A surface water environment water quality monitoring operation and maintenance management method and system

Abstract

This invention relates to the field of water quality monitoring technology and discloses a method and system for the operation and maintenance management of surface water environmental water quality monitoring. The method includes steps S101, calculating the shore-to-shore near-infrared residual offset; step S102, calculating the reconstituted equivalent brightness on both banks; step S103, subtracting the original reflection to obtain corrected reflection; step S104, calculating the water quality evidence field; step S105, calculating segment-level single values; and step S106, outputting panoramic indicators as water quality monitoring results. This invention uses the cross-sectional geometry of the river channel as a driving force, transforming the asymmetric non-intrinsic reflection of the near-infrared channel into quantifiable radiation parameters. Through a physical heuristic model, it achieves the orderly transmission and accurate subtraction of this interference from near-infrared to inversion bands such as blue, green, and red, effectively eliminating directional systematic biases in chlorophyll and turbidity inversion. This improves the spatiotemporal comparability and judgment stability of monitoring results, adapting to the precise monitoring needs of complex scenarios in narrow river channels.

Description

Technical Field

[0001] This invention relates to the field of water quality monitoring technology, and more specifically, to a method and system for the operation and maintenance management of surface water environmental water quality monitoring. Background Technology

[0002] Surface water environmental water quality monitoring is widely using multi-source image data such as shore-based cameras and drones to estimate key indicators such as chlorophyll and turbidity. The conventional process includes atmospheric correction and empirical inversion models, which are mature in wide water areas such as lakes and nearshore seas, but have obvious limitations in narrow river scenarios such as cities and valleys.

[0003] In narrow river channels, the width of the river is close to the spatial resolution of the image, as well as the scale of influence from the atmosphere and optical systems. Radiation from the highly reflective land on both banks leaks to the water surface pixels through atmospheric scattering and point diffusion of the imaging system, causing eigenvalue reflections in the near-infrared channel of the water body. When there is a difference in brightness between the two banks, the radiation spillover from the land area along the normal cross-section consists of two attenuating components that decrease from both banks towards the interior of the water surface. Because the river width and the scale of influence are close, these components overlap significantly on the cross-section, causing the location of the near-infrared minimum value to deviate from the geometric midline, with the offset direction moving away from the brighter bank. Factors such as automatic exposure and white balance variations, thin clouds and cloud shadows, and seasonal differences in shoreline vegetation further amplify this asymmetric spillover, causing the blue, green, and red bands used for inversion to inherit and strengthen the eigenvalue components.

[0004] Existing proximity correction and inversion methods are mostly designed for open water surfaces, relying on fixed parameters or external atmospheric information. They cannot adaptively handle the cross-sectional geometry and brightness differences between the banks of narrow channels. Cross-band propagation is not geometrically constrained, and non-intrinsic components are introduced into the inversion input of chlorophyll and turbidity, resulting in a directional systematic bias from near-infrared to visible bands and then to the inversion results. At the same time, the selection of sampling and alarm locations on the operation and maintenance side lacks quantitative basis for cross-sectional asymmetric pollution, resulting in poor spatiotemporal comparability of monitoring results and unstable judgments, making it difficult to meet the needs of accurate water quality monitoring in narrow channels. Summary of the Invention

[0005] This invention provides a method and system for monitoring, operating and managing surface water environmental water quality, thereby solving the technical problems mentioned in the background.

[0006] This invention provides a method for monitoring, operating, and managing surface water environmental water quality, comprising the following steps: Step S101: Obtain the water body mask, the edge lines on both banks and the normal cross section. Sample near-infrared reflection based on the normal cross section to determine the geometric centerline position and the minimum near-infrared position, and calculate the shore-directed near-infrared residual offset. Step S102: Obtain the near-infrared equivalent brightness of both banks on the buffer zone between the two banks, establish a dual-bank convolution and baseline model to obtain the baseline term and attenuation scale, determine the target asymmetry ratio by combining the bank-to-bank near-infrared residual offset, and calculate the re-normalized equivalent brightness of both banks under the condition that the total brightness of both banks remains unchanged. Step S103: Calculate the cross-band average reflection to obtain the propagation coefficient, generate the neighboring residual based on the equivalent brightness and attenuation scale of the renormalized banks, obtain the baseline residual by weighting the propagation baseline term according to the bank spectrum, superimpose to obtain the total residual field, and subtract the original reflection to obtain the corrected reflection. Step S104: Construct algal color proxy and turbidity proxy based on corrected reflection, calculate segment-level energy statistics to obtain adaptive weights, and fuse the three to obtain a water quality evidence field. Step S105: Calculate the normalized value of the adjacent risk based on the total residual field, and calculate the bank-directed geometric risk coefficient by combining the bank-directed near-infrared residual offset and the river section width. Based on this, generate the pixel reliability weight and perform robust aggregation of the water quality evidence field to obtain segment-level single values. Step S106: Weighted aggregation of segment-level individual values ​​for each segment according to segment length weight, and output panoramic indicators as water quality monitoring results.

[0007] Furthermore, a water mask is obtained from the input image. The water mask is a set of image pixels. A pixel value of 1 indicates that the corresponding area is water, and a pixel value of 0 indicates that the corresponding area is land. The bank edges are extracted from the input image. The bank edges are broken lines that represent the boundaries between the water body and the land. A normal cross section is constructed along the river centerline parameters. The normal cross section is perpendicular to the tangent of the river centerline, and its two ends intersect the bank edges. The width of the river segment is determined between the two intersection points of the normal cross section and the bank edges. The width of the river segment is the length of the line segment of the normal cross section located inside the water mask. On the normal transverse line, the sampling interval is defined with the intersection of the normal transverse line and the left bank edge as the starting point and the intersection of the normal transverse line and the right bank edge as the ending point; the reflectance of the near-infrared band is collected only at the pixel position with a pixel value of 1 in the water mask along the sampling interval. The reflectance of the near-infrared band is the water-free reflectance after atmospheric correction. Based on the determined width of the river segment, calculate the position of the geometric centerline, which is the cross-sectional coordinate corresponding to the midpoint of the river segment width; Within the sampling interval, the position with the lowest reflectivity in the near-infrared band is identified as the minimum near-infrared position. Subtract the cross-sectional coordinates corresponding to the geometric midline position from the cross-sectional coordinates corresponding to the minimum near-infrared position to obtain the shore-side near-infrared residual offset.

[0008] Furthermore, a left bank buffer zone is formed by extending a set width along the left bank edge into the land area, and a right bank buffer zone is formed by extending a set width along the right bank edge into the land area. Both the left and right bank buffer zones contain only land area pixels. The average values ​​of the near-infrared reflectance of all pixels in the left bank buffer zone and the near-infrared reflectance of all pixels in the right bank buffer zone are calculated. The two average values ​​are used as the near-infrared equivalent brightness of the left bank and the near-infrared equivalent brightness of the right bank, respectively. The near-infrared reflectance is the water-leaving reflectance after atmospheric correction.

