Pollution research and water quality prediction method based on river segmented water environment capacity analysis

Abstract

The invention discloses a pollution research and water quality prediction method based on river segmented water environment capacity analysis. The method comprises the following steps: accurately dividing a drainage basin into water environment capacity accounting units; obtaining a river water body image through satellite remote sensing; identifying and quantifying the positions and the intensities of point source and non-point source pollution sources in the drainage basin by utilizing the water quality parameters of the river obtained by the remote sensing image; establishing a river water quality prediction model; calculating the water environment capacity of each river reach under different surface water environment quality standards; comparing the estimated pollutant discharge amount with the water environment capacity under each water quality standard, and judging whether the pollutant discharge amount exceeds the capacity or not; and predicting future water quality change by using the model. According to the method, remote sensing, GIS and machine learning technologies are utilized, rapid analysis of the river segmented water environment capacity and accurate prediction of the water quality are achieved, dependence on a large amount of field sampling and complex software simulation is reduced, and the working efficiency is greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of environmental science and engineering technology, and in particular to a pollution research and water quality prediction method based on river segment water environment capacity analysis. Background Art

[0002] At present, the calculation of river water environment capacity mainly adopts actual river water sampling test or software simulation method, such as MIKE11. The actual sampling test method requires a lot of manpower, material and financial resources, and needs to collect and analyze water samples at different sections of the river. For larger river basins, this method may lead to inaccurate water environment capacity calculation due to insufficient sampling representativeness, and it is time-consuming and labor-intensive. The software simulation method requires a lot of basic data support, including detailed hydrological, meteorological and pollution source information. In practical applications, it is often difficult to establish a model or the simulation results are unreliable due to insufficient data.

[0003] The problem with existing technologies is that they consume a lot of resources and require a lot of data, making it difficult to efficiently and accurately calculate the water environment capacity of rivers, especially in large river basins or areas where data is scarce. This makes it difficult to predict water quality in river basins and control the total amount of pollutants, and it is impossible to provide an effective scientific basis for river basin management.

[0004] Due to the lack of accurate water environment capacity accounting and water quality prediction methods, it is difficult to control the total amount of pollutants, and the governance and improvement of the current water environment are hindered. Rivers may continue to be polluted, and water quality may not meet environmental standards, which will in turn affect the health of the ecosystem and human water safety. This not only restricts the sustainable development of the basin, but may also cause more serious environmental and social problems. Summary of the invention

[0005] In order to overcome the shortcomings of the prior art, the present invention proposes a pollution research and water quality prediction method based on river segment water environment capacity analysis, and uses remote sensing, GIS and machine learning technology to achieve rapid analysis of river segment water environment capacity and accurate prediction of water quality, reducing dependence on a large number of on-site sampling and complex software simulations, greatly improving work efficiency; reducing the investment in manpower, material and financial resources, and avoiding the high cost of conducting a large number of sampling in a large area of ​​the basin.

[0006] To achieve the above object, the present invention provides a pollution research and water quality prediction method based on river segment water environment capacity analysis, comprising the following steps: Step 1: Using GIS and remote sensing data, the river basin is accurately divided into water environment capacity accounting units based on river topography, hydrological characteristics and pollution source locations; Step 2: Obtain high-resolution river water images through satellite remote sensing and pre-process the images to extract effective information; Step 3: Using the spectral characteristics of remote sensing images, a machine learning algorithm is used to invert river water quality parameters, such as suspended matter concentration and chlorophyll content; Step 4: Use remote sensing and GIS technology to identify and quantify the location and intensity of point and non-point source pollution sources within the watershed; Step 5: Based on the water quality parameters and pollution source data obtained by inversion, a water quality prediction model for the river is established using a machine learning algorithm; Step 6: Using the water quality model, calculate the water environment capacity of each river section under different surface water environment quality standards; Step 7: Compare the estimated pollutant emissions with the water environment capacity under each water quality standard to determine whether the capacity is exceeded; Step 8: Use the model to predict future water quality changes and provide early warning information for water quality exceeding the standard; Step 9: Based on the analysis results, propose targeted pollution control measures and water environment management recommendations.

[0007] Furthermore, the specific method of step 1 is: Step 1.1: Obtain digital elevation model (DEM) data of the watershed to analyze terrain undulations and water flow direction; collect river network data, including information such as the location, length, and flow direction of rivers; obtain land use and land cover data to understand human activities and the natural environment in the area; Step 1.2: Use DEM data to calculate the river's slope, drainage area, and drainage path; determine the main tributaries and confluence points, and identify the river's watershed; Step 1.3: Collect known pollution source data, including industrial discharge outlets, sewage treatment plants, agricultural discharge points, etc.; mark the location of pollution sources on the GIS platform to form a pollution source distribution map; Step 1.4: Determine the standards for segmentation. You can set the length range of each river section based on administrative regions, tributary confluence points, pollution source locations, etc. to ensure the rationality and operability of the division; Step 1.5: Divide the river into several sections from upstream to downstream according to the established principles, ensuring that each section has relatively uniform geographical and hydrological characteristics; Step 1.6: Consider special geographical features or important ecological protection areas, and adjust the initially divided river sections to ensure that the pollution source concentration areas are accurately divided to facilitate subsequent analysis; Step 1.7: Check the length, area and number of pollution sources of each river section to ensure that the division is balanced, and discuss the division plan with relevant experts or departments to obtain professional opinions; Step 1.8: Based on the verification and feedback, the river segmentation scheme is finally determined, and each river segment is assigned a unique number and attribute information for easy management and analysis.

