Multi-index remote sensing monitoring method for water quality of inland water body based on mapping knowledge domain

Abstract

The invention discloses an inland water body water quality multi-index remote sensing monitoring method based on a knowledge graph. The method comprises the following steps: collecting multi-source remote sensing image data, water body appearance and inherent optical characteristic data and historical water quality monitoring data of a target water body area; analyzing a response relationship between the optical characteristics of the water body and water quality components by using the obtained remote sensing reflectivity-optical characteristic-water quality parameter combined data set, and determining key information elements of remote sensing inversion of each water quality parameter of the water body; and constructing an inland water quality multi-index remote sensing inversion knowledge graph, and realizing time-space continuous monitoring of multiple water quality indexes of the water body by utilizing remote sensing images. According to the method, remote sensing monitoring results are complemented by means of a map reasoning technology, so that the completeness of monitoring data is remarkably improved; the defects of the prior art in the aspects of multi-index comprehensive monitoring and knowledge association expression are overcome, and novel technical means and decision support are provided for lake and river ecological environment management.

Description

Technical Field

[0001] The present invention relates to the field of water quality monitoring technology, and in particular to a multi-index remote sensing monitoring method for inland water quality based on a knowledge graph. Background Art

[0002] Against the backdrop of global climate change and increasing human activity, the stability of inland water ecosystems is under serious threat. Water quality monitoring of inland water bodies, such as lakes and rivers, is a critical component of environmental protection and water resources management. Traditional water quality monitoring methods rely heavily on manual sampling and laboratory analysis. While highly accurate, they suffer from long lead times, high costs, and limited coverage, making them inadequate for large-scale, high-frequency, and dynamic monitoring. In recent years, remote sensing technology, owing to its advantages of wide spatial coverage and rapid timeliness, has gained widespread application in water quality monitoring. However, remote sensing monitoring typically captures only a limited number of surface parameters, such as chlorophyll concentration, suspended matter concentration, and water transparency. Furthermore, relying on empirical modeling or statistical regression methods, it is difficult to fully explore the inherent connections and evolutionary patterns between different water quality indicators. Furthermore, most current remote sensing water quality monitoring methods focus on a single indicator and lack the ability to integrate heterogeneous multi-source data and conduct comprehensive analysis of multiple indicators, limiting the systematicity and accuracy of monitoring results.

[0003] In recent years, knowledge graphs, a key research direction in artificial intelligence, have been gradually introduced into environmental science due to their unique advantages in modeling complex relationships and organizing knowledge. However, existing research has primarily focused on tracing the source of water environmental events and identifying pollutants. A systematic knowledge modeling framework for remote sensing water quality monitoring has yet to be established, nor is a mechanism for deeply integrating remote sensing data with multi-metric water quality information. Therefore, a new multi-metric water quality monitoring method combining remote sensing observations with knowledge graphs is urgently needed. Summary of the Invention

[0004] Purpose of the invention: The purpose of the present invention is to provide a multi-index remote sensing monitoring method for inland water quality based on knowledge graph.

[0005] Technical solution: The knowledge graph-based multi-index remote sensing monitoring method for inland water quality of the present invention comprises the following steps: (1) Collect multi-source remote sensing image data, water surface and inherent optical property data, and historical water quality monitoring data of the target water area; (2) Using the acquired remote sensing reflectance-optical properties-water quality parameter joint data set, analyze the relationship between the optical properties of water bodies and the response of water quality components, and determine the key information elements for remote sensing inversion of various water quality parameters; (3) Construct a knowledge map of remote sensing inversion of multiple water quality indicators for inland water bodies, and use remote sensing images to achieve spatiotemporal continuous monitoring of multiple water quality indicators.

[0006] Furthermore, the step (1) includes: (1.1) Obtain and preprocess long-sequence remote sensing image data of the target water body to form a long-sequence remote sensing reflectance image and water body inherent optical property image dataset; (1.2) Obtain data on the surface optical properties and intrinsic optical properties of the target water body; (1.3) Obtain water quality data of target water bodies and conduct data quality control; (1.4) Using the remote sensing reflectance images, water optical property data, and water quality data obtained in steps (1.1)-(1.3), construct a remote sensing reflectance-optical property-water quality parameter joint dataset.

