Multi-band full-polarization SAR vegetation root zone profile soil water content inversion method

Abstract

The invention discloses a multiband full-polarization SAR vegetation root zone profile soil water content inversion method, and relates to the technical field of remote sensing image processing, and the method comprises the steps: employing multiband full-polarization SAR time sequence observation data, constructing a multiband multi-polarization SAR observation combined time sequence inversion algorithm, introducing soil humidity dielectric attribute space-time constraint, and obtaining high-precision soil humidity; inverting a surface complex dielectric constant by using a multi-band and multi-temporal polarization target decomposition model, and further calculating the SAR surface penetration depth; calculating a scale factor of penetration depth by combining an SAR root zone soil humidity inversion result and an actual section soil humidity observation value; according to the soil humidity inversion result, the earth surface penetration depth and the scale factors of different wave bands, the function relation between the soil humidity and the depth of the vegetation root zone profile is determined. Therefore, by adopting the method, high-resolution vegetation root zone profile soil moisture content inversion can be realized, and an accurate and reliable scientific basis is provided for precision agriculture drought monitoring and soil moisture content monitoring.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing image processing, and in particular to a method for retrieving soil moisture content in the vegetation root zone profile using multi-band fully polarized SAR. Background Art

[0002] Soil moisture content is a key parameter in hydrological cycle, agricultural production and ecosystem research. Although traditional ground measurement methods can obtain relatively accurate soil moisture content data, their measurement range is limited, making it difficult to meet the monitoring needs of large-scale areas, and there are limitations such as destructiveness, limited spatial coverage and difficulty in dynamic monitoring. Synthetic Aperture Radar (SAR) technology, especially multi-band fully polarized SAR, has become an effective means for large-scale and non-destructive soil moisture content monitoring due to its all-weather, all-day, high-resolution and penetration capabilities.

[0003] However, existing SAR soil moisture inversion studies often assume the surface soil layer as a homogeneous half-space with uniform surface dielectric properties, ignoring the vertical variability of soil moisture, and it is difficult to be directly applied to vegetated areas, especially for retrieving the soil moisture content in the vegetation root zone profile. The vegetation canopy and stems will strongly scatter and attenuate the microwave signal, masking the weak signal from the root zone soil, resulting in a large deviation in the inversion results. Existing methods mostly rely on simple vegetation correction models, which are difficult to accurately separate the scattering contributions of vegetation and soil, especially in areas with dense vegetation or complex structures. Moreover, different bands and polarization modes have different sensitivities to vegetation and soil, increasing the inversion difficulty and being unfavorable for achieving high-precision and large-scale monitoring of soil moisture content in the vegetation root zone. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for retrieving soil moisture content in the vegetation root zone profile using multi-band fully polarized SAR, which can utilize the differences in the penetration depths of SAR signals in different bands to achieve high-precision inversion of the soil moisture content in the vegetation root zone profile, and overcome the defect that the polarization SAR observation data in the prior art cannot accurately decouple the surface and vegetation scattering.

[0005] To achieve the above purpose, the present invention provides a method for retrieving soil moisture content in the vegetation root zone profile using multi-band fully polarized SAR, including:

[0006] S1. Collect and preprocess the time-series multi-band fully polarized SAR data of the vegetated area to obtain the filtered coherence matrix T′3;

[0007] S2. Using the polarimetric target decomposition model, accurately decouple the surface and vegetation scattering of the filtered coherence matrix T'3 obtained in step S1 to obtain the coherence matrix of the surface scattering component, the model adaptive optimal parameters, and the backscattering coefficient of the surface scattering, which are used to construct a soil moisture inversion algorithm;

[0008] S3. Based on the surface scattering component obtained in step S2, add the alpha approximation model observables to construct a temporal inversion algorithm for joint multi-band and multi-polarization SAR observations;

[0009] S4. Collect passive microwave data corresponding to the time and regional scope in step S1, construct a spatio-temporal domain soil dielectric constant constraint mechanism, and embed it into the temporal inversion algorithm constructed in step S3 to obtain high-precision soil moisture in the vegetation root zone;

[0010] S5. Increase the observables by combining multi-band and multi-temporal observation data, expand the polarimetric target decomposition model in step S2, construct a multi-band and multi-temporal polarimetric target decomposition model, and invert the complex dielectric constant of the surface;

[0011] S6. Using the complex dielectric constant of the surface inverted in step S5, calculate the penetration depth of SAR in different bands in the vegetation root zone;

[0012] S7. Determine the scaling factor according to the measured data of the surface profile soil moisture, the soil moisture obtained in step S4, and the penetration depth calculated in step S6;

[0013] S8. Based on steps S4, S6, and S7, calculate the soil moisture, penetration depth, and their scaling factors inverted by the profile SAR in different bands, determine the profile soil temperature analysis model in the vegetation root zone of the observation area, and solve the profile soil water content.

