Water quality prediction methods, devices, and equipment that integrate remote sensing inversion and hydrodynamic models

Abstract

This invention provides a water quality prediction method, apparatus, and equipment that integrates remote sensing inversion and hydrodynamic modeling, relating to the field of water quality prediction technology. The method includes: inverting data based on satellite remote sensing data at an initial time to obtain pollutant concentration and water depth inversion values, used to set the initial operating conditions of the hydrodynamic model; running the hydrodynamic model at the current time to perform water quality simulation, obtaining time-series pollutant concentration and water depth predicted values; inverting data based on satellite remote sensing data at time k to obtain new pollutant concentration and water depth inversion values; performing assimilation processing to obtain assimilated pollutant concentration and water depth values; if it is determined that hydrodynamic model optimization should continue, updating the initial operating conditions of the hydrodynamic model using the assimilated values, re-running the hydrodynamic model for water quality simulation, until it is determined that model optimization and water quality simulation should be stopped, obtaining the time-series water quality prediction result. This invention can effectively improve the accuracy of pollutant simulation using hydrodynamic models.

Description

Technical Field

[0001] This invention relates to the field of water quality prediction technology, and in particular to a water quality prediction method, apparatus and equipment that integrates remote sensing inversion and hydrodynamic model. Background Technology

[0002] Hydrodynamic models play a crucial role in water pollution simulation. By simulating water flow and the transport, diffusion, and transformation of pollutants, they provide a scientific basis for water quality management and pollution control. Key applications include pollution diffusion and migration in rivers, lakes, coastal and marine environments; urban water pollution simulation; impact assessment of agricultural and industrial pollution emissions; and water quality management of reservoirs and drinking water sources. Hydrodynamic models can predict changes in pollutant concentrations, evaluate the effectiveness of pollution control measures, and support responses to sudden pollution events and long-term water quality change predictions.

[0003] However, water quality simulation requires high-resolution data such as flow rate, velocity, water depth, and pollutant concentration. In practical applications, however, data acquisition is difficult, and the spatial resolution of the observation data is often insufficient, leading to inaccurate initial and boundary conditions of the hydrodynamic model and affecting the reliability of the simulation results. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a water quality prediction method, apparatus and equipment that integrates remote sensing inversion and hydrodynamic model, which can effectively improve the accuracy of pollutant simulation by hydrodynamic model.

[0005] In a first aspect, embodiments of the present invention provide a water quality prediction method that integrates remote sensing inversion and hydrodynamic models, comprising:

[0006] Acquire satellite remote sensing data corresponding to the study area;

[0007] Inversion is performed based on satellite remote sensing data at the initial moment to obtain pollutant concentration inversion values ​​and water depth inversion values, which are then used to set the initial operating conditions of the hydrodynamic model.

[0008] The hydrodynamic model is run at the current time to simulate water quality and obtain time-series pollutant concentration predictions and water depth predictions, including pollutant concentration predictions and water depth predictions at time k; and, based on the satellite remote sensing data at time k corresponding to the current time, new pollutant concentration inversion values ​​and new water depth inversion values ​​are obtained through inversion.

[0009] The predicted pollutant concentration and the new pollutant concentration inversion value at time k, as well as the predicted water depth and the new water depth inversion value at time k, are subjected to data assimilation processing to obtain the assimilated pollutant concentration value and the assimilated water depth value.

[0010] If it is determined that hydrodynamic model optimization should continue, then time k will be taken as the new current time. The initial operating conditions of the hydrodynamic model will be updated using the pollutant concentration assimilation value and the water depth assimilation value. The hydrodynamic model will continue to run at the new current time to simulate water quality until it is determined that model optimization and water quality simulation should be stopped.

[0011] In one implementation, the predicted pollutant concentration value and the new inverted pollutant concentration value at time k, as well as the predicted water depth value and the new inverted water depth value at time k, are respectively subjected to data assimilation processing to obtain assimilated pollutant concentration values ​​and assimilated water depth values, including:

[0012] The new pollutant concentration inversion value is corrected using the measured pollutant concentration data corresponding to the study area. The corrected pollutant concentration inversion value and the predicted pollutant concentration value at time k are then assimilated to obtain the assimilated pollutant concentration value.

[0013] Furthermore, the new water depth inversion value is corrected using the measured water depth data corresponding to the study area, and the corrected water depth inversion value and the predicted water depth value at time k are assimilated to obtain the assimilated water depth value.

[0014] In one implementation, the new pollutant concentration inversion value is corrected using measured pollutant concentration data corresponding to the study area, including:

[0015] The support vector machine model was trained using measured pollutant concentration data corresponding to the study area to obtain a pollutant concentration correction model.

[0016] The new pollutant concentration inversion value is input into the pollutant concentration correction model, so that the new pollutant concentration inversion value is corrected by the pollutant concentration correction model to obtain the corrected pollutant concentration inversion value.

[0017] In one implementation, the corrected pollutant concentration inversion value and the predicted pollutant concentration value at time k are subjected to data assimilation processing to obtain the assimilated pollutant concentration value, including:

[0018] The fusion weights are determined based on the corrected pollutant concentration inversion values ​​and the predicted pollutant concentration values ​​at time k.

[0019] By using fusion weights, the corrected pollutant concentration inversion value and the pollutant concentration prediction value at time k are weighted and summed to achieve data assimilation and obtain the assimilated pollutant concentration value.

[0020] In one implementation, the fusion weights are determined based on the corrected pollutant concentration inversion value and the predicted pollutant concentration value at time k, including:

[0021] Determine the first error value between the corrected pollutant concentration inversion value and the measured pollutant concentration data, and determine the second error value between the predicted pollutant concentration value at time k and the measured pollutant concentration data;

[0022] The fusion weights are determined based on the comparison between the first error value and the second error value.

