Remote sensing inversion method for carbon concentration of snow black

Abstract

The invention relates to the technical field of accumulated snow remote sensing modeling, in particular to an accumulated snow black carbon concentration remote sensing inversion method. Acquiring remote sensing topographic data of the target area and performing topographic correction; the reflectivity of a snow area in the remote sensing topographic data after topographic correction is obtained; establishing a pure snow optical model by adopting an analytic asymptotic radiation transfer theory according to the reflectivity of the accumulated snow area; introducing an ice-black carbon composite particle absorption characteristic into the pure snow optical model, constructing a dirty snow optical model, and performing optimization solution on the dirty snow optical model by adopting a strict radiation transmission model fitting result; and according to the optimized dirty snow optical model, carrying out inversion on the black carbon concentration of the accumulated snow area. According to the snow remote sensing inversion method provided by the invention, the influence of terrain shadow can be effectively reduced, the black carbon absorption coefficient is modeled into a wavelength function based on the snow radiation transmission model, the calculation complexity is greatly reduced, and the calculation efficiency and precision are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of snow remote sensing modeling, and in particular to a remote sensing inversion method for snow black carbon concentration. Background Art

[0002] As a key component of the Earth system, snow has a significant high albedo characteristic and can reflect 80% to 90% of the solar radiation back to the atmosphere, playing an irreplaceable role in regulating the heat exchange of the Earth-Atmosphere system and maintaining the global energy balance. However, the black carbon aerosols that are ubiquitous in the atmosphere are transported across borders through atmospheric circulation, and are deposited on the surface of the snow or enter the snow layer through dry / wet deposition, significantly changing the optical properties of the snow; at the microscopic scale, due to the increased probability of interaction with black carbon particles, the scattering behavior of photons in the snow is more frequent than in the air, resulting in the darkening of snow. Factors such as black carbon content, source composition, particle size distribution characteristics, and ice-black carbon particle mixing mode can affect the reflective properties of snow by changing the penetration depth of solar radiation in the snow layer. This microscopic change ultimately leads to a significant change in the optical properties of snow at the macroscopic scale, which in turn affects the climate system at the regional and even global scales.

[0003] Studies have shown that the presence of black carbon in the cryosphere not only enhances solar radiation absorption, accelerates the snowmelt process and changes the albedo of the snow layer, but also further produces positive radiative forcing, triggering a series of complex climate feedback mechanisms. When black carbon is deposited on the surface of snow and glaciers, it will significantly reduce the snow / ice albedo, increase the absorption of solar radiation energy, and induce positive radiative forcing, thereby further triggering the snow / ice albedo feedback, causing more solar radiation energy to be absorbed by the snow / ice layer, accelerating the melting process of the snow / ice layer; more importantly, the climate change caused by the radiative forcing induced by black carbon will in turn affect the transmission path of black carbon and the dry and wet deposition rates in the atmosphere, and further affect the black carbon deposition concentration in snow / ice, forming a complex circular feedback mechanism. When acting in synergy with dust, the snow albedo can drop by as much as 0.423. This significant albedo change directly affects the regional energy balance.

[0004] The presence of black carbon in snow not only exacerbates snowmelt, but also significantly shortens the duration of snow accumulation, increases peak flood flow, changes regional water cycle and runoff characteristics, and may induce a series of secondary disasters, posing a potential threat to regional ecological security and socioeconomic development. Specifically, the impact of black carbon on snow is mainly reflected in three aspects: albedo change, snowmelt process, and radiative forcing. Studies have shown that the snowmelt effect of black carbon in some areas even exceeds the impact of greenhouse gases. When these light-absorbing particles (LAPs) are deposited in the cryosphere, they can significantly affect the process of global climate change. On a time scale, relevant studies have shown that black carbon significantly accelerates snowmelt, both at the basin scale and the global scale, and shortens the duration of snow accumulation on a time scale of days to weeks.

