A multi-satellite cooperative carbon monitoring method and system and a storage medium

Abstract

The application discloses a kind of multi-satellite coordination carbon monitoring method, system and storage medium, it is related to satellite remote sensing monitoring technical field, the method according to the real-time orbit data of each target satellite, load performance, each target grid corresponding grid priority and the environmental state data in each target grid, determine each target satellite is located in the carbon monitoring distribution task of three-dimensional monitoring network;Control each target satellite according to respective corresponding carbon monitoring distribution task, obtain carbon monitoring data;According to the carbon monitoring data monitored by each target satellite and the ground monitoring data and meteorological data of same time registration, determine carbon concentration and carbon sink flux data.The method and system provided by the application, based on multiple target satellites cooperate in the carbon monitoring data in each target grid in three-dimensional monitoring network, and according to each target satellite orbit real-time parameter and grid priority, dynamically distribute observation task, shorten carbon monitoring period, improve monitoring accuracy.

Description

Technical Field

[0001] This invention relates to the field of satellite remote sensing monitoring technology, and in particular to a multi-satellite collaborative carbon monitoring method, system, and storage medium. Background Technology

[0002] Carbon monitoring is a monitoring system that uses integrated observation, numerical simulation, statistical analysis and other technical means to obtain data on greenhouse gas emission intensity, environmental concentration and carbon sink capacity, in order to serve the research and management of climate change.

[0003] Existing technologies utilize remote sensing platforms such as satellites and drones, equipped with high-precision greenhouse gas monitoring equipment, to achieve large-scale observation of the Earth's atmosphere and obtain information on the distribution of greenhouse gas concentrations. However, there are only a few carbon monitoring satellites currently in existence, and their capabilities vary significantly, with some being active and others passive (some satellites are unusable at night). This results in long carbon monitoring cycles and low monitoring efficiency.

[0004] Atmospheric carbon statistics are affected by a combination of factors, including carbon emissions and carbon absorption. Even though professional carbon monitoring satellite data is very accurate, it is still unable to account for the influence of human activities (such as thermal power generation), natural disasters (forest fires), and natural factors (such as atmospheric currents, typhoons, and volcanic eruptions) due to limitations in its orbit, monitoring swath width, and number of satellites.

[0005] To implement global carbon monitoring, a wider scope, more factors, and more real-time data synchronization are needed to achieve accurate statistics.

[0006] Therefore, the existing technology needs further improvement. Summary of the Invention

[0007] In view of the shortcomings of the prior art, the purpose of this invention is to provide a multi-satellite collaborative carbon monitoring method, system and storage medium to overcome the defects of long monitoring cycle and low monitoring efficiency in the prior art carbon monitoring technology.

[0008] The technical solution adopted by this invention to solve the technical problem is as follows:

[0009] Firstly, this application provides a multi-satellite collaborative carbon monitoring method, which includes:

[0010] Acquire real-time orbital data and payload performance data of each target satellite within the target monitoring area; wherein, the target satellites include carbon monitoring satellites, and at least one of hyperspectral satellites, multispectral satellites, visible light satellites, and SAR radar satellites;

[0011] Based on the real-time orbital data, payload performance, grid priority corresponding to each target grid, and environmental status data within each target grid, the carbon monitoring assignment tasks for each target satellite within the three-dimensional monitoring network are determined. The three-dimensional monitoring network is constructed based on carbon monitoring analysis information and includes multiple divided target grids. The carbon monitoring analysis information includes: distribution data of atmospheric targets, carbon emission or carbon sink monitoring priority setting data, and distribution grid data of different types of orbital satellites.

[0012] The target satellites are controlled to acquire carbon monitoring data according to their respective carbon monitoring assignment tasks.

[0013] Based on the carbon monitoring data monitored by each target satellite and the ground monitoring data and meteorological data registered at the same time, the carbon concentration and carbon flux data of the target monitoring area are determined.

[0014] Optionally, the construction steps of the three-dimensional monitoring network include:

[0015] A three-dimensional monitoring grid is constructed in horizontal and vertical dimensions. In the horizontal dimension, the target monitoring area is dynamically divided according to the preset carbon emission or carbon sink monitoring priority, resulting in multiple horizontal dimension grids. The location of each horizontal dimension grid is updated according to a preset period or in real time. In the vertical dimension, the atmosphere is divided into multiple vertical dimension levels based on the distribution data of target objects in the atmosphere and the distribution grid data of different types of orbital satellites. The three-dimensional monitoring grid is constructed based on each horizontal dimension grid and each vertical dimension level.

[0016] Optionally, the step of determining the carbon monitoring assignment task for each target satellite within the three-dimensional monitoring network based on the real-time orbital data, payload performance, grid priority corresponding to each target grid, and environmental state data within each target grid includes:

[0017] Based on the real-time orbital data and environmental status data of each target satellite, as well as the preset optimal observation window selection criteria, the effective observation windows corresponding to each target satellite in different target grids are selected.

[0018] Based on the selected effective observation windows, the payload performance and grid priority of each target satellite, and according to the preset dynamic task allocation algorithm, the carbon monitoring assignment tasks of each target satellite within each target grid are determined.

[0019] Optionally, the preset dynamic task allocation algorithm includes: a load balancing algorithm and a conflict elimination rule; the step of determining the carbon monitoring task allocation for each target satellite within each target grid based on the selected effective observation window, the payload performance of each target satellite, and the grid priority, and according to the preset dynamic task allocation algorithm, includes:

[0020] Based on the payload performance of each target satellite, a load balancing algorithm is used to assign daily monitoring tasks to each target satellite; wherein, the payload performance includes: remaining energy, storage capacity, and working duration;

[0021] In accordance with the conflict elimination rules, when multiple target satellites cover the same target grid, the priority of payload accuracy, orbit adaptability, and observation window duration is reduced in that order, and the most matching satellite with an error lower than the preset error value is selected.

[0022] Based on the most matched satellite within each target grid and the functional type of each most matched satellite, the carbon monitoring assignment task for each most matched satellite within its corresponding target grid is determined.

[0023] Optionally, the step of determining carbon concentration and carbon flux data based on carbon monitoring data monitored by each of the target satellites and simultaneously registered ground monitoring data and meteorological data includes:

[0024] Acquire ground monitoring data and meteorological data;

[0025] The carbon monitoring data, ground monitoring data, and meteorological data monitored by each of the target satellites are fused and calculated using a multi-source data fusion algorithm to obtain fused multi-satellite observation data;

[0026] The multi-satellite observation data is input into the trained carbon monitoring and prediction model to obtain the carbon concentration and carbon flux data output by the carbon monitoring and prediction model.

[0027] Optionally, the step of using a multi-source data fusion algorithm to fuse the carbon monitoring data, ground monitoring data, and meteorological data monitored by each of the target satellites to obtain the fused multi-satellite observation data includes:

[0028] For the carbon monitoring data monitored by target satellites with different spectral carriers, an adaptive spectral matching algorithm is used to extract signals and perform noise reduction to obtain satellite networking data.

[0029] Based on meteorological reanalysis data and ground-based wind station data, a preset wind speed correction model is used to correct the monitoring concentration of the target objects in each target grid in the satellite network data, and outlier removal is performed to obtain corrected satellite data for wind speed and concentration.

[0030] By aligning the satellite observation time in the wind speed and concentration correction satellite data with the day-night time series data of the target monitoring area, multi-satellite detection data is obtained.

[0031] Optionally, the training method for the carbon monitoring and prediction model includes:

[0032] A training dataset was constructed using measured data from key ground-based emission sources, data from ecological carbon sink sample plots, and data from calibration stations as the raw data.

[0033] Based on the training dataset, a closed-loop iterative training is performed on the preset neural network model. By dynamically adjusting the inversion parameters to minimize the prediction error, a trained carbon monitoring prediction model is obtained.

[0034] Optionally, the method further includes:

[0035] Acquire real-time orbital data, payload performance data, historical monitoring data accuracy, and data transmission protocols for one or more newly added target satellites;

[0036] Based on the real-time orbit data, payload performance data, historical monitoring data accuracy, and data transmission protocol of each newly added target satellite, compatibility and performance verifications are performed.

