A collaborative inversion method for atmospheric CO2 concentration using spaceborne lidar and hyperspectral instrument

Abstract

The present invention relates to a coordinated inversion method for atmospheric CO2 concentration for satellite-borne laser radar and hyperspectrometer. First, the aerosol atmospheric profile is obtained by inverting the laser radar 1064nm echo signal, the CO2 column weighted concentration is calculated according to the laser radar 1572nm echo signal, the atmospheric radiation is modeled by sciatran, the solar spectrum obtained by observation is least squares fitted with the simulated value output by the sciatran model, and the CO2 vertical concentration profile is obtained. Then, the vertical profile is used as an absolute constraint to construct a loss function, and the weight of the loss function is set by a priori observations to obtain an optimized solution that minimizes the loss function, update the atmospheric profile, and calculate the final XCO2 product. The present invention integrates the advantages of high precision and high availability of laser radar detection with the advantages of light coverage and high resolution of hyperspectrometer detection to obtain an atmospheric CO2 column concentration product with high resolution, high coverage, high precision and high availability.

Description

Technical Field

[0001] The invention belongs to the technical field of satellite remote sensing earth observation, and in particular relates to a coordinated inversion method of atmospheric CO2 concentration for a satellite-borne laser radar and a hyperspectrometer. Background Art

[0002] The space-based CO2 observation system is an important component of anthropogenic carbon emission monitoring, with the characteristics of large spatial coverage and high temporal resolution. Since the 21st century, satellite remote sensing has also begun to play an important role. With the successful launch of GOSAT, OCO-2 and Carbon Satellite, people's understanding of CO2 emissions has deepened. In recent years, the successive launch of multiple satellites at home and abroad, such as Fengyun-3-04, Gaofen-5, OCO-3, GOSAT-2 and GHGSat series, has further set off a wave of space-based atmospheric greenhouse gas monitoring. At present, these satellite products have demonstrated good application effects in many fields such as CO2 flux optimization inversion, artificial point source positioning and natural disaster carbon emission assessment. However, the existing space-based CO2 observation system cannot fully meet the new needs of anthropogenic carbon emission monitoring. Existing monitoring requires that the space-based CO2 observation system should have the characteristics of high resolution, high coverage, high precision and high availability, but the existing space-based CO2 observation system is mainly a passive observation system. When accurate atmospheric conditions cannot be provided, data inversion will be difficult to converge. At the same time, the data products of the existing space-based CO2 observation system are also subject to signal quality, and cannot provide effective products when the solar altitude angle is small. The direct consequence of these unfavorable factors is that the current efficiency of the passive remote sensing satellite products for CO2 observation is only 2%-5%. Although active remote sensing observation systems, such as detection satellites equipped with lidar, can well avoid the reconstruction of complex radiation transmission processes and have high product efficiency and detection performance, the existing hardware technology does not support space scanning observations and can only perform sub-satellite point measurements in the form of "dotting", which results in its CO2 concentration product not having surface attributes. Therefore, proposing an active and passive collaborative space-based CO2 detection mechanism fusion algorithm is a major problem that needs to be solved urgently in this field. Summary of the invention

[0003] In view of the shortcomings of the prior art, the present invention provides a coordinated inversion method for atmospheric CO2 concentration for satellite-borne lidar and hyperspectrometer, which integrates the advantages of high precision and high availability of lidar detection with the advantages of light coverage and high resolution of hyperspectrometer detection, and obtains an atmospheric CO2 column concentration product with high resolution, high coverage, high precision and high availability characteristics.

[0004] In order to achieve the above object, the technical solution provided by the present invention is a coordinated inversion method of atmospheric CO2 concentration for a space-borne laser radar and a hyperspectral instrument, comprising the following steps:

[0005] Step 1: Invert the atmospheric aerosol profile using the 1064nm laser echo signal of the lidar and the Fernald algorithm;

[0006] Step 2, using the laser radar 1572nm laser echo signal to invert the CO2 column weighted concentration;

[0007] Step 3, using the aerosol profile outputted from step 1 as the input of the sciatran model, and then performing least square fitting between the observed solar spectrum and the simulated value outputted from the sciatran model to obtain the CO2 concentration profile;

[0008] Step 4: compare the weighted concentration of the CO2 column output in step 2 with the CO2 concentration profile output in step 3. When the difference exceeds the threshold ε, the new CO2 concentration profile is obtained by minimizing the loss function and input into the sciatran model in step 3 for updating. Repeat this process until the difference between the CO2 concentration profile output in step 3 and the CO2 column weighted concentration output in step 2 is less than or equal to the threshold ε.

