Soil carbon sequestration and emission reduction detection method combined with isotopic tracing technology

Abstract

The invention discloses a soil carbon sequestration and emission reduction detection method combined with an isotopic tracing technology, and relates to the technical field of environmental science and agriculture, and the method comprises the following steps: building an online coupling system comprising a front-end column chromatography separation module, a middle-end isotope ratio mass spectrometer and a rear-end AI algorithm; and carrying out front-end sampling and column chromatography separation in the target area to obtain classified gas components. According to the on-line coupling system combined with the isotope tracing technology, through cooperative work of the front-end column chromatography separation module, the middle-end isotope ratio mass spectrometer and the rear-end AI algorithm, monitoring data can be received and processed in real time, an emission source contribution thermodynamic diagram can be rapidly generated, the data acquisition and processing time is greatly shortened, and the detection accuracy is improved. Therefore, management personnel can timely acquire accurate emission information, quick response is realized, effective management and control measures are taken, missing of an optimal intervention window due to data lag is avoided, and the efficiency of soil carbon sequestration emission reduction monitoring and the data timeliness are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the fields of environmental science and agricultural technology, and in particular to a soil carbon sequestration and emission reduction detection method combined with isotope tracing technology. Background Art

[0002] Soil is one of the largest carbon pools in terrestrial ecosystems and has enormous carbon sequestration potential. Agriculture is one of the important sources of greenhouse gas emissions. The emissions of gases such as carbon dioxide, methane and nitrous oxide in the soil contribute significantly to the greenhouse effect. Through soil carbon sequestration and emission reduction testing, the emission characteristics and dynamic changes of greenhouse gases during wheat and corn cultivation can be clarified, providing a basis for formulating scientific and reasonable emission reduction measures, thereby effectively reducing the impact of agricultural activities on climate change.

[0003] In the existing technology, soil carbon sequestration and emission reduction detection relies heavily on offline sampling-laboratory analysis processes, with a long data acquisition cycle, making it difficult to capture short-term emission pulses and daily / seasonal scale fluctuations, which in turn causes subsequent management measures to miss the optimal intervention window due to data lags. Therefore, how to build an online coupling system including column chromatography, isotope tracing and AI algorithm real-time analysis to generate a heat map of emission source contributions and realize an integrated soil carbon sequestration and emission reduction detection process of separation-detection-traceability is the problem to be solved by the present invention. To this end, a soil carbon sequestration and emission reduction detection method combined with isotope tracing technology is proposed. Summary of the Invention

[0004] The purpose of the present invention is to provide a soil carbon sequestration and emission reduction detection method combined with isotope tracing technology to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0006] A soil carbon sequestration and emission reduction detection method combined with isotope tracing technology comprises the following steps:

[0007] Step 1: Build an online coupling system consisting of a front-end column chromatography separation module, a mid-end isotope ratio mass spectrometer, and a back-end AI algorithm;

[0008] Step 2: Perform front-end sampling and column chromatography separation in the target area to obtain classified gas components, and perform mid-end isotope ratio mass spectrometry detection to obtain gas isotope characteristic data;

[0009] Step 3: The back-end AI algorithm receives mass spectrometer data in real time, analyzes gas concentrations and isotope fingerprint characteristics, and identifies emission source types and contribution ratios;

[0010] Step 4: Based on the analysis results, the AI ​​algorithm generates a heat map of soil greenhouse gas emission source contributions, showing the distribution and intensity of emission sources in different regions;

[0011] Step 5: Combine heat maps with isotope data to trace the source of greenhouse gas emissions, conduct continuous dynamic monitoring, and capture short-term emission pulses and daily / seasonal fluctuations;

[0012] In step 6, the system automatically matches the preset warning threshold, triggers the warning signal and adjusts agricultural management measures to conduct control interventions to improve carbon sequestration and emission reduction effects.

[0013] A further improvement of the technical solution of the present invention is that: the step 1 comprises:

[0014] Select wheat and corn planting plots where soil carbon sequestration and emission reduction testing is required as target areas, set monitoring targets, and then select column chromatography separation modules, isotope ratio mass spectrometers, and AI algorithms as needed;

[0015] Based on the selected column chromatography separation module, isotope ratio mass spectrometer and AI algorithm, an online coupling system suitable for the target area is built, including a front-end column chromatography separation module, a mid-end isotope ratio mass spectrometer and a back-end AI algorithm. Ensure that the hardware connections of each module are correct and the data transmission lines are unobstructed. Optimize the column chromatography separation conditions and adjust the isotope ratio mass spectrometer parameters to achieve optimal detection sensitivity. At the same time, initialize and configure the back-end AI algorithm to ensure that it can receive and process data in real time.

[0016] A further improvement of the technical solution of the present invention is that: step 2 includes:

[0017] Deploy gas sampling devices in a 10m x 10m grid in the target area. The sampling devices introduce gas samples into the column chromatography separation module through the gas path system. Ensure that the sampling devices and column chromatography modules are tightly connected and the gas path is unobstructed to avoid gas leakage or contamination.

[0018] After receiving the gas sample, the column chromatography separation module separates the gas according to its physical and chemical properties. By optimizing the column temperature and carrier gas flow rate, different greenhouse gas components are effectively separated. The separated gas components are then sequentially fed into the isotope ratio mass spectrometer to provide pure gas samples for subsequent testing, ensuring the accuracy of the test results.

[0019] The isotope ratio mass spectrometer receives the separated gas components and uses a dual-inlet method to simultaneously detect isotope characteristic data. The conventional inlet introduces sample gas, and the high-concentration inlet injects standard gas with a known isotope ratio for mass discrimination correction. At the same time, the chromatographic separation time axis is aligned through the dynamic time warping algorithm to generate a concentration-isotope data stream, and then push the data to the back-end AI algorithm in real time.

[0020] A further improvement of the technical solution of the present invention is that: step 3 includes:

[0021] The backend AI algorithm receives the concentration-isotope data stream pushed by the mass spectrometer in real time via the MQTT protocol, analyzes the timestamp, gas composition, and isotope ratio, segments the data using a sliding window method with a window length of 15 minutes, applies a Kalman filter to smooth noise, fills in missing values ​​using the KNN algorithm to ensure the spatiotemporal continuity of the data, and standardizes the data and converts it into a unified format.

[0022] The backend AI algorithm uses a pre-trained LSTM-Random Forest hybrid model to analyze the standardized data, extracting the dynamic characteristics and isotope fingerprint characteristics of the gas concentration time series. It then compares them with a known isotope signature database to identify the emission source types of different greenhouse gases.