[0037] Based on the compatibility and performance verification results, the target grid is determined and the carbon monitoring assignment tasks for each target satellite in the three-dimensional monitoring network are updated.

[0038] Secondly, this application also provides a multi-satellite collaborative carbon monitoring system, comprising:

[0039] The satellite information acquisition module is used to acquire real-time orbital data and payload performance data of various target satellites within the target monitoring area; including at least one of hyperspectral satellites, multispectral satellites, visible light satellites, and SAR radar satellites.

[0040] The task allocation module is used to determine the carbon monitoring allocation tasks for each target satellite within the three-dimensional monitoring network based on the real-time orbital data, payload performance, grid priority corresponding to each target grid, and environmental status data within each target grid. The three-dimensional monitoring network is constructed based on carbon monitoring analysis information and includes multiple divided target grids. The carbon monitoring analysis information includes: distribution data of atmospheric targets, carbon emission or carbon sink monitoring priority setting data, and distribution grid data of different types of orbital satellites.

[0041] The carbon monitoring data acquisition module is used to control each of the target satellites to acquire carbon monitoring data according to their respective carbon monitoring assignment tasks;

[0042] The data statistics module is used to determine the carbon concentration and carbon flux data of the target monitoring area based on the carbon monitoring data monitored by each target satellite and the ground monitoring data and meteorological data registered at the same time.

[0043] Thirdly, this application also provides a computer storage medium, which is a computer-readable storage medium, wherein a computer program is stored in the computer storage medium, and when the computer program is executed by a computer, the computer is used to execute the multi-satellite collaborative carbon monitoring method.

[0044] Beneficial effects:

[0045] This invention provides a multi-satellite collaborative carbon monitoring method, system, and storage medium. It acquires real-time orbital data and payload performance data of each target satellite; determines carbon monitoring assignment tasks for each target satellite within a three-dimensional monitoring network based on the real-time orbital data, payload performance, grid priority of each target grid, and environmental state data within each target grid; controls each target satellite to acquire carbon monitoring data according to its corresponding carbon monitoring assignment task; and determines carbon concentration and carbon flux data based on the carbon monitoring data monitored by each target satellite and simultaneously registered ground monitoring data and meteorological data. The method and system provided by this invention utilize a constructed three-dimensional monitoring network to collaboratively monitor carbon data within each target grid of the three-dimensional monitoring network using multiple target satellites. Furthermore, it dynamically allocates observation tasks based on the real-time orbital parameters of each target satellite and grid priority, thereby shortening the carbon monitoring cycle and improving monitoring efficiency. Attached Figure Description

[0046] Figure 1 A flowchart illustrating the steps of a multi-satellite collaborative carbon monitoring method provided by this invention;

[0047] Figure 2 A schematic diagram illustrating the principle of constructing the three-dimensional monitoring network provided by this invention;

[0048] Figure 3 The present invention provides a multi-satellite collaborative observation time allocation diagram combining active and passive satellites;

[0049] Figure 4 This is a schematic diagram of the model training process provided by the present invention;

[0050] Figure 5 A schematic diagram illustrating the process of adaptive access to a new target satellite provided by the present invention;

[0051] Figure 6 The present invention provides a principle block diagram of a multi-satellite collaborative carbon monitoring system. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0053] It should be noted that although a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than that shown in the flowchart. The terms "first," "second," etc., used in the specification, claims, and the foregoing drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used herein is for the purpose of describing embodiments of the invention only and is not intended to limit the invention.

[0054] The core of current carbon neutrality accounting is a two-dimensional statistical approach involving both emissions and carbon sinks. Existing carbon neutrality accounting technologies primarily employ the following two types of statistical methods:

[0055] (1) Emission-side statistical methods, including inventory method, point source measurement method and satellite inversion method.

[0056] The inventory method is calculated by multiplying "activity data (such as fossil fuel consumption) by emission factors" and is the mainstream method for carbon accounting in various countries.

[0057] The point source measurement method involves installing online monitoring systems at key emission sources (thermal power plants, cement plants) to record emission data in real time with an accuracy of ±0.1ppm. However, it only covers a single source and lacks regional representativeness.

[0058] Satellite inversion method uses CO2 satellite observation data to infer regional emissions, making up for the lag in the inventory method (traditional inventories need to be updated every 1-2 years). However, the current single-satellite inversion accuracy is only ±1.2-1.5ppm, and the regional resolution is low (10km level).

[0059] (2) Statistical methods for carbon sinks, including: statistical methods combining field measurements and model estimation, remote sensing model methods and satellite carbon sink inversion methods.

[0060] The method of combining field measurements with model estimation involves setting up quadrats in forests or wetlands to measure biomass and soil carbon content, and then extrapolating regional carbon sinks using models such as InVEST (Integrated Assessment Model for Ecosystem Services and Trade-offs) and CENTURY, with an accuracy of ±8-15%. However, this method is time-consuming, labor-intensive, and results in uneven coverage.

[0061] Remote sensing modeling uses vegetation indices from satellites such as MODIS and Sentinel to estimate biomass carbon sinks. It has wide coverage, but its accuracy is greatly affected by vegetation type and climate.

[0062] Satellite carbon sink retrieval involves inverting carbon sink flux by fusing CO2 column concentration (XCO2) with ecological parameters. This CO2 column concentration corresponds to the concentration in a three-dimensional region between the ground and the satellite. Current technologies can only achieve quarterly-scale estimations, making it difficult to match the dynamic monitoring needs at the emission end.

[0063] While using satellites for carbon neutrality statistics offers numerous advantages, such as global coverage (addressing regional blind spots in traditional methods, including remote areas and transnational regions), dynamic monitoring (achieving near real-time (daily) data updates, superior to the annual lag of inventory methods), objectivity and neutrality (reducing human error in calculations and improving the credibility of carbon trading data), and integrated monitoring (simultaneously covering emission sources (cities, industrial areas) and carbon sinks (forests, wetlands), the number of satellites with carbon monitoring capabilities currently available is limited. Furthermore, the capabilities of these satellites vary significantly, with some satellites unusable for carbon monitoring at night. This results in long monitoring cycles and low monitoring efficiency.

[0064] Furthermore, in existing technologies, the revisit cycle of a single carbon monitoring satellite is 16-30 days, which cannot capture intraday fluctuations in emissions or carbon sinks, such as peak nighttime industrial emissions and daytime photosynthesis in vegetation, making it difficult to support dynamic carbon neutrality accounting. Moreover, passive satellites cannot operate at night, and the number of active satellites is limited (only 3 are operational globally), resulting in missing monitoring during peak human activity periods or insufficient coverage of key areas. Additionally, single-satellite data is affected by wind speed, clouds, and aerosols, leading to low accuracy in distinguishing emissions / carbon sinks and failing to meet the ±1ppm accuracy requirement for carbon trading. Therefore, existing carbon monitoring methods suffer from drawbacks such as long monitoring cycles, low accuracy, and uneven coverage.

[0065] To address the aforementioned problems in existing technologies, this invention proposes a multi-satellite collaborative carbon monitoring method, system, and storage medium. It constructs a three-dimensional monitoring network based on carbon monitoring analysis information; acquires real-time orbital data and payload performance data of each target satellite; determines the carbon monitoring allocation tasks for each target satellite within the three-dimensional monitoring network based on the real-time orbital data, payload performance, grid priority corresponding to each target grid, and environmental state data within each target grid; controls each target satellite to acquire carbon monitoring data according to its corresponding carbon monitoring allocation task; and determines carbon concentration and carbon flux data based on the carbon monitoring data monitored by each target satellite, concurrently registered ground monitoring data, and remote sensing time-series data reflecting the intensity of human activities. This invention, by constructing a three-dimensional monitoring network containing multiple target grids and integrating active and passive satellites to collaboratively and dynamically allocate observation tasks to each target grid, overcomes the limitations of traditional single-satellite or planar networking, thereby improving monitoring efficiency. On the other hand, due to the numerous interfering factors during carbon monitoring, the carbon measurement results of a single carbon monitoring satellite have significant deviations and cannot meet the accuracy requirements. Therefore, the method provided in this invention employs multi-satellite collaborative monitoring to eliminate interfering factors and improve the accuracy of carbon monitoring.

