A method, prediction model and application for predicting available cadmium content in soil.

Abstract

This invention discloses a method, model, and application for predicting the available cadmium content in soil, relating to the field of soil nutrient prediction technology. The method includes the following steps: (1) collecting multiple farmland soil samples and measuring key nutrient indicators; (2) recording, statistically analyzing, and plotting the measured data; (3) performing correlation analysis on all statistical data to determine indicators significantly correlated with available cadmium; (4) establishing a prediction model for available cadmium in soil using multiple linear regression analysis; and calculating the available cadmium content in the soil based on the prediction model. The prediction model and method provided by this invention are simple, have few variable parameters, and high prediction accuracy, effectively predicting the available cadmium content in soil and providing a reference for the scientific management of cadmium-polluted farmland.

Description

Technical Field

[0001] This invention relates to the field of soil nutrient prediction technology, specifically to a method, prediction model, and application for predicting the available cadmium content in soil. Background Technology

[0002] Farmland soil is an important production resource, providing crops with 18 essential nutrients for their survival, including C, H, O, N, S, P, B, Si, K, Na, Ca, Mg, Mn, Cl, Fe, Cu, Zn, and Mo. While crop root hair cells absorb these elements through cation exchange, they also absorb some toxic heavy metals, such as Cd, Cr, Pb, Hg, and Ni, with Cd being particularly prone to accumulation in rice.

[0003] The complex interactions between crop nutrients and heavy metals in farmland soil cause their content to be constantly changing. Taking Cd as an example, within an agricultural cycle, Cd is mainly input into farmland through fertilizers, irrigation water, and atmospheric deposition, and is output from farmland through irrigation water and rice harvesting. Currently, the publicly known ways influencing Cd content include: 1. Foliar spraying of silicon and selenium fertilizers can reduce rice's absorption of Cd, trapping Cd in the soil; 2. Applying biochar, such as rice straw char, can significantly reduce the content of available Cd in the soil that can be absorbed by rice; 3. Lowering soil pH can increase the content of available Cd in the soil; 4. Applying high amounts of ammonium nitrogen fertilizer, such as urea (amide N), can significantly increase the Cd content in rice grains; 5. Soil buffering capacity also affects rice's absorption of Cd, thus affecting the Cd content in the soil; 6. Overuse of phosphorus fertilizer (especially from phosphate rock) can inhibit rice growth and also increase the risk of heavy metal pollution such as Cd, Cu, Mn, and Pb in the soil; 7. Among potassium fertilizers, KCl can promote rice's absorption of Cd, while K2SO4 reduces the content of available Cd in the soil; 8. S fertilizer combined with water management can reduce rice's absorption of Cd; 9. High concentrations of Fe... 2+ Mn fertilizer and foliar Zn fertilizer can effectively inhibit rice from absorbing Cd from the soil; 10. Organic fertilizer can reduce or promote Cd accumulation in rice, but long-term overuse of organic fertilizer will promote rice's absorption of Cd.

[0004] Currently, regression analysis is increasingly being used to establish models for assessing soil heavy metal content. These models typically estimate the ionic and dissolved heavy metal content in soil based on the soil's heavy metal content and basic physicochemical properties. Furthermore, these models are characterized by requiring fewer parameters, ease of modeling, and high accuracy.

[0005] However, there are no existing methods for predicting the available cadmium content in soil, which cannot provide a reference for the scientific management of cadmium-polluted farmland. Summary of the Invention

[0006] This invention provides a method, model, and application for predicting the available cadmium content in soil, aiming to solve the problems existing in the above-mentioned background art.

[0007] To achieve the above-mentioned technical objectives, the present invention mainly adopts the following technical solutions:

[0008] In a first aspect, the present invention discloses a method for predicting the available cadmium content in soil, comprising the following steps:

[0009] (1) Collect soil samples from multiple farmlands and measure key nutrient indicators respectively;

[0010] (2) Record the measured data, and compile and plot the data;

[0011] (3) Conduct correlation analysis on all statistical data to determine the indicators that are significantly correlated with available cadmium;

[0012] (4) A prediction model for available cadmium in soil was established using multiple linear regression analysis.