[0023] Based on the identified emission source types, the contribution ratio of each emission source to greenhouse gas concentration is further analyzed to quantify the relative contribution of each emission source.

[0024] A further improvement of the technical solution of the present invention is that the analysis process of the emission source type and contribution ratio is as follows:

[0025] Use a pre-trained LSTM model to analyze the standardized gas concentration time series data, capturing the dynamic features in the time series data and extracting the pattern of gas concentration changes over time. At the same time, it also extracts the isotope fingerprint features, that is, the characteristics of different isotope ratios in the gas.

[0026] The extracted isotope fingerprint characteristics are compared with a known isotope signature database, where the database stores typical isotope signatures of different emission sources. The characteristic identification of each emission source is determined through comparison, that is, the unique isotope fingerprint of each emission source. Based on the comparison results, the random forest model is used to classify and identify the emission source type, and the identification results are output to clarify the type of greenhouse gas emission source at each sampling point;

[0027] Based on the identified emission source types, we further analyzed the contribution ratio of each emission source to greenhouse gas concentrations. We used the output probability of the random forest model to quantify the relative contribution of each emission source. Combined with the gas concentration time series data, we analyzed the changes in the contribution of different emission sources in different time periods.

[0028] A further improvement of the technical solution of the present invention is that: step 4 includes:

[0029] The AI ​​algorithm aggregates the analyzed emission source contribution data into 10m×10m grids, calculates the mean contribution ratio of each emission source within each grid, and uses the Kriging interpolation algorithm, combined with spatial autocorrelation characteristics, to convert the discrete grid data into a continuous spatial distribution field. This generates a basic contribution density map covering the entire target area, ensuring spatial continuity.

[0030] Based on the interpolation results, thematic contribution maps were generated for each emission source type and merged into a comprehensive heat map using a weighted overlay algorithm. The contribution intensity was divided into five levels (low / medium-low / medium / medium-high / high) using the natural breakpoint method. These maps were visualized using a color gradient from blue to yellow to red. The dominant emission source type in each region was also labeled to enhance information readability.

[0031] The system receives new data every 15 minutes and triggers incremental updates, recalculating only the grid areas affected by the new data, refreshing the heat map through GPU-accelerated rendering, and ultimately outputting a visual interface that supports interactive operations.

[0032] A further improvement of the technical solution of the present invention is that the generation process of the basic contribution density map is:

[0033] The emission source contribution data analyzed by the AI ​​algorithm is aggregated according to 10m×10m grids. The data of all sampling points in each grid are summarized to ensure the data in each grid is complete. For each grid, the mean contribution ratio of each emission source is calculated. The mean contribution ratio of emission source calculated in each grid is stored as grid contribution data. Each grid corresponds to a set of mean contribution ratios of emission sources, forming a discrete grid data set.

[0034] The Kriging interpolation algorithm is used to perform spatial interpolation on the grid contribution data in the discrete grid data set, estimate the value of the unknown point based on the value of the known point, and analyze the spatial autocorrelation of the grid contribution data to determine the parameters of the Kriging interpolation;

[0035] According to the Kriging interpolation results, the discrete grid data is converted into a continuous spatial distribution field, and the continuous spatial distribution field generated by the Kriging interpolation is plotted into a basic contribution density map covering the entire target area. The generated basic contribution density map is verified by comparing it with the known emission source distribution information to ensure its rationality and accuracy. Based on the verification results, the Kriging interpolation parameters or AI algorithm are optimized to further improve the accuracy of the basic contribution density map.

[0036] A further improvement of the technical solution of the present invention is that: step 5 includes:

[0037] The generated heat map is combined with the isotope data for analysis. The heat map provides spatial distribution information, and the isotope data provides emission source characteristic identification. The isotope fingerprint characteristics are associated with the emission source type. The spatial resolution of the heat map and the density of the isotope sampling points are synchronously calibrated to ensure the spatial and temporal matching of the data, forming a multi-dimensional emission source characteristic database.

[0038] A Bayesian hybrid model is used to integrate the contribution ratio of the heat map with isotope concentration data in real time, reversely inferring the instantaneous contribution intensity of each emission source. Wavelet analysis is then used to extract high-frequency pulses and low-frequency periods in the time series, obtain emission pulse characteristics, and mark the spatiotemporal coordinates of abnormal emission events. When an abnormal emission event is detected, a high-precision resampling verification mechanism is triggered.

[0039] Based on the historical emission pulse characteristics, the heat map update frequency and isotope sampling strategy are dynamically adjusted, and the LSTM neural network is used to predict the emission trend in the next 24 hours.

[0040] A further improvement of the technical solution of the present invention is that the process of using the LSTM neural network to predict the emission trend in the next 24 hours is:

[0041] Collect historical emission data, including the contribution ratio of emission sources in the thermal map, isotope concentration data, and time series data; organize the data; extract historical emission pulse characteristics; use wavelet analysis to extract high-frequency pulse characteristics and low-frequency periodic characteristics from historical emission data; analyze emission pulse characteristics in different regions; identify areas and time periods with frequent emission activities; and then store the extracted emission pulse characteristics in a feature database, including the spatiotemporal coordinates of each sampling point and pulse characteristic parameters;

[0042] Based on the historical emission pulse characteristics, the target area is divided into sub-areas of different priorities. High-priority areas are areas with frequent and large emission pulses; low-priority areas are areas with less emission activities. Based on the regional priority, the update frequency of the heat map is dynamically adjusted, and the isotope sampling strategy is adjusted at the same time. Then, a dynamic adjustment strategy is implemented, and the heat map update and isotope sampling processes are monitored in real time. Based on real-time data feedback, the adjustment strategy is further optimized to ensure the effective use of resources and the accuracy of data.

[0043] Use historical emission data to train the LSTM neural network model and optimize model parameters to ensure that the model can accurately capture the dynamic characteristics of the time series. Input the latest emission data into the trained LSTM neural network model to predict the emission trend for the next 24 hours. The prediction results include the emission concentration and emission source contribution ratio of each sampling point. The prediction results are then visualized to generate a heat map of the emission trend for the next 24 hours.