[0005] Considering that the refractive index, morphology, particle size and mixing state of black carbon with other particles will change with the emission source and existence time, it is difficult to compare black carbon samples in different time and space. Therefore, the black carbon concentration of snow has become a universal indicator for quantifying the black carbon-induced albedo effect. At present, the methods for obtaining the black carbon concentration of snow mainly include field surveys, numerical simulations and remote sensing inversion. Among them, although traditional field surveys can accurately obtain the black carbon content and optical properties of snow at a point scale, and combine various aerosol traceability models to clarify the source, migration, deposition and aging of black carbon and provide important data support for numerical simulation and remote sensing inversion, continuous field surveys require a lot of resource investment and the sampling scale is limited (usually 10 1 ~10 2 In contrast, the numerical simulation of climate models can more systematically reveal the transport, evolution and removal mechanisms of black carbon aerosols. Its advantages in large-scale research are beyond the reach of traditional single-point observations. However, the resolution of driving data of climate models is relatively coarse, and its grid simulation results in degrees are difficult to reflect the detailed changes in snow black carbon concentration under complex terrain. Given the significant impact of black carbon on snow and ice albedo and radiation intensity, the snow and ice albedo feedback mechanism in climate change research urgently needs a set of high-precision snow black carbon remote sensing inversion algorithms. At present, most studies are still in the single-point sampling stage, lacking a mature snow black carbon concentration remote sensing inversion algorithm system. The key technical difficulties in the current snow black carbon remote sensing inversion are: first, the high complexity of the snow medium and the multiple scattering characteristics of black carbon particles make it difficult to accurately simulate the radiation transmission process; second, the mixing modes of ice-black carbon composite particles are diverse, and the optical properties under different mixing states are significantly different, which brings challenges to the accurate calculation of extinction characteristics; third, factors such as terrain effects and underlying surface heterogeneity in actual observations increase the uncertainty of the inversion process. Summary of the invention

[0006] In order to solve the problems existing in the prior art, the present invention provides a remote sensing inversion method for snow black carbon concentration. It includes: obtaining remote sensing terrain data of the target area and performing terrain correction; obtaining the reflectivity of the snow area in the remote sensing terrain data after terrain correction; establishing a pure snow optical model based on the reflectivity of the snow area using analytical asymptotic radiation transfer theory; introducing the absorption characteristics of ice-black carbon composite particles into the pure snow optical model, constructing a dirty snow optical model, and optimizing and solving the dirty snow optical model using the fitting results of a rigorous radiation transfer model; inverting the black carbon concentration in the snow area based on the optimized optical model. The snow remote sensing inversion method proposed in the present invention can effectively reduce the influence of terrain shadows, and model the black carbon absorption coefficient as a function of wavelength based on the snow radiation transmission model, which greatly reduces the computational complexity and improves the computational efficiency and accuracy.

[0007] The present invention adopts the following technical scheme, a remote sensing inversion method for snow black carbon concentration, comprising:

[0008] Acquire remote sensing terrain data of the target area and perform terrain correction;

[0009] Obtain the reflectivity of snow-covered areas in remote sensing terrain data after terrain correction;

[0010] A pure snow optical model is established using an analytical asymptotic radiation transfer model according to the reflectivity of the snow area;

[0011] The absorption characteristics of ice-black carbon composite particles are introduced into the pure snow optical model to construct a dirty snow optical model, and the dirty snow optical model is optimized and solved using the fitting results of a rigorous radiation transfer model.

[0012] The black carbon concentration in the snowy area is inverted based on the optimized optical model.

[0013] Furthermore, remote sensing terrain data of the target area is obtained and terrain correction is performed, including:

[0014] The improved SCS+C correction model is used to perform terrain correction on the remote sensing terrain data. The corrected reflectivity is expressed as:

[0015]

[0016] c snow,i =b snow,i / a snow,i

[0017] Among them, ρ COR is the surface reflectance after SCS+C terrain correction, ρ L2 is the L2-level satellite surface reflectivity, cosi is the cosine of the local solar incidence angle, θ s is the solar zenith angle, S is the surface slope, is the solar azimuth, is the terrain slope; c snow,i is the empirical parameter of the linear regression equation between the pixel reflectance of the ith band in the snow area and the cosine of the solar zenith angle, a snow,i and b snow,i are the intercept and slope of the regression equation of the ith band in the snow area, respectively.

[0018] Furthermore, the reflectivity of the snow area in the remote sensing terrain data after terrain correction is obtained, including:

[0019] Obtaining a normalized snow cover index of each pixel in the remote sensing terrain data, and determining a snow cover area in the remote sensing terrain data according to the normalized snow cover index of each pixel;

[0020] When the normalized snow accumulation index is greater than a set threshold, the pixel is determined to be a snow accumulation pixel, and all pixels are determined in turn to obtain the snow accumulation area in the remote sensing terrain data.

[0021] Furthermore, the normalized snow cover index of each pixel in the remote sensing terrain data is obtained, which is expressed as:

[0022] NDSI=(ρ COR,GREEN -ρ COR,NIR ) / (ρ COR,GREEN +ρ COR,NIR )

[0023] Among them, ρ COR,GREEN and ρ COR,NIR They are the surface reflectances of the green band and near-infrared band in the remote sensing terrain data after terrain correction.

[0024] Furthermore, an optical model of the snow-covered area is established by using an analytical asymptotic radiation transfer model according to the reflectivity of the snow-covered area, including:

[0025] According to the analytical asymptotic radiation transfer model, the relationship between the snow reflectivity, spherical albedo and Lambertian reflectivity in the snow area is established, which is expressed as:

[0026]

[0027] Where R is the reflectivity of snow, r s is the spherical albedo, R0 is the Lambertian reflectivity, μ and μ0 are the cosines of the solar zenith angle and the observation zenith angle, respectively; u(μ0) and u(μ) are escape functions, which characterize the angular distribution of photons escaping when the light source is located at an infinite depth inside the medium, and together they constitute the angular function f.