[0023] In one implementation, the initial operating conditions of the hydrodynamic model are set using pollutant concentration inversion values ​​and water depth inversion values, including:

[0024] Set the water level and flow velocity parameters of the hydrodynamic model at the initial moment, and use the pollutant concentration inversion value as the water quality parameter of the hydrodynamic model at the initial moment to obtain the initial conditions;

[0025] Furthermore, the water body extent at the initial moment is determined based on the water depth inversion value, and the boundary conditions of the hydrodynamic model at the initial moment are set based on the water body extent.

[0026] The initial operating conditions include initial conditions and boundary conditions.

[0027] Secondly, embodiments of the present invention also provide a water quality prediction device that fuses remote sensing inversion with a hydrodynamic model, comprising:

[0028] The data acquisition module is used to acquire satellite remote sensing data corresponding to the study area;

[0029] The model configuration module is used to perform inversion based on satellite remote sensing data at the initial time to obtain pollutant concentration inversion values ​​and water depth inversion values, so as to set the initial operating conditions of the hydrodynamic model using the pollutant concentration inversion values ​​and water depth inversion values.

[0030] The prediction and inversion module is used to run the hydrodynamic model to simulate water quality at the current time and obtain the predicted values ​​of time-series pollutant concentration and water depth, including the predicted values ​​of pollutant concentration and water depth at time k; and to perform inversion based on the satellite remote sensing data at time k corresponding to the current time to obtain new inverted values ​​of pollutant concentration and new inverted values ​​of water depth.

[0031] The assimilation module is used to perform data assimilation processing on the predicted pollutant concentration value and the new pollutant concentration inversion value at time k, as well as the predicted water depth value and the new water depth inversion value at time k, to obtain the assimilated pollutant concentration value and the assimilated water depth value.

[0032] The water quality prediction module is used to update the initial operating conditions of the hydrodynamic model by taking time k as the new current time if it is determined to continue hydrodynamic model optimization. The module then continues to run the hydrodynamic model for water quality simulation at the new current time until it is determined to stop model optimization and water quality simulation.

[0033] In one implementation, the assimilation module is specifically used for:

[0034] The new pollutant concentration inversion value is corrected using the measured pollutant concentration data corresponding to the study area. The corrected pollutant concentration inversion value and the predicted pollutant concentration value at time k are then assimilated to obtain the assimilated pollutant concentration value.

[0035] Furthermore, the new water depth inversion value is corrected using the measured water depth data corresponding to the study area, and the corrected water depth inversion value and the predicted water depth value at time k are assimilated to obtain the assimilated water depth value.

[0036] Thirdly, embodiments of the present invention also provide an electronic device, including a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement any of the methods provided in the first aspect.

[0037] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing computer-executable instructions, which, when invoked and executed by a processor, cause the processor to implement any of the methods provided in the first aspect.

[0038] The water quality prediction method, apparatus, and equipment fusion of remote sensing inversion and hydrodynamic model provided in this invention, after acquiring satellite remote sensing data corresponding to the study area, firstly performs inversion based on the satellite remote sensing data at the initial time to obtain pollutant concentration inversion values ​​and water depth inversion values, so as to set the initial operating conditions of the hydrodynamic model using the pollutant concentration inversion values ​​and water depth inversion values; then, it runs the hydrodynamic model at the current time to simulate water quality and obtains pollutant concentration prediction values ​​and water depth prediction values, including the pollutant concentration prediction value and water depth prediction value at time k; and finally, it performs inversion based on the satellite remote sensing data at time k corresponding to the current time to obtain new pollutant concentration prediction values. The pollutant concentration inversion value and the new water depth inversion value are obtained. Then, the pollutant concentration prediction value and the new pollutant concentration inversion value at time k, as well as the water depth prediction value and the new water depth inversion value at time k, are assimilated to obtain the pollutant concentration assimilation value and the water depth assimilation value. If it is determined that model optimization should continue, time k is taken as the new current time. The initial operating conditions of the hydrodynamic model are updated using the pollutant concentration assimilation value and the water depth assimilation value. The hydrodynamic model continues to run at the new current time to simulate water quality until it is determined that water quality prediction should be stopped. Finally, the pollutant concentration assimilation value at the time when water quality prediction is stopped is taken as the water quality prediction result. The above method uses satellite remote sensing data to obtain pollutant concentration inversion values ​​and water depth inversion values, which are then used to set the initial operating conditions of the hydrodynamic model, compensating for the deficiencies of insufficient and unevenly distributed ground observation points. At the same time, satellite remote sensing data can comprehensively reflect the conditions of large-scale water bodies and watersheds, and cover a wide area at the same time scale, thus ensuring data consistency. In addition, by dynamically updating the initial operating conditions of the hydrodynamic model, the initial operating conditions gradually approach the actual physical conditions, thereby gradually improving the accuracy of pollutant concentration simulation in the hydrodynamic model.

[0039] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained in accordance with the structures particularly pointed out in the description, claims and drawings.

[0040] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0041] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0042] Figure 1 This is a flowchart illustrating a water quality prediction method that integrates remote sensing inversion and hydrodynamic models, as provided in an embodiment of the present invention.

