A method and device for dynamic simulation of water ecosystem at all scales in a watershed

Abstract

The present invention provides a method and device for dynamic simulation of a water ecosystem at all scales in a watershed, which is applied to the field of environmental monitoring technology, including: obtaining remote sensing images of a water ecological environment to be simulated, constructing a remote sensing big data inversion model, and obtaining the output of the remote sensing big data inversion model; resampling the output of the remote sensing big data inversion model using an improved Kalman filter, verifying and calibrating the output of the FVCOM hydrodynamic model, and dynamically and continuously simulating the water ecological environment to be simulated according to a preset frequency. This application aims at the problem that the simulation results of existing water quality models have large uncertainty, and provides a method that can establish a remote sensing big data inversion model by combining GA and ANN, which can be applied to long-term, multi-scale, and large-scale dynamic monitoring of water ecological quality.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental monitoring, and in particular to a method and device for dynamic simulation of a full-scale water ecosystem in a watershed. Background Art

[0002] Surface water is an important natural resource on which human beings depend for survival. The quality of water depends on the influence of multiple natural and human factors. Water quality will be affected by regional natural processes and human activities at any time.

[0003] The water quality of a river basin is one of the important objects of water environment supervision. It needs to be continuously monitored in time and space in order to promptly detect water quality problems in the river basin and then take rapid and scientific disposal measures. The water quality model is one of the basic methods to solve water quality problems. It is a method of estimating and predicting water pollution using mathematical simulation technology. However, the water quality model is affected by limiting factors such as the lack of experience of users, lack of calibration of environmental parameters, and errors in basic data, and the simulation results have great uncertainty. Its uncertainty is mainly manifested in the input errors of multi-source heterogeneous data, parameter uncertainty, and structural errors caused by the inability of specific model structures to reproduce physical mechanisms.

[0004] In recent years, with the implementation of the Overall Plan for the Construction of Ecological and Environmental Big Data, my country's ecological and environmental big data industry has flourished. An ecological monitoring system covering nine land types, including farmland, grassland, forest, bay, desert, karst, lake, swamp and city, involving more than 600 observation stations and more than 280 observation indicators has been built; an environmental quality monitoring network covering elements such as air, water and soil, involving more than 93,000 monitoring stations has been built. A three-dimensional "sky-ground" monitoring system integrating ecosystem, environmental pollution, meteorology, hydrology, frost and resource and environmental remote sensing has gradually been formed.

[0005] In the past half century, with the vigorous development of remote sensing technology, many researchers at home and abroad have developed water ecological index inversion methods based on remote sensing big data. These methods can be roughly divided into two categories: physics-driven methods and data-driven methods. In fact, remote sensing monitoring is mostly carried out for specific applications and scientific problems, which often requires coupling the above-mentioned multiple methods for cross-research.

[0006] The physical driven method mainly simulates the radiation transmission process of water and air in different optical scenes according to the composition of water bodies, and then infers its imaging characteristics on remote sensing images; this type of method mainly includes radiation transmission models and numerical simulations. Because it takes into account physical conservation, momentum conservation, and the inherent absorption and scattering characteristics of water body components, this type of method has clear physical meaning, rigorous causal relationships, and strong portability, and has been widely used in remote sensing estimation of water ecological indicators.

[0007] Remote sensing estimation models of water quality parameters combined with numerical simulation are another common physical driving method. These methods use high-precision remote sensing inversion results as data-driven fields, based on numerical simulation models, and activate long-term and large-scale physical simulations on a three-dimensional scale of water ecology through the calibration of parameters such as turbulence, convection, and turbulence. Quantitative remote sensing inversion combined with numerical simulation has significant advantages in three-dimensional estimation of water quality and overcoming the lack of image availability caused by cloud and rain cover, providing theoretical support for long-term, multi-scale, and large-scale dynamic monitoring of water ecological quality.

[0008] However, these methods have certain problems. First, the calculations of this type of method are complex and the model running time is long. In the actual operation and parameter calibration process, it is often necessary to calculate complex partial differential equations. Secondly, the final simulation effect of the model is greatly affected by the remote sensing inversion results, and the accuracy of remote sensing inversion is very high. In addition, this method requires a large amount of hydrodynamic data and high-precision underwater topography data, and is less applicable to river sections with insufficient data or dam interference. In addition, the spatial scale effect between the remote sensing inversion results and the numerical simulation is significant; when the two methods do not match well, the simulation results may be inaccurate and difficult to interpret. Summary of the invention

[0009] The present application provides a method for dynamic simulation of water ecosystems at all scales in a watershed. The simulation results of existing water quality models have a large uncertainty problem. The present disclosure can not only be applied to long-term, multi-scale, and large-scale dynamic monitoring of water ecological quality, but also solve the problems of input errors of multi-source heterogeneous data, parameter uncertainty, and structural errors of specific models. Accurately and timely discover water quality problems in the watershed. In order to solve the above invention purpose, the technical solution is as follows:

[0010] On the one hand, the embodiment of the present application provides a method for dynamic simulation of a watershed full-scale water ecosystem, comprising the following steps:

[0011] S1: Acquire the remote sensing image of the water ecological environment to be simulated, preprocess the remote sensing image of the selected area, and obtain the preprocessed remote sensing image;

