Dynamic monitoring method and system for lignin-regulated soil Cd bio-availability

Abstract

The invention belongs to the field of monitoring, and particularly relates to a dynamic monitoring method for lignin-regulated soil Cd bio-availability, which comprises the following steps: acquiring Micro-XRF element surface distribution data of a to-be-detected soil sample; for each pixel point, calculating a first spatial correlation between a Cd signal gradient and a carbon signal gradient and a second spatial correlation between the Cd signal gradient and a gradient of a weighted sum of silicon, aluminum and iron signals in a neighborhood of the pixel point; when the first spatial correlation is greater than a preset positive threshold value and the second spatial correlation is smaller than a preset negative threshold value, determining the pixel point as a target micro-domain; performing NanoSIMS analysis on the target micro-domain to obtain a multi-dimensional image data cube; based on a pre-established reference ion mass spectrum library of pure lignin, a pure mineral phase and a Cd standard substance, unmixing the multi-dimensional image data cube to obtain a pure abundance distribution diagram of the lignin, the minerals and the Cd in the target micro-domain; and generating a high-resolution soil Cd occurrence form probability graph according to the pure abundance distribution graph.

Description

Technical Field

[0001] This application belongs to the field of control, and in particular relates to a dynamic monitoring method and system for lignin-regulated soil Cd bioavailability. Background Technology

[0002] Cadmium (Cd) pollution in soil is highly toxic and bioaccumulative. The migration, transformation, and bioavailability of Cd in soil depend not only on its total content but also, to a large extent, on its speciation within the soil microenvironment. Soil is a complex system composed of minerals, organic matter such as lignin, and their interactions. Lignin, as a major product of plant residue degradation, contains abundant oxygen-containing functional groups and is a key component regulating soil Cd activity. Methods for analyzing the speciation of heavy metals in soil mainly rely on chemical sequential extraction methods such as the biochemical re-extraction (BCR) method. This method is essentially a destructive macroscopic statistical approach, failing to preserve the original physical structure and micro-regional distribution information of the soil. Furthermore, the re-adsorption or speciation of heavy metals during extraction can easily occur, leading to distorted analytical results. In addition, it is difficult to elucidate the microscopic mechanisms by which organic matter retains Cd.

[0003] With the development of in-situ high-resolution imaging techniques such as synchrotron X-ray fluorescence spectroscopy (Micro-XRF) and nano-secondary ion mass spectrometry (NanoSIMS), researchers can directly observe the distribution characteristics of heavy metals at the microscale. However, Micro-XRF data cannot accurately locate target micro-domains with specific interactions, such as organic-inorganic complexes. Furthermore, during high-resolution NanoSIMS analysis of these micro-domains, the highly complex soil microenvironment often leads to signal aliasing of different components at the pixel level, forming mixed pixels and making it difficult to resolve pure component distributions. Moreover, it is impossible to determine and distinguish the specific adsorption probability and retention strength of cadmium within lignin, minerals, and at their interfaces, thus limiting in-depth research on the microscopic occurrence of cadmium in soil. Summary of the Invention

[0004] To address the aforementioned issues, in the first aspect, a dynamic monitoring method for lignin-regulated soil Cd bioavailability is proposed, comprising the following steps: Micro-XRF elemental surface distribution data of the soil sample to be tested are acquired. The elements include Cd, carbon (C) representing organic matter, and silicon (Si), aluminum (Al), and iron (Fe) representing minerals. For each pixel, the first spatial correlation between the Cd signal gradient and the carbon signal gradient, and the second spatial correlation between the Cd signal gradient and the gradient of the weighted sum of the silicon, aluminum, and iron signals are calculated in its neighborhood. When the first spatial correlation is greater than a preset positive threshold and the second spatial correlation is less than a preset negative threshold, the pixel is determined to be the target micro-domain. NanoSIMS analysis was performed on the target micro-domain to obtain a multidimensional image data cube, the data cube including Cd and lignin-representing elements. 12 C - , 12 C 14 N - Ions, representing minerals 28 Si - , 27 Al 16 O - , 56 Fe 16 O - ion; Based on a pre-established reference ion mass spectrometry library of pure lignin, pure mineral phases and Cd standards, the multidimensional image data cube is unmixed to obtain a purity abundance distribution map of lignin, minerals and Cd within the target microdomain. A high-resolution probability map of soil Cd occurrence forms is generated based on the purity abundance distribution map.

