Dynamic inversion method, system, equipment and medium for near-surface greenhouse gas concentration

Abstract

The present invention belongs to the technical field of atmospheric environment monitoring, and specifically discloses a dynamic inversion method, system, device and medium for near-surface greenhouse gas concentration. The method includes: obtaining greenhouse gas concentration data of a ground observation station, greenhouse gas concentration data of satellite remote sensing, meteorological data, geographic information data and CO2 emission inventory data within a set time period in a target area; matching the data of the ground observation station with the remote sensing data of the satellite in terms of time and space; based on the spatio-temporal matching data, using the ST-CatBoost remote sensing inversion model to obtain discrete strip-shaped near-surface greenhouse gas concentration; using the near-surface greenhouse gas concentration spatial prediction model to obtain spatially continuous near-surface greenhouse gas concentration covering the target area, and generating a spatial distribution map of near-surface greenhouse gas concentration. The present invention can ensure the data consistency of the near-surface observation point data and the remote sensing observation data in terms of time and space.

Description

Technical Field

[0001] The present invention relates to the technical field of atmospheric environment monitoring, and in particular to a method, system, equipment and medium for dynamic inversion of near-ground greenhouse gas concentration. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] Under the severe situation of global climate change, greenhouse gas emissions have become a focus of attention. Traditional methods based on site monitoring, statistical calculations and mechanism models have many limitations, such as the very limited spatial coverage of ground-based monitoring sites, and emission inversion combined with emission factor calculations and numerical model simulations (WFR-CHEM, WRF-STILT, etc.), which are computationally intensive, time-sensitive, and have large errors, making it difficult to accurately understand the dynamic characteristics and emission patterns of greenhouse gases.

[0004] With the continuous development of satellite remote sensing technology, it provides a new perspective and means for large-scale, real-time and continuous greenhouse gas monitoring. At the same time, in recent years, satellite remote sensing inversion algorithms have also been continuously developed and optimized, and inversion methods have gradually evolved from empirical algorithms and physical algorithms to artificial intelligence algorithms, providing technical support for the construction of accurate and dynamic greenhouse gas inversion methods.

[0005] Although current satellite remote sensing technology provides a new perspective for greenhouse gas monitoring, it is limited by the observation characteristics of current satellite sensors. The greenhouse gas concentration data obtained often presents a discontinuous strip distribution, making it difficult to achieve comprehensive coverage of the study area.

[0006] In addition, near-ground greenhouse gas concentrations are easily affected by environmental and climatic conditions, emission sources are spatially clustered, and greenhouse gas concentrations in different regions show seasonal variations. It is often impossible to accurately obtain the spatiotemporal distribution characteristics of greenhouse gases using satellite remote sensing technology alone. Summary of the invention

[0007] In order to solve the above problems, the present invention proposes a dynamic inversion method, system, equipment and medium for near-ground greenhouse gas concentration. Combined with GOSAT satellite remote sensing and machine learning technology, it can not only overcome the limitation of discontinuous spatial distribution of satellite data, but also generate spatial and temporal continuous greenhouse gas concentration distribution characteristics covering the entire region, providing more comprehensive and reliable data support for regional-scale greenhouse gas monitoring and assessment.

[0008] In some embodiments, the following technical solutions are adopted:

[0009] A dynamic inversion method for near-ground greenhouse gas concentration, comprising:

[0010] Obtain greenhouse gas concentration data from ground observation stations, satellite remote sensing greenhouse gas concentration data, meteorological data, geographic information data and CO2 emission inventory data within a set time period in the target area;

[0011] In terms of time, the daily average concentration data of the ground observation station is matched with the remote sensing data of the satellite passing on the same day; in terms of space, the satellite remote sensing data that overlaps with the ground observation station within the matching radius is matched;

[0012] The greenhouse gas concentration data, meteorological data, geographic information data and CO2 emission inventory data after time and space matching are used as input, and the trained ST-CatBoost remote sensing inversion model is used to obtain the discrete strip-shaped near-ground greenhouse gas concentration;

[0013] The discrete strip-shaped near-ground greenhouse gas concentration, meteorological data, geographic information data and CO2 emission inventory data are used as input, and the trained near-ground greenhouse gas concentration spatial prediction model is used to obtain the spatially continuous near-ground greenhouse gas concentration covering the target area, and generate a near-ground greenhouse gas concentration spatial distribution map.