[0012] S2: construct a remote sensing big data inversion model, input the preprocessed remote sensing image into the remote sensing big data inversion model, and obtain the output of the remote sensing big data inversion model;

[0013] S3: resampling the output of the remote sensing big data inversion model by using an improved Kalman filter to obtain the resampled output of the remote sensing big data inversion model, wherein the improved Kalman filter is a combination of multi-particle resampling and Kalman filtering;

[0014] S4: Obtain the initial measurement values ​​of boundary conditions, hydrological and meteorological data and water quality spatial distribution characteristic indicators corresponding to the water ecological environment to be simulated, and input them into the FVCOM hydrodynamic model to obtain the output of the FVCOM hydrodynamic model;

[0015] S5: Verify and calibrate the output of the FVCOM hydrodynamic model using the output of the resampled remote sensing big data inversion model to obtain a calibrated FVCOM hydrodynamic model;

[0016] S6: Input the initial measurement values ​​of the boundary conditions, hydrological and meteorological data, and water quality spatial distribution characteristic indicators corresponding to the area to be simulated into the calibrated FVCOM hydrodynamic model, and perform dynamic and continuous simulation of the simulated water ecological environment at a preset frequency.

[0017] Preferably, the step of acquiring a remote sensing image of the water ecological environment to be simulated in S1, preprocessing the remote sensing image of the selected area to obtain the preprocessed remote sensing image, includes:

[0018] S11: Acquire remote sensing images of the water ecological environment to be simulated;

[0019] S12: performing geometric correction on the remote sensing image to obtain a remote sensing image after geometric correction;

[0020] S13: performing radiation calibration and atmospheric correction on the remote sensing image after geometric correction to obtain a remote sensing image after radiation calibration;

[0021] S14: Perform a long-term data consistency check on the remote sensing image after radiometric calibration to obtain a preprocessed remote sensing image.

[0022] Preferably, the step S2 of constructing a remote sensing big data inversion model, inputting the preprocessed remote sensing image into the remote sensing big data inversion model, and obtaining the output of the remote sensing big data inversion model includes:

[0023] S21: Initializing the GA-ANN model, wherein the initializing the GA-ANN model includes: initializing GA parameters and initializing network weights and biases of the ANN model;

[0024] S22: collecting spectrum and water quality parameter data of the water ecological environment to be simulated, and performing regression analysis on the spectrum and water quality parameter data of the water ecological environment to be simulated based on the ANN model;

[0025] S23: optimizing the ANN model by calculating the reverse path based on GA to obtain global optimal parameters of network weights and biases of the ANN model, wherein the GA includes: a chromosome, an initialized population, a selection operation, a crossover operation, and a mutation operation;

[0026] S24: The GA-ANN model is updated based on the global optimal parameters of the network weights and biases of the ANN model, and the updated GA-ANN model is set as the remote sensing big data inversion model.

[0027] Preferably, the GA in S23 optimizes the ANN model by calculating the reverse path to obtain the global optimal parameters of the network weight and bias of the ANN model, and the GA includes: a chromosome, an initialized population, a selection operation, a crossover operation and a mutation operation, including:

[0028] S231: Create an initialized population containing a series of chromosomes;

[0029] S232: Calculate the fitness based on formula (1) to obtain the fitness score of each chromosome:

[0030]

[0031] Where N is the number of sampling points; y i ′ is the prediction of the sampling point, y i is the measured value at the sampling point;

[0032] S233: performing a selection operation on the population based on a selection algorithm, and selecting a chromosome with a fitness score higher than a threshold as a parent chromosome;

[0033] S234: performing a crossover operation on the parent chromosome to generate a child chromosome, wherein the crossover operation includes a single-point crossover, a multi-point crossover or a uniform crossover;

[0034] S235: performing a random mutation operation on the offspring chromosome, wherein the mutation operation is to randomly change the gene value in the chromosome;

[0035] S236: Repeat S232-S235 until a predetermined stopping condition is reached to obtain the global optimal parameters of the network weights and biases of the ANN model, wherein the predetermined stopping condition is the maximum number of iterations or the optimization result of the optimization function reaching a preset threshold.

[0036] Preferably, the S3 uses an improved Kalman filter to resample the output of the remote sensing big data inversion model to obtain the resampled output of the remote sensing big data inversion model, and the improved Kalman filter is a combination of multi-particle resampling and Kalman filtering, including:

[0037] S31: Initializing the particle space based on the output of the remote sensing big data inversion model, defining multiple scales, and setting the particle weight of each scale, resampling the particle space based on the particle weight to obtain a multi-scale particle space, wherein the particle weight is an entropy value measurement of the corresponding scale;

[0038] S32: Take each particle in the multi-scale particle space as the predicted state, and perform Kalman filtering based on formula (2) to obtain the output of the resampled remote sensing big data inversion model:

[0039]

[0040] Among them, x r is the value after filtering, x t-1 is the value of step t-1, U t-1 is the system noise.

