Digital management method for ecological monitoring and investigation of plateau lake

Abstract

The present application relates to the technical field of lake ecological monitoring and digital management, and discloses a digital management method for plateau lake ecological monitoring and investigation, which comprises the following steps: obtaining multi-source ecological monitoring and investigation data of a lake inlet area in a freezing and thawing period and uniformly connecting the data to a digital management platform; dynamically defining the spatial range of a lake inlet mixing zone based on a remote sensing suspended matter concentration spatial gradient field adaptive threshold; decomposing the source pollution concentration components of each sampling point by using a Bayesian Monte Carlo multi-end element mixing model; dynamically updating the anisotropic variation function parameters based on real-time hydrological and meteorological data, and performing a collaborative Kriging spatial interpolation on the source pollution concentration components; and integrating the source pollution spatial distribution and source tracing attribution data to generate monitoring and investigation results and write the results into the management platform.

Description

Technical Field

[0001] This invention relates to the field of lake ecological monitoring and digital management technology, and more specifically, to a digital management method for ecological monitoring and surveying of plateau lakes. Background Technology

[0002] During ecologically sensitive periods such as freeze-thaw cycles, assessing the spatial distribution of water quality and analyzing pollution sources in the mixing zones of inflowing rivers in plateau lakes are crucial aspects of digital management of ecological monitoring surveys. Current digital management processes typically generate a spatial distribution map of the entire lake's water quality using co-kriging interpolation of remote sensing and ground monitoring data. Simultaneously, isotope sampling analysis results are stored as a separate source tracing report on the management platform. These two types of monitoring and survey results are presented separately and then manually compared and interpreted by management personnel.

[0003] The existing technologies described above have the following problems: First, the water quality distribution maps generated by the spatial interpolation module cannot distinguish whether pollutants in the same high-concentration area originate from surface snowmelt runoff channels or freeze-thaw underground runoff infiltration paths, treating all high-value areas equally and leading to misjudgments in pollution flux attribution. Second, isotope source apportionment results are only available at discrete sampling points, and the spacing between sampling points is much larger than the spatial scale of fine gradient changes in the mixing zone, making it impossible to present a continuous spatial differentiation pattern of pollution sources in the management platform. Third, the spatial extent of the mixing zone changes in real time with the inflow into the lake during the freeze-thaw period, and the fixed boundary of the analysis area is not applicable between different monitoring and survey cycles. The above three aspects of information are independent of each other in the digital management process, failing to coordinate the processing of spatial accuracy, source attributes, and dynamic boundaries, thus restricting the level of refined management of ecological monitoring and survey results. Summary of the Invention

[0004] This invention provides a digital management method for ecological monitoring and survey of plateau lakes, which solves the technical problems in related technologies such as the lack of systematic digital management means for the mixing zone of river estuaries flowing into plateau lakes during the freeze-thaw period, and the inability to effectively realize the spatial distribution analysis of pollution concentrations from different sources and the attribution analysis of pollution fluxes.

[0005] This invention discloses a digital management method for ecological monitoring and survey of plateau lakes, including: acquiring satellite remote sensing image reflectance data, measured values ​​of water quality parameters and corresponding coordinates, stable isotope analysis data of water samples, isotope characteristic values ​​of end-member water sources, real-time flow data of rivers flowing into the lake, and wind field data of the lake surface during the freeze-thaw period, and uniformly connecting them to a digital management platform; The concentration of suspended matter in each grid is calculated based on the reflectance data of remote sensing images, and a spatial gradient field of suspended matter concentration is generated. An adaptive threshold is determined according to the cumulative distribution function of the gradient amplitude. Connected regions with gradient amplitudes exceeding the adaptive threshold are defined as dynamic mixing zones at the estuary of the lake. Using the isotope values ​​of water bodies at each sampling point and the isotope characteristic values ​​of the end-member water source as input, the mixing ratio of surface snowmelt runoff and the mixing ratio of freeze-thaw groundwater runoff at each sampling point are solved using a multi-end-member mixing model. The mixing ratio is multiplied by the concentration of water quality parameters to decompose and obtain the pollution concentration component values ​​of surface runoff source and groundwater source at each sampling point. Within the dynamic mixing zone at the estuary of the lake, the pollution concentration components from surface runoff and groundwater runoff are used as the main variables, and the suspended matter concentration field retrieved from remote sensing is used as the covariate. The spatial distribution of pollution concentration from surface runoff and groundwater runoff is generated using the co-kriging interpolation algorithm. By integrating the spatial distribution of pollution concentrations from surface runoff and groundwater runoff, a spatial distribution map of pollution concentrations from different sources and an attribution analysis report of pollution fluxes from different sources are generated and written into the ecological monitoring and survey results database of the digital management platform.

[0006] Furthermore, after integrating multi-source ecological monitoring and survey data into the digital management platform, Z-score standardization was applied to the measured values ​​of water quality parameters, remote sensing image reflectance data, isotope analysis data, real-time flow data of rivers flowing into the lake, and wind field data on the lake surface, so that all types of data could participate in subsequent calculations within a unified dimensionless numerical space. Before being written into the ecological monitoring and survey results database, the standardized data and intermediate calculation results are de-standardized and restored to their original physical dimensions.

[0007] Furthermore, the calculation of suspended matter concentration in each grid cell based on remote sensing image reflectance data and the generation of a spatial gradient field for suspended matter concentration include: calculating the estimated suspended matter concentration of each grid cell using the ratio of near-infrared band reflectance to red band reflectance of the remote sensing image, calculating the gradient amplitude of the spatial distribution of the estimated suspended matter concentration, and generating the spatial gradient field for suspended matter concentration. The gradient magnitudes of all grids in the spatial gradient field of suspended matter concentration are sorted by size. The gradient magnitude corresponding to the cumulative distribution function reaching the preset percentile is taken as the adaptive threshold. Grids with gradient magnitudes exceeding the adaptive threshold are extracted as high gradient regions. Connectivity analysis is performed on the high gradient regions, and the connected regions are defined as the spatial range of the dynamic mixing zone of the estuary at the current monitoring time.

[0008] Furthermore, after defining the spatial range of the dynamic mixing zone at the estuary of the lake, principal component analysis is performed on the spatial gradient field of suspended matter concentration. The principal gradient direction within the dynamic mixing zone at the estuary is extracted as anisotropic direction parameter. The anisotropic direction parameter characterizes the dominant direction of pollutant diffusion after the river runoff flows into the lake, and is used to define the direction of the anisotropic variogram model in the Kriging interpolation algorithm.

