Greenhouse gas point source emission automatic identification method, system, equipment and medium

Abstract

The invention discloses a greenhouse gas point source emission automatic identification method, system and device and a medium, and relates to the technical field of environment monitoring. The method comprises the following steps: acquiring a greenhouse gas emission list and satellite remote sensing data of a to-be-detected area; carrying out emission source preliminary screening according to the greenhouse gas emission list, and determining a preliminarily screened greenhouse gas source; inverting a greenhouse gas concentration distribution diagram by using satellite remote sensing data, and performing target detection and pixel segmentation on the greenhouse gas concentration distribution diagram by using a Mask R-CNN model to determine a greenhouse gas plume; the greenhouse gas plume comprises a specific position and a specific form; and performing quantitative analysis on the greenhouse gas plume by using an IME algorithm, and determining a final emission source in combination with the preliminarily screened greenhouse gas source. According to the invention, the recognition accuracy and real-time performance of the greenhouse gas emission source can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental monitoring, and particularly to an automatic identification method, system, device and medium for greenhouse gas point source emissions. Background Art

[0002] Greenhouse gases, or greenhouse effect gases, refer to the gas components in the atmosphere that cause the greenhouse effect. The greenhouse effect can maintain the temperature of the planet's surface and prevent excessive heat from escaping in the form of radiation. However, an excessive greenhouse effect can also lead to an increase in the planet's temperature and global climate anomalies. Generally speaking, greenhouse gases mainly include the following six types: carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), hydrofluorocarbons (HFCs), perfluorocarbons (PFCs), and sulfur hexafluoride (SF6). With the gradual increase in the degree of global industrialization, the content of greenhouse gases on the Earth's surface has gradually increased with human activities. Among them, the two types of greenhouse gases that contribute the most to the greenhouse effect: carbon dioxide (CO2) and methane (CH4), have shown an even more significant growth trend in recent years.

[0003] Currently, the emission monitoring of greenhouse gases is mainly divided into two technical paths: the bottom-up inventory method and the top-down measurement method.

[0004] The bottom-up inventory method relies on statistical data and mathematical models to estimate the emission activities of emission sources. For example, based on global emission databases such as Carbon Mapper and Climate Trace, the emission flux can be calculated by statistically analyzing emission activities (such as energy use, landfill volume) and emission factors, and using methods such as the first-order decay model. This method has the advantages of convenience and integrity when the data is complete. However, its accuracy highly depends on the quality of statistical data and the reliability of model parameters. For example, key parameters such as methane production rate need to be obtained through on-site experiments, which have a large degree of uncertainty. In addition, the inventory method cannot reflect the emission dynamics in real time and has limited applicability to areas with scarce data.

[0005] The top-down measurement method realizes the inversion of emission amounts by directly detecting the concentration of greenhouse gases in the atmosphere, mainly including two technical paths: ground-based monitoring and spaceborne remote sensing:

[0006] Ground-based monitoring: Using handheld devices or fixed monitoring stations to monitor specific emission sources (such as power plants, landfills). This method has the characteristics of high precision and strong reliability, and can output high-timeliness greenhouse gas concentration data. However, its application scope is limited by the number of monitoring stations and terrain conditions, and it is difficult to cover large-scale or hidden emission sources.

[0007] Spaceborne remote sensing: Remote sensing technology based on satellite platforms detects the absorption characteristics of greenhouse gases in the atmospheric reflection spectrum to invert their concentration distribution. Spaceborne remote sensing has advantages such as a wide coverage area and a short revisit period, and is an important means for current greenhouse gas monitoring. For example, hyperspectral satellites can quickly provide regional emission data. However, the emission plumes in the spectrally identified images are complex, making it difficult to identify emission signals from small emission sources or complex regions.

[0008] Due to its high spectral and high spatial resolution, hyperspectral remote sensing technology can identify greenhouse gas emission hotspots in small or complex scenarios. However, there are still the following difficulties in the process of identifying emission plumes: 1. Strong background interference: Interference signals from high-reflectivity ground objects (such as building roofs and solar panels) often appear in hyperspectral data, affecting the accuracy of gas concentration inversion. 2. Complex plume morphology: Emission plumes are usually affected by factors such as terrain and airflows, with complex and irregular morphologies, making it difficult for traditional image processing methods to effectively extract their boundaries and concentration distributions. 3. Large data volume for processing: Hyperspectral data has many bands and high spatial resolution, posing high requirements for computing resources and algorithm optimization. Although the hyperspectral matching filtering method can extract greenhouse gas concentration information, its adaptability to complex background conditions is weak, especially limited in identifying irregular plume morphologies and quantifying small emission sources. Summary of the Invention

[0009] The purpose of the present invention is to provide an automatic identification method, system, device and medium for greenhouse gas point source emissions, which can improve the accuracy and real-time performance of identifying greenhouse gas emission sources.