[0014] As a further solution, the greenhouse gas concentration data include: satellite remote sensing column concentration data of CO2 and CH4 and ground concentration data of CO2 and CH4 at observation sites; the meteorological data include: temperature, wind speed, wind direction, evaporation, albedo, ground air pressure, boundary layer height, near-sea surface temperature and total rainfall data; geographic information data include topography, population distribution, land use data, soil data and normalized difference vegetation index data; CO2 emission inventory data include CO2 emissions of each location unit in the target area within a set time period.

[0015] As a further solution, the daily average concentration data of the ground observation station is matched with the remote sensing data of the satellite passing on the same day in terms of time, specifically:

[0016] The daily ground observation station data are temporally matched with the satellite remote sensing column concentration data passing through in the same time period.

[0017] As a further solution, spatially overlapping satellite remote sensing data within a set radius of ground observation sites will be matched, specifically:

[0018] On each day, the ground observation site data are matched with the satellite remote sensing column concentration data within a set matching radius centered on the corresponding ground observation site; the specific matching method is as follows:

[0019] If no satellite remote sensing column concentration data that overlaps with the current ground observation site data in space is found within the set matching radius, the current ground observation site data has no matching data and is not included in the data set;

[0020] If there is only one satellite remote sensing data point that overlaps spatially with the data of the current ground observation site within the set matching radius, the greenhouse gas column concentration value of the satellite remote sensing data point will be used as the column concentration matching the ground observation site and included in the data set;

[0021] If there are two or more satellite remote sensing column concentration data points that overlap with the current ground observation site data in space within the set matching radius, the weighted average method is used to match these satellite column concentration data.

[0022] As a further solution, the weighted average method is used to match these satellite column concentration data points, specifically:

[0023] The distance between the ground observation site and each remote sensing column concentration data point is calculated, and the reciprocal of the distance is used as the spatial weight;

[0024] The weighted average of the column concentration observations of each remote sensing data point is calculated based on the weights to obtain the final matching result:

[0025]

[0026] ;

[0027] in, , , …, are the distances between the ground observation site and each remote sensing data point, , , …, are the observed values ​​of column concentration at each remote sensing data point; It is the number of satellite remote sensing column concentration data points corresponding to the data of the current ground observation site within the set matching radius.

[0028] As a further solution, after obtaining the spatial distribution map of greenhouse gas concentration near the ground, it also includes:

[0029] By detecting abnormal points in the spatial distribution map of near-ground greenhouse gas concentrations, information on near-ground high-value areas of greenhouse gases is automatically extracted, output as a spatial distribution map of abnormal greenhouse gas emission sources, and generate a list of high-value points.

[0030] As a further solution, the specific process of detecting abnormal points in the spatial distribution map of near-ground greenhouse gas concentration is as follows:

[0031] Divide the spatial distribution map of near-ground greenhouse gas concentration into multiple spatial grid points, each of which corresponds to a near-ground greenhouse gas concentration data;

[0032] Randomly select a CO2 or CH4 concentration value and use the concentration value as a split point to divide the data space into two sub-parts; for each sub-part, randomly select a CO2 or CH4 concentration value as a split point to divide the sub-part into two sub-parts;

[0033] The above process is iterated repeatedly until the CO2 or CH4 concentration value at a certain spatial grid point is isolated or reaches a predetermined depth;

[0034] The spatial grid points are identified as high-value areas, and near-ground greenhouse gas concentration, longitude and latitude, dimension and location information of the high-value areas are extracted.