[0009] Furthermore, using the near-infrared reflectance collected along the normal cross-section as the fitting object, a dual-bank convolution and baseline model is established and fitted to obtain the baseline term and attenuation scale. The dual-bank convolution and baseline model includes the left bank near-infrared equivalent brightness, the right bank near-infrared equivalent brightness, the cross-sectional coordinates along the normal cross-section, the river segment width, the baseline term, and the attenuation scale. Based on the shore-to-near-infrared residual offset and attenuation scale, the target asymmetry ratio is obtained through exponential calculation; the equivalent brightness of the left shore near-infrared is added to the equivalent brightness of the right shore near-infrared to obtain the total brightness of both shores; keeping the value of the total brightness of both shores unchanged, the total brightness of both shores is proportionally allocated according to the target asymmetry ratio to obtain the re-normalized equivalent brightness of the left shore and the re-normalized equivalent brightness of the right shore respectively.

[0010] Furthermore, a target band set is determined. For each band in the target band set, the average value of the original reflection of the band within the left bank buffer zone and the average value of the original reflection of the band within the right bank buffer zone are calculated. These two average values ​​are used as the left bank cross-band average reflection and the right bank cross-band average reflection of the band, respectively. The left bank cross-band average reflection of each band is divided by the average reflection corresponding to the left bank near-infrared equivalent brightness, and the right bank cross-band average reflection of the band is divided by the average reflection corresponding to the right bank near-infrared equivalent brightness to obtain the left bank propagation coefficient and the right bank propagation coefficient of the band. Using the width range of the river segment from the left bank to the right bank on the normal cross section as the calculation range, for each band in the target band set, combined with the left bank propagation coefficient and the right bank propagation coefficient of the band, the proximity residual of the band at each position on the normal cross section is calculated according to the attenuation law of the distance from the bank.

[0011] Furthermore, for each band in the target band set, the left bank propagation coefficient of that band is multiplied by the renormalized left bank equivalent brightness, and the right bank propagation coefficient of that band is multiplied by the renormalized right bank equivalent brightness. The two products are then added together. The sum is divided by the sum of the renormalized left bank equivalent brightness and the renormalized right bank equivalent brightness to obtain the shore spectral weighting coefficient of that band. The baseline term is multiplied by the shore spectral weighting coefficient of that band to obtain the baseline residual of that band. For each band in the target band set, the neighboring residuals of that band are added to the baseline residuals to obtain the total residual field of that band; The corrected reflection for this band is obtained by subtracting the corresponding pixel value in the total residual field from the original reflection pixel value. If the subtraction result is less than zero, it is taken as zero.

[0012] Furthermore, green band corrected reflection and blue band corrected reflection are selected from the corrected reflection. The pixel value of green band corrected reflection is divided by the sum of the pixel value of blue band corrected reflection and the smallest positive number to obtain the algae color proxy. Select the red band corrected reflection and the green band corrected reflection from the corrected reflection, and divide the pixel value of the red band corrected reflection by the sum of the pixel value of the green band corrected reflection and the smallest positive number to obtain the turbidity proxy. Along the normal cross section, calculate the difference between the corrected reflection pixel value of the green band and the corrected reflection pixel value of the blue band, and the difference between the corrected reflection pixel value of the red band and the corrected reflection pixel value of the green band. Take the non-negative values ​​of the two differences, and then calculate the median of the two non-negative value sets. The two medians are used as the segment-level energy statistics value of algal color correlation and the segment-level energy statistics value of turbidity correlation, respectively. Add the energy statistics of the algal color-related segment to the energy statistics of the turbidity-related segment, and then add the smallest positive number to obtain the denominator for weight calculation; divide the energy statistics of the algal color-related segment by the denominator for weight calculation to obtain the adaptive weight; Multiply the adaptive weight by the algal color proxy, then subtract the adaptive weight from 1 and multiply by the turbidity proxy. Add the two products together to obtain the water quality evidence field.

[0013] Further, the total residual of the near-infrared channel is selected from the total residual field, and the median of the total residual of the near-infrared channel corresponding to all pixels on the normal cross section is obtained; the total residual value of the near-infrared channel of each pixel on the normal cross section is divided by the median to obtain the normalized value of the neighbor risk corresponding to each pixel. Take the absolute value of the near-infrared residual offset along the shore, and divide this absolute value by the width of the river section to obtain the geometric risk coefficient along the shore. For each pixel, the normalized value of the neighbor risk is negative, and then the negative result is exponentially calculated. First, the shore geometric risk coefficient is subtracted from 1, and then the result of the exponential calculation is multiplied by the difference to obtain the pixel reliability weight corresponding to each pixel. On the normal cross section, the value of each pixel in the water quality evidence field is multiplied by the pixel reliability weight of the corresponding pixel to obtain a set of product values. The median of the set of product values ​​is then used as the segment-level single value.

[0014] Furthermore, a corresponding segment length weight is determined for each normal cross section within the same river segment, with the unit of segment length weight being meters; Calculate the sum of the segment length weights corresponding to all normal cross-sections within the same river segment; Divide the segment length weight corresponding to each normal transverse line by the sum of the segment length weights to obtain the normalized weight corresponding to each normal transverse line. Multiply the normalized weight corresponding to each normal cross section by the segment-level single value corresponding to that normal cross section to obtain the weighted contribution value corresponding to each normal cross section. The weighted contribution values ​​corresponding to all normal cross sections are summed to obtain the panoramic index as the water quality monitoring result.

[0015] This invention provides a surface water environment water quality monitoring and operation management system, comprising: The infrared residual offset output module is used to acquire the water body mask, the shoreline and the normal cross section. It samples the near-infrared reflection based on the normal cross section to determine the geometric centerline position and the minimum near-infrared position, and calculates the shore-directed near-infrared residual offset. The equivalent brightness output module for both sides is used to obtain the near-infrared equivalent brightness of both sides on the buffer zone between the two sides, establish a dual-side convolution and baseline model to obtain the baseline term and attenuation scale, combine the shore-to-shore near-infrared residual offset to determine the target asymmetry ratio, and calculate the renormalized equivalent brightness of both sides under the condition that the total brightness of both sides remains unchanged. The corrected reflection output module is used to calculate the cross-band average reflection to obtain the propagation coefficient, generate the neighboring residual based on the equivalent brightness and attenuation scale of the renormalized banks, obtain the baseline residual by weighting the propagation baseline term according to the bank spectrum, and obtain the total residual field by superposition. The original reflection is subtracted to obtain the corrected reflection. The water quality evidence field output module is used to construct algal color proxy and turbidity proxy based on corrected reflection, calculate segment-level energy statistics to obtain adaptive weights, and fuse the three to obtain the water quality evidence field. The segment-level single-value output module is used to calculate the neighboring risk normalization value based on the total residual field, combine the shore-side near-infrared residual offset with the river segment width to calculate the shore-side geometric risk coefficient, generate the pixel reliability weight, and robustly converge the water quality evidence field to obtain the segment-level single value. The panoramic index output module is used to weight and aggregate the segment-level single values ​​of each segment according to the segment length weight, and output the panoramic index as the water quality monitoring result.

