CN115310550A - Method and system for calculating concentration of atmospheric carbon dioxide dry air column - Google Patents

Abstract

The invention provides a method and a system for calculating the dry air column concentration of atmospheric carbon dioxide, and relates to the technical field of atmospheric carbon dioxide (CO ₂ ) monitoring. The specific steps are as follows: collect XCO ₂ datasets and environmental covariate data from three remote sensing monitoring in the study area, and perform preprocessing such as gridding and space-time matching; use environmental covariates to fuse the three XCO ₂ datasets , obtain the XCO ₂ fusion data; take the XCO ₂ fusion data as the dependent variable and the environmental covariates as the independent variables, establish a machine learning model based on the XGBoost algorithm; input the environmental covariate data into the XGBoost model, and calculate the comprehensive domain of XCO ₂ The spatiotemporal distribution reconstructs the dataset. The present invention fuses three XCO ₂ data sets monitored by remote sensing, and uses a machine learning model to reconstruct the full-domain temporal and spatial distribution of XCO ₂ to provide support for the carbon accounting of the "Double Carbon Action".

Description

A method and system for calculating the concentration of atmospheric carbon dioxide dry air column

technical field

The invention relates to the technical field of monitoring atmospheric carbon dioxide dry air column concentration (XCO ₂ ), in particular to a calculation method and system for atmospheric carbon dioxide dry air column concentration.

Background technique

The use of sensors mounted on satellites or space stations for remote sensing monitoring is an important means of obtaining the temporal and spatial distribution of XCO ₂ , mainly including Orbital Carbon Observation 2 (OCO-2), the Greenhouse Gas Observation Satellite (GOSAT), and satellites mounted on the International Space Station Sensors on Orbiting Carbon Observatory 3 (OCO-3). The sensors of OCO-2 and OCO-3 are almost identical. Both are equipped with three high-resolution spectrometers, which measure the reflected sunlight spectra near 0.76, 1.61, and 2.06 microns respectively. GOSAT is equipped with a spectrometer that measures the reflected solar spectrum at 0.76, 1.61 and 2.06 microns, and also measures the reflected solar spectrum in the 5.56-14.3 micron band.

XCO ₂ monitored by satellites is relatively sparse in space, and the method of using model simulation can fill in the missing part. The current method of reconstructing XCO ₂ in the whole area is the Carbon Tracker model simulation, which uses the CO ₂ concentration monitored by the ground, airships, and satellites to correct the bottom-up estimated atmospheric CO ₂ concentration. However, due to the limitation of computer computing power, the resolution of the simulated XCO ₂ data is low, usually only 3°×2° on the global scale. At the same time, this method relies on the carbon emission inventory. However, the statistics, collection and correction of the emission inventory data take a long time, so there is a certain lag in the calculation of the temporal and spatial distribution of the current _XCO2 model simulation.

In recent years, studies have used statistical methods combined with satellite monitoring XCO ₂ data to reconstruct the spatial and temporal distribution of XCO ₂ . However, due to the relatively sparse distribution of _{XCO 2} satellite monitoring data in space, the accurate Relatively low. According to previous studies, the temporal and spatial distribution of atmospheric pollutants reconstructed by machine learning is often more accurate than kriging interpolation. Using the correlation between air pollutants and environmental covariates such as meteorology, altitude, and land use type, the comprehensive spatial and temporal distribution of air pollutants is reconstructed through machine learning modeling. At the same time, machine learning methods are often used for fusion applications of homogeneous data, which can well make up for the limitations of a single data source.

The invention applies the machine learning method to the fusion of multiple remote sensing monitoring XCO ₂ data sets, and reconstructs the overall domain spatio-temporal distribution of XCO ₂ to overcome the problem of a large number of missing data in XCO ₂ monitoring.

Contents of the invention

The purpose of the present invention is to provide a method and system for calculating atmospheric carbon dioxide dry air column concentration to overcome the problem of a large number of missing data in _XCO2 monitoring in the prior art.

