Greenhouse gas emission calculation method, device and electronic equipment

Abstract

The embodiments of the present invention provide a method, device and electronic device for calculating greenhouse gas emissions, and relate to the technical field of greenhouse gas emission accounting. The method for calculating greenhouse gas emissions includes: determining an accounting data set based on a greenhouse gas database, multiple in-situ observation data of gas concentrations and multiple in-situ observation data of gas fluxes; establishing a random forest model of prediction factors and greenhouse gas concentrations; predicting greenhouse gas concentrations of multiple reservoirs based on the random forest model; calculating gas exchange coefficients; determining predicted gas exchange coefficients through the random forest model; determining the greenhouse gas diffusion flux and total emissions of the reservoir based on the greenhouse gas concentrations and the predicted gas exchange coefficients; determining the greenhouse gas footprint before water storage based on the accounting data set; and determining the net greenhouse gas emissions of each reservoir based on the total emissions after water storage and the greenhouse gas footprint before water storage. Through the solution of the present invention, the accuracy of calculating the net greenhouse gas emissions of the reservoir is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of greenhouse gas emission accounting, and in particular to a method, device and electronic equipment for calculating greenhouse gas emissions. Background Art

[0002] Carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O) are three important greenhouse gases, and rising atmospheric concentrations are a key factor contributing to global warming. Damming and impounding water inevitably results in a certain degree of land inundation. The retention of organic matter in soil and vegetation makes reservoirs important sites for carbon burial and transformation.

[0003] Currently, existing greenhouse gas emission accounting methods have the following key flaws: Most studies focus on greenhouse gas emissions from individual reservoirs, or simply extrapolate based on the average of three greenhouse gases. This method relies heavily on the balance of the original data. However, the collected literature data is not evenly distributed, and most studies may be concentrated in certain regions. Therefore, the reservoir emissions calculated using the average extrapolation method may have large deviations. Furthermore, most research on reservoir greenhouse gases focuses on large reservoirs, but the proportion of small reservoirs is far greater than that of large reservoirs. Linear regression analysis of only a single environmental factor for large reservoirs cannot reasonably explain the main influencing factors and differences in the reservoir greenhouse gas emission process.

[0004] Therefore, in order to improve the accuracy of greenhouse gas emission accounting for reservoirs, how to propose a method for calculating greenhouse gas emissions for reservoirs has become an urgent problem to be solved. Summary of the Invention

[0005] The purpose of the technical solution of the present invention is to provide a method, device and electronic equipment for calculating greenhouse gas emissions, which can solve the technical problem in the existing technology that when calculating the greenhouse gas emissions of reservoirs, only linear regression analysis of a single environmental factor is performed on large reservoirs, which affects the accuracy of the calculation of the greenhouse gas emissions of reservoirs.

[0006] In view of this, the technical solution of the first aspect of the present invention provides a method for calculating greenhouse gas emissions.

[0007] The technical solution of the second aspect of the present invention provides a device for calculating greenhouse gas emissions.

[0008] The technical solution of the third aspect of the present invention provides an electronic device.

[0009] In order to achieve the above-mentioned purpose, the technical solution of the first aspect of the present invention provides a method for calculating greenhouse gas emissions, which includes: obtaining a greenhouse gas database, in-situ observation data of gas concentrations corresponding to greenhouse gas concentrations of multiple reservoirs, and in-situ observation data of gas fluxes corresponding to multiple reservoirs; determining a calculation data set based on the greenhouse gas database, multiple in-situ observation data of gas concentrations, and multiple in-situ observation data of gas fluxes; establishing a random forest model of prediction factors and greenhouse gas concentrations based on the calculation data set in combination with a random forest algorithm; and predicting greenhouse gas concentrations of multiple reservoirs based on the random forest model. degree; calculate the gas exchange coefficient when the Schmidt constant is equal to 600 under standard conditions based on the thin film boundary layer theory; predict the gas exchange coefficient using the random forest model to determine the predicted gas exchange coefficient; determine the reservoir greenhouse gas diffusion flux based on the greenhouse gas concentration and the predicted gas exchange coefficient; determine the reservoir greenhouse gas emissions based on the reservoir greenhouse gas diffusion flux and the area of ​​the reservoir, and determine the total emissions corresponding to multiple reservoirs based on the reservoir greenhouse gas emissions; determine the greenhouse gas footprint of each reservoir before water storage based on the accounting data set; and determine the net greenhouse gas emissions of each reservoir based on the total emissions after water storage and the greenhouse gas footprint before water storage.

[0010] According to the greenhouse gas emission accounting method provided by the present invention, the net greenhouse gas emissions corresponding to multiple reservoirs are determined by calculating the greenhouse gas concentrations and greenhouse gas fluxes of the reservoirs. Greenhouse gases include carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O). Specifically, a multidimensional database is constructed using in situ observation data on reservoir greenhouse gas concentrations and fluxes from relevant literature, wherein the greenhouse gas databases include GRanD and HydroLAKES. After determining the multidimensional database, prediction factors are determined based on the publicly available reservoir dam information dataset. The prediction factors include reservoir characteristic factors, climate characteristic factors, watershed environmental factors, and socioeconomic factors. Based on the prediction factors and the multidimensional database, an accounting dataset is determined. A random forest model is constructed for the gas concentrations of CO2, CH4, and N2O based on the data in the accounting dataset, and the parameters of the random forest model are optimized and trained. The gas exchange coefficient under standard conditions when the Schmidt constant is equal to 600 is calculated based on the thin film boundary layer theory. The gas exchange coefficient under standard conditions when the Schmidt constant is equal to 600 is predicted using a random forest model to determine the predicted gas exchange coefficient. Finally, the greenhouse gas diffusion flux corresponding to each reservoir is determined based on the predicted gas exchange coefficient and the greenhouse gas concentration of the reservoir predicted by the random forest model. The greenhouse gas emissions determined based on the predicted gas exchange coefficient and the greenhouse gas concentration of the reservoir predicted by the random forest model are the total greenhouse gas emissions after the reservoir is filled with water. The greenhouse gas footprint of multiple reservoirs before filling is determined based on the surface area and greenhouse gas emission coefficient of the land cover subunit corresponding to each reservoir. The net greenhouse gas emissions of each reservoir, i.e., the net flux, are determined based on the total emissions after the reservoir is filled with water and the greenhouse gas footprint of the reservoir before filling with water. Among them, the net emission flux before and after the reservoir construction is defined as the average value of the total emissions per unit area of ​​each reservoir after removing the greenhouse gas footprint before filling with water.

[0011] In some technical solutions, optionally, the accounting data set includes a multidimensional feature matrix, which includes reservoir characteristic factors, climate characteristic factors, watershed environmental factors and socio-economic factors, and the reservoir characteristic factors, climate characteristic factors, watershed environmental factors and socio-economic factors are prediction factors.

[0012] In this approach, a system of predictive factors influencing reservoir greenhouse gas emissions is constructed by integrating multidimensional data, thereby determining a multidimensional feature matrix. Specifically, characteristic parameters corresponding to multiple reservoirs are determined based on publicly available reservoir and dam information datasets. An open-source Geographic Information System (GIS) is then acquired, and a watershed environmental factor assimilation platform is constructed using open-source GIS technology to obtain environmental characteristic information from multiple different watersheds. Based on the environmental characteristic information from multiple different watersheds and the characteristic parameters corresponding to multiple reservoirs, reservoir characteristic factors, climate characteristic factors, watershed environmental factors, and socioeconomic factors are determined. These reservoir characteristic factors, climate characteristic factors, watershed environmental factors, and socioeconomic factors are used as predictors, which are structured and stored to form a multidimensional feature matrix that is input into the machine learning model. Each row in the multidimensional feature matrix represents a reservoir sample, and each column represents a characteristic factor.