[0010] To achieve the above purpose, the present invention provides the following solutions:

[0011] An automatic identification method for greenhouse gas point source emissions includes:

[0012] Obtain a greenhouse gas emission inventory and satellite remote sensing data of the area to be measured;

[0013] Perform preliminary screening of emission sources according to the greenhouse gas emission inventory to determine the preliminarily screened greenhouse gas sources;

[0014] Invert the greenhouse gas concentration distribution map using the satellite remote sensing data, and use the Mask R-CNN model to perform object detection and pixel segmentation on the greenhouse gas concentration distribution map to determine the greenhouse gas plume; the greenhouse gas plume includes specific locations and morphologies;

[0015] Perform quantitative analysis on the greenhouse gas plume using the IME algorithm, and determine the final emission sources in combination with the preliminarily screened greenhouse gas sources.

[0016] Optionally, the processing process of performing preliminary screening of emission sources according to the greenhouse gas emission inventory specifically includes:

[0017] Judging the data in the greenhouse gas emission inventory according to the screening criteria, and determining the greenhouse gas sources with large emissions and obvious emission characteristics that meet the screening criteria as the preliminarily screened greenhouse gas sources; the greenhouse gas emission inventory uses at least the global public greenhouse gas emission inventories of CarbonMapper and Climate Trace.

[0018] Optionally, the calculation formula for retrieving the greenhouse gas concentration distribution map using satellite remote sensing data includes:

[0019]

[0020] where α i is the methane concentration increment, with the unit of ppm·m, x a is the average radiance value of the entire scene image, Σ is the covariance matrix of the selected calculation bands of the image, k is the unit absorption spectrum of methane, and x is the retrieved concentration value.

[0021] Optionally, before using the Mask R-CNN model to perform object detection and pixel segmentation on the greenhouse gas concentration distribution map, it further includes: training the Mask R-CNN model;

[0022] The process of training the Mask R-CNN model specifically includes:

[0023] Obtaining training data; the training data includes historical gas concentration distribution maps and corresponding plume labels;

[0024] Constructing a pre-trained network based on a convolutional neural network and a region proposal network; the convolutional neural network is used to extract the spatial features of the remote sensing image; the region proposal network is used to generate regions of interest;

[0025] Inputting the training data into the pre-trained network, training with the goal of minimizing the loss between the network output and the plume label, and determining the network that meets the accuracy evaluation as the final Mask R-CNN model.

[0026] Optionally, the calculation formula for quantitatively analyzing the greenhouse gas plume using the IME algorithm includes:

[0027]

[0028] where: Q is the emission flux, ΔC is the concentration increment, A is the plume area, U is the wind speed, and L is the plume length.

[0029] The present invention also provides an automatic identification system for greenhouse gas point source emissions, including:

[0030] A data acquisition unit for obtaining a greenhouse gas emission inventory and satellite remote sensing data of a region to be measured;

[0031] A preliminary screening unit for preliminarily screening emission sources according to the greenhouse gas emission inventory to determine the preliminarily screened greenhouse gas sources;

[0032] A plume identification unit for retrieving a greenhouse gas concentration distribution map using the satellite remote sensing data, and performing object detection and pixel segmentation on the greenhouse gas concentration distribution map using a Mask R-CNN model to determine a greenhouse gas plume; the greenhouse gas plume includes specific positions and shapes;

[0033] A quantitative analysis unit for quantitatively analyzing the greenhouse gas plume using the IME algorithm and determining the final emission sources in combination with the preliminarily screened greenhouse gas sources.

[0034] The present invention also provides an electronic device including a memory and a processor, the memory is used for storing a computer program, and the processor runs the computer program to enable the electronic device to execute the automatic identification method for greenhouse gas point source emissions according to the above.

[0035] The present invention also provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the automatic identification method for greenhouse gas point source emissions as described above is implemented.