[0009] Moreover, the detection formula of the laser radar in step 1 is as follows:

[0010]

[0011] Where P(Z) is the energy of the atmospheric backscatter echo signal received by the lidar at the height Z, E is the transmit energy of the lidar, C is the radar constant, β(Z) is the atmospheric backscatter coefficient, and σ(Z′) is the atmospheric extinction coefficient.

[0012] There are two unknown quantities β(Z) and σ(Z) in equation (1). The Fernald algorithm is used to solve the lidar equation to obtain the aerosol profile. The Fernald algorithm can treat aerosol and molecular components separately, including forward and backward parts, as shown in equations (2) and (3):

[0013]

[0014]

[0015] A(i)=β1(i)[S1(i)-S2(i)][β2(i)+β2(i+1)]Δr (4)

[0016] where i is the layer number, β1(i) and β2(i) are the backscattering coefficients of aerosols and molecules determined according to the U.S. Standard Atmospheric Model, S1(i) is the lidar ratio of aerosols, S2(i) is the lidar ratio of molecules, Δr is the interlayer distance, and X(i) is the distance correction signal.

[0017] According to equations (1) to (4), the aerosol profile can be obtained by data inversion. The aerosol profile is the atmospheric extinction coefficient at different altitudes.

[0018] Moreover, the calculation formula of the CO2 column concentration in step 2 is as follows:

[0019]

[0020] Where XCO 2LIDAR is the CO2 weighted concentration, P0 is the emitted laser intensity, P is the received laser intensity, R is the detection distance, λ on is the wavelength of a laser beam near the absorption peak of the gas to be measured, λ off is the wavelength of a laser beam near the absorption valley, WF(P) is the weight function, and p plane and p surface They represent the air pressure at the top and bottom of the atmosphere respectively. These parameters can be obtained by radar observations.

[0021] Moreover, in step 3, sciatran is used to model the atmospheric radiation to simulate the spectrum of the solar spectrum under certain atmospheric conditions. The simulation formula of the solar spectrum is as follows:

[0022]

[0023] Where lnI′ is the simulated normalized solar spectrum, I Toa is the simulated spectrum of SCIATRAN, I oλ is the solar spectrum, k λ (z) is the spectral absorption cross section of the gas, n λ (z) is the molecular number concentration.

[0024] Absorption coefficient a λ (z) can be expressed as a λ (z) = n λ (z)k λ (z), since the absorption coefficient is related to the vertical profile of the molecular number concentration, the normalized solar spectrum is related to the vertical profile of the molecular number concentration of the absorber CO2. When the linearization error is ignored, the normalized solar spectrum can be used to calculate the absorption coefficient a λ (z) is expanded by first-order Taylor. Since the absorption coefficient can be approximated as the change in molecular number concentration within the range of Δz, equation (6) can be converted into the following equation:

[0025]

[0026] Where lnI is the normalized spectrum of the actual measurement, is the normalized spectrum simulated by SCIATRAN, λ is the wavelength, z is the height, is the vertical profile of the prior molecular number concentration, δn λ (z) is the change in molecular number concentration, n λ (z) is the molecular number concentration, and z0 represents the height from the surface to the top of the atmosphere.

[0027] Assume that the actual gas concentration profile v is consistent with the prior gas concentration profile There is a proportionality factor, at which the actual molecular number concentration n λ (z) also obeys this scaling factor, i.e. make Considering that there are many gases in the atmosphere, other factors still have a great influence on the atmospheric composition. Each gas should be given a weight, and the gas molecules also have a scattering effect. If the scattering weight is expressed as a low-order polynomial, equation (7) can be expanded to the following equation:

[0028]

[0029] In the formula, I is the number of gas types, b is j is the scattering weight, λ j is the wavelength of various gases.

[0030] In actual observation, the band is discretized, and CO2 molecules have their own unique atmospheric absorption band, so we divide equation (8) by the solar spectrum I 0λ Discretize to get:

[0031]

[0032] In the formula, λ represents the wavelength of the selected absorption band, v i is the concentration profile of the ith gas.

