A farmland soil risk rapid screening method and system based on gamma spectrum

Abstract

The application discloses a farmland soil risk rapid screening method and system based on gamma spectrum, and belongs to the field of agricultural information technology and environmental monitoring technology. The method comprises the following steps: obtaining boundary information of a target farmland area, and dividing the farmland area into detection units; obtaining gamma spectrum data at the detection point positions corresponding to the detection units and performing pretreatment; constructing a soil risk characteristic parameter system comprising a nuclide concentration abnormality characteristic parameter, a nuclide ratio characteristic parameter, a comprehensive intensity characteristic parameter of a spectrum and a spectrum structure change characteristic parameter; based on the soil risk characteristic parameters, a soil risk rapid screening model is constructed to distinguish the farmland soil risk, and the farmland soil risk is classified according to the model output result. The method realizes rapid, non-destructive and large-scale screening of the farmland soil risk, reduces the detection cost, improves the screening efficiency, and provides technical support for farmland risk management and fine agricultural decision-making.

Description

Technical Field

[0001] This invention relates to the fields of agricultural information technology and environmental monitoring technology, and in particular to a method and system for rapid, non-contact screening and classification assessment of potential risks in farmland soil using gamma spectroscopy detection technology, specifically a rapid screening method and system for farmland soil risks based on gamma spectroscopy. Background Technology

[0002] With the increasing intensification of agriculture, farmland soils face multiple risks, including heavy metal accumulation, soil degradation, nutrient imbalance, and localized radioactive anomalies. Currently, farmland soil risk assessment mainly relies on laboratory chemical analysis methods, such as atomic absorption spectrometry and ICP-MS. These methods have the following shortcomings: long detection cycles, making it difficult to meet the needs of large-scale, rapid screening; high sampling costs and strong destructiveness, which are not conducive to long-term continuous monitoring; discrete sampling points, making it difficult to reflect the continuous spatial distribution characteristics of soil; and most methods target a single indicator, lacking the ability to quickly assess comprehensive risks.

[0003] Gamma spectroscopy can detect naturally occurring radionuclides in soil (such as...) 40 K, 238 U series 232 The gamma rays emitted by the Th series indirectly reflect soil composition, structure, and some pollution characteristics, offering advantages such as non-contact, rapid detection, and suitability for field operations. However, existing technologies are mostly focused on geological surveys or single soil property inversion, lacking a rapid screening method and indicator system for farmland soil risks.

[0004] Therefore, there is an urgent need to propose a rapid screening method and system for farmland soil risks based on gamma spectral data, so as to achieve efficient identification and classification of farmland soil risks over a large area. Summary of the Invention

[0005] The purpose of this invention is to provide a rapid screening method and system for farmland soil risk based on gamma spectroscopy. By processing and analyzing the characteristic parameters of farmland soil gamma spectroscopy, a soil risk screening model is constructed to achieve rapid identification and classification of potential risks in farmland soil, thereby overcoming the problems of long detection cycles, high costs and difficulty in large-scale application of existing technologies.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] On the one hand, this invention provides a rapid risk screening method for farmland soil based on gamma spectroscopy, comprising the following steps:

[0008] S1. Obtain the boundary information of the target farmland area, and divide the farmland area into detection units according to the farmland scale, farming method and plot shape. The detection unit corresponds to the gamma spectrum detection point, and the location of the detection point is used to represent the soil condition of the corresponding detection unit.

[0009] S2. At each detection point, a gamma spectrometer is used to conduct in-situ measurements of farmland soil to obtain gamma spectrum data;

[0010] S3. Preprocess the raw gamma spectrum data obtained in step S2. The preprocessing includes energy scale correction, background and environmental interference correction, time normalization, and noise suppression and smoothing.

[0011] S4. Based on the preprocessed gamma spectral data, a feature parameter system for rapid soil risk screening is constructed. The feature parameters include nuclide concentration anomaly feature parameters, nuclide ratio feature parameters, comprehensive energy spectrum intensity feature parameters, and energy spectrum structure change feature parameters.

[0012] S5. Based on the soil risk characteristic parameters, construct a rapid soil risk screening model, wherein the screening model includes one or more of the following: a rapid discrimination model based on threshold rules, a comprehensive risk index model based on weighted scoring, and a model based on machine learning algorithms.

[0013] S6. Based on the model output results, classify the risk levels of farmland soil into three categories: low risk, medium risk, and high risk.