In a first aspect, an embodiment of the present application provides a method for calculating the concentration of an atmospheric carbon dioxide dry air column, comprising the following steps:

Collect _XCO2 data sets and environmental covariate data from three remote sensing monitoring in the study area, and preprocess the above data sets;

Fusion of three XCO ₂ data sets monitored by remote sensing to obtain XCO ₂ fusion data;

Use XCO ₂ fusion data and environmental covariate data to train the XGBoost model;

The environmental covariate data was input into the XGBoost model, and the comprehensive spatial-temporal distribution reconstruction dataset of XCO ₂ was calculated.

In the above implementation process, by collecting the XCO ₂ data sets and environmental covariate data of the three remote sensing monitoring in the study area, and performing preprocessing such as gridding and spatio-temporal matching _; Obtain the XCO ₂ fusion data; use the XCO ₂ fusion data and environmental covariate data to train the XGBoost model; input the environmental covariate data into the XGBoost model, and calculate the comprehensive spatial and temporal distribution reconstruction data set of XCO ₂ . The invention fuses the XCO ₂ data of three remote sensing monitoring, and utilizes the correlation between XCO ₂ and environmental covariates such as meteorology, altitude, land use type, road density, vegetation normalized index, etc., and models through machine learning , to calculate the comprehensive spatial-temporal distribution reconstruction data set of XCO ₂ to overcome the problem of a large number of missing data in XCO ₂ monitoring.

Based on the first aspect, in some embodiments of the present invention, data fusion is performed on three _XCO2 data sets monitored by remote sensing, and the steps of obtaining _XCO2 fusion data are as follows:

enter:

OCO-2 XCO ₂ monitoring data A={a ₁ ,...,a _j ,...,a _J };

OCO-3 XCO ₂ monitoring data B={b ₁ ,...,b _j ,...,b _J };

GOSAT XCO ₂ monitoring data C={c ₁ ,…,c _j ,…,c _J };

Environmental covariate data set D = {d ₁ ,...,d _j ,...,d _J };

Note: where d _j represents the value set of all environmental covariates at j, not the jth environmental covariate. d _j and a _j , b _j and c _j correspond to each other in time and space, that is, the observation values of the three sensors on the same day and the same grid. Since OCO-2, OCO-3 and GOSAT data are missing to varying degrees, there may be missing values in a _j , b _j and c _j , but not at the same time.

Output: fusion data set E={e ₁ ,...,e _j ,...,e _J };

method:

① Merge A and B into O:

Where O={o ₁ ,...,o _j ,...,o _J }, o _j =f ₁ (a _j , b _j ), where at least one of a _j and b _j is non-empty;

f1 is the fusion function _:

② Linear transformation of GOSAT dataset C:

S={j|O _j is not empty and C _j is not empty};

M={c _s |s∈S}, N={o _s |s∈S};

Establish a linear model f ₃ : N←M;

Transform c _j based on f ₃ :

j from 1 to J:

when c _j is not empty and

c′ _j = f ₃ (c _j );

The transformed data set is C′={c′ ₁ ,...,c′ _j ,...,c′ _J };

③ Take the union of data set O and data set C′ to get the final XCO ₂ fusion data set E.

Based on the first aspect, in some embodiments of the present invention, the steps of using _XCO fusion data and environmental covariate data to establish an XGBoost model are as follows:

Taking the XCO ₂ fusion data as the dependent variable and the environmental covariate as the independent variable, the XGBoost model was trained.

