Collaborative retrieval of ozone profile from FY-3F satellite nadir and limb ultraviolet hyperspectral measurements

Abstract

The application discloses a FY-3F satellite OMS-N nadir and OMS-L limb ultraviolet hyperspectral ozone collaborative inversion method, and relates to the field of satellite atmospheric remote sensing. The method acquires FY-3F satellite OMS-N nadir and OMS-L limb synchronous ultraviolet spectral data, sets an ozone state vector from the ground to the top of the atmosphere, simulates spectral radiation by using a nadir and limb radiation transmission model, constructs a joint cost function fusing double-source observation residual and prior constraint, and inverses the full-atmosphere ozone profile by Levenberg-Marquardt iteration optimization. The application is based on deep fusion of synchronous original spectral information of the same platform, overcomes the limitations of insufficient resolution of the upper layer of the nadir-type stratosphere and insensitivity of the lower layer of the limb-type troposphere, eliminates the time and space matching errors of multi-satellite fusion, realizes high-precision and high-vertical-resolution ozone profile inversion from the troposphere to the stratosphere, and can provide reliable data for atmospheric environment monitoring and climate research.

Description

Technical Field

[0001] This invention belongs to the field of satellite atmospheric remote sensing and ozone profile inversion technology, and relates to a method for the coordinated inversion of nadir and limb ultraviolet hyperspectral ozone from the Fengyun-3F satellite. Specifically, it relates to an algorithm that uses the ultraviolet hyperspectral ozone detector carried by the Fengyun-3F satellite (FY-3F) to achieve high-precision inversion of the entire atmospheric ozone profile from the troposphere to the stratosphere through coordinated observation of the nadir module (OMS-N) and the limb module (OMS-L). Background Technology

[0002] Ozone is a crucial trace gas in the atmosphere, and its vertical distribution has a decisive impact on ultraviolet radiation absorption, atmospheric chemical processes, climate change, and air quality. Accurately obtaining the vertical profile of ozone from the troposphere to the stratosphere is an urgent need for atmospheric environmental monitoring and climate research. Satellite ultraviolet hyperspectral remote sensing, due to its high spectral resolution and global coverage, has become the mainstream method for vertical ozone detection.

[0003] my country's Fengyun-3F satellite (FY-3F), successfully launched in 2023, carries an advanced ultraviolet hyperspectral ozone detector. This instrument innovatively integrates two independent detection modules: the nadir-type module (OMS-N) and the limb-type module (OMS-L). OMS-N observes ultraviolet radiation reflected and scattered by the Earth's atmosphere and surface from a nadir perspective, acquiring spectral data with high signal-to-noise ratio and wide horizontal coverage. Its ozone profile products can reflect the ozone distribution from the troposphere to the lower stratosphere well, but its vertical resolution is limited, especially above the ozone peak layer in the stratosphere. OMS-L scans the atmospheric tangential layer in a limb-type manner, inverting upper-level ozone by detecting atmospheric limb-scattered radiation. It can provide a high-vertical-resolution ozone profile from the top of the troposphere to the middle and upper stratosphere. However, its detection altitude is limited, it is not sensitive to ozone in the middle and lower troposphere, especially the boundary layer, and its horizontal coverage is relatively low.

[0004] It is evident that relying solely on any one of the detection modules is insufficient to independently achieve high-precision ozone profile monitoring across the entire atmosphere, from the troposphere to the stratosphere. Existing international multi-source ozone profile fusion products, such as those combining inversion results from nadir observations (e.g., OMI) and lateral edge observations (e.g., MLS), typically originate from different satellite platforms, resulting in spatiotemporal matching errors. Furthermore, these are post-inversion statistical fusions, failing to jointly utilize the original radiation information during the inversion process, thus limiting the consistency and accuracy of the full-profile ozone profile. The unique advantage of the Fengyun-3F satellite's dual-module observations on the same platform makes deep collaboration from the signal source possible. However, currently, there is a lack of an algorithm that can fully utilize OMS-N and OMS-L synchronous ultraviolet hyperspectral radiation data and achieve complementary advantages between the two within a unified inversion framework. This prevents the production of high-precision, high-vertical-resolution full-atmosphere ozone profile products from the troposphere to the stratosphere.

[0005] Therefore, there is an urgent need in this field for a new method for ozone co-retrieval that deeply integrates nadir and edge observations.