[0014] Furthermore, the sources of the boundary information include geographic information systems, remote sensing images, and farmland management systems, and the division methods of the detection units include one or more of the following: regular grid division method, division method based on actual field boundaries, and adaptive division method combining crop type, irrigation conditions, and management units.

[0015] Furthermore, the gamma-ray spectral data includes 40 Counting of energy spectrum peaks of K nuclide 238 Spectral peak counts of U-series nuclides 232 The energy spectrum peak counts of Th series nuclides, the total energy spectrum count rate per unit time, and the count distribution information in different energy ranges are obtained. The gamma energy spectrum measurement process is set according to farmland conditions, including measurement height, single-point measurement time, and number of repeated measurements.

[0016] Furthermore, in S3, the energy calibration calibrates the energy spectrum channel with the actual energy, eliminating the influence of instrument drift on the data;

[0017] The background and environmental interference correction can eliminate the interference of environmental background radiation, cosmic rays and surrounding facilities on the measurement results;

[0018] The time normalization process converts the energy spectrum counts at different measurement times into a uniform count rate per unit time.

[0019] The noise suppression and smoothing process applies a smoothing filter to the energy spectrum curve to improve the stability of feature extraction.

[0020] Furthermore, the nuclide concentration anomaly characteristic parameter in S4 compares the nuclide concentration value at the detection point with the regional background value and historical reference value to calculate the nuclide concentration anomaly index, which is used to characterize the degree of anomaly of radionuclides in the soil, including soil parent material anomaly, potential heavy metal associated anomaly, and anomaly changes caused by human disturbance.

[0021] The nuclide ratio characteristic parameters are constructed based on the relative relationships between different nuclides, including: 238 U / 232 Th ratio 40 K / 232 Th ratio and 40 K / 238 U ratio;

[0022] The comprehensive energy spectrum intensity characteristic parameter is constructed based on the total energy spectrum count rate and the cumulative count value of a specific energy band per unit time.

[0023] The energy spectrum structure change characteristic parameters are constructed by analyzing the count ratio distribution and morphological changes of the energy spectrum curve in different energy ranges.

[0024] Furthermore, the rapid discrimination model based on threshold rules compares various soil risk characteristic parameters with preset thresholds. When a parameter exceeds the corresponding threshold, it determines that the corresponding detection unit has soil risk. The threshold is determined based on regional statistical background, historical monitoring data, and empirical rules.

[0025] Furthermore, the comprehensive risk index model based on weighted scoring is constructed based on multidimensional soil risk characteristic parameters, and its calculation method is as follows:

[0026] ,in, This is the comprehensive soil risk score. For the first Soil risk characteristic parameters, These are the corresponding weighting coefficients.

[0027] Furthermore, the soil risk model based on machine learning algorithms utilizes historical farmland monitoring data to construct a machine learning model. The model input is soil risk characteristic parameters, and the output is either a soil risk category or a risk probability. The machine learning algorithm includes one of support vector machines, decision trees, random forests, or neural network algorithms.

[0028] On the other hand, the present invention also provides a rapid screening system for farmland soil risk based on gamma spectrum, including a farmland spatial information acquisition module, a detection unit division module, a gamma spectrum data acquisition module, a spectrum data preprocessing module, a soil risk characteristic parameter construction module, a rapid screening model module for soil risk, and a risk level determination and result output module.

[0029] The farmland spatial information acquisition module includes GIS vector boundary data, farmland outlines generated from remote sensing images, and plot numbers and boundary information from the agricultural plot management system.

[0030] The division criteria for the detection unit division module include farmland area size, plot shape regularity, and farming method.

[0031] Furthermore, the soil risk characteristic parameter construction module includes sub-modules for extracting anomaly characteristic parameters of nuclide concentration, nuclide ratio characteristic parameters, comprehensive intensity characteristic parameters of energy spectrum, and characteristic parameters of energy spectrum structure change.

[0032] Furthermore, the rapid soil risk screening model module includes a threshold rule discrimination sub-model, a comprehensive risk index sub-model, and a machine learning prediction sub-model, and is equipped with a model fusion unit for fusing and judging the output results of multiple sub-models. The fusion judgment combines the output results of the threshold model, the weighted scoring model, and the machine learning model. When the judgment results of different models are inconsistent, the result with the higher risk level is adopted as the final judgment result.