[0066] The following description, in conjunction with the accompanying drawings, provides a more detailed account of the multi-satellite collaborative carbon monitoring method, system, and storage medium disclosed in this embodiment.

[0067] Firstly, this application also provides a multi-satellite collaborative carbon monitoring method, such as... Figure 1 As shown, it includes:

[0068] Step S1: Obtain real-time orbital data and payload performance data of each target satellite within the target monitoring area; wherein, the target satellite includes a carbon monitoring satellite, and at least one of a hyperspectral satellite, a multispectral satellite, a visible light satellite, and a SAR radar satellite.

[0069] The target monitoring area in this step can be a specific region, such as a province or a large city, or it can be carbon monitoring on a global scale.

[0070] The target monitoring area may contain various types of remote sensing satellites, such as carbon monitoring satellites, multispectral satellites, hyperspectral satellites, visible light satellites, and SAR (Synthetic Aperture Radar) satellites. Among these, carbon monitoring satellites, multispectral satellites, and hyperspectral satellites share the core commonality of retrieving carbon-related information based on optical remote sensing. Their core differences lie in spectral resolution, spatial resolution, monitoring accuracy, and application scenarios. The core performance of carbon monitoring satellites lies in dedicated hyperspectral imaging and atmospheric column concentration measurement, while the core performance of hyperspectral satellites is general hyperspectral imaging and fine-grained component analysis or carbon sink analysis. The core performance of multispectral satellites is broad-spectral coverage and macroscopic carbon sink analysis.

[0071] The aforementioned carbon monitoring satellites, multispectral satellites, and hyperspectral satellites are all optical remote sensing satellites and share certain commonalities. For example, they all utilize the principle of solar radiation-atmosphere or surface reflection or absorption-satellite reception to retrieve carbon-related parameters through spectral characteristics. They all serve core carbon monitoring targets such as atmospheric CO2 concentration, terrestrial ecosystem carbon sinks (vegetation carbon sequestration), and land use or cover change (impacting the carbon cycle). They all require radiometric calibration, atmospheric correction, cloud or aerosol removal, and inversion models to obtain carbon concentration or carbon storage from the raw spectra. Furthermore, they all support carbon neutrality, carbon verification, carbon cycle research, and climate change assessment. However, these satellites differ in the following ways in their functions:

[0072] 1. There are differences in satellite positioning and payload design:

[0073] A. Carbon monitoring satellite (dedicated).

[0074] Positioning: A dedicated satellite for high-precision monitoring of global atmospheric CO2 column concentration.

[0075] Payload: The main payload is an Atmospheric Carbon dioxide Grating Spectrometer (ACGS), focusing on the CO2 absorption spectrum (0.76μm, 1.6μm, 2.06μm); the auxiliary payload is a Cloud and Aerosol Polarimetric Imager (CAPI), specifically designed to eliminate interference.

[0076] Bands: 3 core narrow bands + auxiliary multi-bands, with extremely high spectral resolution (0.04nm level).

[0077] B. Hyperspectral satellite (general purpose).

[0078] Positioning: General-purpose fine-spectral detection satellite, not dedicated to carbon detection.

[0079] Payload: Imaging spectrometer covering tens to hundreds of consecutive narrow bands (5–10 nm intervals), with fine sampling across the entire spectrum.

[0080] Bands: Continuous hyperspectral coverage (visible light – near-infrared – short-wave infrared), no dedicated carbon absorption channel, but can extract CO2, CH4 and vegetation biochemical components.

[0081] C. Multispectral satellite (general purpose).

[0082] Positioning: General-purpose broadband imaging satellite, primarily for macroscopic monitoring.

[0083] Payload: Multispectral camera with 3–13 discrete widebands (blue, green, red, near-infrared, short-wave infrared, etc.).

[0084] Bandwidth: Broad spectrum, few channels, discontinuous, lacks the ability to sample fine molecular absorption features.

[0085] D. Visible light satellite (general).

[0086] Positioning: A general-purpose optical imaging satellite, primarily designed to acquire the true colors and textures of ground features, focusing on visual interpretation and basic geographic information collection. Payload: A high-resolution visible light camera, typically employing a combination of panchromatic and multispectral imaging, with stereo imaging capabilities. Bands: Panchromatic band (0.45–0.90 μm, high spatial resolution) and standard red, green, and blue multispectral channels, with extremely high spatial resolution (sub-meter level), lower spectral resolution, and no gas detection capabilities.

[0087] E. SAR radar satellite (general purpose).

[0088] Positioning: Active microwave imaging satellite, capable of all-weather, day-and-night Earth observation, primarily used for deformation monitoring, surface classification, and penetrating imaging. Payload: Synthetic aperture radar, acquiring the backscattering coefficient of ground features by transmitting and receiving pulse signals, unaffected by clouds or rain. Bands: Typically operates at single or dual frequencies (e.g., L-band, C-band, X-band), utilizing polarization methods (single-polarization, dual-polarization, full polarization) to extract ground feature structure information; lacks spectral detection capabilities.

[0089] 2. There is a difference between spectral and spatial resolution.

[0090] Carbon monitoring satellites have extremely high spectral resolution (0.04 nm level), enabling them to accurately capture CO2 absorption lines. Hyperspectral satellites have high spectral resolution (5-10 nm), continuous narrow bands, and can identify material composition. Multispectral satellites, on the other hand, have low spectral resolution (tens to hundreds of nm), wide bands, and only macroscopic spectral features. Visible light satellites typically contain only 3-4 wide bands, including panchromatic (Pan) and red, green, and blue (RGB), with single-band bandwidths reaching tens to hundreds of nanometers. They cannot distinguish fine spectral features of ground objects and are mainly used to acquire natural true-color images. SAR radar satellites have no spectral resolution and do not collect ground object reflectance spectra, but they have high spatial resolution. Through pulse compression and synthetic aperture technology, their spatial resolution can reach meter-level to sub-meter level (1-3 m), and is unaffected by clouds, rain, or day / night cycles. However, unlike visible light satellites, their high resolution is reflected in the fine depiction of ground object structure, texture, deformation, and dielectric properties, rather than the visual clarity of optical images.

[0091] 3. The carbon monitoring capabilities and accuracy differ.

[0092] The core capability of carbon monitoring satellites is high-precision inversion of global atmospheric CO2 column concentration, with an accuracy better than 4 ppm. Therefore, their advantages include: dedicated channels and interference removal, resulting in the highest accuracy and best global consistency in atmospheric CO2 monitoring. Limitations include: low spatial resolution, making it unable to precisely monitor surface carbon sinks or emission sources; and difficulty in distinguishing between surface and anthropogenic emissions when only measuring atmospheric column concentration.

[0093] The core capabilities of hyperspectral satellites include: detailed inversion of vegetation biochemical parameters (chlorophyll, leaf area index, biomass) and carbon sink estimation. They can identify methane (CH4) and anthropogenic emission sources; and support detailed monitoring of soil organic carbon and wetland carbon. Their advantages include: high spectral resolution, the ability to distinguish different vegetation types and carbon pool components, and higher carbon sink estimation accuracy than multispectral satellites; they can also detect trace greenhouse gases. Limitations include: large data volume and complex processing; lower spatial resolution than multispectral satellites; and lower accuracy in atmospheric CO2 column concentration compared to dedicated carbon satellites.

[0094] The core capabilities of multispectral satellites include: macroscopic vegetation cover, normalized difference vegetation index (NDVI) or enhanced vegetation index (EDI), and estimation of regional carbon sink potential based on land use classification; large-scale carbon cycle monitoring. Their advantages include: high spatial resolution, fast revisit times, easy data acquisition, simple processing, and low cost; suitable for large-scale, long-term dynamic monitoring. Their limitations include: low spectral resolution, inability to capture fine CO2 absorption lines, and inability to directly retrieve atmospheric CO2 concentration; limited accuracy in vegetation carbon sink estimation, and difficulty in distinguishing fine land cover from carbon components.

