Three-dimensional aquatic vegetation canopy reflection characteristic modeling method based on real scene

Abstract

The invention relates to a three-dimensional aquatic vegetation canopy reflection characteristic modeling method based on a real scene. The method comprises the following steps that leaf reflection and transmission spectrums are obtained through calculation of a plant leaf radiation transmission model; in combination with an actual aquatic vegetation scene, vegetation canopy reflection characteristics are described through canopy structure parameters and leaf BRDF; calibrating model parameters by adopting four-flow approximation and linear approximation, and applying the model parameters to approximate solution of vegetation canopy reflectivity; finally, a LuxCoreRender physical rendering engine is used for coupling input parameters such as incident radiation, vegetation canopy reflection factors, water optical attribute parameters and observation geometry, and the vegetation canopy reflectivity is calculated in combination with an aquatic vegetation scene. According to the method, a high-precision three-dimensional aquatic vegetation canopy reflection model is established, remote sensing radiation modeling precision and application value can be improved, an important tool platform is provided for wetland remote sensing monitoring and ecological protection and coastal wetland blue carbon sink function evaluation, and the method has popularization and application prospects.

Description

(I) Technical field

[0001] The invention relates to a three-dimensional aquatic vegetation canopy reflectance characteristic modeling method based on real scenes, belongs to the field of optical remote sensing, and has important significance in wetland ecological research and quantitative crop monitoring applications. (II) Background technology

[0002] The three-dimensional vegetation radiation transfer model can more accurately simulate the propagation and reflection process of light in the vegetation canopy, especially after taking into account details such as leaf angle, canopy structure and light source position, which can improve the accuracy of remote sensing data inversion. This is crucial for extracting parameters such as chlorophyll content, vegetation cover and biomass. For agriculture, the three-dimensional radiation transfer model can be used to simulate the growth of crops under different vegetation density, structure and irrigation conditions, which helps to scientifically manage crops, optimize the use of fertilizers and water, increase yields and optimize agricultural management.

[0003] In the three-dimensional vegetation radiation transfer modeling, scene construction is one of the core steps, which directly affects the accuracy and application effect of the model. Scene construction mainly involves the detailed simulation of the real physical structure of vegetation, environment, lighting and terrain. The real scene can accurately describe the vegetation structure. The vegetation structure in the three-dimensional model should be as close to the actual situation as possible, including the size, shape, arrangement of leaves, density and height distribution of the canopy layer, and the spatial distribution between different vegetation types. This detailed description can effectively improve the accuracy of the radiation transfer model, so that the inversion parameters (such as LAI, chlorophyll content, etc.) are more consistent with the real data. In addition, the real scene construction can simulate diverse lighting conditions. The real scene often includes diverse terrains and non-vegetated areas. These factors will significantly affect the scattering, absorption and reflection characteristics of radiation. By incorporating information such as environmental background and terrain characteristics into the three-dimensional radiation transfer model, the radiation transfer path in the real scene can be better reflected and the uncertainty in radiation modeling can be reduced.

[0004] After constructing a realistic three-dimensional scene, a simulation data set can be generated in combination with a radiation transfer model to provide a data calibration and verification benchmark for the development of remote sensing inversion algorithms. This simulated data set plays an important role in algorithm development, model comparison, and error assessment, especially in the absence of actual observation data, and can serve as an ideal comparison benchmark. Three-dimensional radiation transfer modeling based on real scenes supports multi-scale and multi-angle observation and analysis. By constructing scenes of different scales and perspectives, the effects of observation angles and resolution on radiation transfer can be better studied. Especially when observing from multiple angles, three-dimensional scenes can reveal the light reflection and transmission characteristics of each layer of the canopy, which is crucial for the development of multi-angle remote sensing inversion algorithms and correction algorithms. (III) Summary of the invention

[0005] The present invention relates to a three-dimensional aquatic vegetation canopy reflectance characteristic modeling method based on a real scene, and the steps are as follows: obtaining leaf spectrum and canopy spectrum through PROSPECT 5 model calculation; describing vegetation canopy reflectance characteristics through canopy structure parameters and leaf BRDF in combination with actual aquatic vegetation scene; calibrating model parameters by using four-flow approximation and linear approximation and applying them to approximately solve vegetation canopy reflectance; finally, using LuxCoreRender physical rendering engine to couple all input parameters, including incident radiation, vegetation canopy reflectance factor, and water optical properties, to calculate vegetation canopy reflectance, and the specific steps are as follows:

[0006] 1 A three-dimensional aquatic vegetation canopy reflectance characteristic modeling method based on a real scene, characterized by comprising the following steps:

[0007] (1) The leaf reflection and transmission spectra were calculated using the PROSPECT 5 plant leaf reflectance model, and the aquatic vegetation scene was constructed by combining the canopy spectrum based on measured data. The canopy reflectance characteristics of aquatic vegetation in a specific scene were described using canopy structure parameters and leaf BRDF.

