A method for retrieving atmospheric N2O column concentration based on thermal infrared satellite remote sensing

Abstract

This invention discloses a method for inverting atmospheric N2O column concentration based on thermal infrared satellite remote sensing, belonging to the field of atmospheric environment remote sensing technology. The method includes: acquiring and preprocessing satellite observation data; selecting a specific spectral range as the inversion characteristic spectral region; introducing a fast radiative transfer model to construct a forward simulation operator to simulate the radiance of the top atmospheric layer; selecting N2O-sensitive characteristic bands with high signal-to-noise ratio and low interference based on a channel optimization strategy using the N2O / CO2 Jacobi intensity ratio; constructing an extended state vector, using surface temperature as an unknown parameter and performing joint optimal estimation iterative inversion with the N2O profile to correct the surface temperature in the reanalysis data, thereby achieving high-precision inversion of N2O atmospheric column concentration. This invention transforms surface temperature from a fixed input parameter into a state variable to be inverted, effectively decoupling surface temperature error from the N2O signal, eliminating systematic bias, and significantly improving the inversion accuracy of N2O column concentration.

Description

Technical Field

[0001] This invention relates to the field of atmospheric environment remote sensing technology, and in particular to a method for inverting atmospheric N2O column concentration based on thermal infrared satellite remote sensing. Background Technology

[0002] Nitrous oxide (N2O) has a warming potential approximately 298 times that of CO2 and participates in stratospheric ozone depletion. With increasing emissions from industrial and agricultural sources, global atmospheric N2O concentrations continue to rise. Therefore, high-precision, high-spatial-temporal resolution global monitoring of atmospheric N2O concentrations will provide crucial scientific data support for climate change research and the formulation of greenhouse gas emission reduction policies.

[0003] Satellite remote sensing, due to its global coverage, long-term data series, and high spatiotemporal resolution, has been widely used for atmospheric composition monitoring. Hyperspectral infrared (HIR) detection technology is the primary means of monitoring atmospheric N2O concentration. Currently, atmospheric N2O retrieval research has been conducted based on satellite HIR instruments (such as IASI and CrIS). The HIRAS-II instrument carried by my country's FY-3E satellite, as the core payload of the world's first operational meteorological satellite in a dawn / dusk orbit, provides a unique data source that complements existing morning / afternoon satellites for global atmospheric composition monitoring.

[0004] However, retrieving atmospheric N2O concentration based on satellite monitoring information faces numerous technical challenges. Firstly, the characteristic absorption spectrum of N2O significantly overlaps with other atmospheric components (e.g., at 4.5 μm (~2150–2250 cm⁻¹)). -1 (1) The characteristic absorption band overlaps with the absorption bands of gases such as CO2 and H2O. Traditional channel selection methods are insufficient to effectively separate the contributions of multiple gases. Secondly, the accuracy of N2O inversion is highly dependent on the accuracy of the input data, especially surface temperature, which has a significant impact on inversion accuracy. ERA5 reanalysis data provides high spatiotemporal resolution atmospheric parameters globally, but its spatial resolution is relatively coarse, making it difficult to represent the true surface temperature at the satellite pixel scale, especially in areas with complex topography or heterogeneous underlying surfaces. This uncertainty is directly transmitted to the N2O inversion results, leading to systematic bias. If the prior surface temperature is too low, the algorithm will incorrectly lower the inverted N2O concentration to fit the observed brightness temperature, resulting in a significant negative bias. Summary of the Invention

[0005] The purpose of this invention is to provide a method for inverting atmospheric N2O column concentration based on thermal infrared satellite remote sensing, which solves the problem of systematic deviation caused by cross-interference of CO2 spectrum and surface temperature error in the existing technology.