[0009] Furthermore, a multi-terminal component mixing model is used to solve for the mixing ratio of surface snowmelt runoff and freeze-thaw groundwater runoff at each sampling point. This includes: taking the oxygen and deuterium isotope values ​​of the water body at each sampling point as input, and the isotopic characteristic values ​​of the surface snowmelt runoff end-member and freeze-thaw groundwater runoff end-member as constraints, and using a Bayesian Monte Carlo multi-terminal component mixing model to solve for the mixing ratio of surface snowmelt runoff and freeze-thaw groundwater runoff at each sampling point. The sum of the mixing ratio of surface snowmelt runoff and the mixing ratio of freeze-thaw groundwater runoff is equal to one, and both belong to a closed interval from zero to one. The expected value of the posterior probability distribution of the mixing ratio of each endmember is taken as the mixing ratio of each endmember at each sampling point.

[0010] Furthermore, after obtaining the pollution concentration components from surface runoff and groundwater at each sampling point, the source is identified in the preset pollution source isotope feature domain map using the dissolved nitrate nitrogen isotope value and dissolved nitrate oxygen isotope value of each sampling point, and the dominant pollution source type label of each sampling point is determined. The preset pollution source isotope feature domain map includes three types of feature intervals: soil organic nitrogen mineralization domain, fertilizer domain, and livestock manure domain. The feature interval into which the dual isotope values ​​of each sampling point fall corresponds to the dominant pollution source type of that sampling point. The dominant pollution source type label is superimposed on the corresponding spatial location on the sub-source pollution concentration spatial distribution map.

[0011] Furthermore, before executing the co-kriging interpolation algorithm, the characteristic length and characteristic width of the estuary runoff diffusion are calculated based on the real-time flow data of the rivers flowing into the lake and the wind field data of the lake surface, including: calculating the average flow velocity of the estuary cross section by dividing the real-time flow of the rivers flowing into the lake by the product of the average water depth of the estuary cross section and the width of the estuary cross section. Based on the average flow velocity and average water depth of the estuary section, the longitudinal diffusion coefficient and the transverse diffusion coefficient are estimated according to Elder's formula. The longitudinal diffusion coefficient is equal to the product of the longitudinal dimensionless empirical coefficient and the average flow velocity and average water depth of the estuary section, and the transverse diffusion coefficient is equal to the product of the transverse dimensionless empirical coefficient and the average flow velocity and average water depth of the estuary section. The mean normalization based on the range is applied to the mean of the average flow velocity at the estuary section, the components of the wind field in each direction on the lake surface, the longitudinal diffusion coefficient, and the transverse diffusion coefficient, and then mapped to a dimensionless numerical space. After normalization, the characteristic length is equal to the longitudinal diffusion coefficient divided by the sum of the average flow velocity at the estuary and the component of the lake surface wind field along the main diffusion direction of the runoff; the characteristic width is equal to the quotient of the transverse diffusion coefficient divided by the sum of the transverse wind field component perpendicular to the main diffusion direction and the smallest positive number, multiplied by the characteristic propagation time obtained by dividing the characteristic length by the sum of the average flow velocity at the estuary and the component of the lake surface wind field along the main diffusion direction of the runoff. The feature length is used as the major axis range parameter of the anisotropic variogram model, and the feature width is used as the minor axis range parameter. Combined with the anisotropic direction parameter, a parameter set of the anisotropic variogram model that is dynamically updated with hydrological and meteorological conditions is generated.

[0012] Furthermore, the co-Kriging interpolation algorithm is executed based on the anisotropic variogram model parameter set, including: inputting the feature length and the feature width as the major axis range and minor axis range of the anisotropic variogram into the variogram fitting process, respectively, and inputting the anisotropic direction parameter as the principal axis direction angle of the variogram. When constructing the Kriging equations, the solver of the co-Kriging interpolation algorithm calculates the spatial weight of each sampling point to the target grid point based on the anisotropic variogram model. Sampling points that are within the feature length range of the target grid point along the anisotropic direction parameter direction receive higher weights, while sampling points that are more than the feature width away from the target grid point perpendicular to the anisotropic direction parameter direction receive lower weights. This makes the interpolation result reflect the directional spatial structure of estuarine runoff diffusion.

[0013] Furthermore, after generating the spatial distribution map of pollution concentration from different sources, the pollution concentrations from surface runoff and groundwater at each grid point are multiplied by the corresponding velocity field components obtained based on real-time flow data of rivers flowing into the lake and wind field data on the lake surface, respectively, to calculate the spatial distribution of pollutant flux density from each source. The flux density of pollutants from various sources in the dynamic mixing zone of the river estuary flowing into the lake is summarized to generate data on the total source pollution flux and spatial hotspot distribution of surface runoff and groundwater runoff. The total pollution flux from different sources and the spatial hotspot distribution data are integrated into the pollution flux attribution analysis report, so that the pollution flux attribution analysis report written into the ecological monitoring survey results database includes a spatial distribution map of pollution concentration from different sources, spatial hotspot distribution of pollution from different sources, labels of the dominant pollution source type for each sampling point, and statistics on pollution flux attribution.

[0014] This invention also provides a digital management system for ecological monitoring and surveying of plateau lakes, used to execute the above-mentioned digital management method, characterized in that it includes: The multi-source data acquisition and access module is used to acquire satellite remote sensing image reflectance data of the river estuary area during the freeze-thaw period, measured values ​​of water quality parameters and corresponding coordinates, stable isotope analysis data of water samples, isotope characteristic values ​​of end-source water sources, real-time flow data of rivers flowing into the lake and wind field data of the lake surface, and to uniformly access the multi-source ecological monitoring and survey data into the digital management platform. The dynamic mixing zone delineation module is used to calculate the suspended matter concentration of each grid based on the remote sensing image reflectance data and generate a spatial gradient field of suspended matter concentration. It determines an adaptive threshold based on the cumulative distribution function of the gradient amplitude and defines the connected regions with gradient amplitudes exceeding the adaptive threshold as the dynamic mixing zone of the lake estuary. The source concentration component decomposition module is used to take the water isotope values ​​and end-member water source isotope characteristic values ​​of each sampling point as input, and use a multi-end-member mixing model to solve the mixing ratio of surface snowmelt runoff and the mixing ratio of freeze-thaw groundwater runoff at each sampling point. The mixing ratio is multiplied by the water quality parameter concentration to decompose and obtain the surface runoff source pollution concentration component value and the groundwater source pollution concentration component value of each sampling point. The co-kriging spatial interpolation module is used to generate the spatial distribution of surface runoff pollution concentration and the spatial distribution of groundwater pollution concentration in the dynamic mixing zone of the river estuary within the lake, using the surface runoff pollution concentration component and the groundwater runoff pollution concentration component as the main variables and the suspended matter concentration field retrieved from remote sensing as the covariate, and employing the co-kriging interpolation algorithm. The results integration and writing module is used to integrate the spatial distribution of pollution concentrations from surface runoff sources and the spatial distribution of pollution concentrations from groundwater sources, generate spatial distribution maps of pollution concentrations from different sources and attribution analysis reports of pollution fluxes from different sources, and write them into the ecological monitoring and survey results database of the digital management platform.