[0035] In other embodiments, the following technical solutions are adopted:

[0036] A dynamic inversion system for near-ground greenhouse gas concentration, comprising:

[0037] The data acquisition module is used to obtain greenhouse gas concentration data from ground observation stations within a set time period in the target area, greenhouse gas concentration data from satellite remote sensing, meteorological data, geographic information data, and CO2 emission inventory data;

[0038] The spatiotemporal matching module is used to match the daily average concentration data of the ground observation site with the remote sensing data of the satellite passing on the same day in terms of time; and to match the satellite remote sensing data that overlaps spatially within the matching radius set by the ground observation site in terms of space;

[0039] The remote sensing inversion module is used to take the greenhouse gas concentration data, meteorological data, geographic information data and CO2 emission inventory data after time and space matching as input, and use the trained ST-CatBoost remote sensing inversion model to obtain discrete strip-shaped near-ground greenhouse gas concentrations;

[0040] The spatial prediction module is used to take the discrete strip-shaped near-ground greenhouse gas concentration, meteorological data, geographic information data and CO2 emission inventory data as input, and use the trained near-ground greenhouse gas concentration spatial prediction model to obtain the spatially continuous near-ground greenhouse gas concentration covering the target area, and generate a near-ground greenhouse gas concentration spatial distribution map.

[0041] In other embodiments, the following technical solutions are adopted:

[0042] A terminal device comprises a processor and a memory, wherein the processor is used to implement instructions; the memory is used to store a plurality of instructions, wherein the instructions are suitable for being loaded by the processor and executing the above-mentioned dynamic inversion method of near-ground greenhouse gas concentration.

[0043] In other embodiments, the following technical solutions are adopted:

[0044] A computer-readable storage medium stores a plurality of instructions, wherein the instructions are suitable for being loaded by a processor of a terminal device and executing the above-mentioned dynamic inversion method for near-ground greenhouse gas concentration.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] (1) The present invention matches the ground observation point data with the remote sensing column concentration data in time and space respectively, which can ensure the data consistency between the near-ground observation point data and the remote sensing observation data in time and space, and improve the accuracy and reliability of model training.

[0047] When performing spatial data matching, the specific matching result is determined through a weighted average scheme based on the number of remote sensing observation data points corresponding to the near-ground observation point data within the matching radius, thereby improving the accuracy of the matching result.

[0048] (2) The present invention addresses the technical difficulties in remote sensing inversion of greenhouse gas concentrations. It makes comprehensive use of multi-source data such as near-surface observation point data, satellite remote sensing data, meteorological data, geographic information data, and CO2 emission inventory data, combined with machine learning algorithms, to achieve high-precision near-surface greenhouse gas concentration inversion. At the same time, through the spatiotemporal data matching method, the spatiotemporal information matching of multi-source data is enhanced, reducing the error impact of the multi-source information input model.

[0049] (3) This invention combines the ST-CatBoost remote sensing inversion model and the near-surface greenhouse gas concentration spatial prediction model to effectively integrate the model inversion results with the remote sensing satellite observation results, overcoming the limitations of traditional satellite remote sensing data, which presents strip-like distribution and discontinuous spatial coverage, and achieves the reconstruction of the spatiotemporal continuous distribution characteristics of greenhouse gas concentrations in the research area, providing comprehensive and reliable data support for regional-scale greenhouse gas monitoring and assessment. This method not only improves the spatiotemporal integrity of greenhouse gas concentration products, but also improves the accuracy of monitoring results through the fusion of multi-source data.

[0050] Other features and advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1It is a schematic diagram of the dynamic inversion of global near-surface greenhouse gas concentration and the high-value area identification process in an embodiment of the present invention;

[0052] Figure 2 Schematic diagram showing that the GOSAT satellite CO2 / CH4 column concentration does not match the WDCGG site observation data in an embodiment of the present invention;

[0053] Figure 3 Schematic diagram of single-point matching between GOSAT satellite CO2 / CH4 column concentration and WDCGG site observation data in an embodiment of the present invention;

[0054] Figure 4 Schematic diagram of multi-point matching between GOSAT satellite CO2 / CH4 column concentration and WDCGG station observation data in an embodiment of the present invention;

[0055] Figure 5 This is a graph showing the accuracy evaluation results of the CO2 gas concentration inversion result and the WDCGG station observation data in an embodiment of the present invention;