[0007] Furthermore, the preprocessing in step (1.1) includes radiation correction, atmospheric correction, geometric correction, water body extraction and cloud detection.

[0008] Furthermore, the step (2) includes: (2.1) Analyze the intrinsic optical properties of water bodies using the intrinsic optical properties and water quality parameter matching data from the remote sensing reflectance-optical properties-water quality parameter joint dataset; (2.2) Quantitative analysis of the driving mechanisms of the apparent optical properties of water bodies will be conducted using paired data of remote sensing reflectance, apparent optical properties, and water quality parameters from the remote sensing reflectance-optical properties-water quality parameter joint dataset.

[0009] Furthermore, the step (2.1) includes: (2.1.1) Calculate the relative absorption contributions of total suspended particulate matter, pigmented particulate matter, non-pigmented particulate matter, and colored dissolved organic matter in the water to identify the key components that dominate the changes in the optical environment of the water. The calculation formula for the absorption contribution is as follows: , in, Indicates the i Class components at wavelength The contribution rate of the total absorption coefficient; Indicates the wavelength of the corresponding component The absorption coefficient at represents the total absorption coefficient at the corresponding wavelength; (2.1.2) Based on the statistical relationship between the absorption and scattering coefficients of each optical component and water quality indicators, a correlation analysis method is used to quantify its sensitivity. The correlation analysis method is expressed as: , in, r is the correlation coefficient, X represents the explanatory variable, Yrepresents the concentration of lake water components, and are their average values ​​respectively.

[0010] Furthermore, the step (3) includes collecting inland water body optical signals, water quality signals, and inversion model information, extracting key information from inland water body water quality remote sensing inversion, quantitatively characterizing the relationship between water quality elements and key information, and applying the inland water body water quality remote sensing quantitative inversion model.

[0011] Furthermore, the collection of inland water optical signals, water quality signals, and inversion model information includes: (3.1) Remote sensing reflectance feature normalization: extract remote sensing reflectance data matching the target water body from processed multi-source, multi-temporal remote sensing images and standardize and label them by sensor type, time, region, and band; (3.2) Water optical parameter normalization: Merge the apparent optical properties and intrinsic optical properties of the water at the corresponding sampling points to form absorption coefficients and scattering coefficients in different bands and their source material components; (3.3) Consolidation of water quality monitoring indicators: standardization of concentrations of various water quality indicators in measured water quality data and spatiotemporal labeling; (3.4) Integration and annotation of linked data: the three types of information from steps (3.1) to (3.3) are fused and matched based on a unified time-space index to construct a three-dimensional remote sensing reflectance-optical properties-water quality index for knowledge graph modeling. Combine databases and add necessary semantic attribute labels; (3.5) Propose remote sensing inversion knowledge for studying nutrient status indicators of water bodies, including empirical methods, semi-empirical methods, and intelligent inversion algorithms.

[0012] Furthermore, the extraction of key information from remote sensing inversion of inland water quality includes: (3.6) Integrate the multi-source information knowledge collected in steps (1) and (2), select key information as entities in the knowledge graph, and based on the intrinsic relationship between the water quality indicators and the optical properties of the water body obtained in step (2.1), select the remote sensing reflectance, surface optical properties of the water body, inherent optical properties, and concentrations of related water quality elements that have a strong correlation with each water quality indicator as the entities corresponding to each indicator, and clarify their attribute definitions, data types, and semantic boundaries; (3.7) Extract quantitative or qualitative relationships between entities from the results of physical analysis; (3.8) Rule pattern construction: extracting regular descriptions from statistical laws or empirical formulas, defining them as logical rule nodes in the graph for reasoning; (3.9) Extract cross-modal links, establish semantic mapping relationships between different information types, and promote knowledge graphs to support intelligent retrieval and interpretation across data modalities.