[0095] Therefore, based on the differences between carbon monitoring satellites, hyperspectral satellites, multispectral satellites, visible light satellites, and SAR radar satellites, they can be applied to different scenarios. For example: carbon monitoring satellites are suitable for global or regional atmospheric CO2 concentration mapping, carbon source-flux inversion, international carbon verification, and climate change research. Hyperspectral satellites are suitable for detailed vegetation carbon sink monitoring, wetland or forest carbon pool assessment, methane emission source identification, soil organic carbon monitoring, and point source emission monitoring in mining areas or industries. Multispectral satellites are suitable for estimating total carbon emissions in a specific region, the impact of land use change on the carbon cycle, macro-ecosystem carbon dynamics, and long-term carbon trend monitoring. Visible light satellites can monitor forest cover and wildfires, while SAR radar satellites can observe vegetation growth and distinguish tree species. Additionally, visible light satellites can detect cloud images to estimate sunlight and rainfall.

[0096] Step S2: Based on the real-time orbital data, payload performance, grid priority corresponding to each target grid, and environmental status data within each target grid, determine the carbon monitoring allocation task for each target satellite within the three-dimensional monitoring network; wherein, the three-dimensional monitoring network is constructed based on carbon monitoring analysis information and includes multiple divided target grids; the carbon monitoring analysis information includes: distribution data of atmospheric targets, carbon emission or carbon sink monitoring priority setting data, and distribution grid data of different types of orbital satellites.

[0097] Before proceeding to this step, the process also includes: constructing a three-dimensional monitoring network based on carbon monitoring and analysis information; wherein, the carbon monitoring and analysis information includes: distribution information of atmospheric targets, priority information for carbon emission or carbon sink monitoring, and distribution area information of different types of orbital satellites; the three-dimensional monitoring network consists of multiple target grids.

[0098] This step establishes a three-dimensional monitoring network to incorporate the aforementioned different types of remote sensing satellites, enabling collaborative carbon monitoring by satellites with different functions. In specific implementation, the satellites in step S1 include active and passive satellites. Active satellites include lidar satellites, such as the "Atmosphere-1" satellite. Passive satellites include high-resolution or grating satellites, such as GOSAT-2, OCO-3, TanSat, Sentinel-5P, as well as hyperspectral, multispectral, and visible light satellites.

[0099] Furthermore, the three-dimensional monitoring network constructed in this step comprises multiple grids, including horizontal grids and vertical layers. The steps for constructing the three-dimensional monitoring network based on carbon monitoring and analysis information include:

[0100] A three-dimensional monitoring grid is constructed in horizontal and vertical dimensions. In the horizontal dimension, the target monitoring area is dynamically divided according to the preset carbon emission or carbon sink monitoring priority, resulting in multiple horizontal dimension grids. The location of each horizontal dimension grid is updated according to a preset period or in real time. In the vertical dimension, the atmosphere is divided into multiple vertical dimension levels based on the distribution data of target objects in the atmosphere and the distribution grid data of different types of orbital satellites. The three-dimensional monitoring grid is constructed based on each horizontal dimension grid and each vertical dimension level.

[0101] The three-dimensional monitoring network disclosed in this embodiment includes grid division based on the horizontal dimension and hierarchical division based on the vertical dimension. The division is based on carbon monitoring and analysis information, which is data used to assess and analyze the state, characteristics, and trends of carbon activity during carbon monitoring. This includes: distribution data of atmospheric targets, priority setting data for carbon emission or carbon sink monitoring, and grid data of different types of orbital satellites. The distribution data of atmospheric targets refers to the distribution information of atmospheric targets (such as gas composition, particulate matter, clouds, etc.), which exhibits significant stratification characteristics with altitude. For example, the atmosphere can be divided into a 0-2km near-surface layer, a 2-12km upper troposphere, and a 12-50km stratosphere, with differences in nitrogen, oxygen, and particulate matter in each layer. Priority setting data for carbon emission or carbon sink monitoring includes regional priority, industry priority, or technology priority. Regional priority can be urban and regional carbon monitoring priority or ecological restoration regional priority, etc. Industry priorities could prioritize key industries, such as those with concentrated emission sources like thermal power, steel, and cement. Technology priorities could prioritize high-precision monitoring technologies or the use of multi-source data fusion technologies for data analysis.

[0102] In detail, in one implementation, such as Figure 2 As shown, the horizontal grid division is dynamically based on carbon emission or carbon sink priorities. For example, the grid size for key emission source areas is 1km×1km×1km, for urban areas it is 5km×5km×5km, for ecological carbon sink areas it is 5km×5km×5km, and for oceans or deserts it is 10km×10km×10km. Vertically, the grid is stratified by matching atmospheric CO2 distribution with the spatial altitude of different types of orbiting satellites. For example, the layers matching atmospheric CO2 distribution correspond to: 0-2km near-surface layer, 2-12km upper troposphere, 12-50km stratosphere, with low-Earth orbit satellites (200-1000km) covering 0-12km and medium-high Earth orbit satellites (1000-35000km) covering 12-50km.

[0103] Target satellites are distributed within the defined target grid to perform the next carbon monitoring task. In this embodiment, the target satellites include both active and passive satellites. Figure 3 As shown, active satellites are carbon monitoring satellites, which can be lidar satellites. These lidar satellites are responsible for all-day observation of nighttime industrial emission areas and cloudy areas, providing vertical profile data. They can monitor key areas day and night, covering nighttime industrial emissions and cloudy areas to provide vertical profile detection data. Passive satellites are hyperspectral or grating satellites, which can only achieve daytime carbon monitoring, that is, observations during the peak period of human activity, from one hour after sunrise to one hour before sunset.

[0104] In step S1, after acquiring the real-time orbital data and payload performance data of each target satellite, carbon monitoring tasks are allocated accordingly based on preset grid priorities or environmental status data. Specifically, the real-time orbital data of each target satellite includes altitude, latitude and longitude, inclination, flight speed, and remaining fuel. The payload performance data of each target satellite includes spectral range, swath width, accuracy, and operational status. The environmental status data includes information such as cloud cover and wind speed within each target grid.

[0105] Furthermore, the step of determining the carbon monitoring assignment task for each target satellite within the three-dimensional monitoring network based on the real-time orbital data, payload performance, grid priority corresponding to each target grid, and environmental state data within each target grid includes:

[0106] Step S21: Based on the real-time orbital data and environmental status data of each target satellite, and the preset optimal observation window selection criteria, select the effective observation windows corresponding to each target satellite in different target grids.

[0107] In this step, the optimal observation window for each target satellite to each target grid is calculated using real-time satellite orbit data and environmental status data. In other words, the target satellite with the most accurate monitoring data is matched for each target grid, so as to obtain the carbon monitoring data of the target grid using the most matched target satellite.

[0108] Furthermore, when calculating the optimal observation window based on real-time satellite orbit data, an orbital dynamics model can be used. The conditions that the optimal observation window needs to meet include: longitude overlap greater than or equal to 90%, latitude coverage greater than or equal to 90%, solar altitude angle greater than or equal to 15° when it is a passive satellite, and cloud cover less than or equal to 30% within the current target grid.

[0109] The formula for determining observation window coverage is as follows:

[0110] ;

[0111] in, This represents the geocentric angle between the satellite and the target grid. or Indicates the latitude and longitude of the satellite's nadir point. and Indicates the latitude and longitude of the target grid center. This indicates the maximum coverage geocentric angle. In practical applications, the maximum coverage geocentric angle for active satellites is 0.5 degrees, and for passive satellites it is 1.2 degrees.

[0112] Step S22: Based on the selected effective observation windows, the payload performance and grid priority of each target satellite, and according to the preset dynamic task allocation algorithm, determine the carbon monitoring task allocation results of each target satellite within each target grid.

