Water pollutant stratified collection method based on multifunctional monitoring ship

Abstract

The application relates to the technical field of data processing, and discloses a water body pollutant layered collection method based on a multifunctional monitoring ship. The method comprises the following steps: determining the stability of a deep layer in combination with a Richardson number, setting a pycnocline precision sampling grid on and under a pycnocline interface to form a sampling depth sequence; correcting cable tilt errors through a depth encoder and a three-axis tilt angle sensor, collecting water samples at each depth layer position and recording three-dimensional coordinates to generate a sample information set; performing multi-index detection on the water samples to construct a pollutant concentration matrix; calculating a vertical direction concentration gradient vector and a second-order difference, identifying a pollutant concentration mutation interface, and determining a main pollutant accumulation layer position and a source factor spatial distribution. The application establishes an intelligent sampling depth planning method based on the layered structure of water body density, solves the problems of lack of pertinence in traditional sampling depth determination and insufficient sampling precision, and improves the identification accuracy of pollutant vertical distribution characteristics and the scientificity of pollutant source tracing.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method for stratified collection of water pollutants based on a multi-functional monitoring vessel. Background Technology

[0002] Stratified sampling of water pollutants is an important technical means for water environment monitoring. Traditional methods mainly rely on fixed-point, fixed-depth sampling or stratified sampling at equal intervals. After the sampling vessel is fixed in a specific location, water samples are collected at a preset fixed depth using a water sampler, or sampling points are set up throughout the water column at uniform depth intervals. These methods are relatively simple to operate and can obtain water samples at different depths for subsequent laboratory analysis, and are widely used in routine water quality monitoring.

[0003] However, existing technologies have significant shortcomings, primarily in the lack of specificity in determining sampling depth based on the structural characteristics of the water body. Traditional fixed-point, fixed-depth sampling relies on experience to select depth, which cannot adapt to the differences in the stratified structure of different water bodies. While stratified sampling at equal intervals covers the entire water column, the sampling density is insufficient in areas with drastic changes in pollutant concentration, while it is too dense in areas with uniform pollutant distribution, resulting in wasted resources. Furthermore, the lack of precise correction for cable tilt during sampling allows water flow to cause deviations between the actual and target depths of the sampler, affecting sampling accuracy. Finally, post-sampling testing and analysis are conducted independently, lacking comprehensive analysis of multi-indicator data, making it difficult to identify the vertical distribution patterns of pollutants and the characteristics of pollution sources.

[0004] Based on the aforementioned shortcomings, a series of technical problems need to be solved. First, how to dynamically determine the sampling depth according to the density stratification structure of the water body, increasing sampling density in areas where pollutants easily accumulate, such as density gradient layers, and reducing the number of sampling points in homogeneous layers, to achieve optimal allocation of sampling resources. Second, how to correct cable tilt errors in real time during sampling to ensure the sampler reaches the precise target depth and improve the spatial positioning accuracy of stratified sampling. Third, how to construct a vertical distribution matrix of pollutant concentrations based on multi-index detection data, quantitatively identify abrupt changes in pollutant concentrations through gradient analysis and differential operations, and reveal the accumulation patterns of pollutants in different water layers. Finally, how to perform source apportionment analysis on the pollutant concentration matrix to separate the characteristic components and spatial distribution of different pollution sources, trace the source and migration path of pollutants, and provide a scientific basis for pollution control. Summary of the Invention

[0005] This application provides a method for stratified sampling of water pollutants based on a multi-functional monitoring vessel. By establishing an intelligent sampling depth planning method based on the water density stratification structure, it solves the problems of insufficient targeting and sampling accuracy in traditional sampling depth determination, and improves the accuracy of identifying the vertical distribution characteristics of pollutants and the scientific nature of tracing pollution sources.

[0006] This application provides a method for stratified collection of water pollutants based on a multi-functional monitoring vessel, the method comprising:

[0007] Step S1: Collect acoustic Doppler frequency shift signals and CTD probe data along the preset route, calculate the density stratification structure of the water body in the vertical direction according to the density gradient equation, identify the depth position of the density stratification interface, and generate a three-dimensional density field grid.

[0008] Step S2: Based on the depth position of the density jump layer interface in the three-dimensional density field grid, and combined with the Richardson number, determine the stability of each depth layer, set up a jump layer precision sampling grid above and below the density jump layer interface, and set up sampling points with equal spacing in the uniform layer between the jump layers to form a sampling depth sequence.

[0009] Step S3: Lower the multi-channel stratified water sampling device to the target depth in the sampling depth sequence. After correcting the cable tilt error by using a depth encoder and a three-axis tilt sensor, trigger the electromagnetic drive valve to collect water samples from each depth layer and record the three-dimensional coordinates, collection time and environmental parameters to generate a sample information set.

[0010] Step S4: Filter and separate the water samples in the sample information set and perform multi-index detection to obtain the measured concentration values ​​of heavy metals, nutrients and organic pollutants in each layer, and construct a pollutant concentration matrix;

[0011] Step S5: Calculate the vertical concentration gradient vector and second-order difference based on the pollutant concentration matrix, identify the pollutant concentration abrupt change interface, calculate the stratum pollution load index, and determine the main accumulation strata of pollutants and the spatial distribution of source factors.

[0012] The technical solution provided in this application collects acoustic Doppler frequency shift signals and CTD probe data along a preset route, calculates the vertical density stratification structure of the water body based on the density gradient equation, identifies the depth position of the density gradient interface, and generates a three-dimensional density field grid. This transforms the determination of sampling depth from empirical judgment to scientific planning based on the physical structure of the water body, overcoming the limitations of traditional methods that cannot adapt to different water body stratification characteristics. Based on the depth position of the density gradient interface in the three-dimensional density field grid, and combined with the Richardson number to determine the stability of each depth layer, a precise sampling grid for the gradient is set above and below the density gradient interface. Sampling points with equal spacing are set in the uniform layers between the gradient layers to form a sampling depth sequence. This strategy increases the sampling density in stable stratification areas where pollutants easily accumulate, and reduces the number of sampling points in areas where pollutants are evenly distributed, achieving optimized allocation of sampling resources and accurate capture of the vertical distribution characteristics of pollutants. A multi-channel stratified water sampling device was lowered to the target depth in the sampling depth sequence. After correcting for cable tilt errors using a depth encoder and a triaxial tilt sensor, an electromagnetically driven valve was triggered. This depth-fixing mechanism eliminated depth deviations caused by water flow, ensuring the spatial positioning accuracy of water samples collected at each depth level. Simultaneously, three-dimensional coordinates, sampling time, and environmental parameters were recorded to generate a sample information set, providing comprehensive data support for subsequent correlation analysis of pollutant concentrations and environmental conditions. The water samples in the sample information set were filtered, separated, and subjected to multi-index detection to obtain the measured concentration values ​​of heavy metals, nutrients, and organic pollutants at each layer, constructing a pollutant concentration matrix. This matrix established a quantitative relationship between pollutant type, concentration, and depth location, laying a data foundation for in-depth analysis of the vertical distribution patterns of pollutants.

[0013] Based on the pollutant concentration matrix, the vertical concentration gradient vector and second-order difference are calculated to identify abrupt changes in pollutant concentration. This method quantitatively characterizes the vertical variation of pollutant concentration through mathematical operations, transforming ambiguous stratification phenomena into precisely localizable abrupt interfaces and revealing the accumulation mechanism of pollutants at physical barriers such as density gradients. The stratigraphic pollution load index is calculated to determine the main accumulation layers of pollutants. This index comprehensively assesses the multi-indicator pollution status of a single layer, identifying the most severely polluted critical layers and providing a scientific basis for the formulation of targeted remediation measures. A positive definite matrix factorization algorithm is used to perform source analysis on the pollutant concentration matrix, decomposing it into a source component matrix and a source contribution matrix. Through iterative optimization, the main pollution source factors and the characteristic pollutant composition, spatial distribution range, and relative contribution rate of each source factor are obtained. The application of this algorithm in the field of stratified water pollutant collection has achieved a leap from single pollutant concentration detection to comprehensive analysis of multi-source pollution. Through mathematical decomposition, complex mixed pollution signals are separated into pollution source factors with clear physical meaning. The characteristic pollutant composition of each source factor reflects the emission characteristics of the source, the spatial distribution range reveals the influence area of ​​the source, and the relative contribution rate quantifies the importance of the source to the overall pollution. The non-negativity constraint of the algorithm ensures the physical rationality of the decomposition results, and the iterative optimization process guarantees the mathematical optimality of the source analysis. This transforms pollution source tracing from qualitative speculation to quantitative calculation, providing solid technical support for pollutant source identification, migration path analysis, and pollution responsibility allocation. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a schematic diagram of one embodiment of the water pollutant stratification collection method based on a multi-functional monitoring vessel in this application. Detailed Implementation

[0016] This application provides a method for stratified collection of water pollutants based on a multi-functional monitoring vessel. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data used can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0017] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the water pollutant stratification collection method based on a multi-functional monitoring vessel in this application includes:

[0018] Step S1: Collect acoustic Doppler frequency shift signals and CTD probe data along the preset route, calculate the density stratification structure of the water body in the vertical direction according to the density gradient equation, identify the depth position of the density stratification interface, and generate a three-dimensional density field grid.

[0019] Step S2: Based on the depth position of the density jump layer interface in the three-dimensional density field grid, and combined with the Richardson number, determine the stability of each depth layer, set up a jump layer precision sampling grid above and below the density jump layer interface, and set up sampling points with equal spacing in the uniform layer between the jump layers to form a sampling depth sequence.