[0016] The beneficial effects of this invention are as follows: Driven by the cross-sectional geometry of the river channel, this invention transforms the asymmetric non-intrinsic reflection of the near-infrared channel into quantifiable radiation parameters. Through a physical heuristic model, it achieves the orderly transmission and accurate subtraction of this interference from the near-infrared to the blue, green, and red inversion bands, effectively eliminating the directional systematic bias in chlorophyll and turbidity inversion. Simultaneously, by leveraging pixel reliability weighting and adaptive weighting of algal color and turbidity, it ensures that the water quality evidence field and segment-level single values ​​truly reflect the intrinsic condition of the water body. The final aggregated panoramic single indicator integrates information from the entire river section, providing clear quantitative basis for sampling location selection and alarm determination on the operation and maintenance side, thereby improving the spatiotemporal comparability and judgment stability of monitoring results, and adapting to the precise monitoring needs of complex scenarios in narrow rivers. Attached Figure Description

[0017] Figure 1 This is a flowchart of a surface water environment water quality monitoring, operation and maintenance management method according to the present invention; Figure 2 This is a schematic diagram of a surface water environment water quality monitoring and operation management system according to the present invention.

[0018] In the figure: Infrared residual offset output module 201, equivalent brightness output module for both banks 202, corrected reflection output module 203, water quality evidence field output module 204, segment-level single value output module 205, panoramic index output module 206. Detailed Implementation

[0019] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0020] It should be noted that, unless otherwise defined, the technical or scientific terms used in one or more embodiments of the present invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in one or more embodiments of the present invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" indicate that the element or object preceding the term encompasses the elements or objects listed following the term and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0021] like Figures 1-2 As shown, a method for monitoring, operating, and managing surface water environmental quality includes the following steps: Step S101: Obtain the water body mask, the edge lines on both banks and the normal cross section. Sample near-infrared reflection based on the normal cross section to determine the geometric centerline position and the minimum near-infrared position, and calculate the shore-directed near-infrared residual offset. Step S102: Obtain the near-infrared equivalent brightness of both banks on the buffer zone between the two banks, establish a dual-bank convolution and baseline model to obtain the baseline term and attenuation scale, determine the target asymmetry ratio by combining the bank-to-bank near-infrared residual offset, and calculate the re-normalized equivalent brightness of both banks under the condition that the total brightness of both banks remains unchanged. Step S103: Calculate the cross-band average reflection to obtain the propagation coefficient, generate the neighboring residual based on the equivalent brightness and attenuation scale of the renormalized banks, obtain the baseline residual by weighting the propagation baseline term according to the bank spectrum, superimpose to obtain the total residual field, and subtract the original reflection to obtain the corrected reflection. Step S104: Construct algal color proxy and turbidity proxy based on corrected reflection, calculate segment-level energy statistics to obtain adaptive weights, and fuse the three to obtain a water quality evidence field. Step S105: Calculate the normalized value of the adjacent risk based on the total residual field, and calculate the bank-directed geometric risk coefficient by combining the bank-directed near-infrared residual offset and the river section width. Based on this, generate the pixel reliability weight and perform robust aggregation of the water quality evidence field to obtain segment-level single values. Step S106: Weighted aggregation of segment-level individual values ​​for each segment according to segment length weight, and output panoramic indicators as water quality monitoring results.

[0022] In one embodiment of the present invention, a water body mask is obtained on the input image. The water body mask is a set of image pixels. When the pixel value is 1, it indicates that the corresponding area of ​​the pixel is water, and when the pixel value is 0, it indicates that the corresponding area of ​​the pixel is land. The bank edges are extracted from the input image. The bank edges are broken lines that define the boundary between the water body and the land. A normal cross section is constructed along the river centerline parameters. The normal cross section is perpendicular to the tangent of the river centerline, and the two ends of the normal cross section intersect the bank edges respectively. The river segment width is determined between the two intersection points of the normal cross section and the bank edges. The river segment width is the length of the line segment of the normal cross section located inside the water body mask. On the normal transverse line, the sampling interval is defined with the intersection of the normal transverse line and the left bank edge as the starting point and the intersection of the normal transverse line and the right bank edge as the ending point; the reflectance of the near-infrared band is collected only at the pixel position with a pixel value of 1 in the water mask along the sampling interval. The reflectance of the near-infrared band is the water-free reflectance after atmospheric correction. Based on the determined width of the river segment, calculate the position of the geometric centerline, which is the cross-sectional coordinate corresponding to the midpoint of the river segment width; Within the sampling interval, the position with the lowest reflectivity in the near-infrared band is identified as the minimum near-infrared position. Subtract the cross-sectional coordinates corresponding to the geometric midline position from the cross-sectional coordinates corresponding to the minimum near-infrared position to obtain the shore-side near-infrared residual offset.

[0023] It should be noted that the input image is a remote sensing image used to acquire surface water environment information. It must contain near-infrared band data, and the spatial resolution of the image must meet the accuracy requirements for water body mask extraction and shoreline extraction. For example, the resolution should not be less than 10 meters, and the specific resolution should be determined according to the width of the monitored river section. The radiometric accuracy must support the subsequent quantitative analysis of near-infrared reflectance, and the image must be preprocessed, such as radiometric calibration and geometric correction, to ensure the accuracy of the location of ground objects and reflectance data. The water body mask is a set of pixels extracted from the input image through image processing technology to distinguish between water and land areas. Its function is to limit the range of subsequent near-infrared reflectance sampling, and to sample only in water areas with a pixel value of 1. The image processing technology can be based on the near-infrared band thresholding method. Since water has low reflectance and land has high reflectance in the near-infrared band, a reflectance threshold is set, and pixels below the threshold are marked as water.

[0024] It should be noted that the shoreline is represented by a polyline of the water-land boundary in the input image. The left shoreline is the boundary between the left side of the water body and the land, and the right shoreline is the boundary between the right side of the water body and the land. The shoreline extraction is achieved through edge detection algorithms, such as the Canny edge detection algorithm. First, the boundary pixels between the water body and the land are detected, and then the boundary pixels are connected in sequence to form a polyline. The inflection point of the polyline needs to be determined according to the boundary direction change to ensure that the polyline can accurately reflect the actual shoreline shape, thus providing a boundary basis for the construction of the normal cross-section and the delineation of the sampling interval. In addition, when extracting the shoreline, if there is vegetation cover or building occlusion in the image that causes the boundary to be blurred, the image needs to be denoised first (such as Gaussian filtering), and then the boundary is optimized through morphological operations (such as dilation and erosion) to ensure that the extracted shoreline is continuous and accurate. The polyline processing of the shoreline needs to set an inflection point determination threshold. For example, when the angle between the line connecting adjacent boundary pixels and the overall boundary direction exceeds 15 degrees, it is determined to be an inflection point to avoid the polyline being too complex or simplified. This will not be elaborated here.

[0025] It should be noted that the river centerline parameters are used to mark different locations of the river centerline. Each parameter value corresponds to a feature point on the centerline. The parameter values ​​must cover the entire monitored river section and can be evenly distributed according to the length of the centerline. For example, a parameter value can be set every 50 meters. The specific interval is determined according to the length of the river section and the image resolution to ensure that the normal cross-section corresponding to adjacent parameters can completely cover the river. Specifically, the river centerline is obtained by using the distance transformation skeleton extraction method, that is, calculating the distance from each pixel in the river area to the water boundary (both banks), selecting the pixel with the largest distance, and connecting these pixels according to the river direction to form the river centerline. Then, the centerline is divided into segments from the start end to the end end according to the length. The endpoint of each segment is a feature point, and the index of each feature point or the distance to the start end is the corresponding river centerline parameter.