[0013] Furthermore, the quantitative characterization of the relationship between the water quality elements and key information includes: For each water quality indicator, the method obtained through the information collection step describes the mathematical relationship between the key information elements of the remote sensing inversion of each water quality parameter and the concentration level of the water quality indicator; all matching data are randomly divided into two groups, 70% are used to train the model, and the remaining 30% are used to verify the application effect of the model, and it is ensured that both the training set and the verification set contain matching data of high, medium and low concentration levels, and the remote sensing inversion accuracy of each water quality indicator is evaluated; time attributes and spatial identifiers are introduced to support the adaptive evolution and fusion of water quality knowledge in different times and regions, and a knowledge graph query and visualization interface is constructed.

[0014] Furthermore, the application of the inland water quality remote sensing quantitative inversion model includes: (3.10) For periods covered by remote sensing imagery, the inversion rule path is constructed by combining the knowledge triples of the response mechanism of remote sensing bands, optical properties, and water quality indicators, the patterns of sensitive bands, and the identification results of the dominant absorption components in the atlas; (3.11) For time periods where remote sensing images are missing, the water quality indicators for the target time period are intelligently supplemented through a graph-based reasoning model, relying on the spatiotemporal patterns embedded in the graph, the evolution path between data from adjacent time periods, and known water quality parameters.

[0015] Beneficial effects: Compared with the existing technology, the present invention has the following significant advantages: the present invention integrates remote sensing images and water quality observation data to construct a knowledge graph, and expresses the association, evolution and influence mechanism between multi-index water quality factors through entity-relationship modeling. On this basis, with the help of graph reasoning technology, the remote sensing monitoring results are supplemented, which significantly improves the completeness of the monitoring data; this method not only makes up for the shortcomings of the existing technology in multi-index comprehensive monitoring and knowledge association expression, but also provides a new technical means and decision-making support for the ecological environment management of lakes and rivers. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a flow chart of the present invention; Figure 2 This is a schematic diagram of remote sensing inversion knowledge of various water quality indicators of inland water bodies. DETAILED DESCRIPTION

[0017] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0018] like Figure 1As shown, the multi-index remote sensing monitoring method for inland water quality based on knowledge graph of the present invention includes the following steps: (1) Collect multi-source remote sensing image data, water surface and inherent optical property data, and historical water quality monitoring data of the target water area; (1.1) Obtain and preprocess long-sequence remote sensing image data of the target water body to form a long-sequence remote sensing reflectance image and water body inherent optical property image dataset; Remote sensing data include multispectral or hyperspectral images, such as Landsat, MODIS, MERIS, OLCI, GOCI, etc., which can be selected according to the spatiotemporal resolution requirements of water quality monitoring and the type of monitoring indicators.

[0019] After acquiring remote sensing image data, the original image needs to be preprocessed, which mainly includes radiation correction, atmospheric correction, geometric correction, water body extraction and cloud shadow removal. Radiation correction converts the digital signal received by the sensor into physical radiation brightness or Rayleigh-corrected reflectance value to ensure the comparability of data obtained at different times or by different sensors.

[0020] Atmospheric correction eliminates the effects of atmospheric scattering and absorption on images. Physical model-based methods (such as the 6S model) or empirical linear methods can be used to form remote sensing reflectance image data that can be used for water quality monitoring.

[0021] Geometric correction is the spatial registration of remote sensing images to ensure that they have consistent spatial positioning accuracy with georeferenced data or other data sources.

[0022] Water body extraction can accurately separate water body areas from remote sensing images through methods such as NDWI (water body index) and MNDWI, providing a clear target area for subsequent water quality inversion.

[0023] Cloud detection, in order to ensure monitoring accuracy, also requires the detection and removal of clouds and cloud shadows. Detection techniques based on thresholds or classification algorithms are often used to ensure that the final image used for analysis has reliable quality and complete coverage.

[0024] After the remote sensing reflectance image data is obtained through preprocessing in the above steps, the semi-analytical model can be used to calculate the inherent optical characteristics of the water body (including the absorption coefficient, scattering coefficient and backscattering coefficient of each water quality component in the water body).

[0025] (1.2) Obtain data on the surface optical properties and intrinsic optical properties of the target water body; The surface optical property data of water bodies, namely the off-water reflectance data, can be measured using the ASSD FieldSpec Pro portable spectrometer from ASSD using the surface method in the 350-1050 nm band.