[0028] Furthermore, the optical model also includes:

[0029] The optical model is simplified according to the spherical albedo in the optical model:

[0030] The spherical albedo is obtained as follows:

[0031] r s =e -y ,

[0032] Among them, ω0 is the single scattering albedo, g is the asymmetry factor, β is the probability of photon absorption, and y is an intermediate variable. When the snow in the snow-covered area does not contain black carbon, the calculation method is:

[0033] k abs =Bαc,

[0034] Among them, k abs is the snow mass absorption coefficient, k ext is the extinction coefficient, α is the absorption coefficient of ice, λ is the wavelength, c is the snow density, χ is the imaginary part of the complex refraction coefficient of ice particles at wavelength λ, B is the shape factor, which is used to simulate the morphology of snow particles. The effective absorption length l is expressed as:

[0035] l=ξd,

[0036] Where ξ is the morphology-dependent parameter, d is the snow particle size, l is the effective absorption length;

[0037] In summary, the simplified optical model is:

[0038]

[0039] Where λ1 and λ2 are the wavelengths of 865 nm and 1020 nm, respectively, R(λ2) is the satellite reflectivity at a wavelength of 1020 nm, u(μ0) and u(μ) are the escape functions, and R(λ i ) is the reflectivity of snow at wavelength i, α(λ i ) is the absorption coefficient of ice, and R0 is the Lambertian reflectivity.

[0040] Furthermore, ice-black carbon composite particles are introduced into the optical model, specifically:

[0041] When the particle size distribution follows Jungian power law distribution, The pure black carbon absorption coefficient is simplified by the exponent, which is expressed as:

[0042]

[0043] k0≡k bc (λ0), λ0=1μm

[0044]

[0045] Where m represents BC absorption The coefficient, μ, represents the cosine of the solar zenith angle, represents the absorption coefficient of snow containing black carbon, c ice represents the volume concentration of ice particles, c bc represents the volume concentration of black carbon, k bc represents the volume absorption coefficient of pure black carbon, k ice is the volume absorption coefficient of pure ice, α ice represents the absorption coefficient of pure ice, k0 represents the volume absorption coefficient of pure black carbon at 1 μm, λ is the wavelength, λ0 is 1 μm, and F bc is the intermediate variable, n pol =1.75, χ pol =0.47;

[0046] Based on the analytical asymptotic radiative transfer model, the probability of a photon being hit by an ice-black carbon composite particle is expressed as:

[0047]

[0048] in, Characterizes the probability of photons being absorbed by ice particles, β pol Characterizes the probability of photons being absorbed by black carbon.

[0049] The reflectivity and spherical albedo of snow containing black carbon are expressed as:

[0050]

[0051] Where R is the reflectivity of snow, r s is the albedo of the snow sphere, R0 is the Lambertian reflectivity of the snow, x is the angle function, α is the pure ice absorption coefficient, l is the effective absorption length, f and is the intermediate variable,

[0052] Furthermore, the dirty snow optical model is optimized and solved by using the rigorous radiation transfer model fitting results, and the expression is:

[0053]

[0054] A=1.247,B=1.186,C=5.157

[0055] p(θ)=11.1exp(-0.087θ)+1.1exp(-0.014θ)

[0056] cosθ=-cosθ0cosθv +sinθ0sinθ v cosφ r

[0057] Where R0 is the Lambertian reflectance, μ and μ0 are the cosines of the solar zenith angle and the observation zenith angle, respectively, p(θ) is the phase function, cosθ is the intermediate variable, AC is a constant, θ0 is the solar zenith angle, θ v represents the observation zenith angle, φ r Relative azimuth

[0058] The beneficial effects of the present invention are as follows: the present invention proposes a terrain correction algorithm specifically for snow remote sensing inversion, which can effectively reduce the influence of terrain shadows; in addition, by introducing snow area ratio data, the secondary error generated in the correction process due to ignoring the high reflective characteristics of snow is improved. These improvements significantly reduce the error of snow black carbon concentration inversion, thereby improving the accuracy of inversion.

[0059] Compared with conventional field surveys, the cost of the present invention is greatly reduced, and at the same time, it can provide large-scale snow black carbon concentration distribution that can be continuously monitored even in periods without field surveys. Compared with climate model simulation results, the present invention does not require a large amount of driving data, has a fast calculation speed, and a higher spatial resolution, and can obtain a more detailed snow black carbon spatial distribution map. In addition, the remote sensing retrieval results can more accurately reflect the actual concentration of black carbon deposited in the snow, rather than the concentration of black carbon aerosols suspended in the atmosphere. Therefore, the present invention shows obvious advantages in comparison with traditional climate models.

