A method and system for evaluating saline-alkali soil treatment effect based on multi-source data

Abstract

This invention discloses a method and system for evaluating the effectiveness of saline-alkali land remediation based on multi-source data. The method includes: acquiring amendment data, monitoring data, and remediation standards for coastal saline soil; performing temporal registration and feature decomposition on the amendment data and monitoring data; extracting boundary salt control indicators based on the amendment data and water-salt transport conditions of unit plots; constructing a water-salt transport network for plot units using a directed acyclic graph and training a Bayesian network to obtain the optimal indicators for the application rate, depth, and time of the amendment; and calculating a confidence matrix based on the optimal indicators and the upper and lower bounds of the remediation results to obtain the confidence level and difference in remediation effectiveness. This method utilizes spatial stratified sampling, variable-point temporal decomposition, and a confidence matrix to assess the reliability of remediation results, distinguishing differences in remediation effectiveness, making the evaluation of saline-alkali land remediation more refined and dynamic, and enhancing the rational allocation of remediation resources.

Description

Technical Field

[0001] This invention relates to the field of saline-alkali land management, and in particular to a method and system for evaluating the effectiveness of saline-alkali land management based on multi-source data. Background Technology

[0002] Soil salinization is one of the core environmental problems globally that restricts agricultural production and threatens ecological security. These saline-alkali lands, due to high soil salinity, pH imbalance, and soil compaction, hinder vegetation growth, resulting not only in the waste of arable land resources but also exacerbating ecological chain reactions such as desertification and soil erosion. This impacts regional agricultural sustainable development and national ecological security. With increasing emphasis on arable land protection, ecological restoration, and food security, saline-alkali land management has become a key objective for ecological restoration and agricultural quality improvement. Water conservancy reform plays a crucial role in saline-alkali land management. A series of treatment technologies, including chemical improvement, biological improvement, and comprehensive improvement, have been widely used in different scenarios. Coastal saline-alkali land often adopts a combination of underground drainage and desulfurized gypsum improvement, while inland severely saline-alkali land relies on leaching irrigation, biochar improvement, and salt-tolerant plant planting technology. Farmland saline-alkali land uses drip irrigation for salt control, compound amendments, and salt-tolerant crop rotation as the core of treatment. While saline-alkali land treatment technologies are being applied on a large scale, the scale and demand for treatment of coastal saline-alkali land are gradually increasing. The scientific evaluation of treatment effects is gradually becoming a core link in ensuring treatment quality and optimizing treatment strategies.

[0003] In the assessment of saline-alkali land management, manual on-site sampling and testing are the main methods. Data such as soil salinity, pH value, and vegetation coverage are collected for accurate judgment. However, manual sampling and testing have limitations such as small monitoring range, long cycle, and high time and labor consumption, making it difficult to meet the dynamic monitoring needs of management. In recent years, with the development of modern information technologies such as remote sensing, sensor technology, and big data modeling, the technology for assessing the effectiveness of saline-alkali land management has gradually upgraded towards intelligence and precision. UAV hyperspectral remote sensing can quickly invert common indicators such as large-scale vegetation coverage and soil salinity. Soil sensors can collect soil physicochemical data in real time to achieve continuous monitoring. Currently, multi-source data assessment technology has been applied in scenarios such as agricultural saline-alkali land acceptance, ecological restoration project monitoring, and comparative tests of management technologies, providing certain data support for adjusting management plans and evaluating project effectiveness. However, a systematic and full-chain assessment system has not yet been formed. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for evaluating the effectiveness of saline-alkali land management based on multi-source data.

[0005] To achieve the above objectives, the present invention is implemented according to the following technical solution: The first aspect of this invention provides a method for evaluating the effectiveness of saline-alkali land remediation based on multi-source data, comprising: Data on amendments, monitoring data, and treatment standards for coastal saline soils were obtained. Based on these treatment standards, treatment indicators for plot units were determined through spatial stratified sampling according to the type of salinity and the type of water and salt transport. The improver data and monitoring data are time-series registered, and the trend component, seasonal component and regression component are decomposed. The residual uncertainty of the regression component is used as the fluctuation feature. Based on the governance indicators, fluctuation characteristics, trend components and seasonal components, the water and salt transport conditions of the plot unit are inverted, and the boundary salt control index is extracted based on the amendment data and the transport depth in the water and salt transport conditions. Based on the boundary salt control index, a water and salt transport network for plot units is constructed using a directed acyclic graph. A Bayesian network is trained based on the water and salt transport network and amendment data to obtain the optimal indexes for the amount, depth, and time of application of the amendment. Based on the governance indicators and monitoring data, the governance results are obtained. Based on the upper and lower bounds of the optimal indicators and governance results, the confidence matrix is ​​calculated to obtain the confidence level of the governance effect assessment and the difference in governance effect among different plot units in coastal saline soil.

[0006] Furthermore, the method for obtaining the governance indicators includes: Coastal saline soil samples were collected from various spatial locations in the target area. The soil texture of the coastal saline soil samples included sandy, loamy, and clayey soils. Based on the samples, data on soil amendments and monitoring data during the saline-alkali treatment process in the past two quarters were obtained. Based on the monitoring data, the saline-alkali type of each spatial location was determined, and the treatment standards for saline soil were obtained based on the saline-alkali type. Based on the monitoring data, the water and salt transport type is determined by quarterly soil electrical conductivity distribution data and quarterly soil water and salt profile distribution data. If the peak electrical conductivity coincides with the peak time of the water and salt profile distribution and the electrical conductivity of the 0-10cm soil surface layer is higher than that of the 20-30cm deep layer, it is determined to be the evaporative salt accumulation type. If the quarterly electrical conductivity difference is greater than or equal to 30% and the deep layer electrical conductivity is higher than that of the surface layer, it is determined to be the leaching desalination type. If the electrical conductivity increases linearly with the water and salt profile distribution, it is determined to be the capillary ascent type. Based on the target area, fixed-size plots are divided by spatial stratification sampling according to the type of salinity and the type of water-salt transport. The desalination threshold of the plots is determined by the treatment standards to obtain the treatment index of the plots. The sampling combination consists of more than or equal to 6 plots and is evenly distributed in the target area. The salinity types include chloride type, sulfate type, soda type and mixed type.