[0056] Figure 6 This is a diagram showing the accuracy evaluation results of the CH4 gas concentration inversion result and the WDCGG site observation data in an embodiment of the present invention. DETAILED DESCRIPTION

[0057] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0058] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0059] Embodiment 1

[0060] In one or more embodiments, a method for dynamic inversion of near-surface greenhouse gas concentrations is disclosed, combining Figure 1 , specifically including the following process:

[0061] (1) Obtain greenhouse gas concentration data from near-surface observation stations, greenhouse gas concentration data from satellite remote sensing, meteorological data, geographic information data, and CO2 emission inventory data within a set time period in the target area.

[0062] In this embodiment, the greenhouse gas concentration data includes CO2 and CH4 gas concentration data.

[0063] As a specific example, the ground station data is obtained from the WDCGG (World Data Centre for Greenhouse Gases) monitoring station, collecting daily observation data of CO2 and CH4 gas concentrations at the WDCGG ground station.

[0064] Satellite remote sensing column concentration data can be selected from the global discrete strip data from the GOSAT L2 FTS (Greenhouse Gases Observing Satellite Level2 Fourier Transform Spectrometer) sensor, with a temporal and spatial resolution of monthly scale and 0.25°, to obtain the column concentration data of CO2 and CH4 (XCO2 and XCH4).

[0065] Meteorological data are derived from ERA5 meteorological reanalysis data, including temperature, wind speed, wind direction, evaporation, albedo, surface pressure, boundary layer height (BLH), sea surface temperature (SST), total rainfall and other meteorological parameters, with temporal and spatial resolutions of day and 0.25° respectively.

[0066] Geographic information data include: topography, population distribution, land use data, soil data, and NDVI (Normalized Difference Vegetation Index) data. The topographic data uses the SRTM (Shuttle Radar Topography Mission) 30-meter spatial resolution DEM (Digital Elevation Model) data; the population distribution data is obtained through the WorldPop dataset of global population data, and the land use data is obtained from the ESRI land use dataset with a spatial resolution of 10 meters; the soil data comes from the HWSD global soil dataset; and the NDVI data comes from the MODIS13Q1 data product with a spatial resolution of 1 km.

[0067] The CO2 emission inventory data comes from ODIAC Fossil Fuel CO2 Emissions Datasets, with a temporal resolution of monthly and a spatial resolution of 1 km.

[0068] (2) In terms of time, the daily average concentration data of the ground observation station is matched with the remote sensing data of the satellite passing on the same day; in terms of space, the satellite remote sensing data that overlaps with the ground observation station within the matching radius is matched.

[0069] In this embodiment, in order to ensure the consistency of all data in terms of time range and spatial resolution, a matching algorithm that takes both time and space weights into consideration is used to perform spatiotemporal matching to form a feature data set of unified spatiotemporal scale.

[0070] Specifically, the ground observation site data are matched with the satellite remote sensing data in time, specifically: the daily average concentration data of the ground observation site and the satellite remote sensing column concentration data passing in the same time period are matched in time.

[0071] Spatially, the satellite remote sensing data with spatial overlap near the ground observation site (within a radius of 0.25° longitude and latitude) are matched. The specific process is as follows:

[0072] On each day, the daily average concentration data of the ground observation station is matched with the satellite remote sensing data within a radius of 0.25° centered on the corresponding ground observation station; Figure 2-Figure 4 , the specific matching rules are as follows:

[0073] ① No match: If there is no spatially overlapping satellite remote sensing column concentration data within a matching radius of a ground observation site, the data of the ground observation site is considered to have no match and is not included in the dataset;

[0074] ② Single-point matching: If there is only one satellite remote sensing column concentration data point that overlaps with the observation value of the ground observation site within a matching radius, the remote sensing data is directly used for matching, that is, Single-joint: ; Among them, V wdcgg is the column concentration data matching the ground observation station data, V satellite It is the column concentration data from satellite remote sensing.