[0033] The beneficial effects of this invention are:

[0034] This invention sets up gamma-ray spectrometry detection points within farmland detection units and uses a gamma-ray spectrometer to perform in-situ measurements of farmland soil. This eliminates the need to collect large amounts of soil samples and send them to a laboratory for analysis, avoiding the sampling, transportation, and experimental analysis steps present in traditional detection methods. This significantly shortens the soil risk assessment cycle. By performing energy calibration, background and environmental interference correction, time normalization, and noise suppression and smoothing on the gamma-ray spectrometry data, the data collected on-site can be directly used for subsequent risk analysis, enabling rapid screening and real-time determination of farmland soil risks.

[0035] Using gamma spectral data as the core input information, it is possible to cover a large area of ​​farmland in a single field operation, avoiding high-frequency, multi-point laboratory chemical analysis, and significantly reducing labor costs, testing costs and time costs. By dividing the detection unit, the farmland space is discretized into multiple representative detection units, so as to reflect the soil condition of a large area of ​​farmland with a small number of detection points, thereby improving the efficiency and economy of farmland soil risk monitoring as a whole.

[0036] Based on gamma spectral data, a multi-dimensional characteristic parameter system was constructed, consisting of anomaly characteristic parameters of nuclide concentration, characteristic parameters of nuclide ratio, characteristic parameters of comprehensive intensity of energy spectrum, and characteristic parameters of energy spectrum structure change. It does not rely solely on a single nuclide or a single indicator for risk assessment. Through the synergistic effect of multiple characteristic parameters, it can more comprehensively reflect various influencing factors such as soil parent material conditions, weathering degree, human disturbance, and potential pollution, thereby improving the stability and accuracy of farmland soil risk identification and reducing the possibility of misjudgment due to fluctuations in a single indicator.

[0037] By introducing a rapid discrimination model based on threshold rules, a comprehensive risk index model based on weighted scoring, and a model based on machine learning algorithms to conduct multi-model collaborative or fusion judgment, the applicability limitations of a single model under different farmland conditions are effectively reduced, and the reliability and consistency of risk screening results are improved.

[0038] Based on the model output, farmland soil risk is divided into low-risk, medium-risk, and high-risk levels, and the risk level is correlated with the spatial location of the detection unit to form farmland soil risk distribution results. Through the risk level classification output, intuitive and clear risk information can be provided to agricultural management departments and farmers, guiding the formulation of subsequent refined sampling, pollution control, or farmland management measures, avoiding blind and comprehensive detection or control, and significantly improving the pertinence and scientific nature of farmland soil risk management.

[0039] The method of this invention has clear steps and complete logic, low dependence on equipment, and can be adapted to different types of gamma spectrometers. It is easy to deploy in different farmland areas and different agricultural management scenarios. At the same time, it can be easily expanded into a farmland soil risk assessment system or continuous monitoring platform based on gamma spectrum, and has a good foundation for engineering implementation and prospects for promotion and application. Attached Figure Description

[0040] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0041] Figure 1 This is a flowchart of the rapid risk screening method for farmland soil based on gamma spectroscopy according to the present invention;

[0042] Figure 2This is a schematic diagram illustrating the relationship between soil risk characteristic parameters and the rapid soil risk screening model in this invention;

[0043] Figure 3 This is a structural diagram of the rapid risk screening system for farmland soil based on gamma energy spectrum of the present invention;

[0044] Figure 4 This is a schematic diagram of the division of farmland detection units and the layout of gamma-ray spectrum detection points in this invention. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0046]

Example 1

[0047] like Figure 1 As shown in this embodiment, a rapid risk screening method for farmland soil based on gamma spectroscopy is provided, which consists of the following steps:

[0048] S1. Farmland Area Division and Monitoring Point Layout: A typical contiguous farmland area is selected as the target farmland area, and the boundary information of the farmland area is obtained through a geographic information system. Based on the farmland area and cultivation method, the target farmland area is divided into monitoring units according to a regular grid of 50 m × 50 m. Each monitoring unit is equipped with a gamma-ray spectroscopy monitoring point, which is located at the center of the monitoring unit and is used to represent the soil condition of that monitoring unit.

[0049] S2. In-situ acquisition of gamma-ray spectrometer data: At each detection point, a portable gamma-ray spectrometer was used to conduct in-situ measurements of farmland soil. During the measurement process, the gamma-ray spectrometer detector was set at a height of approximately 1 m above the ground surface, and the measurement time at each detection point was set to 180 s to acquire the corresponding gamma-ray spectrometer data. The collected data includes: 40 K nuclide count data 238 U-series nuclide counting data 232 Th series nuclide count data, total count rate per unit time, and count distribution information for different energy ranges.