[0033] Solve using nonlinear least squares method:

[0034]

[0035] In the formula, lnI′ i is the simulation value, lnI i is the observed value, and m represents the number of equations in the spectral interval at a certain spectral resolution, that is, m is obtained by dividing the length of the spectral interval by the spectral resolution.

[0036] The passive CO2 vertical concentration profile v obtained under the prior condition of obtaining aerosol profiles by active detection model is obtained. i .

[0037] Moreover, the loss function calculation formula in step 4 is as follows:

[0038]

[0039] in,

[0040]

[0041]

[0042] In the formula, the weight value λ is iteratively selected through the previous satellite observation data, XCO2(p) is the atmospheric profile, and XCO 2LIDAR is the weighted volume ratio of atmospheric CO2 column, WF(p) is the weight function, which describes the absorption capacity of specific atmospheric molecules at different pressures to light, IWF(p) represents the integral of the weight function, σ on (p)-σ off (p) represents the absorption cross-sectional area difference, m dryair represents the molecular mass of H2O and dry air, XH2O(p) represents the volume ratio of H2O at pressure p, g is the gravitational acceleration, p plane and p surface Represents the air pressure at the top and bottom of the atmosphere respectively.

[0043] When the difference between the weighted concentration of the CO2 column output in step 2 and the CO2 concentration profile output in step 3 exceeds the threshold ε, the optimal CO2 concentration profile under the existing conditions can be obtained by minimizing the loss function, which is input into the sciatran model in step 3 to obtain a new CO2 concentration profile, which is compared with the weighted concentration of the CO2 column output in step 2. This process is repeated until the difference between the new CO2 concentration profile output in step 3 and the weighted concentration of the CO2 column output in step 2 is less than or equal to the threshold ε, the iteration is stopped, and the optimal CO2 concentration profile is obtained.

[0044] Compared with the prior art, the present invention has the following advantages: 1) combining active detection with passive detection to make full use of detection information; 2) innovatively proposing an active and passive detection fusion mechanism to achieve complementary advantages in the four dimensions of coverage, availability, resolution, and accuracy to obtain CO2 concentration products. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 is a flow chart of an embodiment of the present invention.

[0046] Figure 2 This is a WF distribution diagram of an embodiment of the present invention.

[0047] Figure 3 This is the CO2 concentration profile calculated by the embodiment of the present invention.

[0048] Figure 4 This is a conceptual diagram of space-based active and passive collaborative detection in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The present invention provides a coordinated inversion method for atmospheric CO2 concentration for a space-borne laser radar and a hyperspectrometer. Firstly, the aerosol atmospheric profile is obtained by inverting the laser radar 1064nm echo signal, the CO2 column weighted concentration is calculated according to the laser radar 1572nm echo signal, the atmospheric radiation is modeled by sciatran, the solar spectrum obtained by observation is fitted with the simulated value output by the sciatran model by least squares, the CO2 vertical concentration profile is obtained, and then the vertical profile is used as an absolute constraint to construct a loss function, the weight of the loss function is set by a priori observation values, the optimization solution that minimizes the loss function is obtained, the atmospheric profile is updated, and the final XCO2 product with high resolution, high coverage, high precision and high availability characteristics is calculated.

[0050] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0051] like Figure 1 As shown, the process of the embodiment of the present invention includes the following steps:

[0052] Step 1: Use the 1064nm laser echo signal of the lidar and the Fernald algorithm to invert the atmospheric aerosol profile.

[0053] The detection formula of LiDAR is as follows:

[0054]

[0055] Where P(Z) is the energy of the atmospheric backscatter echo signal received by the lidar at the height Z, E is the transmit energy of the lidar, C is the radar constant, β(Z) is the atmospheric backscatter coefficient, and σ(Z′) is the atmospheric extinction coefficient.

[0056] There are two unknown quantities β(Z) and σ(Z) in equation (1). The Fernald algorithm is used to solve the lidar equation to obtain the aerosol profile. The Fernald algorithm can treat aerosol and molecular components separately, including forward and backward parts, as shown in equations (2) and (3):

[0057]

[0058]

[0059] A(i)=β1(i)[S1(i)-S2(i)][β2(i)+β2(i+1)]Δr (4)

[0060] Where i is the layer number; β1(i) and β2(i) are the backscattering coefficients of aerosols and molecules determined according to the U.S. Standard Atmospheric Model; S1(i) is the laser radar ratio of aerosols, which is 50sr in this embodiment; S2(i) is the laser radar ratio of molecules, which is Δr is the inter-layer distance; X(i) is the distance correction signal.