[0014] Preferably, step S1 includes:

[0015] S11. Using the image data of three polarization modes, namely HH, VV, HV, or VH, of the polarimetric SAR remote sensing image, generate the coherence matrix T3;

[0016] S12. Use the boxcar filtering algorithm to perform coherence filtering on T3 to obtain the filtered coherence matrix T'3.

[0017] Preferably, in step S2, the surface, even-order, and volume scattering models of the polarimetric target decomposition model respectively adopt the improved X-Bragg model, the improved double Fresnel scattering model, and the adaptive volume scattering model GVSM;

[0018] Among them, the expression of the improved X-Bragg model is as follows:

[0019]

[0020] In the formula, T S_X-Bragg is the polarization coherence matrix of the improved X - Bragg model, f s is the surface scattering power coefficient, β is the Bragg scattering coefficient, and δ is the surface roughness depolarization angle;

[0021] The expression of the improved double - Fresnel scattering model is as follows:

[0022]

[0023]

[0024] In the formula, T XD_Fresnel is the coherence matrix of the improved double - Fresnel scattering, f d is the even - order scattering power, T XD11 , T XD22 , T XD12 , T XD21 , T XD33 are the elements of the even - order scattering polarization coherence matrix;

[0025] The expression of the adaptive volume scattering model GVSM is as follows:

[0026]

[0027] In the formula, γ, ρ, e are the vegetation scattering model parameters, e = 2<|S HV | 2 > / <|S VV | 2 >, which can be obtained under the condition of the rotation - symmetry hypothesis of <S HH S VV * >. f v is the power of the vegetation volume scattering component.

[0028] Preferably, step S2 includes:

[0029] S21. Use the improved double - Fresnel scattering model and the adaptive volume scattering model GVSM to construct the cost equations for the filtered coherence matrix T′3 and the theoretical scattering of T'3. Through polarization target decomposition processing, solve the cost equations, eliminate the vegetation volume scattering and even - order scattering components, extract the coherence matrix of the surface scattering component, calculate the polarization coherence matrix of the improved X - Bragg model, obtain the S′ HH , S′ VV , S′ HV backscattering coefficients, and calculate the surface roughness depolarization angle;

[0030] S22. Substitute the depolarization angle of the surface roughness obtained in step S21 into the polarization target decomposition model to perform the second-stage polarization target decomposition process, eliminate surface scattering and even-order scattering, extract the coherence matrix of the vegetation body scattering model, and perform unitary matrix transformation to calculate the corresponding covariance matrix C volume , and then according to C volume matrix elements, calculate the anisotropy and shape parameters of the vegetation body scattering model;

[0031] S23. Substitute the depolarization angle of the surface roughness obtained in step S21 into the improved X-Bragg model, and substitute the anisotropy and shape parameters of the vegetation body scattering model obtained in step S22 into the adaptive volume scattering model GVSM model to construct the filtered coherence matrix T'3 and the cost equation of the T′3 theoretical scattering model. Through step-by-step iteration and least squares solution, calculate the optimal co-polarization phase difference parameter;

[0032] S24. Based on the polarization target decomposition model, according to the obtained depolarization angle of the surface roughness, co-polarization phase difference, vegetation anisotropy and shape parameters, extract the HH and VV polarimetric backscattering coefficients of the surface scattering component and as follows:

[0033]

[0034] In the formula, "*" represents the complex conjugate, T S_X-Bragg_11 , T S_X-Bragg_12 , T S_X-Bragg_22 are the respective elements of the surface scattering coherence matrix.

[0035] Preferably, in step S3, the alpha approximation model eliminates the scattering contributions of vegetation cover and rough surface by using the ratio of backscattering coefficients of adjacent time phases, and by combining multi-band observation data and different polarization observation information, increases the number of observables, expands the observation matrix, and realizes the positive definite solution of unknown parameters.