[0008] (2) The four-flow approximation and linear approximation are used to calibrate the model parameters describing the reflectance of the vegetation canopy and water-soil substrate and apply them to approximate the reflectance of the vegetation canopy;

[0009] (3) LuxCoreRender graphics rendering engine is used to couple all input parameters, including incident radiation, vegetation canopy reflectance factor, and water optical property parameters, to calculate vegetation canopy reflectance.

[0010] 2. A three-dimensional aquatic vegetation canopy reflectance characteristic modeling method based on a real scene according to claim 1, characterized in that: the "leaf reflection and transmission spectra are calculated by the PROSPECT 5 plant leaf reflectance model, and the aquatic vegetation scene is constructed in combination with the canopy spectrum based on measured data, and the canopy structure parameters and leaf BRDF are used to describe the aquatic vegetation canopy reflectance characteristics of a specific scene" described in step (1); the specific method is as follows:

[0011] According to the flat plate model theory, a single leaf is divided into several layers, and then the spectral reflectance and transmittance of a single leaf are calculated; through field experiments, the morphological parameters of the plants are measured, including plant height, leaf area index and leaf inclination distribution; the biochemical parameters of aquatic vegetation leaves, including mesophyll structure parameters, chlorophyll concentration, carotenoid concentration, water content and dry matter concentration, are input into the PROSPECT 5 leaf radiation transmission model to calculate the reflectance and transmittance spectral data of a single leaf in the full band of 400-2500nm, that is, the leaf BRDF. The specific calculation method is as follows:

[0012] Assume that the blade consists of N layers of the same type, which are separated by N-1 layers of air; the first layer receives incident light with an incident angle of α, assuming that the light wave is isotropic, and ρ α represents the reflectivity, τ α represents the transmittance; inside the blade, ρ 90 and τ 90 Represent the reflectivity and transmittance of each layer of elements inside the blade respectively; then the total reflectivity R of N layers N,α and transmittance T N,α for:

[0013]

[0014] The scene model is composed of a set of 3D models of multiple randomly distributed single plants within a certain spatial range and a plane grid used to represent the water-soil background (hereinafter collectively referred to as "bottom"). In nature, the growth and reproduction characteristics of plants require them to satisfy certain topological relationships in three-dimensional space, which means that a certain growth spacing needs to be maintained between the various components of two plants, and there will be no complete three-dimensional space overlap, intersection, crossing or sharing of the same three-dimensional space point. For example, the independent rhizome parts of two plants will not completely overlap or intersperse; the branches, roots or leaves of plants will at most be close to each other, and will not occupy the same position in three-dimensional space at the same time. According to the plant density data, the distribution of single plant models in the entire field is set, and then the reflectance factor of the vegetation canopy is calculated; by simulating the canopy reflection under different lighting and observation geometries, a high-precision vegetation canopy reflectance spectrum is obtained.

[0015] 3. A three-dimensional aquatic vegetation canopy reflectance characteristic modeling method based on real scenes according to claim 1, characterized in that: the "using four-flow approximation and linear approximation to calibrate the model parameters describing the vegetation canopy and water-soil substrate reflectance and apply them to approximate the vegetation canopy reflectance" described in step (2) is specifically implemented as follows: the model is calibrated for a given canopy structure ∑ and observation geometry Ω, and it consists of two nested submodules, the first submodule describes the relationship between the vegetation canopy and the substrate reflectance, which depends on a physical parameter that needs to be calibrated (called S), and the second submodule describes the correlation between the wavelength of the S item and the optical properties of the blade, and the required calibration parameter is called F S ;

[0016] Step 1: Four-stream approximation calibration of the parameter S describing the relationship between vegetation canopy and bottom reflectance:

[0017] The canopy BRF (R) is a bidirectional component The weighted sum of Hemispherical reflection factor in:

[0018]

[0019] Among them, E Sun Represents the ground solar irradiance, E Sky Represents the ground sky irradiance; in order to improve the model rendering efficiency, diffuse light is not considered here, so it is assumed that E Sky is zero:

[0020]

[0021]

[0022] where r is the bottom reflectance assumed to be Lambertian, τ xy and ρ xy denote the transmittance and reflectance of the canopy, respectively. The incoming and outgoing directions are indicated by the subscripts x and y, respectively, and can be s (source direction), o (observation direction), or d (diffuse downward or upward direction). ssoo is the two-way gap probability, which is larger than τ in the near-hotspot region due to correlated photon paths ss τ oo , the formula can be expressed as:

[0023]

[0024] Where T1 = τ ssoo +τ ss τ do +τ sd τ do +τ sd τ oo And T2 = τ ssoo -τ ss τ oo , ρ so The terms correspond to canopy reflectance simulated with different substrate types:

[0025] r=r0=0.0=>R0=ρ so

[0026] ρ dd The terms T1 and T2 can be derived from the simulation of canopy reflectance, [R1, R2, R3] corresponding to the canopy reflectance of the simulated scenarios under three different substrate types, and the substrate reflectance is represented by r = [r1, r2, r3]. Then the term ρ can be calculated dd :

[0027]

[0028] Among them, M1 = (R1-R0) / r1, M2 = (R2-R0) / r2, M3 = (R3-R0) / r3, the derived four-flow parameter S[S 4s ]=[ρ dd , ρ so , T1, T2] depends on the canopy structure, while only three of them [ρ so ,T1,T2] also depends on the observation configuration defined by the viewing angle and the illumination direction, which is due to ρ dd corresponds to the diffuse flux, and in addition, [ρ dd ,ρ so ,T1] also depends on the leaf optical properties, while T2 should depend only on the gap fraction; therefore, the proposed solution to describe the correlation of canopy reflectance to substrate reflectance needs to be applied to different canopy structures ∑, observation configurations Ω, and leaf optical properties [ρ l , τ l ], for each combination [∑, Ω, [ρ l , τ l ]], using four different types of substrate reflectance reference values: [r] ref = [r0, r1, r2, r3], which contains the typical variation range of the bottom reflectivity, and calculates the four-flow parameter S[S 4s ];

[0029] Step 2: Linear approximation calibration of the parameter S describing the relationship between vegetation canopy and substrate reflectance:

[0030] The S term can also be calculated based on the linear approximation. By ignoring the nonlinear interaction terms T1 and T2 in the four-flow approximation, a further simplification can be introduced to obtain a linear approximation for the canopy reflectance, simplifying the S term to [S lin ]=[ρ so ,A].

[0031] R=ρ so +rA

[0032] A can be derived by ρ so And simulate the canopy reflection of another substrate type r1 for analysis and calculation,

[0033]

[0034] Similar to the four-flow solution, we need to l , τ l ]]Combined calculation [S lin ] term, therefore, a set of previous canopy reflectance simulations is needed to evaluate the accuracy of this linear approximation method: for each [∑,Ω,[ρ l , τ l, calculate [S lin = [ρ so , A];

[0035] Step 3: Calibrate the correlation parameter F between the wavelength of the S term and the leaf optical properties S :

[0036] By calibrating Fs(K) to determine its correlation with the total leaf absorption coefficient K, the leaf optical properties are affected by many factors, including surface characteristics, the internal structure of the mesophyll, the distribution of its biochemical components in the leaf volume, and the complex refractive index of these components. The real part of the refractive index n that drives the light scattering process changes little in the 450 - 2200 nm band, being between 1.3 < n < 1.5. Therefore, the spectral variation of the leaf optical properties is mainly driven by the imaginary part of the refractive index, which corresponds to the absorption coefficient. The total absorption coefficient K is determined by the specific absorption coefficients k b (λ) of each biochemical component, namely chlorophyll, dry matter, water, and the corresponding component contents C b as follows:

[0037] K = ∑k b (λ)C b

[0038] The variation range of the K value is calculated based on the minimum and maximum values of k b (λ) in PROSPECT5 and the upper and lower limits of C of the main absorbing substances (chlorophyll, water, and dry matter) in the LOPEX and ANGERS datasets. For a given leaf mesophyll structure parameter N, refractive index n, canopy structure v, and observation configuration Ω, the spectral variation of S based on the four - stream solution [S b = [ρ 4s , ρ dd , T1, T2], or in the linear approximation solution [S so = [ρ lin , A]) can be described by a smooth and monotonically increasing function that depends on K. For each canopy structure, observation configuration, mesophyll structure index, and refractive index, the dependence of the S term on the K value can be approximately expressed as the function Fs(K): so , A]) can be described by a smooth and monotonically increasing function that depends on K. For each canopy structure, observation configuration, mesophyll structure index, and refractive index, the dependence of the S term on the K value can be approximately expressed as the function Fs(K):

[0039]

[0040] For the 450 - 2200 nm band, select six values [K] ref .

[0041] 4. A three-dimensional aquatic vegetation canopy reflectance characteristic modeling method based on a real scene according to claim 1, characterized in that: the "using the LuxCoreRender graphics rendering engine to couple all input parameters, including incident radiation, vegetation canopy reflectance factor, and water body optical property parameters, to calculate vegetation canopy reflectance" in step (3) is specifically implemented as follows:

[0042] For typical aquatic vegetation canopy scenes, the leaf area index (LAI) generally ranges from 0.5 to 5, the average leaf angle (ALA) ranges from 20° to 45°, and the coverage ranges from 5% to 80%. For each set of scenes, the canopy reflectance under different observation geometries (VZA = [0°, 45°] and SZA = [20°, 35°]) as well as three substrate properties and six sets of leaf optical properties (corresponding to the central wavelengths of the six bands of the Sentinel-2 satellite sensor) were simulated. LuxCoreRender was used to calculate the reference canopy reflectance of 5184 scenes (36 canopy structures × 4 observation geometries × 3 substrate properties × 6 leaf feature combinations × 2 bands), while considering the spectral correlation of the refractive index n(λ). The six optical properties of the leaves were taken according to the LOPEX dataset and input into PROSPECT 5 model, and the water body bottom property parameters are calculated by the water body optical property parameterization step. Given the canopy structure ∑, observation geometry Ω and leaf mesophyll structure index N, the calibrated model is applied to calculate the canopy reflectance. First, the specific absorption coefficient k of the known leaf component is used b (λ), given leaf biochemical components C b and wavelength λ to calculate the total absorption coefficient K, and then use the previously calibrated function F S Calculate each S term for the considered K value; finally, for the actual substrate reflectance r, use the S term to calculate the vegetation canopy reflectance to approximately reflect the correlation of the canopy reflectance R to different substrate properties.

[0043] The advantages of the present invention compared with the prior art are:

[0044] (1) Current research is usually based on the assumption of canopy homogeneity, and it is assumed that the pixel contains only two components: leaves and soil. This simplified assumption is applicable to areas with high vegetation coverage, but for aquatic vegetation in habitats such as wetlands, the source of spectral reflectance signals is not limited to vegetation and soil, but also includes factors such as water scattering. In this case, the traditional inversion method will no longer be applicable. By introducing the spectral unmixing method, the present invention can more accurately invert the spectral characteristics of leaves, thereby improving the accuracy of vegetation reflectance simulation in wetlands and similar environments.

[0045] (2) Background noise can affect the accurate measurement of vegetation biophysical parameters, while saturation effects can weaken the response to parameter changes. In order to improve the accuracy of scene simulation, the present invention effectively reduces the impact of background noise and saturation effects, ensures a higher sensitivity to vegetation biophysical parameters, and significantly improves the performance of the model under complex environmental conditions. (IV) Description of the drawings

[0046] Figure 1 The technical process of the present invention; Figure 2 Comparison between LuxCoreRender reflection simulation and LESS simulation values; Figure 3 Comparison of LuxCoreRender BRF simulation and LESS model results under different VZAs. (V) Specific implementation methods

[0047] In order to better illustrate the three-dimensional aquatic vegetation canopy reflectance characteristic modeling method based on real scenes involved in the present invention, the model of the present invention and the LESS model were used to conduct tests and analyses, and good results were achieved. The specific implementation method is as follows:

[0048] (1) The leaf reflection and transmission spectra were calculated using the PROSPECT 5 plant leaf reflectance model, and the aquatic vegetation scene was constructed by combining the canopy spectrum based on measured data. The canopy reflectance characteristics of aquatic vegetation in a specific scene were described using canopy structure parameters and leaf BRDF.