[0013] In some technical solutions, optionally, a random forest model of prediction factors and greenhouse gas concentrations is established based on the accounting data set in combination with a random forest algorithm, including: constructing an initial model for greenhouse gas concentrations based on the accounting data set; dividing multiple data in the accounting data set into intervals to determine a first interval and a second interval, with the first interval being larger than the second interval; determining that the data in the first interval in the accounting data set is valid data; dividing the valid data into a training set and a test set, with the number of data in the training set being larger than the number of data in the test set; training the initial model based on the training set and the prediction factors to determine the random forest model; wherein, the first model result of the random forest model is determined by constructing a parameter network and combining cross-validation; the test set is characterized by the coefficient of determination and the root mean square error, and the characterization result is used to determine the prediction performance of the random forest model.

[0014] In this solution, a random forest model is constructed for the gas concentrations of CO2, CH4 and N2O based on the R language, and model training, hyperparameter adjustment and performance evaluation are performed. Specifically, in the process of model building, in order to avoid the influence of extreme values, data outside the first interval are eliminated. Multiple groups of data in the first interval are taken as valid data, and the valid data are randomly divided. The valid data are randomly allocated according to 75% training set and 25% test set, ensuring that the number of data in the training set is significantly larger than the number of data in the test set to meet the sample size requirements of machine learning. All predictors are constructed into a multidimensional feature matrix, which is input into the initial model for training to determine the random forest model. The optimal parameters in the random forest model, that is, the selection of the first model result, is achieved by constructing a parameter grid and combining it with cross-validation. Constructing the parameter grid includes setting the number of decision trees, the depth of the tree and the minimum number of leaf node samples. The performance of different parameter combinations is evaluated on the training set by cross-validation, and the optimal parameters are selected as the output result of the random forest model.

[0015] In some technical solutions, optionally, the greenhouse gas footprint of each reservoir before water storage is determined based on the accounting data set, including: determining the land cover subunit of each reservoir; determining the surface area of ​​the land cover subunit; determining the greenhouse gas emission coefficient of each reservoir before water storage based on the accounting data set; and determining the greenhouse gas footprint of the reservoir before water storage based on the greenhouse gas emission coefficient and the surface area.

[0016] In this scenario, the pre-filling greenhouse gas footprint is assessed by multiplying the surface area of ​​each land cover subunit by a greenhouse gas emission factor. For channel-type reservoirs, the pre-filling geomorphology consists of submerged land and the original channel; for non-channel-type reservoirs, the pre-filling geomorphology is considered to consist entirely of submerged land. The pre-filling area is divided into multiple land cover subunits based on land cover types, such as forest, cropland, wetland, grassland, and natural river channels. The length of the channel submerged by the reservoir is estimated by the overlap between the reservoir polygon in the China Reservoir Dataset (CRD) and the river network in the Natural Runoff Simulation Dataset (GRADES). The width of the channel submerged by the reservoir is estimated by combining the hydraulic geometry of downstream stations with the flow of the river segment in GRADES. The surface area of ​​the submerged channel is determined by combining the length and width of the channel submerged by the reservoir.

[0017] In some technical solutions, optionally, after determining the net greenhouse gas emissions of each reservoir based on the total emissions after water storage and the greenhouse gas footprint before water storage, it also includes: determining the reservoir area and soil climate type of each reservoir; determining the area of ​​each soil climate type based on the soil climate type and the reservoir area; determining the greenhouse gas emission coefficient corresponding to the soil climate type; and determining the greenhouse gas balance based on the reservoir area, the area of ​​the soil climate type and the greenhouse gas emission coefficient corresponding to the soil climate type.

[0018] In this scenario, the actual water surface area after reservoir construction and the total area of ​​the surrounding impact zone are obtained. The reservoir area of ​​each reservoir is determined using GIS data, and the reservoir area is divided into multiple subunits based on soil climate type and climate zone. The greenhouse gas balance is determined based on the reservoir area, the area of ​​the soil climate type, and the greenhouse gas emission coefficient corresponding to the soil climate type.

[0019] In some technical solutions, optionally, the accounting data set further includes: geospatial coordinates, temporal metadata, and temporal resolution of the sampling points corresponding to each reservoir.

[0020] In this solution, the data in the accounting dataset also includes temporal dimension information, namely the geospatial coordinates, temporal metadata, and temporal resolution of each reservoir sampling point. The geospatial coordinates are the latitude and longitude coordinates of each reservoir sampling point, which are used to map the reservoir sampling points to geographic space and support overlay analysis with raster data such as climate, soil, and land use rate. The temporal metadata includes the monitoring date and duration corresponding to the sampling, which is used to track seasonal changes in greenhouse gas concentrations or dynamic responses after water level regulation. The temporal resolution includes the time interval of data sampling, such as hourly, daily, monthly, quarterly, or annually, which is used to match the time scale of data from different sources.

[0021] The technical solution of the second aspect of the present invention provides a greenhouse gas emission accounting device, which includes: a data unit for obtaining a greenhouse gas database, in-situ observation data of gas concentrations corresponding to greenhouse gas concentrations of multiple reservoirs, and in-situ observation data of gas fluxes corresponding to multiple reservoirs; determining an accounting data set based on the greenhouse gas database, multiple gas concentration in-situ observation data, and multiple gas flux in-situ observation data; a model unit for establishing a random forest model of prediction factors and greenhouse gas concentrations based on the accounting data set in combination with a random forest algorithm; a prediction unit for predicting the greenhouse gas concentrations of multiple reservoirs based on the random forest model. ; Calculate the gas exchange coefficient when the Schmidt constant is equal to 600 under standard conditions based on the thin film boundary layer theory; Predict the gas exchange coefficient through the random forest model to determine the predicted gas exchange coefficient; Determine a unit for determining the greenhouse gas diffusion flux of the reservoir based on the greenhouse gas concentration and the predicted gas exchange coefficient; Determine the greenhouse gas emissions of the reservoir based on the greenhouse gas diffusion flux of the reservoir and the area of ​​the reservoir, and determine the total emissions corresponding to multiple reservoirs based on the greenhouse gas emissions of the reservoir; Determine the greenhouse gas footprint of each reservoir before water storage based on the accounting data set; Accounting unit for determining the net greenhouse gas emissions of each reservoir based on the total emissions after water storage and the greenhouse gas footprint before water storage.

[0022] In some technical solutions, optionally, the model unit is also used to: construct an initial model for greenhouse gas concentration based on the accounting data set; divide multiple data in the accounting data set into intervals to determine a first interval and a second interval, and the first interval is larger than the second interval; determine that the data in the first interval in the accounting data set is valid data; divide the valid data into a training set and a test set, and the number of data in the training set is larger than the number of data in the test set; train the initial model based on the training set and the prediction factor to determine the random forest model; wherein, the first model result of the random forest model is determined by constructing a parameter network and combining cross-validation; the test set is characterized by the determination coefficient and the root mean square error, and the characterization result is used to determine the prediction performance of the random forest model.