[0007] Furthermore, the method for obtaining the fluctuation characteristics includes: Based on land parcel units, amendment data and monitoring data are obtained. The amendment data includes amendment type, application rate, and application time. The amendment data and monitoring data are aligned to a weekly time scale according to timestamps. Data with non-overlapping timestamps are linearly interpolated to obtain time-series registration data. Bayesian change-point time-series decomposition is performed on the time-series registration data to obtain trend components, seasonal components, and regression components. The variance of the regression component residuals is used as the fluctuation characteristic. The Bayesian change-point time-series decomposition formula is as follows: ; in land parcel unit exist Fitted values ​​of monitoring data at time points. land parcel unit The The weights of each trend term were obtained by fitting the registered data according to the AIC information criterion. land parcel unit The The rate of change of each trend item land parcel unit The The inflection point time of each trend item land parcel unit Time window weight, land parcel unit The governance time window is determined according to the uniform weighting rule. The total number of trend items. For logistic functions, For the first The amplitude of the sinusoidal term of the seasonal fluctuation. For the first The angular frequency of seasonal fluctuations This represents the total number of seasonal items, expressed in weeks. for The transpose of the explanatory variable matrix at time points, where the explanatory variables are obtained from improver data and monitoring data. The regression coefficient vector is obtained through least squares regression fitting. for The return residual of time, by and The monitoring data was obtained at any given time.

[0008] Furthermore, the method for obtaining the water-salt transport conditions includes: Based on the data obtained from the plot unit, including treatment indicators, fluctuation characteristics, trend components, seasonal components, amendment data, and quarterly soil water and salt profile distribution data, the application data of amendments and the quarterly soil water and salt profile distribution data are used as process driving terms, and the treatment indicators are used as target constraints. Based on the process driving terms and target constraints, the basic state of water and salt balance of the plot unit is calculated through the soil water and salt balance equation. The basic state is corrected by the water and salt fluctuation characteristics of the seasonal components within a unit period, and the balance result of salt input and salt output within a unit period is obtained. Based on the quarterly soil water and salt profile distribution data, the initial salt concentration distribution is used as the initial condition. Based on the equilibrium result and the initial condition, a salt transport simulation model is constructed using the convection diffusion equation. The actual salt data of the quarterly soil water and salt profile distribution data is used as a reference. Based on the reference, the soil saturated hydraulic conductivity and salt diffusion coefficient in the simulation model are iteratively inverted using the SCE-UA optimization algorithm until the root mean square error between the iterative simulation value and the reference is less than 0.3 dS / m, at which point the inversion is terminated. The transport rate and transport depth are obtained and used as the water and salt transport conditions for the plot unit.

[0009] Furthermore, the method for obtaining the boundary salt control index includes: The boundary salinity control index is calculated based on the amendment data and the migration depth in the water-salt migration conditions. The calculation formula for the boundary salinity control index is as follows: ; in land parcel unit Boundary salt control indicators, The correction coefficient for the effect of the modifier is determined based on the slope of the trend component. land parcel unit At application depth The concentration of the modifier at the location, This refers to the application range of the improver. land parcel unit In depth Soil hydraulic conductivity at that location, land parcel unit In depth The soil moisture gradient is obtained based on the soil moisture content data from monitoring.

[0010] Furthermore, the method for obtaining the optimal index includes: Based on the plot unit, the water and salt transport boundary and the application point of the amendment are used as network nodes, the direction of water and salt transport is used as the direction of the edge between the network nodes, and the boundary salt control index is used as the edge weight. A directed acyclic graph is constructed according to the network nodes, edge weights and directions. The monitoring data of the topology and the amendment data are obtained based on the directed acyclic graph. The change in soil electrical conductivity per unit period after the application of amendment is used as the label. Training set data is generated based on the monitoring data of the topology and the amendment data. The training set data dimension is used as the input layer data dimension of the Bayesian network. The Bayesian network is trained using the training set and labels. The nonlinear relationship between the input variables and labels is fitted based on the hidden layer until the deviation between the predicted salt control effect and the actual salt control effect is minimized. The deviation probability distribution of the salt control effect and the trained Bayesian network are obtained. The optimal application index of the amendment is obtained by optimizing the objective function based on the deviation probability distribution. The application index includes the application amount, application depth, and application time. The optimization objective function formula is as follows: ; in This is a set of parameters for the application index of the amendment. For the feasible domain of the parameter, for Divergence, representing the probability distribution of the bias. With the target probability distribution Differences For input variables, land parcel unit Probability distribution of changes in target soil electrical conductivity under treatment indicators The regularization coefficient is . land parcel unit The weights of the constraint terms are obtained based on the characteristic contributions in the water-salt transport conditions. For the trace operation of a matrix, is the inverse of the covariance matrix of the parameter set.

[0011] Furthermore, the method for obtaining the confidence level of the governance effect assessment and the difference in governance effects includes: Based on the land parcel unit, the treatment indicators and monitoring data are obtained. The actual desalination rate is calculated according to the desalination threshold in the treatment indicators and the soil electrical conductivity in the monitoring data. The actual desalination rate is used as the treatment result. Based on the optimal indicators, the expected deviation probability distribution of each land parcel unit is output through the Bayesian network. The mean of the expected deviation probability distribution is used as the expected desalination rate. The fluctuation range of the actual desalination rate is calculated based on the fluctuation characteristics. The actual desalination rate plus the fluctuation range is used as the upper bound of the treatment result, and the actual desalination rate minus the fluctuation range is used as the lower bound of the treatment result. The plot unit is used as the row, and the confidence level is used as the column. A confidence matrix is ​​constructed based on the rows and columns. The elements of the confidence matrix indicate whether the expected desalination rate is included in the upper and lower bounds of the treatment result at the corresponding confidence level. If it is included, it is marked as 1; if it is not included, it is marked as 0. Based on the element values ​​of the confidence matrix, the percentage of times the plot unit contains the expected desalination rate at different confidence levels is statistically analyzed. The percentage of times is used as the confidence level for evaluating the treatment effect. The absolute value of the difference between the actual desalination rate and the expected desalination rate in the plot unit is used as the treatment effect difference. If the confidence level for evaluating the treatment effect and the treatment effect difference are less than a preset threshold, the application index of the amendment is adjusted according to the optimal index.