[0026] Intrinsic optical properties of water include absorption and backscattering coefficients for total suspended particulate matter, pigmented and non-pigmented particles, and the absorption coefficient of colored dissolved organic matter. Absorption coefficients can be measured using quantitative filter technology within the range of 350–800 nm. Backscattering coefficients are measured using the Hydroscat-6P scatterometer produced by HOBI Labs. The Hydroscat-6P has six spectral channels covering the visible to near-infrared range: 442 nm, 488 nm, 532 nm, 590 nm, 767 nm, and 852 nm.

[0027] In order to ensure the temporal and spatial matching between water body optical data and water quality parameter data, water body optical characteristic parameters can be measured synchronously with water quality parameter data.

[0028] (1.3) Obtain water quality data of target water bodies and conduct data quality control; Water quality monitoring data comes from ground monitoring stations or automatic monitoring equipment, including typical water quality parameters such as chlorophyll, suspended matter, transparency, total nitrogen, total phosphorus, etc. that are suitable for remote sensing monitoring.

[0029] After data collection, data quality control is required, primarily including missing value identification, outlier detection, and consistency checks. Missing value identification utilizes time series integrity analysis, while outlier detection combines statistical analysis (such as box plot analysis and the Z-score method) with empirical thresholds to eliminate observations that significantly deviate from the normal range. Consistency checks primarily involve cross-comparisons across multiple sites and indicators to ensure that logical relationships and temporal trends between data are reasonable. For cases where individual data are missing or outliers but are recoverable, temporal interpolation and spatial proximity interpolation methods can be used to complete the data.

[0030] (1.4) Using the remote sensing reflectance images, water optical property data, and water quality data obtained in steps (1.1)-(1.3), construct a remote sensing reflectance-optical property-water quality parameter joint dataset.

[0031] After completing the collection and preprocessing of remote sensing images and water quality monitoring data, spatiotemporal registration operations are required to form remote sensing reflectance data that is consistent with the time and location of ground water quality observation data.

[0032] Specifically, the reflectance values ​​of matching remote sensing imagery pixels are extracted based on the observation time and spatial location of the ground monitoring data. Temporally, the reflectance characteristics are correlated with the water quality status by selecting the remote sensing observation time closest to the water quality sampling date (preferably the same day; if not, high-quality imagery from the three days before and after is used). Spatially, the pixel values ​​at the corresponding location are extracted from the remote sensing imagery using Geographic Information System (GIS) tools based on the latitude and longitude coordinates of the monitoring station. If the station falls on the boundary of multiple pixels, the average value of the surrounding pixels, weighted averaging, or spatial interpolation can be used to enhance the matching accuracy.

[0033] (2) Using the acquired remote sensing reflectance-optical properties-water quality parameter joint data set, analyze the relationship between the optical properties of water bodies and the response of water quality components, and determine the key information elements for remote sensing inversion of various water quality parameters; (2.1) Analyze the inherent optical properties of water bodies using the intrinsic optical properties and water quality parameter matching data in the remote sensing reflectance-optical properties-water quality parameter joint dataset. This includes the following steps: (2.1.1) Calculate the relative absorption contributions of total suspended particulate matter, pigmented particulate matter, non-pigmented particulate matter, and colored dissolved organic matter in the water to identify the key components that dominate the changes in the optical environment of the water. The calculation formula for the absorption contribution is as follows: , in, Indicates the i Class components at wavelength The contribution rate of the total absorption coefficient; Indicates the wavelength of the corresponding component The absorption coefficient at It represents the total absorption coefficient at the corresponding wavelength (i.e. the sum of the absorption coefficients of all components); commonly used sensitive bands include the blue light band (about 440 nm), the green light band (about 550 nm) and the red light band (about 670 nm).

[0034] By performing the above-mentioned ratio calculations in multiple bands, the absorption dominance of each component substance at different spectral positions is analyzed, and combined with the spectral response trend, the dominant optical component that most significantly affects the water surface reflectivity is identified, providing a basis for subsequent model construction and establishment of atlas semantic relationships.