[0023] In some technical solutions, optionally, the accounting unit is also used to: determine the reservoir area and soil climate type of each reservoir; determine the area of ​​each soil climate type based on the soil climate type and the reservoir area; determine the greenhouse gas emission coefficient corresponding to the soil climate type; determine the greenhouse gas balance based on the reservoir area, the soil climate type area and the greenhouse gas emission coefficient corresponding to the soil climate type.

[0024] The technical solution of the third aspect of the present application provides an electronic device, including a processor, a memory, and a program or instruction stored in the memory and executable on the processor. When the program or instruction is executed by the processor, the steps of the greenhouse gas emission accounting method in the first aspect are implemented.

[0025] Additional aspects and advantages of the technical solutions of the present invention will become apparent in the following description or will be understood through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 A schematic diagram showing a flow chart of a method for calculating greenhouse gas emissions according to an embodiment of the present application is shown;

[0027] Figure 2 A partial flow chart of a method for calculating greenhouse gas emissions according to an embodiment of the present application is shown;

[0028] Figure 3 A partial flow chart of a method for calculating greenhouse gas emissions according to an embodiment of the present application is shown;

[0029] Figure 4 A partial flow chart of a method for calculating greenhouse gas emissions according to an embodiment of the present application is shown;

[0030] Figure 5 A schematic block diagram of the structure of a device for calculating greenhouse gas emissions according to an embodiment of the present application is shown;

[0031] Figure 6 A schematic structural block diagram of an electronic device according to an embodiment of the present application is shown;

[0032] Figure 7 A diagram showing the locations of sampling points for collecting greenhouse gas concentrations in a reservoir according to one embodiment of the present application is shown;

[0033] Figure 8 A diagram showing the locations of sampling points for collecting greenhouse gas diffusion fluxes from a reservoir according to one embodiment of the present application is shown;

[0034] Figure 9 shows a spatial distribution diagram of carbon dioxide gas concentration according to one embodiment of the present application;

[0035] Figure 10 shows a spatial distribution diagram of the diffusion flux of carbon dioxide according to one embodiment of the present application;

[0036] Figure 11 shows a spatial distribution diagram of methane gas concentration according to one embodiment of the present application;

[0037] Figure 12 shows a spatial distribution diagram of the diffusion flux of methane according to one embodiment of the present application;

[0038] Figure 13 shows a spatial distribution diagram of nitrous oxide gas concentration according to one embodiment of the present application;

[0039] Figure 14 FIG. 4 shows a spatial distribution diagram of the diffusion flux of nitrous oxide according to an embodiment of the present application.

[0040] in, Figure 5 and Figure 6 The corresponding relationship between the reference numerals and component names is as follows:

[0041] 900: greenhouse gas emission accounting device; 902: data unit; 904: model unit; 906: prediction unit; 908: determination unit; 910: accounting unit; 1000: electronic device; 1109: memory; 1110: processor. DETAILED DESCRIPTION

[0042] In order to more clearly understand the above-mentioned purposes, features and advantages of the embodiments of the present invention, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings and specific implementation methods. It should be noted that the embodiments of the present application and the features therein can be combined with each other in the absence of conflict.

[0043] In the following description, many specific details are set forth to facilitate a full understanding of the present application. However, the embodiments of the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present application is not limited to the specific embodiments disclosed below.

[0044] The following is combined with Figures 1 to 14 , the greenhouse gas emission calculation method, device and electronic device provided in the embodiments of the present application are described in detail through specific embodiments and their application scenarios.

[0045] This embodiment provides a method for calculating greenhouse gas emissions. Figure 1 As shown, including:

[0046] Step S100: Acquire a greenhouse gas database, in-situ observation data of greenhouse gas concentrations corresponding to a plurality of reservoirs, and in-situ observation data of greenhouse gas fluxes corresponding to a plurality of reservoirs;

[0047] Step S102: determining a calculation data set based on a greenhouse gas database, a plurality of gas concentration in-situ observation data, and a plurality of gas flux in-situ observation data;

[0048] Step S104: establishing a random forest model of prediction factors and greenhouse gas concentrations based on the calculation data set and the random forest algorithm;

[0049] Step S106: predicting greenhouse gas concentrations of multiple reservoirs based on the random forest model;

[0050] Step S108: Calculating the gas exchange coefficient when the Schmidt constant is equal to 600 under standard conditions according to the thin film boundary layer theory;

[0051] Step S110: predicting the gas exchange coefficient using a random forest model to determine the predicted gas exchange coefficient;

[0052] Step S112: determining the greenhouse gas diffusion flux of the reservoir according to the greenhouse gas concentration and the predicted gas exchange coefficient;

[0053] Step S114: determining the greenhouse gas emissions of the reservoir according to the greenhouse gas diffusion flux of the reservoir and the area of ​​the reservoir, and determining the total emissions corresponding to the multiple reservoirs according to the greenhouse gas emissions of the reservoir;

[0054] Step S116: determining the greenhouse gas footprint of each reservoir before water storage based on the accounting data set;

[0055] Step S118: Determine the net greenhouse gas emissions of each reservoir based on the total emissions after water storage and the greenhouse gas footprint before water storage.

[0056] According to the greenhouse gas emissions accounting method provided by the present invention, net greenhouse gas emissions corresponding to multiple reservoirs are determined by calculating reservoir greenhouse gas concentrations and greenhouse gas fluxes. Greenhouse gases include carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O). Specifically, a multidimensional database is constructed using in situ observational data on reservoir greenhouse gas concentrations and fluxes from the CO2 Measurements Database, the Global River Methane Database, and relevant literature. Greenhouse gas databases include GRanD and HydroLAKES. After determining the multidimensional database, prediction factors are determined based on publicly available reservoir and dam information datasets. The prediction factors include reservoir characteristics, climate characteristics, watershed environmental factors, and socioeconomic factors. Based on the prediction factors and the multidimensional database, an accounting dataset is determined. A random forest model is constructed for CO2, CH4, and N2O concentrations based on the data in the accounting dataset, and the random forest model is parameter optimized and trained. The gas exchange coefficients were calculated based on thin-film boundary layer theory for a Schmidt constant of 600 under standard conditions. These coefficients were then predicted using a random forest model to determine the predicted gas exchange coefficients. Finally, the greenhouse gas diffusion fluxes corresponding to each reservoir were determined based on the predicted gas exchange coefficients and the greenhouse gas concentrations predicted by the random forest model. The greenhouse gas emissions calculated based on the predicted gas exchange coefficients and the greenhouse gas concentrations predicted by the random forest model were the total greenhouse gas emissions after the reservoir was filled. The greenhouse gas footprints of multiple reservoirs before filling were determined based on the surface area and greenhouse gas emission coefficients of the land cover subunits corresponding to each reservoir. The net greenhouse gas emissions (net flux) of each reservoir were determined based on the total emissions after filling and the greenhouse gas footprint before filling. The net emission flux before and after reservoir construction was defined as the average per unit area of ​​the total emissions after removing the greenhouse gas footprint before filling.

[0057] Understandably, by integrating multi-source data, machine learning modeling, and a dual accounting benchmark, the net greenhouse gas emissions of reservoirs before and after construction are precisely quantified, extending data source coverage to a global scale. This addresses the sampling bias of existing international models, such as the Reservoir Greenhouse Gas Net Flux Assessment Model (G-res Tool), when applied to specific regions, and the low frequency of monitoring coverage for small reservoirs. By combining a random forest model with multi-dimensional predictors, the nonlinear mechanisms of greenhouse gas generation are captured, and greenhouse gas emissions are predicted from multiple perspectives. This avoids the uncertainty of greenhouse gas accounting caused by extrapolation based on average values ​​and improves the robustness of reservoir greenhouse gas emissions accounting.