[0006] To achieve the above objectives, this invention provides a method for inverting atmospheric N2O column concentration based on thermal infrared satellite remote sensing, comprising the following steps: Step 100: Acquire atmospheric apparent radiation spectrum data observed by the spaceborne hyperspectral infrared detector, and perform spectral calibration and radiometric calibration on the atmospheric apparent radiation spectrum data to obtain preprocessed radiation spectrum data; at the same time, acquire atmospheric background parameters, including atmospheric temperature profile, atmospheric humidity profile, atmospheric pressure profile, initial value of surface temperature, and prior gas concentration profile. Step 200: Select wavenumbers ranging from 2150 to 2250 cm⁻¹ within the 4.5 μm mid-infrared band from the preprocessed radiation spectral data. -1 The spectral range is used as the inversion characteristic spectral region, which covers the N2O molecule. Fundamental frequency vibration absorption band; Step 300: Input atmospheric background parameters into the RTTOV fast radiative transfer model, construct a forward simulation operator, and calculate the simulated brightness temperature; construct an extended state vector, which includes... Stratified atmospheric N2O concentration and surface temperature; Step 400: Calculate the Jacobian matrix of each candidate channel in the inverted characteristic spectral region for N2O and CO2 based on the RTTOV fast radiative transfer model, calculate the N2O sensitivity score and CO2 sensitivity score of each candidate channel, construct the N2O to CO2 sensitivity ratio index, and screen the inverted channels according to the N2O sensitivity score and sensitivity ratio index to obtain the characteristic channel set. Step 500: Construct a cost function containing observation constraints and prior constraints using the optimal estimation method, and use the Levenberg-Marquardt iterative algorithm to minimize the cost function, while simultaneously obtaining the corrected surface temperature and N2O atmospheric vertical concentration profiles; Step 600: Perform quality control on the N2O atmospheric vertical concentration profile, discard inversion results that do not meet the quality criteria, and perform column concentration integration calculation on the N2O atmospheric vertical concentration profile that has passed the quality control to obtain the atmospheric N2O column concentration inversion result.

[0007] Furthermore, in step 200, after selecting the inversion characteristic spectral region, spectral channels with a signal-to-noise ratio lower than a preset signal-to-noise ratio threshold and abnormal spectral channels affected by detector blind elements are removed from the inversion characteristic spectral region.

[0008] Furthermore, in step 300, the forward simulation formula for the RTTOV fast radiative transfer model is: ; in, For satellite observation of brightness temperature vector, This is the forward simulation function for the RTTOV fast radiative transfer model. To expand the state vector, This is the atmospheric background parameter vector. This is the sum of instrument noise and model error.

[0009] Furthermore, in step 400, the formula for calculating the Jacobian matrix is: ; in, For Jacobian matrices, This is the simulated brightness temperature vector output by the RTTOV fast radiative transfer model. To expand the state vector, It is a partial differential operator.

[0010] Furthermore, in step 400, the formula for calculating the N2O sensitivity score is as follows: ; The formula for calculating the CO2 sensitivity score is: ; in, For the first N2O sensitivity score for each channel, For the first CO2 sensitivity scores for each channel, For the first The first channel in the The N2O Jacobian value of the layer, For the first The first channel in the CO2 Jacobian value of the layer This represents the total number of vertical atmospheric strata.

[0011] Furthermore, in step 400, the formula for calculating the sensitivity ratio index is: ; in, For the first The N2O to CO2 sensitivity ratio index for each channel.

[0012] Furthermore, in step 400, the method for selecting inversion channels based on N2O sensitivity scores and sensitivity ratios to obtain the feature channel set is as follows: all candidate channels are selected according to their N2O sensitivity scores. Sort from high to low, select a set number of items that appear at the top of the sort. The channels form a set of characteristic channels.

[0013] Furthermore, in step 500, the expression for the cost function is: ; in, Let cost function be This is the simulated brightness temperature vector output by the RTTOV fast radiative transfer model. The observation error covariance matrix, The prior state vector, Let be the prior error covariance matrix. This is a transpose.

[0014] Furthermore, in step 500, the state vector update formula for the Levenberg-Marquardt iterative algorithm is: ; in, For the first The extended state vector of the next iteration For the first The extended state vector of the next iteration For the first Jacobian matrix at the next iteration The damping factor, For the first The simulated brightness temperature vector output by the RTTOV fast radiative transfer model in the next iteration; the cost function of two adjacent iterations. When the relative change is less than the preset convergence threshold, the iteration stops and the current extended state vector is output as the final inversion result.