[0075] ③Multi-point matching situation: If two or more remote sensing column concentration data points spatially overlap with the observation values ​​of the ground observation site within a matching radius, they will be weighted according to the spatial distance.

[0076] Specifically, the distance between the ground observation site and each matching satellite remote sensing observation point (such as D1, D2, etc.) is calculated, and the inverse of the distance is used as the spatial weight; combined with the spatial weight, the concentration values ​​of multiple matching points are weighted averaged to obtain the final matching result, that is:

[0077] Multi-join:

[0078] ;

[0079] in, , , …, are the distances between the ground observation site and each remote sensing data observation point, , , …, are the observation values ​​of each remote sensing data observation point; is the number of satellite remote sensing data points corresponding to the data of the current ground observation site within the set matching radius; n≥2.

[0080] (3) A dataset was constructed using the spatiotemporally matched greenhouse gas concentration data (satellite remote sensing column concentration, ground concentration at observation sites), meteorological data, geographic information data, and CO2 emission inventory data. The spatiotemporal-classification gradient boosting decision tree (ST-CatBoost) algorithm was used for training and modeling to obtain the ST-CatBoost remote sensing inversion model, which outputs discrete strip-shaped near-ground greenhouse gas concentrations.

[0081] In this embodiment, the data set is first normalized and divided into a training set, a validation set, and a test set in a ratio of 7:2:1, and the spatiotemporal-classification gradient boosting decision tree (ST-CatBoost) remote sensing inversion model is trained; the model framework is constructed using the Python sklearn tool library, and the GridSearch method is used to optimize and determine the hyperparameters. The ten-fold cross-validation method is used to verify and evaluate the model prediction accuracy and determine the optimal ST-CatBoost remote sensing inversion model.

[0082] Among them, the spatiotemporal-classification gradient boosting decision tree (ST-CatBoost) remote sensing inversion model of this embodiment is based on the classic CatBoost (classification gradient boosting decision tree) algorithm, and incorporates time and space information, that is, variables representing time (day of the year) and space (longitude, latitude, altitude) are added to the CatBoost model to improve the simulation effect of the model.

[0083] Figure 5 and Figure 6 The accuracy evaluation results of the comparison between the CO2 and CH4 gas concentration inversion results and the WDCGG station observation data in this embodiment are given respectively, where RMSE represents the root mean square error, and the smaller the value, the higher the inversion accuracy of the model; 2 The index represents the determination coefficient. The closer the value is to 1, the higher the inversion accuracy of the model; N represents the number of data points for accuracy verification; Y=aX+b is the straight line fitted according to the data points, a and b are the fitting coefficients, respectively. The closer the straight line is to the baseline (dashed line in the figure), the higher the inversion accuracy of the model.

[0084] pass Figure 5 and Figure 6 It can be seen that the CO2 and CH4 concentration inversion results obtained by the ST-CatBoost remote sensing inversion model in this embodiment are R 2The indicators reached 0.84 and 0.87 respectively, and the difference between the fitting straight line and the baseline was small, indicating that the inversion accuracy of the model was high.

[0085] (4) A dataset was constructed using the discrete strip-shaped near-ground greenhouse gas concentrations output by the ST-CatBoost remote sensing inversion model, as well as meteorological data, geographic information data, and CO2 emission inventory data. The ST-CatBoost algorithm was used for training to obtain a spatial prediction model for near-ground greenhouse gas concentrations, which outputs a spatially continuous near-ground greenhouse gas concentration covering the target area.

[0086] In this embodiment, the spatial prediction model of near-ground greenhouse gas concentration is also trained and modeled based on the spatiotemporal-classification gradient boosting decision tree (ST-CatBoost) algorithm, with a symmetric decision tree as the base learner, adding time (day of the year) and space (longitude, latitude, altitude) information to the feature variables, and iterative optimization through gradient boosting.

[0087] The model training and optimization process is consistent with the method in step (3) and will not be described in detail.