[0035] (2.1.2) Based on the statistical relationship between the absorption and scattering coefficients of each optical component (such as total suspended particulate matter, pigmented particulate matter, non-pigmented particulate matter, and colored dissolved organic matter) and water quality indicators, a correlation analysis method is used to quantify the sensitivity of each component. The correlation analysis method is expressed as: , in, ris the correlation coefficient, X represents the explanatory variable, Y represents the concentration of lake water components, and are their average values ​​respectively. r Between -1 and 1, if r A positive value indicates a positive correlation. r A negative value indicates a negative correlation. It is generally believed that | r When |>0.6, the two groups of variables have a strong correlation, and when 0.4<| r When |<0.6, there is a moderate correlation. r When |<0.4, the correlation is weak.

[0036] (2.1.3) Sort the relative absorption contribution rates of the water components obtained above from high to low, and sort the correlation coefficients between the absorption and scattering coefficients of the water components obtained above and the water quality indicators from strong to weak. Select the inherent optical characteristic parameters with high relative absorption contribution rates and strong correlation coefficients for the subsequent construction of the water quality remote sensing inversion model.

[0037] (2.2) Quantitative analysis of the driving mechanisms of the apparent optical properties of water bodies will be conducted using paired data of remote sensing reflectance, apparent optical properties, and water quality parameters from the remote sensing reflectance-optical properties-water quality parameter joint dataset.

[0038] The following steps are involved: First, a synchronous zonal reduction method was used to screen for representative scenarios using paired samples of water body off-water remote sensing reflectance or remote sensing reflectance and water quality indicators (such as chlorophyll a, total phosphorus, and total nitrogen). This method, based on principal component analysis and statistical clustering techniques, retains representative band combinations and sampling scenarios while eliminating redundant and unstable band information. Then, the off-water reflectance or remote sensing reflectance data in the typical scenarios screened out are visualized as a function curve of the concentration of each water quality indicator (i.e., a band response diagram is drawn with water quality concentration as the horizontal axis and the reflectance of different bands as the vertical axis), so as to intuitively identify the band areas that are most sensitive to changes in the concentration of each water quality indicator.

[0039] Finally, based on the identification of sensitive bands, correlation and regression analyses are conducted between the reflectance of the corresponding bands and various water quality indicators. Correlation analysis can employ methods such as the Pearson correlation coefficient, partial correlation analysis, and mutual information analysis to quantify the intensity of the response between sensitive bands and water quality indicators. Regression analysis can employ methods such as linear regression, polynomial regression, support vector regression, or semi-analytical models based on physical mechanisms to establish a mapping model from reflectance to water quality concentration, thereby elucidating the driving mechanisms of water quality changes driven by remote sensing reflectance characteristics.

[0040] Through the above analysis process, we can systematically identify the sensitive band areas that dominate the surface optical changes of water bodies, clarify the response relationship between the reflectance changes of each band and the changes of water quality indicators, and thus provide data support and physical interpretation for the subsequent entity attribute construction and semantic relationship modeling of "reflectance-water quality indicators" in the water quality remote sensing inversion knowledge graph.

[0041] (3) Construct a knowledge graph for remote sensing inversion of inland water quality indicators, and use remote sensing images to achieve spatiotemporal continuous monitoring of multiple water quality indicators. This mainly includes four steps: collecting information such as inland water optical signals, water quality signals, and inversion models; extracting key information from inland water quality remote sensing inversion; quantitatively characterizing the relationship between water quality elements and key information; and applying the quantitative inversion model for inland water quality remote sensing. Information collection on inland water optical signals, water quality signals, inversion models, etc. includes: According to the key information elements of remote sensing inversion of water quality parameters determined in step (2), the key information elements of remote sensing reflectivity and water body optical characteristics are sorted out as the data basis for conducting water quality remote sensing monitoring.