[0044] A further improvement of the technical solution of the present invention is that: step 6 includes:

[0045] The system monitors emission data in real time and automatically compares the current emission indicators of each emission source with the preset warning threshold. When the monitored value exceeds the warning threshold, the warning signal is immediately triggered and the management personnel are notified through the visual interface, SMS or email to ensure timely response;

[0046] Based on the early warning signals, the causes of exceeding the standards are analyzed and corresponding control strategies are matched from the knowledge base to generate targeted management measures and recommendations. Managers can then adjust agricultural activities accordingly.

[0047] Managers implement control interventions in accordance with the adjusted measures, and continuously monitor the effects of the interventions, compare data before and after the interventions, evaluate the effectiveness of the measures, provide reference for subsequent management, and ensure that agricultural activities meet carbon sequestration and emission reduction goals.

[0048] Due to the adoption of the above technical solution, the present invention has the following technical advancements compared to the prior art:

[0049] 1. The present invention provides a soil carbon sequestration and emission reduction detection method combined with isotope tracing technology, and an online coupling system combined with isotope tracing technology. Through the collaborative work of the front-end column chromatography separation module, the mid-end isotope ratio mass spectrometer and the back-end AI algorithm, it can receive and process monitoring data in real time, quickly generate emission source contribution heat maps, greatly shorten the time for data acquisition and processing, and ensure that management personnel can obtain accurate emission information in a timely manner, so as to quickly respond and take effective management and control measures, avoid missing the optimal intervention window due to data lag, and significantly improve the efficiency of soil carbon sequestration and emission reduction monitoring and data timeliness.

[0050] 2. The present invention provides a soil carbon sequestration and emission reduction detection method combined with isotope tracing technology. It utilizes the association between isotope fingerprint characteristics and emission source types, and analyzes gas concentrations and isotope characteristics through AI algorithms. It can accurately identify different emission source types and their contribution ratios. It can not only distinguish different emission sources such as soil organic matter decomposition, microbial activity, and fertilizer application, but also quantify the relative contribution of each emission source to greenhouse gas concentrations. By dynamically monitoring and analyzing the spatiotemporal changes of emission sources, managers can identify high-emission areas and major emission sources, thereby formulating targeted emission reduction strategies, optimizing agricultural management measures, effectively reducing greenhouse gas emission risks, and improving carbon sequestration and emission reduction effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments described in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0052] Figure 1 It is a schematic diagram of the workflow of the present invention;

[0053] Figure 2 Schematic diagram of the method of the present invention. DETAILED DESCRIPTION

[0054] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0055] Example 1, as Figure 1 、 Figure 2 As shown, the present invention provides a soil carbon sequestration and emission reduction detection method combined with isotope tracing technology, comprising the following steps:

[0056] Step 1: Build an online coupling system including a front-end column chromatography separation module, a mid-end isotope ratio mass spectrometer, and a back-end AI algorithm. Select wheat and corn planting plots where soil carbon sequestration and emission reduction detection is required as target areas, set monitoring targets, and then select column chromatography separation modules, isotope ratio mass spectrometers, and AI algorithms as needed. According to the selected column chromatography separation modules, isotope ratio mass spectrometers, and AI algorithms, build an online coupling system suitable for the target area, including a front-end column chromatography separation module, a mid-end isotope ratio mass spectrometer, and a back-end AI algorithm. Ensure that the hardware connections of each module are correct and the data transmission lines are unobstructed, optimize the column chromatography separation conditions, adjust the isotope ratio mass spectrometer parameters to achieve optimal detection sensitivity, and at the same time, initialize and configure the back-end AI algorithm to ensure that it can receive and process data in real time.

[0057] The specific workflow of step 1 is as follows:

[0058] Select plots of land with continuous wheat-corn rotation for more than three years, covering different soil types and terrains, ensure that the plots have stable power supply, 4G / 5G signal coverage and irrigation systems, and set monitoring targets. The core indicators are 、 、 Emission flux and its isotopic characteristics ( 、 、 ), for the temporal and spatial resolution, the spatial resolution is 10 m×10 m, the time resolution is 15 minutes, and it is encrypted to 5 minutes at night. The selection basis of the front-end column chromatography separation module is that it needs to be completed within 5 minutes 、 、 Baseline separation, adaptable to high humidity and dusty gas environments, the selection basis of mid-range isotope ratio mass spectrometer is multi-isotope detection capability: simultaneous measurement is required 、 、 , and the accuracy is ≤0.3‰, adapting to gas concentration fluctuations. The selection basis of the back-end AI algorithm is that the data analysis and thermal map generation of 10 monitoring points must be completed within 1 minute; build an online coupling system, and according to the selected column chromatography separation module, isotope ratio mass spectrometer and AI algorithm, connect the outlet of the column chromatography separation module to the inlet of the isotope mass spectrometer through a 1 / 8-inch stainless steel capillary (length 2m), and install a three-way valve in the middle for standard gas calibration. The mass spectrometer pushes the concentration-isotope data stream to the server of the AI ​​algorithm in real time through the TCP / IP protocol, and performs data transmission testing to ensure that the column chromatography separation module can transmit the separated gas component data to the isotope ratio mass spectrometer, and ensure that the isotope ratio mass spectrometer can transmit the detected isotope characteristic data to the AI ​​algorithm in real time; optimize the column chromatography separation conditions, adjust the column chromatography separation conditions, ensure that the greenhouse gas components can be effectively separated, and pass the standard Gas samples are tested to optimize separation conditions to ensure optimal separation results. The parameters of the isotope ratio mass spectrometer are adjusted according to the isotopic characteristics of the target gas. Standard isotope gases are used for calibration to ensure optimal detection sensitivity and accuracy of the mass spectrometer. The AI ​​algorithm is initialized and configured, including the data receiving interface, data processing model, and warning threshold settings, to ensure that the AI ​​algorithm can receive and process data transmitted by the mass spectrometer in real time and generate thermal maps and traceability results. Actual sampling tests are conducted in the target area, and the entire online coupling system is run to check whether the system can operate normally, including sampling, separation, detection, data transmission, and analysis. The system detection results are compared and verified with the known emission source characteristics to ensure the accuracy and reliability of the system. The results of offline sampling and laboratory analysis are compared to verify the performance advantages of the online coupling system.