[0013] (6) Calculate the available cadmium content in the soil based on the prediction model.

[0014] Secondly, the present invention discloses a prediction model for the available cadmium content in soil obtained by the prediction method described in the first aspect.

[0015] Thirdly, the present invention discloses the application of a prediction model as described in the second aspect in predicting the available cadmium content in soil.

[0016] Compared with the prior art, the present invention has the following beneficial effects:

[0017] The prediction model and method provided by this invention are simple, have few variable parameters, and have high prediction accuracy. They can effectively predict the content of available cadmium in the soil, providing a reference for the scientific management of cadmium-polluted farmland. Attached Figure Description

[0018] Figure 1 This is a frequency distribution characteristic diagram of farmland soil physicochemical indicators provided in the embodiments of the present invention;

[0019] Figure 2 This is a frequency distribution density map of soil physicochemical properties provided in this embodiment of the invention;

[0020] Figure 3This section shows the Spearman correlation coefficients of Zhejiang farmland soil parameters provided in this embodiment of the invention; where *** indicates significance at the P < 0.001 level; ** indicates significance at the P < 0.01 level; and * indicates significance at the P < 0.05 level. The area above the diagonal is the correlation coefficient matrix; the diagonal lines are histograms of the distribution of each variable (soil parameter); the area below the diagonal is a scatter plot, where the curves are local regression curves (loess curves); the point at the center of the ellipse is the point determined by the means of the two variables.

[0021] Figure 4 This section shows the Pearson correlation coefficients of the soil parameters from farmland in Zhejiang Province provided in this embodiment of the invention; *** indicates significance at the P < 0.001 level; ** indicates significance at the P < 0.01 level; * indicates significance at the P < 0.05 level. The area above the diagonal is the correlation coefficient matrix; the diagonal lines are histograms of the distribution of each variable (soil parameter); the area below the diagonal is a scatter plot, where the curves are local regression curves (loess curves); the point at the center of the ellipse is the point determined by the means of the two variables.

[0022] Figure 5 This is an effective cadmium (Cd) regression diagnostic chart based on ordinary least squares provided in an embodiment of the present invention. Detailed Implementation

[0023] The present invention will be further described below with reference to embodiments. These embodiments are descriptive and not limiting, and should not be construed as limiting the scope of protection of the present invention. Unless otherwise specified, the raw materials used in the present invention are all commercially available products. Unless otherwise specified, the methods used in the present invention are conventional methods in the art. The mass of each substance used in the present invention is a conventionally used mass.

[0024] This invention discloses a method for predicting the available cadmium content in soil, comprising the following steps:

[0025] (1) Collect soil samples from multiple farmlands and measure key nutrient indicators respectively;

[0026] (2) Record the measured data, and compile and plot the data;

[0027] (3) Conduct correlation analysis on all statistical data to determine the indicators that are significantly correlated with available cadmium;

[0028] (4) A prediction model for available cadmium in soil was established using multiple linear regression analysis.

[0029] (5) Calculate the available cadmium content in the soil based on the prediction model.

[0030] In a preferred embodiment of the present invention, in step (1), soil samples are taken from each plot of farmland in an “S” pattern, with a soil sampling depth of 8-15 cm.

[0031] In a preferred embodiment of the present invention, in step (1), the key nutrient indicators include soil pH and component content, wherein the components include organic matter, total nitrogen, available nitrogen, available phosphorus, available potassium, slow-release potassium, boron, molybdenum, iron, manganese, copper, zinc and available cadmium.

[0032] In a preferred embodiment of the present invention, in step (2), the measured data are recorded, statistically analyzed and plotted, including plotting the number of sampling points in various locations, histograms or frequency distribution maps, frequency density maps, multivariate correlation scatter matrix maps, establishing linear regression based on ordinary least squares, diagnosing the regression model, adding or deleting strong influence points in abnormal observations, and selecting variables and establishing robust regression for full subset regression.

[0033] In a preferred embodiment of the present invention, in step (3), the Spearman correlation coefficient is used, or when the observed value of the variable exceeds 500, the Pearson correlation coefficient is used for correlation analysis.