[0043] Figure 2 This is an overall flowchart of a water quality prediction method that integrates remote sensing inversion and hydrodynamic model according to an embodiment of the present invention;

[0044] Figure 3 This is a schematic diagram of the structure of a water quality prediction device that integrates remote sensing inversion and hydrodynamic model according to an embodiment of the present invention;

[0045] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] Currently, water quality prediction using hydrodynamic models suffers from low reliability due to difficulties in data acquisition and insufficient spatial resolution of observational data. Furthermore, the calibration and validation of hydrodynamic models require a large amount of observational data, which is scarce or incomplete in many regions, limiting the application and accuracy of these models. Remote sensing data possesses many unique characteristics, giving it a significant advantage in compensating for the lack of data in hydrodynamic models. With its wide coverage, high spatial resolution, multispectral observation, and long-term series capabilities, remote sensing data effectively compensates for the shortcomings of hydrodynamic models in terms of monitoring data acquisition and accuracy. Based on this, this invention provides a water quality prediction method, apparatus, and equipment that integrates remote sensing inversion and hydrodynamic models, effectively improving the accuracy of pollutant simulation using hydrodynamic models.

[0048] To facilitate understanding of this embodiment, a detailed description of the water quality prediction method that integrates remote sensing inversion and hydrodynamic models, as disclosed in this embodiment of the invention, will be provided first. (See [link to relevant documentation]). Figure 1 The diagram shows a flow chart of a water quality prediction method that integrates remote sensing inversion and hydrodynamic models. The method mainly includes the following steps S102 to S110:

[0049] Step S102: Obtain satellite remote sensing data corresponding to the study area.

[0050] Step S104: Based on the satellite remote sensing data at the initial time, inversion is performed to obtain the inversion values ​​of pollutant concentration and water depth, so as to set the initial operating conditions of the hydrodynamic model using the inversion values ​​of pollutant concentration and water depth.

[0051] The current time or a time adjacent to the current time can be used as the initial time. The initial operating conditions include initial conditions and boundary conditions. The initial conditions are used to set the parameters of the hydrodynamic model such as water level, flow velocity, and water quality at the initial time. The boundary conditions are used to define the boundaries of the hydrodynamic model region, including inflow boundary, outflow boundary, and water level boundary.

[0052] In one example, pollutant concentration inversion can be performed based on satellite remote sensing data at the initial time to obtain pollutant concentration inversion values; water body range inversion can be performed based on satellite remote sensing data at the initial time to obtain water body range inversion values; and water body range inversion values ​​can be superimposed with high-precision DEM (Digital Elevation Model) data corresponding to the study area to obtain water depth inversion values ​​(i.e., water depth or water level inversion values).

[0053] In one example, by using the pollutant concentration inversion value to configure the water quality parameters in the initial conditions, and setting the water level, flow velocity, etc. of the hydrodynamic model at the initial moment; and by using the water depth inversion value to set the boundary conditions, the initial operating conditions required to run the hydrodynamic model can be obtained.

[0054] Step S106: Run the hydrodynamic model at the current time to simulate water quality and obtain the predicted values ​​of time-series pollutant concentration and water depth, including the predicted values ​​of pollutant concentration and water depth at time k; and perform inversion based on the satellite remote sensing data at time k corresponding to the current time to obtain new inverted values ​​of pollutant concentration and new inverted values ​​of water depth.

[0055] The current time can be denoted as time t0. In one example, the time step δt is set according to business needs and computing power conditions, and the runtime of the hydrodynamic model is set. The hydrodynamic model starts running from time t0 to obtain the time series predicted values ​​of river pollutant concentration, water body range, and water depth with a time resolution of δt.

[0056] Here, the k-th time can be denoted as t kTime. In one example, based on t k By inverting pollutant concentrations from satellite remote sensing data at a given time, new pollutant concentration inversion values ​​can be obtained; based on t k By performing water body range inversion on the satellite remote sensing data at the current time, a new water body range inversion value can be obtained. By overlaying the new water body range inversion value with the high-precision DEM data corresponding to the study area, a new water depth inversion value can be obtained.

[0057] Step S108: Perform data assimilation processing on the predicted pollutant concentration value and the new pollutant concentration inversion value at time k, as well as the predicted water depth value and the new water depth inversion value at time k, to obtain the assimilated pollutant concentration value and the assimilated water depth value.

[0058] In one example, regarding pollutant concentration, the fusion method for the predicted pollutant concentration output by the hydrodynamic model, the inverted pollutant concentration value from remote sensing, and the surface water quality monitoring results (i.e., measured pollutant concentration data) can be achieved through data assimilation technology to obtain the assimilated pollutant concentration value; similarly, the assimilated values ​​for water body extent and water body depth can be obtained. This embodiment of the invention utilizes assimilation technology to combine the surface water quality monitoring results with the predicted pollutant concentration output by the hydrodynamic model, optimizing the hydrodynamic model's state and improving its accuracy. The data assimilation content mainly includes pollutant concentration, water body extent, and water body depth.

[0059] Step S110: If it is determined to continue hydrodynamic model optimization, then the k-th time is taken as the new current time t0. The initial operating conditions of the hydrodynamic model are updated using the pollutant concentration assimilation value and the water depth assimilation value. The hydrodynamic model continues to run at the new current time to simulate water quality until it is determined to stop model optimization and water quality simulation.

[0060] In one example, the decision to continue data assimilation and water quality prediction can be made based on the accuracy of the water quality prediction results from the hydrodynamic model. If it is determined to continue model optimization, the initial operating conditions of the hydrodynamic model will be reset using the assimilated pollutant concentration and water depth values ​​at the current moment, so that the initial operating conditions gradually approach the real physical conditions. Steps S106 to S108 are repeated until it is determined to stop model optimization and water quality simulation.