Based on the first aspect, in some embodiments of the present invention, preprocessing the environmental covariate data of the research area includes the following steps:

Spatial division of the research area to obtain a spatial grid with a resolution of 1 km;

Use spatial resampling and other methods to assign environmental covariate data to a predefined grid;

In a second aspect, an embodiment of the present application provides a system for calculating the concentration of an atmospheric carbon dioxide dry air column, including:

Data acquisition module for collecting _XCO2 datasets and environmental covariate data from three remote sensing monitoring in the study area;

The data preprocessing module performs preprocessing such as gridding and space-time matching on the XCO ₂ data sets and environmental covariate data of the three remote sensing monitoring in the study area, and obtains the preprocessed data sets;

The data fusion module is used for data fusion of the XCO ₂ data sets of three remote sensing monitoring to obtain the XCO ₂ fusion data set;

The XGBoost model building module is used to obtain the XGBoost model based on XCO ₂ fusion data and environmental covariate data training;

The concentration calculation module is used to input the environmental covariate data into the XGBoost model, and calculate the comprehensive spatial and temporal distribution reconstruction data set of XCO ₂ .

In the above implementation process, by collecting the XCO ₂ data sets and environmental covariate data of the three remote sensing monitoring in the study area, and performing preprocessing such as gridding and spatio-temporal matching _; Obtain the XCO ₂ fusion data; use the XCO ₂ fusion data and environmental covariate data to train the XGBoost model; input the environmental covariate data into the XGBoost model, and calculate the comprehensive spatial and temporal distribution reconstruction data set of XCO ₂ . The present invention fuses the XCO ₂ data of three remote sensing monitoring, and utilizes the correlation between XCO ₂ and environmental covariates such as meteorology, altitude, land use type, road density, vegetation normalized index, etc., and models through machine learning , to calculate the comprehensive spatial-temporal distribution reconstruction data set of XCO ₂ to overcome the problem of a large number of missing data in XCO ₂ monitoring.

Embodiments of the present invention have at least the following advantages or beneficial effects:

The embodiment of the present invention provides a method and system for calculating the concentration of atmospheric carbon dioxide dry air column, by collecting three remote sensing monitoring _XCO2 data sets and environmental covariate data in the research area, and performing preprocessing such as gridding and time-space matching; Data fusion of three XCO ₂ data sets monitored by remote sensing is carried out to obtain XCO ₂ fusion data; XGBoost model is obtained by training using XCO ₂ fusion data and environmental covariate data; input environmental covariate data set into XGBoost model, and calculate A comprehensive domain spatiotemporal distribution reconstruction dataset for XCO ₂ . The present invention fuses three XCO ₂ data sets monitored by remote sensing, and utilizes the correlation between XCO ₂ and environmental covariates such as meteorology, altitude, land use type, road density, vegetation normalization index, etc. Based on the model, the reconstruction data set of the comprehensive spatial and temporal distribution of XCO ₂ can be calculated to overcome the problem of a large number of missing data in XCO ₂ monitoring.

Description of drawings

In order to illustrate the technical solutions of the embodiments of the present invention more clearly, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention, and thus It should be regarded as a limitation on the scope, and those skilled in the art can also obtain other related drawings based on these drawings without creative work.

Fig. 1 is a flow chart of a calculation method for atmospheric carbon dioxide dry air column concentration provided by an embodiment of the present invention;

FIG. 2 is a schematic diagram of a data fusion process provided by an embodiment of the present invention;

Fig. 3 is a schematic diagram of the spatial and temporal distribution of remote sensing data and fusion data provided by the embodiment of the present invention;

Fig. 4 is a structural block diagram of an atmospheric carbon dioxide dry air column concentration calculation system provided by an embodiment of the present invention.

Icons: 110-data acquisition module; 120-data preprocessing module; 130-data fusion module; 140-XGBoost model building module; 150-concentration calculation module.

Detailed ways

In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments It is a part of the embodiments of this application, not all of them. The components of the embodiments of the application generally described and illustrated in the figures herein may be arranged and designed in a variety of different configurations.

Accordingly, the following detailed description of the embodiments of the application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of the application. Based on the embodiments in this application, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of this application.

Example

Some implementations of the present application will be described in detail below in conjunction with the accompanying drawings. In the case of no conflict, each of the following embodiments and each feature in the embodiments can be combined with each other.