[0034] In a preferred embodiment of the present invention, in step (3), the indicators that are significantly correlated with available cadmium are the contents of soil copper, zinc, iron, manganese, total nitrogen, boron, molybdenum and available phosphorus.

[0035] Preferably, in step (4), a correlation analysis is performed using the robust regression M estimation method to establish a prediction model for available cadmium in the soil.

[0036] Furthermore, in step (4), the prediction model is:

[0037] Cd=0.0570-0.0481B-0.283Mo-0.0459Cu:B+0.0418tN:Mo+0.818B:Mo-0.122Zn:B:Mo-0.0411t N:B:Mo

[0038] Where Cd represents the content of available cadmium in the soil, the unit is mg·kg -1 B, Mo, Cu, tN, and Zn represent the contents of boron, molybdenum, copper, total nitrogen, and zinc in the soil, respectively, in mg·kg⁻¹. -1 The colon ":" indicates an interactive or interactive relationship.

[0039] The present invention also discloses a prediction model for the available cadmium content in soil obtained by the prediction method described above.

[0040] The present invention also discloses the application of the prediction model described above in predicting the available cadmium content in soil.

[0041] The following is a description through specific embodiments.

[0042] Example

[0043] 1. Experimental subjects: This invention uses typical farmland soil in Zhejiang Province as the subject. Soil samples were collected from Zhejiang Soil Fertility and Environmental Monitoring Stations from 2008 to 2018 and measured in the same year.

[0044] 2. Test methods:

[0045] A method for predicting the available cadmium content in soil includes the following steps:

[0046] (1) Collect soil samples from multiple farmlands and measure key nutrient indicators respectively.

[0047] This invention collects, processes, and stores soil samples according to the agricultural industry standard NYT 1121.1-2006, Soil Testing Part 1. Specifically, an "S"-shaped sampling method is used, identifying 15 evenly distributed sampling points. 500g of soil is collected from each point at a depth of 8-15cm. The soil samples from these 15 points are mixed thoroughly, and then 1kg is collected as a final sample. A total of 355 soil samples are collected from different sampling points.

[0048] In the laboratory, the samples were cleaned, air-dried, ground, and then passed through sieves of 2.0, 0.25, and 0.15 mm respectively. The samples were then sealed in plastic bags and stored for testing.

[0049] The key nutrient indicators in this invention include soil pH and component content, wherein the components include organic matter, total nitrogen, available nitrogen, available phosphorus, available potassium, slow-release potassium, boron, molybdenum, iron, manganese, copper, zinc, and available cadmium. A normality statistical test was performed on the soil pH and the content of 13 nutrient indicators, with a sample size of 3564, which is greater than 2000.

[0050] For each soil sample, soil pH and the contents of the above-mentioned nutrients were determined according to the following methods: the determination methods for soil pH, organic matter, total nitrogen, available nitrogen, and available phosphorus were in accordance with "Soil Agrochemical Analysis"; available potassium and slow-release potassium were determined according to the method of agricultural industry standard NY / T 889-2004; boron in the soil was determined by boiling extraction-methyleneimine-H colorimetric method (NY / T 1121.8-2006); molybdenum was determined by oxalic acid-ammonium oxalate extraction-polarography method (NY / T 1121.9-2012); iron, manganese, copper, zinc, and available cadmium were determined by diethylenetriaminepentaacetic acid (DTPA) extraction-inductively coupled plasma atomic emission spectrometry (HJ 804-2016).

[0051] The data of key soil nutrient indicators measured are shown in Table 1.

[0052] Table 1. Characteristics of soil index contents measured.

[0053]

[0054]

[0055] As shown in Table 1, the average pH of farmland soil in Zhejiang Province was 5.81, close to the median of 5.68, with a coefficient of variation of 12.8%. The average contents of soil organic matter, total nitrogen (tN), and available nitrogen (aN) were 34.1 g·kg⁻¹, 2.06 g·kg⁻¹, and 2.06 g·kg⁻¹, respectively. -1 and 157 mg·kg -1 The coefficient of variation was 34.5%–36.7%. The mean contents of available potassium (aK), slow-release potassium (saK), and iron (Fe) in the soil were 101, 332, and 248 mg·kg⁻¹, respectively. -1 The coefficient of variation was 70.5%–90.7%, which is considered moderate. The coefficients of variation for available phosphorus (aP), manganese (Mn), copper (Cu), zinc (Zn), boron (B), molybdenum (Mo), and cadmium (Cd) in the soil were greater than 100%, indicating that the arithmetic mean representing their contents does not accurately reflect their true content.