[0060] Based on the snow radiation transfer model, the present invention models the black carbon absorption coefficient as a function of wavelength, which greatly reduces the calculation complexity and improves the calculation efficiency and accuracy. The present invention also takes into account the impact of poor satellite imaging quality on the inversion accuracy, and uses a rigorous radiation transfer model fitting result to optimize the snow Lambertian reflectivity, thereby improving the inversion accuracy. Compared with the traditional empirical model, this algorithm has a solid physical foundation and stronger generalization ability, and therefore has more advantages in application. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0062] Figure 1 A schematic diagram of a process flow of a remote sensing inversion method for snow black carbon concentration according to an embodiment of the present invention; Figure 2A schematic diagram of snow black carbon remote sensing inversion accuracy verification result according to an embodiment of the present invention. DETAILED DESCRIPTION

[0065] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0066] Black carbon in snow can cause a strong extinction effect, and the extinction intensity at different wavelengths varies greatly. Clarifying the difference in extinction effect caused by black carbon on snow reflectivity at different wavelengths is the basic condition for remote sensing inversion of snow black carbon. Only by clarifying the reflective characteristics of snow and the extinction effect caused by black carbon can we select the appropriate band reflectivity to carry out snow black carbon inversion. On this basis, through the assumptions about the optical thickness, heterogeneity, close stacking and mixing state of snow, the extinction term of black carbon is introduced into the scattering characteristics of ice particles, and finally the optical properties of snow containing black carbon can be simulated. The inverse solution in the above process is the basic principle of remote sensing inversion of snow black carbon. Therefore, the present invention provides a remote sensing inversion method for snow black carbon concentration. The method takes the analytical asymptotic radiation transfer theory as the core, introduces the black carbon extinction coefficient into the calculation of the extinction characteristics of ice-black carbon composite particles, and uses the extinction bands of ice particles and black carbon to solve the snow black carbon concentration. The present invention can effectively improve the quantitative application level of snow remote sensing, quantify the spatial distribution pattern of snow black carbon, and has practical significance for accurately grasping the law of snow formation and disappearance, and protecting regional agricultural and animal husbandry development and ecological environment. Specifically, it includes:

[0067] Acquire remote sensing terrain data of the target area and perform terrain correction; obtain the reflectivity of the snow-covered area in the remote sensing terrain data after terrain correction; establish an optical model of pure snow based on the reflectivity of the snow-covered area using an analytical asymptotic radiation transfer model; introduce the absorption characteristics of ice-black carbon composite particles into the pure snow optical model to construct a dirty snow optical model, and use the fitting results of a rigorous radiation transfer model to optimize and solve the dirty snow optical model; and invert the black carbon concentration in the snow-covered area based on the optimized optical model.

[0068] In a specific embodiment of the present invention, a schematic flow chart of a remote sensing inversion method for snow black carbon concentration in an embodiment of the present invention is as follows: Figure 1 As shown, including:

[0069] Acquire remote sensing terrain data of the target area and perform terrain correction;

[0070] The present invention combines digital terrain elevation and snow area ratio data, and uses the improved SCS+C correction model to perform terrain correction on L2-level satellite remote sensing data based on MOD09GA. The relationship between the corrected reflectivity and the original reflectivity is expressed as follows:

[0071]

[0072] c snow,i =b snow,i / a snow,i

[0073] In the formula, ρ COR is the surface reflectance after SCS+C terrain correction, ρ L2 is the L2-level satellite surface reflectivity, cosi is the cosine of the local solar incidence angle, θ s is the solar zenith angle, S is the surface slope, is the solar azimuth,

[0074] is the terrain slope, is the solar azimuth. Since snow has extremely high reflectivity, directly performing full-band correction on the ground object will result in a significant overestimation of the reflectivity of the snow-covered area. The present invention aims to remotely invert the black carbon concentration in snow. Therefore, it is very important to accurately obtain the true surface reflectivity of the snow-covered area. To this end, the present invention introduces the snow area ratio data MOD10A1 to identify snow-covered ground objects. In the formula, c snow,i It is the empirical parameter of the linear regression equation between the pixel reflectance of the ith band in the snow area and the cosine of the solar zenith angle, a snow,i and b snow,i are the intercept and slope of the regression equation of the ith band in the snow-covered area. Considering that the daily-scale growth and decline of snow has a significant impact on the reflectivity, in order to ensure the stability of the correction result, the present invention further adopts the ridge regression algorithm and enhances the robustness of the regression by introducing a regularization term.

[0075] Obtain the reflectivity of snow-covered areas in remote sensing terrain data after terrain correction.