[0059] Step 2: perform front-end sampling and column chromatography separation in the target area to obtain classified gas components, and perform mid-end isotope ratio mass spectrometry detection to obtain isotope characteristic data of the gas. The gas sampling device is deployed in a 10m×10m grid layout in the target area. The sampling device introduces the gas sample into the column chromatography separation module through the gas path system, ensuring that the sampling device is tightly connected to the column chromatography module and the gas path is unobstructed to avoid gas leakage or contamination, and ensure that the collected gas sample can accurately reflect the emission situation in the target area. After receiving the gas sample, the column chromatography separation module separates the gas according to its physical and chemical properties. By optimizing the column temperature and carrier gas flow rate, different greenhouse gas components are effectively separated. The separated gas components are then fed into the isotope ratio mass spectrometer in sequence to provide pure gas samples for subsequent testing and ensure the accuracy of the test results. The isotope ratio mass spectrometer receives the separated gas components and uses a dual-inlet method to simultaneously detect isotope characteristic data. The conventional inlet introduces sample gas, and the high-concentration inlet injects standard gas with a known isotope ratio for mass discrimination correction. At the same time, the chromatographic separation time axis is aligned through the dynamic time warping algorithm to generate a concentration-isotope data stream, which is then pushed to the back-end AI algorithm in real time.

[0060] The specific workflow of step 2 is as follows:

[0061] In the target area (wheat or corn planting plot), sampling points are divided into a 10m×10m grid layout, and the location of each sampling point is marked to ensure that the sampling points are evenly distributed and fully cover the target area. A gas sampling device, such as a soil gas diffusion tube, a gas chromatography injector, or other equipment suitable for soil gas collection is installed at each sampling point. Ensure that the sampling device is firmly installed to prevent the device from loosening or being damaged by external factors (such as wind, animals, etc.). The sampling device is connected to the column chromatography separation module through the gas path system. Check whether the gas path connection is tight to ensure that the gas path is unobstructed to avoid gas leakage or contamination. Perform a sealing test on the gas path system and use soapy water to check whether there are bubbles at the connection to ensure that the gas sample can accurately reflect the emission situation in the target area. Start the gas sampling device to collect soil gas samples in the target area. Ensure that the sampling frequency and sampling time meet the detection requirements. Set it to sample once an hour to capture short-term emission pulses. Introduce the collected gas samples into the column chromatography separation module through the gas path system. The column chromatography separation module optimizes the column chromatography separation conditions according to the physical and chemical properties of the gas. Set the appropriate column temperature according to the boiling point and polarity of the target gas. For example, for 、 、 The column temperature is set between 40°C and 60°C. At the same time, a suitable carrier gas, such as helium or nitrogen, is selected, and the flow rate is adjusted to ensure that the gas components can be effectively separated. The carrier gas flow rate is set between 1.0mL / min and 2.0mL / min. Through the optimized column chromatography conditions, different greenhouse gas components are effectively separated to ensure the purity of the separated gas components; the gas components separated by column chromatography are sequentially sent to the isotope ratio mass spectrometer. The isotope ratio mass spectrometer adopts a dual-inlet method to synchronously detect isotope characteristic data. The conventional inlet introduces sample gas for isotope ratio measurement, and the high-concentration inlet injects standard gas with a known isotope ratio for mass discrimination correction. The chromatographic separation time axis is aligned through the dynamic time warping algorithm to ensure the accuracy of the data, and a concentration-isotope data stream is generated, including gas concentration and isotope ratio information, which is then pushed to the back-end AI algorithm in real time;

[0062] Step 3: Receive mass spectrometer data in real time through the back-end AI algorithm, analyze gas concentrations and isotope fingerprint characteristics, and identify emission source types and contribution ratios. The back-end AI algorithm receives the concentration-isotope data stream pushed by the mass spectrometer in real time through the MQTT protocol, parses the timestamp, gas composition, and isotope ratio, and segments the data using the sliding window method with a window length of 15 minutes. Kalman filtering is applied to smooth the noise, and missing values ​​are filled using the KNN algorithm to ensure the spatiotemporal continuity of the data. The data is standardized and converted into a unified format. The back-end AI algorithm uses a pre-trained LSTM-random forest hybrid model to analyze the standardized data, extract the dynamic characteristics and isotope fingerprint characteristics of the gas concentration time series, and then compare them with the known isotope characteristic database to identify the emission source types of different greenhouse gases. Based on the identified emission source types, the contribution ratio of each emission source to the greenhouse gas concentration is further analyzed to quantify the relative contribution of each emission source;

[0063] In addition, the analysis process of emission source types and contribution ratios is as follows:

[0064] A pre-trained LSTM model is used to analyze the standardized gas concentration time series data to capture the dynamic features in the time series data and extract the pattern of gas concentration changes over time. At the same time, the isotope fingerprint features, that is, the features of different isotope ratios in the gas, are extracted and compared with a known isotope feature database. The database stores typical isotope features of different emission sources. The characteristic identifier of each emission source is determined by comparison, that is, the unique isotope fingerprint of each emission source. Based on the comparison results, a random forest model is used to classify and identify the emission source type. The identification results are output to clarify the greenhouse gas emission source type of each sampling point. Based on the identified emission source type, the contribution ratio of each emission source to greenhouse gas concentration is further analyzed. The output probability of the random forest model is used to quantify the relative contribution of each emission source. Combined with the gas concentration time series data, the contribution changes of different emission sources over different time periods are analyzed.

[0065] The calculation formula for the output probability of the random forest model is as follows:

[0066] ;

[0067] Where, Represents the input sample Emission source type The probability of represents the number of decision trees in the random forest model, Indicates the A decision tree for input samples The prediction results, is an indicator function, if ,but ;otherwise The output probability of the random forest model is determined by the voting results of each decision tree. For a given input sample, each decision tree in the random forest model will give a prediction result. The final output probability of the random forest model is the average of the prediction results of all decision trees.