[0015] This invention addresses the technical problem in existing processes where spatial interpolation modules cannot distinguish the pollution contributions from different runoff channels within the same high-concentration region, leading to misattribution of pollution flux, by using the source pollution concentration components obtained from isotope endmember mixing ratio decomposition as independent master variables in co-kriging spatial interpolation. This achieves the technical effect that the spatial distribution map of source pollution concentrations output by the interpolation carries source differentiation information at each grid point. Furthermore, by performing co-kriging interpolation with a high-resolution remote-sensed suspended matter concentration field as a covariate, it solves the technical problem that isotope source apportionment results are only usable at discrete sampling points and cannot cover the fine spatial scale of the mixing area. This achieves the technical effect of presenting a continuous spatial differentiation pattern of source pollution concentrations in the management platform even under sparse sampling conditions. Finally, by dynamically defining the mixing area based on an adaptive threshold of the remote-sensed suspended matter concentration spatial gradient field and using real-time hydrological and meteorological data to drive the dynamic updating of the anisotropic variogram parameters, it solves the technical problem that fixed analysis area boundaries are not applicable between different monitoring and survey cycles. This achieves the technical effect that the analysis area boundaries and interpolation structure parameters automatically adjust with changes in inflow into the lake. Attached Figure Description

[0016] Figure 1 This is a flowchart of the digital management method for ecological monitoring and surveying of plateau lakes provided in this embodiment of the invention; Figure 2 This is a schematic diagram showing the decomposition and comparison of total nitrogen source concentration at each sampling point provided in the embodiments of the present invention; Figure 3 This is a schematic diagram of the mixing ratio of surface snowmelt runoff and freeze-thaw groundwater runoff at each sampling point provided in the embodiments of the present invention; Figure 4 This is an isotope bivariate scatter plot provided in an embodiment of the present invention. and A diagram illustrating the relationship. Figure 5 This is provided by the embodiments of the present invention. and A schematic diagram of a scatter plot for identifying pollution sources; Figure 6 This is a schematic diagram of the distribution of suspended matter concentration gradient amplitude and adaptive threshold provided in an embodiment of the present invention; Figure 7 This is a schematic diagram comparing the interpolation results of typical grid-based total nitrogen concentration from different sources provided in this embodiment of the invention; Figure 8 This is a schematic diagram of the source pollution flux attribution summary provided in an embodiment of the present invention; Figure 9 This is a schematic diagram illustrating the relationship between hydrological and meteorological driving parameters and anisotropic variability function parameters provided in an embodiment of the present invention. Detailed Implementation

[0017] Digital management of ecological monitoring and surveys of plateau lakes requires the integration of multi-source monitoring information, including satellite remote sensing imagery, measured water quality data from fixed stations and mobile surveys, water sample isotope analysis data, and hydrological and meteorological data. During ecologically sensitive periods such as freeze-thaw cycles, it is necessary to conduct refined spatial distribution assessments of water quality and pollution source attribution analyses in the mixing zones of inflowing rivers. Current digital management processes typically generate a spatial distribution map of the entire lake's water quality using co-kriging interpolation of remote sensing and ground monitoring data. Simultaneously, isotope sampling analysis results are stored as a separate source tracing report on the management platform. The two types of monitoring and survey results are presented separately and then manually compared and interpreted by management personnel.

[0018] The following issues hinder refined management in this process: First, while the water quality distribution map generated by the spatial interpolation module can reflect the concentration gradient structure within the mixing zone of the river estuary flowing into the lake, it cannot distinguish whether pollutants in the same high-concentration area originate from surface snowmelt runoff channels or freeze-thaw underground runoff infiltration paths. Treating all high-value areas the same leads to misjudgments in pollution flux attribution in the ecological monitoring and survey report. Second, the isotope source apportionment module can identify the contribution ratio of each water source and the dominant pollution source type at discrete sampling points, but the spacing between sampling points is much larger than the spatial scale of the fine gradient changes in the mixing zone, making it impossible to present the continuous spatial differentiation pattern of pollution sources in the management platform. Third, the spatial extent of the mixing zone changes in real time with the inflow during the freeze-thaw period, and the fixed boundary of the analysis area is not applicable between different monitoring and survey cycles. These three aspects of information are independent of each other in the digital management process, failing to coordinate the processing of spatial accuracy, source attributes, and dynamic boundaries, thus restricting the level of refined management of ecological monitoring and survey results.

[0019] According to an embodiment of this implementation, the main body executing this method is the data processing server of the digital management platform. The data processing server has the computing capabilities of remote sensing image interpretation, geostatistical analysis and isotope mixing calculation, and is connected to satellite remote sensing data interface, fixed monitoring station data acquisition terminal, mobile survey terminal and hydrological and meteorological data service interface through the data aggregation layer.

[0020] At least one embodiment of the present invention discloses a digital management method for ecological monitoring and surveying of plateau lakes, such as... Figure 1 As shown, it includes the following steps: Step 1: Obtain multi-source ecological monitoring and survey data of the river estuary area flowing into the lake during the freeze-thaw period and integrate them into the digital management platform; High-resolution satellite remote sensing images of the river estuary area flowing into the lake during the freeze-thaw period were acquired, along with reflectance data for each band, measured values ​​of water quality parameters and corresponding GPS coordinates of fixed monitoring stations and mobile survey points in the estuary area, stable isotope analysis data of water samples from multiple sampling points in the estuary mixing zone, isotope characteristic values ​​of known end-source water sources, real-time flow data of rivers flowing into the lake, and wind field data of the lake surface. All of the above multi-source ecological monitoring and survey data were then integrated into the data aggregation layer of the digital management platform.

[0021] It should be noted that the stable isotope analysis data of the water samples include the water bodies at each sampling point. Value and Value, and dissolved nitrate Value and Values. Known isotopic characteristic values ​​of end-member water sources include those of surface snowmelt runoff end-members and freeze-thaw groundwater runoff end-members, each at [value missing]. and The characteristic range in space. Measured values ​​of water quality parameters include, but are not limited to, total nitrogen concentration, total phosphorus concentration, and suspended solids concentration.

[0022] After data access, various data are preprocessed: measured values ​​of water quality parameters (total nitrogen concentration, total phosphorus concentration, suspended solids concentration, etc.), remote sensing image reflectance data, and isotope analysis data. , , , The real-time flow data of rivers flowing into the lake and the wind field data of the lake surface were processed using Z-score standardization to eliminate the impact of differences in the dimensions and numerical magnitudes of various physical quantities on the subsequent joint calculation of multi-source data, so that all types of data can participate in subsequent calculations in a unified dimensionless numerical space. The concentration component decomposition, spatial interpolation and flux calculation involved in each subsequent step were all performed in the standardized space. The final output results were de-standardized and restored to the original physical dimensions before being written into the results database.