[0061] According to equations (1) to (4), the aerosol profile can be obtained by data inversion. The aerosol profile is the atmospheric extinction coefficient at different altitudes.

[0062] Step 2: Use the 1572nm laser echo signal of the lidar to invert the weighted concentration of the CO2 column.

[0063] The calculation formula for CO2 column concentration is as follows:

[0064]

[0065] Where XCO 2LIDAR is the CO2 weighted concentration, P0 is the emitted laser intensity, P is the received laser intensity, R is the detection distance, λ on is the wavelength of a laser beam near the absorption peak of the gas to be measured, λ off is the wavelength of a laser beam near the absorption valley, WF(P) is the weight function, and p plane and p surface They represent the air pressure at the top and bottom of the atmosphere respectively. These parameters can be obtained by radar observations.

[0066] Step 3: Use the aerosol profile output from step 1 as the input of the sciatran model, and then perform least squares fitting on the observed solar spectrum and the simulated value output by the sciatran model to obtain the CO2 concentration profile.

[0067] Sciatran is used to model atmospheric radiation and simulate the solar spectrum under certain atmospheric conditions. The simulation formula of the solar spectrum is as follows:

[0068]

[0069] Where lnI′ is the simulated normalized solar spectrum, I Toa is the simulated spectrum of SCIATRAN, I oλ is the solar spectrum, k λ (z) is the spectral absorption cross section of the gas, n λ (z) is the molecular number concentration.

[0070] Absorption coefficient a λ (z) can be expressed as a λ (z) = nλ (z)k λ (z), since the absorption coefficient is related to the vertical profile of the molecular number concentration, the normalized solar spectrum is related to the vertical profile of the molecular number concentration of the absorber CO2. When the linearization error is ignored, the normalized solar spectrum can be used to calculate the absorption coefficient a λ (z) is expanded by first-order Taylor. At the same time, since the absorption coefficient can be approximated as the change in molecular number concentration within the range of Δz, equation (6) can be converted into the following equation:

[0071]

[0072] Where lnI is the normalized spectrum of the actual measurement, is the normalized spectrum simulated by SCIATRAN, λ is the wavelength, z is the height, is the vertical profile of the prior molecular number concentration, δn λ (z) is the change in molecular number concentration, and z0 represents the height from the surface to the top of the atmosphere.

[0073] Assume that the actual gas concentration profile v is consistent with the prior gas concentration profile There is a proportionality factor, at which the actual molecular number concentration n λ (z) also obeys this scaling factor, i.e. make Considering that there are many gases in the atmosphere, other factors still have a great influence on the atmospheric composition. Each gas should be given a weight, and the gas molecules also have a scattering effect. If the scattering weight is expressed as a low-order polynomial, equation (7) can be expanded to the following equation:

[0074]

[0075] In the formula, I is the number of gas types, b is j is the scattering weight, λ j is the wavelength of various gases.

[0076] In actual observation, the band is discretized, and CO2 molecules have their own unique atmospheric absorption band, so we divide equation (8) by the solar spectrum I 0λ Discretize to get:

[0077]

[0078] In the formula, λ represents the wavelength of the selected absorption band, v i is the concentration profile of the i-th gas, and in this embodiment specifically refers to the concentration profile of CO2.

[0079] Solve using nonlinear least squares method:

[0080]

[0081] In the formula, I′ i is the analog value, I i is the observed value, and m represents the number of equations in the spectral interval at a certain spectral resolution, that is, m is obtained by dividing the length of the spectral interval by the spectral resolution.

[0082] The passive CO2 vertical concentration profile v is obtained under the prior condition of obtaining the aerosol profile by the active detection model. i .

[0083] Step 4: compare the weighted concentration of the CO2 column output in step 2 with the CO2 concentration profile output in step 3. When the difference exceeds the threshold ε, the new CO2 concentration profile is obtained by minimizing the loss function and input into the sciatran model in step 3 for updating. Repeat this process until the difference between the CO2 concentration profile output in step 3 and the CO2 column weighted concentration output in step 2 is less than or equal to the threshold ε.