[0041] Preferably, the step S4 obtains the initial measurement values ​​of boundary conditions, hydrological and meteorological data and water quality spatial distribution characteristic indicators corresponding to the water ecological environment to be simulated, and inputs them into the FVCOM hydrodynamic model to obtain the output of the FVCOM hydrodynamic model, including:

[0042] S41: Obtain boundary conditions, hydrological and meteorological data, and water quality spatial distribution characteristic indicators corresponding to the area to be simulated, wherein the hydrological and meteorological data include temperature, runoff, flow velocity, and wind speed, and the water quality spatial distribution characteristic indicators include chlorophyll, total nitrogen, total phosphorus, and ammonia nitrogen;

[0043] S42: generating grid data of a hydrodynamic model based on preset parameters, and setting initial condition data;

[0044] S43: Input the boundary conditions, hydrological and meteorological data, water quality spatial distribution characteristic indicators, grid data, and initial condition data corresponding to the area to be simulated into the FVCOM model to obtain the output of the FVCOM model.

[0045] Preferably, the S5 verifies and calibrates the output of the FVCOM hydrodynamic model using the output of the resampled remote sensing big data inversion model to obtain the calibrated FVCOM hydrodynamic model, including:

[0046] S51: Set the simulation time range and time step in the FVCOM model;

[0047] S52: Run the FVCOM model, compare the output of the resampled remote sensing big data inversion model with the output of the FVCOM model through the objective function, and use the Kalman filter to adjust the model parameters until the objective function reaches the optimal value. The objective function is used to calculate the degree of fit between the output of the resampled remote sensing big data inversion model and the observed data.

[0048] In a second aspect, the present application provides a device for simulating the dynamics of a water ecosystem at a full scale in a watershed, including the following steps:

[0049] Remote sensing image unit: used to obtain remote sensing images of the water ecological environment to be simulated, pre-process the remote sensing images of the selected area, and obtain pre-processed remote sensing images;

[0050] Inversion model unit: used to construct a remote sensing big data inversion model, input the preprocessed remote sensing image into the remote sensing big data inversion model, and obtain the output of the remote sensing big data inversion model;

[0051] Resampling unit: used to resample the output of the remote sensing big data inversion model by using an improved Kalman filter to obtain the resampled output of the remote sensing big data inversion model, wherein the improved Kalman filter is a combination of multi-particle resampling and Kalman filter;

[0052] Hydrodynamic model unit: used to obtain the initial measurement values ​​of boundary conditions, hydrological and meteorological data and water quality spatial distribution characteristic indicators corresponding to the water ecological environment to be simulated, and input them into the FVCOM hydrodynamic model to obtain the output of the FVCOM hydrodynamic model;

[0053] Calibration unit: used to verify and calibrate the output of the FVCOM hydrodynamic model using the output of the resampled remote sensing big data inversion model to obtain a calibrated FVCOM hydrodynamic model;

[0054] Continuous simulation unit: It is used to input the initial measurement values ​​of the boundary conditions, hydrological and meteorological data, and water quality spatial distribution characteristic indicators corresponding to the area to be simulated into the calibrated FVCOM hydrodynamic model, and perform dynamic and continuous simulation of the simulated water ecological environment at a preset frequency.

[0055] In a third aspect, an embodiment of the present application provides an electronic device, characterized in that the electronic device comprises: a shell, a processor, a memory, a circuit board and a power supply circuit, wherein the circuit board is placed inside the space enclosed by the shell, and the processor and the memory are arranged on the circuit board; the power supply circuit is used to supply power to various circuits or devices of the above-mentioned electronic device; the memory is used to store executable program code; the processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute any of the methods described in the foregoing.

[0056] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement any of the methods described above.

[0057] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0058] Compared with the prior art, the above technical solution has at least the following beneficial effects: The above solution, a method for dynamic simulation of a water ecosystem at all scales in a watershed, combines GA and ANN to establish a remote sensing big data inversion model, and then accurately simulates the water quality based on the FVCOM hydrodynamic model. This solution has the following advantages. First, the calculation is not complicated and the model running time is short. There is no need to calculate complex partial differential equations. Secondly, the final simulation effect of the model does not require high accuracy of remote sensing inversion. The spatial scale effect between remote sensing inversion results and numerical simulations can be solved. In addition, a large amount of hydrodynamic data and high-precision underwater topography data are not required, and it is more applicable to river sections with insufficient data or dam interference. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0060] Figure 1 It is a flow chart of a method for dynamic simulation of a watershed full-scale water ecosystem provided by an embodiment of the present invention;

[0061] Figure 2 is a scatter plot of measured and predicted values ​​of the remote sensing big data inversion model provided by an embodiment of the present invention;

[0062] Figure 3 It is a block diagram of a dynamic simulation device for a watershed full-scale water ecosystem provided by an embodiment of the present invention;

[0063] Figure 4 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0064] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0065] The embodiment of the present invention provides a method for dynamic simulation of a water ecosystem at a full scale in a watershed. The method can be implemented by an electronic device, which can be a terminal or a server. Figure 1 The flowchart of a method for dynamic simulation of a watershed full-scale water ecosystem is shown in FIG. The processing flow of the method may include the following steps:

[0066] S1: Acquire the remote sensing image of the water ecological environment to be simulated, preprocess the remote sensing image of the selected area, and obtain the preprocessed remote sensing image;

[0067] Preferably, the S1 comprises:

[0068] S11: Acquire remote sensing images of the water ecological environment to be simulated;

[0069] S12: performing geometric correction on the remote sensing image to obtain a remote sensing image after geometric correction;

[0070] S13: performing radiation calibration and atmospheric correction on the remote sensing image after geometric correction to obtain a remote sensing image after radiation calibration;

[0071] S14: Perform a long-term data consistency check on the remote sensing image after radiometric calibration to obtain a preprocessed remote sensing image.