[0036] Preferably, the multi-band multi-polarization SAR observation joint time series inversion algorithm includes obtaining the surface dielectric constant of each time phase by using the least squares method or a look-up table, and the expression is as follows:

[0037]

[0038]

[0039] In the formula, M HH_VV is the time series observation matrix, A HH_VV is the time series alpha coefficient vector, λ1, λ2 represent different bands and λ1>λ2, N represents the number of time phases of the time series observation, ε soilis the soil dielectric constant, ξ is the incident angle, α HH , α VV respectively represent the horizontal polarization and vertical polarization Fresnel scattering coefficients.

[0040] Preferably, step S4 includes:

[0041] S41. Collect passive microwave data within the time and area ranges corresponding to step S1, calculate the minimum and maximum values of the soil moisture data product within the selected time and area ranges, and back-calculate the minimum and maximum values of the corresponding surface dielectric constant;

[0042] S42. Use the minimum and maximum values of the surface dielectric constant calculated in step S41 as the solution constraints for the unknown parameters of the surface dielectric constant, substitute them into the multi-band and multi-polarization SAR observation joint time-series inversion algorithm constructed in step S3, solve the surface dielectric constant under high-resolution observation, and convert it into soil moisture to obtain a high-precision soil moisture inversion result.

[0043] Preferably, in step S5, the polarization target decomposition model for multi-band and multi-temporal observation joint is as follows:

[0044]

[0045] where, [T] O1 and [T] O2 are the polarization coherence matrices of two different band or different temporal polarization SAR observations respectively, [T S_X-Bragg , [T XD_Fresnel , [T GV GVSM are the corresponding surface scattering, even-order scattering and volume scattering models respectively, and the superscripts 1 and 2 represent the corresponding bands or temporal phases.

[0046] Preferably, in step S7, under different band observations, the calculation formula of the scale factor K SAR is as follows:

[0047]

[0048] In the formula, K C , K L , K P are the scale factors for C, L, and P bands respectively, and M msr_fit is the relationship formula between the actual depth and the measured soil moisture.

[0049] Preferably, in step S8, the profile soil moisture analysis model is:

[0050] M (i,j) = a (i,j) ξ 2 + b​(i,j) ξ + d (i,j) ;

[0051] In the formula, M (i,j) is the soil moisture at the pixel (i, j), ξ is the penetration depth; a (i,j) , b (i,j) , d (i,j) are polynomial coefficients, which are determined by a multivariate observation equation constructed from the soil moisture, penetration depth, and scale factor retrieved from SAR of different bands.

[0052] Therefore, the present invention adopts the above method for retrieving the soil water content of the vegetation root zone profile using multi - band fully - polarized SAR, and has the following technical effects:

[0053] The present invention utilizes multi - band, multi - temporal fully - polarized SAR data, combines passive microwave data constraints, retrieves high - precision soil moisture in the vegetation root zone, combines the measured data of soil moisture in the vegetation root zone, establishes an analytical model of soil moisture in the root zone profile, realizes the retrieval of soil water content in the vegetation root zone profile, and provides a method for high - precision, large - scale, non - destructive monitoring of soil water content in the vegetation root zone and fine - agricultural soil moisture monitoring.

[0054] Next, through the drawings and embodiments, the technical solutions of the present invention will be further described in detail. Description of the Drawings

[0055] Figure 1 is a flowchart of a method for retrieving the soil water content of the vegetation root zone profile using multi - band fully - polarized SAR;

[0056] Figure 2 is a flowchart of constraining the soil dielectric constant with passive microwave data in an embodiment of a method for retrieving the soil water content of the vegetation root zone profile using multi - band fully - polarized SAR;

[0057] Figure 3 is a schematic diagram of the electromagnetic wave soil penetration depth in an embodiment of a method for retrieving the soil water content of the vegetation root zone profile using multi - band fully - polarized SAR;

[0058] Figure 4 is a flowchart of solving the scale factor in an embodiment of a method for retrieving the soil water content of the vegetation root zone profile using multi - band fully - polarized SAR. Detailed Embodiments

[0059] The present invention can be more detailedly explained through the following embodiments. The purpose of disclosing the present invention is to protect all changes and improvements within the scope of the present invention. The present invention is not limited to the following embodiments.

[0060] Please refer to Figure 1, the present invention provides a method for retrieving soil moisture content in the vegetation root zone profile of multi-band fully polarized SAR, including the following steps:

[0061] S1. Collect and preprocess the multi-band fully polarized SAR data in the vegetation-covered area over time series to obtain the filtered coherence matrix T′3. The specific steps are as follows:

[0062] S11. Generate the coherence matrix T3 using the image data of the HH, VV, HV, or VH polarization modes of the polarimetric SAR remote sensing image.