[0084] The loss function calculation formula is as follows:

[0085]

[0086] in,

[0087]

[0088]

[0089] In the formula, the weight value λ is iteratively selected through the previous satellite observation data, XCO2(p) is the atmospheric profile, and XCO 2LIDAR is the weighted volume ratio of atmospheric CO2 column, WF(p) is the weight function, which describes the absorption capacity of specific atmospheric molecules at different pressures to light, IWF(p) represents the integral of the weight function, σ on (p)-σ off (p) represents the absorption cross-sectional area difference, m dryair represents the molecular mass of H2O and dry air, XH2O(p) represents the volume ratio of H2O at pressure p, g is the gravitational acceleration, p plane and p surface Represents the air pressure at the top and bottom of the atmosphere respectively.

[0090] In the actual sciatran model, the atmosphere can usually be divided into 28 layers, which can be integrated layer by layer. Since the atmosphere of layers 3-28 in the atmospheric model is relatively stable, only the atmospheric profiles of layers 1-2 need to be adjusted during the modification process. When the difference between the CO2 column weighted concentration output in step 2 and the CO2 concentration profile output in step 3 exceeds the threshold ε, the optimal CO2 concentration profile under the existing conditions can be obtained by minimizing the loss function, which is input into the sciatran model in step 3 to obtain a new CO2 concentration profile, and the new CO2 concentration profile is compared with the CO2 column weighted concentration output in step 2. This process is repeated until the difference between the new CO2 concentration profile output in step 3 and the CO2 column weighted concentration output in step 2 is less than or equal to the threshold ε, and the iteration is stopped to obtain the optimal CO2 concentration profile.

[0091] During specific implementation, the above process can be automatically operated using computer software technology.

[0092] The specific embodiments described herein are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in similar ways, but they will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.

Claims

1. A coordinated inversion method for atmospheric CO2 concentration using space-borne laser radar and hyperspectrometer, characterized in that: The steps include: Step 1: Invert the atmospheric aerosol profile using the 1064nm laser echo signal of the lidar and the Fernald algorithm; Step 2, using the laser radar 1572nm laser echo signal to invert the CO2 column weighted concentration; Step 3, using the aerosol profile outputted from step 1 as the input of the sciatran model, and then performing least square fitting between the observed solar spectrum and the simulated value outputted from the sciatran model to obtain the CO2 concentration profile; Step 4, compare the weighted concentration of CO2 column outputted in step 2 with the CO2 concentration profile outputted in step 3. When the difference exceeds the threshold ε, a new CO2 concentration profile is obtained by minimizing the loss function and inputted into the sciatran model in step 3 for updating. Repeat this process until the difference between the CO2 concentration profile outputted in step 3 and the CO2 column weighted concentration outputted in step 2 is less than or equal to the threshold ε. The loss function calculation formula is as follows: in, In the formula, the weight value λ is iteratively selected through the previous satellite observation data, XCO2(p) is the atmospheric profile, and XCO 2LIDAR is the weighted volume ratio of atmospheric CO2 column, WF(p) is the weight function, which describes the absorption capacity of specific atmospheric molecules at different pressures to light, IWF(p) represents the integral of the weight function, σ on (p)-σ off (p) represents the absorption cross-sectional area difference, m dryair represents the molecular mass of H2O and dry air, XH2O(p) represents the volume ratio of H2O at pressure p, g is the gravitational acceleration, p plane and p surface Represents the air pressure at the top and bottom of the atmosphere respectively.

2. The method for coordinated inversion of atmospheric CO2 concentration for space-borne laser radar and hyperspectrometer according to claim 1, characterized in that: The detection formula of the laser radar in step 1 is as follows: Where P(Z) is the energy of the atmospheric backscatter echo signal received by the lidar at the height Z, E is the transmit energy of the lidar, C is the radar constant, β(Z) is the atmospheric backscatter coefficient, and σ(Z′) is the atmospheric extinction coefficient; There are two unknown quantities β(Z) and σ(Z) in equation (1). The Fernald algorithm is used to solve the lidar equation to obtain the aerosol profile. The Fernald algorithm can treat aerosol and molecular components separately, including forward and backward parts, as shown in equations (2) and (3): A(i)=β1(i)[S1(i)-S2(i)][β2(i)+β2(i+1)]Δr (4) Where i is the layer number, β1(i) and β2(i) are the backscatter coefficients of aerosols and molecules determined according to the US Standard Atmospheric Model, S1(i) is the lidar ratio of aerosols, S2(i) is the lidar ratio of molecules, Δr is the interlayer distance, and X(i) is the distance correction signal; According to equations (1) to (4), the aerosol profile can be obtained by data inversion. The aerosol profile is the atmospheric extinction coefficient at different altitudes.