[0023] Step 2: Define the spatial extent of the dynamic mixing zone at the estuary of the lake based on the remote sensing suspended matter concentration gradient field; Using the ratio of near-infrared to red band reflectance in high-resolution remote sensing images, the estimated suspended matter concentration of each grid is calculated. The gradient amplitude of the spatial distribution of the estimated suspended matter concentration is calculated to generate a spatial gradient field of suspended matter concentration. The gradient amplitudes of all grids in the spatial gradient field of suspended matter concentration are sorted by magnitude, and the gradient amplitude corresponding to the cumulative distribution function reaching a preset percentile is taken as an adaptive threshold. Grids with gradient amplitudes exceeding the adaptive threshold are extracted as high gradient regions. Connectivity analysis is performed on the high gradient regions, and the connected regions are defined as the spatial range of the dynamic mixing zone of the estuary at the current monitoring time, so as to separate the estuary mixing region with significant gradient changes from the open lake surface region with gentle gradient changes.

[0024] Furthermore, in order to incorporate directional information reflecting the diffusion morphology of the dynamic mixing zone in the estuary into subsequent spatial interpolation, after defining the spatial extent of the dynamic mixing zone, principal component analysis was performed on the spatial gradient field of suspended solids concentration to extract the principal gradient direction within the dynamic mixing zone as anisotropic directional parameters. ,in It characterizes the dominant direction of pollutant diffusion after river runoff flows into the lake, and is used to define the direction of the anisotropic variogram model in subsequent steps.

[0025] Step 3: Decompose the source pollution concentration components at each sampling point based on the isotope mixing model; Based on the sampling points of each isotope Value and The values ​​are used as inputs, and the isotopic characteristics of the surface snowmelt runoff endmembers and freeze-thaw subsurface runoff endmembers are used as constraints. The mixing ratio of surface snowmelt runoff at each sampling point is solved using a Bayesian Monte Carlo multi-endmember mixing model. Mixing ratio with freeze-thaw groundwater runoff ,in ,and , The expected value of the posterior probability distribution of the mixing ratio of each endmember is taken as the mixing ratio of each endmember at each sampling point. The mixing ratio of each sampling point is then compared with the measured concentration of the water quality parameter at each sampling point. Multiply and decompose to obtain the surface runoff source pollution concentration component values ​​at each sampling point. and the component values ​​of pollution concentration from underground runoff sources .

[0026] It should be noted that the aforementioned Bayesian Monte Carlo multi-endmember mixture model is a conventional source analysis method for handling isotope endmember uncertainties. The inputs to the Bayesian Monte Carlo multi-endmember mixture model are the water isotope measurements at each sampling point and the distribution range of isotope eigenvalues ​​for each endmember. The output of the Bayesian Monte Carlo multi-endmember mixture model is the posterior probability distribution of the mixing proportions of each endmember. Because... and All are dimensionless mixing ratios. The dimension of is concentration unit, therefore and Dimensions and Consistent, the dimensions of the above decomposition operations are self-consistent.

[0027] Furthermore, in order to further obtain the source type attributes of pollutants at each sampling point, after decomposing and obtaining the concentration components of pollutants from different sources, the dissolved nitrates at each sampling point were also utilized. Value and The values ​​are used to identify the dominant pollution source type label for each sampling point within a pre-defined pollution source isotope feature domain map. This pre-defined map includes three feature intervals: soil organic nitrogen mineralization, fertilizer, and livestock manure. The feature interval into which the dual isotope values ​​of each sampling point fall corresponds to the dominant pollution source type for that point. The dominant pollution source type label, as discrete attribute information, is superimposed on the source spatial distribution map in the final output to aid in attribution analysis.

[0028] Step 4: Perform co-kriging spatial interpolation on the source pollution concentration components within the dynamic mixing zone to generate the spatial distribution of source pollution concentrations; Within the dynamic mixing zone of the river estuary flowing into the lake, the surface runoff source pollution concentration component values ​​at each sampling point were used. A set of main variables, pollution concentration components from freeze-thaw groundwater runoff. Using the suspended matter concentration field retrieved from remote sensing as another set of main variables, and the co-variable as the suspended matter concentration field, the spatial distribution of surface runoff source pollution concentration and groundwater source pollution concentration of each high-resolution grid point in the dynamic mixing zone of the river estuary into the lake is calculated using the co-kriging interpolation algorithm.

[0029] It should be noted that co-kriging interpolation is a conventional geostatistical method that utilizes the spatial cross-correlation between the main variables and covariates to improve interpolation accuracy. The input to co-kriging interpolation is the source pollution concentration components at discrete sampling points and a high-resolution remote-sensed suspended particulate matter concentration field covering the entire dynamic mixing zone of the estuary. The output is the estimated source pollution concentration for each grid point within the dynamic mixing zone of the estuary. Differential spatial correlations exist between remote-sensed suspended particulate matter concentration and the pollution concentration components from surface runoff and groundwater runoff. This differential correlation allows co-kriging interpolation to obtain different spatial distribution patterns when interpolating the two types of source components separately, thus maintaining the physical rationality of source estimation even in grid areas without isotope sampling coverage. Since Z-score standardization was performed on all types of data in step 1, the main variables and covariates are in a unified dimensionless numerical space, eliminating the influence of their inconsistent dimensions on the fitting of the cross-variogram function in co-kriging interpolation.

[0030] Furthermore, to ensure that the variogram parameters of the spatial interpolation reflect the dynamic changes in the diffusion pattern of the dynamic mixing zone at the lake estuary under hydrological and meteorological conditions, before performing co-kriging interpolation, the characteristic length of estuarine runoff diffusion was calculated based on real-time flow data of the inflowing rivers and lake surface wind field data, using empirical relations based on dimensional analysis—derived from the two-dimensional shallow water diffusion equation with appropriate simplification. and feature width It should be noted that the above formula is an empirically simplified model, applicable to estimating the magnitude of mixing and diffusion in shallow estuaries, rather than a rigorous theoretical exact solution. Among these, and The calculation method is as follows: Assume the real-time flow rate of the river flowing into the lake is... The average water depth at the river mouth section is The width of the river mouth section is The average flow velocity at the river mouth section is:

[0031] in, The average flow velocity at the river mouth section. The dimension is volumetric flow rate (V / V). ), The dimension of this quantity is length ( ), The dimension of this quantity is length ( ), The dimension is velocity ( The dimensions on both sides of the formula are consistent.