[0050] S3. Gamma Spectrum Data Preprocessing: The acquired raw gamma spectrum data is preprocessed as follows:

[0051] Perform energy calibration on the energy spectrum channels;

[0052] The influence of ambient background radiation on the measurement results was excluded;

[0053] Convert the count data into a count rate per unit time;

[0054] The energy spectrum curve is smoothed and filtered to reduce random noise interference.

[0055] S4. Soil Risk Characteristic Parameter Extraction: Based on the preprocessed gamma spectral data, a soil risk characteristic parameter system is constructed, including:

[0056] Nuclide concentration anomaly characteristic parameters: obtained by comparing the nuclide concentration at each detection point with the regional background value;

[0057] Nuclide ratio characteristic parameters: including 238 U / 232 Th、 40 K / 232 Th ratio;

[0058] The overall intensity characteristic parameters of the energy spectrum are characterized by the total count rate per unit time.

[0059] Energy spectrum structure variation characteristic parameters: obtained by counting ratios of different energy bands.

[0060] S5. Rapid screening model based on threshold rules: Input the above soil risk characteristic parameters into the rapid discrimination model based on threshold rules. When any nuclide anomaly index or nuclide ratio parameter exceeds the preset threshold, the corresponding detection unit is determined to have soil risk.

[0061] S6. Soil Risk Level Classification: Based on the model's judgment results, the risk of farmland soil corresponding to the detection unit is classified:

[0062] Detection units that do not exceed the threshold are classified as low-risk;

[0063] Exceeding a single threshold is considered a medium risk.

[0064] Exceeding multiple thresholds is considered high risk, ultimately resulting in the risk distribution of farmland soil.

[0065]

Example 2

[0066] See Figure 2 Based on Example 1, this embodiment introduces a weighted scoring model and a machine learning model to construct a multi-model fusion risk screening method.

[0067] S1. Acquisition of characteristic parameters: Soil risk characteristic parameters are acquired according to the method in Example 1.

[0068] S2. Weighted Scoring Model Construction: Based on the acquired feature parameters, a comprehensive risk index model is constructed.

[0069] ,in, These are the characteristic parameters of radionuclide anomalies. The characteristic parameter of the nuclide ratio, These are the comprehensive intensity characteristic parameters of the energy spectrum. The weighting coefficients are determined based on historical monitoring data and are characteristic parameters of energy spectrum structure changes.

[0070] S3. Machine Learning Model Construction: Using historical farmland monitoring data, a soil risk classification model based on the random forest algorithm is constructed. The model input is soil risk characteristic parameters, and the output is soil risk category.

[0071] S4. Multi-model fusion judgment: The output results of the threshold model, weighted scoring model and machine learning model are fused. When the judgment results of different models are inconsistent, the result with the higher risk level is adopted as the final judgment result.

[0072] S5. Risk Level Output: Based on the output results of the fusion model, farmland soil risks are divided into low risk, medium risk and high risk, and corresponding spatial distribution results are generated.

[0073]

Example 3

[0074] This embodiment selects three different types of farmland areas, including dryland farmland, paddy field farmland, and facility agriculture farmland, and uses the method of this invention for rapid soil risk screening in each area. Under different farmland conditions, by adjusting the detection unit scale and model threshold parameters, effective identification and classification of soil risks can be achieved, indicating that the method of this invention has good regional adaptability and stability.

[0075]

Example 4

[0076] like Figure 3-4 As shown, this embodiment provides a mobile, gamma-ray spectroscopy-based rapid risk screening system for farmland soil, suitable for rapid on-site detection and initial risk screening of single plots or small-scale farmland, including:

[0077] Farmland spatial information acquisition module: The boundary information of the target farmland area is acquired by manual input or by calling a pre-stored farmland boundary file, and the boundary information is transmitted to the detection unit division module;

[0078] Detection unit division module: Based on farmland boundary information and preset unit area thresholds, the farmland is divided into multiple detection units, and corresponding detection point coordinates are generated for each detection unit to guide on-site detection;

[0079] Gamma Spectrum Data Acquisition Module: Operators carry a mobile gamma spectrometer and perform in-situ measurements at each detection point according to the detection point coordinates to acquire the raw gamma spectrum data of the corresponding detection unit.