[0005] In the second aspect, a dynamic monitoring system for lignin-regulated soil Cd bioavailability is proposed, comprising the following units: The target micro-domain determination unit is used to acquire Micro-XRF elemental surface distribution data of the soil sample to be tested. The elements include Cd, carbon (C) representing organic matter, and silicon (Si), aluminum (Al), and iron (Fe) representing minerals. For each pixel, the first spatial correlation between the Cd signal gradient and the carbon signal gradient, and the second spatial correlation between the Cd signal gradient and the gradient of the weighted sum of the silicon, aluminum, and iron signals are calculated in its neighborhood. When the first spatial correlation is greater than a preset positive threshold and the second spatial correlation is less than a preset negative threshold, the pixel is determined to be the target micro-domain. The NanoSIMS analysis unit is used to perform NanoSIMS analysis on the target microdomain to obtain a multidimensional image data cube, the data cube including Cd and lignin-representing elements. 12 C - , 12 C 14 N - Ions, representing minerals 28 Si - , 27 Al 16 O - , 56 Fe 16 O - ion; The demixing unit is used to demix the multidimensional image data cube based on a pre-established reference ion mass spectrometry library of pure lignin, pure mineral phases and Cd standards, to obtain a purity abundance distribution map of lignin, minerals and Cd within the target microdomain. The probability map generation unit is used to generate a high-resolution soil Cd occurrence probabilistic map based on the purity abundance distribution map.

[0006] This invention combines the macroscopic screening advantages of Micro-XRF with the high-resolution microscopic imaging capabilities of NanoSIMS, solving the problem of locating the occurrence state of heavy metals in complex and heterogeneous soil environments. Through a gradient-based spatial correlation screening mechanism, key micro-regions of Cd interaction with organic matter and minerals can be quickly identified, avoiding the inefficiency of blind observation. A non-negative matrix factorization algorithm is used to spectrally unmix multidimensional image data, eliminating background interference and signal crosstalk, thereby accurately obtaining the true abundance distribution of lignin, minerals, and Cd. Furthermore, by constructing an interface retention index model incorporating Euclidean distance and a normalized scale factor, not only is the spatial location of Cd determined, but the specific proportions of Cd adsorption on the lignin surface, mineral retention, and their interfacial complexation can be quantitatively analyzed from a probabilistic perspective. This provides data support and visualization for a deeper understanding of the microscopic regulatory mechanism of lignin on soil heavy metals, improving the reliability and scientific rigor of soil environmental monitoring results. Attached Figure Description

[0007] Figure 1 This is a flowchart of the present invention; Figure 2 This is a schematic diagram illustrating the abundance obtained from nonnegative matrix decomposition. Detailed Implementation

[0008] To make the objectives, technical solutions, and advantages of this specification clearer, the technical solutions of this specification will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of them. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this specification.

[0009] See Figure 1 The method for dynamic monitoring of lignin-regulated soil Cd bioavailability, as shown, includes the following steps: S1. Acquire Micro-XRF elemental surface distribution data of the soil sample to be tested. The elements include Cd, carbon (C) representing organic matter, and silicon (Si), aluminum (Al), and iron (Fe) representing minerals. For each pixel, calculate the first spatial correlation between the Cd signal gradient and the carbon signal gradient, and the second spatial correlation between the Cd signal gradient and the gradient of the weighted sum of the silicon, aluminum, and iron signals in its neighborhood. When the first spatial correlation is greater than a preset positive threshold and the second spatial correlation is less than a preset negative threshold, determine the pixel as the target micro-domain. Freshly collected soil samples were freeze-dried to remove moisture. The dried soil samples were then vacuum-impregnated with epoxy resin. After embedding and curing, the surface of the embedded block was ground and polished until the soil cross-section was exposed, with a surface roughness of less than 1 micrometer. The prepared samples were placed on the sample stage of a Micro-XRF spectrometer. The excitation voltage was set to 50 kV, the current to 600 μA, and the spot size to 20 micrometers. The scanning area was selected using the instrument's built-in mapping function, and the pixel dwell time was set to 10 ms to 50 ms. The fluorescence signals of the L-series of Cd, and the K-series of C, Si, Al, and Fe were sequentially acquired. The two-dimensional intensity matrix data of each element within the scanning area were then exported.

[0010] A sliding window of 3×3 or 5×5 pixels is used to traverse the Micro-XRF image. For each pixel, the gradient vectors of the Cd and C element intensity matrices within the window are calculated using the Sobel operator. The Si, Al, and Fe element intensity matrices are normalized and summed to obtain the total mineral signal matrix. The gradient vector of this total mineral signal matrix is ​​then calculated using the Sobel operator. The Pearson correlation coefficient between the Cd and C gradient vectors is used as the first spatial correlation, and the Pearson correlation coefficient between the Cd gradient vector and the total mineral signal gradient vector is used as the second spatial correlation.

[0011] The positive threshold for the first spatial correlation is set between 0.6 and 0.8, and the negative threshold for the second spatial correlation is set between -0.6 and -0.8. The calculation results of all pixels are traversed to filter out the position coordinates that simultaneously meet the correlation coefficient conditions. The physical regions corresponding to these coordinates are marked as regions of interest to be analyzed. The position of the target micro-domain is marked on the sample surface using laser etching or physical indentation for subsequent instrument positioning.

[0012] To compensate for the limitations of Micro-XRF in terms of carbon visibility and resolution, in another embodiment, atomic number contrast in BSE images is used to distinguish organic matter (dark areas) from minerals (light areas), while EDS is used to quickly screen out areas with high Cd element signals. By overlaying images, the interface regions where "organic matter-minerals-Cd" coexist or come into contact are found, and platinum is deposited or positioning markers are cut under SEM to determine the coordinates of the target micro-domain.