[0012] A second aspect of the present invention provides a system for evaluating the effectiveness of saline-alkali land management based on multi-source data, comprising: Data acquisition module: used to acquire data on amendments, monitoring data and treatment standards for coastal saline soil, and to determine treatment indicators for plot units based on the treatment standards according to the type of salinity and the type of water and salt transport through spatial stratified sampling; Data feature analysis module: used to perform time-series registration of the improver data and monitoring data, and decompose trend components, seasonal components and regression components, and use the residual uncertainty of the regression components as fluctuation characteristics; Salt control index extraction module: used to invert the water and salt transport conditions of the plot unit based on the treatment index, fluctuation characteristics, trend components and seasonal components, and extract the boundary salt control index based on the amendment data and the transport depth in the water and salt transport conditions; Amendment optimization index fitting module: used to construct a water-salt transport network of plot units through a directed acyclic graph based on the boundary salt control index, and train a Bayesian network based on the water-salt transport network and amendment data to obtain the optimal index of application amount, application depth and application time of the amendment. The governance effect evaluation module is used to obtain governance results based on the governance indicators and monitoring data, calculate the confidence matrix based on the upper and lower bounds of the optimal indicators and governance results, and obtain the confidence of governance effect evaluation and the difference in governance effect among different plot units in coastal saline soil.

[0013] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects: This invention spatially stratifies and samples monitoring data of saline-alkali land by salt and alkali type and water-salt transport type, selects effective treatment indicators, obtains actually available treatment data, reflects the heterogeneity of different plots, and performs time-series registration of actual applied amendment data and monitoring data through Bayesian variable point time-series decomposition to extract treatment trends and fluctuation characteristics, thereby improving the analytical depth of monitoring data. Combining the obtained plot heterogeneity, a water-salt transport network is constructed through a directed acyclic graph and a Bayesian network is trained to obtain the theoretical optimal application amount, application depth, and application time of the amendment, providing data support for treatment regulation. The reliability of the treatment results of coastal saline soil is evaluated through a confidence matrix, distinguishing the differences and uncertainties in treatment effects of different plot units, enhancing the rational allocation of treatment resources and risk management, and making the evaluation of saline-alkali land treatment effects more refined and dynamic. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating the steps of a method for evaluating the effectiveness of saline-alkali land management based on multi-source data, as described in an embodiment of the present invention. Detailed Implementation