[0015] Therefore, the present invention employs the above-mentioned method for inverting atmospheric N2O column concentration based on thermal infrared satellite remote sensing, which has the following beneficial effects: This method overcomes the limitations of traditional approaches that treat land surface temperature as a fixed known quantity. By incorporating land surface temperature into the state vector for joint inversion, it effectively decouples land surface temperature errors from N2O signals, significantly improving the reliability and accuracy of N2O products. Employing a novel channel optimization strategy based on the N2O / CO2 Jacobi ratio, compared to traditional methods based solely on N2O sensitivity, it more effectively eliminates the influence of interfering gases such as CO2, ensuring that the spectral information used in the inversion primarily originates from variations in N2O itself, thus enhancing the physical consistency and stability of the inversion process. Combining the RTTOV fast radiative transfer model with an efficient optimization iterative algorithm, it significantly improves computational efficiency while maintaining physical inversion accuracy, enabling operational N2O monitoring on a global scale and providing high-precision, continuous scientific data support for global greenhouse gas balance assessment and climate change research.

[0016] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0017] Figure 1 This is a flowchart of an atmospheric N2O column concentration inversion method based on thermal infrared satellite remote sensing according to the present invention; Figure 2 This is the N2O vertical profile diagram obtained by N2O inversion in this invention; Figure 3 This is a graph showing the spectral fitting results of the N2O inversion profile only in this invention; Figure 4 This is the N2O vertical profile obtained by the combined temperature inversion of this invention; Figure 5 This is a graph showing the spectral fitting results of the profile corresponding to the joint temperature inversion of this invention; Figure 6 This is a comparison chart of the percentage distribution of deviations between N2O inversion and combined temperature inversion in this invention. Detailed Implementation

[0018] The following detailed description of embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely illustrates selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0019] It should be noted that the specific terms used in this embodiment are explained as follows: RTTOV (Radiative Transfer for TOVS) is a fast radiative transfer model developed by the European Centre for Medium-Range Weather Forecasts (ECMWF) that can quickly simulate the radiation brightness temperature of the top of the atmosphere as observed by spaceborne infrared detectors. The Optimal Estimation Method (OEM) is an inversion algorithm that fuses observational information with prior knowledge through a Bayesian framework. The Jacobian matrix is ​​a partial derivative matrix that describes how sensitive the model output is to changes in the input parameters. The Levenberg-Marquardt algorithm (LM algorithm) is a nonlinear least squares iterative optimization algorithm that combines the global search capability of gradient descent with the local convergence speed of Gauss-Newton method. ERA5 is the fifth-generation global atmospheric reanalysis dataset provided by the European Centre for Medium-Range Weather Forecasts. The MIPAS extended atmospheric model is a global atmospheric model built by IMK / IAA (Institute for Meteorology and Climatology, Karlsruhe Institute of Technology / Institute for Astrophysics, Andalusia) based on MIPAS observations. It covers the vertical profiles of various trace gases, including N2O, from the ground to an altitude of 200 km, and distinguishes typical spatiotemporal scenarios such as equatorial, mid-latitude, and polar seasonality.

[0020] Please see Figure 1 A method for inverting atmospheric N2O column concentration based on thermal infrared satellite remote sensing includes the following steps: Step 100: Acquire satellite observation data and auxiliary information, and perform preprocessing; This embodiment selects L1-level hyperspectral infrared atmospheric apparent radiation spectral data acquired by the HIRAS-II detector aboard the FY-3E satellite throughout 2023 as the satellite observation data source. The HIRAS-II detector covers three bands: long-wave infrared, mid-wave infrared, and short-wave infrared, with a spectral resolution of 0.625 cm⁻¹. -1 .

[0021] Simultaneously, the following auxiliary data were obtained as atmospheric background parameters: atmospheric temperature profile, atmospheric humidity profile, atmospheric pressure profile, and initial surface temperature values ​​from ERA5 reanalysis data; and N2O profile data from the MIPAS atmospheric extended model were used as prior gas concentration profiles.

[0022] The acquired HIRAS-II L1 level data underwent spectral and radiometric calibration to confirm the accuracy of the observed spectral wavenumbers and the quantitative precision of the radiative intensity. ERA5 reanalysis data were temporally and spatially matched with satellite observation pixels for interpolation to obtain the atmospheric background parameters corresponding to each observation pixel. This yielded preprocessed radiometric spectral data and the matched atmospheric background parameters.