[0072] In some embodiments, obtaining remote sensing images of the water ecological environment to be simulated may include collecting measured spectra synchronously / quasi-synchronously with the transit time of Sentinel-2MSI and Landsat TM / ETM+ / OLI. The measured spectra on the ground can be collected on-site using the American ASD Field Spec3 single-channel portable spectrometer. In actual measurement, according to the absorption, reflection and scattering characteristics of water and sky light, the following matters should be noted: (1) Considering that the field of view of the spectrometer is 25°, when measuring the radiation brightness of the gray plate, the distance from the gray plate should be between 20 and 35 cm, and at the same time, it should be ensured that there is no obvious probe shadow in the gray plate; when measuring the water body and sky light, the side of the ship should be used to avoid the shadow, and a reflective panel should be used instead of direct measurement of the incident radiation. The height of the probe and the water body should be kept at an appropriate distance, generally between 0.5 and 2 meters. (2) In order to reduce the uncertainty caused by the disturbance of the water body by the hull in the cruising experiment, the spectral measurement procedures such as gray plate, water body and sky light should be carried out in sequence after the ship is anchored and the water surface is stable. (3) In order to reduce the spectral measurement errors caused by solar flares, direct sunlight, the surrounding environment and the shadow of the ship, the azimuth angle between the spectrometer probe and the solar incident plane on the horizontal projection plane should be kept at about 135°, and the observation zenith angle between the spectrometer probe and the water surface normal should be ensured to be between 35° and 45°. (4) When measuring the two elements of water body and sky light, the probe should be in the same solar incident plane, and the probe direction angle should be about 90° during the two measurements. (5) ASD calculates the reflectivity of the ground object based on the integral of the radiation reflected by the ground object (the measurement time interval is 0.1s, and the integration time of the instrument is defined as 172ms). Three important parameters are measured at each sampling point: water surface radiation, sky light and reference plate (reflectivity is 30%). Choose clear and cloudless weather with low wind speed (usually less than level 3) to carry out the water body spectrum measurement experiment; the dress and operation of the experimenters are strictly in accordance with the spectral measurement specifications of ASD. Spectral measurement is synchronized with in-situ water sample collection. The ASD FieldSpec3 spectrometer operates in the band of 350 to 2500nm, and the field of view of the fiber optic probe sensor is 25°. Its high spectral resolution is very suitable for the analysis of physical and chemical properties, and has become one of the important means of analyzing the optical characteristics of water components. However, the spectral resolution of different bands is different: the spectral resolution of the band less than 1000nm is 3nm, and the sampling interval is 1.4nm; the spectral resolution of the band greater than 1000nm is 10nm, and the sampling interval is 2nm. According to the signal composition received by the sensor, the measurement of the true off-water reflectance of the water body requires the collection of standard gray plates at each sampling point in turn.

[0073] It should be noted that the consistency test of long-term time series data is a statistical method used to evaluate the stability and consistency of time series data. Its purpose is to determine whether the data maintains a consistent trend and nature within a given time period.

[0074] It should be further explained that long time series data usually involves data observations at multiple time points or time periods. When performing consistency checks, it is necessary to determine whether the data has similar statistical properties, such as mean, variance, correlation, etc. Specifically, it includes: stationarity test, autocorrelation test and identical distribution test. If the data is not consistent, appropriate preprocessing methods (such as difference or smoothing) are needed to deal with the inconsistency of the data.

[0075] S2: construct a remote sensing big data inversion model, input the preprocessed remote sensing image into the remote sensing big data inversion model, and obtain the output of the remote sensing big data inversion model;

[0076] Preferably, the S2 comprises:

[0077] S21: Initializing the GA-ANN model, wherein the initializing the GA-ANN model includes: initializing GA parameters and initializing network weights and biases of the ANN model;

[0078] S22: collecting spectrum and water quality parameter data of the water ecological environment to be simulated, and performing regression analysis on the spectrum and water quality parameter data of the water ecological environment to be simulated based on the ANN model;

[0079] S23: optimizing the ANN model by calculating the reverse path based on GA to obtain global optimal parameters of network weights and biases of the ANN model, wherein the GA includes: a chromosome, an initialized population, a selection operation, a crossover operation, and a mutation operation;

[0080] S24: The GA-ANN model is updated based on the global optimal parameters of the network weights and biases of the ANN model, and the updated GA-ANN model is set as the remote sensing big data inversion model.

[0081] Preferably, the S23 includes:

[0082] S231: Create an initialized population containing a series of chromosomes;

[0083] S232: Calculate the fitness based on formula (1) to obtain the fitness score of each chromosome:

[0084]

[0085] Where N is the number of sampling points; y i ′ is the prediction of the sampling point, y i is the measured value at the sampling point;

[0086] S233: performing a selection operation on the population based on a selection algorithm, and selecting a chromosome with a fitness score higher than a threshold as a parent chromosome;

[0087] S234: performing a crossover operation on the parent chromosome to generate a child chromosome, wherein the crossover operation includes a single-point crossover, a multi-point crossover or a uniform crossover.