Please refer to Figures 1-2, Figure 1 is a flowchart of a method for calculating the concentration of atmospheric carbon dioxide dry air column provided by an embodiment of the present invention, and Figure 2 is a schematic diagram of a data fusion process provided by an embodiment of the present invention. The method for calculating the atmospheric carbon dioxide dry air column concentration comprises the following steps:

Step S110: Collect remote sensing monitoring XCO ₂ data and environmental covariate data in the study area. The above-mentioned XCO ₂ data include three XCO ₂ datasets monitored by remote sensing, including the XCO 2 data from the Orbiting Carbon Observatory-2 (OCO-2), the XCO ₂ data from the Greenhouse Gas Observatory Satellite (GOSAT), and the XCO ₂ data from the XCO ₂ data from the sensor Orbiting Carbon Observatory 3 (OCO-3) on the International Space Station. The above datasets include meteorology, population density, planetary boundary layer height, land use type, normalized difference vegetation index, elevation, road information, etc. Among them, the meteorological data comes from the European Center for Medium-Range Weather Forecast, the land use type data comes from the Climate Change Institute of the European Space Agency, the altitude data comes from the Space Shuttle Radar Topography Mission, and the population density data come from the gridded world population.

Preprocess the environmental covariate data for the study area. The above preprocessing is to process the environmental covariate data into a predefined 1km grid by means of gridding and space-time matching for training and calculation of machine learning models. This data set usually includes dozens to hundreds of variables . Specifically include the following steps:

First, the study area is spatially divided to obtain a spatial grid with a resolution of 1 km.

Then, the environmental covariate data were assigned to the predefined 1km grid using methods such as gridding and space-time matching to obtain the preprocessed environmental covariate data set.

Step S120: Perform data fusion on the three XCO ₂ data sets monitored by remote sensing to obtain XCO ₂ fusion data, the steps are as follows:

enter:

OCO-2 XCO ₂ monitoring data A={a ₁ ,...,a _j ,...,a _J };

OCO-3 XCO ₂ monitoring data B={b ₁ ,...,b _j ,...,b _J };

GOSAT XCO ₂ monitoring data C = {c ₁ ,…,c _j ,…,c _J };

Environmental covariate data set D = {d ₁ ,...,d _j ,...,d _J };

Note: where d _j represents the value set of all environmental covariates at j, not the jth environmental covariate. d _j and a _j , b _j and c _j correspond to each other in time and space, that is, the observation values of the three sensors on the same day and the same grid. Since OCO-2, OCO-3 and GOSAT data are missing to varying degrees, there may be missing values in a _j , b _j and c _j , but not at the same time.

Output: fusion data set E={e ₁ ,...,e _j ,...,e _J };

method:

① Merge A and B into O:

Where O={o ₁ ,...,o _j ,...,o _J }, o _j =f ₁ (a _j , b _j ), where at least one of a _j and b _j is non-empty;

f1 is the fusion function _:

② Linear transformation of GOSAT dataset C:

S={j|O _j is not empty and C _j is not empty};

M={c _s |s∈S}, N={o _s |s∈S};

Establish a linear model f ₃ : N←M;

Transform c _j based on f ₃ :

j from 1 to J:

when c _j is not empty and

c′ _j = f ₃ (c _j );

The transformed data set is C'={c' ₁ ,...,c' _j ,...,c' _J };

③ Take the union of data set O and data set C′ to get the final XCO ₂ fusion data set E.

Step S130: using the XCO ₂ fusion data as the dependent variable and the environmental covariate as the independent variable, training a machine learning model based on the XGBoost algorithm;

Step S140: input the environmental covariate data into the XGBoost model, and calculate and obtain the reconstructed data set of XCO ₂ 's comprehensive domain spatio-temporal distribution.

In the above implementation process, by collecting the XCO ₂ data sets and environmental covariate data of the three remote sensing monitoring in the study area, and performing preprocessing such as gridding and spatio-temporal matching _; Obtain the XCO ₂ fusion data; use the XCO ₂ fusion data and environmental covariate data to train the XGBoost model; input the environmental covariate data into the XGBoost model, and calculate the comprehensive spatial and temporal distribution reconstruction data set of XCO ₂ . The present invention fuses three XCO ₂ data sets monitored by remote sensing, and utilizes the correlation between XCO ₂ and environmental covariates such as meteorology, altitude, land use type, road density, vegetation normalization index, etc. Model, calculated to obtain the XCO ₂ comprehensive domain spatio-temporal distribution reconstruction data set.