[0076] The present invention further uses the Normalized Difference Snow Index (NDSI) of the MOD10A1 product to determine the snow area in the remote sensing terrain data. The NDSI calculation method is as follows:

[0077] NDSI=(ρ COR,GREEN -ρ COR,NIR ) / (ρ COR,GREEN +ρ COR,NIR )

[0078] Among them, ρ COR,GREENand ρ COR,NIR They are the surface reflectance of the MOD10A1 green band and near infrared band after terrain correction. When NDSI>0.4, the present invention considers that the pixel can be determined as a snow pixel and can be used for snow black carbon remote sensing inversion.

[0079] According to the reflectivity of the snow-covered area, a pure snow optical model of the snow-covered area is established by using an analytical asymptotic radiation transfer model;

[0080] The present invention models the optical properties of pure snow based on the Analytical Asymptotic radiative transfer theory (ART). The basic assumptions of ART include the following points: the optical thickness of snow is so great that the influence of the underlying surface can be ignored. For the visible light band, the snow thickness is required to be greater than 10 to 30 cm to meet this assumption; for the near-infrared band, the penetration depth is less than 2 to 5 cm; the heterogeneity of the snow layer in the horizontal and vertical directions is ignored, and it is assumed that it is a plane-parallel medium, and the influence of dense stacking on the scattering properties of snow is ignored, and all inversion parameters are snow surface parameters; it is assumed that the snow particle size is much larger than the wave of the incident light in the visible light and near-infrared range, and when the snow contains black carbon, it is assumed that the black carbon is mixed with ice particles.

[0081] The ART model gives the relationship between snow reflectivity, spherical albedo and Lambertian reflectivity:

[0082]

[0083] Where R is the snow reflectivity, r s is the spherical albedo, R0 is the Lambertian reflectivity, μ and μ0 are the cosines of the solar zenith angle and the observation zenith angle, respectively; u(μ0) and u(μ) are escape functions, which characterize the angular distribution of photons escaping when the light source is located at an infinite depth inside the medium, and together they constitute the angular function f.

[0084] Spherical albedo r s The calculation involves the single scattering albedo ω0 and the asymmetry factor g, which are expressed as follows:

[0085] r s =e -y ,

[0086] Among them, β is also called the probability of photon absorption, and the sum of it and the single scattering albedo ω0 is 1. y is an intermediate parameter, which is affected by the optical properties of ice particles and black carbon. When there is no black carbon in the snow, β is the ice particle absorption coefficient κ abs and the extinction coefficient κ ext Ratio of:

[0087]

[0088] α is the absorption coefficient of ice, λ is the wavelength, and c is the snow density (parameterized as 1 / 3 g / cm 3 ), χ is the imaginary part of the complex refractive index of ice particles at wavelength λ. Therefore, when the wavelength is determined, the absorption coefficient α of ice can be considered as a known value. In traditional snow radiation transmission models, snow is usually regarded as a medium layer composed of large spherical particles, and MIE scattering is mostly used to calculate its extinction efficiency, scattering efficiency and absorption efficiency. Snow particles in nature are affected by many factors such as temperature, humidity and pressure. Their geometric shapes are complex and changeable, which is far from the assumption of spherical particles. In ART, the value of the shape factor B is controlled to achieve the simulation of the irregular shape of snow particles.

[0089] The present invention introduces the effective absorption length l, which can be expressed as the product of the morphology-dependent parameter ξ and the snow particle size d:

[0090] l=ξd,

[0091] Combining the above formulas, we can get the spherical albedo r s The simplified result is:

[0092]

[0093] Among them, α is the absorption coefficient of ice, l is the effective absorption length, μ and μ0 are the cosine of the solar zenith angle and the observation zenith angle respectively; u(μ0) and u(μ) are escape functions; it can be found that the above formula contains two unknown quantities: the pure Schramberg reflectivity R0 and the absorption length l, which can be solved by using the reflectivity of two bands. These two bands need to meet two points: sensitive to inversion parameters; and rarely affected by atmospheric scattering and absorption.

[0094]

[0095] Where λ is the wavelength that affects the snow reflectance characteristics, which should usually be within the visible light range. Due to satellite band limitations, 865nm and 1020nm can usually be used, that is, λ1 and λ2 are wavelengths of 865nm and 1020nm respectively, R(λ2) is the satellite reflectivity at the corresponding wavelength of 1020nm, u(μ0) and u(μ) are escape functions, R is the snow reflectivity, and R0 is the Lambertian reflectivity.

[0096] By obtaining the pure Scheibert reflectivity R0 and absorption length l, basic modeling of snow properties such as snow reflectivity, spherical albedo and snow particle size can be achieved.