[0088] S235: Perform a random mutation operation on the offspring chromosome, wherein the mutation operation randomly changes the gene value in the chromosome.

[0089] S236: Repeat S232-S235 until a predetermined stopping condition is reached to obtain the global optimal parameters of the network weights and biases of the ANN model, wherein the predetermined stopping condition is the maximum number of iterations or the optimization result of the optimization function reaching a preset threshold.

[0090] In some embodiments, the inversion problem refers to determining the parameters (or model parameters) that characterize the characteristics of the problem based on the results and certain general principles (or models). In engineering applications, inversion problems are widely found in fields such as geophysics, biology, medicine, and architecture. The inversion problem exists relative to the forward problem, and its general working procedure is data-forward model-estimated values ​​of model parameters. Therefore, the forward calculation model is the basis of the inversion problem. Only when the numerical solution to the forward problem is obtained can we consider solving the inversion problem. The remote sensing big data inversion model is a model that uses remote sensing data for reverse analysis or reverse inference to estimate relevant parameters or variables of the surface or atmosphere. It infers the characteristics and properties of the surface or atmosphere by analyzing information such as spectrum, radiation or scattering in remote sensing data.

[0091] It should be noted that the steps of the remote sensing big data inversion model include data acquisition, preprocessing, feature extraction, and inversion calculation. First, it is necessary to obtain high-quality remote sensing data, including remote sensing images, radiation data, or scattering data. Then, the data is preprocessed, such as radiation correction, atmospheric correction, and geometric correction, to ensure the accuracy and consistency of the data. Next, using the extracted features or indicators, an inversion model is established, and appropriate algorithms and methods are selected for inversion calculations. This may involve techniques such as machine learning, statistical analysis, physical models, or simulations. Finally, the inversion results are interpreted and analyzed to evaluate the reliability and confidence of the parameters or variables.

[0092] S3: resampling the output of the remote sensing big data inversion model by using an improved Kalman filter to obtain the resampled output of the remote sensing big data inversion model, wherein the improved Kalman filter is a combination of multi-particle resampling and Kalman filtering;

[0093] Preferably, the S3 comprises:

[0094] S31: Initializing the particle space based on the output of the remote sensing big data inversion model, defining multiple scales, and setting the particle weight of each scale, resampling the particle space based on the particle weight to obtain a multi-scale particle space, wherein the particle weight is an entropy value measurement of the corresponding scale;

[0095] S32: Take each particle in the multi-scale particle space as the predicted state, and perform Kalman filtering based on formula (2) to obtain the output of the resampled remote sensing big data inversion model:

[0096]

[0097] Among them, x r is the value after filtering, x t-1 is the value of step t-1, U t-1 is the system noise.

[0098] In some embodiments, resampling is a statistical method for dealing with uncertainty and variability in data. It generates a new data set by resampling the original data set to obtain more reliable estimates or evaluation results. The main purpose of resampling is to evaluate the stability and reliability of statistics, as well as the performance and prediction accuracy of the estimation model by generating multiple data sets. It can help us model and analyze the data more accurately and reduce the bias and errors caused by the limited size of the data set or the influence of sample selection. In the resampling process, different methods can be used, such as bootstrap, jackknife, and cross-validation. These methods can be selected according to the specific problem and data type.

[0099] It should be noted that the multi-particle resampling process can solve the linear assumption limitation in the Kalman filter and is more suitable for nonlinear and non-Gaussian systems. Multi-particle resampling can better utilize the measurement data and improve the accuracy and robustness of the filter. The Kalman filter part of the particle filter can effectively handle the dynamic changes and measurement noise of the system.

[0100] It should be further explained that the method combining multi-particle resampling and Kalman filtering can flexibly adapt to different problems and data characteristics, and has good adaptability and robustness.

[0101] S4: Obtain the initial measurement values ​​of boundary conditions, hydrological and meteorological data and water quality spatial distribution characteristic indicators corresponding to the water ecological environment to be simulated, and input them into the FVCOM hydrodynamic model to obtain the output of the FVCOM hydrodynamic model;

[0102] Preferably, the S4 comprises:

[0103] S41: Obtain boundary conditions, hydrological and meteorological data, and water quality spatial distribution characteristic indicators corresponding to the area to be simulated, wherein the hydrological and meteorological data include temperature, runoff, flow velocity, and wind speed, and the water quality spatial distribution characteristic indicators include chlorophyll, total nitrogen, total phosphorus, and ammonia nitrogen;

[0104] S42: generating grid data of a hydrodynamic model based on preset parameters, and setting initial condition data;

[0105] S43: Input the boundary conditions, hydrological and meteorological data, water quality spatial distribution characteristic indicators, grid data, and initial condition data corresponding to the area to be simulated into the FVCOM model to obtain the output of the FVCOM model.

[0106] In some embodiments, the FVCOM hydrodynamic model is the abbreviation of Finite-Volume Coastal Ocean Model. It is a numerical model for simulating the hydrodynamic processes of oceans and lakes. The FVCOM model is based on the finite volume method, which divides the domain into discrete volume units and calculates the mass, momentum and energy exchange in these units. It can simulate hydrodynamic processes such as tides, currents, waves, water temperature, salinity and dissolved oxygen distribution in oceans and lakes. The FVCOM hydrodynamic model is a numerical model for simulating the hydrodynamic processes of oceans and lakes.