Please refer to FIG. 3 . FIG. 3 is a schematic diagram of a data fusion process provided by an embodiment of the present invention, the horizontal axis represents space, and the vertical axis represents time. In the schematic diagram, C=G+S represents the XCO ₂ data set monitored by GOSAT, which has data in the entire time interval P; O=T+S represents the XCO ₂ data set monitored by OCO (OCO-2 and OCO-3), Data exists only in time interval _P1 . In the P ₁ time interval, both the XCO ₂ monitored by GOSAT and the XCO ₂ monitored by OCO have data point distribution. Due to the orbits of satellites and space stations and unfavorable meteorological conditions, the daily-scale satellite data are missing in space. The XCO ₂ monitored by OCO only covers the time-space interval of O, while the XCO ₂ monitored by GOSAT only covers C in the time period P. In this space-time interval, the part where the two overlap in space is the S space-time interval. The GOSAT monitoring value and the OCO monitoring value in the S space-time interval were linearly fitted, and the GOSAT monitoring value was used as the independent variable, and the OCO monitoring value was used as the dependent variable to establish a linear model. Using this linear model, the GOSAT monitoring value of the G part is converted into the OCO monitoring value, so that the OCO monitoring value approximately covers the G space interval, and its time interval covers the entire P period, realizing the fusion of multiple XCO ₂ monitoring data.

Based on the same inventive concept, the present invention also proposes a calculation system for atmospheric carbon dioxide dry air column concentration, please refer to FIG. 4 , which is a structural block diagram of an atmospheric carbon dioxide dry air column concentration calculation system provided by an embodiment of the present invention. The atmospheric carbon dioxide dry air column concentration calculation system includes:

The data acquisition module 110 is used to collect XCO ₂ data sets and environmental covariate data of three remote sensing monitoring in the research area;

The data preprocessing module 120 is used to preprocess the _XCO2 data and environmental covariate data in the research area to obtain preprocessed _XCO2 data sets and environmental covariate data sets;

XGBoost model building module 140, for building an XGBoost model based on XCO ₂ fusion data and environmental covariate data;

The concentration calculation module 150 is used for inputting the environmental covariate data into the XGBoost model, and calculating the reconstructed data set of the comprehensive spatial-temporal distribution of XCO ₂ .

In the above implementation process, by collecting the XCO ₂ data sets and environmental covariate data of the three remote sensing monitoring in the study area, and performing preprocessing such as gridding and spatio-temporal matching _; Obtain the XCO ₂ fusion data; use the XCO ₂ fusion data and environmental covariate data to train the XGBoost model; input the environmental covariate data into the XGBoost model, and calculate the comprehensive spatial and temporal distribution reconstruction data set of XCO ₂ . The invention fuses the XCO ₂ data of three remote sensing monitoring, and utilizes the correlation between XCO ₂ and environmental covariates such as meteorology, altitude, land use type, road density, vegetation normalized index, etc., and models through machine learning , to obtain the comprehensive spatial-temporal distribution reconstruction data set of XCO ₂ .

The above descriptions are only preferred embodiments of the present application, and are not intended to limit the present application. For those skilled in the art, there may be various modifications and changes in the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this application shall be included within the protection scope of this application.

It will be apparent to those skilled in the art that the present application is not limited to the details of the exemplary embodiments described above, but that the present application can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application. Therefore, the embodiments should be regarded as exemplary and not restrictive in all points of view, and the scope of the application is defined by the appended claims rather than the foregoing description, and it is intended that the scope of the present application be defined by the appended claims rather than by the foregoing description. All changes within the meaning and range of equivalents of the elements are embraced in this application. Any reference sign in a claim should not be construed as limiting the claim concerned.