[0020] Step S3: Lower the multi-channel stratified water sampling device to the target depth in the sampling depth sequence. After correcting the cable tilt error by using a depth encoder and a three-axis tilt sensor, trigger the electromagnetic drive valve to collect water samples from each depth layer and record the three-dimensional coordinates, collection time and environmental parameters to generate a sample information set.

[0021] Step S4: Filter and separate the water samples in the sample information set and perform multi-index detection to obtain the measured concentration values ​​of heavy metals, nutrients and organic pollutants in each layer, and construct a pollutant concentration matrix;

[0022] Step S5: Calculate the vertical concentration gradient vector and second-order difference based on the pollutant concentration matrix, identify the pollutant concentration abrupt change interface, calculate the stratum pollution load index, and determine the main accumulation strata of pollutants and the spatial distribution of source factors.

[0023] It is understood that the implementing entity of this application can be a water pollutant stratification and collection system based on a multi-functional monitoring vessel, or it can be a terminal or a server; the specific implementation is not limited here. This application's embodiments use a server as an example for illustration.

[0024] Specifically, the stratification of water bodies is identified through the collaborative work of an acoustic Doppler current profiler and a CTD probe. The acoustic Doppler current profiler emits acoustic pulses into the water. When these pulses encounter suspended particles or plankton, they are reflected. Due to the different velocities of water layers at different depths relative to the instrument, the frequency of the reflected sound waves exhibits a Doppler frequency shift. The velocity vector of each depth layer is calculated using this frequency shift. CTD, an abbreviation for "Conductivity-Temperature-Depth," is a comprehensive sensor device used to simultaneously measure the conductivity, temperature, and depth of water. Conductivity data is used to calculate salinity, and temperature and salinity together determine water density. Depth data provides the vertical coordinates of the sampling location. The CTD probe simultaneously measures these three parameters: conductivity, temperature, and depth. Conductivity reflects salinity, while temperature and salinity together determine water density. The density gradient equation is based on two physical processes: thermal expansion and salinity contraction. The thermal expansion coefficient describes the degree of density decrease caused by an increase in temperature, while the salinity contraction coefficient describes the degree of density increase caused by an increase in salinity. The actual density value at each depth is calculated by multiplying the temperature gradient and salinity gradient by their respective coefficients, integrating the results, and then superimposing the surface water density benchmark value. When the density difference between two adjacent layers exceeds a set threshold, vertical mixing of the fluid at that location is strongly suppressed, forming a density-skipping interface. Kriging interpolation is a geostatistical method that performs optimal linear unbiased estimation of unknown locations based on the spatial correlation of known sampling points. The algorithm first calculates the semivariance function between sampling points, fits a theoretical semivariance model, then determines the interpolation weights based on the model, and finally generates a continuous three-dimensional density field grid. Each node in the grid stores the spatial coordinates and density value of that location.

[0025] Precise sampling locations are planned based on the identified density stratification interfaces. The Richardson number is a dimensionless parameter describing the stability of water stratification. The numerator is the square of the Brunt-Visala frequency, which characterizes the buoyancy oscillation frequency generated by density stratification. It is calculated by multiplying the gravitational acceleration by the vertical density gradient and then dividing by the reference density. The denominator is the square of the vertical gradient of the horizontal velocity, characterizing the intensity of shear turbulence. When the Richardson number exceeds a set threshold, the buoyancy stabilizing effect is stronger than the shear turbulence mixing effect, making it difficult for pollutants to migrate vertically at this layer, thus becoming a key area for pollutant accumulation. The precise sampling grid for the density stratification interface sets multiple sampling points above and below the interface to capture the abrupt change in pollutant characteristics at the interface. The upper sampling points are distributed from near to far from the interface, and the lower sampling points use the same strategy, forming a dense sampling array centered on the interface. The pollutant distribution in the homogeneous layers between the mesas is relatively gentle. A proportional spacing strategy is adopted, where the spacing between adjacent sampling points increases with the sampling point number. The spacing between adjacent sampling points is calculated by multiplying the base spacing by a power of the sampling point number. In this application, the power is set to 1, meaning the spacing between the nth sampling point and the (n-1)th sampling point is n times the base spacing. The spacing between sampling points closer to the mesas is smaller, while the spacing gradually increases further away from the mesas. This approach reduces the number of sampling points while ensuring sample representativeness. The generated sampling depth sequence includes the vertical distance of each sampling point from the water surface, arranged from shallowest to deepest.

[0026] This system enables precise depth determination and synchronous data acquisition for a multi-channel, stratified water sampling device. The fiber optic sensing circuitry embedded in the main control cable operates on the principle of total internal reflection. When the light signal propagates through the fiber, bending or stress on the cable alters the light transmission characteristics. By detecting these changes in the light signal, the cable's status and depth position are monitored in real time. The depth encoder employs absolute encoding technology, with multiple coding tracks engraved on the encoder disk. Each position corresponds to a unique coding value, which is read by a photoelectric sensor and converted into the cable release length. A three-axis tilt sensor incorporates a MEMS accelerometer to measure the gravitational acceleration components in three orthogonal directions. The cable's tilt angle is calculated based on the angle between the gravitational direction and the sensor's coordinate system. When the cable deviates from its vertical direction due to water flow, the release length may not equal the actual depth. This is corrected by multiplying the measured cosine of the tilt angle by the release length to arrive at the actual vertical depth. When the difference between the actual depth and the target depth in the sampling depth sequence is less than the set accuracy threshold, the control system applies a trigger current to the corresponding channel's electromagnetic drive valve. The electromagnetic coil generates a magnetic field that drives the valve core to move, and the valve fully opens within a set time. The pressure difference inside and outside the sampling bottle drives the water sample to fill the sampling bottle, and the valve automatically closes to form a seal after it is full. Each sampling bottle is equipped with a GPS positioning module to record latitude and longitude coordinates, a depth encoder to record depth coordinates, a real-time clock to record the acquisition time, and temperature and pressure sensors to record environmental parameters. All information is associated with the sample number and stored to form a sample information set.

[0027] Water samples undergo pretreatment and multi-index detection. The filtration process removes suspended particulate matter and plankton from the water sample using a filter membrane with a pore size smaller than the molecular size of the target analytes. Dissolved pollutants pass through the membrane into the filtrate, while particulate matter is retained on the membrane surface. After filtration, the water samples are packaged according to different detection requirements. For heavy metal detection, acidic reagents are added to lower the pH value to prevent metal ion hydrolysis and precipitation. For nutrient detection, preservatives are added to inhibit microbial activity and prevent sample deterioration. For organic pollutant detection, samples are sealed and refrigerated to inhibit volatilization and degradation. A portable inductively coupled plasma mass spectrometer (ICP-MS) atomizes the sample and introduces it into a high-temperature plasma. Element atoms are ionized to form charged ions, which are separated according to their mass-to-charge ratio in the mass spectrometer. The detector records the ion signal intensity at different mass-to-charge ratios; the signal intensity is proportional to the element concentration. The standard curve method uses a series of standard solutions of known concentrations to establish a linear relationship between signal intensity and concentration. After measuring the sample signal intensity, the blank signal intensity is subtracted, and the element concentration in the sample is calculated based on the slope of the standard curve. A continuous flow analyzer uses a peristaltic pump to mix the sample and chromogenic reagent in a fixed ratio, and a colorimetric reaction occurs in a constant-temperature reaction tube. The reaction product passes through a colorimetric cell, and the absorbance at a specific wavelength is measured by a detector. According to the Lambert-Beer law, absorbance is proportional to concentration, and the nutrient concentration is calculated. Solid-phase extraction (SPE) utilizes the selective adsorption of target analytes by packing material in a solid-phase extraction column. When a water sample passes through the extraction column, organic pollutants are adsorbed onto the surface of the packing material, while water and other interfering substances elute. The target analytes are then eluted with an organic solvent, and the eluent is concentrated by nitrogen blowing before entering a gas chromatography-mass spectrometry (GC-MS) system. GC separates different compounds based on their boiling points and polarity differences, while mass spectrometry identifies and quantifies compounds based on the mass-to-charge ratio of molecular ions and fragment ions. The measured concentration values ​​of each pollutant at each layer are arranged according to sampling depth and pollutant type to construct a pollutant concentration matrix.