[0023] Step 200: Determine the inversion characteristic spectral region; Due to the mid-infrared 4.5μm (~2150–2250cm) -1 ) Covered N2O molecules The fundamental frequency vibration absorption band exhibits a strong absorption cross-section for N2O, demonstrating high sensitivity to changes in N2O concentration in the lower troposphere. The FY-3E satellite operates in a dawn-dusk orbit, during which the solar zenith angle is large, and the interference from solar reflected radiation in the mid-infrared 4.5μm band is less than that of morning and afternoon satellites, which is beneficial for improving the stability of daytime inversion results in this band. Therefore, the HIRAS mid-infrared 4.5μm band was selected, with the spectral range defined as 2150–2250 cm⁻¹. -1 After selecting the inversion characteristic spectral region, spectral channels with a signal-to-noise ratio lower than the preset signal-to-noise ratio threshold are removed, and the effective candidate channel set is retained.

[0024] Step 300: Construct the forward simulation operator and calculate the simulated brightness temperature; Atmospheric background parameters are input into the RTTOV fast radiative transfer model to construct a forward simulation operator. The RTTOV fast radiative transfer model, based on a pre-calculated optical thickness lookup table, rapidly calculates the radiative brightness temperature of the upper atmosphere under given atmospheric and surface conditions. The forward simulation relationship is expressed as: ; in, For satellite observation of brightness temperature vector, The input to the forward analog function is and Simulated brightness temperature vector at time, To expand the state vector, These are atmospheric background parameters (including atmospheric temperature profile, atmospheric humidity profile, atmospheric pressure profile, and a priori gas concentration profile). This is the sum of instrument noise and model error. In subsequent inversion iterations, the atmospheric background parameter vector... Keeping it constant, the forward simulation function is therefore abbreviated as: .

[0025] Define extended state vector for: ; in, For the first Atmospheric N2O concentration , This represents the total number of vertical atmospheric layers. For surface temperature. Extended state vector. The dimension is , among which the former The component corresponds to the N2O concentration in each layer, the first component... One component is surface temperature. This is a transpose. Surface temperature is incorporated into the extended state vector, allowing the inversion algorithm to adaptively adjust the surface temperature based on spectral information during the iteration process, thereby eliminating the systematic inversion bias introduced by surface temperature deviations in the reanalysis data.

[0026] Step 400: Channel optimization based on the N2O to CO2 Jacobi sensitivity ratio index, specifically including the following sub-steps: Step 401: Calculate the Jacobian matrix; The prior state vector Input atmospheric background parameters into the RTTOV fast radiative transfer model, call the K-model module of the RTTOV fast radiative transfer model, calculate the partial derivatives of the observed brightness temperature with respect to each component of the extended state vector, and obtain the Jacobian matrix. .

[0027] It should be noted that the prior state vector The N2O profile provided by the MIPAS atmospheric extension model and the initial surface temperature value provided by the ERA5 reanalysis data together constitute the N2O profile.

[0028] The formula for calculating the Jacobian matrix is: ; in, For Jacobian matrices, This is the simulated brightness temperature vector output by the RTTOV fast radiative transfer model. To extend the state vector. Jacobian matrix. The The element represents the first element. The simulated brightness temperature of the first channel affects the second channel. The partial derivatives of each state variable. Extracting each candidate channel under a given background atmospheric condition. In all vertical layers Jacobian ratio of N2O Jacobian ratio to CO2 , It is a partial differential operator.

[0029] Step 402: Calculate the N2O sensitivity score and CO2 sensitivity score for each candidate channel; The N2O sensitivity score is the sum of the absolute values ​​of the N2O Jacobian values ​​across all vertical layers, reflecting the overall response strength of the channel to changes in N2O concentration throughout the atmosphere. The CO2 sensitivity score is calculated in the same way. The calculation formulas are as follows: ; ; in, For the first N2O sensitivity score for each channel, For the first CO2 sensitivity scores for each channel, For the first The first channel in the The N2O Jacobian value of the layer, For the first The first channel in the CO2 Jacobian value of the layer This represents the total number of vertical atmospheric strata.

[0030] Step 403: Calculate the N2O to CO2 sensitivity ratio for each candidate channel; this ratio reflects the channel's dominance in response to N2O relative to CO2 interference. The calculation formula is: ; in, For the first The N2O to CO2 sensitivity ratio index for each channel. The larger the value, the more dominant the channel's response to N2O is relative to CO2 interference.

[0031] Step 404: Filter the inversion feature channels; All candidate channels were scored based on N2O sensitivity. Sort from highest to lowest, and select the one at the top of the list. The channels that meet the above criteria are used to form a feature channel set. In this embodiment, 25 channels that meet the above criteria are ultimately selected to construct the feature channel set. The Jacobian response of the selected channel is concentrated in the lower troposphere, and its N2O to CO2 sensitivity ratio index All values ​​are greater than 1, indicating that the selected channel responds more strongly to N2O interference than to CO2 interference.