[0026] It should be noted that the normal cross-section represents the line segment corresponding to each channel centerline parameter (feature point). The direction of the line segment is perpendicular to the tangent of the channel centerline at that feature point. Both ends of the line segment need to extend to intersect with the bank edges to ensure that the line segment can cover the entire channel cross-section. First, form two line segments with the feature point and the previous and next feature points, and calculate the average slope of the two line segments. The direction corresponding to this slope is the tangent. Then, draw line segments from the feature point along the normal direction on both sides until the two ends of the line segment intersect with the left bank edge and the right bank edge, respectively. The intersection of the bank lines is the endpoint of the normal cross-section, and the line segment between the two points is the normal cross-section. The width of the river segment is the length of the line segment covering the water area on the normal cross-section. The sampling interval represents the range of the line segment on the normal cross-section from the intersection point (starting point) of the left bank line to the intersection point (ending point) of the right bank line. Near-infrared reflectance is sampled only for water pixels (pixel value of 1) within this interval to ensure that the sampling data comes only from the target water area and excludes interference from land pixels. The sampling interval can be set to 1 pixel.

[0027] It should be noted that atmospherically corrected water-leaving reflectance represents the reflectance reflecting only the water body's own reflective properties after removing atmospheric scattering, absorption, and other interferences. It is dimensionless and its value ranges from 0 to 1. Specifically, the radiometric calibration data of the input image is first obtained, and the image DN value is converted into apparent radiance. Then, an atmospheric correction model (such as the 6S model) is used, with atmospheric parameters at the time of image capture (such as atmospheric aerosol optical thickness and water vapor content, which can be obtained from metadata attached to the image or meteorological station data) to remove the influence of the atmosphere on radiance and obtain the surface reflectance. Finally, for water bodies, a water reflectance separation algorithm (such as based on the difference in reflectance between the near-infrared band and the visible light band) is used to separate the reflectance generated only by the water body itself from the surface reflectance, which is the atmospherically corrected water-leaving reflectance.

[0028] It should be noted that the cross-sectional coordinates start at the sampling interval with 0 and increase towards the end of the sampling interval; the minimum near-infrared position is the cross-sectional coordinate position corresponding to the pixel with the lowest near-infrared reflectance value among all water pixels in the sampling interval, and this position is least affected by the proximity effect between the two banks; the shore-to-near-infrared residual offset is used to quantify the direction and degree of offset of the minimum near-infrared position relative to the geometric midline. A positive difference indicates that the offset is towards the end of the sampling interval, i.e., the right bank; a negative difference indicates that the offset is towards the start of the sampling interval, i.e., the left bank.

[0029] In one embodiment of the present invention, a left bank buffer zone is formed by extending a predetermined width along the left bank edge into the land area, and a right bank buffer zone is formed by extending a predetermined width along the right bank edge into the land area. Both the left bank buffer zone and the right bank buffer zone contain only land area pixels. The average values ​​of the near-infrared reflectance of all pixels in the left bank buffer zone and the near-infrared reflectance of all pixels in the right bank buffer zone are calculated respectively. The two average values ​​are used as the near-infrared equivalent brightness of the left bank and the near-infrared equivalent brightness of the right bank, respectively. The near-infrared reflectance is the water-leaving reflectance after atmospheric correction.

[0030] It should be noted that the left bank buffer zone is an area of ​​a specific width extending from the left bank edge (the broken line of the water-land boundary) towards the land side. Only land area pixels are retained within this area. The width of the buffer zone can be set according to the image spatial resolution; for example, 3 to 5 pixels (30 to 50 meters) for a 10-meter resolution image, and 2 to 3 pixels (60 to 90 meters) for a 30-meter resolution image, ensuring coverage of the main land area types near the shoreline. The width of the buffer zone can also be set according to the shoreline type: 50 to 80 meters for vegetated shorelines and 30 to 50 meters for building / bare soil shorelines, avoiding the introduction of non-shoreline-related land area signals due to excessive width. The right bank buffer zone is an area of ​​a specific width extending from the right bank edge towards the land side, containing only land area pixels. The left bank near-infrared equivalent brightness represents the average radiation intensity of the left bank land area in the near-infrared band; the right bank near-infrared equivalent brightness represents the average radiation intensity of the right bank land area in the near-infrared band.

[0031] In one embodiment of the present invention, a near-infrared profile composed of near-infrared reflectance collected along the normal transverse line is used as the fitting object. A dual-bank convolution and baseline model is established and fitted to obtain the baseline term and attenuation scale. The dual-bank convolution and baseline model includes the left bank near-infrared equivalent brightness, the right bank near-infrared equivalent brightness, the transverse coordinates along the normal transverse line, the river segment width, the baseline term, and the attenuation scale. Based on the shore-to-near-infrared residual offset and attenuation scale, the target asymmetry ratio is obtained through exponential calculation; the equivalent brightness of the left shore near-infrared is added to the equivalent brightness of the right shore near-infrared to obtain the total brightness of both shores; keeping the value of the total brightness of both shores unchanged, the total brightness of both shores is proportionally allocated according to the target asymmetry ratio to obtain the re-normalized equivalent brightness of the left shore and the re-normalized equivalent brightness of the right shore respectively.

[0032] It should be noted that the calculation formula for fitting the near-infrared reflectance profile using dual-bank convolution and baseline models is as follows:

[0033] in This represents the near-infrared reflectance obtained from the fitting. This represents the baseline term, expressed in units of reflectance. This indicates the near-infrared equivalent brightness of the left bank. The value represents the near-infrared equivalent brightness of the right bank, x represents the cross-sectional coordinate, and s represents the channel centerline parameter. Indicates the attenuation scale. denoted by , and e represents the natural constant.

[0034] It should be noted that the least squares method is used for fitting, specifically including the following steps: 1. Set initial values ​​for the baseline term, for example, 0.8 to 1 times the minimum reflectance value in the near-infrared profile; set initial values ​​for the attenuation scale, for example, 1 / 5 to 1 / 3 of the river width; 2. Substitute these values ​​into the model to calculate the predicted reflectance at each cross-sectional coordinate; 3. Calculate the sum of squared errors between the predicted and actual reflectance; 4. Iteratively adjust the baseline term and the attenuation scale until the sum of squared errors is minimized, and output the final parameters. The baseline term is an additive bias parameter in the bi-bank convolution and baseline model, with units of reflectance, representing a system bias independent of the brightness of the land on both banks; the attenuation scale is the attenuation length parameter in the bi-bank convolution and baseline model, with units consistent with the river width (meters or pixels), representing the rate of attenuation of the near-infrared brightness of the land on both banks with distance from the shore. The larger the value, the slower the brightness attenuation and the wider the influence range.