[0068] The formula for calculating the contribution ratio of each emission source to greenhouse gas concentration is as follows:

[0069] ;

[0070] Where, Indicates the type of emission source Contribution to greenhouse gas concentrations, represents the number of sampling points, Indicates the The gas at the sampling point belongs to the emission source type The probability of Indicates the The gas concentration at each sampling point, represents the total number of emission source types, represents the total weighted contribution of all emission source types to the gas concentration at all sampling points;

[0071] The calculation formula for the contribution changes of different emission sources in different time periods is as follows:

[0072] ;

[0073] Where, Indicates that in the time window Internal, emission source type Contribution to greenhouse gas concentrations, Represents a time window, including all sampling points within the window;

[0074] The specific workflow of step 3 is as follows:

[0075] The back-end AI algorithm receives the concentration-isotope data stream pushed by the isotope ratio mass spectrometer in real time through the MQTT protocol, ensuring the stability and real-time performance of data transmission, avoiding data loss or delay, and parsing the timestamps, gas components and isotope ratios in the data stream. 、 、 , isotope ratios such as 、 The parsed data is stored in a temporary data buffer, and the sliding window method is used to segment the data. The window length is set to 15 minutes. The data in each window is used as an independent data segment for subsequent analysis, and the Kalman filter algorithm is applied to smooth the data in each data segment to remove noise. The KNN algorithm is used to fill the missing values ​​in the data segment to ensure the spatiotemporal continuity of the data, and then the processed data is standardized and converted into a unified format; the pre-trained LSTM (long short-term memory network) model is used to analyze the standardized gas concentration time series data, and the LSTM model is used to capture the dynamic features in the time series data, extract the pattern of gas concentration changes over time, and extract the features of different isotope ratios in the gas, namely the isotope fingerprint features, to distinguish different emission sources; the extracted isotope fingerprint features are compared with the known isotope feature database, and the isotope fingerprint features are compared with the known isotope feature database. The isotope signature database stores typical isotope signatures of different emission sources, including soil organic matter decomposition, microbial activity, and fertilizer application. Through comparison, the unique isotope fingerprint of each emission source is determined. Based on the comparison results, the random forest model is used to classify and identify the emission source type. The random forest model outputs the greenhouse gas emission source type of each sampling point based on the isotope fingerprint characteristics and known emission source characteristics. The output result clearly defines the emission source type of each sampling point, such as soil organic matter decomposition, microbial activity, or fertilizer application. Based on the identified emission source type, the contribution ratio of each emission source to the greenhouse gas concentration is further analyzed. The output probability of the random forest model is used to quantify the relative contribution of each emission source. Then, combined with the gas concentration time series data, the contribution changes of different emission sources in different time periods are analyzed, and a time series graph of the emission source contribution ratio is generated to intuitively display the dynamic changes of each emission source.

[0076] Step 4: Based on the analysis results, the AI ​​algorithm generates a heat map of soil greenhouse gas emission source contributions, showing the distribution and intensity of emission sources in different regions;

[0077] Step 5: Combine heat maps with isotope data to trace the source of greenhouse gas emissions, conduct continuous dynamic monitoring, and capture short-term emission pulses and daily / seasonal fluctuations;

[0078] In step 6, the system automatically matches the preset warning threshold, triggers the warning signal and adjusts agricultural management measures to conduct control interventions to improve carbon sequestration and emission reduction effects.

[0079] Example 2, as Figure 1 、 Figure 2 As shown, based on Example 1, the present invention provides a technical solution: preferably, step 4 includes:

[0080] The AI ​​algorithm aggregates and analyzes emission source contribution data on a 10m×10m grid, calculates the mean contribution ratio of each emission source within each grid, and uses the Kriging interpolation algorithm, combined with spatial autocorrelation characteristics, to convert the discrete grid data into a continuous spatial distribution field, generating a basic contribution density map covering the entire target area to ensure spatial continuity. Based on the interpolation results, thematic contribution maps are generated by emission source type and merged into a comprehensive heat map using a weighted overlay algorithm. The natural breakpoint method is used to divide the contribution intensity into five levels: low / medium-low / medium / medium-high / high, and visualized with different color gradients from blue to yellow to red. The dominant emission source type in each area is also labeled to enhance information readability. The system receives new data every 15 minutes and triggers incremental updates. Only the grid areas affected by the new data are recalculated, and the heat map is refreshed through GPU-accelerated rendering. The final output is a visualization interface that supports interactive operations. Users can zoom, pan, or switch time dimensions to simultaneously view the spatiotemporal evolution trends of emission source contributions.

[0081] In addition, the generation process of the basic contribution density map is:

[0082] The emission source contribution data analyzed by the AI ​​algorithm are aggregated according to 10m×10m grids, and all sampling point data in each grid are summarized to ensure that the data in each grid is complete. For each grid, the mean contribution ratio of each emission source is calculated, and the mean emission source contribution ratio calculated in each grid is stored as grid contribution data. Each grid corresponds to a set of emission source contribution ratio mean values, forming a discrete grid data set. The Kriging interpolation algorithm is used to spatially interpolate the grid contribution data in the discrete grid data set, and the values ​​of unknown points are estimated based on the values ​​of known points. The spatial autocorrelation of the grid contribution data is analyzed and the parameters of the Kriging interpolation are determined. Based on the Kriging interpolation results, the discrete grid data are converted into a continuous spatial distribution field. The continuous spatial distribution field generated by the Kriging interpolation is plotted as a basic contribution density map covering the entire target area. The generated basic contribution density map is verified by comparing it with the known emission source distribution information to ensure its rationality and accuracy. Based on the verification results, the Kriging interpolation parameters or the AI ​​algorithm are optimized to further improve the accuracy of the basic contribution density map.

[0083] The calculation formula for the mean contribution ratio of each emission source is as follows:

[0084] ;

[0085] Where, Indicates the The average contribution ratio of the emission sources in the current grid, Indicates the number of sampling points in the current grid, Indicates the The first grid sampling point Contribution ratio of various emission sources;

[0086] The calculation formula of Kriging interpolation is as follows:

[0087] ;

[0088] Where, Indicates the target point The estimated value of Represents a known point The actual value of Represents the weight coefficient, which is calculated based on spatial autocorrelation. Indicates the number of known points;

[0089] The specific workflow of step 4 is as follows:

[0090] The target area is spatially divided using a quadtree structure, and the discrete sampling point data are aggregated into 10m×10m grid units. Each grid is defined with a unique ID, and the grid to which the sampling point belongs is calculated based on the coordinates of the sampling point. For the sampling points located at the grid boundary, that is, the sampling points with a distance of <1m from the boundary, the nearest neighbor assignment is used to avoid duplicate attribution. Then, all the sampling point data in each grid are summarized, and the mean contribution ratio of each emission source is calculated. At the same time, the mean contribution ratio of the emission source calculated in each grid is stored as grid contribution data to form a discrete grid data set. Based on the grid contribution data, the experimental variation function is calculated, where the coverage is set. The maximum lag distance of the grid range is set, and the lag interval is set to determine the number of distance bands. For each type of emission source, the semi-variance of each distance band is calculated, and then the theoretical variation function is fitted. The semi-variance model is used to fit the parameters using the least squares method. The evaluation indicators including the residual sum of squares and the coefficient of determination are calculated, and then cross-validation optimization is performed. Each time one grid data is excluded, the grid value is predicted by interpolation of the remaining grids, and the root mean square error is calculated. If the root mean square error is greater than 15%, the variation function parameters are adjusted; according to the Kriging interpolation results, the discrete grid data is converted into a continuous spatial distribution field, and a basic contribution density map covering the entire target area is drawn to make The contribution density is represented by color gradients. The generated basic contribution density map is verified by comparing it with the known emission source distribution information to ensure its rationality and accuracy. According to the verification results, the Kriging interpolation parameters or AI algorithm are optimized to further improve the accuracy of the basic contribution density map. Based on the interpolation results, thematic contribution maps are generated by emission source type. A thematic map is generated for each emission source type to show its contribution distribution in the target area. The weighted overlay algorithm is used to merge the thematic contribution maps of different emission source types into a comprehensive heat map. Different weights are assigned according to the importance or contribution ratio of each emission source. The system uses a natural breakpoint method to divide the contribution intensity into five levels: low / medium-low / medium / medium-high / high. It uses a blue→yellow→red color gradient for visualization. The dominant emission source type in each area is marked on the comprehensive heat map. The system receives new data every 15 minutes and triggers an incremental update. Only the grid area affected by the new data is recalculated to avoid recalculating the entire dataset. The heat map is refreshed through GPU-accelerated rendering to ensure the smoothness of the visualization interface. The output supports interactive operations. Users can zoom, pan, or switch time dimensions to simultaneously view the spatiotemporal evolution trend of emission source contributions.

[0091] Step 5 includes:

[0092] The generated heat map is combined with the isotope data for analysis. The heat map provides spatial distribution information, and the isotope data provides emission source characteristic identification. The isotope fingerprint characteristics are associated with the emission source type, and the spatial resolution of the heat map and the density of the isotope sampling points are synchronously calibrated to ensure the spatiotemporal matching of the data. A multi-dimensional emission source characteristic database is formed. A Bayesian hybrid model is used to integrate the heat map contribution ratio and isotope concentration data in real time, and the instantaneous contribution intensity of each emission source is reversely calculated. The high-frequency pulses and low-frequency periods in the time series are extracted by wavelet analysis to obtain the emission pulse characteristics and mark the spatiotemporal coordinates of abnormal emission events. When an abnormal emission event is detected, a high-precision resampling verification mechanism is triggered. Based on the historical emission pulse characteristics, the heat map update frequency and isotope sampling strategy are dynamically adjusted, and the LSTM neural network is used to predict the emission trend in the next 24 hours.

[0093] The process of using LSTM neural network to predict the emission trend in the next 24 hours is as follows:

[0094] Collect historical emission data, including the contribution ratio of emission sources, isotope concentration data and time series data in the thermal map, organize the data, extract historical emission pulse characteristics, use wavelet analysis methods to extract high-frequency pulse characteristics and low-frequency periodic characteristics in historical emission data, analyze the emission pulse characteristics of different regions, identify areas and time periods with frequent emission activities, and then store the extracted emission pulse characteristics in the feature database, including the spatiotemporal coordinates and pulse characteristic parameters of each sampling point. According to the historical emission pulse characteristics, the target area is divided into sub-areas with different priorities. The high-priority area is the area with frequent emission pulses and large amplitudes; the low-priority area is the area with less emission activities. According to the regional priority, the update frequency of the thermal map is dynamically adjusted. For high-priority areas, the update frequency of the thermal map is increased to update every 15 minutes, and for low-priority areas, the update frequency of the thermal map is reduced to update every 1 hour. At the same time, the isotope sampling strategy was adjusted. In high-priority areas, the isotope sampling density was increased, the number of sampling points was increased, or the sampling interval was shortened. In low-priority areas, the original sampling density was maintained or the number of sampling points was appropriately reduced. A dynamic adjustment strategy was then implemented, and the heat map update and isotope sampling process were monitored in real time. Based on real-time data feedback, the adjustment strategy was further optimized to ensure the effective use of resources and the accuracy of the data. The LSTM neural network model was trained using historical emission data including time series data, heat map contribution ratio, and isotope concentration. The model parameters were optimized to ensure that the model could accurately capture the dynamic characteristics in the time series. The latest emission data was input into the trained LSTM neural network model to predict the emission trend in the next 24 hours. The prediction results included the emission concentration and emission source contribution ratio of each sampling point. The prediction results were then visualized to generate a heat map of the emission trend in the next 24 hours.

[0095] The specific workflow of step 5 is as follows:

[0096] The generated heat map (providing spatial distribution information) is combined with the isotope data (providing emission source characteristic identification) for analysis to ensure that the spatial resolution of the heat map matches the density of the isotope sampling points so that the data can accurately correspond in time and space. According to the fingerprint characteristics in the isotope data, the characteristic identification of each emission source is clarified, and the isotope fingerprint characteristics are associated with the emission source type to form a one-to-one correspondence. The spatial distribution information of the heat map and the emission source characteristic identification in the isotope data are integrated into a database. The database contains multi-dimensional information such as the geographical location, emission source type, isotope fingerprint characteristics, contribution ratio, etc. of each sampling point; the Bayesian hybrid model is used to fuse the contribution ratio data and isotope concentration data in the heat map in real time to reversely infer the emission sources. The instantaneous contribution intensity of the emission source is analyzed, and wavelet analysis is performed on the time series data of the emission source to extract high-frequency pulse and low-frequency periodic features, obtain emission pulse features, mark the spatiotemporal coordinates of abnormal emission events, and trigger a high-precision resampling verification mechanism when an abnormal emission event is detected, and then perform high-precision sampling on the spatiotemporal coordinates of the abnormal event to verify the authenticity of the abnormal event; based on the historical emission pulse features, the update frequency of the heat map and the isotope sampling strategy are dynamically adjusted, and the LSTM neural network is used to analyze the time series data of the emission source to predict the emission trend in the next 24 hours. The prediction results are visualized to show the spatial distribution of the emission trend in the next 24 hours, and a time dimension switching function is provided in the visualization interface to facilitate users to view the prediction results of different time periods;

[0097] Step 6 includes:

[0098] The system monitors emission data in real time and automatically compares the current emission indicators of each emission source with the preset warning threshold. When the monitoring value exceeds the warning threshold, the warning signal is immediately triggered and the management personnel are notified through the visual interface, SMS or email to ensure timely response. According to the warning signal, the reasons for the exceedance are analyzed and the corresponding control strategies are matched from the knowledge base to generate targeted management measures. The management personnel adjust agricultural activities accordingly, such as optimizing the amount of fertilizer, changing the irrigation method or adjusting the planting pattern to reduce emission risks and improve carbon sequestration and emission reduction effects. The management personnel implement control interventions according to the adjusted measures and continuously monitor the intervention effects, compare the data before and after the intervention, evaluate the effectiveness of the measures, provide reference for subsequent management, and ensure that agricultural activities meet the carbon sequestration and emission reduction goals;

[0099] The specific workflow of step 6 is as follows:

[0100] The system receives and processes emission monitoring data in real time, including greenhouse gas concentrations, emission source contribution ratios, etc. The data sources include online monitoring equipment and AI algorithm analysis results, and compares the current emission indicators of each emission source with the preset warning thresholds in real time. The preset warning thresholds are set according to emission standards and management objectives. When the monitoring value exceeds the warning threshold range, the warning signal is triggered immediately. The warning signal notifies management personnel in a variety of ways, including: visual interface, SMS notification and email notification. Among them, for the visual interface, the warning information is displayed in the system interface, and the exceeding areas and indicators are highlighted. For SMS notification, a text message is sent to the management personnel, including warning information, exceeding indicators and recommended measures. For email notification, an email is sent to the management personnel, detailing the warning situation and recommended measures, recording detailed information for each warning, and generating a warning log; according to the warning signal, the specific reasons for exceeding the standard are analyzed, including: excessive fertilization, improper irrigation or improper soil management, etc. Excessive fertilization can easily lead to Increased emissions and improper irrigation can easily lead to Increased emissions and improper soil management can easily lead to Emissions increase, and then the management and control strategies corresponding to the causes of exceeding the standard are matched from the knowledge base. The knowledge base stores response measures and suggestions for different causes of exceeding the standard. Based on the matching results, targeted management measures are generated, such as optimizing the amount of fertilizer, reducing the amount of nitrogen fertilizer applied, increasing the proportion of organic fertilizer, changing the irrigation method, using drip irrigation or sprinkler irrigation, reducing flooding time, adjusting the planting pattern, using rotation or intercropping, and improving soil structure. The management measures are notified to management personnel through a visual interface, text message or email, etc.; management personnel adjust agricultural activities according to the recommended measures, continuously monitor emission data after intervention, record the effect of intervention, compare data before and after intervention, and evaluate the effectiveness of measures. The effectiveness of intervention measures is evaluated based on monitoring data. For example, a decrease in emission concentration indicates that the measures are effective, and no change or increase in emission concentration indicates that the measures are ineffective and need further adjustment. The evaluation results are fed back to the knowledge base to optimize future management and control strategies. The intervention effect evaluation results are used as a reference for subsequent management, and agricultural activities are continuously optimized to ensure that agricultural activities meet carbon sequestration and emission reduction goals and reduce greenhouse gas emissions.

[0101] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A soil carbon sequestration and emission reduction detection method combined with isotope tracing technology, characterized in that: The following steps are involved: Step 1: Build an online coupling system consisting of a front-end column chromatography separation module, a mid-end isotope ratio mass spectrometer, and a back-end AI algorithm; Step 2: Perform front-end sampling and column chromatography separation in the target area to obtain classified gas components, and perform mid-end isotope ratio mass spectrometry detection to obtain gas isotope characteristic data; Step 3: The back-end AI algorithm receives mass spectrometer data in real time, analyzes gas concentrations and isotope fingerprint characteristics, and identifies emission source types and contribution ratios; Step 4: Based on the analysis results, the AI ​​algorithm generates a heat map of soil greenhouse gas emission source contributions, showing the distribution and intensity of emission sources in different regions; Step 5: Combine heat maps with isotope data to trace the source of greenhouse gas emissions, conduct continuous dynamic monitoring, and capture short-term emission pulses and daily / seasonal fluctuations; In step 6, the system automatically matches the preset warning threshold, triggers the warning signal, adjusts agricultural management measures, and conducts control intervention.

2. The soil carbon sequestration and emission reduction detection method combined with isotope tracing technology according to claim 1, characterized in that: The step 1 comprises: Select wheat and corn planting plots where soil carbon sequestration and emission reduction testing is required as target areas, set monitoring targets, and then select column chromatography separation modules, isotope ratio mass spectrometers, and AI algorithms as needed; Based on the selected column chromatography separation module, isotope ratio mass spectrometer and AI algorithm, an online coupling system suitable for the target area is built, including a front-end column chromatography separation module, a mid-end isotope ratio mass spectrometer and a back-end AI algorithm. The column chromatography separation conditions are optimized, and the isotope ratio mass spectrometer parameters are adjusted to achieve the best detection sensitivity. At the same time, the back-end AI algorithm is initialized and configured.

3. The soil carbon sequestration and emission reduction detection method combined with isotope tracing technology according to claim 1, characterized in that: The step 2 includes: Gas sampling devices are deployed in a 10m x 10m grid in the target area. The sampling devices introduce gas samples into the column chromatography separation module through the gas path system. After receiving the gas sample, the column chromatography separation module separates the gas according to its physical and chemical properties. By optimizing the column temperature and carrier gas flow rate, different greenhouse gas components are effectively separated, and the separated gas components are then sent to the isotope ratio mass spectrometer in sequence. The isotope ratio mass spectrometer receives the separated gas components and uses a dual-inlet method to simultaneously detect isotope characteristic data. The conventional inlet introduces sample gas, and the high-concentration inlet injects standard gas with a known isotope ratio for mass discrimination correction. At the same time, the chromatographic separation time axis is aligned through the dynamic time warping algorithm to generate a concentration-isotope data stream, and then push the data to the back-end AI algorithm in real time.

4. The soil carbon sequestration and emission reduction detection method combined with isotope tracing technology according to claim 3 is characterized by: The step 3 comprises: The backend AI algorithm receives the concentration-isotope data stream pushed by the mass spectrometer in real time via the MQTT protocol, analyzes the timestamp, gas composition, and isotope ratio, segments the data using a sliding window method with a window length of 15 minutes, applies a Kalman filter to smooth noise, fills missing values ​​using the KNN algorithm, and normalizes the data to convert it into a unified format. The backend AI algorithm uses a pre-trained LSTM-Random Forest hybrid model to analyze the standardized data, extracting the dynamic characteristics and isotope fingerprint characteristics of the gas concentration time series. It then compares them with a known isotope signature database to identify the emission source types of different greenhouse gases. Based on the identified emission source types, the contribution ratio of each emission source to greenhouse gas concentration is further analyzed to quantify the relative contribution of each emission source.