[0008] Furthermore, the specific process of step 2 is: Step 2.1: Determine the spatial resolution and spectral band required for the study area and select appropriate satellite data, such as Landsat 8 (30-meter resolution) or Sentinel-2 (10-meter resolution); Step 2.2: Download remote sensing images of the study area from public satellite data platforms, and select clear and cloudless images to avoid cloud and atmospheric influences; Step 2.3: Radiometric Correction: Convert the digital value (DN value) to radiance or surface reflectance, correcting for differences in sensor response, including: Atmospheric correction: Use atmospheric correction algorithms (such as FLAASH and 6S models) to eliminate the effects of atmospheric scattering and absorption; Geometric correction: Correct the spatial geometric distortion of the image so that it accurately matches the geographic coordinate system; Orthorectification: Use DEM data to eliminate the impact of terrain undulations on images and ensure the accuracy of the location of objects; Step 2.4: According to the boundaries of the study area, crop the image to extract the area of ​​interest. If necessary, stitch multiple images together to form a complete coverage area. Step 2.5: Detect the cloud and shadow areas in the image, create a cloud mask, and try to use images from other times to supplement the cloud-covered areas; Step 2.6: Check the preprocessed image for banding, noise or other anomalies to ensure that the spectral characteristics of the image are accurate and suitable for subsequent analysis; Step 2.7: Save the processed image data in a standard format (such as GeoTIFF) and back it up.

[0009] Furthermore, the specific process of step 3 is as follows: Step 3.1: According to the research requirements, select the water quality parameters that need to be inverted, such as suspended solids concentration (TSS), chlorophyll-a concentration (Chl-a), dissolved organic carbon (DOC), etc.; Step 3.2: Obtain measured water quality data that is synchronized with or close in time to the remote sensing image, ensuring that the location of the measured point is accurate and corresponds to the image pixel; Step 3.3: Extract the multi-band reflectance values ​​corresponding to the measured point locations from the preprocessed remote sensing image, and record the reflectance of each measured point in each band.

[0010] Step 3.4: Based on the reflectance value, calculate the spectral index related to the water quality parameters, such as NDVI, NDWI, Turbidity Index, etc., and select the index that best reflects the target water quality parameters; Step 3.5: Combine the measured water quality parameters with the corresponding spectral reflectance and spectral index to form a training sample set, ensuring that the number of samples is sufficient to cover different water quality conditions; Step 3.6: Select an appropriate machine learning algorithm, such as random forest, support vector machine, or neural network, train the model using the training dataset, and adjust the parameters to improve model performance; Step 3.7: Use the cross-validation method to evaluate the predictive ability of the model and calculate evaluation indicators such as root mean square error (RMSE), coefficient of determination (R²), etc. Step 3.8: Apply the trained model to the entire image data, calculate the value of the target water quality parameter pixel by pixel, and generate a spatial distribution map of the water quality parameter; Step 3.9: Analyze the spatial characteristics of the inversion results, identify the high-pollution areas, and use GIS software to create a water quality distribution map to visually display the results; if possible, use additional measured data or historical data to verify the accuracy of the inversion results.

[0011] Furthermore, the specific process of step 4 is as follows: Step 4.1: Use high-resolution remote sensing images to locate possible point sources of pollution, such as industrial areas, factories, and sewage treatment plants, and confirm and supplement information based on the list of pollution sources published by the government; Step 4.2: Based on the land use data, determine the non-point source pollution areas such as agricultural land, livestock farms, and urban expansion areas, and draw the boundaries of the non-point source pollution areas in GIS; Step 4.3: Obtain detailed information on point source emissions, types of emissions, emission time, etc., and collect agricultural activity data in non-point source areas, such as fertilizer and pesticide usage, livestock and poultry breeding scale, etc.; Step 4.4: Using known emission factors or direct emission data, calculate the pollutant emissions from each point source, taking into account the temporal and spatial variation characteristics of emissions; Step 4.5: Use empirical formulas or models (such as the outlet coefficient method) to estimate pollutant emissions from nonpoint source areas, taking into account the impact of factors such as rainfall, topography, and land management measures on nonpoint source emissions; Step 4.6: Organize the information of all pollution sources into a database, including spatial location and attribute data; Step 4.7: On the GIS platform, the pollution source data is associated with the geographic spatial information to generate a pollution source distribution map to show the spatial distribution and intensity of the pollution sources; Step 4.8: Classify the pollution sources according to the type of pollutant, such as nitrogen pollution sources, phosphorus pollution sources, organic pollution sources, etc.