[0107] It should be noted that the input data of the FVCOM model has certain format requirements: the FVCOM model requires a file that defines the domain grid. This grid file is usually a text file that contains the coordinates and connection information of the grid nodes; the FVCOM model needs to define boundary conditions, including inflow, outflow, and boundary reflection. These boundary conditions usually require data such as water level, flow velocity, and salinity; the FVCOM model requires an initial condition file to define the initial state of the simulation. This file usually contains initial values ​​such as water level, flow velocity, and salinity; the FVCOM model can accept the input of external forcing items, such as wind fields, tides, and river inputs. These external forcing items usually require the provision of corresponding spatiotemporal variation data; the FVCOM model has some internal parameters that need to be set, such as bottom friction coefficient, water depth, and vertical mixing coefficient. These parameters usually need to be adjusted and optimized according to specific problems and research purposes.

[0108] S5: Verify and calibrate the output of the FVCOM hydrodynamic model using the output of the resampled remote sensing big data inversion model to obtain a calibrated FVCOM hydrodynamic model;

[0109] Preferably, the S5 comprises:

[0110] S51: Set the simulation time range and time step in the FVCOM model;

[0111] S52: Run the FVCOM model, compare the output of the resampled remote sensing big data inversion model with the output of the FVCOM model through the objective function, and use the Kalman filter to adjust the model parameters until the objective function reaches the optimum, wherein the objective function is used to calculate the degree of fit between the output of the resampled remote sensing big data inversion model and the observed data. In some embodiments, in the FVCOM hydrodynamic model, parameter calibration refers to a method of making the output of the model more consistent with the actual observation results by adjusting the model parameters according to the observed data.

[0112] It should be noted that the FVCOM model is run with the default model parameters and the observed data is compared with the model output. The comparison requires the definition of an objective function to measure the difference between the model output and the observed data. Common objective functions include root mean square error (RMSE), correlation coefficient, etc.

[0113] S6: Input the initial measurement values ​​of the boundary conditions, hydrological and meteorological data, and water quality spatial distribution characteristic indicators corresponding to the area to be simulated into the calibrated FVCOM hydrodynamic model, and perform dynamic and continuous simulation of the simulated water ecological environment at a preset frequency.

[0114] In some embodiments, the FVCOM model is run using the prepared input data and parameter configurations. During the simulation, the model will calculate the mass, momentum, and energy exchange at each time step to simulate the hydrodynamic process.

[0115] The above is an introduction to the method embodiment. The following is a further explanation of the scheme described in this application through an apparatus embodiment.

[0116] like Figure 3 As shown, the embodiment of the present application provides a basin full-scale water ecosystem dynamic simulation device 300, comprising the following steps:

[0117] Remote sensing image unit 310: used to obtain remote sensing images of the water ecological environment to be simulated, pre-process the remote sensing images of the selected area, and obtain pre-processed remote sensing images;

[0118] Inversion model unit 320: used to construct a remote sensing big data inversion model, input the preprocessed remote sensing image into the remote sensing big data inversion model, and obtain the output of the remote sensing big data inversion model;

[0119] The resampling unit 330 is used to resample the output of the remote sensing big data inversion model by using an improved Kalman filter to obtain the resampled output of the remote sensing big data inversion model, wherein the improved Kalman filter is a combination of multi-particle resampling and Kalman filter;

[0120] The hydrodynamic model unit 340 is used to obtain the initial measurement values ​​of the boundary conditions, hydrological and meteorological data and water quality spatial distribution characteristic indicators corresponding to the water ecological environment to be simulated, and input them into the FVCOM hydrodynamic model to obtain the output of the FVCOM hydrodynamic model;

[0121] Calibration unit 350: used to verify and calibrate the output of the FVCOM hydrodynamic model using the output of the resampled remote sensing big data inversion model to obtain a calibrated FVCOM hydrodynamic model;

[0122] Continuous simulation unit 360: used to input the initial measurement values ​​of boundary conditions, hydrological and meteorological data, and water quality spatial distribution characteristic indicators corresponding to the area to be simulated into the calibrated FVCOM hydrodynamic model, and perform dynamic and continuous simulation of the simulated water ecological environment at a preset frequency.

[0123] An electronic device for dynamic simulation of full-scale water ecosystem in a watershed, characterized in that the electronic device comprises: a shell, a processor, a memory, a circuit board and a power supply circuit, wherein the circuit board is arranged inside the space enclosed by the shell, and the processor and the memory are arranged on the circuit board; the power supply circuit is used to supply power to various circuits or devices of the above-mentioned electronic device; the memory is used to store executable program code; the processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute any of the methods described in the foregoing.

[0124] A computer-readable storage medium for dynamic simulation of a water ecosystem at full scale in a watershed, characterized in that the computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement any of the methods described above.

[0125] This application aims to solve the problem that the simulation results of existing water quality models have great uncertainty. It provides a method that can accurately simulate water quality conditions by combining GA and ANN to establish a remote sensing big data inversion model, and then based on the FVCOM hydrodynamic model, it can be applied to long-term, multi-scale, and large-scale dynamic monitoring of water ecological quality.