[0006] The information disclosed in this background section is intended only to enhance the understanding of the overall background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0007] The purpose of this invention is to provide a new method for ozone co-retrieval that integrates nadir and fringe observation depth.

[0008] To achieve the above objectives, the present invention provides the following solution: A method for co-retrieval of nadir and lateral ultraviolet hyperspectral ozone from Fengyun-3F satellite includes the following steps: Step S1: Obtain the observed spectral radiance vector y from the OMS-N nadir module, synchronously collected by the ultraviolet hyperspectral ozone detector carried by Fengyun-3F satellite in the target geographic region. N The observed spectral radiation vector y of the adjacent module OMS-L L Wherein, the observed spectral radiation vector y of the OMS-N N Covering the 265–330 nm ultraviolet band, with a spectral sampling interval of no more than 0.5 nm; the observed spectral radiative vector y of the OMS-L L Covering the same 265–330 nm band, corresponding to multiple tangent heights from the Earth's surface to approximately 80 km, with adjacent tangent heights having a step size of 1–2 km; Step S2: Define a state vector x representing the vertical distribution of ozone from the Earth's surface to the top of the atmosphere. This state vector divides the atmosphere into multiple layers in the vertical direction, covering the entire altitude range from the troposphere to the stratosphere. The elements of the state vector x are the natural logarithm of the ozone volume mixing ratio of each vertical layer. The vertical coordinates are given in terms of pressure levels, extending from the reference surface pressure of 1000 hPa to the top of the atmosphere of 0.1 hPa, dividing the atmosphere into 45 to 72 layers. The layer thickness in the troposphere is 0.8 to 1.2 km, and the layer thickness in the stratosphere is 1.5 to 2.5 km. In the ozone peak region of 10 to 50 hPa, the layer thickness is further increased to 1 km. Step S3: Using the forward model of nadir radiative transfer F N (x) and the forward model of lateral radiative transfer F L(x), respectively, the nadir simulated spectral radiation and the limb simulated spectral radiation are obtained by simulating the state vector x; wherein, the nadir forward model is based on the vector discrete coordinate radiative transfer model and includes the absorption of gases such as ozone, NO2, SO2, and O2-O2, as well as Rayleigh scattering and Lambetian surface reflection, and at the same time, the Ring spectrum is used as a pseudo-absorber in the simulation to correct the Ring effect; the limb forward model is based on the spherical atmospheric multiple scattering radiative transfer model, and introduces atmospheric refraction and non-uniformity parameterization schemes caused by cloud ice crystal / water droplet scattering in the radiative transfer calculation; Step S4: Construct a joint cost function that simultaneously integrates nadir and fringe observations. J (x), represented as: J (x)=[y N -F N (x)] T S [y N -F N (x)]+[y L -F L (x)] T S [y L -F L (x)]+[xx a ] T S [xx a ]; In the formula, y N Let F be the nadir observation spectral radiation vector. N (x) represents the simulated spectral radiation of the nadir forward model for the state vector x, S N The covariance matrix of the nadir measurement error is taken as a diagonal matrix, whose diagonal elements are determined by the square of the noise equivalent radiance of each ultraviolet spectral channel, and the additional error contributions caused by residual cloud contamination and surface inhomogeneity are considered; y L Spectral radiation vector of the near edge observation, F L (x) represents the simulated spectral radiation of the adjacent forward model on the state vector x, S L The covariance matrix of the edge measurement error includes diagonal measurement noise terms and off-diagonal forward model error terms. The off-diagonal elements are constructed using an exponentially decaying correlation model based on the spatial correlation length (taken as 5–8 km) of atmospheric inhomogeneity and cloud scattering. a Let S be the prior ozone profile vector. a The prior error covariance matrix adopts a Gaussian correlation model, and its variance is given by the monthly average variance profile of ozone climatology. The correlation length is set to 3-5 km in the troposphere and 7-10 km in the stratosphere.

[0009] Step S5: Using x as the parameter to be inverted, minimize the joint cost function through an iterative optimization algorithm. J (x), the iteration termination condition is that the relative change of the cost function is less than 10. -3 Or the maximum relative change of the state vector is less than 10. -3 The maximum number of iterations does not exceed 20 to obtain the optimal estimate of the ozone profile. This optimal estimate provides a full atmospheric ozone profile from the troposphere to the stratosphere, and simultaneously outputs the inversion error of each layer and the vertical resolution information represented by the average kernel matrix.