[0049] (2) The four-flow approximation and linear approximation are used to calibrate the model parameters describing the reflectance of the vegetation canopy and water-soil substrate and apply them to approximate the reflectance of the vegetation canopy;

[0050] (3) LuxCoreRender graphics rendering engine is used to couple all input parameters, including incident radiation, vegetation canopy reflectance factor, and water optical property parameters, to calculate vegetation canopy reflectance.

[0051] Single scattering verification and multiple scattering verification were carried out on the main plane of the sun, and the BRF simulation results of the virtual photon model based on the LESS model extension in three commonly used land scenarios, HET10, HOM23 and HOM33, were compared. Figure 2 As shown in Figure 2, the results show that LuxCoreRender is in very good agreement with the RAMI reference data, with the root mean square error being 0.0017 and 0.006 for single scattering and multiple scattering, respectively. 2 The values ​​for single scattering and multiple scattering are 99.46% and 92.61% respectively. Figure 3The comparison results of LuxCoreRender and LESS simulation data under forward and backward observation angles. The red line in the figure is the BRF simulation result of LuxCoreRender, and the black line is the LESS result, which respectively shows the comparison results of single scattering and multiple scattering. The single scattering radiance simulated by LuxCoreRender is similar to the LESS model, and is also sampled using the ray tracing method, so there will be no low vegetation hotspots, and the hotspots of the two methods can also be completely matched. Therefore, it can be concluded that the LuxCoreRender rendering simulation and its implementation in this study have achieved the same level of accuracy as the most advanced 3D models participating in the RAMI test.

[0052] The method for simulating the real scene of aquatic vegetation and its canopy reflectance characteristics in the present invention is helpful to explore the relationship between the canopy reflectance spectrum of aquatic vegetation and solar radiation, vegetation structural parameters, and water-soil substrate reflectance characteristics in a more in-depth and scientific manner. At the same time, the present invention significantly improves the spatiotemporal continuous simulation capability of remote sensing modeling, accurately inverts the leaf spectral characteristics, and significantly improves the simulation accuracy of wetland vegetation reflectance. By effectively reducing the interference of background noise and saturation effect, the sensitivity to the physical and chemical parameters of vegetation biomass is enhanced, ensuring the versatility and reliability of the model.

Claims

1. A three-dimensional aquatic vegetation canopy reflectance characteristic modeling method based on real scenes, characterized in that The following steps are involved: (1) The leaf reflection and transmission spectra were calculated using the PROSPECT 5 plant leaf reflectance model, and the aquatic vegetation scene was constructed by combining the canopy spectrum based on measured data. The canopy reflectance characteristics of aquatic vegetation in a specific scene were described using canopy structure parameters and leaf BRDF. (2) The four-flow approximation and linear approximation were used to calibrate the model parameters describing the reflectance of the vegetation canopy and water-soil substrate, and then applied to approximate the reflectance of the vegetation canopy; (3) LuxCoreRender physical rendering engine is used to couple all input parameters, including incident radiation, vegetation canopy reflectance factor, and water optical property parameters, to calculate the vegetation canopy reflectance.