[0088] (5) Using the trained ST-CatBoost remote sensing inversion model and the near-surface greenhouse gas concentration spatial prediction model, remote sensing inversion and spatial prediction are performed based on the latest GOSAT satellite remote sensing column concentration data, meteorological data, geographic information data, and CO2 emission inventory data. The specific process is as follows:

[0089] Obtain the latest GOSAT satellite remote sensing column concentration data, meteorological data, geographic information data, and CO2 emission inventory data;

[0090] In terms of time, the daily average concentration data of the ground observation station is matched with the remote sensing data of the satellite passing on the same day; in terms of space, the satellite remote sensing data that overlaps with the ground observation station within the matching radius is matched;

[0091] The greenhouse gas concentration data, meteorological data, geographic information data and CO2 emission inventory data after time and space matching are used as input, and the trained ST-CatBoost remote sensing inversion model is used to obtain the discrete strip-shaped near-ground greenhouse gas concentration;

[0092] The discrete strip-shaped near-ground greenhouse gas concentration, meteorological data, geographic information data and CO2 emission inventory data are used as input, and the trained near-ground greenhouse gas concentration spatial prediction model is used to obtain the spatially continuous near-ground greenhouse gas concentration covering the target area, and generate a spatial distribution map of near-ground greenhouse gas concentration.

[0093] In this embodiment, the inversion and prediction results of CO2 and CH4 gas concentrations are synchronously visualized by calling Basemap and Matplotlib tool libraries to output a spatial distribution map of near-ground CO2 and CH4 greenhouse gas concentrations.

[0094] As an optional solution, by detecting abnormal points in the spatial distribution map of near-ground greenhouse gas concentrations, information on near-ground high-value areas of greenhouse gases can be automatically extracted, output as a spatial distribution map of abnormal greenhouse gas emission sources, and generate a list of high-value points; detailed information on the high-value point list includes: administrative divisions, longitude, latitude, concentration, greenhouse gas types, detailed locations, etc.

[0095] The specific process of detecting abnormal points in the spatial distribution map of near-ground greenhouse gas concentration is as follows:

[0096] (5-1) Divide the spatial distribution map of near-surface greenhouse gas concentration into multiple spatial grid points, each of which corresponds to a near-surface greenhouse gas concentration data;

[0097] (5-2) randomly selecting a CO2 or CH4 concentration value and taking the concentration value as a split point to divide the data space into two sub-parts; for each sub-part, randomly selecting a CO2 or CH4 concentration value as a split point to divide the sub-part into two sub-parts;

[0098] (5-3) The process of (5-2) is iterated repeatedly until the CO2 or CH4 concentration value at a certain spatial grid point is isolated or reaches a predetermined depth, that is, the number of segmentation times reaches a preset threshold (usually log2N based on experience, where N is the amount of data).

[0099] (5-4) Identify the spatial grid points as high-value areas and extract information such as near-ground greenhouse gas concentration, longitude, latitude, and location of the high-value areas.

[0100] The present invention combines the ST-CatBoost remote sensing inversion model and the near-ground greenhouse gas concentration spatial prediction model to effectively fuse the model inversion results with the remote sensing satellite observation results, overcoming the limitations of traditional satellite remote sensing data that present strip-like distribution and discontinuous spatial coverage, and realizing the reconstruction of the spatiotemporal continuous distribution characteristics of greenhouse gas concentrations in the research area, providing comprehensive and reliable data support for regional scale greenhouse gas monitoring and assessment. This method not only improves the spatiotemporal integrity of greenhouse gas concentration products, but also improves the accuracy of monitoring results through the fusion of multi-source data.

[0101] Embodiment 2

[0102] In one or more embodiments, a dynamic inversion system for near-ground greenhouse gas concentration is disclosed, specifically comprising:

[0103] The data acquisition module is used to obtain greenhouse gas concentration data from ground observation stations within a set time period in the target area, greenhouse gas concentration data from satellite remote sensing, meteorological data, geographic information data, and CO2 emission inventory data;

[0104] The spatiotemporal matching module is used to match the daily average concentration data of the ground observation site with the remote sensing data of the satellite passing on the same day in terms of time; and to match the satellite remote sensing data that overlaps spatially within the matching radius set by the ground observation site in terms of space;