[0010] Optionally, the state vector x in step S2 represents the distribution of the natural logarithm of the ozone volume mixing ratio with altitude, with the vertical stratification range from the surface to 0.1 hPa (approximately 60 km), wherein the stratification thickness in the troposphere is 1–2 km and the stratification thickness in the stratosphere is 2–3 km.

[0011] Optionally, the nadir measurement error covariance matrix S in step S4 N The matrix is ​​taken as a diagonal matrix, and the diagonal elements are determined by the noise equivalent radiance of each ultraviolet spectral channel at the nadir; the lateral measurement error covariance matrix S L It includes measurement noise at the height of each tangent in the edge scan and error terms caused by non-uniform cloud scattering.

[0012] Optionally, the prior ozone profile vector x mentioned in step S4 a and its prior error covariance matrix S a Constructed from at least one of the following sources: climatological ozone database, climatological monthly mean profiles from satellite rim surveys, and ozone fields from reanalysis data.

[0013] Optionally, the joint cost function described in step S4 J The residual term of the near-edge observation of (x) is further multiplied by a highly correlated dynamic weighting factor W( z ), to enhance the contribution of upper stratospheric ozone signal, where W( z () is a diagonal weighted matrix that varies with the tangent height, and its diagonal elements w ( z i Take as , z i For the first i The height of the adjacent tangent line, z p The reference height for ozone peak is taken as 30–40 km. The control weight expansion is set at 8-15 km, and the enhancement coefficient α ranges from 1 to 5.

[0014] Optionally, the iterative update form of the iterative optimization algorithm described in step S5 is as follows: x i+1 =x i +(K S K i +S + D) -1 {K S [yF(x i )]-S [x i -x a ]}; In the formula, x i For the first i The state vector of the next iteration, K i Joint observation vector y=[y ,y ] T For state vector x at x i The Jacobian matrix at the datum is given by the Jacobian K at the datum. N And adjacent Jacobi K L S is pieced together The joint error covariance matrix includes nadir error and edge error, where λ is the damping factor and D is the diagonal scaling matrix.

[0015] Optionally, the synchronous observation of the nadir module OMS-N and the lateral edge module OMS-L in step S1 needs to meet the spatiotemporal matching conditions: the ground distance between the center of the nadir pixel and the lateral edge scanning tangent point is less than or equal to 50 km, and the observation time difference is less than or equal to 30 seconds.

[0016] Optionally, in step S3, the forward model incorporates Ring effect correction and polarization factor when performing radiative transfer calculations. The Ring effect involves using the Ring spectrum as a pseudo-absorber in the simulation, setting the vertical profile of the pseudo-absorber to be proportional to the atmospheric molecular number density, and using the total column scaling factor as an additional fitting parameter. The polarization correction factor is compensated by using the ratio of the pre-calculated vector radiative transfer simulation result to the scalar simulation result as a spectral function.

[0017] Optionally, step S5 involves obtaining the optimal estimate of the ozone profile. Simultaneously, output the inversion error covariance matrix. The average kernel matrix A is used to characterize the vertical resolution and sensitivity of the profile inversion information, where the average kernel matrix A is calculated according to A= Calculate and derive the degree-of-freedom components of each vertical layer. d i = Aii And the posterior estimates and uncertainties of the total tropospheric ozone column and the total stratospheric ozone column.