[0061] The water quality prediction method integrating remote sensing inversion and hydrodynamic model provided in this invention obtains pollutant concentration inversion values ​​and water depth inversion values ​​from satellite remote sensing data to set the initial operating conditions of the hydrodynamic model, thus compensating for the deficiencies of insufficient and unevenly distributed ground observation points. Simultaneously, satellite remote sensing data can comprehensively reflect the conditions of large-scale water bodies and watersheds, covering a wide area at the same time scale, thereby ensuring data consistency. Furthermore, by dynamically updating the initial operating conditions of the hydrodynamic model, the initial operating conditions gradually approach the actual physical conditions, thereby gradually improving the accuracy of pollutant concentration simulation by the hydrodynamic model.

[0062] To facilitate understanding, this invention provides a specific implementation of a water quality prediction method that integrates remote sensing inversion and hydrodynamic models. It includes:

[0063] (I) Data Preparation and Processing: The initial conditions of the hydrodynamic model are fundamental to its operation and directly affect the accuracy and reliability of the simulation. Within the study area, remote sensing technology is used to provide various initial conditions for the hydrodynamic model, especially in large areas where hydrological data is scarce, where remote sensing can provide crucial data support. By integrating remote sensing data and ground observation data, the accuracy and reliability of the hydrodynamic model can be improved.

[0064] It mainly includes three parts: remote sensing inversion of pollutant concentration, inversion of water body extent and depth, and other data collection.

[0065] (1.1) Remote sensing inversion of pollutant concentration:

[0066] 1) Obtain high-resolution multispectral satellite remote sensing data and measured concentrations of water pollutants, such as suspended solids (TSS), chlorophyll (Chl-a), and chemical oxygen demand (COD);

[0067] 2) Extract features related to pollutant concentration from satellite remote sensing data, including reflectance of each band, band combination and texture features, etc.

[0068] 3) Select feature variables extracted from satellite remote sensing data as model inputs, and select the measured concentration of water pollutants as target variables;

[0069] 4) Train the random forest model using the training set data. Specific steps include:

[0070] a. Set the parameters of the random forest, including the number of trees (n_estimators), maximum depth (max_depth), minimum number of sample splits (min_samples_split), etc.;

[0071] b. Fit the model using the training data, generate multiple decision trees, and obtain the final prediction result through majority voting (or average value in regression problems);

[0072] 5) Use the trained random forest model to predict pollutant concentrations from the initial satellite remote sensing data and generate pollutant concentration inversion values ​​for the study area.

[0073] (1.2) Inversion of water body extent and depth:

[0074] 1) Threshold-based water extraction:

[0075] The main steps for identifying water bodies based on the unsupervised classification K-means method include: feature selection: selecting features for classification, such as reflectance of each band and water body index; clustering: applying clustering algorithms to classify the image, clustering pixels into several classes; post-processing: manually or automatically identifying the water body categories in the clustering results and generating the final water body range extraction results.

[0076] 2) Water depth calculation:

[0077] The water body extent extraction results are overlaid with a high-precision DEM to calculate the water depth inversion value at each point in the study area.

[0078] (1.3) Other data collection:

[0079] 1) Hydrological and hydrodynamic parameters, including: flow rate and velocity: flow rate and velocity of rivers or water bodies.

[0080] 2) Meteorological parameters, including: Rainfall: Rainfall intensity and distribution, used as input for water volume in the model; Evaporation: Surface evaporation rate, used to calculate water balance; Wind speed and direction: Affect water surface fluctuations and water mixing.

[0081] 3) Topographic and geomorphological parameters, including: Digital Elevation Model (DEM): Topographic elevation data that defines the topographic features of a watershed or body of water; River morphology: Geometric parameters such as river width, depth, and slope.

[0082] 4) Flow rate at upstream and downstream sections, and pollutant concentration data at upstream and downstream sections.

[0083] (ii) Two-dimensional hydrodynamic modeling, including: converting collected and processed terrain data into a hydrodynamic model-compatible format, correcting the data, interpolating irregular data, and generating regular grid data. Specifically:

[0084] (2.1) Import terrain data: Import high-precision digital elevation model (DEM) data.

[0085] (2.2) Generate triangular mesh: Use the mesh generation tool to create a triangular mesh suitable for model calculation.

[0086] (2.3) Set mesh parameters: Set the mesh refinement level according to the study area and simulation accuracy requirements to obtain a triangular mesh file compatible with the hydrodynamic model.

[0087] (III) Setting hydrodynamic model parameters, including:

[0088] (3.1) Initial and boundary conditions:

[0089] Initial conditions: Set parameters such as water level, flow velocity, and water quality at the initial moment of the model. In the specific implementation, the water level and flow velocity parameters of the hydrodynamic model at the initial moment are set, and the pollutant concentration inversion value is used as the water quality parameter of the hydrodynamic model at the initial moment to obtain the initial conditions.

[0090] Boundary conditions: Define the boundary conditions of the model region, including inflow boundary, outflow boundary, and water level boundary. In specific implementation, the initial water body extent is determined based on the water depth inversion value, and the boundary conditions of the hydrodynamic model at the initial time are set based on the water body extent.

[0091] (3.2) Model parameters:

[0092] Physical parameters: Set the river friction coefficient (such as Manning roughness coefficient), turbulence parameters, diffusion coefficient, etc.

[0093] Water quality parameters: set biochemical reaction rate, sedimentation rate, resuspension rate, etc.

[0094] (iv) Running the hydrodynamic model: After creating the triangular mesh file and setting the model parameters, determine the time step based on business needs and computing power. The runtime period for the model is typically determined based on research requirements and data availability. The hydrodynamic model is run starting at time t0, simulating the time-series model state set, which includes predicted pollutant concentrations. Predicted water body extent and water depth prediction values Where n is the total number of time points.