[0036] According to the specific embodiments provided by the present invention, the following technical effects are disclosed by the present invention:

[0037] The present invention discloses an automatic identification method, system, device and medium for greenhouse gas point source emissions. The method includes obtaining a greenhouse gas emission inventory and satellite remote sensing data of a region to be measured; preliminarily screening emission sources according to the greenhouse gas emission inventory to determine the preliminarily screened greenhouse gas sources; retrieving a greenhouse gas concentration distribution map using the satellite remote sensing data, and performing object detection and pixel segmentation on the greenhouse gas concentration distribution map using a Mask R-CNN model to determine a greenhouse gas plume; the greenhouse gas plume includes specific positions and shapes; quantitatively analyzing the greenhouse gas plume using the IME algorithm and determining the final emission sources in combination with the preliminarily screened greenhouse gas sources. The present invention can improve the identification accuracy and real-time performance of greenhouse gas emission sources. Description of the Drawings

[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0039] Figure 1 It is a schematic flowchart of the method for automatically identifying greenhouse gas point source emissions of the present invention. Specific embodiments

[0040] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0041] The purpose of the present invention is to provide a method, system, device and medium for automatically identifying greenhouse gas point source emissions, which can improve the accuracy and real-time performance of identifying greenhouse gas emission sources.

[0042] To make the above objects, features and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the drawings and specific embodiments.

[0043] As Figure 1 shown, the present invention provides a method for automatically identifying greenhouse gas point source emissions, aiming to combine the global greenhouse gas emission inventory, spaceborne hyperspectral imager data and deep learning technology to solve the problems of wide distribution of greenhouse gas emission sources, high difficulty in identification and complex quantification of emission amounts. By comprehensively using statistical inventories and hyperspectral remote sensing data, a method for automatically identifying and quantitatively measuring greenhouse gas point source emissions based on hyperspectral images and deep learning is provided, providing technical support for the precise monitoring of greenhouse gases and the formulation of emission reduction policies. The main steps include:

[0044] Step 100: Obtain the global greenhouse gas emission inventory through public platforms (such as Carbon Mapper, Climate Trace, etc.), and collect hyperspectral data covering the target area from satellite remote sensing data sources (such as PRISMA, GF-5, etc.). The obtained remote sensing data includes hyperspectral images, covering the absorption bands of methane (1600–2500 nm) and carbon dioxide (2000–2200 nm), providing data support for subsequent analysis and target detection. This step not only ensures the diversity of data, but also enables the acquisition of key information from multiple emission sources globally.

[0045] Step 200: Conduct a preliminary screening of emission sources through the greenhouse gas emission inventory to identify key emission sources to be analyzed. Focus on point sources with relatively high methane and carbon dioxide emission intensities, such as landfills, coal mines, chemical plants, oil fields, etc. These sources may have obvious plume characteristics and emission intensities exceeding the predetermined threshold. The screening criteria for emission sources include emission intensity, significance of emission characteristics, and data integrity and time coverage.

[0046] Step 300: Invert the greenhouse gas concentration distribution map using satellite remote sensing data, and use the Mask R-CNN model to perform object detection and pixel segmentation on the greenhouse gas concentration distribution map to determine the greenhouse gas plume; the greenhouse gas plume includes specific locations and shapes.

[0047] Step 400: Use the IME algorithm to quantitatively analyze the greenhouse gas plume, and combine it with the preliminarily screened greenhouse gas sources to determine the final emission sources.

[0048] Based on the above technical solutions, the following embodiments are provided.

[0049] First aspect: A method for screening key emission sources and constructing a training dataset based on the global greenhouse gas emission inventory

[0050] Step 1.1 Obtain global greenhouse gas emission inventory data

[0051] Obtain emission data of facilities or regions from global public greenhouse gas emission inventories such as Carbon Mapper and Climate Trace. These data provide key information such as the geographical location, emissions, and emission characteristics of emission sources, which are used to screen key targets for subsequent hyperspectral identification.

[0052] Step 1.2 Screen key emission sources

[0053] Screen the emission inventory data obtained in Step 1.1, and select point sources with relatively high methane and carbon dioxide emission intensities, including facilities such as landfills, coal mines, chemical plants, and oil fields as analysis targets.

[0054] The screening criteria include: whether the emission intensity exceeds the set threshold; whether the emission characteristics are clear and obvious (such as significant plume characteristics); whether the data integrity and coverage meet the requirements of subsequent analysis.

[0055] Step 1.3 Construct a training dataset

[0056] Collect methane and carbon dioxide plume data related to the screening targets from the Carbon Mapper platform as the training data source for the deep learning model.

[0057] Data cleaning: Remove missing values, outliers, and noise data to ensure data quality.

[0058] Data annotation: Annotate the spatial distribution and concentration values of the plume to provide high-quality inputs for model training.

[0059] Data augmentation: Augment the data by rotation, scaling, and noise addition to improve the model's adaptability in different scenarios.

[0060] Preferably, step 1.3 includes:

[0061] Step 1.3.1 Data grouping

[0062] Classify the plume data according to the emission source type (such as landfill, coal mine, oil field) and store them in groups for targeted optimization during model training.