[0028] Pollution stratification characteristics are calculated using a pollutant concentration matrix. The concentration gradient vector represents the rate of change of pollutant concentration in the vertical direction. It is calculated by dividing the concentration difference between two adjacent layers by the depth difference. A positive gradient value indicates that the concentration increases with increasing depth, while a negative gradient value indicates that the concentration decreases with increasing depth. The absolute value of the gradient reflects the degree of concentration change. Second-order difference calculation is performed on the concentration gradient vector again to reflect the rate of gradient change. When the absolute value of the second-order difference is large, it indicates that there is an abrupt change in the gradient within that depth interval, and the pollutant concentration distribution has reached an inflection point. The standard deviation of the second-order difference for all depth intervals is calculated. The standard deviation reflects the dispersion of the second-order difference. Multiplying the standard deviation by a set factor serves as a judgment threshold. Depth intervals with absolute values ​​of second-order differences exceeding the threshold are identified as interfaces of abrupt changes in pollutant concentration. The stratum pollution load index comprehensively assesses the multi-indicator pollution status of a single stratum. It is calculated by summing the ratios of the concentrations of each pollution indicator to the corresponding environmental quality standard limits and dividing by the total number of indicators. A ratio greater than 1 indicates that the indicator exceeds the standard, and the larger the ratio, the more severe the exceedance. The stratum with the largest index value is the main accumulation layer of pollutants. The positive definite matrix factorization algorithm decomposes the pollutant concentration matrix into the product of a source component matrix and a source contribution matrix. Each column of the source component matrix represents the characteristic spectrum of a pollution source, describing the relative composition of various pollutants emitted by that source. Each row of the source contribution matrix represents the contribution of a pollution source at each sampling point. The algorithm iteratively optimizes to minimize the difference between the reconstructed matrix and the original concentration matrix. During the iteration process, the elements of both the source component matrix and the source contribution matrix are kept non-negative. Finally, the main pollution source factors are identified. Each source factor includes the characteristic pollutant composition, spatial distribution range, and relative contribution rate. The characteristic pollutant composition is determined by the proportion of elements in the corresponding column of the source component matrix. The spatial distribution range is determined by the position of the non-zero element in the corresponding row of the source contribution matrix. The relative contribution rate is calculated by the proportion of the contribution of the factor to the total contribution.

[0029] In one specific embodiment, step S1 includes:

[0030] The acoustic Doppler current profiler is controlled to travel along a preset route at a set speed, and emits sound pulses of a set frequency at set intervals. It receives Doppler frequency shift signals reflected back from different depth layers of the water body and calculates the current velocity vector data within a set water depth range.

[0031] Temperature, conductivity, and depth data are collected synchronously using a CTD probe at a set sampling frequency. The temperature and conductivity data are substituted into the density gradient equation, and combined with the surface water density, salinity contraction coefficient, and thermal expansion coefficient, the density stratification structure of the water body in the vertical direction is calculated.

[0032] Determine whether the density difference between two adjacent layers exceeds a set threshold, identify the depth location where there is a significant density jump layer, and obtain the depth coordinates of each density jump layer interface;

[0033] Based on the collected spatial point data, a three-dimensional density field grid with a set resolution is generated by the Kriging interpolation algorithm, and the three-dimensional density field grid is stored in the form of a matrix containing spatial coordinates and density values.

[0034] Specifically, the acoustic Doppler current profiler is installed on the bottom of the multi-functional monitoring vessel. The monitoring vessel travels at a constant speed along a pre-planned route. The onboard control system determines the vessel's position based on GPS positioning. Whenever the vessel moves to a preset distance interval, the control system triggers the acoustic Doppler current profiler to emit acoustic pulses underwater. The frequency of the acoustic pulses is set according to the water depth and monitoring requirements. When the sound waves propagate in the water and encounter suspended particles or plankton, they are scattered and reflected. Due to the difference in motion speed between different depth layers of the water relative to the instrument, the frequency of the reflected sound waves shifts relative to the transmission frequency. The receiving transducer array captures the Doppler frequency shift signals returned from different depth layers. The signal processing unit extracts the frequency shift amount of each depth layer. According to the Doppler effect principle, the frequency shift amount is proportional to the radial velocity. Combining the transmission frequency, sound speed, and frequency shift amount, the radial velocity of each depth layer is calculated. Then, the horizontal and vertical velocity components are obtained through multi-beam measurement and combined to form three-dimensional velocity vector data. The CTD probe is fixed to the water sampling device and lowered with it. The temperature sensor inside the probe uses a thermistor or platinum resistance thermometer to measure the water temperature. The conductivity sensor reflects the salinity of the water by measuring the conductivity of the water sample between two electrodes. The pressure sensor calculates the depth based on the hydrostatic pressure. The three sensors synchronously collect data according to the set sampling frequency and transmit it to the data processing unit in real time. The density gradient equation is based on the seawater state equation. Increased temperature causes the water to expand thermally and decrease in density, while increased salinity causes the water to contract and increase in density. The data processing unit subtracts the reference temperature from the measured temperature value to obtain the temperature deviation. Multiplying the temperature deviation by the coefficient of thermal expansion, it obtains the effect of temperature on density. At the same time, it converts the salinity value based on the conductivity data, subtracts the reference salinity value to obtain the salinity deviation, and multiplies the salinity deviation by the salinity contraction coefficient to obtain the effect of salinity on density. Using the surface water density as a reference value, the temperature and salinity effects are superimposed to calculate the actual density value at each depth, forming density vertical profile data. The data processing unit traverses the density profile data in depth order, extracts the density values ​​of two adjacent depth layers, calculates the difference between the density of the upper layer and the density of the lower layer, compares the absolute value of the difference with a preset density difference threshold, and when the density difference exceeds the threshold, it indicates that there is a significant density discontinuity at that location, the vertical mixing of the fluid is significantly suppressed, it is determined to be a density jump interface, and the depth coordinate value of the interface is recorded.Kriging interpolation is a spatial statistical interpolation method. The algorithm first calculates the spatial distance between known sampling points, then calculates the variance of the sampling point density values ​​at different distance intervals, and plots the semivariogram function. The semivariogram increases with distance, reflecting spatial autocorrelation. A suitable theoretical model is selected to fit the semivariogram function; commonly used models include the spherical model, exponential model, and Gaussian model. After determining the fitting parameters, for each grid node to be interpolated, the algorithm calculates the distance from that node to each known sampling point. Based on the semivariogram model, it determines the weight coefficient of each sampling point to that node; the weight coefficient decreases with distance. The density value of each sampling point is multiplied by its corresponding weight coefficient and summed to obtain the interpolated density value of that grid node. This process is repeated for all grid nodes in the 3D space to complete the interpolation. The generated 3D density field grid is divided according to a set horizontal and vertical resolution. Each grid node stores longitude, latitude, and depth coordinates and the corresponding density interpolation result. The data is stored in matrix form, with row indices corresponding to depth layers and column indices corresponding to horizontal positions. The matrix element value represents the value at that position. Density values ​​are obtained by analyzing the surface flow velocity of a monitored water area during sampling. Acoustic Doppler current profilers detect a higher surface flow velocity, which decreases significantly below five meters. Temperature data collected simultaneously by the CTD probe shows that the surface water temperature is relatively high and gradually decreases with depth. Conductivity data shows that salinity increases rapidly between four and six meters. The data processing unit substitutes the temperature and conductivity data into the density gradient equation, using the surface water density as the initial baseline. The density increment caused by the temperature decrease is superimposed with the density increment caused by the salinity increase to calculate the density value of each depth layer. When the density difference between adjacent layers exceeds a set threshold at a depth of five meters, this depth is identified as a density gradient interface. Density data from multiple spatial sampling points are input into the Kriging interpolation algorithm. The algorithm assigns interpolation weights to the grid nodes in three-dimensional space based on the spatial distribution characteristics of the sampling points and the correlation of density values, generating a three-dimensional density field grid covering the entire monitoring area. The density gradient interface in the grid appears as a dense region of density isosurfaces. The three-dimensional visualization clearly reflects the water stratification structure and the spatial distribution characteristics of the gradient.

[0035] In one specific embodiment, step S2 includes:

[0036] Read the depth position of each density hopping layer interface identified in the three-dimensional density field grid, and extract the horizontal and vertical velocity components of each depth layer in the velocity profile data;

[0037] Richardson number for each depth layer is calculated based on the vertical gradient of the horizontal flow velocity and the Brunt-Vesala frequency. When the Richardson number is greater than a set stability threshold, the corresponding depth layer is determined to be a strongly stable layer region.

[0038] Multiple sampling points are set above and below the density gradient interface to form a gradient precision sampling grid. The sampling points are located at different distances above the interface and different distances below the interface.

[0039] Sampling points are set up in the uniform layer between the layers according to the equal spacing strategy. The spacing between each sampling point is calculated based on the basic spacing and the sampling point number in the layer, and a sampling depth sequence containing all sampling depths is generated.