[0012] Step 4.9: Check the rationality of the pollution source data, correct possible errors, and update the database regularly to reflect the latest pollution source information; organize the pollution source data into the format required by the model, including time series and spatial distribution, to ensure that the data matches the subsequent water quality model and capacity calculation.

[0013] Furthermore, the specific process of step 5 is as follows: Step 5.1: Organize the spatial distribution data of water quality parameters obtained by inversion in step 3, ensure the uniformity of data format, collect the pollution source data identified and quantified in step 4, including spatial location and pollutant emissions; obtain hydrological data, such as river flow and flow velocity, and meteorological data, such as rainfall and temperature; Step 5.2: Select a suitable machine learning algorithm, such as random forest, support vector machine, or neural network, based on data characteristics and research needs, taking into account the interpretability and predictive ability of the model to ensure that the model is suitable for water quality prediction; Step 5.3: Clean the input data, handle missing values ​​and outliers, ensure data quality, standardize or normalize numerical data, and eliminate the impact of different dimensions; Step 5.4: Use correlation analysis to determine the characteristic variables that are closely related to water quality parameters, such as pollutant emissions, flow, and temperature, eliminate redundant or low-correlation features, simplify the model, and improve efficiency; Step 5.5: Divide the data set into training set and test set, generally in a ratio of 8:2, to ensure the reliability of model evaluation, the randomness and representativeness of data division, and avoid data leakage; Step 5.6: Use the training set data to input the selected machine learning algorithm, train the water quality prediction model, adjust the model parameters, and use the cross-validation method to optimize the model performance; Step 5.7: Use the test set data to evaluate the prediction accuracy and generalization ability of the model and calculate evaluation indicators such as root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R²); Step 5.8: Adjust the model structure and parameters based on the evaluation results. You may need to reselect features or algorithms, and repeatedly train and evaluate until the model achieves satisfactory performance. Step 5.9: Apply the trained model to the entire river area, predict the water quality of each river section, and generate a spatial distribution map of water quality prediction to provide a basis for subsequent analysis.

[0014] Furthermore, the specific process of step 6 is as follows: Step 6.1: Identify the surface water quality standards that the river needs to meet, such as Class III, Class IV or Class V water quality, and obtain the pollutant concentration limits under each water quality standard as the basis for calculation; Step 6.2: Based on the definition of water environment capacity, formulate a calculation formula: water environment capacity = (water quality standard limit - upstream water quality background value) × river flow × time; consider the natural attenuation coefficient of pollutants, modify the formula and improve accuracy; Step 6.3: Obtain actual flow data for each river section, ensure the temporal and spatial matching of the data, determine the upstream water quality background value, which may require actual measurement or reference to historical data, determine the attenuation coefficient of pollutants, and refer to literature or experimental data; Step 6.4: For each river section, substitute the corresponding parameters and calculate the water environment capacity under different water quality standards. Considering the changes in different seasons or specific periods, it may be necessary to calculate separately; Step 6.5: Organize the capacity results of each river section under different water quality standards into a table, draw a spatial distribution map of water environment capacity on the GIS platform, and intuitively display the capacity size; Step 6.6: Compare with other research or authoritative data to verify the rationality of the calculation. If there are large differences, check the calculation process and parameters and correct the errors; Step 6.7: Record the calculation steps, formulas, and parameter sources in detail to ensure the repeatability of the results, and save the calculation tables and spatial distribution maps to facilitate subsequent analysis and report writing.

[0015] Furthermore, the specific process of step 7 is as follows: Step 7.1: Collect pollutant emission data from all pollution sources and summarize them by river section to ensure the temporal and spatial consistency of emission data and match the time scale of capacity calculation; Step 7.2: For each river section, calculate the difference between the actual pollutant discharge and the water environment capacity: Difference = actual discharge - water environment capacity; positive values ​​indicate that the discharge exceeds the capacity, and negative values ​​indicate that it is within the capacity range; Step 7.3: Determine the water quality prediction category of the river section based on the difference: if the difference is less than 0, the predicted water quality meets the target standard; if the difference is greater than 0, the predicted water quality does not meet the standard and the main pollution factors need to be identified; Step 7.4: Analyze the types of pollutants that exceed the capacity and determine the pollution factors that have the greatest impact on water quality. It may be necessary to calculate and compare each pollutant separately; Step 7.5: Visualize the difference results on the GIS platform and draw a distribution map of river sections with excessive pollution to intuitively display the spatial location and degree of pollution exceeding the standard; Step 7.6: If there are multiple periods of data, analyze the time trend of pollutant emissions and capacity to determine whether the pollution situation is improving or worsening year by year, and provide a basis for governance; Step 7.7: Compare with the actual water quality monitoring data to verify the accuracy of the prediction results. If the prediction is inconsistent with the actual measurement, analyze the reasons and it may be necessary to adjust the model or data; Step 7.8: Organize the results of the comparative analysis and write a report including text descriptions, tables and charts to provide decision-making support for relevant departments and clarify governance priorities.