[0042] (3.1) Remote sensing reflectance feature normalization: extract remote sensing reflectance data matching the target water body from processed multi-source, multi-temporal remote sensing images and standardize and label them by sensor type, time, region, and band; (3.2) Water optical parameter normalization: Merge the apparent optical properties and intrinsic optical properties of the water at the corresponding sampling points to form a data set including absorption coefficients and scattering coefficients in different bands and their source material components (such as pigment particles, non-pigment particles, CDOM, etc.); (3.3) Consolidation of water quality monitoring indicators: standardization of concentrations of various water quality indicators in measured water quality data and spatiotemporal labeling; (3.4) Integration and annotation of linked data: the three types of information from steps (3.1) to (3.3) are fused and matched based on a unified time-space index to construct a three-dimensional remote sensing reflectance-optical properties-water quality index for knowledge graph modeling. Combine databases and add necessary semantic attribute tags (such as water area type, observation time, pollution level, etc.); (3.5)Collect and sort out the existing knowledge in the field of remote sensing inversion of various water quality indicators of inland waters and the field of remote sensing inversion for target waters, including inversion strategies, algorithm forms, function forms, characteristic bands and combination methods, etc. In addition, considering that the inversion methods proposed in existing research generally have poor universality, especially the quantitative inversion of indicators such as chlorophyll a, total nitrogen, and total phosphorus is significantly affected by the differences in water body regional characteristics, it is necessary to propose remote sensing inversion knowledge of trophic state indicators for the studied water body based on exploring the response laws between the optical characteristics of the studied lake water body and water quality components, such as the form of empirical / semi-empirical methods and intelligent inversion algorithms, as Figure 2 shown.

[0043] The extraction of key information for remote sensing inversion of inland water quality includes: (3.6)Integrate and process the multi-source information knowledge collected in steps (1) and (2), screen the key information among them as entities in the knowledge graph, and select the remote sensing reflectance, apparent optical characteristics of the water body (such as water-leaving reflectance), inherent optical characteristics (such as absorption and scattering coefficients of key component substances), and relevant water quality element concentrations (such as considering the influence of chlorophyll concentration when inverting nitrogen and phosphorus concentrations) that have strong correlations with each water quality indicator as the corresponding entities for each indicator, and clarify their attribute definitions, data types, and semantic boundaries; (3.7)Extract the quantitative or qualitative relationships between entities from the physical analysis results; for example, "Band B665 is highly sensitive to changes in chlorophyll a", "CDOM is the dominant absorbing substance affecting the reflectance of blue-green bands", "The scattering coefficient of total suspended particles is positively correlated with the total phosphorus concentration", and convert them into structured triples in the form of <B665, highly correlated, chlorophyll a> or <CDOM, dominant absorption, blue-green bands>; (3.8)Construct rule patterns, extract the regular descriptions in statistical laws or empirical formulas (such as "When the CDOM absorption coefficient is higher than threshold A, the reflectance of the blue band drops significantly"), and define them as logical rule nodes for reasoning in the graph; (3.9)Extract cross-modal links, establish semantic mapping relationships between different information types, such as mapping "band combination features" to "water quality indicator concentrations", and promote the knowledge graph to support intelligent retrieval and interpretation across data modalities.

[0044] The quantitative characterization of the relationship between water quality elements and key information includes: The relationships between entities are quantitatively expressed. For each water quality indicator, analytical / semi-analytical methods such as the band and band ratio algorithm, baseline algorithm, characteristic index algorithm, and segmentation algorithm obtained through the information collection step, as well as intelligent methods such as BP neural network (BP), support vector machine (SVR), and stepwise linear regression (SLR), as well as proposed innovative methods, are used to describe the mathematical relationship between the key information elements of remote sensing inversion of each water quality parameter and the concentration level of the water quality indicator.

[0045] For analytical and semi-analytical methods, parameter regionalization is performed using collected remote sensing reflectance and water optical information data and matching field observations of various water quality indicators. For intelligent methods, all matching data are randomly divided into two groups: 70% is used for model training and the remaining 30% is used to verify the model's application effect. Both the training and validation sets are guaranteed to contain matching data at high, medium, and low concentration levels. Evaluation metrics such as the coefficient of determination (R²), relative root mean square error (RMSE), mean absolute error (MAE), and relative error (MRE) can be used to assess the accuracy of remote sensing inversion of various water quality indicators.

[0046] At the same time, the introduction of time attributes and spatial identifiers supports the adaptive evolution and integration of water quality knowledge at different times and regions. A knowledge graph query and visualization interface is also built to support functions such as graph structure display, path query, and band inversion path tracing.