[0035] It should be noted that the target asymmetry ratio is a dimensionless parameter representing the degree of near-infrared brightness asymmetry between the left and right banks. A value greater than 1 indicates that the left bank brightness is greater than the right bank brightness, and a value less than 1 indicates that the right bank brightness is greater than the left bank brightness. The renormalized left bank equivalent brightness represents the left bank near-infrared brightness allocated according to the target asymmetry ratio, assuming the total brightness of both banks remains unchanged. The unit is reflectance. This eliminates the random error in the original brightness statistics and better reflects the brightness asymmetry law reflected by the offset. The specific calculation process is as follows: multiply the sum of the total brightness of both banks by the target asymmetry ratio to obtain the numerator; add 1 to the target asymmetry ratio to obtain the denominator; divide the numerator by the denominator to obtain the renormalized left bank equivalent brightness. The specific calculation process for the renormalized right bank equivalent brightness is as follows: add 1 to the target asymmetry ratio to obtain the denominator; divide the sum of the total brightness of both banks by the denominator to obtain the renormalized right bank equivalent brightness.

[0036] Specifically, the target asymmetry ratio The calculation formula is as follows: ,in This represents the shore-to-near-infrared residual offset. denoted by , where s represents the channel centerline parameter, and exp represents the exponential function with the natural constant e as the base.

[0037] In one embodiment of the present invention, a target band set is determined. For each band in the target band set, the average value of the original reflection of the band within the left bank buffer zone and the average value of the original reflection of the band within the right bank buffer zone are calculated. The two average values ​​are used as the left bank cross-band average reflection and the right bank cross-band average reflection of the band, respectively. The left bank cross-band average reflection of each band is divided by the average reflection corresponding to the left bank near-infrared equivalent brightness, and the right bank cross-band average reflection of the band is divided by the average reflection corresponding to the right bank near-infrared equivalent brightness to obtain the left bank propagation coefficient and the right bank propagation coefficient of the band. Using the width range of the river segment from the left bank to the right bank on the normal cross section as the calculation range, for each band in the target band set, combined with the left bank propagation coefficient and the right bank propagation coefficient of the band, the proximity residual of the band at each position on the normal cross section is calculated according to the attenuation law of the distance from the bank.

[0038] It should be noted that the target band set represents a specific combination of visible light bands used for water quality parameter (algae color, turbidity) inversion. It must cover the wavelength range sensitive to algae color and turbidity, specifically including: blue band (wavelength 450 to 520 nm, sensitive to chlorophyll in water), green band (wavelength 520 to 600 nm, sensitive to both algae color and turbidity), and red band (wavelength 620 to 680 nm, sensitive to turbidity). If the input image is a multispectral image, directly select the bands corresponding to the above wavelengths; if it is a hyperspectral image… Select one band corresponding to the center wavelength from each of the above wavelength ranges (e.g., 490nm for the blue band, 560nm for the green band, and 650nm for the red band); the left bank cross-band average reflectance is the arithmetic mean of the original reflectance of the target band within the left bank buffer zone (the uncorrected reflectance pixel value of this band in the input image), in units of reflectance, representing the average radiation intensity of the left bank land area in this target band; the left bank propagation coefficient represents the dimensionless scaling factor that maps the proximity effect intensity of the left bank near-infrared band to the target band.

[0039] It should be noted that the proximity residual represents the residual component in the target band caused by radiation leakage from the land on both sides of the shore. It decreases with increasing distance from the shore, and the unit is reflectance. The attenuation law with increasing distance from the shore adopts exponential decay, based on the radiative propagation characteristics of atmospheric scattering and the point spread function of the imaging system, i.e., land radiation decreases exponentially with increasing distance from the shore. Specifically, the proximity residual of band b... The calculation formula is as follows:

[0040] in This represents the equivalent brightness of the left bank after renormalization. This represents the equivalent brightness of the right bank after renormalization. Represents the left bank propagation coefficient. Represents the right bank propagation coefficient. Indicates the attenuation scale. denoted by , e represents the natural constant, x represents the cross-sectional coordinate, s represents the channel centerline parameter, and b represents the waveband.

[0041] In one embodiment of the present invention, for each band in the target band set, the left bank propagation coefficient of the band is multiplied by the renormalized left bank equivalent brightness, and the right bank propagation coefficient of the band is multiplied by the renormalized right bank equivalent brightness. The two product results are added together. The sum is divided by the sum of the renormalized left bank equivalent brightness and the renormalized right bank equivalent brightness to obtain the shore spectrum weighting coefficient of the band. The baseline term is multiplied by the shore spectrum weighting coefficient of the band to obtain the baseline residual of the band. For each band in the target band set, the neighboring residuals of that band are added to the baseline residuals to obtain the total residual field of that band; The corrected reflection for this band is obtained by subtracting the corresponding pixel value in the total residual field from the original reflection pixel value. If the subtraction result is less than zero, it is taken as zero.

[0042] It should be noted that the shore spectrum weighting coefficient represents the dimensionless coefficient used to map the baseline term across the near-infrared band to the target band, taking into account both shore radiation intensity and band mapping ratio; the baseline residual represents the residual component in the target band that is independent of the distance from the shore and is caused by system bias (such as atmospheric correction residuals and sensor null drift); the total residual field represents the sum of the residuals of all non-water body intrinsic radiation in the target band, including the neighbor residuals that vary with the distance from the shore and the fixed baseline residuals; and the corrected reflection is the water reflectance after deducting all non-water body intrinsic radiation residuals.

[0043] Specifically, the shore spectrum weighting coefficients of band b The calculation formula is as follows: ,in This represents the equivalent brightness of the left bank after renormalization. This represents the equivalent brightness of the right bank after renormalization. Represents the left bank propagation coefficient. This represents the right bank propagation coefficient.

[0044] In one embodiment of the present invention, green band corrected reflection and blue band corrected reflection are selected from the corrected reflection, and the pixel value of green band corrected reflection is divided by the sum of the pixel value of blue band corrected reflection and the smallest positive number to obtain algae color proxy. Select the red band corrected reflection and the green band corrected reflection from the corrected reflection, and divide the pixel value of the red band corrected reflection by the sum of the pixel value of the green band corrected reflection and the smallest positive number to obtain the turbidity proxy. Along the normal cross section, calculate the difference between the corrected reflection pixel value of the green band and the corrected reflection pixel value of the blue band, and the difference between the corrected reflection pixel value of the red band and the corrected reflection pixel value of the green band. Take the non-negative values ​​of the two differences, and then calculate the median of the two non-negative value sets. The two medians are used as the segment-level energy statistics value of algal color correlation and the segment-level energy statistics value of turbidity correlation, respectively. Add the energy statistics of the algal color-related segment to the energy statistics of the turbidity-related segment, and then add the smallest positive number to obtain the denominator for weight calculation; divide the energy statistics of the algal color-related segment by the denominator for weight calculation to obtain the adaptive weight; Multiply the adaptive weight by the algal color proxy, then subtract the adaptive weight from 1 and multiply by the turbidity proxy. Add the two products together to obtain the water quality evidence field.