[0040] Specifically, the data processing unit reads the depth location data of the identified density stratification interfaces from the three-dimensional density field grid matrix, and simultaneously extracts the velocity information of each depth layer corresponding to the density grid from the velocity profile database collected by the acoustic Doppler current profiler. The velocity vector contains three directional components. The data processing unit decomposes the velocity vector, extracting the east-west and north-south velocity components in the horizontal plane. The two horizontal components are combined to obtain the horizontal velocity. At the same time, the vertical velocity components, either upward or downward, are extracted. The horizontal velocity reflects the horizontal transport capacity of the water body, and the vertical velocity reflects the vertical mixing intensity of the water body. The Brunt-Visara frequency characterizes the buoyancy oscillation frequency generated by density stratification. In the calculation, the density values ​​of adjacent depth layers are first extracted, the density difference is calculated and divided by the depth difference to obtain the vertical density gradient, the gravitational acceleration is multiplied by the vertical density gradient and then divided by the reference density value at that depth, and the square root of the result is the Brunt-Visara frequency. The larger the frequency value, the stronger the density stratification and the greater the buoyancy restoring force. The method for calculating the vertical gradient of horizontal velocity is to extract the horizontal velocity values ​​of adjacent depth layers, subtract the horizontal velocity of the lower layer from the horizontal velocity of the upper layer to obtain the velocity difference, divide the velocity difference by the depth difference between the two layers to obtain the rate of change of horizontal velocity in the vertical direction. The gradient value reflects the intensity of shearing action. Shearing action produces turbulent mixing that tends to disrupt density stratification. Richardson number calculation uses the square of the Brunt-Visala frequency as the numerator, which represents the strength of buoyancy stabilization, and the square of the vertical gradient of the horizontal velocity as the denominator, which represents the strength of shear mixing. Dividing the numerator by the denominator yields the Richardson number, a dimensionless parameter whose value reflects the relative strength of buoyancy stabilization and shear mixing. After calculating the Richardson number for each depth layer, the data processing unit compares it with a set stability threshold. When the Richardson number for a certain depth layer is greater than the threshold, it indicates that the buoyancy stabilization effect at that layer is much stronger than the shear mixing effect, and vertical fluid exchange is significantly suppressed. Pollutants are difficult to diffuse upward or downward across this layer, and the depth layer is determined to be a strongly stable layer and its attributes are marked. The precision sampling grid for density gradient interfaces is set up specifically for density gradient interfaces. The data processing unit reads the depth coordinates of the density gradient interface and sets multiple sampling points above the interface depth value. The first sampling point is at a smaller vertical distance from the interface, the second sampling point is at a larger vertical distance from the interface than the first sampling point, and subsequent sampling points are arranged in ascending order. The sampling points below the interface adopt a symmetrical setting strategy, with the first sampling point located at a smaller distance below the interface and subsequent sampling points being at progressively larger vertical distances from the interface. The number of sampling points above and below the interface is determined according to the thickness of the gradient and the monitoring accuracy requirements. Multiple sampling points form a dense sampling array centered on the interface. The spacing between sampling points is smaller near the interface and gradually increases further away from the interface. The precision sampling grid for density gradient interfaces can accurately capture the abrupt changes and gradient distribution patterns of pollutant concentrations near the gradient interface.The density distribution of the uniform layers between the mesolayers is relatively uniform, and the pollutant concentration changes relatively smoothly. A proportional spacing strategy is adopted for sampling points, where the distance between adjacent sampling points increases according to the sampling point number. The data processing unit first determines the basic spacing as an initial reference value. The distance between the first sampling point within a layer and the mesolayer interface is equal to the basic spacing; the distance between the second sampling point and the first sampling point is equal to the basic spacing multiplied by the sampling point number within the layer; the distance between the third sampling point and the second sampling point is equal to the basic spacing multiplied by the sampling point number. The number increases with depth, and the sampling point spacing increases with the number. The sampling point spacing is smaller near the mesolayer to ensure sampling resolution in the transition area, while the spacing is larger in the uniform layers farther from the mesolayer. The central area has a larger spacing between sampling points to reduce redundant sampling. The data processing unit merges the depth values ​​of all sampling points in the precision sampling grid with the depth values ​​of the uniformly spaced sampling points, arranging them in order of depth from shallow to deep to generate a complete sampling depth sequence. Each element in the sequence corresponds to the vertical depth value of one sampling point from the water surface. When a water body monitoring task is carried out, the three-dimensional density field grid shows a clear density gradient interface at a depth of 5 meters, and a significant difference between the density values ​​at a depth of 4 meters and 6 meters. The velocity profile data shows that the horizontal velocity is relatively high from the surface to a depth of 4 meters, decreases sharply from 4 meters to 6 meters, and tends to stabilize below 6 meters. The data processing unit extracts various... The horizontal velocity value at depth is calculated. The difference in horizontal velocity between depths of 4 meters and 5 meters is divided by the depth interval between the two layers to obtain the vertical gradient of the horizontal velocity. Simultaneously, the vertical gradient of density is calculated based on the density values ​​at depths of 4 meters and 5 meters. The vertical gradient of density is multiplied by the acceleration due to gravity, divided by the reference density value, and the square root is taken to obtain the Brundt-Vesala frequency. The square of the Brundt-Vesala frequency is divided by the square of the vertical gradient of horizontal velocity to obtain the Richardson number of that depth layer. If the Richardson number is greater than the set stability threshold, the area near depth 5 meters is determined to be a strongly stable stratification region. The data processing unit sets sampling points above the 5-meter interface. The first sampling point is located at depth 4.8 meters, the second sampling point is located at depth 4.5 meters, and the third sampling point is located at... Sampling points are set at a depth of 4.2 meters below the interface. The first sampling point is located at a depth of 5.2 meters, the second sampling point at a depth of 5.5 meters, and the third sampling point at a depth of 5.8 meters, forming a hierarchical precision sampling grid. Points are arranged at equal intervals in the uniform layer from the surface to a depth of 4 meters. After the basic spacing is set, the depth of the first sampling point is determined according to the basic spacing. For the nth sampling point (n is greater than or equal to 2), the spacing between it and the (n-1)th sampling point is the basic spacing multiplied by n. For example, the spacing of the second sampling point is the basic spacing multiplied by 2, and the spacing of the third sampling point is the basic spacing multiplied by 3. The same strategy is used to set sampling points in the uniform layer from a depth of 6 meters to the bottom layer. All sampling point depth values ​​are merged and sorted to generate a sampling depth sequence.

[0041] In one specific embodiment, step S3 includes:

[0042] The multi-channel stratified water sampling device is lowered via a main control cable, and the main control cable is embedded with an optical fiber sensing line to transmit depth position signals in real time.

[0043] The actual depth position of the device is calculated by measuring the cable release length using a depth encoder, combining the cable tilt angle measured by a triaxial tilt sensor, correcting the cable tilt error caused by water flow, and calculating the actual depth position of the device.

[0044] When the difference between the actual depth position of the device and the target depth in the sampling depth sequence is less than the set accuracy threshold, a trigger current is sent to the electromagnetic drive valve of the corresponding channel to open the valve and allow the water sample to fill the sampling bottle under the drive of environmental pressure.

[0045] Record the longitude, latitude, depth, sampling time, ambient temperature, and in-situ pressure of each sampling bottle at the time of sampling completion, and generate a sample information set containing sample number and multidimensional information.

[0046] Specifically, the multi-channel stratified water sampling device is connected to the winch system on the deck of the monitoring vessel via a main control cable. Fiber optic sensing lines are embedded inside the main control cable. These lines operate based on the principle of total internal reflection of light. When the light signal is transmitted through the fiber, it is sensitive to changes in the cable's stress, bending, and temperature. When the cable is deformed or bent, the refractive index distribution inside the fiber changes, causing measurable changes in the phase, intensity, or wavelength of the light signal. The photoelectric detection unit continuously monitors the characteristic parameters of the light signal and converts the changes in the light signal into cable depth position information. The fiber optic sensing lines transmit the depth position signal to the shipboard data processing unit in real time. The data processing unit receives and records the depth change trajectory of the device. A depth encoder is mounted on the winch's drum shaft. The encoder employs absolute photoelectric encoding technology, with multiple precision coding tracks engraved on the encoder disk surface. Each angular position corresponds to a unique binary code value. As the drum rotates, a photoelectric sensor array reads the coding pattern on the encoder disk, converting the optical signal into a digital code signal. The control unit calculates the number of drum rotations and the angle based on the code value, and combines this with the drum's circumference to calculate the cumulative length of the released cable. The cable release length data is transmitted to the data processing unit in real time. A three-axis tilt sensor is fixed to the frame of the water sampling device. The sensor integrates a microelectromechanical system (MEMS) accelerometer. The accelerometer contains three orthogonal sensitive axes, measuring the gravitational acceleration components in the east-west, north-south, and vertical directions, respectively. The projection ratio of the gravity vector along the three axes reflects the tilt angle of the device relative to the horizontal plane and the vertical line. The data processing unit receives the acceleration component values ​​along the three axes and calculates the ratio of the vertical axis acceleration component to the total gravitational acceleration. The inverse cosine of this ratio is the tilt angle of the cable deviating from the vertical direction. The tilt angle data is collected synchronously with the cable release length data. The water flow exerts a lateral drag force on the cable, causing it to deviate from the vertical direction and form an arc. The cable release length is equal to the length of the arc, while the actual depth of the device is equal to the vertical distance between the device and the water surface. The data processing unit extracts the cable release length and tilt angle values, multiplies the release length by the cosine of the tilt angle (which represents the cosine of the angle between the tilt direction and the vertical direction), and the product is the actual vertical depth position of the device. This correction calculation eliminates the horizontal offset error caused by the water flow, obtaining the accurate depth coordinates of the device.The data processing unit continuously compares the actual depth position of the device with the next target depth value in the sampling depth sequence, calculates the difference between the two and takes the absolute value. When the absolute value of the difference is less than the preset accuracy threshold, it is determined that the device has reached the target sampling depth. The data processing unit immediately sends a trigger command to the electromagnetic drive valve control circuit of the corresponding depth channel in the multi-channel stratified water sampling device. The control circuit converts the low-pressure control signal into the working current of the solenoid valve. The trigger current flows through the solenoid coil to generate a magnetic field. The magnetic field acts on the permanent magnet or magnetic material in the valve core, driving the valve core to move axially against the preload of the return spring. The movement of the valve core causes the valve to change from the closed state to the open state. After the valve opens, the inside of the sampling bottle is connected to the external water body. The pressure inside the sampling bottle is lower than the external ambient pressure. The pressure difference drives the water sample to flow from the valve inlet into the inner cavity of the sampling bottle. The water sample continues to flow until the pressure inside the sampling bottle is balanced with the external pressure. After the sampling bottle is completely filled, the control circuit cuts off the current of the solenoid coil. The valve core returns to the initial position under the action of the return spring, and the valve closes to achieve a seal. The water sample in the sampling bottle is sealed and preserved.The GPS positioning module is installed on the upper part of the water sampling device. It receives positioning signals from multiple satellites and calculates the device's longitude and latitude coordinates by measuring the time difference of signal propagation. The longitude coordinates reflect the device's geographical location in the east-west direction, and the latitude coordinates reflect its geographical location in the north-south direction. The actual depth position calculated by the depth encoder and tilt sensor is used as the depth coordinate value. The real-time clock chip provides the timestamp information of the sampling time, including year, month, day, hour, minute, and second. The temperature sensor measures the ambient water temperature at the sampling point, and the pressure sensor measures the in-situ hydrostatic pressure at the sampling point. The in-situ pressure reflects the water column pressure and water density information at the sampling depth. The data acquisition unit synchronously reads the longitude and latitude data from the GPS module, the depth data from the depth encoder, the time data from the clock chip, the temperature data from the temperature sensor, and the pressure data from the pressure sensor the moment each sampling bottle valve is closed. It associates and stores the unique number of the sampling bottle with the above multi-dimensional data, generating a sample information record containing the sample number, longitude coordinates, latitude coordinates, depth coordinates, sampling time, ambient temperature, and in-situ pressure. The information records from multiple sampling bottles are summarized to form a complete sample information set. When a water pollutant sampling operation is carried out... During the sampling depth sequence, the first target depth is 2 meters below the surface. The winch releases the cable to a length of 2.3 meters as indicated by the depth encoder reading. The triaxial tilt sensor measures the cosine value corresponding to the cable's tilt angle. Multiplying the release length by the cosine value yields an actual depth of 2.0 meters. Since the absolute value of the difference between the actual depth and the target depth is less than the set accuracy threshold, the control unit sends a trigger current to the first channel solenoid valve. After the valve opens, external water flows into the sampling bottle under pressure. Once the sampling bottle is full, the valve closes. The GPS module records the longitude and latitude coordinates to determine the horizontal position of the sampling point. The location, with an actual depth of 2.0 meters as the depth coordinate, is recorded by a clock chip at the exact moment the data acquisition is completed. A temperature sensor measures the water temperature at that depth, and a pressure sensor measures the hydrostatic pressure at that depth. The data acquisition unit packages and stores the number of the first sampling bottle along with the longitude, latitude, depth, time, temperature, and pressure data to form the first sample information record. The device continues to descend to the next target depth and performs the same fixed-depth sampling and data recording process until all target depths in the sampling depth sequence have been sampled and recorded. The information records of all sampling bottles are then summarized to form a sample information set.