[0080] Energy spectrum data preprocessing module: Performs energy calibration correction, background and environmental interference correction, time normalization, and noise suppression and smoothing on the collected raw energy spectrum data in sequence to form standardized energy spectrum data;

[0081] Soil risk characteristic parameter construction module: Based on standardized energy spectrum data, it extracts anomaly characteristic parameters of nuclide concentration, nuclide ratio characteristic parameters, comprehensive energy spectrum intensity characteristic parameters, and energy spectrum structure change characteristic parameters to form soil risk characteristic vectors corresponding to the detection units;

[0082] Soil Risk Rapid Screening Model Module: Input the soil risk feature vector into the built-in risk screening model based on a combination of threshold rules and weighted scoring to calculate the corresponding soil risk index;

[0083] Risk level determination and result output module: Based on the correspondence between the risk index and the preset risk level range, the risk level of the detection unit is determined, and the screening results are output on the terminal interface in the form of text or color marking.

[0084] This embodiment, through a highly integrated system structure, enables rapid on-site screening of farmland soil risks, significantly reducing testing costs and improving testing efficiency, making it suitable for rapid screening applications in agricultural production.

[0085]

Example 5

[0086] This embodiment provides a distributed, gamma-ray spectrum-based rapid screening system for farmland soil risks, suitable for unified management and risk assessment of regional or large-scale farmland. It includes a front-end data acquisition subsystem and a back-end data analysis and management platform, which are connected via wireless or wired communication.

[0087] The front-end data acquisition subsystem includes a gamma spectrum data acquisition module, a positioning module, and a data caching and transmission module, which are used to complete the acquisition of spectrum data in the field and upload the acquired data to the back-end platform.

[0088] The backend data analysis and management platform includes a farmland spatial information acquisition module, a detection unit division module, an energy spectrum data preprocessing module, a soil risk characteristic parameter construction module, a soil risk rapid screening model module, and a risk level determination and result output module.

[0089] The implementation method of this embodiment is as follows:

[0090] The backend platform first obtains the farmland spatial boundary information of the target area and completes the unified division of detection units;

[0091] The front-end acquisition subsystem collects and uploads gamma spectrum data at each detection point;

[0092] The backend platform performs centralized preprocessing and feature construction on the uploaded energy spectrum data;

[0093] The backend platform uses machine learning algorithms to build a rapid soil risk screening model and predict the risk of the detection units.

[0094] The backend platform outputs the screening results in the form of risk level charts, statistical reports, or interface data.

[0095] This embodiment achieves centralized analysis and unified management of large-scale farmland risks through a front-end and back-end separated system architecture, improving the system's scalability and model update capabilities.

[0096]

Example 6

[0097] This embodiment further provides a multi-model fusion intelligent screening system for farmland soil risks, based on the aforementioned embodiments.

[0098] In the rapid soil risk screening model module, a threshold rule discrimination sub-model, a weighted scoring risk index sub-model, and a machine learning prediction sub-model are set up, and a model fusion unit is set up to fuse and determine the output results of each sub-model.

[0099] When the threshold rule discrimination sub-model identifies a risk as high, the system directly outputs a high-risk result;

[0100] When the outputs of the sub-models are inconsistent, the model fusion unit generates the final risk level based on the confidence level or weighting strategy.

[0101] Output the risk level results after fusion.

[0102] This embodiment improves the stability, robustness, and adaptability of soil risk screening results through multi-model collaboration, making it particularly suitable for risk identification in complex farmland environments.

[0103] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A rapid risk screening method for farmland soil based on gamma spectroscopy, characterized in that, Includes the following steps: S1. Obtain the boundary information of the target farmland area, and divide the farmland area into detection units according to the farmland scale, farming method and plot shape. The detection unit corresponds to the gamma spectrum detection point, and the location of the detection point is used to represent the soil condition of the corresponding detection unit. S2. At each detection point, a gamma spectrometer is used to conduct in-situ measurements of farmland soil to obtain gamma spectrum data; S3. Preprocess the raw gamma spectrum data obtained in step S2. The preprocessing includes energy scale correction, background and environmental interference correction, time normalization, and noise suppression and smoothing. S4. Based on the preprocessed gamma spectral data, a feature parameter system for rapid soil risk screening is constructed. The feature parameters include nuclide concentration anomaly feature parameters, nuclide ratio feature parameters, comprehensive energy spectrum intensity feature parameters, and energy spectrum structure change feature parameters. S5. Based on the soil risk characteristic parameters, construct a rapid soil risk screening model, wherein the screening model includes one or more of the following: a rapid discrimination model based on threshold rules, a comprehensive risk index model based on weighted scoring, and a model based on machine learning algorithms. S6. Based on the model output results, classify the risk levels of farmland soil into three categories: low risk, medium risk, and high risk.