[0032] Let the component of the lake surface wind field in the main runoff diffusion direction be... The transverse component perpendicular to the main diffusion direction is The longitudinal diffusion coefficient is The transverse diffusion coefficient is Before substituting into subsequent formulas, for , , , , All of these quantities are normalized to mean based on range, mapping them uniformly to a dimensionless numerical space. and Both are dimensionless characteristic scale parameters, comparable within the same dimensionless space, and used to characterize the relative range ratio between the major and minor axes of the variogram. After normalization, the characteristic length... and feature width The following formulas are given respectively:

[0033] in, To prevent the introduction of extremely small positive numbers due to a denominator of zero, all the above quantities are normalized dimensionless quantities. and The dimensions are consistent. and Estimate the values ​​based on the estuary water depth and flow velocity using Elder's formula, i.e. , , and These are the dimensionless empirical coefficients for the longitudinal and transverse diffusion coefficients, respectively. Characterizing the distance of influence of runoff along the main diffusion direction, Characterizes the lateral spread distance of runoff along a direction perpendicular to the main diffusion direction. As the major axis range parameter of the anisotropic variogram model, As a short-axis range parameter, combined with the anisotropic direction parameter extracted in step 2. Generate a set of anisotropic variogram model parameters that are dynamically updated with hydrological and meteorological conditions. ).

[0034] Furthermore, in order to clarify The physical meaning of the time dimension in the formula needs to be explained. expression middle, Represents the characteristic length of runoff particles propagating along the main diffusion direction. Required feature propagation time ,Right now ; Indicates the lateral convection velocity Driven transverse diffusion coefficient The corresponding lateral feature expansion scale; the product of the two That is, during the feature propagation time Within, the runoff's characteristic lateral width, formed by the combined effects of lateral convection and diffusion perpendicular to the main diffusion direction, thus... The calculation explicitly reflects the constraint of the time dimension on the horizontal expansion range.

[0035] The co-kriging interpolation algorithm is based on the anisotropic variogram model parameter set that is dynamically updated with hydrological and meteorological conditions. This allows the spatial structure of the variogram to be adjusted with changes in inflow and wind field conditions in different monitoring and survey periods, thus avoiding interpolation distortion caused by using fixed range parameters.

[0036] Furthermore, in order to clarify the parameter set of the anisotropic variogram model ( The specific role of ) in the co-kriging interpolation algorithm is as follows: after generating the parameter set of the anisotropic variogram model, and The major and minor axis ranges of the anisotropic variogram are input into the variogram fitting process, and... As the input of the principal axis direction angle of the variogram, the spatial correlation differences along the runoff diffusion dominance direction and the vertical direction within the dynamic mixing zone of the estuary are explicitly encoded into the variogram structure. The solver of the co-Kriging interpolation algorithm calculates the spatial weights of each sampling point to the target grid point based on the aforementioned anisotropic variogram model when constructing the Kriging equations, thus ensuring that along the... Directional distance from target grid point Sampling points within the range receive higher weights, while those perpendicular to the range receive lower weights. Directional distance exceeds The weight of the sampling points is reduced accordingly, so that the directional spatial structure of estuarine runoff diffusion can be reflected in the interpolation results.

[0037] Step 5: Integrate the spatial distribution and source attribution data of pollution sources, generate ecological monitoring and survey results, and write them into the digital management platform; The spatial distributions of pollution concentrations from surface runoff and groundwater runoff sources are integrated to generate a high-resolution spatial distribution map of source pollution concentrations in the mixed zone of the estuary during the freeze-thaw period. The dominant pollution source type labels for each sampling point are overlaid onto the corresponding spatial locations on the high-resolution spatial distribution map. The source concentration distribution data for the entire dynamic mixing zone of the estuary are summarized to generate a source pollution flux attribution analysis report, which is then written into the ecological monitoring and survey results database of the digital management platform. Before outputting the final results, all data and intermediate calculation results from step 1 are de-standardized and restored to their original physical dimensions to ensure that all physical quantities in the high-resolution spatial distribution map of source pollution concentrations and the source pollution flux attribution analysis report written into the ecological monitoring and survey results database are presented in their original units.

[0038] Furthermore, to provide more intuitive spatial hotspot information on pollution fluxes within the management platform to support management decisions, after generating high-resolution spatial distribution maps of pollution concentrations from different sources, the pollution concentrations from surface runoff and groundwater sources at each grid point are multiplied by their corresponding velocity field components to calculate the spatial distribution of pollutant flux density from each source. The velocity field components are obtained based on real-time flow data of rivers flowing into the lake and wind field data over the lake surface. Based on the spatial distribution of pollutant flux density from each source, the pollutant flux densities from each source across the entire dynamic mixing zone of the river estuary are summarized to generate total pollution flux from different sources and spatial hotspot distribution data for surface runoff and groundwater runoff. The total pollution flux from different sources and the spatial hotspot distribution data are integrated into the pollution flux attribution analysis report. This ensures that the final pollution flux attribution analysis report included in the ecological monitoring survey results database simultaneously contains high-resolution spatial distribution maps of pollution concentrations from different sources, spatial hotspot distributions of pollution from different sources, dominant pollution source type labels for each sampling point, and attribution statistics for pollution from different sources, forming a complete digital management result for the ecological monitoring survey of the river estuary mixing zone during the freeze-thaw period.

[0039] This implementation method decomposes the measured water quality concentration at isotope sampling points into two concentration components based on endmember mixing ratios: surface runoff source and groundwater runoff source. Then, using each concentration component value as an independent principal variable, co-kriging spatial interpolation is performed within the dynamic mixing zone of the estuary. Therefore, the source attribute information is embedded into the input of the spatial interpolation within the digital management process, so that the high-resolution spatial distribution map of the source pollution concentration output by the interpolation carries source differentiation information at each grid point. This overcomes the limitation of the spatial interpolation module in the existing process, which cannot distinguish the pollution contribution of different runoff channels in the same high-concentration area, thereby avoiding misjudgment of pollution flux attribution caused by treating all high-value areas equally.

[0040] Meanwhile, since the co-kriging interpolation algorithm uses the remote sensing suspended matter concentration field as a covariate and uses the differential spatial cross-correlation between the remote sensing suspended matter concentration and the concentration components of the two sources for joint estimation, it can still present the continuous spatial differentiation pattern of pollution concentrations from different sources in the management platform even when the spacing between isotope sampling points is much larger than the spatial scale of fine gradient changes in the dynamic mixing zone of the estuary. This overcomes the limitation of existing processes where isotope source apportionment results are only available at discrete sampling points and cannot cover fine spatial scales.

[0041] Furthermore, this implementation method dynamically defines the spatial range of the dynamic mixing zone at the estuary of the lake at each monitoring time by using an adaptive threshold of the spatial gradient field of remotely sensed suspended matter concentration. It also uses a two-dimensional shallow water diffusion equation to drive the update of the anisotropic variogram model parameter set based on real-time hydrological and meteorological data. Therefore, the structural parameters of the analysis area boundary and spatial interpolation can be automatically adjusted with the change of inflow during the freeze-thaw period, overcoming the problem of the fixed analysis area boundary failing between different monitoring and survey cycles.