[0047] In one specific embodiment, step S4 includes:

[0048] The water sample in the sample information set is filtered through a filter membrane with a set pore size to remove suspended particulate matter. The filtered water sample is then divided into multiple portions and different preservation reagents are added to each portion.

[0049] The first water sample was tested for heavy metals using a portable inductively coupled plasma mass spectrometer. The heavy metal concentration was calculated using the standard curve method based on the measured signal intensity, blank signal intensity, and the slope of the standard curve.

[0050] The second water sample was analyzed using a continuous flow analyzer with colorimetric and reduction methods to determine the concentrations of total phosphorus, ammonia nitrogen, and nitrate nitrogen.

[0051] Organic pollutants were detected in the third water sample using solid-phase extraction-gas chromatography-mass spectrometry (SPE-GC-MS) to obtain the concentration values ​​of polycyclic aromatic hydrocarbons (PAHs) and polychlorinated biphenyls (PCBs). The measured concentration values ​​of each pollutant index at each layer were then summarized to construct a pollutant concentration matrix.

[0052] Specifically, the water samples from the sample information collection are first subjected to solid-liquid separation using a vacuum filtration device or a pressure filtration device. The filter membrane material is usually polycarbonate or polyethersulfone, and the pore size is selected according to the detection requirements. A commonly used pore size is 0.45 micrometers. This pore size can effectively trap suspended particulate matter in the water sample, including silt particles, phytoplankton cells, and suspended organic debris. The molecular size of dissolved pollutants is much smaller than the filter membrane pore size, allowing dissolved heavy metal ions, nutrient ions, and organic pollutant molecules to pass smoothly through the filter membrane into the filtrate. Particulate matter is intercepted on the filter membrane surface. After filtration, the filtrate is collected as the water sample to be tested. The filtered water sample is then aliquoted into multiple sample bottles according to the preservation requirements of different detection items. The first water sample is used for heavy metal detection, and nitric acid or salt is added to the sample bottle. Acidic reagents, such as acids, lower the pH of the water sample to below 2. The acidic environment inhibits the hydrolysis and precipitation reactions of heavy metal ions, preventing them from precipitating out as hydroxides. Simultaneously, the acidic conditions reduce the adsorption loss of heavy metal ions on the container walls. The second water sample is used for nutrient detection. Preservatives, such as sodium azide, are added to the sample bottle. Sodium azide inhibits the respiratory and metabolic activities of microorganisms in the water sample. Microbial metabolism consumes or produces nutrients, causing concentration changes. After the preservative is added, the nutrient concentration of the water sample remains stable during transportation and storage. The third water sample is used for organic pollutant detection. The sample bottle is sealed and placed in a low-temperature refrigerator. The low temperature environment reduces the volatilization and degradation rates of organic pollutants, maintaining their original concentration levels. Portable inductively coupled plasma mass spectrometers (ICP-MS) consist of a nebulization system, a plasma torch, a mass analyzer, and a detector. The nebulization system converts liquid water samples into fine droplets to form aerosols. These aerosols are transported to the plasma torch by a carrier gas. The torch temperature reaches 6000 to 10000 degrees Celsius, causing the water to evaporate instantly and the solute molecules to break down into atoms. These atoms lose their outer electrons at high temperatures to form positive ions. The ion beam enters the vacuum chamber through a sampling cone and a retrieval cone. An ion lens system focuses the ion beam and guides it to the mass analyzer. The mass analyzer uses quadrupole or time-of-flight technology to separate ions based on their mass-to-charge ratio. Ions with different mass-to-charge ratios move along different trajectories under specific electric or magnetic fields and arrive at the detector sequentially. The detector records the number of ions corresponding to each mass-to-charge ratio, and the number of ions is converted into a signal intensity value. The standard curve method is used to establish a quantitative relationship between concentration and signal intensity. A series of heavy metal standard solutions with known concentrations are prepared, with a concentration gradient covering the expected concentration range of the water sample to be tested. Each standard solution is sequentially introduced into the mass spectrometer to measure the signal intensity. A scatter plot is plotted with the standard solution concentration as the abscissa and the measured signal intensity as the ordinate. A straight line is fitted using the least squares method, and the slope of the line represents the signal intensity increment corresponding to a unit concentration increment. The blank solution is pure water or reagent blank without the analyte. The signal intensity of the blank solution is measured as the background signal intensity. The background signal originates from instrument noise and trace impurities in the reagent. The net signal intensity is obtained by subtracting the blank signal intensity from the measured signal intensity of the water sample to be tested.The concentration of heavy metals in the water sample is calculated by dividing the net signal intensity by the slope of the standard curve. This measurement and calculation process is repeated for water samples at each depth level to obtain heavy metal concentration data for each layer. A continuous flow analyzer uses a multi-channel peristaltic pump to continuously pump water samples and chemical reagents into a mixing pan at a preset flow rate ratio. After thorough mixing, the water sample and reagents enter a thermostatic reaction tube, where the temperature is controlled at the optimal reaction temperature. Total phosphorus detection uses a potassium persulfate oxidation digestion method combined with the molybdenum blue colorimetric method. Potassium persulfate oxidizes and decomposes organic phosphorus and condensed phosphorus in the water sample into orthophosphate at high temperatures. Orthophosphate reacts with ammonium molybdate to form phosphomolybdic acid, which is then reduced by ascorbic acid to a blue molybdenum blue complex. Ammonia nitrogen detection uses a hypochlorite oxidation method combined with the indophenol blue colorimetric method. Ammonia in the water sample reacts with hypochlorite to form chloramine, which condenses with phenol under alkaline conditions to form the blue indophenol dye. Nitrate nitrogen detection uses a cadmium column reduction method. The water sample passes through a cadmium-copper alloy reduction column, where nitrates are reduced. The nitrite reacts with p-aminobenzenesulfonamide in a diazotization reaction. The diazo compound then couples with naphthylethylenediamine to form a purple-red azo dye. The colored products generated by each reaction flow through a colorimetric cell. A light source and a photodetector are installed at both ends of the colorimetric cell. The light source emits monochromatic light of a specific wavelength that passes through the colorimetric cell. The colored products absorb some of the light energy, reducing the intensity of the transmitted light. The photodetector measures the intensity of the transmitted light. According to Beer-Lambert's law, absorbance is equal to the logarithm of the ratio of incident light intensity to transmitted light intensity. Absorbance is directly proportional to the concentration of the colored product in the solution, which in turn is directly proportional to the concentration of the nutrient to be measured. The instrument is pre-calibrated using a standard nutrient solution of known concentration to establish a calibration curve between absorbance and concentration. After measuring the absorbance of the water sample, the concentrations of total phosphorus, ammonia nitrogen, and nitrate nitrogen in the water sample are calculated based on the calibration curve. Solid-phase extraction (SPE) technology uses an extraction column packed with adsorbent to enrich organic pollutants in a water sample. The extraction column packing is typically C18 reversed-phase silica gel. The surface of the C18 packing is bonded with long octadecyl chains, exhibiting strong hydrophobicity. The water sample passes through the extraction column at a constant flow rate. Non-polar organic pollutants such as polycyclic aromatic hydrocarbons (PAHs) and polychlorinated biphenyls (PCBs) are retained by the adsorbent due to hydrophobic interactions, while polar water molecules and inorganic salt ions are not adsorbed and flow out directly. After enrichment by the extraction column, a small amount of organic solvent such as methanol or dichloromethane is used for elution. The organic solvent disrupts the hydrophobic interactions between the organic pollutants and the packing, causing the organic pollutants to dissolve in the eluent and be flushed out of the extraction column. The eluent volume is much smaller than the original water sample volume, resulting in enrichment of organic pollutants. The eluent is concentrated using a nitrogen evaporator, where a nitrogen stream purges the surface of the eluent under heating conditions, causing rapid evaporation of organic solvents. The concentrated sample is then injected into a gas chromatography-mass spectrometry (GC-MS) instrument. The GC column is packed with a stationary phase, and the sample passes through the column propelled by a carrier gas. Different organic compounds have different partition coefficients between the mobile and stationary phases; compounds with low boiling points and low polarity elute first, while compounds with high boiling points and high polarity elute later. The column outlet is connected to a mass spectrometry ion source, where the separated compound molecules are bombarded and ionized by electrons.Molecules lose electrons to form molecular ions, which further fragment to produce characteristic fragment ions. A mass analyzer separates various ions based on their mass-to-charge ratio (MTR). A detector records the ion signal intensity at different time points with different MTRs, generating a total ion current chromatogram and a mass spectrum. Compound types are identified by chromatographic retention time and mass spectrometric fragment ion characteristics. Concentrations are quantitatively calculated using chromatographic peak areas, yielding the concentration values ​​of each component of polycyclic aromatic hydrocarbons (PAHs) and polychlorinated biphenyls (PCBs). The data processing unit reads the concentration detection results of each pollution indicator at each depth layer, arranging the data according to depth layer as the row index and pollution indicator type as the column index. The first row of the matrix stores the concentrations of each pollution indicator in the surface water sample, the second row stores the concentrations of each pollution indicator in the second depth layer, and so on until the last row stores the concentrations of each pollution indicator in the deepest layer. The first column of the matrix stores the concentration of a specific heavy metal element in each layer, the second column stores the concentration of another heavy metal element in each layer, and subsequent columns store nutrient concentrations and organic pollutant concentrations, respectively. The matrix element values ​​are the measured concentration values ​​of the corresponding pollution indicator at the corresponding depth, thus constructing a pollutant concentration matrix. The array contains concentration information for all depth layers and all pollution indicators. When a water sampling operation is performed, the surface water sample is filtered and then divided. The first sample is treated with nitric acid to adjust the pH value for heavy metal detection. The mass spectrometer measures the signal intensity of this sample, and the heavy metal concentration is obtained by subtracting the blank signal intensity and dividing by the slope of the standard curve. The second sample is treated with sodium azide as a preservative for nutrient detection. The sample reacts with a colorimetric reagent to generate a colored product, and the absorbance value is measured in the colorimetric cell. The concentration corresponding to the absorbance according to the calibration curve is the nutrient concentration. The third sample... After refrigeration, water samples were used for organic matter detection. Organic pollutants were enriched using a solid-phase extraction column. The eluent was concentrated and then injected into a gas chromatography-mass spectrometry (GC-MS) system. The chromatographic column separated the components, and the mass spectrometer recorded the signals of each component. The concentrations of polycyclic aromatic hydrocarbons (PAHs) and polychlorinated biphenyls (PCBs) were calculated from the peak area versus concentration curve. Water samples from other depths were processed using the same method to obtain the concentrations of heavy metals, nutrients, and organic pollutants at each depth. All concentration data were then entered into the corresponding positions in a matrix according to depth and index type, completing the construction of the pollutant concentration matrix.