[0113] Furthermore, the preset dynamic task allocation algorithm includes: a load balancing algorithm and a conflict elimination rule; the step of determining the carbon monitoring task allocation result of each target satellite within each target grid based on the selected effective observation window, the payload performance of each target satellite, and the grid priority, and according to the preset dynamic task allocation algorithm, includes:

[0114] Based on the payload performance of each target satellite, a load balancing algorithm is used to assign daily monitoring tasks to each target satellite. Following conflict elimination rules, when multiple target satellites cover the same target grid, the priority of payload accuracy, orbit adaptability, and observation window duration is prioritized in descending order. The most matched satellite with an error lower than a preset error value is selected. Based on the most matched satellite within each target grid and its functional type, the carbon monitoring assignment task for each most matched satellite within its corresponding target grid is determined. The payload performance includes: remaining energy, storage capacity, and operating time.

[0115] The load balancing algorithm described in this step limits the number of observations per day for a single satellite to ≤ 70% of its total load to avoid overloading satellite energy and storage resources; the conflict resolution rules select observation satellites in order of "load accuracy > orbit adaptability > observation window duration", giving priority to satellites with monitoring errors ≤ 1ppm.

[0116] Step S3: Control each target satellite to acquire carbon monitoring data according to its corresponding carbon monitoring assignment task.

[0117] Once each target satellite has completed its assigned carbon monitoring task, it will collect the corresponding carbon monitoring data according to its assigned task.

[0118] Step S4: Based on the carbon monitoring data monitored by each target satellite and the ground monitoring data and meteorological data registered at the same time, determine the carbon concentration and carbon flux data within the target monitoring area.

[0119] In this step, after obtaining carbon monitoring data from multiple target satellites, the carbon concentration and carbon flux data are determined based on the carbon monitoring data.

[0120] In detail, the steps for determining the carbon concentration and carbon flux data within the target monitoring area based on carbon monitoring data detected by each target satellite and simultaneously registered ground monitoring data and meteorological data include:

[0121] Step S41: Obtain ground monitoring data and meteorological data that are registered with carbon monitoring data at the same time; use a multi-source data fusion algorithm to fuse and calculate the carbon monitoring data, ground monitoring data and meteorological data monitored by each of the target satellites to obtain fused multi-satellite observation data.

[0122] The prerequisite for using multi-source data fusion algorithms to perform fusion calculations on carbon monitoring data in this step is that the received carbon monitoring data needs to be processed first. This data processing includes: using adaptive spectral matching, wind speed interference correction, and other data processing methods, and utilizing day-night time-series data of the target monitoring area to perform fusion calculations on carbon monitoring data, ground monitoring data, and meteorological data monitored by each target satellite, in order to obtain multi-satellite observation data.

[0123] In detail, the step of using a multi-source data fusion algorithm to fuse and calculate the carbon monitoring data, ground monitoring data, and meteorological data monitored by each of the target satellites to obtain the fused multi-satellite observation data includes:

[0124] Step S411: For the carbon monitoring data monitored by target satellites with different spectral carriers, an adaptive spectral matching algorithm is used to extract signals and perform noise reduction to obtain satellite networking data.

[0125] First, the adaptive spectral matching algorithm standardizes the characteristic wavelengths of CO2 to 1594-1624nm and 2050-2060nm, and the characteristic wavelengths of CH4 to 1650-1660nm and 2300-2320nm, adapting to different types of payloads such as gratings, hyperspectral instruments, and FTS (Fourier Transform Spectrometer). In practical implementation, a combined algorithm of "wavelet transform + adaptive threshold filtering" is used to suppress background noise, ensuring that the signal-to-noise ratio (SNR) of the observed data is ≥1000:1, effectively removing instrument noise, cosmic ray interference, and background light interference.

[0126] Furthermore, different target satellites acquire corresponding carbon monitoring data. In this step, the acquired carbon monitoring data are first fused based on the different types of target satellites, and then adaptive spectral matching is used on the fused carbon monitoring data to obtain noise-processed satellite networking data.

[0127] Furthermore, in one implementation, a portion of the raw data monitored by satellites can be used as primary indicator data. This primary indicator data is processed as direct monitoring data to obtain satellite network data. The remaining raw data and factors monitored by satellites are transformed through algorithms to generate derived data and factors. These derived data and factors are then used as secondary indicator data and added to the algorithm model for calculating satellite network data, thereby improving the accuracy of the monitoring data. For example, visible light satellites can monitor deforestation to calculate carbon sequestration capacity. Alternatively, SAR radar satellites can monitor forest growth, directly accumulating observed growth height to calculate the increase in carbon sequestration.

[0128] In one implementation, after acquiring carbon monitoring satellite data covering the target monitoring area, as well as carbon monitoring data from hyperspectral satellites, multispectral satellites, visible light satellites, and SAR radar satellites, the multi-source satellite monitoring data is first spatiotemporally registered to align pixels from different data sources spatially and temporally. Secondly, feature parameters related to carbon emissions or carbon sinks are extracted from the multi-source satellite monitoring data. These feature parameters include at least one of spectral index, backscattering coefficient, and polarization decomposition parameters. The feature parameters are then input into a pre-trained deep learning fusion model, which outputs a carbon emission flux or carbon storage distribution map of the target area. The deep learning fusion model employs an attention mechanism to adaptively weight and fuse features from different data sources.

[0129] In the above embodiments, the deep learning fusion model can be generated by setting corresponding weight ratios for historical carbon monitoring data from different types of satellites based on factors such as the accuracy and importance of carbon monitoring data from different types of satellites, generating a training dataset, and then training based on the sample data in the training dataset.

[0130] Step S412: Based on meteorological reanalysis data and ground-based wind station data, a preset wind speed correction model is used to correct the monitoring concentration of the target objects in each target grid in the satellite network data, and outlier removal is performed to obtain corrected satellite data for wind speed and concentration.

[0131] Secondly, the wind speed correction model establishes a dynamic relationship of "grid CO2 concentration = observed value - wind speed × diffusion coefficient" by integrating ERA5 meteorological reanalysis data and ground wind station data. The diffusion coefficient is updated once per hour using the random forest algorithm.

[0132] Step S413: Align the satellite observation time in the wind speed and concentration correction satellite data with the day and night time series data in the target monitoring area to obtain multi-satellite detection data.

[0133] In this step, the day and night time series data within the target monitoring area are based on the global sunrise and sunset database, with observation periods dynamically divided by latitude and season: 6:00-20:00 (daytime) and 20:00-6:00 (nighttime) in the Northern Hemisphere summer, and 7:00-17:00 (daytime) and 17:00-7:00 (nighttime) in the Northern Hemisphere winter.

[0134] In practice, multi-satellite detection data includes nighttime multi-source remote sensing data, which includes: nighttime light data involving human activities, sea surface temperature data, tidal data, ocean current data, etc. Therefore, when calculating carbon emissions, nighttime light data involving human activities can be fused with nighttime sea surface temperature and ocean current data to distinguish between anthropogenic carbon emissions and natural carbon emissions.

[0135] The above steps sequentially extract, correct, and time-align the carbon monitoring data from each target satellite to obtain multi-satellite observation data.

[0136] Step S42: Input the multi-satellite observation data into the trained carbon monitoring and prediction model to obtain the carbon concentration and carbon flux data output by the carbon monitoring and prediction model.

[0137] The multi-satellite observation data calculated in the above steps are input into the trained carbon monitoring and prediction model to obtain the carbon concentration and carbon flux data output by the carbon monitoring and prediction model.

[0138] Furthermore, the training method for the carbon monitoring and prediction model includes:

[0139] Step S401: Construct a training dataset using measured data from key ground emission sources, ecological carbon sink sample plot data, and calibration station data as raw data.

[0140] Building a high-quality training dataset is a crucial step in achieving accurate carbon emission monitoring and assessment, and in contributing to the attainment of carbon neutrality goals. The following details the specific steps for constructing a training dataset using measured data from key ground-based emission sources, ecological carbon sink sample plots, and calibration station data as raw data:

[0141] The first aspect is the collection of measured data from key ground-based emission sources.

[0142] The scope of emission sources was determined, encompassing thousands of key ground-based emission sources across major industries such as power, steel, chemicals, and building materials. The selection of these emission sources took into full consideration factors such as their emission scale, industry representativeness, and the degree of their environmental impact.