[0032] Step 405: Calculate the degrees of freedom.

[0033] Degrees of freedom (DFS) are an important indicator for measuring the information content of the observation system regarding the target parameters. A higher DFS value indicates a stronger observation-driven inversion result and less influence from prior constraints, and can be used to evaluate the vertical resolution of the inversion. This study uses the Jacobian matrix calculated by RTTOV based on optimal estimation theory. Prior error covariance matrix and instrument noise covariance matrix The degrees of freedom can be calculated. The calculation formula is as follows: ; in, Indicates degrees of freedom. The trace operation is represented; in this embodiment, with 25 observation channels, the degree of freedom is calculated to be 1.5, indicating that the inversion system can independently extract about 1.5 vertical structure features, which is consistent with the limited amount of information that can only be inverted from the N2O profile and surface temperature.

[0034] Step 500: Construct the cost function and perform joint optimal estimation iterative inversion; specifically including the following sub-steps: Step 501: Construct the cost function; The optimal estimation method is used to construct a cost function that includes observation constraints and prior constraints. Its expression is: ; in, For input The cost function at that time, For satellite observation brightness temperature vector, i.e. The vectorized representation of, The input for the RTTOV fast radiative transfer model is The simulated brightness temperature vector output at that time. The instrument noise covariance matrix is... The prior state vector, Let be the prior error covariance matrix. It is constructed based on the relative error and vertical correlation length of the prior profile, and is used to control the dependence of the inversion solution on prior knowledge and the vertical smoothness. Instrument noise covariance matrix. The instrument's Noise Equivalent Temperature Difference (NEDT) is set as a diagonal matrix to represent the random error of the observed brightness temperature, and thus determines the weight of the observed data in the inversion cost function.

[0035] The first term of the cost function is the observation constraint term, which measures the weighted residual between the simulated brightness temperature and the observed brightness temperature; the second term is the prior constraint term, which ensures that the inversion result does not deviate too far from prior knowledge. The goal of the optimal estimation method is to find the term that makes the cost function... Extended state vector that achieves minimum value .

[0036] Step 502: Perform the Levenberg-Marquardt iterative solution.

[0037] With prior state vector as initial value for iteration In each iteration, the current state vector is calculated using the RTTOV fast radiative transfer model. Corresponding simulated brightness temperature vector And Jacobi matrix Update the extended state vector according to the following formula: ; in, For the first The extended state vector of the next iteration For the first The extended state vector of the next iteration For the first Jacobian matrix at the next iteration The damping factor, The instrument noise covariance matrix is... Let be the prior error covariance matrix. The prior state vector, For satellite observation of brightness temperature vector, For the first The simulated brightness temperature vector output by the RTTOV fast radiative transfer model at the next iteration. Damping factor. Dynamically adjust during iteration: decrease as the cost function decreases. To accelerate convergence, increase the cost function as it increases. To enhance stability. When At this point, the above formula degenerates into a standard Gauss-Newton iteration. Iteration stops when the relative change in the cost function between two consecutive iterations satisfies the following condition: ; in, and These represent the inputs as follows: and The cost function at that time, To preset the convergence threshold, this embodiment takes... , This represents the rate of change of the cost function. If the convergence condition is met, the iteration stops, and the final extended state vector, containing the corrected surface temperature, is output. The vertical concentration profile of N2O in the atmosphere is obtained; if the convergence condition is not met, the iteration continues until the convergence condition is met.

[0038] Step 600: Quality control and calculation of atmospheric N2O column concentration.

[0039] In embodiments of the present invention, the final extended state vector is sequentially filtered using the following four quality control criteria: (1) Instrument quality identification criteria: Only observation samples with a satellite observation data quality identification QA value of 100 are retained. The QA value represents the instrument working status and data quality level of the HIRAS-II detector at the observation time. QA=100 indicates that the instrument is in the best working state and the observation data has the highest signal-to-noise ratio.

[0040] (2) Temperature deviation threshold criterion: The surface temperature obtained by calculation and inversion With prior surface temperature difference Eliminate abnormal temperature deviations ( The observed samples show that a large deviation in surface temperature indicates that the inversion process of that pixel is affected by cloud contamination or surface heterogeneity.