[0047] Applications of quantitative inversion models for inland water quality remote sensing include: (3.10) For periods covered by remote sensing imagery, the system constructs an inversion rule path by combining the knowledge triples of "remote sensing band - optical properties - water quality index" in the atlas, including the response mechanism, sensitive band patterns, and identification of dominant absorbing components. For example, when inputting remote sensing imagery from a certain period, the system can query the atlas for sensitive bands, matching water body types, and water quality factors. It then uses predefined empirical formulas or intelligent algorithm models in the atlas to perform pixel-by-pixel inversion of water quality indicators and output a spatial distribution map.

[0048] (3.11) For periods of time when remote sensing imagery is missing, the present invention leverages the spatiotemporal patterns embedded in the graph, the evolutionary paths between data from adjacent time periods and known water quality parameters, and uses graph-based reasoning models (such as path reasoning, graph embedding prediction, and rule propagation) to intelligently supplement water quality indicators for the target period. For example, based on the time series attributes of nodes and the "water quality trajectories" of similar historical scenarios, the possible water quality status during periods without remote sensing coverage can be inferred, and the confidence level of the prediction error can be quantified. This invention can significantly enhance the ability to continuously monitor the spatiotemporal water quality of inland lakes, especially during cloudy and rainy seasons when remote sensing images are frequently missing and the frequency of ground-based monitoring is limited.

Claims

1. A multi-index remote sensing monitoring method for inland water quality based on knowledge graph, characterized in that: The steps include: (1) Collect multi-source remote sensing image data, water surface and inherent optical property data, and historical water quality monitoring data of the target water area; (2) Using the acquired remote sensing reflectance-optical properties-water quality parameter joint data set, analyze the relationship between the optical properties of water bodies and the response of water quality components, and determine the key information elements for remote sensing inversion of various water quality parameters; (3) Construct a knowledge map of remote sensing inversion of multiple water quality indicators for inland water bodies, and use remote sensing images to achieve spatiotemporal continuous monitoring of multiple water quality indicators.

2. The multi-index remote sensing monitoring method for inland water quality based on knowledge graph according to claim 1 is characterized in that: The step (1) comprises: (1.1) Obtain and preprocess long-sequence remote sensing image data of the target water body to form a long-sequence remote sensing reflectance image and water body inherent optical property image dataset; (1.2) Obtain data on the surface optical properties and intrinsic optical properties of the target water body; (1.3) Obtain water quality data of target water bodies and perform data quality control; (1.4) Using the remote sensing reflectance images, water optical property data, and water quality data obtained in steps (1.1)-(1.3), construct a remote sensing reflectance-optical property-water quality parameter joint dataset.

3. The multi-index remote sensing monitoring method for inland water quality based on knowledge graph according to claim 1 is characterized in that: The preprocessing in step (1.1) includes radiation correction, atmospheric correction, geometric correction, water body extraction and cloud detection.

4. The multi-index remote sensing monitoring method for inland water quality based on knowledge graph according to claim 1 is characterized in that: The step (2) includes: (2.1) Analyze the intrinsic optical properties of water bodies using the intrinsic optical properties and water quality parameter matching data from the remote sensing reflectance-optical properties-water quality parameter joint dataset; (2.2) Quantitative analysis of the driving mechanisms of the apparent optical properties of water bodies will be conducted using paired data of remote sensing reflectance, apparent optical properties, and water quality parameters from the remote sensing reflectance-optical properties-water quality parameter joint dataset.

5. The method for multi-index remote sensing monitoring of inland water quality based on knowledge graph according to claim 1 is characterized in that: The step (2.1) includes: (2.1.1) Calculate the relative absorption contributions of total suspended particulate matter, pigmented particulate matter, non-pigmented particulate matter, and colored dissolved organic matter in the water to identify the key components that dominate the changes in the optical environment of the water. The calculation formula for the absorption contribution is as follows: ， in, Indicates the i Class components at wavelength The contribution rate of the total absorption coefficient; Indicates the wavelength of the corresponding component The absorption coefficient at represents the total absorption coefficient at the corresponding wavelength; (2.1.2) Based on the statistical relationship between the absorption and scattering coefficients of each optical component and water quality indicators, a correlation analysis method is used to quantify their sensitivity. The correlation analysis method is expressed as: ， in, r is the correlation coefficient, X represents the explanatory variable, Y represents the concentration of lake water components, and are their average values ​​respectively.