[0016] Furthermore, the specific process of step 8 is as follows: Step 8.1: Determine the temporal and spatial scope of the forecast, such as water quality changes in the next year, and identify pollutants and river sections that need to be focused on; Step 8.2: Obtain future pollution source emission plans, such as new projects and industrial restructuring, collect weather forecast data, and consider the impact of extreme weather events on water quality; Step 8.3: Input the future scenario data into the water quality model, ensuring that the data format and units are consistent. For uncertain data, it may be necessary to set multiple scenarios for simulation; Step 8.4: Use the updated input data to run the water quality prediction model, simulate future water quality changes, and generate water quality time series data for each river section during the prediction period; Step 8.5: Assess whether water quality is likely to exceed standards in each river section in the future, identify high-risk areas, and analyze the extent and duration of possible pollutant exceedances; Step 8.6: According to the forecast results, set the warning level, such as general warning, severe warning, and clarify the conditions and standards for triggering the warning; Step 8.7: Release warning information to relevant departments and the public, including the areas that may be affected and the measures to be taken, and establish information release channels to ensure that warning information is communicated in a timely manner; Step 8.8: Develop corresponding emergency response plans according to the warning level, clarify the responsibilities and action plans of various departments, including pollution source supervision, emergency monitoring, public evacuation and other measures; Step 8.9: During the forecast period, strengthen water quality monitoring of key river sections, verify the accuracy of the forecast, adjust the model and early warning measures based on actual monitoring, and form a feedback loop.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention provides a pollution research and water quality prediction method based on river segment water environment capacity analysis. It uses remote sensing, GIS and machine learning technology to achieve rapid analysis of river segment water environment capacity and accurate prediction of water quality, reduce dependence on large-scale on-site sampling and complex software simulation, and greatly improve work efficiency; reduce investment in manpower, material and financial resources, and avoid the high cost of large-scale sampling in a large range of river basins.

[0018] 2. The present invention provides a pollution research and water quality prediction method based on river segment water environment capacity analysis, which is suitable for river basin environments with scarce data and vast areas, especially in large rivers or areas where data is difficult to obtain, and can effectively carry out water quality analysis and prediction work.

[0019] 3. The present invention provides a pollution research and water quality prediction method based on river segment water environment capacity analysis. By integrating multi-source data and advanced machine learning algorithms, the accuracy of water quality parameter inversion and prediction is improved, and the main pollution sources and pollution factors can be identified more accurately.

[0020] Strong guidance: It provides a scientific basis for watershed pollution research and water environment management, can identify major pollution factors and areas exceeding the standard, and facilitates the formulation of targeted pollution control measures and policies.

[0021] 4. The present invention provides a pollution research and water quality prediction method based on river segment water environment capacity analysis, applies remote sensing technology, GIS and machine learning methods to water environment capacity analysis, and through timely prediction and early warning, measures can be taken in advance to prevent further deterioration of water quality, which helps to achieve sustainable management and protection of the water environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0023] Figure 1 It is a schematic diagram of the steps of the present invention DETAILED DESCRIPTION

[0024] The technical solution of the present invention will be more clearly and completely explained below through description of preferred embodiments of the present invention in combination with the accompanying drawings.

[0025] like Figure 1 As shown, the present invention is specifically: Step 1: Using GIS and remote sensing data, the river basin is accurately divided into water environment capacity accounting units based on river topography, hydrological characteristics and pollution source locations; Step 2: Obtain river water body images through satellite remote sensing and preprocess the images to extract information; Step 3: Using the spectral characteristics of remote sensing images, the machine learning algorithm is used to invert the water quality parameters of the river; Step 4: Use remote sensing and GIS technology to identify and quantify the location and intensity of point and non-point source pollution sources within the watershed; Step 5: Based on the water quality parameters and pollution source data obtained by inversion, a water quality prediction model for the river is established using a machine learning algorithm; Step 6: Using the water quality model, calculate the water environment capacity of each river section under different surface water environment quality standards; Step 7: Compare the estimated pollutant emissions with the water environment capacity under each water quality standard to determine whether the capacity is exceeded; Step 8: Use the model to predict future water quality changes and provide early warning information on water quality exceeding the standard.

[0026] As a specific implementation method, according to the above method, first, a detailed spatial analysis of the target river and its basin is carried out using geographic information system (GIS) and remote sensing data. Basic geographic data such as digital elevation model (DEM), river network, land use type, etc. are obtained, and the entire basin is accurately divided into several water environment capacity accounting units in combination with the topographic and geomorphic characteristics, hydrological characteristics and location of pollution sources of the river. It is ensured that each unit has relatively uniform hydrological and pollution characteristics, laying the foundation for subsequent analysis.

[0027] Next, high-resolution river water images are obtained through satellite remote sensing technology, and suitable satellite data sources such as Landsat 8 or Sentinel-2 are selected. The acquired remote sensing images are strictly preprocessed, including radiation correction, atmospheric correction, and geometric correction to ensure that the images accurately reflect the true spectral information of the surface. Then, the spectral characteristics of the river water body are extracted, and the bands or band combinations related to the target water quality parameters are selected to construct the characteristic variable set.