[0095] (V) Simulation Result Analysis and Data Fusion: The hydrodynamic model is run to output water quality simulation results, namely the aforementioned predicted values ​​of pollutant concentration, water body range, and water depth; satellite remote sensing data of the study area is acquired, and water depth, flow velocity, and pollutant concentration data are obtained through inversion; high temporal resolution and high precision ground monitoring data (including measured pollutant concentration data and measured water depth data) are obtained through automatic hydrological stations and water quality stations.

[0096] The fusion method of water quality simulation results from hydrodynamic models, pollutant concentration inversion values ​​from remote sensing, and ground monitoring data is achieved through data assimilation technology. Data assimilation technology combines observational data with model predictions to optimize the model state and improve simulation accuracy. The data assimilation content mainly includes water depth, water body extent, and pollutant concentration. It mainly falls into the following two categories:

[0097] Scenario 1, regarding pollutant concentration: The new pollutant concentration inversion value is corrected using the measured pollutant concentration data corresponding to the study area. The corrected pollutant concentration inversion value and the predicted pollutant concentration value are then processed by data assimilation to obtain the assimilated pollutant concentration value.

[0098] In the second scenario, regarding water depth, the new water depth inversion value is corrected using the measured water depth data corresponding to the study area. The corrected water depth inversion value and the predicted water depth value are then processed to obtain the assimilated water depth value.

[0099] Taking pollutant concentration as an example, this embodiment of the invention provides a specific step for assimilation:

[0100] (5.1) Correcting the new pollutant concentration inversion value using the measured pollutant concentration data corresponding to the study area, including: training the support vector machine model using the measured pollutant concentration data corresponding to the study area to obtain the pollutant concentration correction model; inputting the new pollutant concentration inversion value into the pollutant concentration correction model to correct the new pollutant concentration inversion value through the pollutant concentration correction model to obtain the corrected pollutant concentration inversion value.

[0101] In one example, t k The satellite remote sensing data at time k is processed by algorithm (I) to obtain the remote sensing inversion results, including new pollutant concentration inversion values. New water depth inversion values The remote sensing inversion results were corrected using ground monitoring data to obtain the corrected pollutant concentration inversion value C′. k Corrected water depth inversion value H′ k The specific method is as follows:

[0102] a. collect t k We used real-time pollutant data from river cross-sections, cleaned the data, removed missing and outlier values, established a data sample set, and divided the dataset into a training set and a test set.

[0103] b. Train an SVM (Support Vector Machine) model using the training set data. The goal of the SVM model is to find an optimal hyperplane that separates data points of different classes.

[0104] 1) You can choose a polynomial kernel function, or other kernel functions;

[0105] 2) Initialize the SVM model using the selected kernel function and parameters;

[0106] 3) Use the training set data to train the SVM model. The SVM uses an optimization algorithm to find a separating surface that maximizes the distance from the support vectors to the hyperplane.

[0107] 4) Use the test set to evaluate the model performance, calculate metrics such as classification accuracy, F1-score, and AUC-ROC, and evaluate the model's classification effect.

[0108] 5) Based on the evaluation results, further adjust the model's hyperparameters (including C value, kernel function type, and parameters), and repeat the training and evaluation process until the model meets the requirements.

[0109] c. Process pollutant concentration inversion values ​​using a trained SVM model. The corrected pollutant concentration inversion value C is obtained. k ′ ;

[0110] Similarly, a water depth SVM model is trained using a similar method, and the model is then used to process new water depth inversion values. The corrected water depth inversion value H is obtained. ′ k .

[0111] (5.2) The corrected pollutant concentration inversion values ​​and the predicted pollutant concentration values ​​at time k are subjected to data assimilation processing to obtain assimilated pollutant concentration values, including:

[0112] a. Determine the fusion weights based on the corrected pollutant concentration inversion values ​​and the predicted pollutant concentration values. Specifically, determine the first error value between the corrected pollutant concentration inversion values ​​and the measured pollutant concentration data, and determine the second error value between the predicted pollutant concentration value and the measured pollutant concentration data at time k; determine the fusion weights based on the comparison results between the first error value and the second error value.

[0113] b. Using fusion weights, the corrected pollutant concentration inversion value and the predicted pollutant concentration value at time k are weighted and summed to achieve assimilation and obtain the assimilated pollutant concentration value.

[0114] In practical applications, for time t k The corrected pollutant concentration inversion value C k ′ Compared with the pollutant concentration predictions output by the hydrodynamic model By performing a weighted summation, we obtain the optimal estimate at that moment, which is also the assimilated value of the pollutant concentration. The formula for assimilating pollutant concentrations is shown below:

[0115]

[0116] Similarly, the water depth assimilation value can be obtained. The formula for water depth assimilation is shown below:

[0117]

[0118] Where α and β are weighting coefficients, 0≤α≤1, 0≤β≤1. α and β are determined based on the error analysis between the remote sensing inversion correction values ​​and the model prediction values, with smaller errors accounting for a larger proportion, and the values ​​of α and β are adjusted accordingly.

[0119] Then based on the water depth assimilation value Adjust the water body extent to update the boundary conditions of the hydrodynamic model.

[0120] (vi) Model status update: If water quality prediction is to continue, output the pollutant concentration distribution, water body range and water depth obtained from (iv) after data assimilation as the initial conditions of the hydrodynamic model and restart the hydrodynamic model water quality simulation.

[0121] (vii) Iterative operation: Repeat (iv) to (vi), repeat the prediction and update steps, the initial conditions of the model gradually approach the real physical conditions, and the accuracy of the hydrodynamic model in simulating pollutant concentration is gradually improved.