[0063] Second aspect: Method for generating greenhouse gas concentration increment map based on hyperspectral imager data

[0064] Step 2.1 Obtain hyperspectral data and perform preprocessing

[0065] 1. Hyperspectral data acquisition

[0066] Use spaceborne hyperspectral imagers (such as PRISMA, GF-5, EMIT) to obtain hyperspectral data covering the target area. Select image data in the methane absorption band (such as 1600–2500 nm) and carbon dioxide absorption band (such as 2000–2200 nm).

[0067] 2. Radiometric calibration

[0068] Convert the digital number (DN) of the hyperspectral image to radiance, and the calculation formula is:

[0069] L = Gain·DN + Bias

[0070] where L is the radiance, with the unit of W / (m 2 ·sr·μm), and Gain and Bias are the calibration coefficients of the hyperspectral instrument.

[0071] 3. Topographic correction

[0072] For mountainous areas and complex terrains, correct the influence of the solar incidence angle on the radiance. The correction formula is:

[0073]

[0074] where L H is the corrected brightness of the flat ground surface, L T is the actual brightness under complex terrain, and cosθs and cosγ s They are the cosine values of the solar incidence angle and the observation angle respectively.

[0075] 4. Rejection of interference data

[0076] Use the cloud mask technology to remove the cloud-covered areas; identify high-reflectivity ground objects (such as building roofs, solar panels), and replace their values with the background brightness.

[0077] Step 2.2 Matching filter inversion of concentration increment

[0078] 1. Generation of unit absorption spectrum

[0079] Use the radiative transfer model combined with the HITRAN molecular absorption library to simulate the unit absorption spectrum of greenhouse gases (such as methane) for concentration increment calculation.

[0080] 2. Inversion of concentration increment

[0081] Invert the concentration increment of greenhouse gases through the matching filter algorithm. The formula is:

[0082]

[0083] where α i is the methane concentration increment (unit: ppm·m), x a is the average radiance value of the entire scene image, Σ is the covariance matrix of the selected calculation band of the image, k is the unit absorption spectrum of methane, and x is the inverted concentration value.

[0084] Step 2.3 Optimize the increment map and generate the result

[0085] 1. Noise rejection

[0086] In the inversion result, there may be some isolated noise points or false detection areas, and these noises do not represent real greenhouse gas emissions. Therefore, it is necessary to use image processing techniques, such as morphological operations, filtering denoising, etc., to remove the noises and ensure the accuracy of the result.

[0087] 2. Generation of increment map

[0088] For the denoised concentration increment data, further process it to generate a high-resolution greenhouse gas concentration increment map. This map can accurately reflect the greenhouse gas concentration changes in each region and provide basic data for subsequent plume automatic identification and emission quantification calculation.

[0089] Third aspect: Automatic plume identification and emission quantification method based on deep learning

[0090] Step 3.1 Model training

[0091] Data Preparation

[0092] Use the hyperspectral concentration increment map generated in the aforementioned second aspect and the plume distribution data extracted from Carbon Mapper to construct the training set of the Mask R-CNN model.

[0093] Data Normalization: Normalize the concentration increment data to ensure the consistency of the model input features.

[0094] Data Annotation: Accurately annotate the greenhouse gas plume regions in the training data, including the spatial location and morphology of the plume. The annotation information will be used for supervised learning to help the model learn how to identify the plume. Model Training

[0095] Adopt the Mask R-CNN deep learning model for training:

[0096] Use a Convolutional Neural Network (CNN) to extract the spatial features in the hyperspectral image using the CNN, and process the image data through convolutional layers and pooling layers.

[0097] Generate Regions of Interest (RoIs) through the Region Proposal Network (RPN), that is, the regions where greenhouse gas plumes may exist.

[0098] Perform classification and regression predictions within each RoI to output the specific location and morphology of the plume.

[0099] Model Validation

[0100] Perform model validation on an independent test set to evaluate the accuracy and robustness of the model in the automatic identification of plumes. Use metrics such as accuracy, recall, and F1-score to evaluate the performance of the model in different scenarios to ensure the universality and efficiency of the model.

[0101] Step 3.2 Model Application

[0102] 1. Plume Identification

[0103] Use the trained Mask R-CNN model to analyze the hyperspectral concentration increment map and automatically identify the location and morphology of greenhouse gas plumes. Through the model output, generate a plume distribution map showing the concentration distribution of greenhouse gases in each region.

[0104] 2. Data Post-processing

[0105] Perform post-processing on the recognition results: Eliminate isolated noise points and retain continuous plume regions. Smooth the plume boundaries to improve the stability of the recognition results.

