is the maximum root respiration rate coefficient.

3. The method for predicting seaweed biomass based on the synergistic effect of multiple environmental factors according to claim 2, characterized in that: Light influence coefficient, leaf temperature influence coefficient, root temperature influence coefficient, plant density influence coefficient, water nutrient concentration influence coefficient and soil nutrient concentration influence coefficient: （7） （8） （9） （10） （11） （12） Where I(t) is the light intensity at time t; k el is the leaf illumination half-saturation constant; T(t) is the sea temperature at time t; T optl and T optr are the optimum growth temperatures for leaves and roots, respectively; tl and tr are the coefficients of dependence of leaf growth and root growth on temperature; n w (t) and n s (t) are the nutrient concentrations of water and soil at time t; n lmin and n lcrit are the minimum internal nitrogen content of leaves and the critical internal nitrogen content of leaves, respectively; n rmin and n rcrit are the minimum internal nitrogen content of the root system and the critical internal nitrogen content of the root system, respectively; B l (t) is the leaf biomass at time t; B lmax is the maximum leaf biomass; k l is the coefficient of leaf growth dependence on space.

4. The method for predicting seaweed biomass based on the synergistic effect of multiple environmental factors according to claim 1, characterized in that: The biomass transfer from roots to leaves at time t is represented by the biomass transfer coefficient, which is calculated as follows: （13） Among them, T rl (t) is the biomass transfer coefficient from roots to leaves at time t, T rlmean is the average biomass transfer coefficient from roots to leaves.

5. The method for predicting seaweed biomass based on the synergistic effect of multiple environmental factors according to claim 1, characterized in that: It also includes a model parameter optimization step, using sensitivity analysis and measured data fitting to synergistically optimize seagrass biomass prediction model parameters.

6. The method for predicting seaweed biomass based on the synergistic effect of multiple environmental factors according to claim 5, characterized in that: The model parameter optimization specifically includes the following steps: a11. Screen key parameters based on sensitivity analysis; A sensitivity experiment was set up to impose relative random errors on the parameters of the seagrass biomass prediction model to simulate the errors that may occur in the actual observation process and calculate the uncertainty of the experimental results. The calculation formula is as follows: （14） Among them, CE is the result of the control experiment, SE is the result of the sensitive experiment, N is the number of simulation days, and RU is the result uncertainty; the key parameters that affect the seagrass biomass prediction model are screened out through the RU value; a12. Use measured data fitting method to optimize key parameters; First, the difference function between simulation and observation is established, denoted as the cost function MSE, which is expressed as: （15） Where C (t) = [B l (t), B r (t)] is the model state variable at time t, C * (t) = [B l * (t), B r * (t)] is the observed data corresponding to the model state variables at time t, and N is the number of simulation time steps; Secondly, the adjoint assimilation method is used to carry out forward simulation and adjoint simulation. The gradient of the cost function with respect to the model parameters is obtained by solving the adjoint equation, thereby determining the direction of parameter updating, and the gradient descent method is used to iteratively adjust the parameters to minimize the cost function.

7. The method for predicting seaweed biomass based on the synergistic effect of multiple environmental factors according to claim 6, characterized in that: To achieve gradient descent, you need to calculate the gradient of the cost function of the model with respect to the parameter θ, so as to obtain the gradient vector corresponding to the parameter , and subtract the parameter θ from To achieve parameter optimization, it is expressed as: （16） Among them, θ(i+1) and θ(i) are the values ​​of the parameter θ for the i+1th iteration and the ith iteration respectively. η represents the learning rate, which is used to determine the size of the gradient descent step. The size of the learning rate η will be adjusted in the specific numerical simulation.

8. The method for predicting seaweed biomass based on the synergistic effect of multiple environmental factors according to claim 7, characterized in that: The adjoint assimilation method is used to optimize each parameter in turn. The error between the observed data and the simulation results drives the continuous operation of the model. The parameters are continuously optimized in the adjoint assimilation model. While optimizing the parameters, the optimal learning rate η of each parameter is also obtained.

9. The method for predicting seaweed biomass based on the synergistic effect of multiple environmental factors according to claim 1, characterized in that: Collect environmental information of the target area, use the data provided by the Copernicus Marine Environment Monitoring Service, and obtain the sea surface light intensity at different times, sea water temperature and water nutrient data information at different locations and depths by calling the data interface of the Copernicus Marine Environment Monitoring Service. Obtain soil nutrient and plant density information through multiple field survey data.

Images