[0063] S12. Perform coherence filtering on T3 using the boxcar filtering algorithm to obtain the filtered coherence matrix T'3.

[0064] S2. Use the polarimetric target decomposition model to accurately decouple the surface and vegetation scattering of the filtered coherence matrix T'3 obtained in step S1, and obtain the coherence matrix of the surface scattering component, the model-adaptive optimal parameters, and the backscattering coefficient of the surface scattering, which are used to construct the soil moisture inversion algorithm.

[0065] In this embodiment, the surface, even-order, and volume scattering models of the polarimetric target decomposition model respectively adopt the improved X-Bragg model, the improved double Fresnel scattering model, and the adaptive volume scattering model GVSM, as follows:

[0066] The expression of the X-Bragg model is:

[0067]

[0068] where T S_X-Bragg is the polarization coherence matrix of the X-Bragg model, f s is the surface scattering power coefficient, β is the Bragg scattering coefficient, δ is the surface roughness depolarization angle, which is related to the circular polarization correlation coefficient (γ RRLL ). The specific forms of the relevant parameters are:

[0069]

[0070]

[0071] In the formula, θ is the incident angle, ε soil is the surface dielectric constant related to the soil moisture, S HH and S VV are the horizontal and vertical polarization backscattering coefficients respectively, S RR and S LL are the right-handed and left-handed circular polarization scattering coefficients respectively.

[0072] The expression of the improved double Fresnel scattering model is:

[0073]

[0074] In the formula, T XD_Fresnel is the improved double Fresnel scattering coherence matrix; S XD is the even-order scattering matrix considering the surface depolarization angle; f d is the even-order scattering power; T XD11 , T XD22 , T XD12 , T XD21 , T XD33 are the elements of the even-order scattering polarization coherence matrix, and each element is a function related to the surface dielectric constant (ε soil ), the vegetation dielectric constant (ε trunk ), and the incident angle (θ); δ is the surface roughness depolarization angle; ε t / s is the vegetation stem dielectric constant; ε soil is the soil dielectric constant; is the co-polarization phase difference parameter, usually set to 0.

[0075] The expression of the adaptive volume scattering model GVSM is:

[0076]

[0077] In the formula, γ, ρ, e are the vegetation scattering model parameters, e = 2<|S HV | 2 > / <|S VV | 2 >, and it can be obtained under the assumption of rotational symmetry of <S HH S VV * >. f v is the power of the vegetation volume scattering component.

[0078] Based on the above model, the specific operation of step S2 is as follows:

[0079] S21. Use the improved double Fresnel scattering model and the adaptive volume scattering model GVSM to construct the cost equations for the filtered coherence matrix T′3 and the theoretical scattering of T′3. Through polarization target decomposition processing, solve the cost equations, eliminate the vegetation volume scattering and even-order scattering components, extract the coherence matrix of the surface scattering component, calculate the polarization coherence matrix of the improved X-Bragg model, obtain the S′ HH , S′ VV , S′ HV backscattering coefficients, and calculate the surface roughness depolarization angle δ', and the formula is as follows:

[0080]

[0081] wherein, S RR and S LL are the right - hand circular polarization and left - hand circular polarization scattering coefficients respectively.

[0082] S22. Substitute the surface roughness depolarization angle obtained in step S21 into the polarization target decomposition model, perform the second - stage polarization target decomposition process, eliminate surface scattering and even - order scattering, extract the coherent matrix T of the vegetation body scattering model volume , and perform a unitary matrix transformation to calculate the corresponding covariance matrix C volume . Then, according to the C volume matrix elements, calculate the anisotropy γ′ and shape parameter ρ′ of the vegetation body scattering model, where γ′ = <|S′ HH | 2 > / <|S′ VV | 2 >, and S′ HH , S′ VV are the backscattering coefficients of the surface scattering components.

[0083] S23. Substitute the surface roughness depolarization angle obtained in step S21 into the improved X - Bragg model, and substitute the anisotropy and shape parameters of the vegetation body scattering model obtained in step S22 into the adaptive volume scattering model GVSM model, construct the filtered coherent matrix T′3 and the cost equation of the T′3 theoretical scattering model, and calculate the optimal co - polarization phase - difference parameter (non - zero).