[0056] For example, the arithmetic mean of cadmium content in the soil was 2.74 mg·kg⁻¹, exceeding the risk control limit of 2.0 mg·kg⁻¹ under the conditions of 5.5 ≤ pH ≤ 6.5, as specified in the "Soil Environmental Quality Standard for Agricultural Land Soil Pollution Risk Control" (GB 15618—2018). -1 The concentration was 1.37 times higher, reaching the level of cadmium pollution that requires strict control measures.

[0057] However, in a total of 3564 samples, none exceeded the standard (2.0 mg / kg). -1 Of the 3468 samples, 97.3% exceeded the standard by 1 to 5 times (2.0 to 10 mg / kg). -1 There were 0 sampling points with a concentration exceeding the standard by 5 to 10 times (10 to 20 mg / kg). -1 There were 0 sampling points with a concentration exceeding the standard by 10 to 50 times (20 to 100 mg / kg). -1 72 sampling points were found to have levels exceeding the standard by 50 to 100 times (100 to 200 mg / kg). -1 24 sampling points were found to contain levels exceeding the standard by more than 100 times (200 mg / kg). -1 There are 0 points.

[0058] (2) Record the measured data, and compile and plot the data.

[0059] Data was recorded using Microsoft Excel 365, and statistical analysis and plotting were performed using R (V4.3.3) software. This included plotting the number of sampling points in various locations, histograms or frequency distribution plots, frequency density plots, and scatter plots of multivariate correlations. A linear regression model based on ordinary least squares (i.e., multiple linear regression with interaction terms) was established. Diagnosis of the regression model was performed (i.e., testing the statistical hypotheses of the OLS regression model, including testing for normality, independence, linearity, and homoscedasticity). Strongly influential points in outlier observations were added or removed. Variable selection and robust regression were also performed for full subset regression.

[0060] The frequency distribution of soil physicochemical properties is shown in the figure. Figure 1 The frequency distribution density of the soil physicochemical properties is shown in the figure. Figure 2 .

[0061] Depend on Figure 1 It can be seen that the pH of farmland soil is mainly distributed between 4.5 and 7, and the organic matter (oC) content is distributed between 20 and 50 g·kg⁻¹. -1 Total nitrogen (tN) is 0–4 g·kg⁻¹ -1 Available nitrogen (aN) is 50–250 mg·kg⁻¹. -1 Available phosphorus (aP) is 50–100 mg / kg. -1 Available potassium (aK) is 0–200 mg / kg. -1 Slow-release potassium (saK) is 36–600 mg / kg. -1 Iron (Fe) is 0–400 mg·kg⁻¹ -1 Manganese (Mn) is 0–200 mg·kg -1 Copper (Cu) is 0–40 mg·kg⁻¹ -1 Zinc (Zn) is 0–500 mg / kg. -1 Boron (B) is 0–5 mg / kg. -1 Molybdenum (Mo) is 0–10 mg·kg⁻¹ -1 Cadmium (Cd) is 0–50 mg / kg. -1 .