[0015] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0016] Reference Figure 1 As shown, this invention provides a method for evaluating the effectiveness of saline-alkali land management based on multi-source data, including: Data on amendments, monitoring data, and treatment standards for coastal saline soils were obtained. Based on these treatment standards, treatment indicators for plot units were determined through spatial stratified sampling according to the type of salinity and the type of water and salt transport. In the actual assessment, the target area was a 100-mu (approximately 6.7 hectares) coastal saline-alkali land plot, mainly composed of chloride-type coastal saline-alkali soil. Spring evaporation was strong, and water-salt accumulation on the surface was significant. Information on soil amendments applied from March to August was obtained, including the type of amendment: desulfurized gypsum, dosage 2000 kg / mu, and humic acid dosage 500 kg / mu, applied from March 15th to August 15th at 12 application points in a 50m×50m grid. Weekly monitoring data from March to August was obtained, including weekly electrical conductivity (dS / m) of the 0-10cm and 20-30cm soil layers, weekly volumetric water content (%) of the 0-30cm soil layer, and quarterly soil water-salt profile distribution data of the 0-50cm soil layer, specifically stratified electrical conductivity data. The treatment standard for the target area was determined to be DB12 / T The "Technical Specification for Improvement of Coastal Saline-Alkali Land (1158-2023)" stipulates that the standard for the treatment of chloride-type saline soil is "soil electrical conductivity of 0-20cm layer ≤4dS / m". Soil samples from 0-20cm depth were collected at 12 spatial locations, and their ionic composition was analyzed. Chloride ions accounted for 78% of the total salt content. Quarterly conductivity data from March to May showed that the average conductivity of the 0-10cm surface layer was 6.2 dS / m, and the average conductivity of the 20-30cm deep layer was 4.1 dS / m, with the surface layer being higher than the deep layer. The peak conductivity of both the surface and deep layers occurred in mid-April, indicating an evaporative salt accumulation type. Stratified sampling was conducted based on chloride-type salinity and evaporative salt accumulation-type water-salt transport. The target area was divided into 6 evenly distributed plots, each approximately 16.7 acres, numbered S1-S6. According to the treatment standard DB12 / T 1158-2023, the desalination threshold for each plot was determined as: "soil conductivity of 0-20cm layer ≤ 4 dS / m". The improver data and monitoring data are time-series registered, and the trend component, seasonal component and regression component are decomposed. The residual uncertainty of the regression component is used as the fluctuation feature. In the actual assessment, outliers and missing values ​​were processed for the monitoring data from 26 weeks from March to August. Missing values ​​were filled in using linear interpolation of adjacent two-week data. One missing value, after being filled in, was 5.3 dS / m, with 5.1 dS / m the week before and 5.5 dS / m the week after. The surface conductivity of unit S5 in the 0-10cm layer during the third week of May was 10.2 dS / m, far exceeding the average of 6.1 dS / m for other units during the same period. It was found that surface salt crust was mistakenly mixed in during sampling. The outlier was removed using the 3σ criterion and replaced with the average of 5.9 dS / m from the three weeks before and after the same period. The application time of the amendment was used as the time node, and the monitoring data from the week before application and the 25 weeks after application were integrated to make the time series continuous, with a total of 26 time nodes. The monitoring data of different dimensions were converted into values ​​in the [0,1] interval by the maximum and minimum normalization. The surface conductivity of 0-10cm in the third week of March in Unit S1 was 0.43 after standardization. Based on the preprocessed data, soil, water, salt, and amendment characteristics were extracted. A feature vector was generated for each plot unit, comprising 12 feature dimensions. The initial salinity of the soil was determined by the average electrical conductivity (S1-S6) of the 0-20cm soil layer during the first week of the three months prior to amendment application: 5.2, 5.8, 6.3, 7.1, 6.5, and 5.6 dS / m, respectively. Soil texture and bulk density were determined by field collection of loam samples from the 0-30cm depth of each plot, with an average bulk density of 1.35 g / cm³. 3 The fluctuation range of S1-S6 is 1.32-1.38 g / cm³. 3 The conductivity difference of water and salt transport characteristics is the difference between the conductivity of the 0-10cm and 20-30cm soil layers each week, which represents the degree of surface salt accumulation. In unit S1, the difference in the first week of April is 1.2 dS / m, indicating the surface salt accumulation characteristics in spring. The coefficient of variation of moisture content is the ratio of the standard deviation to the mean of moisture content in each month. In unit S6, the coefficient of variation in May is 0.12, indicating small moisture fluctuations. The rate of salinity accumulation is the ratio of the difference in conductivity over two consecutive weeks to the interval. In unit S4, the rate of accumulation in the fourth week of March is 0.07 dS / m·d, showing a salt accumulation trend. In terms of soil amendment application characteristics, the application amount of soil amendment is 2000 kg / mu of desulfurized gypsum and 500 kg / mu of humic acid. Due to the uniform soil texture of the plots and the relatively uniform application points across units, the density of application points is 0.12 per mu. For each plot unit, the time series data were analyzed using the Bayesian variable point time series decomposition formula to obtain the trend component, seasonal component, and residual component. The trend component slope of unit S1 was -0.08 dS / m·week, indicating a continuous desalination trend due to the effect of the soil amendment. The average value of the seasonal component was 0.32 dS / m, indicating surface salt accumulation due to strong evaporation in spring. The average value of the seasonal component in July was -0.25 dS / m, indicating desalination due to leaching from summer rainfall, resulting in a salinity lower than the long-term trend. The residual variance of the regression components was 0.015 ≤ 0.1 dS / m, accounting for less than 5% of the average soil electrical conductivity. The monitoring data showed good stability and no significant random interference. Based on the governance indicators, fluctuation characteristics, trend components and seasonal components, the water and salt transport conditions of the plot unit are inverted, and the boundary salt control index is extracted based on the amendment data and the transport depth in the water and salt transport conditions. In the actual assessment, based on the plot unit, the treatment indicators, fluctuation characteristics, trend components, seasonal components, amendment data, and quarterly soil water and salinity profile distribution data were obtained. The application data of the amendment and the quarterly soil water and salinity profile distribution data were used as process-driven terms, and the treatment indicators were used as target constraints. Among them, the amendment application data of the process-driven terms were an application depth of 10 cm, which is the average surface application depth, and a desulfurization gypsum concentration of 15 kg / m³. 3 Humic acid concentration 3.75 kg / m³ 3 In the quarterly water and salt profile data of unit S1, the quarterly electrical conductivity of the soil layer at depths of 0-10cm is 6.2-4.8, 10-20cm is 5.1-4.0, and 20-30cm is 4.1-3.7. The target constraint is a soil layer electrical conductivity of ≤4 dS / m at depths of 0-20cm. Based on the process-driven and target constraints, the basic state of water and salt balance for the plot unit is calculated using the soil water and salt balance equation. The input is atmospheric deposition salt, approximately 0.05 kg / m³ per quarter. 2 The output is the amount of salt leached by rainfall. In Q1, with less rainfall, the leaching amount is 0.02 kg / m³. 2 Q2 saw heavy rainfall, with leaching at a rate of 0.1 kg / m³. 2 The salt content change due to evaporation and leaching is Q1, with an accumulation of 0.1 kg / m³. 2 Q2 desalination 0.08kg / m 2 Therefore, the net salt accumulation in Q1 is 0.13 kg / m³. 2 Q2 net desalination -0.13kg / m 2 Combining the seasonal components of unit S1 with the salt accumulation characteristic of Q1 (+0.32 dS / m) and the desalination characteristic of Q2 (-0.25 dS / m) to correct the water-salt balance, Q1 is corrected to 0.1716 kg / m. 2， Salt deposition is enhanced, Q2 is corrected to -0.0975 kg / m³2 Desalination is weakened. Using the initial salt concentration distribution of Q1 as the initial condition, a CDE convection dispersion equation is constructed: ;in Salt concentration, The salt diffusion coefficient is... This represents the rate of water-salt transport. Given the soil depth and the initial salt concentration distribution, the electrical conductivity-to-salt concentration is calculated as EC(dS / m) = 1.6 × salt concentration (g / L) for chloride-type saline soil. The soil saturated hydraulic conductivity K is then iteratively retrieved using the SCE-UA optimization algorithm. s and salt diffusion coefficient The soil saturated hydraulic conductivity was obtained from the salt concentration and soil moisture content data from monitoring data, until the root mean square error between the simulated value and the actual water-salt profile of Q2 was <0.3 dS / m. The final water-salt transport conditions were a transport velocity of 0.8 cm / d and a transport depth D. max =45cm, the boundary salt control index was calculated according to the boundary salt control index formula, where the concentration of the amendment at an application depth of 10cm was 18.75kg / m³ of a mixture of desulfurized gypsum and humic acid. 3 The soil conditioner was applied at a depth of 20cm. At a depth of 10cm, the soil hydraulic conductivity was 1.2cm / d, and at 45cm, it was 0.5cm / d, decreasing linearly with depth. The moisture content at a depth of 0-20cm decreased from 18.5% to 17.2%, with a gradient of -0.0013cm. 3 / cm 3 ·cm, the boundary salt control index of unit S1 was obtained as 0.85; Based on the boundary salt control index, a water and salt transport network for plot units is constructed using a directed acyclic graph. A Bayesian network is trained based on the water and salt transport network and amendment data to obtain the optimal indexes for the amount, depth, and time of application of the amendment. In the actual assessment, the application sites of the amendment and the water-salt migration boundary were used as network nodes. There were 12 amendment application sites P1-P12, with two sites corresponding to each plot unit. Unit S1 corresponded to P1 and P2. The upper limit of the water-salt migration depth for each plot unit was used as the boundary node. Unit S1 had a migration depth of 45cm, corresponding to boundary node B1. The direction of evaporative salt accumulation water-salt migration was from deep to surface, therefore the edges pointed from the water-salt migration boundary node to the application site node within the unit: B1→P1, B1→P2. Using the boundary salt control index of the corresponding plot unit as the edge weight, the edge weight of unit S1 was 0.85. A directed acyclic graph with 18 nodes and 12 edges, without closed-loop structures, was constructed to obtain the 3D features of each node. Specifically, the data includes the boundary salt control index, soil amendment concentration, and current application rate. The change in soil conductivity in the 0-10cm soil layer per unit week after soil amendment application is used as a label, with negative values ​​indicating desalination. Training set data is obtained. In the training set data, the values ​​for nodes P1 and P2 after 4 weeks of application are: node P1: boundary salt control index 0.85, soil amendment concentration 18.75, current application rate 2500, weekly soil conductivity change -0.4; node P2: boundary salt control index 0.85, soil amendment concentration 18.75, current application rate 2500, weekly soil conductivity change -0.4; node P3: boundary salt control index 0.81, soil amendment concentration 18.75, current application rate 2500, weekly soil conductivity change -0.5. The training set data dimension was used as the input layer data dimension of the Bayesian network. The network was trained using the training set and labels, with a 54-dimensional input layer, two hidden layers (10 neurons each), and a 1-dimensional output layer. The training objective was to minimize the deviation between the predicted and actual changes in soil conductivity. After 50 iterations, the deviation probability distribution showed that the deviation between the predicted and actual values ​​was concentrated within ±0.05 dS / m, indicating network convergence. At this point, the root mean square error was 0.04 dS / m. Based on the deviation probability distribution, the optimal index was solved by optimizing the objective function. The feasible region is [missing information - likely a region of interest]. The application rate of desulfurized gypsum was 1500-2500 kg / mu, the application rate of humic acid was 300-700 kg / mu, the application depth was 5-20 cm, the application time was from late February to early April, the regularization coefficient was 0.01, the target probability distribution was less than or equal to -0.6 dS / m, and the optimal application indexes for each unit were obtained as follows: the initial value of desulfurized gypsum application rate was 2000 kg / mu, the optimal value was 2200 kg / mu, the initial value of humic acid application rate was 500 kg / mu, the optimal value was 450 kg / mu, the initial value of application depth was 10 cm, the optimal value was 12 cm, and the initial value of application time was March 15, the optimal value was March 5. Based on the governance indicators and monitoring data, the governance results are obtained. Based on the upper and lower bounds of the optimal indicators and governance results, the confidence matrix is ​​calculated to obtain the confidence level of the governance effect assessment and the difference in governance effect among different plot units in coastal saline soil.