3. The method for coordinated inversion of atmospheric CO2 concentration for spaceborne laser radar and hyperspectrometer according to claim 1, characterized in that: The calculation formula for the CO2 column concentration in step 2 is as follows: Where XCO 2LIDAR is the CO2 weighted concentration, P0 is the emitted laser intensity, P is the received laser intensity, R is the detection distance, λ on is the wavelength of a laser beam near the absorption peak of the gas to be measured, λ off is the wavelength of a laser beam near the absorption valley, WF(P) is the weight function, and p plane and p surface They represent the air pressure at the top and bottom of the atmosphere respectively. These parameters can be obtained by radar observations.

4. The method for coordinated inversion of atmospheric CO2 concentration for space-borne laser radar and hyperspectrometer according to claim 1, characterized in that: In step 3, sciatran is used to model the atmospheric radiation and simulate the solar spectrum under certain atmospheric conditions. The simulation formula of the solar spectrum is as follows: Where lnI′ is the simulated normalized solar spectrum, I Toa is the simulated spectrum of SCIATRAN, I oλ is the solar spectrum, k λ (z) is the spectral absorption cross section of the gas, n λ (z) is the molecular number concentration; Absorption coefficient a λ (z) can be expressed as a λ (z) = n λ (z)k λ (z), since the absorption coefficient is related to the vertical profile of the molecular number concentration, the normalized solar spectrum is related to the vertical profile of the molecular number concentration of the absorber CO2. When the linearization error is ignored, the normalized solar spectrum can be used to calculate the absorption coefficient a λ (z) is expanded by first-order Taylor. Since the absorption coefficient can be approximated as the change in molecular number concentration within the range of Δz, equation (6) can be converted into the following equation: Where lnI is the normalized spectrum of the actual measurement, is the normalized spectrum simulated by SCIATRAN, λ is the wavelength, z is the height, is the vertical profile of the prior molecular number concentration, δn λ (z) is the change in molecular number concentration, n λ (z) is the molecular number concentration, z0 represents the height from the surface to the top of the atmosphere; Assume that the actual gas concentration profile v is consistent with the prior gas concentration profile There is a proportionality factor, at which the actual molecular number concentration n λ (z) also obeys this scaling factor, i.e. make Considering that there are many gases in the atmosphere, other factors still have a great influence on the atmospheric composition. Each gas should be given a weight, and the gas molecules also have a scattering effect. If the scattering weight is expressed as a low-order polynomial, equation (7) can be expanded to the following equation: In the formula, I is the number of gas types, b is j is the scattering weight, λ j is the wavelength of various gases; In actual observation, the band is discretized, and CO2 molecules have their own unique atmospheric absorption band, so we divide equation (8) by the solar spectrum I 0λ Discretize to get: In the formula, λ represents the wavelength of the selected absorption band, v i is the concentration profile of the ith gas; Solve using nonlinear least squares method: In the formula, lnI i ′ is the simulation value, lnI i is the observed value, m represents the number of equations in the spectral interval at a certain spectral resolution, that is, the length of the spectral interval divided by the spectral resolution gives m; The passive CO2 vertical concentration profile v obtained under the prior condition of obtaining aerosol profiles by active detection model is obtained. i .

5. The method for coordinated inversion of atmospheric CO2 concentration for space-borne laser radar and hyperspectrometer according to claim 1, characterized in that: In step 4, when the difference between the weighted concentration of the CO2 column output in step 2 and the CO2 concentration profile output in step 3 exceeds the threshold ε, the optimal CO2 concentration profile under the existing conditions can be obtained by minimizing the loss function, which is input into the sciatran model in step 3 to obtain a new CO2 concentration profile, and the new CO2 concentration profile is compared with the weighted concentration of the CO2 column output in step 2. This process is repeated until the difference between the new CO2 concentration profile output in step 3 and the weighted concentration of the CO2 column output in step 2 is less than or equal to the threshold ε, the iteration is stopped, and the optimal CO2 concentration profile is obtained.

Images