[0018] Optionally, in step S3, the forward model F of nadir radiative transfer... N (x) Based on the LIDORT or VLIDORT multi-stream discrete coordinate model, the adjacent edge radiative transfer forward model F L (x) Based on the spherical atmospheric limb radiative transfer model SCIATRAN or MCC++. Compared with the prior art, the present invention has the following beneficial effects: This invention utilizes the FY-3F satellite platform. Synchronous ultraviolet hyperspectral observations from the nadir and the OMS-L limb module are directly combined with the original radiative information under a unified optimization estimation and inversion framework. This deep fusion, originating from the signal source, completely eliminates the profile inconsistencies caused by spatiotemporal matching errors in traditional multi-satellite fusion products. It yields a high-precision, self-consistent ozone vertical profile across the entire atmospheric column from the troposphere to the stratosphere. Nadir observations ensure high sensitivity in the lower and middle troposphere, while limb observations provide high vertical resolution in the upper and middle stratospheric layers. The two complement each other in the joint cost function, overcoming the inherent limitations of single-detection modules that are "unclear at higher levels and unreachable at lower levels." The inversion results achieve a vertical resolution of 2–3 km in the ozone peak region, with a tropospheric bias of less than 5% and a mid-stratospheric bias of less than 2%, significantly improving the quality of the ozone profile product. By introducing a highly correlated dynamic weighting factor to enhance the contribution of the stratospheric ozone signal and combining it with a refined forward model that includes parameterization of the Ring effect, polarization correction, and cloud scattering inhomogeneity, the accuracy and stability of the inversion are further improved. Simultaneously, the inversion process outputs the error covariance matrix and the average kernel matrix, which can quantitatively assess the information content and effective vertical resolution at each altitude level, providing comprehensive diagnostic information for data assimilation and scientific applications. This method can be directly integrated into the FY-3F ground application system to realize the operational generation of ozone profile products, strongly supporting atmospheric environmental monitoring, air quality forecasting, and climate change research. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A schematic diagram of the process for the coordinated inversion of nadir and lateral ultraviolet hyperspectral ozone from the Fengyun-3F satellite, provided in an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] The purpose of this invention is to provide a new method for ozone co-retrieval that integrates nadir and fringe observation depth.

[0023] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0024] Example 1: This embodiment provides a method for the joint inversion of nadir and limb ultraviolet hyperspectral ozone from the Fengyun-3F satellite. This method is based on synchronous observation data from the OMS-N nadir ultraviolet hyperspectral detection module and the OMS-L limb ultraviolet hyperspectral detection module carried on the same platform as the FY-3F satellite. It utilizes a unified optimal estimation inversion framework to jointly process the two types of raw spectral radiation information, producing a high-precision full-atmosphere ozone profile from the troposphere to the stratosphere. The method is described in detail below with specific parameters and implementation procedures. Those skilled in the art can fully implement this technical solution based on this embodiment.

[0025] Step 1: Acquire synchronous observation data and perform spatiotemporal matching: OMS-N Level-1B nadir radiance data and OMS-L Level-1B limb radiance data, after radiometric calibration and geolocation, were acquired from the FY-3F ground application system. Both detector modules cover the 265–330 nm ultraviolet band, with a spectral sampling interval of approximately 0.4 nm. The OMS-N data is a sequence of nadir pixels, with each pixel accompanied by information such as latitude and longitude, observation time, and solar zenith angle. The OMS-L data is a set of limb scan spectra at different tangent heights, extending from the surface to approximately 80 km, with an adjacent tangent height step size of 1.5 km.

[0026] To ensure the spatiotemporal consistency of the collaborative inversion, a rigorous matching and screening process was performed on the OMS-N and OMS-L data: data pairs were selected where the distance between the nadir pixel center and the tangent point of the edge scan was no greater than 30 km on the ground, and the observation time difference was no more than 10 seconds. For each data pair that met the criteria, the OMS-N observation spectral radiance vector y was extracted. N (length is) N w(each spectral channel) and the OMS-L observed spectral radiance vector (length is...) N h × N w ,in N h (The height of the tangent layer at the edge) is used as the measurement vector for subsequent inversion. Simultaneously, auxiliary parameters such as temperature profiles, pressure profiles, and surface pressure at the matching location are obtained by interpolation from the reanalysis data.

[0027] Step 2: Define the atmospheric state vector and vertical stratification: The atmospheric state vector x, which characterizes the vertical distribution of ozone, is set as the natural logarithm of the ozone volume mixing ratio at each pressure level, i.e., x = [ln q 1,ln q 2,...,ln q n ] T ,in q k For the first k The ozone volume mixing ratio of the layers. The vertical coordinates adopt the pressure hierarchy, extending from the reference surface pressure of 1000 hPa to the top of the atmosphere of 0.1 hPa, and are divided into 55 layers. The specific layering scheme is shown in the table below. The layer thickness is increased in the altitude range of 10 to 50 hPa where the ozone concentration peak is located, so as to fully capture the fine structure of stratospheric ozone.

[0028] Table 1 Vertical Hierarchical Scheme for Ozone Co-inversion State Vector

[0029] This stratification scheme results in a layer thickness of approximately 0.8–1.2 km in the troposphere and approximately 1.5–2.5 km in the upper stratosphere, meeting the overall requirement of 45–72 layers and ensuring full-height coverage from the troposphere to the stratosphere.