[0097] introducing ice-black carbon composite particles into the optical model;

[0098] Assuming that the black carbon-induced extinction occurs mainly in the visible range, while the counterpart of ice particles occurs mainly in the near-infrared range, when the particle size distribution follows the Jungian power law distribution, the application The exponent to simplify the absorption coefficient of pure black carbon has been shown to be effective:

[0099]

[0100] k0≡k bc (λ0), λ0=1μm

[0101]

[0102] Where m represents BC absorption The coefficient, μ, represents the cosine of the solar zenith angle, represents the absorption coefficient of snow containing black carbon, c ice represents the volume concentration of ice particles, c bc represents the volume concentration of black carbon, k bc represents the volume absorption coefficient of pure black carbon, k ice is the volume absorption coefficient of pure ice, α ice represents the absorption coefficient of pure ice, k0 represents the volume absorption coefficient of pure black carbon at 1 μm, λ is the wavelength, λ0 is 1 μm, and F bc is the intermediate variable, n pol =1.75, χ pol =0.47;

[0103] Combining the basic theory of ART, the probability β of a photon being an ice-black carbon composite particle is expressed as:

[0104]

[0105] The albedo of the snowball surface is thus simplified to:

[0106] r s =e -y ,

[0107] Given parameter z=y 2 :

[0108]

[0109] Where f represents the angle function, represents the volume normalized absorption coefficient;

[0110] The effective absorption length l is:

[0111] l=ξd,

[0112] The asymmetry factor g of ice crystal clouds in the visible light ranges from 0.74 to 0.76. In view of the lack of measured data on the asymmetry factor of snow accumulation in previous studies, the median data of ice crystal clouds g = 0.75 is used in the algorithm of the present invention. The average value of the absorption enhancement coefficient B in existing studies is 1.6, and ξ≈11.38 is obtained by substitution.

[0113] Simplified snow reflectivity R and spherical albedo r s It is expressed as:

[0114]

[0115] Where R is the reflectivity of snow, r s is the albedo of the snow sphere, R0 is the Lambertian reflectivity of the snow, x is the angle function, α is the pure ice absorption coefficient, and m represents the BC absorption coefficient, l is the effective suction length, f and is the intermediate variable,

[0116] The above formula contains four unknown parameters: the effective absorption length l, the Lambertian reflectivity R0 of the semi-infinite non-absorbing snow layer, the parameter f and the absorption Coefficient m; These four parameters can be solved using any four bands of visible light and near infrared. The present invention takes the snow reflectivity R as an example:

[0117]

[0118] In the above formula, numbers 1 to 4 are spectral channels. The selection principle is that the first two channels (λ1, λ2) can ignore the absorption effect of ice, and the last two channels (λ3, λ4) can ignore the absorption effect of pollutants. This assumption is applicable to snow with less serious pollution.

[0119] The optical model is optimized and solved by using the fitting results of a rigorous radiation transfer model;

[0120] When the satellite imaging quality is poor, the snow Lambertian reflectivity R0 will be distorted, which will directly lead to a serious decrease in the inversion accuracy of the above formula. Therefore, the present invention uses the results of rigorous radiation transfer model fitting to simplify the expression of snow Lambertian reflectivity R0:

[0121]

[0122] A=1.247,B=1.186,C=5.157

[0123] p(θ)=11.1exp(-0.087θ)+1.1exp(-0.014θ)

[0124] cosθ=-cosθ0cosθv +sinθ0sinθ v cosφ r

[0125] Where R0 is the Lambertian reflectance, μ and μ0 are the cosines of the solar zenith angle and the observation zenith angle, respectively, p(θ) is the phase function, cosθ is the intermediate variable, θ0 is the solar zenith angle, θ v represents the observation zenith angle, φ r represents the relative azimuth. Thus, only three unknown quantities remain, namely, the effective absorption length l, the absorption Coefficient m and spectral absorption coefficient k of black carbon at 1 μm abs (λ), the present invention uses two wavelength bands in the visible light range to calculate the effective absorption length l and the absorption Coefficient m.

[0126]

[0127] The concentration of black carbon in snow can be expressed as:

[0128]

[0129] K(λ)=Fα pol (λ),

[0130]

[0131] Among them, B is the absorption enhancement parameter, R0 is the Lambertian reflectivity of snow, and R1 can be taken as the reflectivity of snow at 400nm. is the exponential function of wavelength, f is the angle function, l is the absorption length, for black carbon, F is 0.95, α pol (λ0) is the absorption coefficient of black carbon; for black carbon, the real part of the complex refractive index n pol and the imaginary part χ pol The change with wavelength is very small and is usually regarded as a constant, n pol =1.75, χ pol =0.47.