[0013] S2, Perform NanoSIMS analysis on the target micro-domain to obtain a multidimensional image data cube, the data cube including Cd and lignin-representing elements. 12 C - , 12 C 14 N - Ions, representing minerals 28 Si - , 27 Al 16 O - , 56 Fe 16 O - ion; Since Micro-XRF is at the micrometer scale and NanoSIMS is at the nanometer scale, to further identify the target micro-domain in NanoSIMS, the sample is transferred to a scanning electron microscope; the ROI region is navigated to using Micro-XRF coordinates; backscattered electron (BSE) imaging is used to acquire micrometer-scale morphology and compositional contrast images of the region; unique features within the ROI are extracted as micro / nano fingerprints. The sample is then transferred to NanoSIMS; scanning is performed using the instrument's built-in optical CCD or ion-induced secondary electron imaging (SISE); image registration is performed based on the aforementioned micro / nano fingerprints to locate the target position; and high-resolution isotope imaging is performed by narrowing the field of view to within 10 μm.

[0014] A 5-10 nm thick gold or platinum conductive layer is sputtered onto the surface of the sample after Micro-XRF analysis. The sample is then loaded into the sample chamber of the NanoSIMS instrument. Using the built-in optical microscope, the instrument navigates to the marked target microdomain and uses a Cs ion source as the primary ion beam, setting the beam current intensity to 1-5 picoamperes, to bombard the sample surface. Pre-sputtering is performed within the analysis region until the secondary ion signal stabilizes. Multiple receivers are set to detect Cd isotopes with mass numbers of 112 or 114, and... 12 C - , 12 C 14 N - ion, 28 Si - , 27 Al16 O - , 56 Fe 16 O - Ions. A multi-dimensional image data cube is obtained by scanning layer by layer with a resolution of 256×256 or 512×512, acquiring stacked image data containing multiple channels. Among them, 12 C - The carbon skeleton structure representing lignin is the most fundamental indicator for distinguishing between organic and inorganic phases. 12 C 14 N - Although primarily indicating nitrogen-containing organic compounds, lignin is often used as a high-contrast tracer for biological organism profiles because its ionization efficiency in secondary ion mass spectrometry is much higher than that of single-atom carbon. Combining these two methods can extract the organic regions containing lignin from the background of inorganic minerals such as silicon, aluminum, and iron. The characteristic vector of pure lignin exhibits high contrast. 12 C - Signal and low 12 C 14 N - Signal.

[0015] S3, based on the pre-established reference ion mass spectrometry library of pure lignin, pure mineral phase and Cd standard, the multidimensional image data cube is unmixed to obtain the purity abundance distribution map of lignin, mineral and Cd in the target micro-domain; Pure lignin standards, pure kaolin or montmorillonite standards, and cadmium carbonate or Cd-adsorbed mineral standards were pre-selected. Under the same NanoSIMS conditions, their secondary ion mass spectrometry characteristics were acquired as the basis matrix W. The multidimensional image data cube of the target micro-domain was expanded into a two-dimensional matrix V, where rows represent pixels and columns represent the signal intensities of different ions. The equation V was iteratively solved using a nonnegative matrix factorization algorithm, approximately equal to WH. During the iteration process, the basis matrix W was kept constant while solving the coefficient matrix H. The obtained coefficient matrix H was then reconstructed according to the original image resolution to obtain a grayscale image of the relative abundance distribution of lignin minerals and Cd in space. (See [reference needed]). Figure 2 .

[0016] In one embodiment, analytical soda ash lignin, standard kaolinite powder, and artificially prepared adsorbed Cd goethite are selected as three types of standard materials. Small amounts of each powder are pressed onto the surface of a high-purity indium sheet to create three independent standard sample blocks, which are then carbon-plated for conductivity. Experimental parameters identical to those used on the soil sample are set on a NanoSIMS instrument, such as using a Cs+ primary ion beam, and the positions of multiple receivers are adjusted for simultaneous detection. 12 C - , 12 C 14 N - , 28 Si- , 27 Al 16 O - , 56 Fe 16 O - Characteristic secondary ions containing Cd were also identified. Raster scanning imaging was performed on three standard samples. In the obtained image data, avoiding edge effects and surface contamination, regions of interest with uniform signals were manually selected, and the average intensity values ​​of all target ion channels within the region were extracted to obtain three characteristic spectral vectors representing pure lignin, pure mineral, and pure Cd morphology, respectively. After normalization, these vectors were combined into a basis matrix.

[0017] S4. Generate a high-resolution soil Cd occurrence probabilities map based on the purity abundance distribution map.