[0030] Step 3: Construct forward modeling of radiative transfer at the nadir and lateral edges: Nadir forward model F N (x) Implemented based on the vector linearized discrete coordinate model VLIDORT v2.8. The simulation uses 8 discrete flux numbers, and the input atmospheric parameters include ERA5 temperature, air pressure, ozone, and NO. 2 SO 2 O 2 -O 2Prior climatological profiles of trace gases were derived, and the Lambetian surface reflectance assumption was introduced, with surface albedo taken as climatological values ​​varying with geographical location and season, ranging from 0.05 to 0.10. Rayleigh scattering and gas absorption cross sections were obtained from UV data in the HITRAN 2020 spectral database. To correct for the Ring effect, the Ring spectrum was used as a pseudo-absorber in the simulation: its vertical profile was set to be proportional to the atmospheric molecular number density, and the corresponding column scaling factor in the cost function was used. c Ring As an additional fitting parameter, it is simultaneously inverted with the ozone state vector.

[0031] Forward model F at the edge L (x) The model was constructed based on the spherical atmospheric lip radiative transfer model using SCIATRAN v4.5. Spherical refraction was enabled in the simulation, and a non-uniformity parameterization scheme based on scattering from cloud ice crystals and liquid water droplets was introduced into the radiative transfer calculation to describe the non-uniformity of cloud scattering along the line-of-sight direction. The input profile for the lip simulation was consistent with the nadir forward model, and the tangent heights were based on the actual scan geometry input of OMS-L, corresponding to the spectral radiation at each tangent height. For the influence of clouds, a two-dimensional non-uniformity correction was used. By defining the cloud extinction coefficient distribution function along the line-of-sight direction, the radiative variation caused by non-uniform cloud scattering was introduced into the forward model.

[0032] Step 4: Construct the joint cost function and each covariance matrix: Joint cost function J (x) is formally the sum of the nadir observation residuals, the edge observation residuals, and the prior constraint terms: J (x)=[y N -F N (x)] T S [y N -F N (x)]+[y L -F L (x)] T S [y L -F L (x)]+[xx a ] T S [xx a ]; The specific construction of each quantity is as follows: Nadir measurement error covariance matrix S N : Take it as a diagonal matrix, the diagonal elements are the noise equivalent radiance of each spectral channel. NEdL The square of the value is determined, taking into account additional errors caused by residual cloud pollution and surface non-uniformity. In this embodiment, Additional error is estimated at 2% of the observed radiance, i.e. For the first k Measured amplitude brightness of the channel.

[0033] The covariance matrix S of the edge measurement error L This includes diagonal measurement noise terms and off-diagonal forward model error terms. Diagonal elements are determined by the noise-equivalent radiance of the spectral channel corresponding to each tangent height, while off-diagonal elements are parameterized in an exponentially correlated form.

[0034] In the formula, z i , z j The height of the tangent at the adjacent edge. For the first i The standard deviation of the model error at the tangent height is taken as 3% of the simulated radiation value at that height. l c For the effective relevant length, this embodiment uses 6 km.

[0035] Highly correlated dynamic weighting factor W: To enhance the weight of the upper stratospheric ozone signal in the inversion, a diagonal weighting matrix W that varies with tangent height is introduced. z Its diagonal elements are:

[0036] Among them, the ozone peak reference height z p =35km, weight broadening Enhancement coefficient .

[0037] Prior ozone profile vector x a and its error covariance matrix S a The prior ozone profile is derived from the monthly mean climatological dataset of OMPS-LP limb observations and mapped to the 55-layer pressure grid in this embodiment via logarithmic interpolation. The prior error covariance matrix adopts a Gaussian correlation model.

[0038] in, Given the log-variance profile of climatological ozone mixing ratios, the correlation length is... L z A 4 km range was taken in the troposphere (below 200 hPa) and an 8 km range was taken in the stratosphere (above 200 hPa) to reflect the differences in the natural variability of ozone at different altitudes.

[0039] Step 5: Levenberg-Marquardt iterative solution: The Levenberg-Marquardt algorithm is used to minimize the joint cost function. J (x). The joint observation vector is denoted as The joint vector of the forward model is In the first i In the next iteration, the Jacobian matrix is ​​calculated. The nadir-to-sky Jacobian is provided by the VLIDORT analytical model, while the adjacent edge Jacobian is obtained by SCIATRAN using a semi-analytical perturbation method. Joint measurement error covariance matrix. It is a block diagonal matrix, given by S N and weighted WS L W is composed of.