[0042] Therefore, this implementation method integrates spatial accuracy, source attributes, and dynamic boundaries into a digital management process, enabling the spatial distribution map of pollution flux by source and the attribution analysis report of pollution flux by source output in the ecological monitoring and survey results database to have refined spatial resolution and source attribution capabilities, thereby improving the digital management level of ecological monitoring and surveys during the freeze-thaw period of plateau lakes.

[0043] The following is an example of an application of the present invention, such as... Figure 2-9 As shown, the implementation process is as follows: A certain plateau lake (code: Lake Area HY) is located in a plateau region with active freeze-thaw activity. During the freeze-thaw period (March to April 20XX), the melting of snow and ice caused a simultaneous increase in surface runoff and groundwater runoff, resulting in drastic changes in water quality parameters in the estuary mixing zone. A digital management platform (code: Platform DMP) is responsible for the overall management of the ecological monitoring and survey results for this lake area. March 15, 20XX, was selected as the typical monitoring date for this monitoring and survey period. Platform DMP simultaneously accessed satellite remote sensing imagery, fixed-site and mobile survey water quality data, water sample isotope analysis data, and hydrological and meteorological data through a data aggregation layer. The goal was to conduct a refined spatial assessment of pollution concentrations and pollution flux attribution analysis in the estuary mixing zone of River RX flowing into the lake.

[0044] The platform's DMP initiated the data aggregation process at 08:00 on March 15, 20XX. It acquired high-resolution images of the estuary area (near-infrared and red band reflectance raster data) from the satellite remote sensing data interface. It also acquired measured values ​​and GPS coordinates of total nitrogen, total phosphorus, and suspended solids concentrations from eight water quality monitoring points via fixed monitoring station terminals and mobile survey terminals. Finally, it obtained data from six isotope sampling points via the water sample laboratory analysis interface. , , , Analysis values ​​are obtained from the hydrological and meteorological data service interface to obtain the real-time flow rate of the river (RX). Lake surface wind field component , The data also included the isotopic characteristic ranges of surface snowmelt runoff end-units and freeze-thaw groundwater runoff end-units. After all data were integrated, Z-score standardization was performed on water quality parameters, remote sensing band reflectance, isotopic data, flow data, and wind field data to eliminate dimensional differences among the physical quantities.

[0045] Table 1 Examples of Multi-Source Data Access and Z-score Standardization Processing

[0046] Taking the total nitrogen concentration station S1 as an example, the standardized calculation is as follows:

[0047] The platform's DMP uses the ratio of near-infrared to red band reflectance in satellite remote sensing imagery to calculate estimated suspended particulate matter (SPM) concentrations grid-by-grid, generating a raster map of the spatial distribution of SPM concentrations covering the estuary area. Subsequently, it calculates the gradient amplitude field of this spatial distribution, sorts all grid gradient amplitudes, and selects the gradient amplitude corresponding to the 85th percentile of the cumulative distribution function. As an adaptive threshold, grids with gradient magnitudes exceeding this threshold are extracted as high-gradient regions. Connectivity analysis is then used to determine the spatial extent of the estuary dynamic mixing zone at this monitoring time, with an area of ​​approximately [area missing]. Subsequently, principal component analysis was performed on the gradient field within the dynamic mixing region to extract the principal gradient directions and determine the anisotropic direction parameters. (That is, the dominant direction of pollutant diffusion after runoff into the lake is 52° north of east), which is used to define the direction of the subsequent variogram.

[0048] Table 2. Delineation of Suspended Solids Concentration Gradient Field and Dynamic Mixing Zone

[0049] The platform DMP uses 6 isotope sampling points Value and The value is input as the surface snowmelt runoff end-unit ( Feature range: ‰to ‰, Feature range: ‰to ‰) and freeze-thaw underground runoff end element ( Feature range: ‰to ‰, Feature range: ‰to Using ‰ as a constraint, a Bayesian Monte Carlo multi-terminal mixing model is used to solve for the surface snowmelt runoff mixing ratio fs and the freeze-thaw groundwater runoff mixing ratio fg at each sampling point. The expected value of the posterior probability distribution is taken as the final mixing ratio. Taking sampling point P1 as an example, the solution yields fs=0.71 and fg=0.29, satisfying fs+fg=0.71+0.29=1. Multiplying the mixing ratio by the measured total nitrogen concentration C at each sampling point, the pollution concentration components from different sources are obtained:

[0050] At the same time, utilizing each sampling point Value and The values ​​are used to identify the source in the isotopic characteristic domain map of the pollution source and determine the dominant pollution source type label for each sampling point.

[0051] Table 3. Isotope mixing model decomposition results and source type labels

[0052] Before performing co-kriging interpolation, the platform DMP calculates the anisotropic variogram model parameter set based on real-time hydro-meteorological data from the river RX. (Known) Average water depth at the river mouth section River estuary cross-sectional width Calculate the average flow velocity at the river mouth cross section:

[0053] Take empirical coefficient , Estimate the diffusion coefficient using Elder's formula:

[0054] right , , , , After performing mean normalization based on the range, substitute the values ​​into the formula to calculate the dimensionless characteristic scale parameter (denoted as the apostrophe quantity after normalization):

[0055] Combined with the extraction in step 2 The anisotropic variability function model parameter set (major axis variable range) at the time of cost monitoring Short-axis variable range principal axis direction angle ).

[0056] Subsequently, the platform's DMP used the Cs values ​​(total nitrogen concentration component from surface runoff) at 6 sampling points as one set of main variables, the Cg values ​​(total nitrogen concentration component from groundwater runoff) as another set of main variables, and the remotely sensed suspended matter concentration field covering the entire dynamic mixing zone as a covariate, and adjusted the long axis range. Short-axis variable stroke principal axis direction angle The input variogram fitting process constructs an anisotropic variogram model, driving the co-kriging interpolator to calculate the estimated Cs and Cg values ​​for each high-resolution grid point within the dynamic mixing region. Along... The direction is Sampling points within the range receive higher spatial weights, while those perpendicular to this direction exceed... The weights of the sampling points are reduced accordingly, so that the interpolation results reflect the directional spatial structure of runoff diffusion.

[0057] Table 4. Anisotropic variogram model parameters and typical raster interpolation results.