Claims (5)

1. A method for reconstructing spatial-temporal distribution of atmospheric carbon dioxide dry air column concentration (XCO ₂ ), is characterized in that it comprises the following steps: Collect three remote sensing monitoring XCO ₂ datasets (including OCO-2, OCO-3, GOSAT) in the study area and environmental covariate data in the area, and preprocess the above datasets; Fusion of the three XCO ₂ data sets monitored by remote sensing to obtain the XCO ₂ fusion data set; Using the XCO ₂ fusion data as the dependent variable and the environmental covariate as the independent variable, the machine learning model based on the XGBoost algorithm is trained; The environmental covariate data was input into the XGBoost model, and the comprehensive spatial-temporal distribution reconstruction dataset of XCO ₂ was calculated. 2. Atmospheric carbon dioxide dry air column concentration calculation method according to claim 1, is characterized in that, the described XCO of _three remote sensing monitoring Data sets are fused The steps are as follows: enter: OCO-2 XCO ₂ monitoring data A={a ₁ ,...,a _j ,...,a _J }; OCO-3 XCO ₂ monitoring data B={b ₁ ,...,b _j ,...,b _J }; GOSAT XCO ₂ monitoring data C = {c ₁ ,...,c _j ,...,c _J }; Environmental covariate data set D = {d ₁ ,...,d _j ,...,d _J }; Note: where d _j represents the value set of all environmental covariates at j, not the jth environmental covariate. d _j and a _j , b _j and c _j correspond to each other in time and space, that is, the observation values of the three sensors on the same day and the same grid. Since OCO-2, OCO-3 and GOSAT data are missing to varying degrees, there may be missing values in a _j , b _j and c _j , but not at the same time. Output: fusion data set E={e ₁ ,...,e _j ,...,e _J }; method: ① Merge A and B into O: Where O={o ₁ ,...,o _j ,...,o _J }, o _j =f ₁ (a _j , b _j ), where at least one of a _j and b _j is non-empty; f1 is the fusion function _:

② Linear transformation of GOSAT dataset C: S={j|O _j is not empty and C _j is not empty}; M={c _s |s∈S}, N={o _s |s∈S}; Establish a linear model f ₃ : N←M; Transform c _j based on f ₃ : j from 1 to J: when c _j is not empty and

c′ _j =f ₃ (c _j ) The transformed data set is C′={c′ ₁ ,...,c′ _j ,...,c′ _J } ③ Take the union of data set O and data set C′ to get the final XCO ₂ fusion data set E. 3. atmospheric carbon dioxide dry air column concentration calculation method according to claim 2, is characterized in that, described use _XCO Fusion data and environmental covariate set up the step of XGBoost model as follows: Taking the XCO ₂ fusion data as the dependent variable and the environmental covariate as the independent variable, the XGBoost model was trained. 4. atmospheric carbon dioxide dry air column concentration calculation method according to claim 1, is characterized in that comprising carrying out preprocessing to environmental covariate data in research area, comprises the following steps: Spatial division of the research area to obtain a spatial grid with a resolution of 1 km; The environmental covariate data are assigned to a predefined grid using methods such as spatial resampling. 5. A system for calculating the concentration of an atmospheric carbon dioxide dry air column, characterized in that it comprises: Data acquisition module for collecting _XCO2 datasets and environmental covariate data from three remote sensing monitoring in the study area; The data preprocessing module performs preprocessing such as gridding and space-time matching on the XCO ₂ data sets and environmental covariate data of the three remote sensing monitoring in the study area, and obtains the preprocessed data sets; The data fusion module is used for data fusion of three XCO ₂ data sets monitored by remote sensing to obtain the XCO ₂ fusion data set; The XGBoost model building module is used to obtain the XGBoost model based on XCO ₂ fusion data and environmental covariate data training; The concentration calculation module is used to input the environmental covariate data into the _XGBoost model, and calculate the comprehensive spatial and temporal distribution reconstruction data set of XCO2.

Images