[0017] In the actual assessment, treatment indicators and monitoring data were obtained based on plot units. The actual desalination rate was calculated according to the desalination threshold in the treatment indicators and the soil conductivity in the monitoring data. The actual desalination rate was taken as the treatment result, where the actual desalination rate = (initial conductivity - current conductivity) / initial conductivity. The conductivity of the 0-20cm soil layer was the core indicator. Combining the initial data before application and the Q3 monitoring data, the actual desalination rate of each plot unit was calculated as follows: the initial conductivity of plot unit S1 was 5.2, the conductivity of Q3 was 3.8, and the actual desalination rate was... The salinity is 26.9%. The initial electrical conductivity of plot S2 is 5.8, Q3 is 4.2, and the actual desalination rate is 27.6%. The initial electrical conductivity of plot S4 is 7.1, Q3 is 5.2, and the actual desalination rate is 26.8%. The initial electrical conductivity of S6 is 5.6, Q3 is 53.9, and the actual desalination rate is 30.4%. Based on the optimal index, the expected deviation probability distribution of each plot unit is output through the Bayesian network, and its mean is taken as the expected desalination rate: mean of the expected deviation probability distribution of plot unit S1 - 1.5 represents the cumulative desalination range from the initial value to Q3, with an expected desalination rate of 28.8%. The mean of the expected deviation probability distribution for plot S2 is -1.7, with an expected desalination rate of 29.3%. The mean of the expected deviation probability distribution for plot S4 is -2.0 to -1.7, with an expected desalination rate of 28.2%. Based on the aforementioned fluctuation characteristics, the fluctuation range of the actual desalination rate is calculated. The actual desalination rate plus the fluctuation range is used as the upper bound of the treatment result, and the actual desalination rate minus the fluctuation range is used as the lower bound of the treatment result. The residual variance of each plot unit is... The mean is 0.01, corresponding to a desalination rate fluctuation range of ±5% of the actual desalination rate. The fluctuation range for S1 is ±1.35, with an upper limit of 28.25 and a lower limit of 25.55. The fluctuation range for S2 is ±1.38, with an upper limit of 28.98 and a lower limit of 26.22. The fluctuation range for S4 is ±1.35, with an upper limit of 28.35 and a lower limit of 25.65. The fluctuation range for S4 is ±1.34, with an upper limit of 28.14 and a lower limit of 25.46. The fluctuation range for S6 is ±1.52, with an upper limit of 31.92 and a lower limit of 28.88. Using land parcel units as rows and confidence levels (90% / 95% / 99%) as columns, the matrix elements are "1 (expected desalination rate within the upper and lower bounds) / 0 (not within)", a confidence matrix is ​​obtained. S1 is zero at 90% / 95% / 99% confidence levels, S2 is zero at 99% confidence level, and S3 is zero at 95% / 99% confidence levels. Based on the element values ​​of the confidence matrix, the percentage of times the upper and lower bounds of the land parcel unit contain the expected desalination rate at different confidence levels is calculated. This percentage is used as the confidence level for assessing the treatment effect. The absolute value of the difference between the actual desalination rate and the expected desalination rate in the land parcel unit is used as the treatment... Regarding the difference in treatment effectiveness, if the confidence level of the effectiveness assessment and the difference in effectiveness are less than a preset threshold, the application indicators of the plant amendment will be adjusted according to the optimal indicators. The preset threshold for the confidence level of the effectiveness assessment is 60%, and the preset threshold for the difference in effectiveness is 2%. Therefore, in case S1, the confidence level for the effectiveness is 0, and the difference in effectiveness is 1.9%, requiring adjustment of the plant amendment application indicators based on the optimal indicators. In case S2, the confidence level for the effectiveness is 66.7%, and the difference in effectiveness is 1.7%, maintaining the existing application indicators. In cases S3 and S4, the confidence level for the effectiveness is 31%, requiring adjustment of the application indicators. In case S5, the difference in effectiveness is 0, and no adjustment of the plant amendment application indicators is required.