[0028] On this basis, machine learning algorithms, such as random forest or support vector machine, are used to establish an inversion model for water quality parameters. Limited measured water quality data are used as training sets to train and verify the model to ensure that the model has good prediction accuracy. Through the trained model, the remote sensing image data of the entire basin is batch processed to obtain the spatial distribution of water quality parameters, providing a basis for identifying high-pollution areas.

[0029] At the same time, remote sensing data and GIS technology are used to identify and quantify pollution sources within the basin. Combined with high-resolution remote sensing images and land use data, the location and scale of point source pollution sources such as industrial discharge ports and sewage treatment plants, as well as non-point source pollution areas such as agricultural land and farms, are determined. Empirical formulas or models are used to estimate the pollutant emissions of various pollution sources, spatialize the pollution source data, and establish a complete pollution source database to provide accurate input data for the water quality model.

[0030] Then, based on the water quality parameters and pollution source data obtained by inversion, a data-driven water quality prediction model is established using machine learning algorithms. The model should consider multiple factors such as flow, meteorological conditions, and pollutant emissions to improve the accuracy of the prediction. Through model training, verification, and optimization, ensure that the model has good generalization capabilities. Using the trained model, predict future water quality changes and generate a spatial distribution map of water quality prediction to provide a scientific basis for early warning and governance.

[0031] Under different surface water environmental quality standards, the model is used to simulate the water environment capacity of each river section. The water quality standards that the river needs to meet are clarified, the pollutant concentration limits under each standard are obtained, the water environment capacity of each river section is calculated, and factors such as the natural attenuation coefficient of pollutants are considered. The estimated pollutant emissions are compared with the water environment capacity under each water quality standard, the difference is calculated, and it is determined whether the pollutant emissions exceed the capacity. Identify the pollution factors that exceed the standard and determine the pollutants that have the greatest impact on water quality.

[0032] Based on the results of the comparative analysis, we will conduct an in-depth analysis of the sources and influencing factors of the pollution factors that exceed the standard and determine the main pollution sources. We will comprehensively evaluate the water quality of the river basin and identify the river sections that meet and do not meet the water quality standards. We will propose specific pollution control and governance measures for the pollution factors that exceed the standard, such as strengthening industrial emission supervision, promoting ecological agriculture, and improving sewage treatment facilities. We will formulate a detailed action plan, clarify the responsible units, time arrangements, and funding requirements, and ensure the effective implementation of governance measures.

[0033] Through the above specific implementation methods, efficient analysis of the water environment capacity of river sections and accurate prediction of water quality are achieved, providing a scientific and practical solution for river basin pollution research and water environment management.

[0034] As a specific implementation example, take the water environment capacity analysis and water quality prediction of a river basin as an example. The river (hereinafter referred to as the river Z) is located in a certain province, with a total length of 150 kilometers, flowing through multiple industrial and agricultural areas. In recent years, the water quality has declined significantly, mainly affected by industrial wastewater and agricultural non-point source pollution. In order to improve the water quality of the Z River, it is necessary to perform a water environment capacity analysis and water quality prediction. The steps are as follows: First, a detailed spatial analysis of the Z River and its basin was conducted using geographic information systems (GIS) and remote sensing data. The basin's digital elevation model (DEM), river network, land use type and other basic geographic data were obtained. Combined with the river's topographic and geomorphic features, hydrological characteristics and the location of pollution sources, the Z River was divided into eight sections from upstream to downstream, numbered Z1 to Z8. Each section is about 20 kilometers long, ensuring that the hydrological and pollution characteristics are relatively uniform, laying the foundation for subsequent analysis and management.

[0035] Next, high-resolution river water images were obtained through satellite remote sensing technology. Sentinel-2 satellite data was selected with a spatial resolution of 10 meters and 13 spectral bands. Cloud-free remote sensing images during the summer of 2022 were downloaded, covering the entire Z River basin. The acquired remote sensing images were preprocessed, including radiation correction (converting the original DN value into surface reflectance), atmospheric correction (using the 6S model to eliminate the effects of atmospheric scattering and absorption), geometric correction, and orthorectification (ensuring that the image accurately corresponds to the actual geographic location) to accurately reflect the true spectral information of the surface. Then, the spectral characteristics of the river water body were extracted, and spectral indices related to the target water quality parameters, such as the normalized difference water index (NDWI) and the chlorophyll index, were calculated.

[0036] Using limited measured water quality data as a training set, water samples were collected at key locations in eight sections of the Z River from June to August 2022. A total of 40 samples were collected, and the total suspended solids (TSS) and chlorophyll-a (Chl-a) concentrations of each sample were measured, and the GPS coordinates of the sampling points were recorded. The random forest algorithm was used to establish an inversion model for water quality parameters, and the model was trained and verified. The model's coefficient of determination R² reached 0.88, and the root mean square error RMSE was 3.5 mg / L, which had good prediction accuracy. Through the trained model, the remote sensing image data of the entire basin was batch processed to obtain the spatial distribution of TSS and Chl-a, providing a basis for identifying high-pollution areas.