[0062] Depend on Figure 2 It can be seen that the frequencies of pH values ​​of farmland soils of 4.5–5.5, 5.5–6.5, and 6.5–7.5 are 40.4%, 35.9%, and 17.1%, respectively (frequency = frequency density * group moment = 0.404 * 1, the same below), and the organic matter (oC) content is 20–50 g·kg. -1 The frequency was 78% (0.023*10+0.034*10+0.021*10), and the total nitrogen (tN) content was 0-4 g·kg. -1The frequency was 99.6% (0.249*4), and the available nitrogen (aN) content was 50–250 g·kg⁻¹. -1 The frequency was 96.1% (3426 / 3564 calculated by the percentage method), and the available phosphorus (aP) content was 50–150 mg·kg⁻¹. -1 The frequency was 95% (0.019*50), and the available potassium (aK) content was 0–200 mg·kg⁻¹. -1 The frequency was 100% (0.005*200), and the slow-acting potassium (saK) content was 36–600 mg·kg. -1 The frequency was 82.8% (0.002*264+0.001*300), and the iron (Fe) content was 0-400 mg·kg. -1 The frequency was 100% (0.003*200+0.002*200), and the manganese (Mn) content was 0~200mg·kg. -1 The frequency was 100% (0.005*200), and the copper (Cu) content was 0–40 mg·kg⁻¹. -1 The frequency was 100% (0.025*40), and the zinc (Zn) content was 0–500 mg·kg. -1 The frequency was 100% (0.002*500), and the boron (B) content was 0-5 mg·kg. -1 The frequency was 98% (0.2*5), and the molybdenum (Mo) content was 0–10 mg·kg⁻¹. -1 The frequency was 91% (0.091*10), and the cadmium (Cd) content was 0–50 mg·kg. -1 The frequency was 95% (0.019*50).

[0063] (3) Conduct correlation analysis on all statistical data to identify indicators that are significantly correlated with available cadmium.

[0064] In this invention, since the sample size of the 14 variables in the experimental soil was 3564 after normality statistical testing, which is greater than 2000, the KS (Kolmogorov-Smirnov) test was chosen. However, the KS test showed that none of the 14 variables conformed to a normal distribution, as shown in Table 2 below.

[0065] Table 2. KS test results of the contents of 14 components in the experimental soil.

[0066]

[0067] For the 14 variables that do not conform to a normal distribution, pairwise correlation analysis was performed using Spearman correlation coefficient analysis. Spearman correlation coefficient is based on data values, not data values, and is not affected by outliers. Therefore, it is robust to outliers and has no restrictions on data types, such as conforming to a normal distribution or requiring a sufficient sample size. It can handle specific types of nonlinear relationships. The results are shown in […]. Figure 3 .

[0068] from Figure 3 The results show that available cadmium is significantly positively correlated with copper (Cu), zinc (Zn), iron (Fe), manganese (Mn), and total nitrogen (tN) (at the P = 0.001 and 0.001 levels), while it is significantly negatively correlated with molybdenum (Mo) (at the P = 0.05 level), boron (B), and available phosphorus (aP) (at the P = 0.001 level). Therefore, it is inferred that soil copper (Cu), zinc (Zn), iron (Fe), manganese (Mn), total nitrogen (tN), boron (B), molybdenum (Mo), and available phosphorus (aP) may be important factors affecting the content of available cadmium in soil.

[0069] Furthermore, this invention also examines the use of Pearson correlation coefficient for prediction. When the number of observed values ​​for a variable exceeds 500, it can be approximated as a normal distribution. According to the Central Limit Theorem, when the amount of data is sufficiently large, the data can be considered approximately normally distributed. Since there are 3564 observed values ​​for each variable (soil parameter index) in this invention, the data can be approximated as normally distributed. Therefore, Pearson correlation coefficient can also be used for analysis, and the results are shown in […]. Figure 4 The results are basically the same as those of the Spearman correlation coefficient.

[0070] (4) A prediction model for available cadmium in soil was established using multiple linear regression analysis.

[0071] The correlation analysis in step (3) shows that the soil available cadmium content is significantly correlated with soil copper (Cu), zinc (Zn), iron (Fe), manganese (Mn), total nitrogen (tN), boron (B), molybdenum (Mo) and available phosphorus (aP). Therefore, these eight correlation items are used to predict the soil available cadmium (Cd) content.

[0072] The percentage of abnormal values ​​for available cadmium (Cd) was 168 / 3564 = 4.71%, including 72 values ​​ranging from 0.001 to 0.007 and 96 values ​​ranging from 57.7 to 189 mg / kg. -1 These outliers were 27–180 times smaller or 320–1050 times larger than the median of 0.18. After excluding these outliers, a linear regression was performed on the bioavailable cadmium (Cd) using ordinary least squares (OLS) for each of the eight relevant items (each with 3395 observations). Figure 5 As shown, R is then corrected. 2 The adjusted R-squared value increased from 0.57 to 0.9057. This means that the 255 variables, comprised of the eight related terms and their interaction terms, explained 90.57% of the available cadmium (Cd) data.