2. The method for modeling the reflectance characteristics of a three-dimensional aquatic vegetation canopy based on a real scene according to claim 1, characterized in that: The specific method of step (1) is as follows: "the leaf reflection and transmission spectra are calculated by the PROSPECT 5 plant leaf reflectance model, and the aquatic vegetation scene is constructed by combining the canopy spectrum based on the measured data, and the canopy structure parameters and leaf BRDF are used to describe the canopy reflection characteristics of aquatic vegetation in a specific scene." According to the flat plate model theory, a single leaf is divided into several layers, and then the spectral reflectance and transmittance of a single leaf are calculated; through field experiments, the morphological parameters of the plants are measured, including plant height, leaf area index and leaf inclination distribution; the biochemical parameters of aquatic vegetation leaves, including mesophyll structure parameters, chlorophyll concentration, carotenoid concentration, water content and dry matter concentration, are input into the PROSPECT 5 leaf radiation transmission model to calculate the reflectance and transmittance spectral data of a single leaf in the full band of 400-2500nm, that is, the leaf BRDF. The specific calculation method is as follows: Assume that the blade consists of N layers of the same type, which are separated by N-1 layers of air; the first layer receives incident light with an incident angle of α, assuming that the light wave is isotropic, and ρ α represents the reflectivity, τ α represents the transmittance; inside the blade, ρ 90 and τ 90 Represent the reflectivity and transmittance of each layer of elements inside the blade respectively; then the total reflectivity R of N layers N,α and transmittance T N,α for: The scene model is composed of a set of 3D models of multiple randomly distributed single plants within a certain spatial range and a plane grid used to represent the water-soil background. The "water-soil background" is collectively referred to as "bottom" in the following text. In nature, the growth and reproduction characteristics of plants require them to satisfy certain topological relationships in three-dimensional space, which means that a certain growth spacing needs to be maintained between the various components of two plants, and there will be no complete three-dimensional space overlap, intersection, crossing or sharing of the same three-dimensional space point. For example, the independent rhizome parts of two plants will not completely overlap or intersperse; the branches, roots or leaves of plants will at most be close to each other, and will not occupy the same position in three-dimensional space at the same time; according to the plant density data, the spatial distribution pattern of the single plant model in the entire field is set, and then the reflection factor of the vegetation canopy is calculated; by simulating the canopy reflection under different illumination and observation geometry, a high-precision vegetation canopy reflection spectrum is obtained.

3. The method for simulating the reflectance characteristics of a three-dimensional aquatic vegetation canopy based on a real scene according to claim 1, characterized in that: The specific implementation method of "using the four-flow approximation and the linear approximation to calibrate the model parameters describing the reflectance of the vegetation canopy and the water-soil substrate, and applying them to the approximate solution of the vegetation canopy reflectance" in step (2) is as follows: the model is calibrated for a given canopy structure ∑ and an observation geometry structure Ω. It consists of two nested submodules. The first submodule describes the relationship between the vegetation canopy and the substrate reflectance, which depends on a physical parameter that needs to be calibrated, called S. The second submodule describes the correlation between the wavelength of the S term and the optical properties of the leaves. The required calibration parameter is called F S ; Step 1: Four-stream approximation calibration of the parameter S describing the relationship between vegetation canopy and bottom reflectance: The canopy directional reflectance factor BRF is a bidirectional component The weighted sum of Hemispherical reflection factor in: Among them, E Sun Represents the solar irradiance received by the ground, E Sky Indicates the sky irradiance received by the ground; in order to improve the rendering efficiency of the model, diffuse light is not considered here, so it is assumed that E Sky is zero: where r is the bottom reflectance assumed to be Lambertian; τ xy and ρ xy denote the transmittance and reflectance of the canopy, respectively; the incoming and outgoing directions are indicated by the subscripts x and y, respectively, which can be s, indicating the source direction; o, indicating the observation direction; or d, indicating the diffusion downward or upward direction; τ ssoo is the two-way gap probability, which is larger than τ in the near-hotspot region due to correlated photon paths ss τ oo , the formula can be expressed as: Where T1 = τ ssoo +τ ss τ do +τ sd τ do +τ sd τ oo And T2 = τ ssoo -τ ss τ oo , ρ so The terms correspond to canopy reflectance simulated with different substrate types: r=r0=0.0=>R0=ρ so ρ dd The terms T1 and T2 can be derived from the simulation of canopy reflectance, [R1, R2, R3] corresponding to the canopy reflectance of the simulated scenarios under three different substrate types, and the substrate reflectance is represented by r = [r1, r2, r3]; then the term ρ can be calculated dd : Among them, M1 = (R1-R0) / r1, M2 = (R2-R0) / r2, M3 = (R3-R0) / r3, the derived four-flow parameter S[S 4s ]=[ρ dd , ρ so , T1, T2] depends on the canopy structure, while only three of them [ρ so ,T1,T2] also depends on the observation configuration defined by the viewing angle and the illumination direction, which is due to ρ dd corresponds to the diffuse flux, and in addition, [ρ dd ,ρ so ,T1] also depends on the leaf optical properties, while T2 should depend only on the gap fraction; therefore, the proposed solution to describe the correlation of canopy reflectance to substrate reflectance needs to be applied to different canopy structures ∑, observation configurations Ω, and leaf optical properties [ρ l , τ l ], for each combination [∑, Ω, [ρ l , τ l ]], using four different types of substrate reflectance reference values: [r] ref = [r0, r1, r2, r3], which contains the typical variation range of the bottom reflectivity, and calculates the four-flow parameter S[S 4s ]; Step 2: Linear approximation calibration of the parameter S describing the relationship between vegetation canopy and substrate reflectance: The S term can also be calculated based on the linear approximation. By ignoring the nonlinear interaction terms T1 and T2 in the four-flow approximation, a further simplification can be introduced to obtain a linear approximation for the canopy reflectance, simplifying the S term to [S lin ]=[ρ so ,A]; R=ρ so +rA A can be derived by ρ so And simulate the canopy reflection of another substrate type r1 for analysis and calculation, Similar to the four-flow solution, we need to l , τ l ]]Combined calculation [S lin ] term, therefore, a set of previous canopy reflectance simulations is needed to evaluate the accuracy of this linear approximation method: for each [∑,Ω,[ρ l , τ l ]], calculate [S lin ]=[ρ so ,A]; Step 3: Calibrate the wavelength of S and the correlation parameter F of the blade optical properties S : By calibrating Fs(K) to determine its correlation with the total leaf absorption coefficient K, the leaf optical properties are affected by many aspects, including surface characteristics, the internal structure of the mesophyll, the distribution of its biochemical components in the leaf volume, and the complex refractive indices of these components. The real part of the refractive index n that drives the light scattering process changes little in the 450 - 2200 nm band and is between 1.3 < n < 1.