[0058] The platform's DMP integrates the spatial distributions of Cs and Cg from each grid within the fully dynamic mixing zone, and overlays the dominant pollution source type labels (soil organic nitrogen mineralization, livestock manure, and fertilizer) from six sampling points onto their corresponding spatial locations, generating a high-resolution spatial distribution map of source pollution concentrations in the river estuary during the freeze-thaw period. Subsequently, the Cs and Cg values ​​of each grid are multiplied by their corresponding velocity field components to calculate the spatial distribution of pollutant flux densities from each source, summarizing the data for the entire dynamic mixing zone (area). The total pollution flux from each source is calculated, and a source pollution flux attribution analysis report is generated. All intermediate calculation results are de-standardized to their original physical dimensions before being written into the ecological monitoring and survey results database of the platform's DMP.

[0059] Table 5 Summary of Source Pollution Flux Attribution Analysis Reports

[0060] In this example, the data stream starts with the unified access and Z-score standardization of multi-source data in step 1, and then proceeds to step 2 to define the dynamic mixing region (area) by adaptive thresholding of the remote sensing suspended object gradient field. , Step 3 involves using an isotope mixing model to decompose the total nitrogen concentration at each sampling point into two components, Cs and Cg, and adding source type labels. Step 4 then uses real-time hydrological and meteorological data... Dynamic calculation of long axis travel Short-axis variable stroke principal axis direction angle The parameter set drives co-kriging interpolation to extend the discrete components into a high-resolution raster distribution across the entire area. Finally, in step 5, the spatial distribution of the sources, source labels, and flux calculation results are integrated and written into the platform's DMP results library after denormalization. The output data of each step is used as direct input for the next step. The dynamic mixing zone boundary, anisotropy parameters, source concentration components, and interpolation results maintain logical consistency throughout the entire process, ensuring the spatial accuracy and source attribute integrity of the source pollution flux attribution analysis report.

[0061] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A digital management method for ecological monitoring and survey of plateau lakes, characterized in that, Includes the following steps: Acquire satellite remote sensing image reflectance data, measured values ​​of water quality parameters and corresponding coordinates, stable isotope analysis data of water samples, isotope characteristic values ​​of end-source water sources, real-time flow data of rivers flowing into the lake, and wind field data of the lake surface in the freeze-thaw period, and integrate them into the digital management platform. The concentration of suspended matter in each grid is calculated based on the reflectance data of remote sensing images, and a spatial gradient field of suspended matter concentration is generated. An adaptive threshold is determined according to the cumulative distribution function of the gradient amplitude. Connected regions with gradient amplitudes exceeding the adaptive threshold are defined as dynamic mixing zones at the estuary of the lake. Using the oxygen and deuterium isotope values ​​of the water at each sampling point as input, and the isotopic characteristics of the surface snowmelt runoff endmember and the freeze-thaw underground runoff endmember as constraints, the mixing ratio of surface snowmelt runoff and freeze-thaw underground runoff at each sampling point is solved using a Bayesian Monte Carlo multi-endmember mixing model. The sum of the mixing ratio of surface snowmelt runoff and the mixing ratio of freeze-thaw underground runoff is equal to one, and both belong to the closed interval from zero to one. The expected value of the posterior probability distribution of the mixing ratio of each endmember is taken as the mixing ratio of each endmember at each sampling point; the mixing ratio of surface snowmelt runoff and the mixing ratio of freeze-thaw groundwater runoff at each sampling point are multiplied by the water quality parameter concentration, respectively, to decompose and obtain the surface runoff source pollution concentration component value and the groundwater source pollution concentration component value at each sampling point. The pollution concentration component value of the underground runoff source ,in The mixing ratio of surface snowmelt runoff, The mixing ratio of freeze-thawed underground runoff, The concentrations of water quality parameters measured at each sampling point; Within the dynamic mixing zone at the estuary of the lake, the pollution concentration components from surface runoff and groundwater runoff are used as the main variables, and the suspended matter concentration field retrieved from remote sensing is used as the covariate. The spatial distribution of pollution concentration from surface runoff and groundwater runoff is generated using the co-kriging interpolation algorithm. By integrating the spatial distribution of pollution concentrations from surface runoff and groundwater runoff, a spatial distribution map of pollution concentrations from different sources and an attribution analysis report of pollution fluxes from different sources are generated and written into the ecological monitoring and survey results database of the digital management platform.

2. The digital management method for ecological monitoring and survey of plateau lakes according to claim 1, characterized in that, After integrating multi-source ecological monitoring and survey data into a unified digital management platform, Z-score standardization was applied to the measured values ​​of water quality parameters, remote sensing image reflectance data, isotope analysis data, real-time flow data of rivers flowing into the lake, and wind field data on the lake surface, so that all types of data could participate in subsequent calculations within a unified dimensionless numerical space. Before being written into the ecological monitoring and survey results database, the standardized data and intermediate calculation results are de-standardized and restored to their original physical dimensions.

3. The digital management method for ecological monitoring and surveying of plateau lakes according to claim 1, characterized in that, Calculating the suspended matter concentration of each grid cell based on remote sensing image reflectance data and generating a spatial gradient field of suspended matter concentration includes: calculating the estimated suspended matter concentration of each grid cell using the ratio of the near-infrared band reflectance to the red band reflectance of the remote sensing image, calculating the gradient amplitude of the spatial distribution of the estimated suspended matter concentration, and generating the spatial gradient field of suspended matter concentration. The gradient magnitudes of all grids in the spatial gradient field of suspended matter concentration are sorted by size. The gradient magnitude corresponding to the cumulative distribution function reaching the preset percentile is taken as the adaptive threshold. Grids with gradient magnitudes exceeding the adaptive threshold are extracted as high gradient regions. Connectivity analysis is performed on the high gradient regions, and the connected regions are defined as the spatial range of the dynamic mixing zone of the estuary at the current monitoring time.

4. The digital management method for ecological monitoring and surveying of plateau lakes according to claim 3, characterized in that, After defining the spatial range of the dynamic mixing zone at the estuary of the lake, principal component analysis is performed on the spatial gradient field of suspended matter concentration. The principal gradient direction within the dynamic mixing zone at the estuary is extracted as anisotropic direction parameter. The anisotropic direction parameter characterizes the dominant direction of pollutant diffusion after the river runoff flows into the lake, and is used to define the direction of the anisotropic variogram model in the Kriging interpolation algorithm.

5. The digital management method for ecological monitoring and surveying of plateau lakes according to claim 1, characterized in that, After obtaining the pollution concentration components of surface runoff and groundwater at each sampling point, the source is identified in the preset pollution source isotope feature domain map using the dissolved nitrate nitrogen isotope value and dissolved nitrate oxygen isotope value of each sampling point, and the dominant pollution source type label of each sampling point is determined. The preset pollution source isotope feature domain map includes three types of feature intervals: soil organic nitrogen mineralization domain, fertilizer domain, and livestock manure domain. The feature interval into which the dual isotope values ​​of each sampling point fall corresponds to the dominant pollution source type of that sampling point. The dominant pollution source type label is superimposed on the corresponding spatial location on the sub-source pollution concentration spatial distribution map.