[0084] S24. Based on the polarization target decomposition model, according to the obtained surface roughness depolarization angle, co - polarization phase - difference vegetation anisotropy γ′ and shape parameter ρ′, extract the HH and VV polarization backscattering coefficients and of the surface scattering component as follows:

[0085]

[0086] wherein, "*" represents the complex conjugate, and T S_X-Bragg_11 , T S_X-Bragg_12 , T S_X-Bragg_22 are the respective elements of the surface scattering coherent matrix.

[0087] S3. Based on the surface scattering component obtained in step S2, add the alpha - approximation model observables to construct a multi - band multi - polarization SAR observation - combined temporal inversion algorithm, that is, a multi - polarization temporal soil moisture inversion algorithm model, specifically as follows:

[0088]

[0089] Among them,

[0090]

[0091] In the formula, M HH_VV is the temporal observation matrix, A HH_VV is the temporal alpha coefficient vector, λ1 and λ2 represent different bands and λ1 > λ2, and N represents the number of temporal phases of the temporal observation.

[0092] Among them, the classical alpha approximation model soil moisture inversion algorithm is:

[0093]

[0094] In the formula, f(VWC, s) represents the scattering from the surface and vegetation cover; f(ε s , PP, θ) represents the surface scattering at PP polarization and incident angle θ; α HH (ε soil , θ) and α VV (ε soil , θ) represent the horizontal polarization and vertical polarization Fresnel scattering coefficients respectively.

[0095] The temporal (N) observation matrix (M PP ) of the polarized backscattering coefficient ratio is:

[0096]

[0097] The corresponding temporal backscattering coefficient vector is:

[0098]

[0099] In summary, according to the temporal observation matrix M PP , the temporal alpha coefficient vector A PP can be solved, and then the surface dielectric constant of each phase can be obtained by using the least squares method or lookup table. Then, based on the dielectric mixing model proposed by Topp et al. in 1980, the surface dielectric constant can be converted into the surface soil volumetric water content.

[0100] S4. Collect passive microwave data in the time and area range corresponding to step S1, construct a spatio-temporal domain soil dielectric constant constraint mechanism, and embed it into the temporal inversion algorithm constructed in step S3 to obtain high-precision soil moisture in the vegetation root zone. Please refer to Figure 2 , and the specific operations are as follows:

[0101] S41. Collect passive microwave data (such as SMAP, SMOS, AMSR2, etc.) within the selected time and regional range, calculate the minimum and maximum values of the soil moisture content data product within the selected time and regional range, and back-calculate the minimum and maximum values of the corresponding surface dielectric constant.

[0102] S42. Use the minimum and maximum values of the surface dielectric constant calculated in step S41 as the solution constraints for the unknown parameters of the surface dielectric constant in the inversion algorithm, substitute them into the multi-polarization time-series soil moisture inversion algorithm model constructed in step S3, solve the surface dielectric constant under high-resolution observation, and convert it into soil moisture to obtain a high-precision soil moisture inversion result.

[0103] S5. Since there is an underdetermined solution problem of the unknown alpha coefficient in the soil moisture inversion algorithm based on the alpha approximation model, it is necessary to jointly use multi-band and multi-temporal observation data to increase the number of observations, expand the polarization target decomposition model in step S2, construct a multi-band and multi-temporal polarization target decomposition model, realize the positive definite solution of the unknown parameters, and invert the surface complex dielectric constant.

[0104] Among them, the framework of the multi-band and multi-temporal polarization target decomposition model is as follows:

[0105]

[0106] Among them, [T] O1 and [T] O2 are the polarization coherence matrices of two different-band or different-temporal polarization SAR observations respectively, and [T S_X-Bragg , [T XD_Fresnel , [T GV GVSM are the corresponding surface scattering, even-order scattering, and volume scattering models respectively, and the superscripts 1 and 2 represent the corresponding bands or time phases.

[0107] Based on the above multi-band and multi-temporal polarization target decomposition model, the solution of unknown parameters such as the surface complex dielectric constant, the dielectric constant of the vegetation vertical structure, and the power of each scattering component can be realized by using multivariate nonlinear equation fitting, as follows:

[0108]

[0109] X = [f s 1 f d 1 f v 1 f s 2 f d 2 f v ​2 ε soil 1 ε soil 2 ε trunk 1 ε trunk 2 ;

[0110] Wherein, The positive definite solution of the unknown parameter vector X can be realized by using the least square fitting algorithm, and the inversion result of the complex permittivity of the ground surface can be obtained.