[0041] (3) Physical range criterion for concentration: Observational samples with N2O column concentrations exceeding the physical background range (400 ppb) obtained from the inversion are excluded. This threshold is determined based on the physical background range of the current global atmospheric N2O concentration.

[0042] (4) Spectral fitting residual criterion (error analysis): Calculate the observed brightness temperature vector With the final simulated brightness temperature vector Chi-square test value between Only retain The observed sample. Chi-square test value. The overall goodness of fit between observations and simulations is measured by a large value. The value indicates that the model failed to adequately fit the observed spectrum, and the inversion result for this pixel is unreliable.

[0043] For the N2O atmospheric vertical concentration profile that passes all the above quality control criteria, the atmospheric N2O column concentration is calculated by weighted integration using the pressure thickness of each layer.

[0044] like Figure 2-5 As shown, this method was compared and validated with observational data from the TCCON Hefei station (32°N, 117°E) on March 14, 2023. The results show that... Figure 2 The spectral fitting results for the traditional inversion scheme, which only retrieves N2O and fixes the ERA5 surface temperature, are shown below. Figure 3 As shown, a certain systematic bias still exists between the spectral simulation and observation, with an average residual of approximately 0.152 K. To mitigate the interference of surface temperature errors on gas inversion, this method incorporates surface temperature as part of the state vector into the optimal estimation inversion framework, solving it jointly with the N2O profile parameters during the inversion process. Utilizing the high sensitivity of the 4.5 μm band to surface thermal radiation, the cost function is iteratively minimized using the LM algorithm, achieving an adaptive correction of the surface temperature from the initial ERA5 value of 294.5 K to 297.0 K. Figure 5 As can be seen, after joint inversion, the spectral fitting residual is significantly reduced to the order of 0.01K, and the agreement between the observed brightness temperature and the simulated brightness temperature is significantly improved; at the same time, by Figure 4 It can be seen that the N2O vertical profile and column concentration obtained by joint inversion maintain good consistency with the TCCON reference value of 320.74 ppb. The above results indicate that incorporating surface temperature into joint inversion can effectively reduce the interference of surface thermal radiation errors on gas inversion, thereby reducing systematic bias and improving the accuracy and stability of N2O inversion results.

[0045] This invention further evaluates the performance of the joint inversion algorithm in improving accuracy by statistically analyzing the percentage of deviation among 32 successfully inverted samples. For example... Figure 6The figure shows a comparison of the percentage distribution of biases for the two inversion methods (N2O inversion only and joint temperature inversion). The joint inversion algorithm significantly outperforms the traditional method in both error control and convergence stability: in the range where the absolute bias is less than 1.5%, the joint inversion uses 17 samples, more than the N2O inversion alone, demonstrating a better trend in error reduction. Statistically, the joint inversion algorithm reduces the mean absolute bias from 2.57% to 2.02%, and simultaneously reduces the standard deviation from 2.44% to 1.74%. This simultaneous decrease in mean and dispersion indicates that simultaneous correction of surface temperature can enhance the consistency of N2O column concentration inversion results under different scenarios. In the range where the bias is greater than 6%, the joint inversion successfully suppresses the occurrence of extreme samples, making the distribution more tightly concentrated in the low-bias range, verifying the important role of this algorithm in reducing surface thermal radiation interference and improving N2O detection accuracy.