[0105] The remote sensing inversion module is used to take the greenhouse gas concentration data, meteorological data, geographic information data and CO2 emission inventory data after time and space matching as input, and use the trained ST-CatBoost remote sensing inversion model to obtain discrete strip-shaped near-ground greenhouse gas concentrations;

[0106] The spatial prediction module is used to take the discrete strip-shaped near-ground greenhouse gas concentration, meteorological data, geographic information data and CO2 emission inventory data as input, and use the trained near-ground greenhouse gas concentration spatial prediction model to obtain the spatially continuous near-ground greenhouse gas concentration covering the target area, and generate a near-ground greenhouse gas concentration spatial distribution map.

[0107] It should be noted that the specific implementation method of the above modules is exactly the same as that in Example 1 and will not be described in detail.

[0108] Embodiment 3

[0109] In one or more embodiments, a terminal device is disclosed, which includes a processor and a memory, wherein the processor is used to implement instructions; the memory is used to store multiple instructions, and the instructions are suitable for being loaded by the processor and executing the dynamic inversion method of near-ground greenhouse gas concentration described in Example 1.

[0110] It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general-purpose processors, digital signal processors DSP, application-specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0111] The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A portion of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0112] In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in a processor or an instruction in the form of software.

[0113] Embodiment 4

[0114] In one or more embodiments, a computer-readable storage medium is disclosed, in which a plurality of instructions are stored, wherein the instructions are suitable for being loaded by a processor of a terminal device and executing the dynamic inversion method of near-ground greenhouse gas concentration described in Example 1.

[0115] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A dynamic inversion method for near-ground greenhouse gas concentration, characterized in that: include: Obtain greenhouse gas concentration data from ground observation stations, satellite remote sensing greenhouse gas concentration data, meteorological data, geographic information data and CO2 emission inventory data within a set time period in the target area; Temporally match the daily average concentration data of ground observation sites with the remote sensing data of satellites passing on the same day; The satellite remote sensing data that overlaps spatially with the ground observation site within the matching radius set will be matched. If there are two or more satellite remote sensing column concentration data points that overlap spatially with the current ground observation site data within the matching radius set, the weighted average method will be used to match these satellite column concentration data, specifically: The distance between the ground observation site and each remote sensing column concentration data point is calculated, and the reciprocal of the distance is used as the spatial weight; The weighted average of the column concentration observations of each remote sensing data point is calculated based on the weights to obtain the final matching result: ； in, , , …, are the distances between the ground observation site and each remote sensing data point, , , …, are the observed values ​​of column concentration at each remote sensing data point; is the number of satellite remote sensing column concentration data points corresponding to the data of the current ground observation station within the set matching radius; The greenhouse gas concentration data, meteorological data, geographic information data and CO2 emission inventory data after time and space matching are used as input, and the trained ST-CatBoost remote sensing inversion model is used to obtain the discrete strip-shaped near-ground greenhouse gas concentration; The discrete strip-shaped near-ground greenhouse gas concentration, meteorological data, geographic information data and CO2 emission inventory data are used as input, and the trained near-ground greenhouse gas concentration spatial prediction model is used to obtain the spatially continuous near-ground greenhouse gas concentration covering the target area, and generate a near-ground greenhouse gas concentration spatial distribution map.

2. A method for dynamic inversion of near-ground greenhouse gas concentrations according to claim 1, characterized in that: The greenhouse gas concentration data include: satellite remote sensing column concentration data of CO2 and CH4 and ground concentration data of CO2 and CH4 at observation sites; the meteorological data include: temperature, wind speed, wind direction, evaporation, albedo, ground air pressure, boundary layer height, near-sea surface temperature and total rainfall data; geographic information data include topography, population distribution, land use data, soil data and normalized difference vegetation index data; CO2 emission inventory data include CO2 emissions of each location unit in the target area within a set time period.

3. A method for dynamic inversion of near-ground greenhouse gas concentrations as claimed in claim 1, characterized in that: The daily average concentration data of the ground observation station is matched with the remote sensing data of the satellite passing on the same day in terms of time, specifically: The daily ground observation station data are temporally matched with the satellite remote sensing column concentration data passing through in the same time period.