[0122] (viii) Model Validation and Adjustment:

[0123] (8.1) Verify the model output using actual observation data such as water level, flow velocity, flow rate, and pollutant concentration. By comparing the model output with the observation data, identify the model's accuracy and error distribution. Use statistical methods such as mean square error and bias analysis to analyze the spatial and temporal distribution of model errors and determine which regions and time periods the model performs poorly.

[0124] (8.2) Through parameter sensitivity analysis, determine the model parameters (such as Manning coefficient, roughness coefficient, boundary conditions, initial conditions, etc.) that have the greatest impact on the output results. By changing the value of a certain parameter, observe its impact on the simulation results and evaluate its sensitivity.

[0125] (8.3) Parameter Adjustment and Optimization: Based on error analysis and sensitivity analysis, prioritize adjusting the parameters that are most sensitive to the simulation results and also contribute significantly to the error. These include model parameters such as roughness coefficient, boundary conditions, and initial conditions, to make the parameters reflect the real physical environment as much as possible.

[0126] (8.4) Model structure correction:

[0127] Mesh optimization: Optimizing the spatial mesh of the model. Increasing the mesh resolution in areas with large errors can improve simulation accuracy. Higher mesh resolution can more accurately simulate small-scale hydrodynamic features, but it also increases computational cost.

[0128] Improvement of water body boundary conditions: By introducing more hydrodynamic boundary condition data or more accurate topographic data, the boundary conditions and water body characteristics of the model are corrected.

[0129] (8.5) Model Evaluation and Optimization Cycle: Compare the adjusted model results with the previous simulation results to evaluate the effectiveness of the accuracy improvement. If the effect is not ideal, parameters can be further adjusted. Through continuous adjustment and verification, perform an optimization cycle until the model reaches the required accuracy level.

[0130] In summary, the embodiments of the present invention provide as follows: Figure 2 The flowchart shown is an overall process for a water quality prediction method that integrates remote sensing inversion and hydrodynamic models. The process includes: acquiring satellite remote sensing data, water quality parameters, hydrological and hydrodynamic parameters, meteorological parameters, and topographic and geomorphological data; inverting pollutant concentrations and water body extent and depth based on satellite remote sensing data, and configuring the hydrodynamic model adoption number in conjunction with water quality parameters, hydrological and hydrodynamic adoption numbers, and meteorological parameters; building a two-dimensional hydrodynamic model using topographic and geomorphological data; after configuring the hydrodynamic model parameters and building the two-dimensional hydrodynamic model, running the hydrodynamic model to simulate water quality and obtain simulation results (including predicted pollutant concentrations and predicted water depths); additionally, for t k Using real-time satellite remote sensing data, spatial distribution data of water bodies (i.e., new inverted values ​​of water body extent), spatial distribution data of pollutant concentration (i.e., new inverted values ​​of pollutant concentration), and spatial distribution data of water depth (i.e., new inverted values ​​of water depth) are obtained through a remote sensing inversion model. Then, water body data, pollutant concentration data, and water depth data are corrected based on ground observation data. This data is then assimilated using the simulation results from the aforementioned model to update the hydrodynamic model parameters. A decision is made whether to continue data assimilation; if the decision is yes, t0 = t... k Then, the model simulation results are retrieved again, and model verification can be performed if the result is negative.

[0131] This invention fully utilizes the wide coverage of remote sensing data to construct a remote sensing inversion model. Data on the water body extent, depth, and pollutant concentration in the study area are obtained through multispectral and hyperspectral satellites and used as input data for the two-dimensional hydrodynamic model. This compensates for the deficiencies of insufficient and unevenly distributed ground observation points. Furthermore, remote sensing data can comprehensively reflect the conditions of large-scale water bodies and watersheds, covering a wide area at the same time scale, ensuring data consistency. The water quality simulation results of the hydrodynamic model, the pollutant concentration obtained through remote sensing inversion, and real-time ground observation data are used for data correction and assimilation, and these are then used as initial conditions to output the hydrodynamic model. This improves the accuracy of the model's initial conditions, thereby enhancing the accuracy of pollutant simulation in the hydrodynamic model.

[0132] Based on the foregoing embodiments, this invention provides a water quality prediction device that integrates remote sensing inversion and hydrodynamic models. (See also...) Figure 3 The diagram shows a water quality prediction device that integrates remote sensing inversion and hydrodynamic models. The device mainly includes the following parts:

[0133] Data acquisition module 302 is used to acquire satellite remote sensing data corresponding to the study area;

[0134] The model configuration module 304 is used to perform inversion based on satellite remote sensing data at the initial time to obtain pollutant concentration inversion values ​​and water depth inversion values, so as to set the initial operating conditions of the hydrodynamic model using the pollutant concentration inversion values ​​and water depth inversion values.

[0135] The prediction and inversion module 306 is used to run the hydrodynamic model at the current time to simulate water quality and obtain the predicted values ​​of time-series pollutant concentration and water depth, including the predicted values ​​of pollutant concentration and water depth at time k; and to perform inversion based on the satellite remote sensing data at time k corresponding to the current time to obtain new inverted values ​​of pollutant concentration and new inverted values ​​of water depth.

[0136] The assimilation module 308 is used to perform data assimilation processing on the predicted pollutant concentration value and the new pollutant concentration inversion value at time k, as well as the predicted water depth value and the new water depth inversion value at time k, to obtain the assimilated pollutant concentration value and the assimilated water depth value.