[0046] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for inverting atmospheric N2O column concentration based on thermal infrared satellite remote sensing, characterized in that, Includes the following steps: Step 100: Acquire atmospheric apparent radiation spectrum data observed by the spaceborne hyperspectral infrared detector, and perform spectral calibration and radiometric calibration on the atmospheric apparent radiation spectrum data to obtain preprocessed radiation spectrum data; at the same time, acquire atmospheric background parameters, including atmospheric temperature profile, atmospheric humidity profile, atmospheric pressure profile, initial value of surface temperature, and prior gas concentration profile. Step 200: Select wavenumbers ranging from 2150 to 2250 cm⁻¹ within the 4.5 μm mid-infrared band from the preprocessed radiation spectral data. -1 The spectral range is used as the inversion characteristic spectral region, which covers the N2O molecule. Fundamental frequency vibration absorption band; Step 300: Input atmospheric background parameters into the RTTOV fast radiative transfer model, construct a forward simulation operator, and calculate the simulated brightness temperature; Construct an extended state vector, which contains Stratified atmospheric N2O concentration and surface temperature; Step 400: Calculate the Jacobian matrix of each candidate channel in the inverted characteristic spectral region for N2O and CO2 based on the RTTOV fast radiative transfer model, calculate the N2O sensitivity score and CO2 sensitivity score of each candidate channel, construct the N2O to CO2 sensitivity ratio index, and screen the inverted channels according to the N2O sensitivity score and sensitivity ratio index to obtain the characteristic channel set. Step 500: Construct a cost function containing observation constraints and prior constraints using the optimal estimation method, and use the Levenberg-Marquardt iterative algorithm to minimize the cost function, while simultaneously obtaining the corrected surface temperature and N2O atmospheric vertical concentration profiles; Step 600: Perform quality control on the N2O atmospheric vertical concentration profile, discard inversion results that do not meet the quality criteria, and perform column concentration integration calculation on the N2O atmospheric vertical concentration profile that has passed the quality control to obtain the atmospheric N2O column concentration inversion result.

2. The method for inverting atmospheric N2O column concentration based on thermal infrared satellite remote sensing according to claim 1, characterized in that, In step 200, after selecting the inversion characteristic spectral region, spectral channels with a signal-to-noise ratio lower than the preset signal-to-noise ratio threshold and abnormal spectral channels affected by detector blind elements are removed from the inversion characteristic spectral region.

3. The method for inverting atmospheric N2O column concentration based on thermal infrared satellite remote sensing according to claim 2, characterized in that, In step 300, the forward simulation formula for the RTTOV fast radiative transfer model is: ； in, For satellite observation of brightness temperature vector, This is the forward simulation function for the RTTOV fast radiative transfer model. To expand the state vector, This is the atmospheric background parameter vector. This is the sum of instrument noise and model error.

4. The method for inverting atmospheric N2O column concentration based on thermal infrared satellite remote sensing according to claim 3, characterized in that, In step 400, the formula for calculating the Jacobian matrix is: ； in, For Jacobian matrices, This is the simulated brightness temperature vector output by the RTTOV fast radiative transfer model. To expand the state vector, It is a partial differential operator.

5. The method for inverting atmospheric N2O column concentration based on thermal infrared satellite remote sensing according to claim 4, characterized in that, In step 400, the formula for calculating the N2O sensitivity score is: ； The formula for calculating the CO2 sensitivity score is: ； in, For the first N2O sensitivity score for each channel, For the first CO2 sensitivity scores for each channel, For the first The first channel in the The N2O Jacobian value of the layer, For the first The first channel in the CO2 Jacobian value of the layer This represents the total number of vertical atmospheric strata.

6. The method for inverting atmospheric N2O column concentration based on thermal infrared satellite remote sensing according to claim 5, characterized in that, In step 400, the formula for calculating the sensitivity ratio index is: ； in, For the first The N2O to CO2 sensitivity ratio index for each channel.

7. The method for inverting atmospheric N2O column concentration based on thermal infrared satellite remote sensing according to claim 6, characterized in that, In step 400, the method for selecting inversion channels based on N2O sensitivity scores and sensitivity ratios to obtain the feature channel set is as follows: all candidate channels are selected according to their N2O sensitivity scores. Sort from high to low, select a set number of items that appear at the top of the sort. The passage It forms a set of feature channels.

8. The method for inverting atmospheric N2O column concentration based on thermal infrared satellite remote sensing according to claim 7, characterized in that, In step 500, the expression for the cost function is: ； in, Let cost function be This is the simulated brightness temperature vector output by the RTTOV fast radiative transfer model. The observation error covariance matrix, Let the prior state vector be... Let be the prior error covariance matrix. This is a transpose.

9. The method for inverting atmospheric N2O column concentration based on thermal infrared satellite remote sensing according to claim 8, characterized in that, In step 500, the state vector update formula of the Levenberg-Marquardt iterative algorithm is: ； in, For the first The extended state vector of the next iteration For the first The extended state vector of the next iteration For the first Jacobian matrix at the next iteration The damping factor, For the first The simulated brightness temperature vector output by the RTTOV fast radiative transfer model in the next iteration; the cost function of two adjacent iterations. When the relative change is less than the preset convergence threshold, the iteration stops and the current extended state vector is output as the final inversion result.

Images