[0037] At the same time, remote sensing data and GIS technology are used to identify and quantify pollution sources within the basin. Combined with high-resolution remote sensing images and land use data, the location and scale of point source pollution sources such as industrial discharge ports and sewage treatment plants, as well as non-point source pollution areas such as agricultural land and livestock and poultry farms are determined. The locations of industrial discharge ports and sewage treatment plants are marked on the GIS platform, and a total of 15 major point sources are identified. Pollution source attribute data are collected, and emission permit data of industrial enterprises, including emission volume and pollutant types, are obtained. The amount of pesticides and fertilizers used in agricultural areas and the scale of livestock and poultry farming are collected. Empirical formulas or models are used to estimate the pollutant emissions of various pollution sources, spatialize the pollution source data, and establish a complete pollution source database to provide accurate input data for the water quality model.

[0038] Then, based on the water quality parameters and pollution source data obtained by inversion, a data-driven water quality prediction model was established using a machine learning algorithm. The gradient boosting decision tree (GBDT) algorithm was used to capture complex nonlinear relationships. The model took into account multiple factors such as flow, meteorological conditions, and pollutant emissions. All variables were standardized, and feature importance analysis was used to select the variables that have the greatest impact on water quality, such as total phosphorus emissions, flow, and rainfall. The data set was divided into 70% training set and 30% test set. The training set was used to train the GBDT model, and the learning rate and the number of trees were adjusted to optimize the model performance. The model was evaluated on the test set, with an R² of 0.83 for TSS and 0.80 for Chl-a, which has good generalization ability. Using the trained model, future water quality changes are predicted, and a spatial distribution map of water quality prediction is generated to provide a scientific basis for early warning and governance.

[0039] Under different surface water environmental quality standards, the model was used to simulate the water environment capacity of each river section. Taking the surface water Class III standard as the target, the limits of TSS and Chl-a were obtained, and the capacity calculation formula was established: the water environment capacity is equal to the water quality standard limit multiplied by the flow minus the upstream section concentration multiplied by the flow. The average flow of each river section and the water quality concentration of the upstream section were obtained. For each river section, the difference between the actual pollutant discharge and the water environment capacity was calculated to determine whether the pollutant discharge exceeded the capacity. It was found that the total phosphorus and ammonia nitrogen emissions in the Z3 to Z6 river sections exceeded the capacity, and agricultural non-point source pollution was serious.

[0040] According to the results of comparative analysis, the sources and influencing factors of the excessive pollution factors are deeply analyzed. The main pollution factors are total phosphorus and ammonia nitrogen, which mainly come from excessive use of fertilizers and livestock and poultry breeding waste. Specific pollution control and governance measures are proposed for the excessive pollution factors, such as promoting soil testing and formula fertilization in agriculture to reduce the application of fertilizers by 20%; building manure treatment facilities in livestock and poultry breeding to realize the resource utilization of waste; in terms of ecological restoration, planting buffer zones on river banks to intercept pollutants in surface runoff. Formulate a detailed action plan to clarify the implementation time, responsible departments and funding investment of various measures to ensure the effective implementation of governance measures.

[0041] The above specific implementations are only descriptions of the preferred implementations of the present invention, and do not limit the protection scope of the present invention. Without departing from the design concept and spirit of the present invention, various modifications, substitutions and improvements made by ordinary technicians in this field to the technical solution of the present invention based on the text description and drawings provided by the present invention should all fall within the protection scope of the present invention. The protection scope of the present invention is determined by the claims.

Claims

1. Pollution research and water quality prediction method based on river segment water environment capacity analysis, characterized in that: The following steps are involved: Step 1: Using GIS and remote sensing data, the river basin is accurately divided into water environment capacity accounting units based on river topography, hydrological characteristics and pollution source locations; Step 2: Obtain river water body images through satellite remote sensing and preprocess the images to extract information; Step 3: Using the spectral characteristics of remote sensing images, the machine learning algorithm is used to invert the water quality parameters of the river; Step 4: Use remote sensing and GIS technology to identify and quantify the location and intensity of point and non-point source pollution sources within the watershed; Step 5: Based on the water quality parameters and pollution source data obtained by inversion, a water quality prediction model for the river is established using a machine learning algorithm; Step 6: Using the water quality model, calculate the water environment capacity of each river section under different surface water environment quality standards; Step 7: Compare the estimated pollutant emissions with the water environment capacity under each water quality standard to determine whether the capacity is exceeded; Step 8: Use the model to predict future water quality changes and provide early warning information on water quality exceeding the standard.

2. The pollution research and water quality prediction method based on river segment water environment capacity analysis according to claim 1 is characterized in that: Step 1: Step 1.1: Obtain digital elevation model data of the watershed to analyze terrain undulations and water flow direction; collect river network data, obtain land use and land cover data, and understand human activities and natural environment in the area; Step 1.2: Use digital elevation model data to calculate the river's slope, drainage area, and drainage path; determine the main tributaries and confluence points, and identify the river's watershed; Step 1.3: Collect known pollution source data, mark the location of pollution sources on the GIS platform, and form a pollution source distribution map; Step 1.4: Determine the segmentation criteria and set the length range of each river section; Step 1.5: Divide the river into several sections from upstream to downstream according to the established principles, ensuring that each section has relatively uniform geographical and hydrological characteristics; Step 1.6: Consider geographical features or important ecological protection areas and adjust the initially divided river sections to ensure that the pollution source concentration areas are accurately divided; Step 1.7: Check the length, area and number of pollution sources of each river section to ensure a balanced division; Step 1.8: Based on the verification and feedback, the river segmentation scheme is finally determined, and each river segment is assigned a unique number and attribute information for easy management and analysis.