[0018] In a preferred embodiment, generating a high-resolution soil Cd occurrence probabilities map based on the purity abundance distribution map specifically involves: Calculate the Euclidean distance d1 between each Cd-containing pixel and the boundary of the nearest lignin region, and the Euclidean distance d2 between each pixel and the boundary of the nearest mineral region; construct the Cd interface retention index I = exp(-(d1+d2) / L), where L is the normalized scaling factor; use the value of the Cd interface retention index I as the interface complexation probability P_interface; and apply the formula P_interface... 木质素 =(1-I)×d2 / (d1+d2) and P 矿物 The lignin adsorption probability P is calculated as (1-I)×d1 / (d1+d2). 木质素 and the probability of mineral retention P 矿物 By combining the Cd purity and abundance of each pixel, a high-resolution probability map of soil Cd occurrence morphology is generated.

[0019] The Otsu algorithm (maximum inter-class variance method) is used to binarize the lignin abundance map and mineral abundance map to extract binary mask images of the lignin and mineral regions. The Canny operator is used to identify the edge pixel sets of the lignin and mineral regions. For each pixel in the abundance map where the Cd signal intensity is greater than the background noise, the Euclidean distance transform algorithm is used to calculate the straight-line distance d1 from the pixel's coordinates to the nearest lignin edge pixel and the straight-line distance d2 from the nearest mineral edge pixel.

[0020] Based on the spatial resolution of NanoSIMS, the normalized scale factor L is set to a value between 50 nm and 100 nm. The calculated d1 and d2 are substituted into the exponential decay formula to calculate the interface retention index I for each Cd-containing pixel. This index value is distributed between 0 and 1. The stored I, d1, and d2 values ​​are retrieved from computer memory. For each pixel, the lignin adsorption probability P is calculated according to algebraic rules. 木质素 and the probability of mineral retention P 矿物 For non-interface regions, i.e., regions with I values ​​close to 0, probability weights are assigned to components that are closer in distance. An RGB three-channel empty matrix of the same size as the original image is constructed, with the Cd purity abundance value of each pixel used as an opacity or brightness weight. The calculated Pinterface value is assigned to the green channel G, Plignin value to the red channel R, and Pmineral value to the blue channel B. A pseudo-color image is synthesized, or three probability distribution heatmaps of Cd complexation at the interface, Cd adsorption in lignin, and Cd immobilization in minerals are output separately to display the chemical speciation distribution of Cd at the soil microscale.

[0021] In a preferred embodiment, acquiring the Micro-XRF elemental surface distribution data of the soil sample to be tested includes: The collected soil samples were dried to constant weight in a 40°C constant temperature oven, and gravel and roots with a diameter greater than 2 mm were removed. The treated soil samples were placed in a mold, and molten low-melting-point sublimed sulfur or metallic indium was added. The samples were then impregnated and embedded in a vacuum environment. After cooling and solidification, a solidified sample was formed. The surface of the solidified sample was polished stepwise using diamond polishing paste, with polishing pastes of 9 μm, 3 μm, 1 μm and 0.25 μm used in sequence, until the surface roughness Ra < 100 nm. On a micro-area X-ray fluorescence spectrometer, the excitation voltage was set to 50 kV, the tube current to 600 μA, and the scanning step size to 20 μm. The polished sample surface was scanned in a vacuum environment, and the characteristic X-ray fluorescence intensities of Cd, C, Si, Al and Fe were collected to generate a two-dimensional gray-scale intensity matrix for each element.

[0022] In the sample pretreatment stage, the collected natural soil is spread evenly in an enamel dish and placed in an electric heating drying oven for more than 48 hours until the weight no longer changes. Then, it is sieved with a nylon screen to remove large particles of impurities. The sieved fine soil is mixed with epoxy resin and curing agent at a mass ratio of 3:1 and placed in a vacuum mounting machine. The mixture is then degassed for 30 minutes under a pressure of -0.1 MPa to remove air bubbles. After the resin has completely cured, a semi-automatic polishing machine is used with velvet polishing cloth and diamond suspensions of different particle sizes for precision polishing. During this process, an atomic force microscope is used to check the surface roughness to ensure that the flatness requirement of Ra < 100 nanometers is met.

[0023] The prepared resin target is placed on the instrument's sample stage, and panoramic photography and navigation are performed using a CCD camera. The system operates in an environment with a vacuum level below 20 to reduce the absorption of characteristic X-rays of light elements by the air. The primary X-ray beam generated by the X-ray tube is focused by a multi-capillary lens and then incident vertically onto the sample surface. The silicon drift detector synchronously records the fluorescence signal excited at each scanning point. The scanning area is divided into a 500×500 pixel grid, and each pixel contains the fluorescence count rate of the corresponding element, thereby constructing five independent two-dimensional gray-scale intensity matrices.

[0024] In a preferred embodiment, calculating the first spatial correlation between the Cd signal gradient and the carbon signal gradient, and the second spatial correlation between the Cd signal gradient and the gradient of the weighted sum of silicon, aluminum, and iron signals, for each pixel, includes: Define a 5×5 pixel rectangular area centered on the current pixel as the neighborhood; Gaussian smoothing filtering was applied to the two-dimensional gray-level intensity matrices of five elements: Cd, C, Si, Al, and Fe. The Sobel operator is used to convolve the filtered Cd matrix and C matrix respectively to obtain the Cd gradient magnitude vector G. Cd and C gradient magnitude vector G C G is calculated using the Pearson correlation coefficient formula. Cd and G C The correlation coefficient between them is denoted as the first spatial correlation. The filtered Si, Al, and Fe matrices are normalized to their maximum values, and then the three normalized matrices are summed with equal weights to obtain the mineral signal weighted sum matrix M. sum ; Using the Sobel operator on M sum Perform convolution operation to obtain the mineral composite gradient magnitude vector G. min G is calculated using the Pearson correlation coefficient formula. Cd and G min The correlation coefficient between them is denoted as the second spatial correlation.