[0018] In this embodiment, the method for obtaining the governance indicators includes: Coastal saline soil samples were collected from various spatial locations in the target area. The soil texture of the coastal saline soil samples included sandy, loamy, and clayey soils. Based on the samples, data on soil amendments and monitoring data during the saline-alkali treatment process in the past two quarters were obtained. Based on the monitoring data, the saline-alkali type of each spatial location was determined, and the treatment standards for saline soil were obtained based on the saline-alkali type. Based on the monitoring data, the water and salt transport type is determined by quarterly soil electrical conductivity distribution data and quarterly soil water and salt profile distribution data. If the peak electrical conductivity coincides with the peak time of the water and salt profile distribution and the electrical conductivity of the 0-10cm soil surface layer is higher than that of the 20-30cm deep layer, it is determined to be the evaporative salt accumulation type. If the quarterly electrical conductivity difference is greater than or equal to 30% and the deep layer electrical conductivity is higher than that of the surface layer, it is determined to be the leaching desalination type. If the electrical conductivity increases linearly with the water and salt profile distribution, it is determined to be the capillary ascent type. Based on the target area, fixed-size plots are divided by spatial stratification sampling according to the type of salinity and the type of water-salt transport. The desalination threshold of the plots is determined by the treatment standards to obtain the treatment index of the plots. The sampling combination consists of more than or equal to 6 plots and is evenly distributed in the target area. The salinity types include chloride type, sulfate type, soda type and mixed type.

[0019] In this embodiment, the method for obtaining the fluctuation characteristics includes: Based on land parcel units, amendment data and monitoring data are obtained. The amendment data includes amendment type, application rate, and application time. The amendment data and monitoring data are aligned to a weekly time scale according to timestamps. Data with non-overlapping timestamps are linearly interpolated to obtain time-series registration data. Bayesian change-point time-series decomposition is performed on the time-series registration data to obtain trend components, seasonal components, and regression components. The variance of the regression component residuals is used as the fluctuation characteristic. The Bayesian change-point time-series decomposition formula is as follows: ; in land parcel unit exist Fitted values ​​of monitoring data at time points. land parcel unit The The weights of each trend term were obtained by fitting the registered data according to the AIC information criterion. land parcel unit The The rate of change of each trend item land parcel unit The The inflection point time of each trend item land parcel unit Time window weight, land parcel unit The governance time window is determined according to the uniform weighting rule. The total number of trend items. For logistic functions, For the first The amplitude of the sinusoidal term of the seasonal fluctuation. For the first The angular frequency of seasonal fluctuations This represents the total number of seasonal items, expressed in weeks. for The transpose of the explanatory variable matrix at time points, where the explanatory variables are obtained from improver data and monitoring data. The regression coefficient vector is obtained through least squares regression fitting. for The return residual of time, by and The monitoring data was obtained at any given time.

[0020] In this embodiment, the method for obtaining the water-salt transport conditions includes: Based on the data obtained from the plot unit, including treatment indicators, fluctuation characteristics, trend components, seasonal components, amendment data, and quarterly soil water and salt profile distribution data, the application data of amendments and the quarterly soil water and salt profile distribution data are used as process driving terms, and the treatment indicators are used as target constraints. Based on the process driving terms and target constraints, the basic state of water and salt balance of the plot unit is calculated through the soil water and salt balance equation. The basic state is corrected by the water and salt fluctuation characteristics of the seasonal components within a unit period, and the balance result of salt input and salt output within a unit period is obtained. Based on the quarterly soil water and salt profile distribution data, the initial salt concentration distribution is used as the initial condition. Based on the equilibrium result and the initial condition, a salt transport simulation model is constructed using the convection diffusion equation. The actual salt data of the quarterly soil water and salt profile distribution data is used as a reference. Based on the reference, the soil saturated hydraulic conductivity and salt diffusion coefficient in the simulation model are iteratively inverted using the SCE-UA optimization algorithm until the root mean square error between the iterative simulation value and the reference is less than 0.3 dS / m, at which point the inversion is terminated. The transport rate and transport depth are obtained and used as the water and salt transport conditions for the plot unit.

[0021] In this embodiment, the method for obtaining the boundary salt control index includes: The boundary salinity control index is calculated based on the amendment data and the migration depth in the water-salt migration conditions. The calculation formula for the boundary salinity control index is as follows: ; in land parcel unit Boundary salt control indicators, The correction coefficient for the effect of the modifier is determined based on the slope of the trend component. land parcel unit At application depth The concentration of the modifier at the location, This refers to the application range of the improver. land parcel unit In depth Soil hydraulic conductivity at that location, land parcel unit In depth The soil moisture gradient is obtained based on the soil moisture content data from monitoring.

[0022] In this embodiment, the method for obtaining the optimal index includes: Based on the plot unit, the water and salt transport boundary and the application point of the amendment are used as network nodes, the direction of water and salt transport is used as the direction of the edge between the network nodes, and the boundary salt control index is used as the edge weight. A directed acyclic graph is constructed according to the network nodes, edge weights and directions. The monitoring data of the topology and the amendment data are obtained based on the directed acyclic graph. The change in soil electrical conductivity per unit period after the application of amendment is used as the label. Training set data is generated based on the monitoring data of the topology and the amendment data. The training set data dimension is used as the input layer data dimension of the Bayesian network. The Bayesian network is trained using the training set and labels. The nonlinear relationship between the input variables and labels is fitted based on the hidden layer until the deviation between the predicted salt control effect and the actual salt control effect is minimized. The deviation probability distribution of the salt control effect and the trained Bayesian network are obtained. The optimal application index of the amendment is obtained by optimizing the objective function based on the deviation probability distribution. The application index includes the application amount, application depth, and application time. The optimization objective function formula is as follows: ; in This is a set of parameters for the application index of the amendment. For the feasible domain of the parameter, for Divergence, representing the probability distribution of the bias. With the target probability distribution Differences For input variables, land parcel unit Probability distribution of changes in target soil electrical conductivity under treatment indicators The regularization coefficient is . land parcel unit The weights of the constraint terms are obtained based on the characteristic contributions in the water-salt transport conditions. For the trace operation of a matrix, is the inverse of the covariance matrix of the parameter set.

[0023] In this embodiment, the method for obtaining the confidence level of the governance effect assessment and the difference in governance effect includes: Based on the land parcel unit, the treatment indicators and monitoring data are obtained. The actual desalination rate is calculated according to the desalination threshold in the treatment indicators and the soil electrical conductivity in the monitoring data. The actual desalination rate is used as the treatment result. Based on the optimal indicators, the expected deviation probability distribution of each land parcel unit is output through the Bayesian network. The mean of the expected deviation probability distribution is used as the expected desalination rate. The fluctuation range of the actual desalination rate is calculated based on the fluctuation characteristics. The actual desalination rate plus the fluctuation range is used as the upper bound of the treatment result, and the actual desalination rate minus the fluctuation range is used as the lower bound of the treatment result. The plot unit is used as the row, and the confidence level is used as the column. A confidence matrix is ​​constructed based on the rows and columns. The elements of the confidence matrix indicate whether the expected desalination rate is included in the upper and lower bounds of the treatment result at the corresponding confidence level. If it is included, it is marked as 1; if it is not included, it is marked as 0. Based on the element values ​​of the confidence matrix, the percentage of times the plot unit contains the expected desalination rate at different confidence levels is statistically analyzed. The percentage of times is used as the confidence level for evaluating the treatment effect. The absolute value of the difference between the actual desalination rate and the expected desalination rate in the plot unit is used as the treatment effect difference. If the confidence level for evaluating the treatment effect and the treatment effect difference are less than a preset threshold, the application index of the amendment is adjusted according to the optimal index.