[0058] Optionally, the multidimensional database further includes a global dam and reservoir database (Grand) and a global lake database (HydroLAKES).

[0059] Furthermore, through the dual accounting benchmark approach, the system automatically separates the greenhouse gas footprint of natural rivers and land emissions before water storage and the greenhouse gas emissions after water storage caused by human engineering, revealing the net climate benefits of hydropower and laying a scientific premise for evaluating the cleanliness of hydropower.

[0060] For example, the GHG footprint measures the net greenhouse gas emissions of a reservoir system over a period of time. For example, the pre-filling GHG footprint is the total amount of greenhouse gases naturally released or absorbed by the original flooded area (land or river) before the reservoir was built; the post-filling GHG footprint is the total amount of greenhouse gas emissions in the water and sediments caused by the reservoir after construction.

[0061] Optionally, when acquiring in situ observation data of greenhouse gas concentrations and fluxes in a reservoir, the geospatial coordinates, temporal metadata, and temporal resolution of the sampling points are simultaneously determined. For example, the geospatial coordinates include the longitude and latitude of the sampling points, the temporal metadata include the monitoring date and duration of the sampling, and the temporal resolution includes the time interval corresponding to the sampling, such as quarterly or annual.

[0062] Furthermore, the greenhouse gas emission accounting method is used to determine the net greenhouse gas flux of reservoirs within a specific regional scope, that is, the net greenhouse gas emissions of each reservoir within a specific regional scope.

[0063] Optionally, the greenhouse gas emission accounting method is applied to the accounting of greenhouse gas emissions from inland water bodies, such as lakes and rivers.

[0064] For example, the thin film boundary layer theory is used to describe the gas transfer rate at the water-air interface. In order to eliminate the influence of temperature, the gas exchange coefficient is standardized to the condition where the Schmidt constant is equal to 600, corresponding to a standard water temperature of 20°C. The gas exchange coefficient under the condition where the Schmidt constant is equal to 600 can be determined based on the greenhouse gas flux and greenhouse gas concentration of in situ observation data. Based on the random forest model, the relationship between the gas exchange coefficient and multiple prediction factors such as wind speed, reservoir area, and water temperature is established to determine the predicted gas exchange coefficient. The thin film boundary layer theory provides a physical basis for the calculation and prediction of the predicted gas exchange coefficient, ensuring the reliability and accuracy of the flux calculation.

[0065] In some embodiments, optionally, the accounting data set includes a multidimensional feature matrix, which includes reservoir characteristic factors, climate characteristic factors, watershed environmental factors and socioeconomic factors, and the reservoir characteristic factors, climate characteristic factors, watershed environmental factors and socioeconomic factors are prediction factors.

[0066] In this embodiment, by integrating multi-dimensional data, a prediction factor system affecting greenhouse gas emissions from reservoirs is constructed, and a multi-dimensional feature matrix is ​​determined. Specifically, characteristic parameters corresponding to multiple reservoirs are determined based on the publicly available reservoir dam information dataset. An open source Geographic Information System (GIS) is obtained, and a watershed environmental factor assimilation platform is constructed based on open source GIS technology to obtain environmental characteristic information of multiple different watersheds. Based on the environmental characteristic information of multiple different watersheds and the characteristic parameters corresponding to multiple reservoirs, reservoir characteristic factors, climate characteristic factors, watershed environmental factors, and socio-economic factors are determined. Reservoir characteristic factors, climate characteristic factors, watershed environmental factors, and socio-economic factors are used as prediction factors, and the prediction factors are structured and stored to form a multi-dimensional feature matrix that is input into the machine learning model. Each row in the multi-dimensional feature matrix represents a reservoir sample, and each column represents a characteristic factor.

[0067] It can be understood that by constructing a "machine learning-environmental factor" model framework through a multidimensional feature matrix, and using the random forest model to capture the nonlinear mechanism of greenhouse gas generation, the limitations of the traditional method based on average value extrapolation are compensated. The multidimensional feature matrix captures nonlinear relationships through the random forest algorithm, improves the scientific nature of the model, thereby improving the accuracy of reservoir greenhouse gas emissions accounting and reducing accounting deviations.

[0068] Optionally, before determining the multidimensional feature matrix, the following steps are also included: spatially aligning data from different sources (literature, database or GIS), unifying units and interpolating missing values.

[0069] For example, the reservoir characteristic factors include basic reservoir characteristics such as the reservoir age, the reservoir depth and the hydraulic retention time.

[0070] For example, climate characteristic factors include weather characteristics such as temperature, rainfall, and wind speed.

[0071] For example, watershed environmental factors include watershed environmental characteristics such as total primary productivity, soil organic carbon and proportion of landscape types.

[0072] For example, socioeconomic factors include human activity characteristics such as population density and gross domestic product (GDP).

[0073] In some embodiments, optionally, as Figure 2As shown, step S104: establishing a random forest model of prediction factors and greenhouse gas concentrations based on the accounting data set and the random forest algorithm, including:

[0074] Step S1040: constructing an initial model for greenhouse gas concentration based on the calculation data set;

[0075] Step S1042: Divide the multiple data in the accounting data set into intervals to determine a first interval and a second interval;

[0076] Step S1044: Determine that the data in the first interval of the accounting data set is valid data;

[0077] Step S1046: Divide the valid data into a training set and a test set;

[0078] Step S1048: training the initial model based on the training set and the prediction factors to determine the random forest model;

[0079] Among them, the first model result of the random forest model is determined by constructing a parameter network and combining it with cross-validation; the test set is characterized by the coefficient of determination and root mean square error, and the characterization results are used to determine the prediction performance of the random forest model.

[0080] In this embodiment, a random forest model is constructed for the gas concentrations of CO2, CH4 and N2O based on the R language, and model training, hyperparameter adjustment and performance evaluation are performed. Specifically, in the process of model building, in order to avoid the influence of extreme values, data outside the first interval are eliminated. Multiple groups of data in the first interval are used as valid data, and the valid data are randomly divided. The valid data are randomly allocated according to 75% training set and 25% test set, ensuring that the number of data in the training set is significantly larger than the number of data in the test set to meet the sample size requirements of machine learning. All predictors are constructed into a multidimensional feature matrix, which is input into the initial model for training to determine the random forest model. The optimal parameters in the random forest model, that is, the selection of the first model result is achieved by constructing a parameter grid and combining cross-validation. Constructing the parameter grid includes setting the number of decision trees, the depth of the tree and the minimum number of leaf node samples. The performance of different parameter combinations is evaluated on the training set by cross-validation, and the optimal parameters are selected as the output result of the random forest model.

[0081] Furthermore, the test set was used to calculate the coefficient of determination and root mean square error (RMSE). The coefficient of determination measures the fit between the model's predicted values ​​and the true values; a closer coefficient of determination is to 1, the stronger the model's explanatory power. The root mean square error quantifies the average deviation between the predicted values ​​and the true values; a smaller RMSE indicates a higher model's prediction accuracy. Using the test set to calculate the coefficient of determination and root mean square error independently validates the model's generalization ability and avoids overfitting.