[0106] Step 3.3 Emission Quantification

[0107] Flux Calculation

[0108] Using the plume distribution map and wind speed information, combined with the IME algorithm to calculate the greenhouse gas emission flux, the calculation formula is:

[0109]

[0110] Where: Q: Emission flux (unit: kg / h); ΔC: Concentration increment (unit: ppm); A: Plume area (unit: m2); U: Wind speed (unit: m / s); L: Plume length (unit: m).

[0111] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same and similar parts between the various embodiments, reference can be made to each other.

[0112] Specific examples are used in this article to elaborate on the principles and implementation manners of the present invention. The descriptions of the above embodiments are only used to help understand the core idea of the present invention; at the same time, for those of ordinary skill in the art, based on the idea of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. An automatic identification method for greenhouse gas point source emissions, characterized in that Including: Obtain the greenhouse gas emission inventory and satellite remote sensing data of the area to be measured; Conduct a preliminary screening of emission sources based on the greenhouse gas emission inventory to determine the preliminarily screened greenhouse gas sources; Invert the greenhouse gas concentration distribution map using satellite remote sensing data, and use the Mask R-CNN model to perform object detection and pixel segmentation on the greenhouse gas concentration distribution map to determine the greenhouse gas plume; the greenhouse gas plume includes the specific location and shape; Use the IME algorithm to quantitatively analyze the greenhouse gas plume, and combine it with the preliminarily screened greenhouse gas sources to determine the final emission sources.

2. The automatic identification method for greenhouse gas point source emissions according to claim 1, wherein The processing process of conducting a preliminary screening of emission sources based on the greenhouse gas emission inventory specifically includes: Judge the data in the greenhouse gas emission inventory according to the screening criteria, and determine the greenhouse gas sources with large emissions and obvious emission characteristics that meet the screening criteria as the preliminarily screened greenhouse gas sources; the greenhouse gas emission inventory at least uses the global public greenhouse gas emission inventories of Carbon Mapper and Climate Trace.

3. The automatic identification method for greenhouse gas point source emissions according to claim 1, characterized in that The calculation formula for inverting the greenhouse gas concentration distribution map using satellite remote sensing data includes: where α i is the increment of methane concentration, with the unit of ppm·m, and x a is the average radiance value of the whole scene image, Σ is the covariance matrix of the selected calculation band of the image, k is the unit absorption spectrum of methane, and x is the concentration value after inversion.

4. The automatic identification method for greenhouse gas point source emissions according to claim 1, wherein Before using the Mask R-CNN model to perform object detection and pixel segmentation on the greenhouse gas concentration distribution map, it also includes: training the Mask R-CNN model; The process of training the Mask R-CNN model specifically includes: Obtain training data; the training data includes historical gas concentration distribution maps and corresponding plume labels; Construct a pre-trained network based on a convolutional neural network and a region proposal network; the convolutional neural network is used to extract the spatial features of the remote sensing image; the region proposal network is used to generate regions of interest; Input the training data into the pre-trained network, train with the goal of minimizing the loss between the network output and the plume label, and determine the final Mask R-CNN model as the network that meets the accuracy evaluation.

5. The automatic identification method for greenhouse gas point source emissions according to claim 1, wherein The calculation formula for quantitatively analyzing the greenhouse gas plume using the IME algorithm includes: Where: Q is the emission flux, ΔC is the concentration increment, A is the area of the plume region, U is the wind speed, and L is the length of the plume.

6. An automatic identification system for greenhouse gas point source emissions, characterized in that, Including: A data acquisition unit for obtaining the greenhouse gas emission inventory and satellite remote sensing data of the area to be measured; A preliminary screening unit for conducting a preliminary screening of emission sources based on the greenhouse gas emission inventory to determine the preliminarily screened greenhouse gas sources; A plume identification unit for inverting the greenhouse gas concentration distribution map using satellite remote sensing data, and using the Mask R-CNN model to perform object detection and pixel segmentation on the greenhouse gas concentration distribution map to determine the greenhouse gas plume; the greenhouse gas plume includes the specific location and shape; A quantitative analysis unit for quantitatively analyzing the greenhouse gas plume using the IME algorithm, and combining it with the preliminarily screened greenhouse gas sources to determine the final emission sources.

7. An electronic device, characterized in that, It includes a memory and a processor. The memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute the automatic identification method for greenhouse gas point source emissions according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, It stores a computer program, and when the computer program is executed by a processor, it implements the automatic identification method for greenhouse gas point source emissions according to any one of claims 1-5.

Images