[0024] A second aspect of the present invention also provides a system for evaluating the effectiveness of saline-alkali land management based on multi-source data, comprising: Data acquisition module: used to acquire data on amendments, monitoring data and treatment standards for coastal saline soil, and to determine treatment indicators for plot units based on the treatment standards according to the type of salinity and the type of water and salt transport through spatial stratified sampling; Data feature analysis module: used to perform time-series registration of the improver data and monitoring data, and decompose trend components, seasonal components and regression components, and use the residual uncertainty of the regression components as fluctuation characteristics; Salt control index extraction module: used to invert the water and salt transport conditions of the plot unit based on the treatment index, fluctuation characteristics, trend components and seasonal components, and extract the boundary salt control index based on the amendment data and the transport depth in the water and salt transport conditions; Amendment optimization index fitting module: used to construct a water-salt transport network of plot units through a directed acyclic graph based on the boundary salt control index, and train a Bayesian network based on the water-salt transport network and amendment data to obtain the optimal index of application amount, application depth and application time of the amendment. The governance effect evaluation module is used to obtain governance results based on the governance indicators and monitoring data, calculate the confidence matrix based on the upper and lower bounds of the optimal indicators and governance results, and obtain the confidence of governance effect evaluation and the difference in governance effect among different plot units in coastal saline soil.

[0025] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined in the claims, all of which should fall within the protection scope of the present invention.

Claims

1. A method for evaluating the effectiveness of saline-alkali land management based on multi-source data, characterized in that, Includes the following steps: Data on amendments, monitoring data, and treatment standards for coastal saline soils were obtained. Based on these treatment standards, treatment indicators for plot units were determined through spatial stratified sampling according to the type of salinity and the type of water and salt transport. The improver data and monitoring data are time-series registered, and the trend component, seasonal component and regression component are decomposed. The residual uncertainty of the regression component is used as the fluctuation feature. Based on the governance indicators, fluctuation characteristics, trend components and seasonal components, the water and salt transport conditions of the plot unit are inverted, and the boundary salt control index is extracted based on the amendment data and the transport depth in the water and salt transport conditions. Based on the boundary salt control index, a water and salt transport network for plot units is constructed using a directed acyclic graph. A Bayesian network is trained based on the water and salt transport network and amendment data to obtain the optimal indexes for the amount, depth, and time of application of the amendment. Based on the governance indicators and monitoring data, the governance results are obtained. Based on the upper and lower bounds of the optimal indicators and governance results, the confidence matrix is ​​calculated to obtain the confidence level of the governance effect assessment and the difference in governance effect among different plot units in coastal saline soil.

2. The method for evaluating the effectiveness of saline-alkali land management based on multi-source data according to claim 1, characterized in that, The method for obtaining the governance indicators includes: Coastal saline soil samples were collected from various spatial locations in the target area. The soil texture of the coastal saline soil samples included sandy, loamy, and clayey soils. Based on the samples, data on soil amendments and monitoring data during the saline-alkali treatment process in the past two quarters were obtained. Based on the monitoring data, the saline-alkali type of each spatial location was determined, and the treatment standards for saline soil were obtained based on the saline-alkali type. Based on the monitoring data, the water and salt transport type is determined by quarterly soil electrical conductivity distribution data and quarterly soil water and salt profile distribution data. If the peak electrical conductivity coincides with the peak time of the water and salt profile distribution and the electrical conductivity of the 0-10cm soil surface layer is higher than that of the 20-30cm deep layer, it is determined to be the evaporative salt accumulation type. If the quarterly electrical conductivity difference is greater than or equal to 30% and the deep layer electrical conductivity is higher than that of the surface layer, it is determined to be the leaching desalination type. If the electrical conductivity increases linearly with the water and salt profile distribution, it is determined to be the capillary ascent type. Based on the target area, fixed-size plots are divided by spatial stratification sampling according to the type of salinity and the type of water-salt transport. The desalination threshold of the plots is determined by the treatment standards to obtain the treatment index of the plots. The sampling combination consists of more than or equal to 6 plots and is evenly distributed in the target area. The salinity types include chloride type, sulfate type, soda type and mixed type.

3. The method for evaluating the effectiveness of saline-alkali land management based on multi-source data according to claim 1, characterized in that, The method for obtaining the wave characteristics includes: Based on land parcel units, amendment data and monitoring data are obtained. The amendment data includes amendment type, application rate, and application time. The amendment data and monitoring data are aligned to a weekly time scale according to timestamps. Data with non-overlapping timestamps are linearly interpolated to obtain time-series registration data. Bayesian change-point time-series decomposition is performed on the time-series registration data to obtain trend components, seasonal components, and regression components. The variance of the regression component residuals is used as the fluctuation characteristic. The Bayesian change-point time-series decomposition formula is as follows: ； in land parcel unit exist Fitted values ​​of monitoring data at time points. land parcel unit The The weights of each trend term were obtained by fitting the registered data according to the AIC information criterion. land parcel unit The The rate of change of each trend item land parcel unit The The inflection point time of each trend item land parcel unit Time window weight, land parcel unit The governance time window is determined according to the uniform weighting rule. The total number of trend items. For logistic functions, For the first The amplitude of the sinusoidal term of the seasonal fluctuation. For the first The angular frequency of seasonal fluctuations This represents the total number of seasonal items, expressed in weeks. for The transpose of the explanatory variable matrix at time points, where the explanatory variables are obtained from improver data and monitoring data. The regression coefficient vector is obtained through least squares regression fitting. for The return residual of time, by and The monitoring data was obtained at any given time.