[0082] Understandably, by removing outliers and optimizing hyperparameters, we eliminate interference from extreme values ​​in the data, allowing the model to focus on the patterns of greenhouse gas concentration changes under normal operating conditions, thereby reducing greenhouse gas concentration prediction errors. Grid search and cross-validation prevent bias from artificially set parameters and ensure the model's generalizability.

[0083] For example, the greenhouse gas concentration data are screened according to the 95% confidence interval (first interval), the top 5% of the maximum or minimum values ​​are eliminated, 95% of the valid data are retained, and outliers caused by instrument errors or extreme environmental events are deleted to avoid abnormal data interfering with model training and improve the robustness of the model.

[0084] Optionally, a random forest model includes a multidimensional feature matrix and a target variable, and the optimal combination is determined through grid search and cross-validation selection. This model captures the nonlinear interactions between multiple predictors, improving the accuracy and data reliability of gas exchange coefficient and greenhouse gas concentration predictions.

[0085] Optionally, a random forest model is constructed by selecting multiple pairs of data that simultaneously report greenhouse gas concentrations and greenhouse gas fluxes in the accounting dataset and combining the reservoir characteristic factors and watershed environmental factors in the prediction factors.

[0086] Furthermore, in the process of screening multiple data in the accounting data set, data that report both greenhouse gas concentrations and greenhouse gas fluxes are considered valid data.

[0087] In some embodiments, optionally, as Figure 3 As shown, step S116: determining the greenhouse gas footprint of each reservoir before water storage based on the accounting data set, including:

[0088] Step S1160: determining the land cover subunit of each reservoir;

[0089] Step S1162: determining the surface area of ​​the land cover sub-unit;

[0090] Step S1164: determining the greenhouse gas emission coefficient of each reservoir before water storage based on the accounting data set;

[0091] Step S1166: Determine the greenhouse gas footprint of the reservoir before water storage based on the greenhouse gas emission coefficient and the surface area.

[0092] In this example, the pre-filling greenhouse gas footprint is assessed by multiplying the surface area of ​​each land cover subunit by a greenhouse gas emission factor. For channel-type reservoirs, the pre-filling geomorphology consists of submerged land and the original channel; for non-channel-type reservoirs, the pre-filling geomorphology is considered to consist entirely of submerged land. The pre-filling area is divided into multiple land cover subunits based on land cover types, such as forest, farmland, wetland, grassland, and natural river channels. The length of the channel submerged by the reservoir is estimated by the overlap between the reservoir polygon in the China Reservoir Dataset (CRD) and the river network in the Natural Runoff Simulation Dataset (GRADES). The width of the channel submerged by the reservoir is estimated by combining the hydraulic geometry of downstream stations with the flow rate of the river segment in GRADES. The surface area of ​​the submerged channel is determined by combining the length and width of the channel submerged by the reservoir.

[0093] It is understandable that by calculating greenhouse gases before water storage, the natural emission baseline before reservoir construction can be clarified, the carbon emissions of the original ecosystem and the greenhouse gases artificially generated by reservoir construction can be separated, and the errors caused by unified accounting can be reduced.

[0094] For example, assuming that the land topography and land cover of the reservoir basin before the reservoir is filled are the same, the emission or absorption of each land cover subunit before the reservoir is filled is calculated based on the spatial distribution characteristics of greenhouse gas emission intensity of terrestrial ecosystems recorded in the literature or dataset or the published greenhouse gas emission coefficient of each land use type.

[0095] In some embodiments, optionally, as Figure 4 As shown, after step S118: determining the net greenhouse gas emissions of each reservoir based on the total emissions after water storage and the greenhouse gas footprint before water storage, the method further includes:

[0096] Step S1200: Determine the reservoir area and soil climate type of each reservoir;

[0097] Step S1202: determining the area of ​​each soil climate type according to the soil climate type and the reservoir area;

[0098] Step S1204: determining a greenhouse gas emission coefficient corresponding to the soil climate type;

[0099] Step S1206: Determine the greenhouse gas balance according to the reservoir area, the soil climate type area, and the greenhouse gas emission coefficient corresponding to the soil climate type.

[0100] In this example, the actual water surface area after the completion of the reservoir construction and the total area of ​​the surrounding impact area are obtained. The reservoir area of ​​each reservoir is determined using GIS data, and the reservoir area is divided into multiple subunits based on soil climate type and climate zone. The greenhouse gas balance is determined based on the reservoir area, soil climate type area, and the greenhouse gas emission coefficient corresponding to the soil climate type. The calculation formula for the greenhouse gas balance is as follows:

[0101] ;

[0102] in, is the greenhouse gas balance, i is the soil climate type, j is the greenhouse gas type, is the jth greenhouse gas emission coefficient of the i-th soil climate type, is the area of ​​soil climate type i, is the reservoir area.

[0103] It can be understood that by distinguishing different soil climate types, combining the areas of different soil climate types and reservoir areas to determine the greenhouse gas balance, the greenhouse gas balance is used as an evaluation indicator of the greenhouse gas emissions of the reservoir, and used to determine the carbon offset benefits of hydropower projects, laying the foundation for evaluating the cleanliness of hydropower in the reservoir.

[0104] For example, the net impact of reservoir construction on greenhouse gas emissions can be determined by calculating the greenhouse gas balance. For example, a positive greenhouse gas balance indicates that greenhouse gas emissions after reservoir impoundment are greater than the pre-impoundment baseline, indicating that the reservoir is a net emission source and requires optimized greenhouse gas emissions management. A negative greenhouse gas balance indicates that greenhouse gas emissions after reservoir impoundment are less than the pre-impoundment baseline, indicating that the reservoir has carbon sink potential, which can strengthen the clean energy status of hydropower in the reservoir.

[0105] In some embodiments, optionally, the accounting dataset further includes: geospatial coordinates, temporal metadata, and temporal resolution of the sampling points corresponding to each reservoir.

[0106] In this embodiment, the data in the accounting dataset also includes temporal dimension information, namely, the geospatial coordinates, temporal metadata, and temporal resolution of each reservoir sampling point. The geospatial coordinates are the latitude and longitude coordinates of each reservoir sampling point, which are used to map the reservoir sampling point to geographic space and support overlay analysis with raster data such as climate, soil, and land use rate. The temporal metadata includes the monitoring date and duration corresponding to the sampling, which is used to track seasonal changes in greenhouse gas concentrations or dynamic responses after water level regulation. The temporal resolution includes the time interval for data sampling, such as hourly, daily, monthly, quarterly, or annual, which is used to match the time scale of data from different sources.

[0107] Understandably, by determining the geospatial coordinates, temporal metadata, and temporal resolution of the sampling points of each reservoir, it is convenient to visualize the accounting data, and combined with the temporal metadata, it can reveal the seasonal patterns of greenhouse gas emissions and improve the traceability of the data.

[0108] In a specific embodiment, optionally, the greenhouse gas emission accounting method is a method of estimating the net emissions of multiple reservoirs within a specific area by establishing a greenhouse gas emission accounting model for reservoirs within the area, including the following steps:

[0109] Step 1: Database Construction. In situ observational data on reservoir greenhouse gas concentrations and fluxes from major domestic and international databases and 169 relevant publications were collected, yielding a total of 3,984 independent spatiotemporal data items, containing 7,342 individual greenhouse gas concentration and flux measurements. The sampling points' geospatial coordinates (latitude / longitude), temporal metadata (monitoring date and duration), and temporal resolution (e.g., quarterly versus annual) were also collected.