[0137] The water quality prediction module 310 is used to update the initial operating conditions of the hydrodynamic model by taking time k as the new current time if it is determined to continue hydrodynamic model optimization, using the pollutant concentration assimilation value and water depth assimilation value, and continue to run the hydrodynamic model for water quality simulation at the new current time until it is determined to stop model optimization and water quality simulation.

[0138] The water quality prediction device integrating remote sensing inversion and hydrodynamic model provided in this invention obtains pollutant concentration inversion values ​​and water depth inversion values ​​from satellite remote sensing data to set the initial operating conditions of the hydrodynamic model, thus compensating for the deficiencies of insufficient and unevenly distributed ground observation points. Simultaneously, satellite remote sensing data can comprehensively reflect the conditions of large-scale water bodies and watersheds, covering a wide area at the same time scale, thereby ensuring data consistency. Furthermore, by dynamically updating the initial operating conditions of the hydrodynamic model, the initial operating conditions gradually approach the actual physical conditions, thereby gradually improving the accuracy of pollutant concentration simulation in the hydrodynamic model.

[0139] In one embodiment, the water quality prediction module 310 is specifically used for:

[0140] The new pollutant concentration inversion value is corrected using the measured pollutant concentration data corresponding to the study area. The corrected pollutant concentration inversion value and the predicted pollutant concentration value at time k are then assimilated to obtain the assimilated pollutant concentration value.

[0141] Furthermore, the new water depth inversion value is corrected using the measured water depth data corresponding to the study area, and the corrected water depth inversion value and the predicted water depth value at time k are assimilated to obtain the assimilated water depth value.

[0142] In one embodiment, the water quality prediction module 310 is specifically used for:

[0143] The support vector machine model was trained using measured pollutant concentration data corresponding to the study area to obtain a pollutant concentration correction model.

[0144] The new pollutant concentration inversion value is input into the pollutant concentration correction model, so that the new pollutant concentration inversion value is corrected by the pollutant concentration correction model to obtain the corrected pollutant concentration inversion value.

[0145] In one embodiment, the water quality prediction module 310 is specifically used for:

[0146] The fusion weights are determined based on the corrected pollutant concentration inversion values ​​and the predicted pollutant concentration values.

[0147] By using fusion weights, the corrected pollutant concentration inversion value and the pollutant concentration prediction value are weighted and summed to achieve assimilation and obtain the assimilated pollutant concentration value.

[0148] In one embodiment, the water quality prediction module 310 is specifically used for:

[0149] Determine the first error value between the corrected pollutant concentration inversion value and the measured pollutant concentration data, and determine the second error value between the predicted pollutant concentration value and the measured pollutant concentration data;

[0150] The fusion weights are determined based on the comparison between the first error value and the second error value.

[0151] In one implementation, the model configuration module 304 is specifically used for:

[0152] Set the water level and flow velocity parameters of the hydrodynamic model at the initial moment, and use the pollutant concentration inversion value as the water quality parameter of the hydrodynamic model at the initial moment to obtain the initial conditions;

[0153] Furthermore, the water body extent at the initial moment is determined based on the water depth inversion value, and the boundary conditions of the hydrodynamic model at the initial moment are set based on the water body extent.

[0154] The initial operating conditions include initial conditions and boundary conditions.

[0155] The device provided in this embodiment of the invention has the same implementation principle and technical effect as the aforementioned method embodiment. For the sake of brevity, any parts not mentioned in the device embodiment can be referred to the corresponding content in the aforementioned method embodiment.

[0156] This invention provides an electronic device, specifically, the electronic device includes a processor and a storage device; the storage device stores a computer program, and the computer program, when run by the processor, executes the method described in any of the above embodiments.

[0157] Figure 4 The present invention provides a schematic diagram of the structure of an electronic device 100, which includes a processor 40, a memory 41, a bus 42 and a communication interface 43. The processor 40, the communication interface 43 and the memory 41 are connected through the bus 42. The processor 40 is used to execute executable modules, such as computer programs, stored in the memory 41.

[0158] The memory 41 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 43 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc.

[0159] Bus 42 can be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 4 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0160] The memory 41 is used to store programs. After receiving an execution instruction, the processor 40 executes the program. The method executed by the device for defining the flow process disclosed in any of the foregoing embodiments of the present invention can be applied to the processor 40 or implemented by the processor 40.

[0161] Processor 40 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of processor 40 or by instructions in software form. Processor 40 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 41. The processor 40 reads the information in memory 41 and, in conjunction with its hardware, completes the steps of the above method.

[0162] The computer program product of the readable storage medium provided in the embodiments of the present invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the foregoing method embodiments. For specific implementation, please refer to the foregoing method embodiments, which will not be repeated here.

[0163] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0164] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A water quality prediction method that integrates remote sensing inversion and hydrodynamic models, characterized in that, include: Acquire satellite remote sensing data corresponding to the study area; Based on the satellite remote sensing data at the initial moment, inversion is performed to obtain pollutant concentration inversion values ​​and water depth inversion values, so as to set the initial operating conditions of the hydrodynamic model using the pollutant concentration inversion values ​​and the water depth inversion values. The hydrodynamic model is run at the current time to simulate water quality and obtain time-series pollutant concentration predictions and water depth predictions, including pollutant concentration predictions and water depth predictions at time k; and, based on the satellite remote sensing data at time k corresponding to the current time, new pollutant concentration inversion values ​​and new water depth inversion values ​​are obtained. The predicted pollutant concentration and the new inverted pollutant concentration at time k, as well as the predicted water depth and the new inverted water depth at time k, are respectively subjected to data assimilation processing to obtain assimilated pollutant concentration values ​​and assimilated water depth values. If it is determined that hydrodynamic model optimization should continue, then the k-th time is taken as the new current time. The initial operating conditions of the hydrodynamic model are updated using the pollutant concentration assimilation value and the water depth assimilation value. The hydrodynamic model continues to run at the new current time to simulate water quality until it is determined that model optimization and water quality simulation should be stopped.