[0040] The iterative update formula is: x i+1 =x i +(K S K i +S + D) -1 {K S [yF(x i )]-S [x i -x a ]}; The damping factor λ is initially set to 0.1. After each iteration, the cost function values ​​before and after the update are compared: if the cost function decreases, λ is divided by a factor of 3, and the next iteration continues; if the cost function increases, λ is multiplied by a factor of 10, and x remains constant. i Recalculate the update step without changing the value, until the cost function decreases. The diagonal scaling matrix D is taken as the identity matrix. The iteration termination condition is: the relative change in the cost function between two consecutive iterations is less than 10. -3 Or the maximum relative change of the state vector max(| x i+1,k - x i,k | / | x i,k |) Less than 10 -3 The maximum number of iterations was set to 20. The final converged result... This is the optimal estimate of the logarithmic profile of the ozone volume mixing ratio. Transforming it back to the ozone volume mixing ratio yields the full atmospheric ozone vertical profile from the troposphere to the stratosphere.

[0041] Step 6: Output of inverted diagnostic parameters: After the iteration converges, the inversion error covariance matrix is ​​calculated and output synchronously. And the average kernel matrix A:

[0042]

[0043] Where K is the Jacobian matrix at the final convergence state. The diagonal elements of the average kernel matrix reflect the degree-of-freedom components and effective resolution of each vertical layer, which can be used to evaluate the vertical sensitivity and information content of the inversion from the lower troposphere to the stratopause at various altitudes. Furthermore, through posterior estimation and error propagation of the ozone column density of each layer, the total amount of tropospheric ozone column and the total amount of stratospheric ozone column and their uncertainties can also be given.

[0044] To verify the technical feasibility of the above method, a simulation experiment was conducted. A known real atmospheric ozone profile was selected as the ground truth (e.g., a full-altitude reference profile formed by stitching together ozone radiosonde data and the OMPS-LP lip profile). These profiles were input into the nadir and lip forward model, respectively, to generate simulated OMS-N and OMS-L radiation spectra. Random errors were then superimposed according to the aforementioned noise model to obtain the simulated measurement vector y. N and y L The ozone profile was then recovered from the simulation measurements using the collaborative inversion method of this embodiment. The inversion results show that the recovered ozone profile is highly consistent with the true value across the entire altitude range, with a tropospheric deviation of less than 5% and a mid-stratospheric deviation of less than 2%. The average kernel matrix shows that the vertical resolution reaches 2–3 km in the 10–50 hPa region and 4–6 km in the free troposphere. Furthermore, the total number of degrees of freedom is higher than the combined value obtained by using only the nadir or fringe for inversion alone, fully verifying the significant advantage of the joint inversion method in achieving high-precision, high-vertical-resolution ozone inversion across the entire tropospheric-stratospheric profile.

[0045] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0046] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for the coordinated inversion of nadir and lateral ultraviolet hyperspectral ozone from Fengyun-3F satellite, characterized in that, Includes the following steps: Step S1: Obtain the observed spectral radiance vector y from the OMS-N nadir module, which is synchronously collected by the ultraviolet hyperspectral ozone detector carried by Fengyun-3F satellite in the target geographic region. N The observed spectral radiation vector y of the adjacent module OMS-L L ; Step S2: Define a state vector x that represents the vertical distribution of ozone from the Earth's surface to the top of the atmosphere. This state vector divides the atmosphere into multiple layers in the vertical direction, covering the entire altitude range from the troposphere to the stratosphere. Step S3: Using the forward model of nadir radiative transfer F N (x) and the forward model of lateral radiative transfer F L (x), respectively, simulated nadir spectral radiation and lateral edge spectral radiation are obtained based on the state vector x; Step S4: Construct a joint cost function that simultaneously integrates nadir and fringe observations. J (x), represented as: J (x)=[y N -F N (x)] T S [y N -F N (x)]+[y L -F L (x)] T S [y L -F L (x)]+[x-x a ] T S [x-x a ]; In the formula, y N Let F be the nadir observation spectral radiation vector. N (x) represents the simulated spectral radiation of the nadir forward model for the state vector x, S N Let y be the covariance matrix of the nadir measurement error. L Spectral radiation vector of the near edge observation, F L (x) represents the simulated spectral radiation of the adjacent forward model on the state vector x, S L Let x be the covariance matrix of the edge measurement error. a Let S be the prior ozone profile vector. a Prior error covariance matrix; Step S5: Using x as the parameter to be inverted, minimize the joint cost function through an iterative optimization algorithm. J (x) yields the optimal estimate of the ozone profile. This optimal estimate provides a full atmospheric ozone profile from the troposphere to the stratosphere.