[0132] In an experimental embodiment of the present invention:

[0133] Based on multiple groups of snow black carbon and organic carbon samples collected in northern Xinjiang between 2016 and 2019, the present invention conducted a comprehensive investigation of the changes in snow black carbon concentration and its radiative forcing effects. These data constitute the key support for the present invention to verify the accuracy of different snow black carbon remote sensing inversion algorithms. Sample collection covers different regions of northern Xinjiang, including meadows, urban areas, highways and industrial areas, and other land cover types. The time span includes both instantaneous sampling and long-term sequence observations in meteorological fields. Usually, contaminated snow samples are sealed in Whirl-Pak bags (capacity of about 2 liters, unmelted) and frozen, and then transported to the laboratory for filtration and analysis. More than 300 snow samples were collected, melted, and filtered, and the black carbon content in each sample was determined using the DRI2015A method.

[0134] Based on the measured data in northern Xinjiang (a total of 127 groups), the inversion accuracy of the snow black carbon inversion algorithm was verified using three evaluation indicators: Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Coefficient of Determination (R²). Figure 2 As shown in the figure, at the spatial resolution of MODIS satellite data (500 meters), the MAE, RMSE and R² values ​​obtained by the inversion algorithm proposed in the present invention are 0.55 μg / g, 0.9 μg / g and 0.88, respectively, showing a high inversion accuracy. The above results indicate that the remote sensing estimation method for snow black carbon concentration based on the radiation transfer model proposed in the present invention has good applicability in the northern Xinjiang region.

[0135] At the same time, the snow black carbon inversion algorithm proposed in the invention is highly flexible and universal. It does not rely on a specific satellite platform or computing environment. The algorithm can be deployed in Python, Google Earth Engine (GEE) or JAVA. The algorithm can adapt to satellite data of different scales and provide key data support for local and even global climate change research, water resources management and ecological protection.

[0136] The present invention does not rely on a specific computing environment, and can be seamlessly connected to cloud computing platforms such as Google Earth Engine, Microsoft Planetary Compute, and PIE-Engine. It is also applicable to local development environments such as Visual Studio Code, Pycharm, and R-Studio. In addition, the present invention is not limited to a specific programming language, and supports multiple languages ​​such as JavaScript, Python, Matlab, R, and IDL, which further enhances the applicability and flexibility of the algorithm. Specifically, since it is not limited to a specific satellite, the present invention can be deployed on a variety of platforms such as ground-based, aerospace, and satellite-based. At the ground-based level, the present invention can use the spectral / image data obtained by a spectrometer or an imaging spectrometer for inversion, and realize accurate inversion at a single-point scale; at the aerospace level, the present invention can be integrated into the airborne platform of an unmanned aerial vehicle or a manned aircraft to obtain a regional-scale spatial distribution map of snow black carbon based on aerial photography data, and due to the algorithm Low computing power requirements, some airborne platforms can even achieve real-time computing; at the satellite level, the algorithm proposed in the present invention can be deployed on hyperspectral or multispectral satellite platforms, including but not limited to hyperspectral platforms such as Resource-1 02D (30-meter resolution), Hyperion (30-meter resolution), EnMAP, PRISMA (30-meter resolution) and CHIME-X (30-meter resolution); and multispectral platforms such as Sentinel-2 (10-meter resolution), Landsat series (30-meter resolution), Sentinel-3 (300-meter resolution) and MODIS (500-meter resolution).

[0137] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A remote sensing inversion method for snow black carbon concentration, characterized in that: include: Acquire remote sensing terrain data of the target area and perform terrain correction; Obtain the reflectivity of snow-covered areas in remote sensing terrain data after terrain correction; A pure snow optical model is established using an analytical asymptotic radiation transfer model according to the reflectivity of the snow area; The pure snow optical model is introduced with ice-black carbon composite particles to construct a dirty snow optical model, and the dirty snow optical model is optimized and solved using the fitting results of a rigorous radiation transfer model; The black carbon concentration in the snowy area is inverted based on the optimized optical model.

2. The remote sensing inversion method for snow black carbon concentration according to claim 1, characterized in that: Acquire remote sensing terrain data of the target area and perform terrain correction, including: The improved SCS+C correction model is used to perform terrain correction on the remote sensing terrain data. The corrected reflectivity is expressed as: c snow,i =b snow,i / a snow,i Among them, ρ COR is the surface reflectance after SCS+C terrain correction, ρ L2 is the L2-level satellite surface reflectivity, cosi is the cosine of the local solar incidence angle, θ s is the solar zenith angle, S is the surface slope, is the solar azimuth, is the terrain slope; c snow,i is the empirical parameter of the linear regression equation between the pixel reflectance of the ith band in the snow area and the cosine of the solar zenith angle, a snow,i and b snow,i are the intercept and slope of the regression equation of the ith band in the snow area, respectively.

3. The remote sensing inversion method for snow black carbon concentration according to claim 1, characterized in that: Obtain the reflectivity of snow-covered areas in remote sensing terrain data after terrain correction, including: Obtaining a normalized snow cover index of each pixel in the remote sensing terrain data, and determining a snow cover area in the remote sensing terrain data according to the normalized snow cover index of each pixel; When the normalized snow accumulation index is greater than a set threshold, the pixel is determined to be a snow accumulation pixel, and all pixels are determined in turn to obtain the snow accumulation area in the remote sensing terrain data.