6. The digital management method for ecological monitoring and surveying of plateau lakes according to claim 4, characterized in that, Before executing the co-kriging interpolation algorithm, the characteristic length and characteristic width of the estuary runoff diffusion are calculated based on the real-time flow data of the rivers flowing into the lake and the wind field data of the lake surface. This includes: calculating the average flow velocity of the estuary cross section by dividing the real-time flow of the rivers flowing into the lake by the product of the average water depth of the estuary cross section and the width of the estuary cross section. Based on the average flow velocity and average water depth of the estuary section, the longitudinal diffusion coefficient and the transverse diffusion coefficient are estimated according to Elder's formula. The longitudinal diffusion coefficient is equal to the product of the longitudinal dimensionless empirical coefficient and the average flow velocity and average water depth of the estuary section, and the transverse diffusion coefficient is equal to the product of the transverse dimensionless empirical coefficient and the average flow velocity and average water depth of the estuary section. The mean normalization based on the range is applied to the mean of the average flow velocity at the estuary section, the components of the wind field in each direction on the lake surface, the longitudinal diffusion coefficient, and the transverse diffusion coefficient, and then mapped to a dimensionless numerical space. After normalization, the characteristic length is equal to the longitudinal diffusion coefficient divided by the sum of the average flow velocity at the estuary and the component of the lake surface wind field along the main diffusion direction of the runoff; the characteristic width is equal to the quotient of the transverse diffusion coefficient divided by the sum of the transverse wind field component perpendicular to the main diffusion direction and the smallest positive number, multiplied by the characteristic propagation time obtained by dividing the characteristic length by the sum of the average flow velocity at the estuary and the component of the lake surface wind field along the main diffusion direction of the runoff. The feature length and feature width are both dimensionless feature scale parameters based on range normalization. In the normalized coordinate space of co-Kriging interpolation, the feature length is used as the major axis range parameter of the anisotropic variogram model, and the feature width is used as the minor axis range parameter. Combined with the anisotropic direction parameter, a set of anisotropic variogram model parameters that are dynamically updated with hydrological and meteorological conditions is generated.

7. The digital management method for ecological monitoring and surveying of plateau lakes according to claim 6, characterized in that, The co-kriging interpolation algorithm is executed based on the anisotropic variogram model parameter set, including: inputting the feature length and the feature width as the major axis range and minor axis range of the anisotropic variogram into the variogram fitting process, respectively, and inputting the anisotropic direction parameter as the principal axis direction angle of the variogram. When constructing the Kriging equations, the solver of the co-Kriging interpolation algorithm calculates the spatial weight of each sampling point to the target grid point based on the anisotropic variogram model. Sampling points that are within the feature length range of the target grid point along the anisotropic direction parameter direction receive higher weights, while sampling points that are more than the feature width away from the target grid point perpendicular to the anisotropic direction parameter direction receive lower weights. This makes the interpolation result reflect the directional spatial structure of estuarine runoff diffusion.

8. The digital management method for ecological monitoring and survey of plateau lakes according to claim 1, characterized in that, After generating the spatial distribution map of pollutant concentrations from different sources, the surface runoff source pollution concentration and the groundwater runoff source pollution concentration of each grid point are multiplied by the corresponding velocity field component obtained based on the real-time flow data of the rivers flowing into the lake and the wind field data of the lake surface, respectively, to calculate the spatial distribution of pollutant flux density from each source. The flux density of pollutants from various sources in the dynamic mixing zone of the river estuary flowing into the lake is summarized to generate data on the total source pollution flux and spatial hotspot distribution of surface runoff and groundwater runoff. The total pollution flux from different sources and the spatial hotspot distribution data are integrated into the pollution flux attribution analysis report, so that the pollution flux attribution analysis report written into the ecological monitoring survey results database includes a spatial distribution map of pollution concentration from different sources, spatial hotspot distribution of pollution from different sources, labels of the dominant pollution source type for each sampling point, and statistics on pollution flux attribution.

9. A digital management system for ecological monitoring and surveying of plateau lakes, used to execute the digital management method described in any one of claims 1 to 8, characterized in that, include: The multi-source data acquisition and access module is used to acquire satellite remote sensing image reflectance data of the river estuary area during the freeze-thaw period, measured values ​​of water quality parameters and corresponding coordinates, stable isotope analysis data of water samples, isotope characteristic values ​​of end-source water sources, real-time flow data of rivers flowing into the lake and wind field data of the lake surface, and to uniformly access the multi-source ecological monitoring and survey data into the digital management platform. The dynamic mixing zone delineation module is used to calculate the suspended matter concentration of each grid based on the remote sensing image reflectance data and generate a spatial gradient field of suspended matter concentration. It determines an adaptive threshold based on the cumulative distribution function of the gradient amplitude and defines the connected regions with gradient amplitudes exceeding the adaptive threshold as the dynamic mixing zone of the lake estuary. The source concentration component decomposition module takes the oxygen and deuterium isotope values ​​of the water at each sampling point as input, and uses the isotopic characteristics of the surface snowmelt runoff endmembers and freeze-thaw groundwater runoff endmembers as constraints. It uses a Bayesian Monte Carlo multi-endmember mixing model to solve for the mixing ratios of surface snowmelt runoff and freeze-thaw groundwater runoff at each sampling point. The sum of the mixing ratios of surface snowmelt runoff and freeze-thaw groundwater runoff is equal to one, and both belong to a closed interval between zero and one. The expected value of the posterior probability distribution of the mixing ratio of each endmember is taken as the mixing ratio of each endmember at each sampling point. The mixing ratios of surface snowmelt runoff and freeze-thaw groundwater runoff at each sampling point are multiplied by the water quality parameter concentrations, respectively, to decompose and obtain the surface runoff source pollution concentration component value and the groundwater source pollution concentration component value at each sampling point. The pollution concentration component value of the underground runoff source ,in The mixing ratio of surface snowmelt runoff, The mixing ratio of freeze-thawed underground runoff, The concentrations of water quality parameters measured at each sampling point; The co-kriging spatial interpolation module is used to generate the spatial distribution of surface runoff pollution concentration and the spatial distribution of groundwater pollution concentration in the dynamic mixing zone of the river estuary within the lake, using the surface runoff pollution concentration component and the groundwater runoff pollution concentration component as the main variables and the suspended matter concentration field retrieved from remote sensing as the covariate, and employing the co-kriging interpolation algorithm. The results integration and writing module is used to integrate the spatial distribution of pollution concentrations from surface runoff sources and the spatial distribution of pollution concentrations from groundwater sources, generate spatial distribution maps of pollution concentrations from different sources and attribution analysis reports of pollution fluxes from different sources, and write them into the ecological monitoring and survey results database of the digital management platform.

Images