2. The method for coordinated inversion of nadir and limb ultraviolet hyperspectral ozone from Fengyun-3F satellite according to claim 1, characterized in that, The state vector x mentioned in step S2 represents the distribution of ozone volume mixing ratio or number density with altitude, with the vertical stratification range from the Earth's surface to about 60 km, where the stratification thickness in the troposphere is 1 to 2 km and the stratification thickness in the stratosphere is 2 to 3 km.

3. The method for coordinated inversion of nadir and limb ultraviolet hyperspectral ozone from Fengyun-3F satellite according to claim 1, characterized in that, The nadir measurement error covariance matrix S in step S4 N The matrix is ​​taken as a diagonal matrix, and the diagonal elements are determined by the noise equivalent radiance of each ultraviolet spectral channel at the nadir; the lateral measurement error covariance matrix S L It includes measurement noise at the height of each tangent in the edge scan and error terms caused by non-uniform cloud scattering.

4. The method for coordinated inversion of nadir and limb ultraviolet hyperspectral ozone from Fengyun-3F satellite according to claim 1, characterized in that, The prior ozone profile vector x mentioned in step S4 a and its prior error covariance matrix S a Constructed from at least one of the following sources: climatological ozone database, climatological monthly mean profiles from satellite edge detection, and ozone fields from reanalysis data.

5. The method for coordinated inversion of nadir and limb ultraviolet hyperspectral ozone from Fengyun-3F satellite according to claim 1, characterized in that, The joint cost function described in step S4 J The residual term of the near-edge observation of (x) is further multiplied by a highly correlated dynamic weighting factor W( z ), to enhance the contribution of upper stratospheric ozone signal, where W( z ) is a diagonal weighted matrix that varies with the cutting height.

6. The method for coordinated inversion of nadir and limb ultraviolet hyperspectral ozone from Fengyun-3F satellite according to claim 1, characterized in that, The iterative update form of the iterative optimization algorithm described in step S5 is as follows: x i+1 =x i +(K S K i +S + D) -1 {K S [y-F(x i )]-S [x i -x a ]}; In the formula, x i For the first i The state vector of the next iteration, K i Joint observation vector y=[y ,y ] T For state vector x at x i The Jacobian matrix at the datum is given by the Jacobian K at the datum. N And adjacent Jacobi K L S is pieced together The joint error covariance matrix includes nadir error and edge error, where λ is the damping factor and D is the diagonal scaling matrix.

7. The method for coordinated inversion of nadir and limb ultraviolet hyperspectral ozone from Fengyun-3F satellite according to claim 1, characterized in that, The synchronous observation of the nadir module OMS-N and the edge module OMS-L in step S1 must meet the spatiotemporal matching conditions: the ground distance between the center of the nadir pixel and the edge scanning tangent point is less than or equal to 50 km, and the observation time difference is less than or equal to 30 seconds.

8. The method for coordinated inversion of nadir and limb ultraviolet hyperspectral ozone from Fengyun-3F satellite according to claim 1, characterized in that, In step S3, the forward model incorporates Ring effect correction and polarization factor when performing radiative transfer calculations. The Ring effect involves simulating the Ring spectrum as a pseudo-absorber.

9. The method for coordinated inversion of nadir and limb ultraviolet hyperspectral ozone from Fengyun-3F satellite according to claim 1, characterized in that, Step S5 obtains the optimal estimate of the ozone profile. Simultaneously, output the inversion error covariance matrix. The average kernel matrix A is used to characterize the vertical resolution and sensitivity of the profile inversion information.

10. The method for coordinated inversion of nadir and limb ultraviolet hyperspectral ozone from Fengyun-3F satellite according to claim 1, characterized in that, In step S3, the forward model F of nadir radiative transfer... N (x) Based on the LIDORT or VLIDORT multi-stream discrete coordinate model, the adjacent edge radiative transfer forward model F L (x) Based on the spherical atmospheric limb radiative transfer model SCIATRAN or MCC++.

Images