[0110] The sampling points for the reservoir greenhouse gas concentration (GHG concentration) are as follows: Figure 7 As shown in the figure, the horizontal axis represents longitude, the vertical axis represents latitude, the yellow circles represent the gas concentration sampling points corresponding to carbon dioxide CO2, a total of 1070; the red circles represent the gas concentration sampling points corresponding to methane CH4, a total of 1167; the blue circles represent the gas concentration sampling points corresponding to nitrous oxide N2O, a total of 437.

[0111] The sampling points of the reservoir greenhouse gas diffusion flux (GHG flux) are as follows: Figure 8 As shown in the figure, the horizontal axis represents longitude, the vertical axis represents latitude, the yellow circles represent the gas diffusion flux sampling points corresponding to carbon dioxide CO2, a total of 2364; the red circles represent the gas diffusion flux sampling points corresponding to methane CH4, a total of 508; the blue circles represent the gas diffusion flux sampling points corresponding to nitrous oxide N2O, a total of 727.

[0112] Furthermore, the sampling point location map was divided according to climatic zones, including plateau and mountain, subtropical, temperate and tropical zones.

[0113] Step 2: Collecting Characteristic Factors. Seven reservoir characteristic parameters were compiled based on publicly available reservoir and dam information datasets. To determine environmental and socioeconomic predictors for each reservoir's catchment area, a basin environmental factor assimilation platform built using open-source GIS technology was used to obtain multivariate environmental characteristic information for 45 watersheds. By integrating all of this data, a database of reservoir greenhouse gas concentrations and emissions, including three greenhouse gases, was constructed.

[0114] Step 3: Establish a reservoir concentration model for the three greenhouse gases. Using the RStudio-2024.12.0-467 program, a random forest model was constructed for CO2, CH4, and N2O concentrations. Model training, hyperparameter adjustment, and performance evaluation were performed. During the modeling process, to avoid the influence of extreme values, data outside the 95% interval were removed, and the remaining data were randomly divided into a training set (75%) and a test set (25%). All the aforementioned predictors were input into the model for training. The optimal model parameters were selected by constructing a parameter grid and combining cross-validation. To measure the model's predictive performance, the coefficient of determination (R2) and root mean square error (RMSE) were used on the test set.

[0115] Step 4: We selected 1197 pairs of data from the reservoir greenhouse gas dataset that reported both concentration and flux, and established the gas exchange coefficients by combining reservoir characteristic parameters and basin environmental characteristics. Random forest model; according to The model prediction results are used, and the greenhouse gas concentrations in the reservoir predicted by the concentration model and the diffusion fluxes of the three greenhouse gases in the reservoir are calculated based on the thin film boundary layer theory to obtain the greenhouse gas emissions after water storage.

[0116] Step 5: Calculate the net greenhouse gas emissions from the reservoir. The net emission flux before and after reservoir construction is defined as the average of the total emissions per unit area of ​​each reservoir after deducting the pre-filling greenhouse gas footprint, i.e., net flux = (cumulative flux after filling - greenhouse gas footprint before filling) / reservoir area. The pre-filling greenhouse gas footprint is assessed by multiplying the surface area of ​​each land cover subunit by the greenhouse gas emission coefficient. For channel-type reservoirs, the pre-filling landform consists of submerged land and the original river channel; for non-channel-type reservoirs, the pre-filling landform is considered to consist entirely of submerged land. Assuming that the pre-filling landform is the same as the land cover of the reservoir basin, the emissions or absorption of each land cover subunit before the reservoir is filled are calculated based on the spatial distribution characteristics of greenhouse gas emission intensity of terrestrial ecosystems reported in the literature or the published greenhouse gas emission coefficients for each land use type. The length of the channel inundated by the reservoir was estimated by the overlap of the CRD reservoir polygon with the GRADES river network, and the channel width was estimated by combining the hydraulic geometry of the downstream stations with the reach discharge from GRADES.

[0117] Finally, the greenhouse gas balance is calculated cumulatively using the following formula:

[0118] ;

[0119] in, is the greenhouse gas balance, i is the soil climate type, j is the greenhouse gas type, is the jth greenhouse gas emission coefficient of the i-th soil climate type, is the area of ​​soil climate type i, is the reservoir area.

[0120] For example, the spatial distribution diagram of greenhouse gas concentration and diffusion flux of reservoirs in a specific area calculated by the present invention is as follows: Figures 9 to 14 As shown in the figure, the horizontal axis represents longitude and the vertical axis represents latitude. Figure 9 The spatial distribution map of carbon dioxide gas concentration (CO2 concentration), carbon dioxide gas concentration (unit: ) The value size is positively correlated with the color distribution of the spatial distribution map. The larger the value, the darker the corresponding color. For example, the color density of the carbon dioxide gas concentration value in the range of 28.80-34.94 is greater than the color density of the carbon dioxide gas concentration value in the range of 18.78-28.79. Figure 10 The spatial distribution diagram of carbon dioxide diffusion flux (CO2flux), carbon dioxide diffusion flux (unit: ) The value size is positively correlated with the color distribution of the spatial distribution map. The larger the value, the darker the corresponding color. For example, the color concentration of the carbon dioxide diffusion flux value in the range of 7.24-10.97 is greater than the color concentration of the carbon dioxide diffusion flux value in the range of 0.73-7.23. Figure 11 is the spatial distribution map of methane gas concentration (CH4concentration), methane gas concentration (unit: ) The value size is positively correlated with the color distribution of the spatial distribution map. The larger the value, the darker the corresponding color. For example, the color density of the methane gas concentration value in the range of 0.14-0.15 is greater than the color density of the methane gas concentration value in the range of 0.13-0.14; Figure 12 is the spatial distribution diagram of methane diffusion flux (CH4flux), methane diffusion flux (unit: ) The value size is positively correlated with the color distribution of the spatial distribution map. The larger the value, the darker the corresponding color. For example, the color concentration of the methane diffusion flux value in the range of 0.08-0.09 is greater than the color concentration of the methane diffusion flux value in the range of 0.06-0.07. Figure 13 is the spatial distribution diagram of nitrous oxide gas concentration (N2O concentration), nitrous oxide gas concentration (unit: ) The value size and the color distribution of the spatial distribution map are positively correlated. The larger the value, the darker the corresponding color. For example, the color density of the nitrous oxide gas concentration value in the range of 27.41-33.16 is greater than the color density of the nitrous oxide gas concentration value in the range of 12.51-27.40. Figure 14 is the spatial distribution diagram of nitrous oxide diffusion flux (N2O flux), nitrous oxide diffusion flux (unit: ) The numerical value is positively correlated with the color distribution of the spatial distribution map. The larger the numerical value, the darker the corresponding color. For example, the color concentration of the nitrous oxide diffusion flux value in the range of 10.71-12.16 is greater than the color concentration of the nitrous oxide diffusion flux value in the range of 7.14-10.70.

[0121] The greenhouse gas emission accounting method provided by the present invention can achieve the following beneficial effects: through systematic data collection and random forest methods, the greenhouse gas concentrations and fluxes of 97,435 reservoirs are estimated, and the net greenhouse gas emissions of reservoirs in the region are obtained by estimating the greenhouse gas footprint before flooding. Compared with the traditional method of estimating the emissions of multiple reservoirs based on collected data and average value extrapolation, the present invention can effectively reduce the uncertainty of accounting; using the method proposed by the present invention, the specific emissions of each reservoir can be calculated, and the contribution of each type of reservoir and the contribution of each greenhouse gas can be obtained, while the traditional method can only calculate the total greenhouse gas emissions; using the greenhouse gas emission accounting method of the present invention, the greenhouse gas concentration and diffusion flux in a specific area can be visualized, the greenhouse gas emissions of local and overall regional reservoirs can be intuitively understood, and the emission hotspots and net emission hotspots of greenhouse gases in reservoirs can be counted and identified, which is of great value for greenhouse gas emission reduction.