[0025] A Gaussian filter with a standard deviation of 1 and a kernel size of 3×3 is used to convolve the original grayscale matrix to smooth high-frequency random noise caused by fluctuations in instrument counting statistics. Then, the Sobel operator is used to perform convolution calculations on the image in both the horizontal and vertical directions, using the formula... The gradient magnitude of each pixel is synthesized. For mineral elements, the intensity matrix values ​​of Si, Al, and Fe are first mapped to the range of 0 to 1, and then superimposed to reflect the overall distribution trend of the soil mineral matrix. In the correlation calculation step, for any central pixel in the image, the gradient magnitudes of 25 pixels in its 5×5 neighborhood are extracted and flattened into a one-dimensional vector of length 25. The Cd gradient vector is used as variable X, and the C gradient vector and the mineral composite gradient vector are used as variables Y. The linear correlation is calculated by substituting them into the Pearson formula. The sliding window algorithm is used to traverse the entire image to generate two correlation coefficient mapping maps with the same size as the original image, representing the correlation between Cd and organic matter and between Cd and mineral matrix in terms of spatial variation trends.

[0026] In a preferred embodiment, determining the pixel as the target micro-domain when the first spatial correlation is greater than a preset positive threshold and the second spatial correlation is less than a preset negative threshold includes: Construct a binary discrimination matrix and retain pixels that satisfy the condition that the first spatial correlation is greater than a preset positive threshold and the second spatial correlation is less than a preset negative threshold; A morphological opening operation is performed on the retained pixel set to remove isolated noise points with an area of ​​less than 3 pixels. The remaining connected regions are marked as target micro-domains to be analyzed, and the geometric center coordinates of the target micro-domains in the Micro-XRF coordinate system and the optical feature images used for relocation are recorded.

[0027] Preferably, the positive threshold is 0.65, and the preset negative threshold is -0.5. The screening logic locates potential organically bound regions based on the positive correlation between Cd and organic matter and the negative correlation with the mineral matrix. The program traverses each position in the correlation coefficient matrix. If the position simultaneously satisfies the conditions of a first spatial correlation greater than 0.65 and a second spatial correlation less than -0.50, then the position is marked as 1 in the binary mask; otherwise, it is marked as 0. For the generated binary mask image, a disk-shaped structuring element with a radius of 1 pixel is used to perform morphological opening operations, i.e., erosion followed by dilation, to eliminate insignificant sporadic noise and smooth region boundaries. All independent preserved regions are identified using a connected component labeling algorithm. The first moment of each connected component is calculated to determine its geometric center coordinates. These coordinates are mapped back to the physical sample stage coordinate system through an affine transformation matrix. Simultaneously, the system automatically captures an optical micrograph within a radius of 1 mm centered on this center, which serves as a feature reference for subsequent searching of the same micro-region in a nano-secondary ion mass spectrometer.

[0028] In a preferred embodiment, performing NanoSIMS analysis on the target micro-domain to obtain a multidimensional image data cube includes: The sample was transferred to the sample chamber of the nanosecondary ion mass spectrometer. Based on the recorded geometric center coordinates, the target microdomain was calibrated and located using an optical microscope and ion imaging. A cesium ion beam was used as the primary ion source, with a beam current intensity set to 10 picoamperes, to pre-sputter the surface of the target microdomain until the secondary ion yield reached a steady state. The image resolution was set to 256 × 256 pixels, and the residence time was 5 milliseconds per pixel, with simultaneous data collection. 12 C - , 12 C 14 N - , 28 Si - , 27 Al 16 O - , 56 Fe 16 O - The multidimensional image data cube is constructed using the signal intensity of characteristic secondary ions containing Cd.

[0029] After the sample is transferred to the high-vacuum sample chamber of the NanoSIMS instrument, the operator compares the optical feature images recorded by Micro-XRF with the images from the NanoSIMS's built-in CCD camera. Using the sample stage stepper motor, the target micro-domain is moved to the ion beam bombardment center. This process is done to eliminate features such as... 28 Si and 12 C 16 Mass spectrometry interference from isotopes like O necessitates adjusting the widths of the inlet and outlet slits to ensure a mass resolution better than 5000. Before data acquisition, a high-energy cesium ion beam pre-sputters within a 30×30 micrometer area to inject cesium ions, enhancing the ionization rate of secondary ions and cleaning surface contaminants. Once the instrument enters formal acquisition mode, the primary ion beam bombards points within a 25×25 micrometer field of view using a grating scan method. Six electron multiplier detectors simultaneously record ion counts for channels with different mass-to-charge ratios. The resulting data structure is a three-dimensional array, with its length and width corresponding to spatial location and its depth corresponding to the signal intensity of the six isotope channels.