[0053] In one specific embodiment, step S5 includes:

[0054] Based on the pollutant concentration values ​​and depth differences between adjacent layers in the pollutant concentration matrix, calculate the concentration gradient vector of each pollution index in the vertical direction;

[0055] Perform a second-order difference operation on the concentration gradient vector. When the absolute value of the second-order difference exceeds a set multiple of the gradient standard deviation, identify the corresponding depth interval as the interface of sudden change in pollutant concentration.

[0056] Based on the ratio of the concentration values ​​of each pollution index at each stratum to the corresponding environmental quality standard limit, the stratum pollution load index is calculated, and the stratum with the largest pollution load index is determined as the main accumulation stratum of pollutants.

[0057] The pollutant concentration matrix is ​​analyzed by positive definite matrix factorization algorithm, which decomposes it into source component matrix and source contribution matrix. Through iterative optimization, the main pollution source factors and the characteristic pollutant composition, spatial distribution range and relative contribution rate of each source factor are obtained.

[0058] Specifically, the data processing unit extracts the pollutant concentration values ​​of two adjacent layers from the pollutant concentration matrix in depth order, calculates the concentration gradient for each pollution index, subtracts the concentration value of the lower layer from the concentration value of the upper layer to obtain the concentration difference, and divides the concentration difference by the depth difference between the two layers to obtain the concentration gradient for that depth interval. A positive concentration gradient indicates that the pollutant concentration increases with increasing depth, while a negative value indicates that the concentration decreases with increasing depth. The magnitude of the absolute value of the gradient reflects the degree of concentration change in the vertical direction. After calculating the concentration gradient for each depth interval of each pollution index, the gradient values ​​are arranged in depth order to form a concentration gradient vector, and each element of the vector corresponds to the gradient value of a depth interval. The second-order difference operation performs a second difference calculation on the concentration gradient vector, extracting the gradient values ​​of adjacent depth intervals in the gradient vector. The second-order difference value is obtained by subtracting the gradient value of the upper layer from the lower layer gradient value. The second-order difference reflects the rate of change of the gradient itself. When the gradient increases or decreases, the second-order difference becomes non-zero; the more drastic the gradient change, the larger the absolute value of the second-order difference. The data processing unit statistically analyzes the second-order difference values ​​for all depth intervals, calculates the arithmetic mean of all second-order difference values ​​as the central trend, calculates the deviation of each second-order difference value from the mean, squares the deviations, sums them, and divides by the total number of depth intervals to obtain the variance. The square root of the variance yields the standard deviation, which quantifies the dispersion and fluctuation of the second-order difference. Multiplying the standard deviation by a set factor yields the second-order difference judgment threshold. Based on the sensitivity requirements, the larger the multiplier, the stricter the judgment standard, and the fewer but more significant abrupt transition interfaces are identified. Conversely, the smaller the multiplier, the more lenient the judgment standard, and the more abrupt transition interfaces are identified, but the risk of noise interference increases. The data processing unit traverses the second-order difference values ​​of each depth interval, calculates the absolute value of the second-order difference values ​​to eliminate the influence of positive and negative signs, and compares the absolute value of the second-order difference with the judgment threshold. When the absolute value of the second-order difference exceeds the judgment threshold, it is determined that abrupt change has occurred in the pollutant concentration gradient of that depth interval. This depth interval is marked as the pollutant concentration abrupt change interface, and the starting depth and ending depth are recorded. The abrupt change interface reflects the inflection point of the pollutant concentration distribution within that depth range. The inflection point location corresponds to the density jump layer blocking the diffusion of pollutants, direct emission from pollution sources, or an active area of ​​biogeochemical processes. The stratigraphic pollution load index comprehensively assesses the multi-indicator pollution status of a single depth stratigraphic layer. The data processing unit reads the environmental quality standard database, which contains the environmental quality standard limits for each pollution indicator. These limits are determined based on water body functional zoning and water quality protection targets. The concentration values ​​of each pollution indicator for a specific layer are extracted from the pollutant concentration matrix. The concentration value of each pollution indicator is divided by the corresponding environmental quality standard limit to obtain the single-indicator pollution coefficient. A pollution coefficient greater than 1 indicates that the concentration of that indicator exceeds the standard limit, which is considered an exceedance state. A pollution coefficient less than 1 indicates that the concentration of that indicator is below the standard limit, which is considered a compliance state. The larger the pollution coefficient value, the more severe the exceedance. The pollution coefficients of all pollution indicators for that layer are summed, and the sum is divided by the total number of pollution indicators to obtain the pollution load index for that layer.The pollution load index is the average of the pollution coefficients of various indicators. The numerical value comprehensively reflects the overall pollution load level of that layer. The data processing unit calculates the pollution load index for each layer in the pollutant concentration matrix, compares the index values ​​of all layers, and identifies the layer with the highest index value as the main accumulation layer for pollutants, indicating the most types of pollutants, the highest concentrations, or the most severe exceedances. The positive definite matrix factorization algorithm decomposes the pollutant concentration matrix into the product of a source component matrix and a source contribution matrix. The number of columns in the source component matrix equals the number of pollution sources, and each column represents the characteristic spectrum of a pollution source, describing the relative proportion of each pollutant indicator emitted by that source. The number of rows in the source contribution matrix equals the number of pollution sources, and each row represents the absolute contribution of a pollution source at each sampling point. During algorithm initialization, initial values ​​for the source component matrix and source contribution matrix are generated empirically or randomly, with all elements of the initial matrix being non-negative. The iterative optimization process calculates the product of the source component matrix and the source contribution matrix to obtain the reconstructed matrix. The reconstructed matrix is ​​then compared with the original pollutant concentration matrix. The residual matrix is ​​obtained by subtracting corresponding elements from the matrix. The objective function value is obtained by summing the squares of all elements in the residual matrix. The objective function value represents the fitting error between the reconstructed matrix and the original matrix. The algorithm adjusts the element values ​​of the source component matrix and source contribution matrix using gradient descent or multiplication update rules to gradually reduce the objective function value. During the update process, all matrix elements are constrained to remain non-negative because pollutant concentration and source contribution do not have negative values. After multiple iterations, the objective function converges to its minimum, yielding the optimal source component matrix and source contribution matrix. Pollution source factors are extracted from the source component matrix, and the element proportions of the corresponding columns for each factor are analyzed to determine the composition of characteristic pollutants. If a factor has a high value at the heavy metal element level but a low value at the nutrient level, then that factor is a heavy metal pollution source. If a factor has high values ​​at the ammonia nitrogen and total phosphorus levels, then that factor is a domestic sewage or agricultural non-point source pollution source. The contribution of each pollution source factor at different depths is extracted from the source contribution matrix. The depth range with the largest contribution is the main influencing layer of that source factor. A profile of the contribution change with depth is plotted to determine the spatial distribution range of the source factors. The total contribution of a pollution source factor is calculated, which is equal to the sum of the contributions of that factor at all sampling points. The sum of the total contributions of all pollution source factors is used as the denominator, and the total contribution of each individual factor is divided by the sum. The relative contribution rate of the source factor is obtained, which quantifies the importance of different pollution sources to the overall pollution. When a vertical distribution analysis of pollutants in a water body is performed, the pollutant concentration matrix shows that the heavy metal concentration is low at the surface, significantly increases at a depth of 5 meters, and decreases at a depth of 10 meters. The data processing unit calculates the concentration gradient from the surface to the 5-meter depth. The gradient value is obtained by dividing the concentration difference by the depth difference. The concentration gradient from the 5-meter to the 10-meter depth is calculated, and the second difference value is obtained by subtracting the two gradient values. If the absolute value of the second difference exceeds the multiple of the standard deviation, the 5-meter depth is identified as the interface of abrupt change in heavy metal concentration.Meanwhile, ammonia nitrogen concentration continued to increase within a depth range of 3 to 7 meters. The gradient vector showed that the concentration gradient in this range was consistently positive and relatively large. The second-order difference value fluctuated little within this range and did not exceed the threshold, indicating that the ammonia nitrogen concentration in this depth range continued to increase without abrupt changes. The data processing unit extracted the concentration values ​​of various pollution indicators at a depth of 5 meters. The pollution coefficients were obtained by dividing the heavy metal concentration by the corresponding standard limit, the ammonia nitrogen concentration by the corresponding standard limit, and the total phosphorus concentration by the corresponding standard limit. The pollution coefficients were summed and divided by the total number of pollution indicators to obtain the pollution load index for this layer. The pollution loads of other layers were then calculated. After comparing the load index, the layer with the highest index value at a depth of 5 meters was identified as the main accumulation layer of pollutants. The positive definite matrix factorization algorithm iteratively decomposed the concentration matrix. The first column of the source component matrix showed a high proportion of heavy metals, indicating an industrial emission source; the second column showed high proportions of ammonia nitrogen and total phosphorus, indicating a domestic sewage source. The first row of the source contribution matrix showed the largest contribution from this source near a depth of 5 meters; the second row showed a significant contribution from this source in the surface to middle layers. The total contribution of the two source factors was calculated, and each was divided by the sum to obtain the relative contribution rate. The contribution rates of industrial emission sources and domestic sewage sources quantified the degree of influence of each source on the overall pollution.