6. The method for multi-index remote sensing monitoring of inland water quality based on knowledge graph according to claim 1 is characterized in that: The step (3) includes collecting inland water body optical signals, water quality signals, and inversion model information, extracting key information from inland water body water quality remote sensing inversion, quantitatively characterizing the relationship between water quality elements and key information, and applying the inland water body water quality remote sensing quantitative inversion model.

7. The method for multi-index remote sensing monitoring of inland water quality based on knowledge graph according to claim 6 is characterized in that: The collection of inland water optical signals, water quality signals, and inversion model information includes: (3.1) Remote sensing reflectance feature normalization: extract remote sensing reflectance data matching the target water body from processed multi-source, multi-temporal remote sensing images and standardize and label them by sensor type, time, region, and band; (3.2) Water optical parameter normalization: Merge the apparent optical properties and intrinsic optical properties of the water at the corresponding sampling points to form absorption coefficients and scattering coefficients in different bands and their source material components; (3.3) Consolidation of water quality monitoring indicators: standardization of concentrations of various water quality indicators in measured water quality data and spatiotemporal labeling; (3.4) Integration and annotation of linked data: the three types of information from steps (3.1) to (3.3) are fused and matched based on a unified time-space index to construct a three-dimensional remote sensing reflectance-optical properties-water quality index for knowledge graph modeling. Combine databases and add necessary semantic attribute tags; (3.5) Propose remote sensing inversion knowledge for studying nutrient status indicators of water bodies, including empirical methods, semi-empirical methods, and intelligent inversion algorithms.

8. The method for multi-index remote sensing monitoring of inland water quality based on knowledge graph according to claim 6 is characterized in that: The key information extraction of inland water quality remote sensing inversion includes: (3.6) Integrate the multi-source information knowledge collected in steps (1) and (2), select key information as entities in the knowledge graph, and based on the intrinsic relationship between the water quality indicators and the optical properties of the water body obtained in step (2.1), select the remote sensing reflectance, surface optical properties of the water body, inherent optical properties, and concentrations of related water quality elements that have a strong correlation with each water quality indicator as the entities corresponding to each indicator, and clarify their attribute definitions, data types, and semantic boundaries; (3.7) Extract quantitative or qualitative relationships between entities from the results of physical analysis; (3.8) Rule pattern construction: extracting regular descriptions from statistical laws or empirical formulas, defining them as logical rule nodes in the graph for reasoning; (3.9) Extract cross-modal links, establish semantic mapping relationships between different information types, and promote knowledge graphs to support intelligent retrieval and interpretation across data modalities.

9. The method for multi-index remote sensing monitoring of inland water quality based on knowledge graph according to claim 6 is characterized in that: The quantitative characterization of the relationship between water quality elements and key information includes: For each water quality indicator, the method obtained through the information collection step describes the mathematical relationship between the key information elements of the remote sensing inversion of each water quality parameter and the concentration level of the water quality indicator; all matching data are randomly divided into two groups, 70% are used to train the model, and the remaining 30% are used to verify the application effect of the model, and it is ensured that both the training set and the verification set contain matching data of high, medium and low concentration levels, and the remote sensing inversion accuracy of each water quality indicator is evaluated; time attributes and spatial identifiers are introduced to support the adaptive evolution and fusion of water quality knowledge in different times and regions, and a knowledge graph query and visualization interface is constructed.

10. The method for multi-index remote sensing monitoring of inland water quality based on knowledge graph according to claim 6 is characterized in that: The application of the inland water quality remote sensing quantitative inversion model includes: (3.10) For periods covered by remote sensing imagery, the inversion rule path is constructed by combining the knowledge triples of the response mechanism of remote sensing bands, optical properties, and water quality indicators, the patterns of sensitive bands, and the identification results of the dominant absorption components in the atlas; (3.11) For time periods where remote sensing images are missing, the water quality indicators for the target time period are intelligently supplemented through a graph-based reasoning model, relying on the spatiotemporal patterns embedded in the graph, the evolution path between data from adjacent time periods, and known water quality parameters.

Images