5. The soil carbon sequestration and emission reduction detection method combined with isotope tracing technology according to claim 4 is characterized by: The analysis process of emission source types and contribution ratios is as follows: Use a pre-trained LSTM model to analyze the standardized gas concentration time series data, capturing the dynamic features in the time series data and extracting the pattern of gas concentration changes over time. At the same time, it also extracts the isotope fingerprint features, that is, the characteristics of different isotope ratios in the gas. The extracted isotope fingerprint characteristics are compared with the known isotope characteristic database to determine the characteristic identification of each emission source, that is, the unique isotope fingerprint of each emission source. Based on the comparison results, the random forest model is used to classify and identify the emission source type, and the identification results are output to clarify the type of greenhouse gas emission source at each sampling point; Based on the identified emission source types, we further analyzed the contribution ratio of each emission source to greenhouse gas concentrations. We used the output probability of the random forest model to quantify the relative contribution of each emission source. Combined with the gas concentration time series data, we analyzed the changes in the contribution of different emission sources in different time periods.

6. The soil carbon sequestration and emission reduction detection method combined with isotope tracing technology according to claim 1, characterized in that: The step 4 comprises: The AI ​​algorithm aggregates the analyzed emission source contribution data into 10m×10m grids, calculates the mean contribution ratio of each emission source within each grid, and uses the Kriging interpolation algorithm, combined with spatial autocorrelation characteristics, to convert the discrete grid data into a continuous spatial distribution field, generating a basic contribution density map covering the entire target area. Based on the interpolation results, thematic contribution maps were generated for each emission source type and merged into a comprehensive heat map using a weighted overlay algorithm. The contribution intensity was divided into five levels: low / medium-low / medium / medium-high / high using the natural breakpoint method. These maps were visualized using a blue→yellow→red color gradient, and the dominant emission source type in each region was labeled. The system receives new data every 15 minutes and triggers incremental updates, recalculating only the grid areas affected by the new data, refreshing the heat map through GPU-accelerated rendering, and ultimately outputting a visual interface that supports interactive operations.

7. The soil carbon sequestration and emission reduction detection method combined with isotope tracing technology according to claim 6, characterized in that: The generation process of the basic contribution density map is: The emission source contribution data analyzed by the AI ​​algorithm are aggregated according to 10m×10m grids. The data of all sampling points in each grid are summarized. For each grid, the mean contribution ratio of each emission source is calculated. The mean contribution ratio of the emission source calculated in each grid is stored as grid contribution data. Each grid corresponds to a set of mean contribution ratios of emission sources, forming a discrete grid data set. The Kriging interpolation algorithm is used to perform spatial interpolation on the grid contribution data in the discrete grid data set, estimate the value of the unknown point based on the value of the known point, and analyze the spatial autocorrelation of the grid contribution data to determine the parameters of the Kriging interpolation; According to the Kriging interpolation results, the discrete grid data is converted into a continuous spatial distribution field, and the continuous spatial distribution field generated by Kriging interpolation is plotted into a basic contribution density map covering the entire target area. The generated basic contribution density map is verified by comparing it with the known emission source distribution information, and then the Kriging interpolation parameters or AI algorithm are optimized according to the verification results.

8. The soil carbon sequestration and emission reduction detection method combined with isotope tracing technology according to claim 1, characterized in that: The step 5 comprises: The generated heat map is combined with isotope data for analysis. The heat map provides spatial distribution information, and the isotope data provides emission source characteristic identification. The isotope fingerprint characteristics are associated with the emission source type, and the spatial resolution of the heat map and the isotope sampling point density are simultaneously calibrated to form a multi-dimensional emission source characteristic database. A Bayesian hybrid model is used to integrate the contribution ratio of the heat map with isotope concentration data in real time, reversely inferring the instantaneous contribution intensity of each emission source. Wavelet analysis is then used to extract high-frequency pulses and low-frequency periods in the time series, obtain emission pulse characteristics, and mark the spatiotemporal coordinates of abnormal emission events. When an abnormal emission event is detected, a high-precision resampling verification mechanism is triggered. Based on the historical emission pulse characteristics, the heat map update frequency and isotope sampling strategy are dynamically adjusted, and the LSTM neural network is used to predict the emission trend in the next 24 hours.

9. The soil carbon sequestration and emission reduction detection method combined with isotope tracing technology according to claim 8, characterized in that: The process of using LSTM neural network to predict the emission trend in the next 24 hours is as follows: Collect historical emission data, including emission source contribution ratios in thermal maps, isotope concentration data, and time series data; organize the data; extract historical emission pulse features; use wavelet analysis to extract high-frequency pulse features and low-frequency periodic features from historical emission data; analyze emission pulse features in different regions; identify areas and time periods with frequent emission activities; and then store the extracted emission pulse features in a feature database; Based on the historical emission pulse characteristics, the target area is divided into sub-areas of different priorities. High-priority areas are areas with frequent and large emission pulses; low-priority areas are areas with less emission activities. Based on the regional priority, the update frequency of the heat map is dynamically adjusted, and the isotope sampling strategy is adjusted at the same time. Then, a dynamic adjustment strategy is implemented, and the heat map update and isotope sampling processes are monitored in real time. Based on real-time data feedback, the adjustment strategy is further optimized. Use historical emission data to train the LSTM neural network model, optimize model parameters, and input the latest emission data into the trained LSTM neural network model to predict the emission trend for the next 24 hours. The prediction results include the emission concentration and emission source contribution ratio of each sampling point. The prediction results are then visualized to generate a heat map of the emission trend for the next 24 hours.

10. The soil carbon sequestration and emission reduction detection method combined with isotope tracing technology according to claim 9, characterized in that: The step 6 comprises: The system monitors emission data in real time and automatically compares the current emission indicators of each emission source with the preset warning threshold. When the monitored value exceeds the warning threshold, the warning signal is immediately triggered and the management personnel are notified through the visual interface, SMS or email. Based on the early warning signals, the causes of exceeding the standards are analyzed and corresponding control strategies are matched from the knowledge base to generate targeted management measures and recommendations. Managers can then adjust agricultural activities accordingly. Managers implement control interventions in accordance with the adjusted measures, and continuously monitor the effects of the interventions, compare data before and after the interventions, evaluate the effectiveness of the measures, and provide reference for subsequent management.

Images