[0143] Install monitoring equipment: Install advanced gas analyzers, flow meters, and other monitoring equipment at the selected emission sources. Gas analyzers can accurately measure the concentration of greenhouse gases such as carbon dioxide and methane in the emitted gases, while flow meters are used to record the emission flow rate in real time. All monitoring equipment must undergo rigorous calibration and verification to ensure that its measurement accuracy meets the requirements of relevant standards.

[0144] Data Acquisition and Transmission: An automated data acquisition system collects monitoring data from emission sources in real time at set time intervals (e.g., every minute or hour). The collected data is transmitted to a data center server for storage and management via wired or wireless communication networks (e.g., Ethernet, 4G / 5G networks). Encryption technology is used during data transmission to ensure data security and integrity.

[0145] The second aspect is the collection of data from ecological carbon sink sample plots.

[0146] Sample plot selection: Based on different climate zones (such as tropical, subtropical, temperate, and cold temperate zones), topography (such as mountains, plains, hills, and basins), and vegetation types (such as forests, grasslands, and wetlands) in China, hundreds of representative ecological carbon sink sample plots were selected. The selection of sample plots followed the principles of random and uniform distribution to ensure that they could comprehensively reflect the carbon sink characteristics of my country's ecosystems.

[0147] Plot survey and monitoring: Multiple survey plots were set up within each plot, and methods such as quadrat surveys and transect surveys were used to conduct detailed investigations into the species, quantity, biomass, and growth status of vegetation. Simultaneously, soil temperature and humidity sensors, soil carbon flux monitors, and other equipment were installed to monitor soil temperature, humidity, organic carbon content, and carbon absorption and release rates in real time. The survey and monitoring work was conducted according to unified standards and specifications to ensure the accuracy and comparability of the data.

[0148] Data Recording and Processing: Detailed records of data obtained from plot surveys and monitoring are kept, including survey time, location, basic plot information, vegetation survey data, and soil monitoring data. The recorded data undergoes preliminary processing and review, eliminating erroneous data and outliers to prepare for subsequent data processing and analysis.

[0149] Thirdly, data collection at calibration sites.

[0150] Calibration site layout: Multiple calibration sites are strategically located in key areas across the country (such as cities along air pollution transmission corridors and ecologically sensitive areas). The site selection for calibration sites fully considers the impact of the surrounding environment, avoiding interference from local pollution sources to ensure that the collected data accurately reflects the environmental background values ​​of the region.

[0151] Monitoring items and equipment: The calibration site is equipped with high-precision meteorological parameter monitoring equipment (such as anemometers, wind vanes, thermometers, hygrometers, etc.) and environmental gas monitoring equipment (such as air quality monitors, etc.) to monitor meteorological parameters (wind speed, wind direction, temperature, humidity, etc.) and the concentration of greenhouse gases in the background environment in real time.

[0152] Data Acquisition and Storage: The data acquisition frequency at calibration sites is matched with that of key ground-based emission sources and ecological carbon sink sample plots to ensure data temporal consistency. Collected data is transmitted to a data center server via a dedicated network for storage, and a robust data backup mechanism is in place to prevent data loss.

[0153] The collected ground-based key emission source measured data, ecological carbon sink sample plot data, and calibration station data were preprocessed to correct or remove outliers, and the data in the original data were formatted to obtain the original dataset. The data in the original dataset were then labeled to obtain sample data. This labeling could involve classifying the data as different carbon emission levels (e.g., high emissions, medium emissions, low emissions) or different carbon sink function types (e.g., strong carbon sink, medium carbon sink, weak carbon sink), etc.

[0154] In this embodiment, carbon concentration data annotation includes spatial information annotation, temporal information annotation, and concentration value annotation. Spatial information annotation includes the method of geographic coordinate annotation and spatial resolution. Geographic coordinate annotation involves accurately labeling the latitude and longitude coordinates of each target grid cell, as well as recording the coordinates of the center point or the four corner points of the corresponding target grid. Spatial resolution is explicitly stated as 1-10km in the original carbon concentration data dataset, and the spatial resolution level of each data point is recorded in the annotation file. For example, data with a 5km resolution is labeled "Spatial Resolution: 5km".

[0155] Time information annotation can be done using timestamps. Timestamp annotations add a precise timestamp to each carbon concentration data point, using the ISO 8601 standard, such as "YYYY-MM-DDTHH:MM:SS", for example, "2024-07-20T14:30:00". Each time information annotation also contains time resolution information, that is, the time resolution is indicated in each data point as 15 minutes to 24 hours, and the time resolution of each data point is recorded in the annotation file. For example, for data with a 1-hour time resolution, it is labeled "Time Resolution: 1 hour".

[0156] Concentration value labeling includes both actual concentration value labeling and precision labeling: Record the actual measured carbon concentration value at each sampling point or time point, in ppm (parts per million). For example, labeling would be "Carbon concentration: 410.5 ppm". Precision labeling specifies in the labeling file that the data precision range is ±0.5-1.0 ppm, and the corresponding precision value is indicated next to each data point. For example, for a data point with a precision of ±0.7 ppm, labeling would be "Precision: ±0.7 ppm".

[0157] In addition, data format adaptation annotations can be performed during data annotation. This means explicitly stating in the dataset that the data supports NetCDF, CSV, and JSON formats. If the data is stored in multiple formats, the format corresponding to each data file is recorded in the annotation file. For example, for a NetCDF format file, it would be labeled "Data Format: NetCDF".

[0158] Carbon sink flux data labeling includes regional identification labeling, flux value labeling, and time range labeling.

[0159] 1. Regional identification and labeling includes ground area labeling and regional boundary description. Geographic area labeling: This clearly identifies the geographic area corresponding to the carbon flux data. This can be a specific administrative division (such as province, city, or county), a nature reserve, or a specific ecosystem area. For example, it could be labeled "Geographic Area: XX City, XX District". Regional boundary description provides the boundary coordinates or a map screenshot to more accurately determine the area's extent.

[0160] Flux value labeling refers to the actual flux value: recording the actual measured value of carbon sink flux for each region, with the unit determined according to specific circumstances, such as grams of carbon per square meter per year (gC / m² / a). For example, it would be labeled "Carbon sink flux: 200 gC / m² / a". Precision labeling specifies in the labeling document that the data precision range is ±3%-8%, and indicates the corresponding precision percentage next to each data point. For example, for a data point with a precision of ±5%, it would be labeled "Precision: ±5%".

[0161] The time range label includes a statistical time label, which is used to indicate the start and end times of the time period. For example, it can be labeled as "Statistical Time: 2020-01-01 to 2020-12-31".

[0162] Similar to carbon concentration data, the metadata specifies the supported data formats (NetCDF, CSV, JSON), and the annotation file records the format of each data file and the carbon trading platform interface adaptation information.

[0163] Step S402: Based on the training dataset, perform closed-loop iterative training on the preset neural network model, and minimize the prediction error by dynamically adjusting the inversion parameters to obtain the trained carbon monitoring prediction model.

[0164] like Figure 4 As shown, after labeling all data points in the original dataset, a training dataset is constructed. This dataset is then used to train a pre-defined neural network model, resulting in a trained carbon monitoring and prediction model. The training data from the dataset, after data matching and processing, is used for closed-loop iterative training to obtain the final carbon monitoring and prediction model. In practice, this pre-defined neural network model can be a fusion of random forest and LSTM (Long Short-Term Memory) models. Input parameters include spectral absorption coefficient, atmospheric correction coefficient, and vertical stratification weights. Iterative optimization is performed every 24 hours, and the carbon concentration monitoring error in key areas converges to ≤0.8 ppm.

[0165] In one embodiment, the data in the training sample set includes: CEMS (Continuous Emission Monitoring System) data for key emission sources (accuracy ±0.1ppm), eddy covariance flux tower data from ecological carbon sink sample plots, TCCON (Total Carbon Column Observing Network) station data (accuracy ±0.3ppm), and UAV remote sensing survey data. Data matching is performed using a three-dimensional alignment of "grid ID-timestamp-vertical stratification". The output data of the carbon monitoring and prediction model includes: carbon concentration data (spatial resolution 1-10km, temporal resolution 15 minutes-24 hours, accuracy ±0.5-1.0ppm); carbon sink flux data (accuracy ±3%-8%); emission source / carbon sink area identification reports with a differentiation accuracy ≥90%; the model output data formats support NetCDF, CSV, and JSON, and are compatible with carbon trading platform data interfaces.