[0126] Figure 4 It is a structural diagram of an electronic device 400 provided in an embodiment of the present invention. The electronic device 400 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 401 and one or more memories 402, wherein the memory 402 stores at least one instruction, and the at least one instruction is loaded and executed by the processor 401 to implement the steps of the above-mentioned Chinese text spelling checking method.

[0127] In an exemplary embodiment, a computer-readable storage medium is also provided, such as a memory including instructions, which can be executed by a processor in a terminal to complete the above-mentioned Chinese text spelling check method. For example, the computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0128] A person skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a disk or an optical disk, etc.

[0129] There are a few points to note:

[0130] (1) The drawings of the embodiments of the present invention only relate to the structures related to the embodiments of the present invention, and other structures may refer to the general design.

[0131] (2) For the sake of clarity, in the drawings used to describe the embodiments of the present invention, the thickness of the layers or regions is exaggerated or reduced, that is, these drawings are not drawn according to the actual scale. It is understood that when an element such as a layer, film, region or substrate is referred to as being "on" or "under" another element, the element may be "directly" "on" or "under" the other element or there may be intermediate elements.

[0132] 3) In the absence of conflict, the embodiments of the present invention and the features therein may be combined with each other to obtain new embodiments.

[0133] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for dynamic simulation of a water ecosystem at all scales in a watershed, characterized in that: include: S1: Acquire the remote sensing image of the water ecological environment to be simulated, preprocess the remote sensing image of the selected area, and obtain the preprocessed remote sensing image; S2: construct a remote sensing big data inversion model, input the preprocessed remote sensing image into the remote sensing big data inversion model, and obtain the output of the remote sensing big data inversion model; S3: resampling the output of the remote sensing big data inversion model by using an improved Kalman filter to obtain the resampled output of the remote sensing big data inversion model, wherein the improved Kalman filter is a combination of multi-particle resampling and Kalman filtering; S4: Obtain the initial measurement values ​​of boundary conditions, hydrological and meteorological data and water quality spatial distribution characteristic indicators corresponding to the water ecological environment to be simulated, and input them into the FVCOM hydrodynamic model to obtain the output of the FVCOM hydrodynamic model; S5: Verify and calibrate the output of the FVCOM hydrodynamic model using the output of the resampled remote sensing big data inversion model to obtain a calibrated FVCOM hydrodynamic model; S6: Input the initial measurement values ​​of the boundary conditions, hydrological and meteorological data, and water quality spatial distribution characteristic indicators corresponding to the area to be simulated into the calibrated FVCOM hydrodynamic model, and perform dynamic and continuous simulation of the simulated water ecological environment according to the preset frequency; Among them, the S3 uses an improved Kalman filter to resample the output of the remote sensing big data inversion model to obtain the resampled output of the remote sensing big data inversion model. The improved Kalman filter is a combination of multi-particle resampling and Kalman filtering, including: S31: Initializing the particle space based on the output of the remote sensing big data inversion model, defining multiple scales, and setting the particle weight of each scale, resampling the particle space based on the particle weight to obtain a multi-scale particle space, wherein the particle weight is an entropy value measurement of the corresponding scale; S32: Take each particle in the multi-scale particle space as the predicted state, and perform Kalman filtering based on formula (2) to obtain the output of the resampled remote sensing big data inversion model: Among them, x r is the value after filtering, x t-1 is the value of step t-1, U t-1 is the system noise.

2. The method for dynamic simulation of a watershed full-scale water ecosystem according to claim 1, characterized in that: The step S1 of acquiring a remote sensing image of the water ecological environment to be simulated, preprocessing the remote sensing image of the selected area, and obtaining the preprocessed remote sensing image includes: S11: Acquire remote sensing images of the water ecological environment to be simulated; S12: performing geometric correction on the remote sensing image to obtain a remote sensing image after geometric correction; S13: performing radiation calibration and atmospheric correction on the remote sensing image after geometric correction to obtain a remote sensing image after radiation calibration; S14: Perform a long-term data consistency check on the remote sensing image after radiometric calibration to obtain a preprocessed remote sensing image.

3. The method for dynamic simulation of a watershed full-scale water ecosystem according to claim 1, characterized in that: The step S2 constructs a remote sensing big data inversion model, inputs the preprocessed remote sensing image into the remote sensing big data inversion model, and obtains the output of the remote sensing big data inversion model, including: S21: Initializing the GA-ANN model, wherein the initializing the GA-ANN model includes: initializing GA parameters and initializing network weights and biases of the ANN model; S22: collecting spectrum and water quality parameter data of the water ecological environment to be simulated, and performing regression analysis on the spectrum and water quality parameter data of the water ecological environment to be simulated based on the ANN model; S23: optimizing the ANN model by calculating the reverse path based on GA to obtain global optimal parameters of network weights and biases of the ANN model, wherein the GA includes: a chromosome, an initialized population, a selection operation, a crossover operation, and a mutation operation; S24: The GA-ANN model is updated based on the global optimal parameters of the network weights and biases of the ANN model, and the updated GA-ANN model is set as the remote sensing big data inversion model.