4. The method for evaluating the effectiveness of saline-alkali land management based on multi-source data according to claim 1, characterized in that, A method for obtaining the water-salt transport conditions includes: Based on the data obtained from the plot unit, including treatment indicators, fluctuation characteristics, trend components, seasonal components, amendment data, and quarterly soil water and salt profile distribution data, the application data of amendments and the quarterly soil water and salt profile distribution data are used as process driving terms, and the treatment indicators are used as target constraints. Based on the process driving terms and target constraints, the basic state of water and salt balance of the plot unit is calculated through the soil water and salt balance equation. The basic state is corrected by the water and salt fluctuation characteristics of the seasonal components within a unit period, and the balance result of salt input and salt output within a unit period is obtained. Based on the quarterly soil water and salt profile distribution data, the initial salt concentration distribution is used as the initial condition. Based on the equilibrium result and the initial condition, a salt transport simulation model is constructed using the convection diffusion equation. The actual salt data of the quarterly soil water and salt profile distribution data is used as a reference. Based on the reference, the soil saturated hydraulic conductivity and salt diffusion coefficient in the simulation model are iteratively inverted using the SCE-UA optimization algorithm until the root mean square error between the iterative simulation value and the reference is less than 0.3 dS / m, at which point the inversion is terminated. The transport rate and transport depth are obtained and used as the water and salt transport conditions for the plot unit.

5. The method for evaluating the effectiveness of saline-alkali land management based on multi-source data according to claim 1, characterized in that, The method for obtaining the boundary salt control index includes: The boundary salinity control index is calculated based on the amendment data and the migration depth in the water-salt migration conditions. The calculation formula for the boundary salinity control index is as follows: ； in land parcel unit Boundary salt control indicators, The correction coefficient for the effect of the modifier is determined based on the slope of the trend component. land parcel unit At application depth The concentration of the modifier at that location, The application range of the improver. land parcel unit In depth Soil hydraulic conductivity at that location, land parcel unit In depth The soil moisture gradient is obtained based on the soil moisture content data from monitoring.

6. The method for evaluating the effectiveness of saline-alkali land management based on multi-source data according to claim 1, characterized in that, The method for obtaining the optimal index includes: Based on the plot unit, the water and salt transport boundary and the application point of the amendment are used as network nodes, the direction of water and salt transport is used as the direction of the edge between the network nodes, and the boundary salt control index is used as the edge weight. A directed acyclic graph is constructed according to the network nodes, edge weights and directions. The monitoring data of the topology and the amendment data are obtained based on the directed acyclic graph. The change in soil electrical conductivity per unit period after the application of amendment is used as the label. Training set data is generated based on the monitoring data of the topology and the amendment data. The training set data dimension is used as the input layer data dimension of the Bayesian network. The Bayesian network is trained using the training set and labels. The nonlinear relationship between the input variables and labels is fitted based on the hidden layer until the deviation between the predicted salt control effect and the actual salt control effect is minimized. The deviation probability distribution of the salt control effect and the trained Bayesian network are obtained. The optimal application index of the amendment is obtained by optimizing the objective function based on the deviation probability distribution. The application index includes the application amount, application depth, and application time. The optimization objective function formula is as follows: ； in This is a set of parameters for the application index of the amendment. For the feasible domain of the parameter, for Divergence, representing the probability distribution of the bias. With the target probability distribution differences For input variables, land parcel unit Probability distribution of changes in target soil electrical conductivity under treatment indicators The regularization coefficient is . land parcel unit The weights of the constraint terms are obtained based on the characteristic contributions in the water-salt transport conditions. For the trace operation of a matrix, is the inverse of the covariance matrix of the parameter set.

7. The method for evaluating the effectiveness of saline-alkali land management based on multi-source data according to claim 1, characterized in that, The method for obtaining the confidence level of the governance effect assessment and the difference in governance effects includes: Based on the land parcel unit, the treatment indicators and monitoring data are obtained. The actual desalination rate is calculated according to the desalination threshold in the treatment indicators and the soil electrical conductivity in the monitoring data. The actual desalination rate is used as the treatment result. Based on the optimal indicators, the expected deviation probability distribution of each land parcel unit is output through the Bayesian network. The mean of the expected deviation probability distribution is used as the expected desalination rate. The fluctuation range of the actual desalination rate is calculated based on the fluctuation characteristics. The actual desalination rate plus the fluctuation range is used as the upper bound of the treatment result, and the actual desalination rate minus the fluctuation range is used as the lower bound of the treatment result. The plot unit is used as the row, and the confidence level is used as the column. A confidence matrix is ​​constructed based on the rows and columns. The elements of the confidence matrix indicate whether the expected desalination rate is included in the upper and lower bounds of the treatment result at the corresponding confidence level. If it is included, it is marked as 1; if it is not included, it is marked as 0. Based on the element values ​​of the confidence matrix, the percentage of times the plot unit contains the expected desalination rate at different confidence levels is statistically analyzed. The percentage of times is used as the confidence level for evaluating the treatment effect. The absolute value of the difference between the actual desalination rate and the expected desalination rate in the plot unit is used as the treatment effect difference. If the confidence level for evaluating the treatment effect and the treatment effect difference are less than a preset threshold, the application index of the amendment is adjusted according to the optimal index.

8. A system for evaluating the effectiveness of saline-alkali land management based on multi-source data, used to execute the method for evaluating the effectiveness of saline-alkali land management based on multi-source data as described in any one of claims 1 to 7, characterized in that, The system includes: Data acquisition module: used to acquire data on amendments, monitoring data and treatment standards for coastal saline soil, and to determine treatment indicators for plot units based on the treatment standards according to the type of salinity and the type of water and salt transport through spatial stratified sampling; Data feature analysis module: used to perform time-series registration of the improver data and monitoring data, and decompose trend components, seasonal components and regression components, and use the residual uncertainty of the regression components as fluctuation characteristics; Salt control index extraction module: used to invert the water and salt transport conditions of the plot unit based on the treatment index, fluctuation characteristics, trend components and seasonal components, and extract the boundary salt control index based on the amendment data and the transport depth in the water and salt transport conditions; Amendment optimization index fitting module: used to construct a water-salt transport network of plot units through a directed acyclic graph based on the boundary salt control index, and train a Bayesian network based on the water-salt transport network and amendment data to obtain the optimal index of application amount, application depth and application time of the amendment. The governance effect evaluation module is used to obtain governance results based on the governance indicators and monitoring data, calculate the confidence matrix based on the upper and lower bounds of the optimal indicators and governance results, and obtain the confidence of governance effect evaluation and the difference in governance effect among different plot units in coastal saline soil.

Images