[0166] To further reduce the monitoring cycle and improve carbon monitoring efficiency, combined with Figure 5 As shown, this embodiment also determines whether there are any new satellites with carbon monitoring capabilities within the space to be monitored. If so, the new target satellite with carbon monitoring capabilities is added to the monitoring network to improve the monitoring efficiency by using a larger number of target satellites. The method further includes:

[0167] Step S01: Obtain real-time orbit data, payload performance data, historical monitoring data accuracy, and data transmission protocol for one or more newly added target satellites.

[0168] Step S02: Based on the real-time orbit data, payload performance data, historical monitoring data accuracy, and data transmission protocol of each newly added target satellite, perform compatibility verification and performance verification.

[0169] In this step, the performance verification standard for adaptive access of the new target satellite is as follows: select three overlapping observation areas (including key emission sources, ecological carbon sink areas, and calibration sites), and the observation data error between the new target satellite and the target satellites in the existing network is ≤1.5ppm; the data transmission protocols supported by the compatibility verification include TCP / IP, UDP (User Datagram Protocol), and FTP (File Transfer Protocol), and are adapted to new satellite orbits such as low Earth orbit constellations and inclined orbits.

[0170] Step S03: Based on the compatibility verification and performance verification results, determine the target grid to be allocated, and update the carbon monitoring task allocation for each target satellite in the three-dimensional monitoring network.

[0171] When the number of satellites in the network reaches 20, including 5 active satellites and 15 passive satellites, the revisit period for key areas is ≤1 hour, and the global average revisit period is ≤6 hours; the observation swath width of active satellites is ≥20km, the observation swath width of passive satellites is ≥50km, and the orbital coverage range is 60°N-60°S.

[0172] It is conceivable that, in practical implementation, the satellite's on-orbit status, payload working status, and data transmission status can be monitored in real time. When a satellite malfunctions or data is abnormal, a task reassignment mechanism will be automatically triggered to ensure the continuity of grid observation.

[0173] Secondly, this application also provides a multi-satellite collaborative carbon monitoring system, such as... Figure 6 ,include:

[0174] The satellite information acquisition module 610 is used to acquire real-time orbital data and payload performance data of each target satellite within the target monitoring area; wherein, the target satellite includes a carbon monitoring satellite, as well as at least one of a hyperspectral satellite, a multispectral satellite, a visible light satellite, and a SAR radar satellite; its function is as described in step S1.

[0175] The task allocation module 620 is used to determine the carbon monitoring allocation tasks for each target satellite within the three-dimensional monitoring network based on the real-time orbital data, payload performance, grid priority corresponding to each target grid, and environmental status data within each target grid. The three-dimensional monitoring network is constructed based on carbon monitoring analysis information and includes multiple divided target grids. The carbon monitoring analysis information includes: distribution data of atmospheric targets, carbon emission or carbon sink monitoring priority setting data, and distribution grid data of different types of orbital satellites. Its function is as described in step S2.

[0176] The carbon monitoring data acquisition module 630 is used to control each of the target satellites to acquire carbon monitoring data according to their respective carbon monitoring assignment tasks; its function is as described in step S3.

[0177] The data statistics module 640 is used to determine the carbon concentration and carbon flux data of the target monitoring area based on the carbon monitoring data monitored by each target satellite and the ground monitoring data and meteorological data registered at the same time. Its function is as in step S4.

[0178] Thirdly, this application also provides a computer storage medium, which is a computer-readable storage medium, wherein a computer program is stored in the computer storage medium, and when the computer program is executed by a computer, the computer is used to execute the multi-satellite collaborative carbon monitoring method.

[0179] This invention provides a multi-satellite collaborative carbon monitoring method, system, and storage medium. Based on the real-time orbital data, payload performance, grid priority of each target grid, and environmental state data within each target grid, the carbon monitoring assignment tasks for each target satellite within a three-dimensional monitoring network are determined. Each target satellite is controlled to acquire carbon monitoring data according to its corresponding carbon monitoring assignment tasks. Based on the carbon monitoring data monitored by each target satellite and simultaneously registered ground monitoring data and meteorological data, carbon concentration and carbon flux data are determined. The method and system provided by this invention construct a three-dimensional monitoring network, collaboratively monitor carbon data within each target grid of the three-dimensional monitoring network based on multiple target satellites, and dynamically allocate observation tasks according to the real-time orbital parameters of each target satellite and grid priority, thereby shortening the carbon monitoring cycle and improving monitoring accuracy.

[0180] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.

[0181] The embodiments described above are merely examples of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application.

Claims

1. A multi-satellite collaborative carbon monitoring method, characterized in that, include: Acquire real-time orbital data and payload performance data of each target satellite within the target monitoring area; wherein, the target satellites include carbon monitoring satellites, and at least two of the following: hyperspectral satellites, multispectral satellites, visible light satellites, and SAR radar satellites; Based on the real-time orbital data, payload performance, grid priority corresponding to each target grid, and environmental status data within each target grid, the carbon monitoring assignment tasks for each target satellite within the three-dimensional monitoring network are determined. The three-dimensional monitoring network, constructed based on carbon monitoring analysis information, includes multiple divided target grids. The carbon monitoring analysis information includes: distribution data of atmospheric targets, carbon emission or carbon sink monitoring priority setting data, and distribution grid data of different types of orbital satellites. Real-time orbital data for each target satellite includes: altitude, latitude and longitude, inclination, flight speed, and remaining fuel data; payload performance data for each target satellite includes spectral range, swath width, accuracy, and operational status data; environmental status data includes cloud cover and wind speed information within each target grid. The construction steps of the three-dimensional monitoring network include: A three-dimensional monitoring grid is constructed in horizontal and vertical dimensions. In the horizontal dimension, target monitoring areas are dynamically divided according to preset carbon emission or carbon sink monitoring priorities, resulting in multiple horizontal dimension grids. The location of each horizontal dimension grid is updated according to a preset period or in real time. In the vertical dimension, the atmosphere is divided into multiple vertical dimension levels based on the distribution data of target objects in the atmosphere and the distribution grid data of different types of orbital satellites. A three-dimensional monitoring grid is constructed based on each horizontal dimension grid and each vertical dimension level. The step of determining the carbon monitoring assignment task for each target satellite within the three-dimensional monitoring network based on the real-time orbit data, payload performance, grid priority corresponding to each target grid, and environmental state data within each target grid includes: Based on the real-time orbital data and environmental status data of each target satellite, and the preset optimal observation window selection criteria, effective observation windows for each target satellite in different target grids are selected. The optimal observation window must meet the following conditions: longitude overlap greater than or equal to 90%, latitude coverage greater than or equal to 90%, solar altitude angle greater than or equal to 15° when the satellite is a passive satellite, and cloud cover less than or equal to 30% within the current target grid. Based on the selected effective observation windows, the payload performance of each target satellite, and grid priority, and according to a preset dynamic task allocation algorithm, the carbon monitoring assignment tasks for each target satellite within each target grid are determined. This includes: assigning daily monitoring tasks to each target satellite based on its payload performance using a load balancing algorithm; wherein, the payload performance includes: remaining energy, storage capacity, and operating time; the load balancing algorithm in the dynamic task allocation limits the number of observations per satellite per day to ≤ 70% of its total load. And according to the conflict resolution rules, when multiple target satellites cover the same target grid, the priority of payload accuracy, orbit adaptability and observation window duration are in descending order, and the best matching satellite with an error lower than the preset error value is selected; wherein, the conflict resolution rules select observation satellites in order of payload accuracy greater than orbit adaptability, orbit adaptability greater than observation window duration, and select satellites with monitoring error ≤1ppm; Based on the most matched satellite within each target grid and the functional type of each most matched satellite, determine the carbon monitoring assignment task for each most matched satellite within its corresponding target grid. The target satellites are controlled to acquire carbon monitoring data according to their respective carbon monitoring assignment tasks. Based on the carbon monitoring data monitored by each target satellite and the ground monitoring data and meteorological data registered at the same time, determine the carbon concentration and carbon flux data within the target monitoring area; The method further includes: Acquire real-time orbital data, payload performance data, historical monitoring data accuracy, and data transmission protocols for one or more newly added target satellites; Based on the real-time orbit data, payload performance data, historical monitoring data accuracy, and data transmission protocol of each newly added target satellite, compatibility and performance verifications are performed. Based on the compatibility and performance verification results, the target grid is determined and the carbon monitoring assignment tasks for each target satellite in the three-dimensional monitoring network are updated. The step of determining the carbon concentration and carbon flux data within the target monitoring area based on the carbon monitoring data monitored by each target satellite and the simultaneously registered ground monitoring data and meteorological data includes: Acquire ground monitoring data and meteorological data; The carbon monitoring data, ground monitoring data, and meteorological data monitored by each of the target satellites are fused and calculated using a multi-source data fusion algorithm to obtain fused multi-satellite observation data; The multi-satellite observation data is input into the trained carbon monitoring and prediction model to obtain the carbon concentration and carbon flux data output by the carbon monitoring and prediction model. The training method for the carbon monitoring and prediction model includes: A training dataset was constructed using measured data from key ground-based emission sources, data from ecological carbon sink sample plots, and data from calibration stations as the raw data. Based on the training dataset, a closed-loop iterative training is performed on the preset neural network model. The prediction error is minimized by dynamically adjusting the inversion parameters, and the trained carbon monitoring prediction model is obtained. The preset neural network model adopts a fusion model of random forest and LSTM, and the carbon monitoring prediction model output supports NetCDF, CSV, and JSON. The carbon monitoring and prediction model assigns corresponding weights to historical carbon monitoring data from different types of satellites to generate a training dataset, and then trains the model based on the sample data in the training dataset.