[0030] In a preferred embodiment, the unmixing of the multidimensional image data cube includes: The multidimensional image data cube is unfolded into a two-dimensional observation matrix. Where m is the number of ion channels and n is the total number of pixels; construct standard mass spectrometry feature vectors containing pure lignin, pure quartz, pure kaolinite, pure hematite, and Cd standards, and use them as the initial values ​​of the basis matrix W after normalization; randomly initialize the non-negative abundance matrix H; minimize the objective function through iterative update rules: ; ; Until the relative rate of change of the objective function value is less than 10 -5 Output the abundance matrix H, reshape it to the image size, and obtain the purity abundance distribution map of lignin, minerals and Cd.

[0031] The 256×256×6 data cube is reshaped into a 6×65536 two-dimensional matrix V, where each column represents the spectral fingerprint of a pixel. Prior knowledge is introduced during model initialization, using the proportion of secondary ion intensities of experimentally measured pure substance standard samples as the initial column vector of the basis matrix W to avoid the algorithm getting trapped in local optima. The optimization process employs a multiplicative update rule. In each iteration, the abundance matrix H is updated according to the gradient direction of the reconstruction error. Simultaneously, L1 regularization is applied to the basis matrix W to maintain its sparsity, assuming that each pixel is mainly composed of a few components. Convergence is determined when the objective function value decreases by less than one ten-thousandth between two consecutive iterations. The output matrix H contains the relative contribution of each endmember component to each pixel. Rearranging it according to the original scan row and column numbers yields an intensity map of chemical components that not only contains spatial distribution information but also eliminates mixed signal interference. Since this invention does not specifically quantify, the intensity map is used as an abundance map for simplicity.

[0032] In a preferred embodiment, calculating the Euclidean distance d1 between each Cd-containing pixel and the nearest lignin region boundary and the nearest mineral region boundary includes: With an abundance threshold of 0.3, the purity abundance distribution maps of lignin and minerals are binarized to generate a lignin binary mask M. lig and mineral binary mask M min The Euclidean distance transformation algorithm is used to calculate the distance from each pixel to M. lig The minimum distance to the non-zero region is d1, and the distance to M is... min The minimum distance between non-zero regions is defined as d2; if a pixel is located inside the mask, the corresponding distance is defined as 0; when calculating the probability, if d1=0 and d2=0 for a certain pixel, the interface bonding probability P is directly set to P. 界面 =1, lignin adsorption probability P 木质素 =0, mineral retention probability P 矿物 =0, to avoid the singularity of a zero denominator.

[0033] Based on statistical distribution characteristics, the abundance map obtained from NMF decomposition is converted into a binary image. Pixels with an abundance value greater than 0.3 are defined as the effective presence region of that component, resulting in two Boolean-type mask images. The entire image is scanned using an Euclidean distance transformation algorithm. For each pixel outside the mask, the linear Euclidean distance between it and the nearest pixel inside the mask is calculated. This distance reflects the spatial proximity of Cd ions to the surface of organic matter or minerals for diffusion or adsorption. To handle overlapping boundary cases, a priority logic is set: when a pixel is simultaneously within the mask range of lignin and minerals, it indicates the presence of an organic-inorganic complex structure, and the chemical morphology of this point is forcibly classified as an interfacial complex state.

[0034] In a preferred embodiment, the step of combining the Cd purity abundance of each pixel to generate a high-resolution soil Cd occurrence probabilities map includes: Set the normalized scale factor L = 50 nm; for each Cd-containing pixel, use the calculated P 木质素 P 矿物 and P 界面 Construct an RGB image matrix; P 矿物 Mapped to the red channel R, P 木质素 Mapped to green channels G, P 界面 Mapped to the blue channel B; the channel values ​​are quantized to 8 bits and a false-color image is synthesized to obtain a probability map of soil Cd occurrence forms showing the distribution probability of Cd in three forms: lignin adsorption, mineral immobilization and interface complexation.

[0035] Probability calculation is based on the principle of spatial proximity and utilizes the exponential decay function. The calculated distances d1 and d2 are converted into probability values, where L = 50 nanometers is used as the characteristic adsorption radius to ensure that the probability value decreases rapidly with increasing distance. The three calculated probability components represent the probability of Cd being adsorbed by minerals, lignin, and complexed at the interface between the two, respectively. To achieve visualization, a three-dimensional array is created as an image container. The values ​​representing the mineral adsorption probability are filled into the first-dimensional red channel, the lignin adsorption probability into the second-dimensional green channel, and the interface complexation probability into the third-dimensional blue channel. The floating-point probability values ​​between 0 and 1 are multiplied by 255 and rounded to the nearest integer, converting them into 8-bit unsigned integers according to computer image standards, thus synthesizing a high-resolution RGB false-color image. The pixels of different colors in the image represent the main chemical occurrence forms of Cd in the micro-region.