[0059] In one specific embodiment, a second-order difference operation is performed on the concentration gradient vector. When the absolute value of the second-order difference exceeds a set multiple of the gradient standard deviation, the corresponding depth interval is identified as a pollutant concentration abrupt change interface, including:

[0060] Extract the concentration gradient values ​​between adjacent depth layers from the concentration gradient vector, calculate the difference between adjacent concentration gradient values, and obtain the second-order difference value for each depth interval;

[0061] The second-order difference values ​​of all depth intervals are statistically analyzed, and the average and standard deviation of the second-order difference values ​​are calculated. The second-order difference judgment threshold is determined by multiplying the standard deviation by a set factor.

[0062] Traverse the second-order difference values ​​of each depth interval, determine whether the absolute value of the second-order difference exceeds the second-order difference judgment threshold, and mark the depth intervals where the absolute value of the second-order difference exceeds the second-order difference judgment threshold as abrupt change intervals.

[0063] Extract the start and end depths of the mutation interval, identify the region between the start and end depths as the pollutant concentration mutation interface, and record the depth location of the pollutant concentration mutation interface and the corresponding pollution index type.

[0064] Specifically, the data processing unit extracts the concentration gradient values ​​between adjacent depth layers sequentially from the concentration gradient vector in depth order. The first element of the concentration gradient vector corresponds to the concentration change rate between the surface layer and the subsurface layer, the second element corresponds to the concentration change rate between the subsurface layer and the third layer, and so on until the bottom layer. The second gradient value is extracted by subtracting the first gradient value to obtain the second difference value of the first depth interval, and the third gradient value is extracted by subtracting the second gradient value to obtain the second difference value of the second depth interval. The entire gradient vector is traversed according to the same calculation rules. The operation of subtracting adjacent gradient values ​​reflects the trend of gradient itself with depth. When the concentration gradient suddenly becomes steep from a gentle slope, the second difference has a large positive value. When the concentration gradient suddenly becomes gentle from a steep slope, the second difference has a large negative value. When the concentration gradient remains stable, the second difference is close to zero. After the second difference values ​​of each depth interval are calculated, they are stored in depth order to form a second difference sequence. The data processing unit reads all values ​​from the second-order difference sequence, sums all the second-order difference values, and divides the sum by the total number of second-order difference values ​​to obtain the arithmetic mean. The arithmetic mean represents the central position and overall trend of the second-order difference. Then, it extracts each second-order difference value one by one and calculates the deviation from the mean. The deviation represents the degree to which the value deviates from the center. The deviation is squared to eliminate the influence of the positive or negative sign. All squared deviations are summed, and the sum is divided by the total number of second-order difference values ​​to obtain the variance. The variance quantifies the dispersion of the second-order difference values. The larger the variance value, the more drastic the fluctuation of the second-order difference; the smaller the variance value, the smoother the fluctuation of the second-order difference. The square root of the variance is then taken to obtain the standard deviation. The standard deviation has the same dimensions as the original data. The standard deviation directly reflects the fluctuation range of the second-order difference value around the mean. The data processing unit multiplies the standard deviation by a preset multiple to obtain the second-order difference judgment threshold. The multiple is determined according to the sensitivity requirements for identifying sudden changes in pollutant concentration. When the multiple is set to 2, about 95% of the normal fluctuation second-order difference values ​​are within the threshold range. When the multiple is set to 3, about 99% of the normal fluctuation second-order difference values ​​are within the threshold range. The larger the multiple, the stricter the judgment standard. Only extreme second-order difference values ​​that significantly deviate from the normal fluctuation are identified as sudden changes. The smaller the multiple, the more lenient the judgment standard. More second-order difference values ​​are identified as sudden changes, but the risk of noise interference increases. The data processing unit traverses each value in the second-order difference sequence in depth order, extracts the second-order difference value for the current depth interval, and performs an absolute value operation on this value to remove the sign. After the absolute value operation, positive and negative second-order differences are uniformly judged according to the degree of deviation. The absolute value of the second-order difference is compared with the second-order difference judgment threshold. When the absolute value of the second-order difference is greater than the judgment threshold, it is determined that the concentration gradient in that depth interval has undergone a significant abrupt change, and the index number or depth range of that depth interval is marked as an abrupt change interval and stored. When the absolute value of the second-order difference is less than or equal to the judgment threshold, it is determined that the concentration gradient change in that depth interval is within the normal fluctuation range and is not marked as an abrupt change interval.The traversal process continues until the last element of the second-order difference sequence. After the traversal is complete, a set of all depth intervals marked as abrupt change intervals is obtained. The data processing unit reads the set of abrupt change intervals, extracts the index information of each interval one by one, and queries the starting and ending depth values ​​of the interval based on the index number. The starting depth value is the depth coordinate corresponding to the upper boundary of the interval, and the ending depth value is the depth coordinate corresponding to the lower boundary. The spatial range between the starting and ending depths is defined as the pollutant concentration abrupt change interface. The depth span of the abrupt change interface is equal to the ending depth minus the starting depth. The depth span reflects the spatial scale of the abrupt change process. The data processing unit records the starting and ending depth values ​​of each abrupt change interface, and also records the pollution index type corresponding to the abrupt change interface, indicating whether it is heavy metal, nutrient, or organic pollution. When a pollutant experiences a sudden concentration change within a certain depth range, the depth of the abrupt change interface varies for different pollutants. For heavy metals, the abrupt change interface corresponds to the location of direct industrial emissions or a sedimentation accumulation layer; for nutrients, it corresponds to the location where a thermocline or densitoid layer hinders vertical mixing; and for organic pollutants, it corresponds to an active biodegradation layer or a location where adsorption-desorption equilibrium changes. By associating the depth of the abrupt change interface with the type of pollutant, when a vertical distribution analysis of pollutants in a water body is performed, the concentration gradient vector shows that the heavy metal concentration gradient is relatively small from the surface to a depth of 3 meters, significantly increases from 3 meters to 5 meters, and then decreases back to a smaller level from 5 meters to 7 meters. The data processing unit extracts the gradient values ​​from the surface to 3 meters as the first gradient, and the gradient values ​​from 3 meters to 5 meters as the second gradient. The second gradient value is significantly larger than the first gradient value. Subtracting the two yields the second difference value for the first depth interval, approximately 3 meters. This second difference value is positive and relatively large. The gradient values ​​from 5 meters to 7 meters are then extracted as the third gradient. The third gradient value is significantly smaller than the second gradient value. Subtracting the two yields the second difference value for the second depth interval, approximately 5 meters. This second difference value is negative and relatively large. The data processing unit continues to calculate the second difference values ​​for other depth intervals. The average value is obtained by summing all the second difference values ​​and dividing by the total number of values. The second difference value is then calculated for each depth interval. The squared deviations of the scores from the average are summed, divided by the total number of scores, and the square root is taken to obtain the standard deviation. The standard deviation is multiplied by a set factor of 3 to obtain the judgment threshold. The second-order difference sequence is iterated. The absolute value of the second-order difference in the depth range near 3 meters is compared with the judgment threshold. If the absolute value exceeds the threshold, the 3-meter depth range is determined as a mutation range and marked. The absolute value of the second-order difference in the depth range near 5 meters is also compared with the judgment threshold. If the absolute value also exceeds the threshold, the 5-meter depth range is determined as a mutation range and marked. The absolute values ​​of the second-order differences in other depth ranges are all less than the judgment threshold and are not marked as mutation ranges. The data processing unit extracts the marked mutation ranges. The starting depth of the 3-meter mutation range is 2.5 meters, and the ending depth is 3.5 meters.This interval was identified as a sudden change in heavy metal concentration, and the starting depth, ending depth, and heavy metal index type were recorded. The starting depth of the 5-meter abrupt change interval was 4.5 meters, and the ending depth was 5.5 meters. This interval was identified as a sudden change in heavy metal concentration, and relevant information was recorded.