[0073] The multiple linear regression equation with 255 independent variables is too large and not concise enough. Therefore, full subset regression is used for variable selection. The recommended variables are adjusted for R... 2 =0.405 was too small and was therefore abandoned. Instead, the weighted least squares (WLS) method, i.e. the M-estimation method of robust regression, was adopted. Bisquare weights were selected for linear fitting, and variables with regression coefficients from large to small were selected. The results are shown in Table 3.

[0074] Table 3 shows the robust regression results of relevant soil parameters on available cadmium.

[0075]

[0076]

[0077] Note: In the table, "*" indicates that all interactive items are included, and ":" indicates an interactive or interaction relationship.

[0078] Statistical analysis using R software revealed that iron, manganese, and phosphorus contributed little to the prediction of available cadmium content in soil; therefore, they were removed to ensure computational simplicity. Based on Table 3, the prediction equation for available cadmium content in soil can be obtained as follows:

[0079] Cd=0.0570-0.0481B-0.283Mo-0.0459Cu:B+0.0418tN:Mo+0.818B:Mo-0.122Zn:B:Mo-0.0411tN:B:Mo.

[0080] (5) Calculate the available cadmium content in the soil based on the prediction model.

[0081] By measuring the contents of boron, molybdenum, total nitrogen, and zinc in farmland soil and substituting them into the prediction equation, the available cadmium content in the soil can be calculated.

[0082] Although embodiments of the invention have been disclosed for illustrative purposes, those skilled in the art will understand that various substitutions, variations, and modifications are possible without departing from the spirit and scope of the invention and the appended claims. Therefore, the scope of the invention is not limited to the contents disclosed in the embodiments.

Claims

1. A method of predicting the effective cadmium content in soil, characterized by, It comprises the following steps: (1) Collecting a plurality of farmland soil samples, and determining key nutrient indexes respectively; in step (1), the key nutrient indexes comprise soil pH and component contents, and the components comprise organic matter, total nitrogen, available nitrogen, available phosphorus, available potassium, slow-acting potassium, boron, molybdenum, iron, manganese, copper, zinc and available cadmium; (2) Recording the determined data, and performing statistics and drawing; (3) Performing correlation analysis on all the statistical data, and determining indexes having significant correlation with available cadmium; the indexes having significant correlation with available cadmium are soil copper, zinc, iron, manganese, total nitrogen, boron, molybdenum and available phosphorus contents; the correlation analysis is performed by using a spearman correlation coefficient or a pearson correlation coefficient when the number of observation values of a variable is more than 500; (4) Establishing a prediction model of available cadmium in soil by a multiple linear regression analysis method; the prediction model is: Cd = 0.0570-0.0481B-0.283Mo-0.0459Cu:B+0.0418tN:Mo+0.818B:Mo-0.122Zn:B:Mo-0.0411tN:B:Mo Wherein, Cd represents the content of effective cadmium in soil, the unit is mg·kg -1 , B, Mo, Cu, tN, Zn respectively represent the content of boron, molybdenum, copper, total nitrogen and zinc in soil, the unit is mg·kg -1 , " indicates the interaction or interaction relationship; (5) Calculating the content of available cadmium in soil according to the prediction model.

2. The method of predicting the effective cadmium content in soil according to claim 1, characterized by, In step (1), each farmland soil is sampled according to an "S" shape, and the soil sampling depth is 8-15 cm.

3. The method of predicting the effective cadmium content in soil according to claim 1, characterized by, In step (2), the determined data are recorded, and statistics and drawing are performed, including drawing a sampling point number diagram of each region, a histogram or frequency distribution diagram, a frequency density diagram, a multivariate correlation scatter matrix diagram, establishing a linear regression based on an ordinary least square method, performing regression model diagnosis, performing addition and deletion processing on strong influence points in abnormal observation values, and performing variable selection and establishing robust regression on a full subset regression.

4. A prediction model of the content of available cadmium in soil obtained by the prediction method according to any one of claims 1-3.

5. Application of the prediction model according to claim 4 in predicting the content of available cadmium in soil.

Images