[0036] It should be understood that in the foregoing description of the embodiments in this specification, various features are combined in a single embodiment, drawing, or description for the purpose of simplifying the description and to aid in understanding a feature. However, this does not mean that the combination of these features is necessary, and those skilled in the art, upon reading this specification, may readily identify some of the devices as separate embodiments. That is, the embodiments in this specification can also be understood as an integration of multiple secondary embodiments. And the content of each secondary embodiment is valid even if it contains fewer than all the features of a single foregoing disclosed embodiment.

Claims

1. A dynamic monitoring method for lignin-regulated soil Cd bioavailability, characterized in that, Includes the following steps: Micro-XRF elemental surface distribution data of the soil sample to be tested are acquired. The elements include Cd, carbon (C) representing organic matter, and silicon (Si), aluminum (Al), and iron (Fe) representing minerals. For each pixel, the first spatial correlation between the Cd signal gradient and the carbon signal gradient, and the second spatial correlation between the Cd signal gradient and the gradient of the weighted sum of the silicon, aluminum, and iron signals are calculated in its neighborhood. When the first spatial correlation is greater than a preset positive threshold and the second spatial correlation is less than a preset negative threshold, the pixel is determined to be the target micro-domain. NanoSIMS analysis was performed on the target micro-domain to obtain a multidimensional image data cube, the data cube including Cd and lignin-representing elements. 12 C - , 12 C 14 N - Ions, representing minerals 28 Si - , 27 Al 16 O - , 56 Fe 16 O - ion; Based on a pre-established reference ion mass spectrometry library of pure lignin, pure mineral phases and Cd standards, the multidimensional image data cube is unmixed to obtain a purity abundance distribution map of lignin, minerals and Cd within the target microdomain. A high-resolution probability map of soil Cd occurrence forms is generated based on the purity abundance distribution map.

2. The method according to claim 1, characterized in that, The process of generating a high-resolution soil Cd occurrence probabilities map based on the purity abundance distribution map specifically involves: Calculate the Euclidean distance d1 between each Cd-containing pixel and the nearest lignin region boundary and the Euclidean distance d2 between each pixel and the nearest mineral region boundary; construct the Cd interface retention index I=exp(-(d1+d2) / L), where L is the normalized scaling factor; The value of the Cd interface retention index I is taken as the interface complexation probability Pinterface; and according to the formula Pinterface 木质素 =(1-I)×d2 / (d1+d2) and P 矿物 The lignin adsorption probability P is calculated as (1-I)×d1 / (d1+d2). 木质素 and the probability of mineral retention P 矿物 By combining the Cd purity and abundance of each pixel, a high-resolution probability map of soil Cd occurrence morphology is generated.

3. The method according to claim 1, characterized in that, The acquisition of Micro-XRF elemental surface distribution data of the soil sample to be tested includes: The collected soil samples were dried to constant weight in a 40°C constant temperature oven, and gravel and roots with a diameter greater than 2 mm were removed. The treated soil samples were placed in a mold, and molten low-melting-point sublimed sulfur or metallic indium was added. The samples were then impregnated and embedded in a vacuum environment. After cooling and solidification, a solidified sample was formed. The surface of the solidified sample was polished stepwise using diamond polishing paste, with polishing pastes of 9 μm, 3 μm, 1 μm and 0.25 μm used in sequence, until the surface roughness Ra < 100 nm. On a micro-area X-ray fluorescence spectrometer, the excitation voltage was set to 50 kV, the tube current to 600 μA, and the scanning step size to 20 μm. The polished sample surface was scanned in a vacuum environment, and the characteristic X-ray fluorescence intensities of Cd, C, Si, Al and Fe were collected to generate a two-dimensional gray-scale intensity matrix for each element.

4. The method according to claim 1, characterized in that, The step of calculating the first spatial correlation between the Cd signal gradient and the carbon signal gradient in the neighborhood of each pixel, and the second spatial correlation between the Cd signal gradient and the gradient of the weighted sum of silicon, aluminum, and iron signals, includes: Define a 5×5 pixel rectangular area centered on the current pixel as the neighborhood; Gaussian smoothing filtering was applied to the two-dimensional gray-level intensity matrices of five elements: Cd, C, Si, Al, and Fe. The Sobel operator is used to convolve the filtered Cd matrix and C matrix respectively to obtain the Cd gradient magnitude vector G. Cd and C gradient magnitude vector G C G is calculated using the Pearson correlation coefficient formula. Cd and G C The correlation coefficient between them is denoted as the first spatial correlation. The filtered Si, Al, and Fe matrices are normalized to their maximum values, and then the three normalized matrices are summed with equal weights to obtain the mineral signal weighted sum matrix M. sum ; Using the Sobel operator on M sum Perform convolution operation to obtain the mineral composite gradient magnitude vector G. min G is calculated using the Pearson correlation coefficient formula. Cd and G min The correlation coefficient between them is denoted as the second spatial correlation.