[0045] It should be noted that the green band corrected reflectance is sensitive to algal color-related substances (such as chlorophyll) in the water. Chlorophyll enhances green light reflection and absorbs blue light, so the difference / ratio between its reflectance and the blue band can represent algal color. The blue band corrected reflectance, in conjunction with the green band reflectance, reflects algal color characteristics through the ratio / difference. The red band corrected reflectance is sensitive to suspended particulate matter (which determines turbidity) in the water. The higher the turbidity, the stronger the red light scattering and the greater the reflectance. In conjunction with the green band reflectance, it can represent turbidity. The algal color proxy represents a dimensionless parameter of algal color characteristics in the water. The larger the value, the higher the content of algal color-related substances (such as chlorophyll) in the water. The very small positive number is used to prevent the denominator from being zero in the reflectance ratio calculation, and is set to 10 to the power of -6. The turbidity proxy represents a dimensionless parameter of turbidity characteristics in the water. The larger the value, the more suspended particulate matter and the higher the turbidity in the water.

[0046] It should be noted that a positive difference in corrected reflectance between the green and blue bands indicates significant algal color characteristics, meaning chlorophyll enhances green light reflection and absorbs blue light, while a negative difference indicates weak algal color characteristics. Similarly, a positive difference in corrected reflectance between the red and green bands indicates significant turbidity characteristics, meaning suspended particulate matter enhances red light scattering, while a negative difference indicates weak turbidity characteristics. Only the positive characteristic signals of algal color and turbidity are retained in the non-negative part to avoid negative differences canceling out effective information. The larger the value of the segment-level energy statistics related to algal color, the more significant the overall algal color characteristics of the cross-section. The larger the value of the segment-level energy statistics related to turbidity, the more significant the overall turbidity characteristics of the cross-section. The adaptive weight represents the dimensionless parameter representing the proportion of algal color surrogate in the water quality evidence field, with a value ranging from 0 to 1. The more significant the algal color characteristics, the closer the weight is to 1; the more significant the turbidity characteristics, the closer the weight is to 0. Furthermore, the adaptive weights directly correspond to the relative strength of algal color features. Subtracting the relative strength of the turbidity features corresponding to the adaptive weights from 1 and adding the products of the two ensures that the water quality evidence field prioritizes the dominant water quality features of the current section. For example, when algal color is dominant, the algal color proxy contribution is large.

[0047] In one embodiment of the present invention, the total residual of the near-infrared channel is selected from the total residual field, and then the median of the total residual of the near-infrared channel corresponding to all pixels on the normal cross section is obtained; the total residual value of the near-infrared channel of each pixel on the normal cross section is divided by the median to obtain the normalized value of the neighbor risk corresponding to each pixel. Take the absolute value of the near-infrared residual offset along the shore, and divide this absolute value by the width of the river section to obtain the geometric risk coefficient along the shore. For each pixel, the normalized value of the neighbor risk is negative, and then the negative result is exponentially calculated. First, the shore geometric risk coefficient is subtracted from 1, and then the result of the exponential calculation is multiplied by the difference to obtain the pixel reliability weight corresponding to each pixel. On the normal cross section, the value of each pixel in the water quality evidence field is multiplied by the pixel reliability weight of the corresponding pixel to obtain a set of product values. The median of the set of product values ​​is then used as the segment-level single value.

[0048] It should be noted that the total residual of the near-infrared channel represents the pixel-level residual value of the near-infrared band in the total residual field, reflecting the total impact of the proximity effect and system bias on each pixel; the proximity risk normalization value is used to quantify the degree of proximity risk of a single pixel relative to the average level of the cross section, with a value greater than 1 indicating a risk higher than average and less than 1 indicating a risk lower than average; the shore-oriented geometric risk coefficient is a dimensionless parameter, and the closer the value is to 1, the more severe the cross section eccentricity and the lower the geometric reliability; the pixel reliability weight is a dimensionless parameter; the segment-level single value is a dimensionless parameter, obtained by pooling the median of the water quality evidence field after reliability weighting, which can robustly reflect the overall water quality status of the cross section and avoid interference from abnormal pixels.

[0049] Specifically, pixel reliability weight The calculation formula is as follows: ,in This represents the normalized value of proximity risk. This represents the onshore geometric risk coefficient.

[0050] In one embodiment of the present invention, a corresponding segment length weight is determined for each normal cross-section within the same river segment, and the unit of the segment length weight is meters; Calculate the sum of the segment length weights corresponding to all normal cross-sections within the same river segment; Divide the segment length weight corresponding to each normal transverse line by the sum of the segment length weights to obtain the normalized weight corresponding to each normal transverse line. Multiply the normalized weight corresponding to each normal cross section by the segment-level single value corresponding to that normal cross section to obtain the weighted contribution value corresponding to each normal cross section. The weighted contribution values ​​corresponding to all normal cross sections are summed to obtain the panoramic index as the water quality monitoring result.

[0051] It should be noted that the definition of the same river segment is based on the natural starting and ending points of the river channel, such as from the source to the estuary, or from the dam to the bridge, to ensure relatively uniform hydrological and water quality conditions within the segment. If the river channel is too long, for example, exceeding 10 kilometers, it can be segmented according to the confluence of tributaries or points of topographic change. The segment length weight is in meters. If the spacing between adjacent normal cross-sections within the same river segment is uniform (e.g., the spacing is 50 meters), then the segment length weight for each cross-section is taken as that spacing value (50 meters). If the spacing is uneven, the average of the spacing between the cross-section and the previous adjacent cross-section and the spacing between the cross-section and the next adjacent cross-section is taken (e.g., if the previous spacing is 40 meters and the next spacing is 60 meters, the segment length weight is 50 meters). In addition, the segment length weight must cover the entire length of the same river segment. The rule for calculating the spacing between the first and last cross-sections is that the first cross-section is taken as the spacing between itself and the second cross-section, and the last cross-section is taken as the spacing between itself and the second to last cross-section, to avoid omitting the river segment length.

[0052] It should be noted that the sum of segment length weights represents the sum of the segment length weights of all normal cross sections within the same river segment, in meters, and is equal to the actual total length of the river segment; the normalized weight represents the proportion of the segment length of a single normal cross section to the total length of the river segment, used to eliminate differences in the absolute values ​​of segment length weights, so that the weighted aggregation result only reflects the contribution of water quality information; the weighted contribution value is a dimensionless parameter, representing the actual contribution of the water quality information of a single normal cross section to the river segment-level panoramic index. The larger the normalized weight and the larger the segment-level single value, the greater the contribution value; the panoramic index represents a single river segment-level index that integrates water quality information from all sections of the same river segment. It is dimensionless, and the larger the value, the stronger the comprehensive spectral evidence of water quality deterioration in the entire river segment.