[0065] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for stratified collection of water pollutants based on a multi-functional monitoring vessel, characterized in that, The method includes: Step S1: Collect acoustic Doppler frequency shift signals and CTD probe data along the preset route, calculate the density stratification structure of the water body in the vertical direction according to the density gradient equation, identify the depth position of the density stratification interface, and generate a three-dimensional density field grid. Step S2: Based on the depth position of the density jump layer interface in the three-dimensional density field grid, and combined with the Richardson number, determine the stability of each depth layer, set up a jump layer precision sampling grid above and below the density jump layer interface, and set up sampling points with equal spacing in the uniform layer between the jump layers to form a sampling depth sequence. Step S3: Lower the multi-channel stratified water sampling device to the target depth in the sampling depth sequence. After correcting the cable tilt error by using a depth encoder and a three-axis tilt sensor, trigger the electromagnetic drive valve to collect water samples from each depth layer and record the three-dimensional coordinates, collection time and environmental parameters to generate a sample information set. Step S4: Filter and separate the water samples in the sample information set and perform multi-index detection to obtain the measured concentration values ​​of heavy metals, nutrients and organic pollutants in each layer, and construct a pollutant concentration matrix; Step S5: Calculate the vertical concentration gradient vector and second-order difference based on the pollutant concentration matrix, identify the pollutant concentration abrupt change interface, calculate the stratum pollution load index, and determine the main accumulation strata of pollutants and the spatial distribution of source factors.

2. The method for stratified collection of water pollutants based on a multi-functional monitoring vessel according to claim 1, characterized in that, Step S1 includes: The acoustic Doppler current profiler is controlled to travel along a preset route at a set speed, and emits sound pulses of a set frequency at set intervals. It receives Doppler frequency shift signals reflected back from different depth layers of the water body and calculates the current velocity vector data within a set water depth range. Temperature, conductivity, and depth data are collected synchronously using a CTD probe at a set sampling frequency. The temperature and conductivity data are substituted into the density gradient equation, and combined with the surface water density, salinity contraction coefficient, and thermal expansion coefficient, the density stratification structure of the water body in the vertical direction is calculated. Determine whether the density difference between two adjacent layers exceeds a set threshold, identify the depth location where there is a significant density jump layer, and obtain the depth coordinates of each density jump layer interface; Based on the collected spatial point data, a three-dimensional density field grid with a set resolution is generated by the Kriging interpolation algorithm, and the three-dimensional density field grid is stored in the form of a matrix containing spatial coordinates and density values.

3. The method for stratified collection of water pollutants based on a multi-functional monitoring vessel according to claim 1, characterized in that, Step S2 includes: Read the depth position of each density hopping layer interface identified in the three-dimensional density field grid, and extract the horizontal and vertical velocity components of each depth layer in the velocity profile data; Richardson number for each depth layer is calculated based on the vertical gradient of the horizontal flow velocity and the Brunt-Vesala frequency. When the Richardson number is greater than a set stability threshold, the corresponding depth layer is determined to be a strongly stable layer region. Multiple sampling points are set above and below the density gradient interface to form a gradient precision sampling grid. The sampling points are located at different distances above the interface and different distances below the interface. Sampling points are set up in the uniform layer between the layers according to the equal spacing strategy. The spacing between each sampling point is calculated based on the basic spacing and the sampling point number in the layer, and a sampling depth sequence containing all sampling depths is generated.

4. The method for stratified collection of water pollutants based on a multi-functional monitoring vessel according to claim 1, characterized in that, Step S3 includes: The multi-channel stratified water sampling device is lowered via a main control cable, and the main control cable is embedded with an optical fiber sensing line to transmit depth position signals in real time. The actual depth position of the device is calculated by measuring the cable release length using a depth encoder, combining the cable tilt angle measured by a triaxial tilt sensor, correcting the cable tilt error caused by water flow, and calculating the actual depth position of the device. When the difference between the actual depth position of the device and the target depth in the sampling depth sequence is less than the set accuracy threshold, a trigger current is sent to the electromagnetic drive valve of the corresponding channel to open the valve and allow the water sample to fill the sampling bottle under the drive of environmental pressure. Record the longitude, latitude, depth, sampling time, ambient temperature, and in-situ pressure of each sampling bottle at the time of sampling completion, and generate a sample information set containing sample number and multidimensional information.

5. The method for stratified collection of water pollutants based on a multi-functional monitoring vessel according to claim 1, characterized in that, Step S4 includes: The water sample in the sample information set is filtered through a filter membrane with a set pore size to remove suspended particulate matter. The filtered water sample is then divided into multiple portions and different preservation reagents are added to each portion. The first water sample was tested for heavy metals using a portable inductively coupled plasma mass spectrometer. The heavy metal concentration was calculated using the standard curve method based on the measured signal intensity, blank signal intensity, and the slope of the standard curve. The second water sample was analyzed using a continuous flow analyzer with colorimetric and reduction methods to determine the concentrations of total phosphorus, ammonia nitrogen, and nitrate nitrogen. Organic pollutants were detected in the third water sample using solid-phase extraction-gas chromatography-mass spectrometry (SPE-GC-MS) to obtain the concentration values ​​of polycyclic aromatic hydrocarbons (PAHs) and polychlorinated biphenyls (PCBs). The measured concentration values ​​of each pollutant index at each layer were then summarized to construct a pollutant concentration matrix.

6. The method for stratified collection of water pollutants based on a multi-functional monitoring vessel according to claim 1, characterized in that, Step S5 includes: Based on the pollutant concentration values ​​and depth differences between adjacent layers in the pollutant concentration matrix, calculate the concentration gradient vector of each pollution index in the vertical direction; Perform a second-order difference operation on the concentration gradient vector. When the absolute value of the second-order difference exceeds a set multiple of the gradient standard deviation, identify the corresponding depth interval as the interface of sudden change in pollutant concentration. Based on the ratio of the concentration values ​​of each pollution index at each stratum to the corresponding environmental quality standard limit, the stratum pollution load index is calculated, and the stratum with the largest pollution load index is determined as the main accumulation stratum of pollutants. The pollutant concentration matrix is ​​analyzed by positive definite matrix factorization algorithm, which decomposes it into source component matrix and source contribution matrix. Through iterative optimization, the main pollution source factors and the characteristic pollutant composition, spatial distribution range and relative contribution rate of each source factor are obtained.

7. The method for stratified collection of water pollutants based on a multi-functional monitoring vessel according to claim 6, characterized in that, The second-order difference operation on the concentration gradient vector, when the absolute value of the second-order difference exceeds a set multiple of the gradient standard deviation, identifies the corresponding depth interval as a pollutant concentration abrupt change interface, including: Extract the concentration gradient values ​​between adjacent depth layers from the concentration gradient vector, calculate the difference between adjacent concentration gradient values, and obtain the second-order difference value for each depth interval; The second-order difference values ​​of all depth intervals are statistically analyzed, and the average and standard deviation of the second-order difference values ​​are calculated. The second-order difference judgment threshold is determined by multiplying the standard deviation by a set factor. Traverse the second-order difference values ​​of each depth interval, determine whether the absolute value of the second-order difference exceeds the second-order difference judgment threshold, and mark the depth intervals where the absolute value of the second-order difference exceeds the second-order difference judgment threshold as abrupt change intervals. Extract the start and end depths of the mutation interval, identify the region between the start and end depths as the pollutant concentration mutation interface, and record the depth location of the pollutant concentration mutation interface and the corresponding pollution index type.

Images