[0122] In addition, the method and model proposed in the present invention are aimed at calculating greenhouse gas emissions from reservoirs, but are not limited to reservoirs. The concept of the present invention can also be applied to the calculation of greenhouse gas emissions from other inland water bodies, such as lakes and rivers.

[0123] like Figure 5 As shown, the embodiment of the present application further provides a greenhouse gas emission accounting device 900, which includes: a data unit 902, which is used to obtain a greenhouse gas database, gas concentration in-situ observation data corresponding to greenhouse gas concentrations of multiple reservoirs, and gas flux in-situ observation data corresponding to greenhouse gas fluxes of multiple reservoirs; determining an accounting data set based on the greenhouse gas database, multiple gas concentration in-situ observation data, and multiple gas flux in-situ observation data; a model unit 904, which is used to establish a random forest model of prediction factors and greenhouse gas concentrations based on the accounting data set in combination with a random forest algorithm; a prediction unit 906, which is used to predict the greenhouse gas concentrations of multiple reservoirs based on the random forest model. volume concentration; calculating the gas exchange coefficient when the Schmidt constant is equal to 600 under standard conditions according to the thin film boundary layer theory; predicting the gas exchange coefficient by using the random forest model to determine the predicted gas exchange coefficient; a determination unit 908 for determining the greenhouse gas diffusion flux of the reservoir according to the greenhouse gas concentration and the predicted gas exchange coefficient; determining the greenhouse gas emissions of the reservoir according to the greenhouse gas diffusion flux of the reservoir and the area of ​​the reservoir, and determining the total emissions corresponding to multiple reservoirs according to the greenhouse gas emissions of the reservoir; determining the greenhouse gas footprint of each reservoir before water storage according to the accounting data set; an accounting unit 910 for determining the net greenhouse gas emissions of each reservoir according to the total emissions after water storage and the greenhouse gas footprint before water storage.

[0124] In some embodiments, optionally, the model unit 904 is also used to: construct an initial model for greenhouse gas concentration based on the accounting data set; divide multiple data in the accounting data set into intervals to determine a first interval and a second interval, and the first interval is larger than the second interval; determine that the data in the first interval in the accounting data set is valid data; divide the valid data into a training set and a test set, and the number of data in the training set is larger than the number of data in the test set; train the initial model based on the training set and the prediction factor to determine a random forest model; wherein, the first model result of the random forest model is determined by constructing a parameter network and combining cross-validation; the test set is characterized by the determination coefficient and the root mean square error, and the characterization result is used to determine the prediction performance of the random forest model.

[0125] In some embodiments, optionally, the accounting unit 910 is also used to: determine the reservoir area and soil climate type of each reservoir; determine the area of ​​each soil climate type based on the soil climate type and the reservoir area; determine the greenhouse gas emission coefficient corresponding to the soil climate type; determine the greenhouse gas balance based on the reservoir area, the soil climate type area and the greenhouse gas emission coefficient corresponding to the soil climate type.

[0126] The greenhouse gas emission accounting device 900 provided by the present invention realizes the greenhouse gas emission accounting method. By integrating multi-source data, machine learning modeling and dual accounting benchmarks, the net greenhouse gas emissions before and after the construction of the reservoir in the region are accurately quantified, and the data source is covered to the global scope. The random forest model is combined with multi-dimensional prediction factors to capture the nonlinear mechanism of greenhouse gas generation, and the greenhouse gas emissions are predicted from multiple aspects, avoiding the uncertainty of greenhouse gas accounting caused by extrapolation based on the average value, and improving the robustness of the reservoir greenhouse gas emission accounting.

[0127] like Figure 6 As shown, an embodiment of the present application also provides an electronic device 1000, including a processor 1110, a memory 1109, and a program or instruction stored in the memory 1109 and executable on the processor 1110. When the program or instruction is executed by the processor 1110, the various processes of the embodiment of the above-mentioned greenhouse gas emission accounting method are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0128] Optionally, the processor 1110 is configured to obtain a greenhouse gas database, in-situ observation data of gas concentrations corresponding to greenhouse gas concentrations of a plurality of reservoirs, and in-situ observation data of gas fluxes corresponding to greenhouse gas fluxes of a plurality of reservoirs;

[0129] Optionally, the processor 1110 is further configured to determine a calculation data set based on a greenhouse gas database, a plurality of gas concentration in-situ observation data, and a plurality of gas flux in-situ observation data;

[0130] Optionally, the processor 1110 is further configured to establish a random forest model of the prediction factors and greenhouse gas concentrations based on the calculation data set and a random forest algorithm;

[0131] Optionally, the processor 1110 is further configured to predict greenhouse gas concentrations in a plurality of reservoirs based on a random forest model;

[0132] Optionally, the processor 1110 is further configured to calculate a gas exchange coefficient when a Schmidt constant is equal to 600 under standard conditions according to thin film boundary layer theory; and predict the gas exchange coefficient using a random forest model to determine a predicted gas exchange coefficient.

[0133] Optionally, the processor 1110 is further configured to determine a reservoir greenhouse gas diffusion flux based on the greenhouse gas concentration and the predicted gas exchange coefficient; determine the reservoir greenhouse gas emissions based on the reservoir greenhouse gas diffusion flux and the area of ​​the reservoir; and determine a total emission corresponding to multiple reservoirs based on the reservoir greenhouse gas emissions; and determine a pre-filling greenhouse gas footprint for each reservoir based on the accounting data set.

[0134] Optionally, the processor 1110 is further configured to determine the net greenhouse gas emissions of each reservoir based on the total emissions after water storage and the greenhouse gas footprint before water storage.

[0135] Memory 1109 can be used to store software programs and various data. Memory 1109 may primarily include a first storage area for storing programs or instructions and a second storage area for storing data. The first storage area may store an operating system, applications or instructions required for at least one function (such as a sound playback function, an image playback function, etc.). Furthermore, memory 1109 may include volatile memory or non-volatile memory, or both. Non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory may be random access memory (RAM), static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct RAM bus random access memory (DRRAM). The memory 1109 in the embodiment of the present application includes but is not limited to these and any other suitable types of memory.

[0136] In the present invention, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. The term "plurality" refers to two or more, unless expressly limited otherwise. Terms such as "installed," "connected," "connected," and "fixed" should be interpreted broadly. For example, "connected" can mean a fixed connection, a detachable connection, or an integral connection; "connected" can mean a direct connection or an indirect connection through an intermediary. Those skilled in the art will understand the specific meanings of these terms in the present invention based on specific circumstances.

[0137] In the description of the present invention, it should be understood that the directions or positional relationships indicated by terms such as "up", "down", "left", "right", "front" and "back" are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or unit referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, they should not be understood as limiting the present invention.