4. A method for dynamic inversion of near-ground greenhouse gas concentrations as claimed in claim 1, characterized in that: Spatially, it will match the satellite remote sensing data that overlaps with the ground observation site within the set radius, specifically: On each day, the ground observation site data are matched with the satellite remote sensing column concentration data within a set matching radius centered on the corresponding ground observation site; the specific matching method is as follows: If no satellite remote sensing column concentration data that overlaps with the current ground observation site data in space is found within the set matching radius, the current ground observation site data has no matching data and is not included in the data set; If there is only one satellite remote sensing data point that spatially overlaps with the data of the current ground observation site within the set matching radius, the greenhouse gas column concentration value of the satellite remote sensing data point will be used as the column concentration matching the ground observation site and included in the data set.

5. A method for dynamic inversion of near-ground greenhouse gas concentrations as claimed in claim 1, characterized in that: After obtaining the spatial distribution map of greenhouse gas concentration near the ground, it also includes: By detecting abnormal points in the spatial distribution map of near-ground greenhouse gas concentrations, information on near-ground high-value areas of greenhouse gases is automatically extracted, output as a spatial distribution map of abnormal greenhouse gas emission sources, and generate a list of high-value points.

6. A method for dynamic inversion of near-ground greenhouse gas concentrations as claimed in claim 5, characterized in that: The specific process of detecting abnormal points in the spatial distribution map of near-ground greenhouse gas concentration is as follows: Divide the spatial distribution map of near-ground greenhouse gas concentration into multiple spatial grid points, each of which corresponds to a near-ground greenhouse gas concentration data; Randomly select a CO2 or CH4 concentration value and use the concentration value as a split point to divide the data space into two sub-parts; for each sub-part, randomly select a CO2 or CH4 concentration value as a split point to divide the sub-part into two sub-parts; The above process is iterated repeatedly until the CO2 or CH4 concentration value at a certain spatial grid point is isolated or reaches a predetermined depth; The spatial grid points are identified as high-value areas, and near-ground greenhouse gas concentration, longitude and latitude, dimension and location information of the high-value areas are extracted.

7. A dynamic inversion system for near-surface greenhouse gas concentration, using a dynamic inversion method for near-surface greenhouse gas concentration as claimed in any one of claims 1 to 6, characterized in that: include: The data acquisition module is used to obtain greenhouse gas concentration data from ground observation stations within a set time period in the target area, greenhouse gas concentration data from satellite remote sensing, meteorological data, geographic information data, and CO2 emission inventory data; The spatiotemporal matching module is used to match the daily average concentration data of the ground observation site with the remote sensing data of the satellite passing on the same day in terms of time; and to match the satellite remote sensing data that overlaps spatially within the matching radius set by the ground observation site in terms of space; The remote sensing inversion module is used to take the greenhouse gas concentration data, meteorological data, geographic information data and CO2 emission inventory data after time and space matching as input, and use the trained ST-CatBoost remote sensing inversion model to obtain discrete strip-shaped near-ground greenhouse gas concentrations; The spatial prediction module is used to take the discrete strip-shaped near-ground greenhouse gas concentration, meteorological data, geographic information data and CO2 emission inventory data as input, and use the trained near-ground greenhouse gas concentration spatial prediction model to obtain the spatially continuous near-ground greenhouse gas concentration covering the target area, and generate a near-ground greenhouse gas concentration spatial distribution map.

8. A terminal device, comprising a processor and a memory, wherein the processor is used to implement instructions; and the memory is used to store multiple instructions, characterized in that: The instructions are suitable for being loaded by a processor and executing the dynamic inversion method for near-ground greenhouse gas concentration according to any one of claims 1-6.

9. A computer-readable storage medium storing a plurality of instructions, characterized in that: The instructions are suitable for being loaded by a processor of a terminal device and executing the dynamic inversion method for near-ground greenhouse gas concentration according to any one of claims 1 to 6.

Images