5. Therefore, the spectral variation of the leaf optical properties is mainly driven by the imaginary part of the refractive index, which corresponds to the absorption coefficient. The total absorption coefficient K is determined by the specific absorption coefficients k b (λ) of each biochemical component, namely chlorophyll, dry matter, water, and the corresponding component contents C b as follows: K=∑k b (λ)C b The range of K values ​​is based on the k b The minimum and maximum values ​​of (λ) and the C of the main absorbing substances—chlorophyll, water, and dry matter—in the LOPEX and ANGERS datasets b The upper and lower limits of the values ​​are calculated. For a given leaf mesophyll structure parameter N, refractive index n, canopy structure ∑ and observation configuration Ω, the spectral variation of S based on the four-flow solution [S 4s ]=[ρ dd , ρ so , T1, T2], or in a linear approximation solution [S lin ]=[ρ so ,A]) can be described by a smooth and monotonically increasing function that depends on K. For each canopy structure, observation configuration, mesophyll structure index and refractive index, the dependence of the S term on the K value can be approximated by the function Fs(K): For the 450-2200nm band, select six values ​​that best represent the relationship between K value and the change in leaf reflection and transmission [K] ref .

4. The method for modeling the reflectance characteristics of a three-dimensional aquatic vegetation canopy based on a real scene according to claim 1, characterized in that: The specific implementation method of "using the LuxCoreRender graphics rendering engine to couple all input parameters, including incident radiation, vegetation canopy reflectance factor, and water optical property parameters, to calculate vegetation canopy reflectance" in step (3) is as follows: For typical aquatic vegetation canopy scenes, the leaf area index (LAI) generally ranges from 0.5 to 5, the average leaf angle (ALA) ranges from 20° to 45°, and the coverage ranges from 5% to 80%. For each set of scenes, the canopy reflectance under different observation geometries (VZA = [0°, 45°] and SZA = [20°, 35°]) as well as three substrate properties and six sets of leaf optical properties (corresponding to the central wavelengths of the six bands of the Sentinel-2 satellite sensor) were simulated. LuxCoreRender was used to calculate the reference canopy reflectance of 5184 scenes (36 canopy structures × 4 observation geometries × 3 substrate properties × 6 leaf feature combinations × 2 bands), while considering the spectral correlation of the refractive index n(λ). The six optical properties of the leaves were taken according to the LOPEX dataset and input into PROSPECT 5 model, and the water body bottom property parameters are calculated by the water body optical property parameterization step. Given the canopy structure ∑, observation geometry Ω and leaf mesophyll structure index N, the calibrated model is applied to calculate the canopy reflectance. First, the specific absorption coefficient k of the known leaf component is used b (λ), given leaf biochemical components C b and wavelength λ to calculate the total absorption coefficient K, and then use the previously calibrated function F S Calculate each S term for the considered K value; finally, for the actual substrate reflectance r, use the S term to calculate the vegetation canopy reflectance to approximately reflect the correlation of the canopy reflectance R to different substrate properties.

Images