[0138] Throughout this specification, terms such as "one embodiment," "some embodiments," and "specific embodiments" mean that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0139] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A method for calculating greenhouse gas emissions, characterized in that: include: acquiring a greenhouse gas database, in-situ observation data of gas concentrations corresponding to greenhouse gas concentrations of a plurality of reservoirs, and in-situ observation data of gas fluxes corresponding to greenhouse gas fluxes of the plurality of said reservoirs; Determining a calculation data set according to the greenhouse gas database, a plurality of in-situ observation data of gas concentrations, and a plurality of in-situ observation data of gas fluxes; Establishing a random forest model of prediction factors and greenhouse gas concentrations based on the accounting data set and the random forest algorithm; predicting greenhouse gas concentrations in a plurality of the reservoirs according to the random forest model; The gas exchange coefficient when the Schmidt constant is equal to 600 under standard conditions is calculated according to the thin film boundary layer theory; Predicting the gas exchange coefficient using the random forest model to determine a predicted gas exchange coefficient; determining a reservoir greenhouse gas diffusion flux based on the greenhouse gas concentration and the predicted gas exchange coefficient; Determining greenhouse gas emissions from the reservoir according to the greenhouse gas diffusion flux of the reservoir and the area of ​​the reservoir, and determining total emissions corresponding to the plurality of reservoirs according to the greenhouse gas emissions from the reservoir; determining a pre-filling greenhouse gas footprint of each of the reservoirs based on the accounting dataset; determining net greenhouse gas emissions of each reservoir based on the total emissions after impoundment and the greenhouse gas footprint before impoundment; Among them, the greenhouse gas footprint before water storage caused by natural rivers and land emissions and the greenhouse gas emissions after water storage caused by human engineering are automatically separated according to the source of greenhouse gas emissions; After determining the net greenhouse gas emissions of each reservoir based on the total emissions after water storage and the greenhouse gas footprint before water storage, the method further includes: determining the reservoir area and soil climate type of each of said reservoirs; Determine the area of ​​each soil climate type according to the soil climate type and the reservoir area; determining a greenhouse gas emission factor corresponding to the soil climate type; determining a greenhouse gas balance according to the reservoir area, the soil climate type area, and a greenhouse gas emission coefficient corresponding to the soil climate type; The calculation formula for greenhouse gas balance is as follows: ； in, is the greenhouse gas balance, i is the soil climate type, j is the greenhouse gas type, is the jth greenhouse gas emission coefficient of the i-th soil climate type, is the area of ​​soil climate type i, is the reservoir area.

2. The method for calculating greenhouse gas emissions according to claim 1, characterized in that: The accounting data set includes a multidimensional feature matrix, which includes reservoir characteristic factors, climate characteristic factors, watershed environmental factors and socioeconomic factors. The reservoir characteristic factors, the climate characteristic factors, the watershed environmental factors and the socioeconomic factors are prediction factors.

3. The method for calculating greenhouse gas emissions according to claim 2, characterized in that: The method of establishing a random forest model of prediction factors and greenhouse gas concentrations based on the calculation data set in combination with a random forest algorithm includes: constructing an initial model for greenhouse gas concentrations based on the accounting data set; Dividing the plurality of data in the accounting data set into intervals to determine a first interval and a second interval, wherein the first interval is larger than the second interval; Determining that data within the first interval in the accounting data set is valid data; Dividing the valid data into a training set and a test set, wherein the amount of data in the training set is greater than the amount of data in the test set; Training the initial model according to the training set and the predictor to determine a random forest model; The first model result of the random forest model is determined by constructing a parameter network and combining cross-validation; The test set is characterized by the coefficient of determination and the root mean square error, and the results of the characterization are used to determine the predictive performance of the random forest model.

4. The method for calculating greenhouse gas emissions according to claim 1, wherein: Determining the greenhouse gas footprint of each reservoir before water storage based on the accounting data set includes: determining land cover subunits for each of said reservoirs; determining a surface area of ​​the land cover subunit; determining a greenhouse gas emission coefficient for each reservoir before water storage based on the accounting data set; A greenhouse gas footprint of the reservoir before water storage is determined based on the greenhouse gas emission coefficient and the surface area.

5. The method for calculating greenhouse gas emissions according to any one of claims 1 to 4, characterized in that: The accounting data set also includes: The geospatial coordinates, temporal metadata, and temporal resolution of the sampling points corresponding to each of the reservoirs.

6. A greenhouse gas emission calculation device, characterized in that: include: a data unit configured to obtain a greenhouse gas database, in-situ observation data of greenhouse gas concentrations corresponding to a plurality of reservoirs, and in-situ observation data of greenhouse gas fluxes corresponding to the plurality of reservoirs; and determine a calculation data set based on the greenhouse gas database, the plurality of in-situ observation data of gas concentrations, and the plurality of in-situ observation data of gas fluxes; A model unit, configured to establish a random forest model of prediction factors and greenhouse gas concentrations based on the accounting data set in combination with a random forest algorithm; a prediction unit, configured to predict greenhouse gas concentrations of the plurality of reservoirs according to the random forest model; calculate a gas exchange coefficient when a Schmidt constant is equal to 600 under standard conditions according to thin film boundary layer theory; and predict the gas exchange coefficient using the random forest model to determine a predicted gas exchange coefficient; a determination unit, configured to determine a reservoir greenhouse gas diffusion flux based on the greenhouse gas concentration and the predicted gas exchange coefficient; determine a reservoir greenhouse gas emission based on the reservoir greenhouse gas diffusion flux and the area of ​​the reservoir, and determine a total emission corresponding to a plurality of the reservoirs based on the reservoir greenhouse gas emission; and determine a pre-water storage greenhouse gas footprint for each of the reservoirs based on the accounting data set; an accounting unit, configured to determine the net greenhouse gas emissions of each reservoir based on the total emissions after water storage and the greenhouse gas footprint before water storage; Among them, the greenhouse gas footprint before water storage caused by natural rivers and land emissions and the greenhouse gas emissions after water storage caused by human engineering are automatically separated according to the source of greenhouse gas emissions; After determining the net greenhouse gas emissions of each reservoir based on the total emissions after water storage and the greenhouse gas footprint before water storage, the method further includes: determining the reservoir area and soil climate type of each of said reservoirs; Determine the area of ​​each soil climate type according to the soil climate type and the reservoir area; determining a greenhouse gas emission factor corresponding to the soil climate type; determining a greenhouse gas balance according to the reservoir area, the soil climate type area, and a greenhouse gas emission coefficient corresponding to the soil climate type; The calculation formula for greenhouse gas balance is as follows: ； in, is the greenhouse gas balance, i is the soil climate type, j is the greenhouse gas type, is the jth greenhouse gas emission coefficient of the i-th soil climate type, is the area of ​​soil climate type i, is the reservoir area.

7. The greenhouse gas emission calculation device according to claim 6, characterized in that: The model unit is also used to: constructing an initial model for greenhouse gas concentrations based on the accounting data set; Dividing the plurality of data in the accounting data set into intervals to determine a first interval and a second interval, wherein the first interval is larger than the second interval; Determining that data within the first interval in the accounting data set is valid data; Dividing the valid data into a training set and a test set, wherein the amount of data in the training set is greater than the amount of data in the test set; Training the initial model according to the training set and the predictor to determine a random forest model; The first model result of the random forest model is determined by constructing a parameter network and combining cross-validation; The test set is characterized by the coefficient of determination and the root mean square error, and the results of the characterization are used to determine the predictive performance of the random forest model.

8. An electronic device, characterized in that: The method comprises a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps of the method for calculating greenhouse gas emissions as described in any one of claims 1 to 5.

Images