3. The pollution research and water quality prediction method based on river segment water environment capacity analysis according to claim 1 is characterized in that: The specific process of step 2 is as follows: Step 2.1: Determine the spatial resolution and spectral band required for the study area and select satellite data; Step 2.2: Download remote sensing images of the study area from public satellite data platforms, and select clear and cloudless images to avoid cloud and atmospheric influences; Step 2.3: Radiometric Correction: Convert the digital values ​​to radiance or surface reflectance, correcting for differences in sensor response, including: Atmospheric correction: Atmospheric correction algorithm is used to eliminate the effects of atmospheric scattering and absorption; Geometric correction: Correct the spatial geometric distortion of the image so that it accurately matches the geographic coordinate system; Orthorectification: Use DEM data to eliminate the impact of terrain undulations on images and ensure the accuracy of the location of objects; Step 2.4: According to the boundaries of the study area, crop the image to extract the area of ​​interest. If necessary, stitch multiple images together to form a complete coverage area. Step 2.5: Detect the cloud and shadow areas in the image, create a cloud mask, and try to use images from other times to supplement the cloud-covered areas; Step 2.6: Check the preprocessed image for banding, noise or other anomalies to ensure the spectral characteristics of the image are accurate; Step 2.7: Save the processed image data in a standard format and back it up.

4. The pollution research and water quality prediction method based on river segment water environment capacity analysis according to claim 1 is characterized in that: The specific process of step 3 is as follows: Step 3.1: Select the water quality parameters to be inverted according to the research requirements; Step 3.2: Obtain measured water quality data that is synchronized with or close in time to the remote sensing image, ensuring that the location of the measured point is accurate and corresponds to the image pixel; Step 3.3: Extract the multi-band reflectance values ​​corresponding to the measured point positions from the pre-processed remote sensing image, and record the reflectance of each measured point in each band; Step 3.4: Based on the reflectance value, calculate the spectral index related to the water quality parameter and select the index that reflects the target water quality parameter; Step 3.5: Combine the measured water quality parameters with the corresponding spectral reflectance and spectral index to form a training sample set; Step 3.6: Select a machine learning algorithm, train the model using the training dataset, and adjust the parameters to improve the model performance; Step 3.7: Use the cross-validation method to evaluate the predictive ability of the model and calculate the evaluation index; Step 3.8: Apply the trained model to the entire image data, calculate the value of the target water quality parameter pixel by pixel, and generate a spatial distribution map of the water quality parameter; Step 3.9: Analyze the spatial characteristics of the inversion results, identify the high-pollution areas, and use GIS software to create a water quality distribution map to visually display the results; use additional measured data or historical data to verify the accuracy of the inversion results.

5. The pollution research and water quality prediction method based on river segment water environment capacity analysis according to claim 1 is characterized in that: The specific process of step 4 is as follows: Step 4.1: Use remote sensing images to locate possible point sources of pollution and confirm and supplement information based on the list of pollution sources published by the government; Step 4.2: Determine the non-point source pollution area based on land use data and draw the boundary of the non-point source pollution area in GIS; Step 4.3: Obtain detailed information on point sources and collect data on agricultural activities in the area of ​​non-point sources; Step 4.4: Using known emission factors or direct emission data, calculate the pollutant emissions from each point source, taking into account the temporal and spatial variation characteristics of emissions; Step 4.5: Use empirical formulas or models to estimate pollutant emissions from nonpoint source areas; Step 4.6: Organize the information of all pollution sources into a database, including spatial location and attribute data; Step 4.7: On the GIS platform, the pollution source data is associated with the geographic spatial information to generate a pollution source distribution map to show the spatial distribution and intensity of the pollution sources; Step 4.8: Classify the pollution sources according to the type of pollutant; Step 4.9: Check the rationality of the pollution source data, correct errors, and update the database regularly to reflect the latest pollution source information; organize the pollution source data into the format required by the model, including time series and spatial distribution, to ensure that the data matches the subsequent water quality model and capacity calculation.