[0053] It should be noted that historical panoramic indicator data of the monitored river section is collected, and extreme outliers are removed to obtain a valid historical data set. For example, abnormally high values ​​caused by a sudden increase in turbidity after heavy rain are eliminated by comparing with meteorological data. The data is then correlated with the water quality levels of the same period's field sampling. If the 20th percentile corresponds to Class I water quality, the 40th percentile to Class II, the 60th percentile to Class III, and the 80th percentile to Class IV, and values ​​greater than 80% correspond to Class V, then each quantile is set as the threshold for the corresponding water quality level. If the calculated quantile results for the historical data of the monitored river section are: 20th percentile corresponds to 0.3, 40th percentile to 0.5, 60th percentile to... 0.7 and 80th percentile correspond to 0.9. Therefore, when the panoramic index is less than 0.3, the water quality level is Class I, suitable for primary protection zones of drinking water sources; when the panoramic index is greater than or equal to 0.3 and less than 0.5, the level is Class II, suitable for secondary protection zones of drinking water sources; when the panoramic index is greater than or equal to 0.5 and less than 0.7, the level is Class III, suitable for fishery water use; when the panoramic index is greater than or equal to 0.7 and less than 0.9, the level is Class IV, suitable for industrial water use; and when the panoramic index is greater than or equal to 0.9, the level is Class V, suitable only for agricultural water use. In operation and maintenance management, if the panoramic index exceeds 0.7 in a certain period, the frequency of on-site sampling needs to be increased to investigate the source of pollution. Furthermore, using the above approach, the statistical values ​​of algal color-related segment-level energy and turbidity-related segment-level energy can be used to determine whether the water quality is dominated by algal color or turbidity, which will not be elaborated here.

[0054] In one embodiment of the present invention, a surface water environment water quality monitoring and maintenance management system includes: Infrared residual offset output module 201 is used to acquire water body mask, shore line and normal cross section, sample near-infrared reflection according to normal cross section, determine geometric center line position and minimum near-infrared position, and calculate shore-directed near-infrared residual offset. The equivalent brightness output module 202 is used to obtain the near-infrared equivalent brightness of both sides on the buffer zone between the two sides, establish a dual-side convolution and baseline model to obtain the baseline term and attenuation scale, determine the target asymmetry ratio by combining the shore-to-shore near-infrared residual offset, and calculate the re-normalized equivalent brightness of both sides under the condition that the total brightness of both sides remains unchanged. The corrected reflection output module 203 is used to calculate the cross-band average reflection to obtain the propagation coefficient, generate the neighboring residual based on the equivalent brightness and attenuation scale of the renormalized banks, obtain the baseline residual by weighting the propagation baseline term according to the bank spectrum, superimpose to obtain the total residual field, and subtract the original reflection to obtain the corrected reflection. The water quality evidence field output module 204 is used to construct algal color proxy and turbidity proxy based on corrected reflection, calculate segment-level energy statistics to obtain adaptive weights, and fuse the three to obtain the water quality evidence field. The segment-level single-value output module 205 is used to calculate the neighboring risk normalization value based on the total residual field, combine the shore-side near-infrared residual offset with the river segment width to calculate the shore-side geometric risk coefficient, generate the pixel reliability weight, and robustly converge the water quality evidence field to obtain the segment-level single value. The panoramic index output module 206 is used to perform weighted aggregation of the segment-level single values ​​of each segment according to the segment length weight, and output the panoramic index as the water quality monitoring result.

[0055] It should be noted that the interval and threshold sizes are set for ease of comparison. The size of the threshold depends on the amount of sample data and the base number set by those skilled in the art for each set of sample data, as long as it does not affect the proportional relationship between the parameter and the quantized value. Furthermore, the above formulas are all dimensionless calculations, and the formulas are derived from software simulations using a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0056] The embodiments of this example have been described above. However, this example is 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 many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. A method for monitoring, operating, and managing surface water environmental water quality, characterized in that, Includes the following steps: Step S101: Obtain the water body mask, the edge lines on both banks and the normal cross section. Sample near-infrared reflection based on the normal cross section to determine the geometric centerline position and the minimum near-infrared position, and calculate the shore-directed near-infrared residual offset. Step S102: Obtain the near-infrared equivalent brightness of both banks on the buffer zone between the two banks, establish a dual-bank convolution and baseline model to obtain the baseline term and attenuation scale, determine the target asymmetry ratio by combining the bank-to-bank near-infrared residual offset, and calculate the re-normalized equivalent brightness of both banks under the condition that the total brightness of both banks remains unchanged. Step S103: Calculate the cross-band average reflection to obtain the propagation coefficient, generate the neighboring residual based on the equivalent brightness and attenuation scale of the renormalized banks, obtain the baseline residual by weighting the propagation baseline term according to the bank spectrum, superimpose to obtain the total residual field, and subtract the original reflection to obtain the corrected reflection. Step S104: Construct algal color proxy and turbidity proxy based on corrected reflection, calculate segment-level energy statistics to obtain adaptive weights, and fuse the three to obtain a water quality evidence field. Step S105: Calculate the normalized value of the adjacent risk based on the total residual field, and calculate the bank-directed geometric risk coefficient by combining the bank-directed near-infrared residual offset and the river section width. Based on this, generate the pixel reliability weight and perform robust aggregation of the water quality evidence field to obtain segment-level single values. Step S106: Weighted aggregation of segment-level individual values ​​for each segment according to segment length weight, and output panoramic indicators as water quality monitoring results.

2. The surface water environmental water quality monitoring, operation and maintenance management method according to claim 1, characterized in that, Obtain a water mask from the input image. The water mask is a set of image pixels. A pixel value of 1 indicates that the corresponding area is water, and a pixel value of 0 indicates that the corresponding area is land. Extract the riverbanks from the input image. The riverbanks are broken lines representing the boundaries between the water body and the land. Construct a normal cross-section along the river centerline parameters. The normal cross-section is perpendicular to the tangent of the river centerline, and its two ends intersect the riverbanks. Determine the river segment width between the two intersection points of the normal cross-section and the riverbanks. The river segment width is the length of the line segment of the normal cross-section inside the water mask. On the normal transverse line, the sampling interval is defined with the intersection of the normal transverse line and the left bank edge as the starting point and the intersection of the normal transverse line and the right bank edge as the ending point; the reflectance of the near-infrared band is collected only at the pixel position with a pixel value of 1 in the water mask along the sampling interval. The reflectance of the near-infrared band is the water-free reflectance after atmospheric correction. Based on the determined width of the river segment, calculate the position of the geometric centerline, which is the cross-sectional coordinate corresponding to the midpoint of the river segment width; Within the sampling interval, the position with the lowest reflectivity in the near-infrared band is identified as the minimum near-infrared position. Subtract the cross-sectional coordinates corresponding to the geometric midline position from the cross-sectional coordinates corresponding to the minimum near-infrared position to obtain the shore-side near-infrared residual offset.

3. The surface water environmental water quality monitoring, operation and maintenance management method according to claim 1, characterized in that, A left bank buffer zone is formed by extending a set width along the left bank edge into the land area, and a right bank buffer zone is formed by extending a set width along the right bank edge into the land area. Both the left and right bank buffer zones contain only land area pixels. The average values ​​of the near-infrared reflectance of all pixels in the left bank buffer zone and the near-infrared reflectance of all pixels in the right bank buffer zone are calculated. The two average values ​​are used as the near-infrared equivalent brightness of the left bank and the near-infrared equivalent brightness of the right bank, respectively. The near-infrared reflectance is the water-leaving reflectance after atmospheric correction.