5. The method according to claim 4, characterized in that, The step of determining the pixel as the target micro-domain when the first spatial correlation is greater than a preset positive threshold and the second spatial correlation is less than a preset negative threshold includes: Construct a binary discrimination matrix and retain pixels that satisfy the condition that the first spatial correlation is greater than a preset positive threshold and the second spatial correlation is less than a preset negative threshold; A morphological opening operation is performed on the retained pixel set to remove isolated noise points with an area of ​​less than 3 pixels. The remaining connected regions are marked as target micro-domains to be analyzed, and the geometric center coordinates of the target micro-domains in the Micro-XRF coordinate system and the optical feature images used for relocation are recorded.

6. The method according to claim 1, characterized in that, The step of performing NanoSIMS analysis on the target micro-domain to obtain a multidimensional image data cube includes: The sample was transferred to the sample chamber of the nanosecondary ion mass spectrometer. Based on the recorded geometric center coordinates, the target microdomain was calibrated and located using an optical microscope and ion imaging. A cesium ion beam was used as the primary ion source, with a beam current intensity set to 10 picoamperes, to pre-sputter the surface of the target microdomain until the secondary ion yield reached a steady state. The image resolution was set to 256 × 256 pixels, and the residence time was 5 milliseconds per pixel, with simultaneous data collection. 12 C - , 12 C 14 N - , 28 Si - , 27 Al 18 O - , 56 Fe 16 O - The multidimensional image data cube is constructed using the signal intensity of characteristic secondary ions containing Cd.

7. The method according to claim 1, characterized in that, The demixing of the multidimensional image data cube includes: The multidimensional image data cube is unfolded into a two-dimensional observation matrix. Where m is the number of ion channels and n is the total number of pixels; construct standard mass spectrometry feature vectors containing pure lignin, pure quartz, pure kaolinite, pure hematite, and Cd standards, and use them as the initial values ​​of the basis matrix W after normalization; randomly initialize the non-negative abundance matrix H; minimize the objective function through iterative update rules: ； ； Until the relative rate of change of the objective function value is less than 10 -5 Output the abundance matrix H, reshape it to the image size, and obtain the purity abundance distribution map of lignin, minerals and Cd.

8. The method according to claim 2, characterized in that, The calculation of the Euclidean distance d1 between each Cd-containing pixel and the nearest lignin region boundary and the nearest mineral region boundary includes: With an abundance threshold of 0.3, the purity abundance distribution maps of lignin and minerals are binarized to generate a lignin binary mask M. lig and mineral binary mask M min The Euclidean distance transformation algorithm is used to calculate the distance from each pixel to M. lig The minimum distance to the non-zero region is d1, and the distance to M is... min The minimum distance between non-zero regions is defined as d2; if a pixel is located inside the mask, the corresponding distance is defined as 0; when calculating the probability, if d1=0 and d2=0 for a certain pixel, the interface bonding probability P is directly set to P. 界面 =1, lignin adsorption probability P 木质素 =0, mineral retention probability P 矿物 =0, to avoid the singularity of a zero denominator.

9. The method according to claim 8, characterized in that, The process of combining the Cd purity abundance of each pixel to generate a high-resolution soil Cd occurrence morphology probability map includes: Set the normalized scale factor L = 50 nm; for each Cd-containing pixel, use the calculated P 木质素 P 矿物 and P 界面 Construct an RGB image matrix; P 矿物 Mapped to the red channel R, P 木质素 Mapped to green channels G, P 界面 Mapped to the blue channel B; the channel values ​​are quantized to 8 bits and a false-color image is synthesized to obtain a probability map of soil Cd occurrence forms showing the distribution probability of Cd in three forms: lignin adsorption, mineral immobilization and interface complexation.

10. A dynamic monitoring system for lignin-regulated soil Cd bioavailability, characterized in that, Includes the following units: The target micro-domain determination unit is used to acquire Micro-XRF elemental surface distribution data of the soil sample to be tested. The elements include Cd, carbon (C) representing organic matter, and silicon (Si), aluminum (Al), and iron (Fe) representing minerals. For each pixel, the first spatial correlation between the Cd signal gradient and the carbon signal gradient, and the second spatial correlation between the Cd signal gradient and the gradient of the weighted sum of the silicon, aluminum, and iron signals are calculated in its neighborhood. When the first spatial correlation is greater than a preset positive threshold and the second spatial correlation is less than a preset negative threshold, the pixel is determined to be the target micro-domain. The NanoSIMS analysis unit is used to perform NanoSIMS analysis on the target microdomain to obtain a multidimensional image data cube, the data cube including Cd and lignin-representing elements. 12 C - , 12 C 14 N - Ions, representing minerals 28 Si - , 27 Al 16 O - , 56 Fe 16 O - ion; The demixing unit is used to demix the multidimensional image data cube based on a pre-established reference ion mass spectrometry library of pure lignin, pure mineral phases and Cd standards, to obtain a purity abundance distribution map of lignin, minerals and Cd within the target microdomain. The probability map generation unit is used to generate a high-resolution soil Cd occurrence probabilistic map based on the purity abundance distribution map.

Images