6. The pollution research and water quality prediction method based on river segment water environment capacity analysis according to claim 1 is characterized in that: The specific process of step 5 is as follows: Step 5.1: Arrange the spatial distribution data of water quality parameters inverted in step 3, collect the pollution source data identified and quantified in step 4, including spatial location and pollutant emissions; obtain hydrological data and meteorological data; Step 5.2: Select a machine learning algorithm based on data characteristics and research needs, consider the interpretability and predictive ability of the model, and ensure that the model is suitable for water quality prediction; Step 5.3: Clean the input data, handle missing values ​​and outliers, ensure data quality, standardize or normalize numerical data, and eliminate the impact of different dimensions; Step 5.4: Use correlation analysis to determine the characteristic variables that are closely related to water quality parameters; Step 5.5: Divide the data set into training set and test set, generally in a ratio of 8:2, to ensure the reliability of model evaluation, the randomness and representativeness of data division, and avoid data leakage; Step 5.6: Use the training set data to input the selected machine learning algorithm, train the water quality prediction model, adjust the model parameters, and use the cross-validation method to optimize the model performance; Step 5.7: Use the test set data to evaluate the prediction accuracy and generalization ability of the model and calculate the evaluation indicators; Step 5.8: According to the evaluation results, adjust the model structure and parameters, and repeatedly train and evaluate until the model achieves satisfactory performance; Step 5.9: Apply the trained model to the entire river area, predict the water quality of each river section, and generate a spatial distribution map of water quality prediction to provide a basis for subsequent analysis.

7. The pollution research and water quality prediction method based on river segment water environment capacity analysis according to claim 1 is characterized in that: The specific process of step 6 is as follows: Step 6.1: Identify the surface water environmental quality standards that the river needs to meet and obtain the pollutant concentration limits under each water quality standard as the basis for calculation; Step 6.2: Based on the definition of water environment capacity, formulate a calculation formula: water environment capacity = (water quality standard limit - upstream water quality background value) × river flow × time; consider the natural attenuation coefficient of pollutants and modify the formula; Step 6.3: Obtain the actual flow data of each river section, ensure the temporal and spatial matching of the data, determine the upstream water quality background value, which requires actual measurement or reference to historical data, determine the attenuation coefficient of pollutants, and refer to literature or experimental data; Step 6.4: For each river section, substitute the corresponding parameters and calculate the water environment capacity under different water quality standards. Considering the changes in different seasons or specific periods, it may be necessary to calculate separately; Step 6.5: Organize the capacity results of each river section under different water quality standards into a table, draw a spatial distribution map of water environment capacity on the GIS platform, and intuitively display the capacity size; Step 6.6: Compare with other research or authoritative data to verify the rationality of the calculation. If there are large differences, check the calculation process and parameters and correct the errors; Step 6.7: Record the calculation steps, formulas, and parameter sources in detail to ensure the repeatability of the results, and save the calculation tables and spatial distribution maps to facilitate subsequent analysis and report writing.

8. The pollution research and water quality prediction method based on river segment water environment capacity analysis according to claim 1 is characterized in that: The specific process of step 7 is as follows: Step 7.1: Collect pollutant emission data from all pollution sources and summarize them by river section to ensure the temporal and spatial consistency of emission data and match the time scale of capacity calculation; Step 7.2: For each river section, calculate the difference between the actual pollutant discharge and the water environment capacity: Difference = actual discharge - water environment capacity; positive values ​​indicate that the discharge exceeds the capacity, and negative values ​​indicate that it is within the capacity range; Step 7.3: Determine the water quality prediction category of the river section based on the difference: if the difference is less than 0, the predicted water quality meets the target standard; if the difference is greater than 0, the predicted water quality does not meet the standard and the main pollution factors need to be identified; Step 7.4: Analyze the types of pollutants that exceed the capacity, determine the pollution factors that have the greatest impact on water quality, and calculate and compare each pollutant separately; Step 7.5: Visualize the difference results on the GIS platform and draw a distribution map of river sections with excessive pollution to intuitively display the spatial location and degree of pollution exceeding the standard; Step 7.6: If there are multiple periods of data, analyze the time trend of pollutant emissions and capacity to determine whether the pollution situation is improving or worsening year by year, and provide a basis for governance; Step 7.7: Compare with the actual water quality monitoring data to verify the accuracy of the prediction results. If the prediction is inconsistent with the actual measurement, analyze the reasons and adjust the model or data; Step 7.8: Organize the results of the comparative analysis and write a report including text descriptions, tables and charts to provide decision-making support for relevant departments and clarify governance priorities.

9. The pollution research and water quality prediction method based on river segment water environment capacity analysis according to claim 1 is characterized in that: The specific process of step 8 is as follows: Step 8.1: Determine the temporal and spatial scope of the forecast and identify the pollutants and river sections that require focus; Step 8.2: Obtain future pollution source emission plans, collect weather forecast data, and consider the impact of extreme weather events on water quality; Step 8.3: Input the future scenario data into the water quality model, ensuring that the data format and units are consistent. For uncertain data, multiple scenarios need to be set for simulation; Step 8.4: Use the updated input data to run the water quality prediction model, simulate future water quality changes, and generate water quality time series data for each river section during the prediction period; Step 8.5: Assess whether water quality is likely to exceed standards in each river section in the future, identify high-risk areas, and analyze the extent and duration of possible pollutant exceedances; Step 8.6: Set the warning level according to the forecast results; Step 8.7: Release warning information to relevant departments and the public, including areas that may be affected and measures to be taken, establish information release channels, ensure that warning information is conveyed in a timely manner, and form a feedback loop.

Images