2. The multi-satellite collaborative carbon monitoring method according to claim 1, characterized in that, The step of fusing and calculating the carbon monitoring data, ground monitoring data, and meteorological data monitored by each of the target satellites using a multi-source data fusion algorithm to obtain the fused multi-satellite observation data includes: For the carbon monitoring data monitored by target satellites with different spectral carriers, an adaptive spectral matching algorithm is used to extract signals and perform noise reduction to obtain satellite networking data. Based on meteorological data and ground-based wind station data, a preset wind speed correction model is used to correct the monitoring concentration of target objects in each target grid in the satellite network data, and outlier removal is performed to obtain corrected satellite data for wind speed and concentration. By aligning the satellite observation time in the wind speed and concentration correction satellite data with the day-night time series data of the target monitoring area, multi-satellite detection data is obtained.

3. A multi-satellite collaborative carbon monitoring system, characterized in that, include: The satellite information acquisition module is used to acquire real-time orbital data and payload performance data of each target satellite in the target monitoring area; wherein, the target satellites include carbon monitoring satellites, and at least two of the following: hyperspectral satellites, multispectral satellites, visible light satellites, and SAR radar satellites; The task allocation module is used to determine the carbon monitoring allocation tasks for each target satellite within the three-dimensional monitoring network based on the real-time orbital data, payload performance, grid priority corresponding to each target grid, and environmental status data within each target grid. The three-dimensional monitoring network is constructed based on carbon monitoring analysis information and includes multiple divided target grids. The carbon monitoring analysis information includes: distribution data of atmospheric targets, carbon emission or carbon sink monitoring priority setting data, and distribution grid data of different types of orbital satellites. Real-time orbital data for each target satellite includes: altitude, latitude and longitude, inclination, flight speed, and remaining fuel data; payload performance data for each target satellite includes spectral range, swath width, accuracy, and operational status data; environmental status data includes cloud cover and wind speed information within each target grid. The construction steps of the three-dimensional monitoring network include: A three-dimensional monitoring grid is constructed in horizontal and vertical dimensions. In the horizontal dimension, target monitoring areas are dynamically divided according to preset carbon emission or carbon sink monitoring priorities, resulting in multiple horizontal dimension grids. The location of each horizontal dimension grid is updated according to a preset period or in real time. In the vertical dimension, the atmosphere is divided into multiple vertical dimension levels based on the distribution data of target objects in the atmosphere and the distribution grid data of different types of orbital satellites. A three-dimensional monitoring grid is constructed based on each horizontal dimension grid and each vertical dimension level. The step of determining the carbon monitoring assignment task for each target satellite within the three-dimensional monitoring network based on the real-time orbit data, payload performance, grid priority corresponding to each target grid, and environmental state data within each target grid includes: Based on the real-time orbital data and environmental status data of each target satellite, and the preset optimal observation window selection criteria, effective observation windows for each target satellite in different target grids are selected. The optimal observation window must meet the following conditions: longitude overlap greater than or equal to 90%, latitude coverage greater than or equal to 90%, solar altitude angle greater than or equal to 15° when the satellite is a passive satellite, and cloud cover less than or equal to 30% within the current target grid. Based on the selected effective observation windows, the payload performance of each target satellite, and grid priority, and according to a preset dynamic task allocation algorithm, the carbon monitoring assignment tasks for each target satellite within each target grid are determined. This includes: assigning daily monitoring tasks to each target satellite based on its payload performance using a load balancing algorithm; wherein, the payload performance includes: remaining energy, storage capacity, and operating time; the load balancing algorithm in the dynamic task allocation limits the number of observations per satellite per day to ≤ 70% of its total load. And according to the conflict resolution rules, when multiple target satellites cover the same target grid, the priority of payload accuracy, orbit adaptability and observation window duration are in descending order, and the best matching satellite with an error lower than the preset error value is selected; wherein, the conflict resolution rules select observation satellites in order of payload accuracy greater than orbit adaptability, orbit adaptability greater than observation window duration, and select satellites with monitoring error ≤1ppm; Based on the most matched satellite within each target grid and the functional type of each most matched satellite, determine the carbon monitoring assignment task for each most matched satellite within its corresponding target grid. The carbon monitoring data acquisition module is used to control each of the target satellites to acquire carbon monitoring data according to their respective carbon monitoring assignment tasks; The data statistics module is used to determine the carbon concentration and carbon flux data within the target monitoring area based on the carbon monitoring data monitored by each target satellite and the ground monitoring data and meteorological data registered at the same time. The system also includes: Acquire real-time orbital data, payload performance data, historical monitoring data accuracy, and data transmission protocols for one or more newly added target satellites; Based on the real-time orbit data, payload performance data, historical monitoring data accuracy, and data transmission protocol of each newly added target satellite, compatibility and performance verifications are performed. Based on the compatibility and performance verification results, the target grid is determined and the carbon monitoring assignment tasks for each target satellite in the three-dimensional monitoring network are updated. The step of determining the carbon concentration and carbon flux data within the target monitoring area based on the carbon monitoring data monitored by each target satellite and the simultaneously registered ground monitoring data and meteorological data includes: Acquire ground monitoring data and meteorological data; The carbon monitoring data, ground monitoring data, and meteorological data monitored by each of the target satellites are fused and calculated using a multi-source data fusion algorithm to obtain fused multi-satellite observation data; The multi-satellite observation data is input into the trained carbon monitoring and prediction model to obtain the carbon concentration and carbon flux data output by the carbon monitoring and prediction model. The training method for the carbon monitoring and prediction model includes: A training dataset was constructed using measured data from key ground-based emission sources, data from ecological carbon sink sample plots, and data from calibration stations as the raw data. Based on the training dataset, a closed-loop iterative training is performed on the preset neural network model. The prediction error is minimized by dynamically adjusting the inversion parameters, and the trained carbon monitoring prediction model is obtained. The preset neural network model adopts a fusion model of random forest and LSTM, and the carbon monitoring prediction model output supports NetCDF, CSV, and JSON. The carbon monitoring and prediction model assigns corresponding weights to historical carbon monitoring data from different types of satellites to generate a training dataset, and then trains the model based on the sample data in the training dataset.

4. A computer storage medium, wherein the storage medium is a computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a computer, implements the multi-satellite collaborative carbon monitoring method as described in any one of claims 1 to 2.

Images