4. The surface water environmental water quality monitoring, operation and maintenance management method according to claim 3, characterized in that, Using the near-infrared reflectance collected along the normal cross-section as the fitting object, a dual-bank convolution and baseline model is established and fitted to obtain the baseline term and attenuation scale. The dual-bank convolution and baseline model includes the left bank near-infrared equivalent brightness, the right bank near-infrared equivalent brightness, the cross-sectional coordinates along the normal cross-section, the river segment width, the baseline term, and the attenuation scale. Based on the shore-to-near-infrared residual offset and attenuation scale, the target asymmetry ratio is obtained through exponential calculation; the equivalent brightness of the left shore near-infrared is added to the equivalent brightness of the right shore near-infrared to obtain the total brightness of both shores; keeping the value of the total brightness of both shores unchanged, the total brightness of both shores is proportionally allocated according to the target asymmetry ratio to obtain the re-normalized equivalent brightness of the left shore and the re-normalized equivalent brightness of the right shore respectively.

5. A method for monitoring, operating, and managing surface water environmental water quality according to claim 1, characterized in that, Determine the target band set. For each band in the target band set, calculate the average original reflection of the band within the left bank buffer zone and the average original reflection of the band within the right bank buffer zone. These two average values ​​are used as the left bank cross-band average reflection and the right bank cross-band average reflection of the band, respectively. Divide the left bank cross-band average reflection of each band by the average reflection corresponding to the left bank near-infrared equivalent brightness, and divide the right bank cross-band average reflection of the band by the average reflection corresponding to the right bank near-infrared equivalent brightness to obtain the left bank propagation coefficient and the right bank propagation coefficient of the band. Using the width range of the river segment from the left bank to the right bank on the normal cross section as the calculation range, for each band in the target band set, combined with the left bank propagation coefficient and the right bank propagation coefficient of the band, the proximity residual of the band at each position on the normal cross section is calculated according to the attenuation law of the distance from the bank.

6. A method for monitoring, operating, and managing surface water environmental water quality according to claim 5, characterized in that, For each band in the target band set, multiply the left bank propagation coefficient of that band by the renormalized left bank equivalent brightness, multiply the right bank propagation coefficient of that band by the renormalized right bank equivalent brightness, and add the two products together; divide the sum by the sum of the renormalized left bank equivalent brightness and the renormalized right bank equivalent brightness to obtain the shore spectral weighting coefficient of that band; multiply the baseline term by the shore spectral weighting coefficient of that band to obtain the baseline residual of that band; For each band in the target band set, the neighboring residuals of that band are added to the baseline residuals to obtain the total residual field of that band; The corrected reflection for this band is obtained by subtracting the corresponding pixel value in the total residual field from the original reflection pixel value. If the subtraction result is less than zero, it is taken as zero.

7. A method for monitoring, operating, and managing surface water environmental water quality according to claim 1, characterized in that, Select green band corrected reflection and blue band corrected reflection from the corrected reflection, and divide the pixel value of green band corrected reflection by the sum of the pixel value of blue band corrected reflection and the smallest positive number to obtain the algae color proxy. Select the red band corrected reflection and the green band corrected reflection from the corrected reflection, and divide the pixel value of the red band corrected reflection by the sum of the pixel value of the green band corrected reflection and the smallest positive number to obtain the turbidity proxy. Along the normal cross section, calculate the difference between the corrected reflection pixel value of the green band and the corrected reflection pixel value of the blue band, and the difference between the corrected reflection pixel value of the red band and the corrected reflection pixel value of the green band. Take the non-negative values ​​of the two differences, and then calculate the median of the two non-negative value sets. The two medians are used as the segment-level energy statistics value of algal color correlation and the segment-level energy statistics value of turbidity correlation, respectively. Add the energy statistics of the algal color-related segment to the energy statistics of the turbidity-related segment, and then add the smallest positive number to obtain the denominator for weight calculation; divide the energy statistics of the algal color-related segment by the denominator for weight calculation to obtain the adaptive weight; Multiply the adaptive weight by the algal color proxy, then subtract the adaptive weight from 1 and multiply by the turbidity proxy. Add the two products together to obtain the water quality evidence field.

8. A method for monitoring, operating, and managing surface water environmental water quality according to claim 1, characterized in that, Select the total residual of the near-infrared channel from the total residual field, and then obtain the median of the total residual of the near-infrared channel corresponding to all pixels on the normal cross section; divide the total residual value of the near-infrared channel of each pixel on the normal cross section by the median to obtain the normalized value of the neighbor risk corresponding to each pixel. Take the absolute value of the near-infrared residual offset along the shore, and divide this absolute value by the width of the river section to obtain the geometric risk coefficient along the shore. For each pixel, the normalized value of the neighbor risk is negativeed, and then the negativeed result is exponentially calculated. First, subtract the shore-side geometric risk coefficient from 1, then multiply the result of the exponential calculation by the difference to obtain the pixel reliability weight corresponding to each pixel. On the normal cross section, the value of each pixel in the water quality evidence field is multiplied by the pixel reliability weight of the corresponding pixel to obtain a set of product values. The median of the set of product values ​​is then used as the segment-level single value.

9. A method for monitoring, operating, and managing surface water environmental water quality according to claim 1, characterized in that, For each normal cross-section within the same river segment, a corresponding segment length weight is determined, with the unit of segment length weight being meters; Calculate the sum of the segment length weights corresponding to all normal cross-sections within the same river segment; Divide the segment length weight corresponding to each normal transverse line by the sum of the segment length weights to obtain the normalized weight corresponding to each normal transverse line. Multiply the normalized weight corresponding to each normal cross section by the segment-level single value corresponding to that normal cross section to obtain the weighted contribution value corresponding to each normal cross section. The weighted contribution values ​​corresponding to all normal cross sections are summed to obtain the panoramic index as the water quality monitoring result.

10. A surface water environment water quality monitoring and operation management system, characterized in that, The method for monitoring, operating, and managing surface water environmental water quality as described in any one of claims 1 to 9 includes: The infrared residual offset output module is used to acquire the water body mask, the shoreline and the normal cross section. It samples the near-infrared reflection based on the normal cross section to determine the geometric centerline position and the minimum near-infrared position, and calculates the shore-directed near-infrared residual offset. The equivalent brightness output module for both sides is used to obtain the near-infrared equivalent brightness of both sides on the buffer zone between the two sides, establish a dual-side convolution and baseline model to obtain the baseline term and attenuation scale, combine the shore-to-shore near-infrared residual offset to determine the target asymmetry ratio, and calculate the renormalized equivalent brightness of both sides under the condition that the total brightness of both sides remains unchanged. The corrected reflection output module is used to calculate the cross-band average reflection to obtain the propagation coefficient, generate the neighboring residual based on the equivalent brightness and attenuation scale of the renormalized banks, obtain the baseline residual by weighting the propagation baseline term according to the bank spectrum, and obtain the total residual field by superposition. The original reflection is subtracted to obtain the corrected reflection. The water quality evidence field output module is used to construct algal color proxy and turbidity proxy based on corrected reflection, calculate segment-level energy statistics to obtain adaptive weights, and fuse the three to obtain the water quality evidence field. The segment-level single-value output module is used to calculate the neighboring risk normalization value based on the total residual field, combine the shore-side near-infrared residual offset with the river segment width to calculate the shore-side geometric risk coefficient, generate the pixel reliability weight, and robustly converge the water quality evidence field to obtain the segment-level single value. The panoramic index output module is used to weight and aggregate the segment-level single values ​​of each segment according to the segment length weight, and output the panoramic index as the water quality monitoring result.

Images