2. The method for rapid screening of farmland soil risk based on gamma spectroscopy according to claim 1, characterized in that, The sources of the boundary information include geographic information systems, remote sensing images, and farmland management systems. The division method of the detection unit includes one or more of the following: regular grid division method, division method based on actual field boundaries, and adaptive division method combining crop type, irrigation conditions, and management unit.

3. The method for rapid screening of farmland soil risk based on gamma spectroscopy according to claim 1, characterized in that, The gamma spectral data includes 40 Counting of energy spectrum peaks of K nuclide 238 Energy spectrum peak count of U-series nuclides 232 The energy spectrum peak counts of Th series nuclides, the total energy spectrum count rate per unit time, and the count distribution information in different energy ranges are included. The parameters of the in-situ gamma spectrum measurement include measurement height, single-point measurement time, and number of repeated measurements. These parameters are adaptively adjusted according to farmland topography and soil type conditions.

4. The method for rapid screening of farmland soil risk based on gamma spectroscopy according to claim 1, characterized in that, The nuclide concentration anomaly characteristic parameter in S4 compares the nuclide concentration value at the detection point with the regional background value and historical reference value to calculate the nuclide concentration anomaly index, which is used to characterize the degree of anomaly of radionuclides in the soil, including soil parent material anomaly, potential heavy metal associated anomaly, and anomaly changes caused by human disturbance. The nuclide ratio characteristic parameters are constructed based on the relative relationships between different nuclides, including: 238 U / 232 Th ratio 40 K / 232 Th ratio and 40 K / 238 U ratio; The comprehensive energy spectrum intensity characteristic parameter is constructed based on the total energy spectrum count rate and the cumulative count value of a specific energy band per unit time. The energy spectrum structure change characteristic parameters are constructed by analyzing the count ratio distribution and morphological changes of the energy spectrum curve in different energy ranges.

5. The method for rapid screening of farmland soil risk based on gamma spectroscopy according to claim 1, characterized in that, The rapid discrimination model based on threshold rules in S5 compares various soil risk characteristic parameters with preset thresholds. When a parameter exceeds the corresponding threshold, it is determined that the corresponding detection unit has soil risk. The threshold is determined based on regional statistical background, historical monitoring data and empirical rules. The weighted scoring-based comprehensive risk index model is constructed based on multidimensional soil risk characteristic parameters. Its calculation method is as follows: ,in, This is the comprehensive soil risk score. For the first Soil risk characteristic parameters, These are the corresponding weighting coefficients; The soil risk model based on machine learning algorithms uses historical farmland monitoring data to construct the machine learning model. The model input is soil risk characteristic parameters, and the output is either a soil risk category or a risk probability.

6. The rapid screening method for farmland soil risk based on gamma spectroscopy according to claim 1, characterized in that, The machine learning algorithm includes one of the following: support vector machine, decision tree, random forest, or neural network algorithm.

7. A rapid risk screening system for farmland soil based on gamma spectroscopy, characterized in that, It includes a farmland spatial information acquisition module, a detection unit division module, a gamma spectrum data acquisition module, a spectrum data preprocessing module, a soil risk characteristic parameter construction module, a soil risk rapid screening model module, and a risk level determination and result output module; The division criteria for the detection unit division module include farmland area size, plot shape regularity, and farming method.

8. The rapid farmland soil risk screening system based on gamma spectroscopy according to claim 7, characterized in that, The farmland spatial information acquisition module includes GIS vector boundary data, farmland outlines generated from remote sensing images, and plot numbers and boundary information from the agricultural plot management system.

9. A rapid risk screening system for farmland soil based on gamma spectroscopy according to claim 7, characterized in that, The soil risk characteristic parameter construction module includes sub-modules for extracting anomaly characteristic parameters of nuclide concentration, nuclide ratio characteristic parameters, comprehensive intensity characteristic parameters of energy spectrum, and characteristic parameters of energy spectrum structure change.

10. A rapid risk screening system for farmland soil based on gamma spectroscopy according to claim 7, characterized in that, The rapid soil risk screening model module includes a threshold rule discrimination sub-model, a comprehensive risk index sub-model, and a machine learning prediction sub-model, and is equipped with a model fusion unit for fusing and judging the output results of multiple sub-models; The fusion judgment combines the outputs of the threshold model, the weighted scoring model, and the machine learning model. When the judgment results of different models are inconsistent, the result with the higher risk level is adopted as the final judgment result.

Images