4. The method for dynamic simulation of a watershed full-scale water ecosystem according to claim 3 is characterized in that: The S23 optimizes the ANN model by calculating the reverse path based on GA to obtain the global optimal parameters of the network weight and bias of the ANN model, and the GA includes: chromosomes, initialized populations, selection operations, crossover operations and mutation operations, including: S231: Create an initialized population containing a series of chromosomes; S232: Calculate the fitness based on formula (1) to obtain the fitness score of each chromosome: Where N is the number of sampling points; y i ′ is the prediction of the sampling point, y i is the measured value at the sampling point; S233: performing a selection operation on the population based on a selection algorithm, and selecting a chromosome with a fitness score higher than a threshold as a parent chromosome; S234: performing a crossover operation on the parent chromosome to generate a child chromosome, wherein the crossover operation includes a single-point crossover, a multi-point crossover or a uniform crossover; S235: performing a random mutation operation on the offspring chromosome, wherein the mutation operation is to randomly change the gene value in the chromosome; S236: Repeat S232-S235 until a predetermined stopping condition is reached to obtain the global optimal parameters of the network weights and biases of the ANN model, wherein the predetermined stopping condition is the maximum number of iterations or the optimization result of the optimization function reaching a preset threshold.

5. The method for dynamic simulation of a watershed full-scale water ecosystem according to claim 1, characterized in that: The step S4 obtains the initial measurement values ​​of the boundary conditions, hydrological and meteorological data and water quality spatial distribution characteristic indicators corresponding to the water ecological environment to be simulated, and inputs them into the FVCOM hydrodynamic model to obtain the output of the FVCOM hydrodynamic model, including: S41: Obtain boundary conditions, hydrological and meteorological data, and water quality spatial distribution characteristic indicators corresponding to the area to be simulated, wherein the hydrological and meteorological data include temperature, runoff, flow velocity, and wind speed, and the water quality spatial distribution characteristic indicators include chlorophyll, total nitrogen, total phosphorus, and ammonia nitrogen; S42: generating grid data of a hydrodynamic model based on preset parameters, and setting initial condition data; S43: Input the boundary conditions, hydrological and meteorological data, water quality spatial distribution characteristic indicators, grid data, and initial condition data corresponding to the area to be simulated into the FVCOM model to obtain the output of the FVCOM model.

6. The method for dynamic simulation of a watershed full-scale water ecosystem according to claim 1, characterized in that: The S5 verifies and calibrates the output of the FVCOM hydrodynamic model using the output of the resampled remote sensing big data inversion model to obtain the calibrated FVCOM hydrodynamic model, including: S51: Set the simulation time range and time step in the FVCOM model; S52: Run the FVCOM model, compare the output of the resampled remote sensing big data inversion model with the output of the FVCOM model through the objective function, and use the Kalman filter to adjust the model parameters until the objective function reaches the optimal value. The objective function is used to calculate the degree of fit between the output of the resampled remote sensing big data inversion model and the observed data.

7. A dynamic simulation device for a full-scale water ecosystem in a watershed, characterized in that: The device is applicable to the method according to any one of claims 1 to 6, and comprises: Remote sensing image unit: used to obtain remote sensing images of the water ecological environment to be simulated, pre-process the remote sensing images of the selected area, and obtain pre-processed remote sensing images; Inversion model unit: used to construct a remote sensing big data inversion model, input the preprocessed remote sensing image into the remote sensing big data inversion model, and obtain the output of the remote sensing big data inversion model; Resampling unit: used to resample the output of the remote sensing big data inversion model by using an improved Kalman filter to obtain the resampled output of the remote sensing big data inversion model, wherein the improved Kalman filter is a combination of multi-particle resampling and Kalman filter; Hydrodynamic model unit: used to obtain the initial measurement values ​​of boundary conditions, hydrological and meteorological data and water quality spatial distribution characteristic indicators corresponding to the water ecological environment to be simulated, and input them into the FVCOM hydrodynamic model to obtain the output of the FVCOM hydrodynamic model; Calibration unit: used to verify and calibrate the output of the FVCOM hydrodynamic model using the output of the resampled remote sensing big data inversion model to obtain a calibrated FVCOM hydrodynamic model; Continuous simulation unit: used to input the initial measurement values ​​of boundary conditions, hydrological and meteorological data, and water quality spatial distribution characteristic indicators corresponding to the area to be simulated into the calibrated FVCOM hydrodynamic model, and perform dynamic and continuous simulation of the simulated water ecological environment according to the preset frequency; The execution steps of the resampling unit include: S31: Initializing the particle space based on the output of the remote sensing big data inversion model, defining multiple scales, and setting the particle weight of each scale, resampling the particle space based on the particle weight to obtain a multi-scale particle space, wherein the particle weight is an entropy value measurement of the corresponding scale; S32: Take each particle in the multi-scale particle space as the predicted state, and perform Kalman filtering based on formula (2) to obtain the output of the resampled remote sensing big data inversion model: Among them, x r is the value after filtering, x t-1 is the value of step t-1, U t-1 is the system noise.

8. An electronic device, characterized in that: The electronic device comprises: a housing, a processor, a memory, a circuit board and a power supply circuit, wherein the circuit board is placed inside the space enclosed by the housing, and the processor and the memory are arranged on the circuit board; the power supply circuit is used to supply power to various circuits or devices of the above-mentioned electronic device; the memory is used to store executable program codes; the processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the method described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the method of any one of claims 1 to 6.

Images