4. The remote sensing inversion method for snow black carbon concentration according to claim 1, characterized in that: Get the normalized snow cover index of each pixel in the remote sensing terrain data, expressed as: NDSI=(ρ COR,GREEN -r COR,NIR ) / (ρ COR,GREEN +r COR,NIR ) Among them, ρ COR,GREEN and ρ COR,NIR They are the surface reflectances of the green band and near-infrared band in the remote sensing terrain data after terrain correction.

5. The remote sensing inversion method for snow black carbon concentration according to claim 1, characterized in that: According to the reflectivity of the snow area, a pure snow optical model is established using an analytical asymptotic radiation transfer model, including: According to the analytical asymptotic radiation transfer model, the relationship between the snow reflectivity, spherical albedo and Lambertian reflectivity in the snow area is established, which is expressed as: Where R is the reflectivity of snow, r s is the spherical albedo, R0 is the Lambertian reflectivity, μ and μ0 are the cosines of the solar zenith angle and the observation zenith angle, respectively; u(μ0) and u(μ) are escape functions, which characterize the angular distribution of photons escaping when the light source is located at an infinite depth inside the medium, and together they constitute the angular function f.

6. The remote sensing inversion method for snow black carbon concentration according to claim 5, characterized in that: The optical model also includes: The optical model is simplified according to the spherical albedo in the optical model: The spherical albedo is obtained as follows: Among them, ω0 is the single scattering albedo, g is the asymmetry factor, β is the probability of photon absorption, and y is an intermediate variable. When the snow in the snow-covered area does not contain black carbon, the calculation method is: Among them, k abs is the snow mass absorption coefficient, k ext is the extinction coefficient, α is the absorption coefficient of ice, λ is the wavelength, c is the snow density, χ is the imaginary part of the complex refraction coefficient of ice particles at wavelength λ, and B is the shape factor, which is used to simulate the morphology of snow particles; the effective absorption length l is expressed as: Among them, ξ is the morphology-dependent parameter, d is the snow particle size; In summary, the simplified pure Scheibert reflectivity R0 and effective absorption length l are: Where λ1 and λ2 are wavelengths of 865 nm and 1020 nm, respectively, R(λ2) is the satellite reflectivity at 1020 nm, u(μ0) and u(μ) are escape functions, and R(λ i ) is the snow reflectivity at wavelength i, α(λ i ) is the absorption coefficient of ice at wavelength i, and R0 is the Lambertian reflectivity.

7. The remote sensing inversion method for snow black carbon concentration according to claim 1, characterized in that: The pure snow optical model introduces ice-black carbon composite particles, specifically: When the particle size distribution follows Jungian power law distribution, The pure black carbon absorption coefficient is simplified by the exponent, which is expressed as: Where m represents BC absorption The coefficient, μ, represents the cosine of the solar zenith angle, represents the absorption coefficient of snow containing black carbon, c ice represents the volume concentration of ice particles, c bc represents the volume concentration of black carbon, k bc represents the volume absorption coefficient of pure black carbon, k ice is the volume absorption coefficient of pure ice, α ice represents the absorption coefficient of pure ice, k0 represents the volume absorption coefficient of pure black carbon at 1 μm, λ is the wavelength, λ0 is 1 μm, and F bc is the intermediate variable, n pol =1.75, χ pol =0.47; Based on the analytical asymptotic radiative transfer model, the probability of a photon being absorbed by an ice-black carbon composite particle is expressed as: in, Characterizes the probability of photons being absorbed by ice particles, β pol Characterizes the probability of photons being absorbed by black carbon. The reflectivity and spherical albedo of snow containing black carbon are expressed as: Where R is the reflectivity of snow, r s is the albedo of the snow sphere, R0 is the Lambertian reflectivity of the snow, x is the angle function, α is the pure ice absorption coefficient, l is the effective absorption length, f and is an intermediate variable.

8. The remote sensing inversion method for snow black carbon concentration according to claim 1, characterized in that: The dirty snow optical model is optimized and solved using the rigorous radiation transfer model fitting results. The expression is: A=1.247,B=1.186,C=5.157 p(θ)=11.1exp(-0.087θ)+1.1exp(-0.014θ) cosθ=-cosθ0cosθ v +sinθ0sinθ v cosφ r Where R0 is the Lambertian reflectance, μ and μ0 are the cosines of the solar zenith angle and the observation zenith angle, respectively, p(θ) is the phase function, cosθ is the intermediate variable, θ0 is the solar zenith angle, θ v represents the observation zenith angle, φ r Indicates relative azimuth.

Images