2. The water quality prediction method based on the fusion of remote sensing inversion and hydrodynamic model according to claim 1, characterized in that, The predicted pollutant concentration and the new retrieved pollutant concentration at time k, as well as the predicted water depth and the new retrieved water depth at time k, are subjected to data assimilation processing to obtain assimilated pollutant concentration values ​​and assimilated water depth values, including: The new pollutant concentration inversion value is corrected using the measured pollutant concentration data corresponding to the study area. The corrected pollutant concentration inversion value and the predicted pollutant concentration value at time k are then assimilated to obtain the assimilated pollutant concentration value. Furthermore, the new water depth inversion value is corrected using the measured water depth data corresponding to the study area, and the corrected water depth inversion value and the predicted water depth value at time k are subjected to data assimilation processing to obtain the assimilated water depth value.

3. The water quality prediction method based on the fusion of remote sensing inversion and hydrodynamic model according to claim 2, characterized in that, The new pollutant concentration inversion values ​​are corrected using measured pollutant concentration data corresponding to the study area, including: The support vector machine model was trained using the measured pollutant concentration data corresponding to the study area to obtain the pollutant concentration correction model; The new pollutant concentration inversion value is input into the pollutant concentration correction model to correct the new pollutant concentration inversion value, thereby obtaining the corrected pollutant concentration inversion value.

4. The water quality prediction method based on the fusion of remote sensing inversion and hydrodynamic model according to claim 2, characterized in that, The corrected pollutant concentration inversion value and the predicted pollutant concentration value at time k are subjected to data assimilation processing to obtain the assimilated pollutant concentration value, including: The fusion weights are determined based on the corrected pollutant concentration inversion values ​​and the predicted pollutant concentration values ​​at time k. Using the fusion weights, the corrected pollutant concentration inversion value and the pollutant concentration prediction value are weighted and summed to achieve data assimilation processing and obtain the pollutant concentration assimilation value.

5. The water quality prediction method based on the fusion of remote sensing inversion and hydrodynamic model according to claim 4, characterized in that, The fusion weights are determined based on the corrected pollutant concentration inversion values ​​and the predicted pollutant concentration values ​​at time k, including: Determine a first error value between the corrected pollutant concentration inversion value and the measured pollutant concentration data, and determine a second error value between the predicted pollutant concentration value at time k and the measured pollutant concentration data; The fusion weight is determined based on the comparison between the first error value and the second error value.

6. The water quality prediction method based on the fusion of remote sensing inversion and hydrodynamic model according to claim 1, characterized in that, The initial operating conditions of the hydrodynamic model are set using the pollutant concentration inversion values ​​and the water depth inversion values, including: The initial conditions are obtained by setting the water level and flow velocity parameters of the hydrodynamic model at the initial time, and using the pollutant concentration inversion value as the water quality parameter of the hydrodynamic model at the initial time. Furthermore, the water body extent at the initial moment is determined based on the water depth inversion value, and the boundary conditions of the hydrodynamic model at the initial moment are set based on the water body extent. The operating conditions include the initial conditions and the boundary conditions.

7. A water quality prediction device that integrates remote sensing inversion and hydrodynamic modeling, characterized in that, include: The data acquisition module is used to acquire satellite remote sensing data corresponding to the study area; The model configuration module is used to perform inversion based on the satellite remote sensing data at the initial time to obtain pollutant concentration inversion values ​​and water depth inversion values, so as to set the initial operating conditions of the hydrodynamic model using the pollutant concentration inversion values ​​and the water depth inversion values; The prediction and inversion module is used to run the hydrodynamic model at the current time to simulate water quality and obtain time-series pollutant concentration prediction values ​​and water depth prediction values, including pollutant concentration prediction values ​​and water depth prediction values ​​at time k; and to perform inversion based on the satellite remote sensing data at time k corresponding to the current time to obtain new pollutant concentration inversion values ​​and new water depth inversion values. The assimilation module is used to assimilate the predicted pollutant concentration value and the new pollutant concentration inversion value at time k, as well as the predicted water depth value and the new water depth inversion value at time k, respectively, to obtain the assimilated pollutant concentration value and the assimilated water depth value. The water quality prediction module is used to, if it is determined that hydrodynamic model optimization should continue, take the k-th time as the new current time, update the initial operating conditions of the hydrodynamic model using the pollutant concentration assimilation value and the water depth assimilation value, and continue to run the hydrodynamic model for water quality simulation at the new current time until it is determined that model optimization and water quality simulation should be stopped.

8. The water quality prediction device that integrates remote sensing inversion and hydrodynamic modeling according to claim 7, characterized in that, The assimilation module is specifically used for: The new pollutant concentration inversion value is corrected using the measured pollutant concentration data corresponding to the study area. The corrected pollutant concentration inversion value and the predicted pollutant concentration value are then assimilated to obtain the assimilated pollutant concentration value. Furthermore, the new water depth inversion value is corrected using the measured water depth data corresponding to the study area, and the corrected water depth inversion value and the predicted water depth value are assimilated to obtain the assimilated water depth value.

9. An electronic device, characterized in that, The method includes a processor and a memory, the memory storing computer-executable instructions executable by the processor, the processor executing the computer-executable instructions to implement the method of any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions that, when invoked and executed by a processor, cause the processor to perform the method according to any one of claims 1 to 6.

Images