[0111] S6. The propagation of electromagnetic waves in a lossy medium (soil) is as Figure 3 shown. Using the complex permittivity of the ground surface inverted in step S5, calculate the penetration depth ξ of SAR at different bands in the vegetation root zone SAR , and the specific steps are as follows:

[0112] S61. Based on the empirical model of the ground surface soil moisture and the complex permittivity, calculate the real part and the imaginary part of the complex permittivity in S5, and the constraint condition is set as the tangent loss ratio tanδ = ε” / ε′ < 0.1, where ε′ is the real part and ε” is the imaginary part.

[0113] S62. According to the definition of the electromagnetic wave penetration depth, calculate the ground surface penetration depth, and the expression is as follows:

[0114]

[0115] S7. Determine the scaling factor K according to the measured data of the ground surface profile soil moisture, the SAR soil moisture inversion result M SAR in step S4 SAR and the penetration depth ξ SAR in step S6. Please refer to Figure 4 , and the specific steps include:

[0116] S71. According to different depths (such as 0 - 40 cm) and their corresponding actual observed values of soil moisture, the relationship model coefficients of soil moisture information and ground surface depth can be obtained by using polynomial fitting, that is, A, B, D, and the expression is:

[0117] M msr_fit = Aξ 2 + Bξ + D;

[0118] S72. Using the soil moisture (M C , M L , M P ) inverted from the SAR observation data of different bands (X / C, L, P) and the estimated ground surface penetration depth (ξ C , ξ L , ξP ) to calculate the proportionality factors (K C , K L , K P ) of the integral of profile soil moisture to depth under observations in different bands:

[0119]

[0120]

[0121] S8. Based on the soil moisture (M C , M L , M P ) retrieved from SAR inversion of different band profiles, the surface penetration depth (ξ C , ξ L , ξ P ) and the proportionality factors (K C , K L , K P ) of the integral of soil moisture to depth, determine the analytical model of profile soil moisture in the vegetation root zone of the observation area, and calculate the profile soil water content according to the model. The specific steps include:

[0122] Combine the soil moisture retrieved from multi - band SAR inversion, the SAR surface penetration depth and the proportionality factor to construct a multivariate observation equation system, and determine the coefficients a, b, d of the analytical model of profile soil moisture information as follows:

[0123]

[0124] where M C (i,j) , M L (i,j) , M P (i,j) are the soil moisture inversion results of C, L, P - band SAR at pixel (i, j); ξ C (i,j) , ξ L (i,j) , ξ P (i,j) are the surface penetration depth estimation results of C, L, P - band SAR at pixel (i, j); K C , K L , K P are the proportionality factors of the integral of profile soil moisture to depth in different bands, which are independent of the pixel position (i, j); a (i,j) , b (i,j) , d (i,j) are the polynomial coefficients of the "moisture - depth" function model at the finally calculated pixel (i, j).

[0125] Finally, the determined analytical model of the profile soil moisture is M (i,j) = a (i,j) ξ 2 + b (i,j) ξ + d (i,j) , and the soil moisture of the vegetation root zone profile at different penetration depths can be obtained.

[0126] Therefore, by adopting the above method for retrieving the soil water content of the vegetation root zone profile by multi-band fully polarized SAR, the invention can realize the retrieval of the soil water content of the vegetation root zone profile with high resolution, and provide accurate and reliable scientific basis for the drought monitoring of precision agriculture and the soil moisture monitoring.

[0127] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that they can still modify or equivalently replace the technical solutions of the present invention, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for retrieving soil moisture content in the vegetation root zone profile of multi-band fully polarized SAR, characterized in that, It includes the following steps: S1. Collect and preprocess the time-series multi-band full-polarization SAR data of the vegetation-covered area to obtain the filtered coherence matrix T′3; S2. Use the polarization target decomposition model to accurately decouple the surface and vegetation scattering of the filtered coherence matrix T′3 obtained in step S1, and obtain the coherence matrix of the surface scattering component, the model-adaptive optimal parameters, and the backscattering coefficient of the surface scattering, which are used to construct the soil moisture inversion algorithm; S3. Add the alpha approximation model observables based on the surface scattering component obtained in step S2 to construct a time-series inversion algorithm for joint multi-band and multi-polarization SAR observations; S4. Collect passive microwave data corresponding to the time and regional scope of step S1, construct a spatio-temporal domain soil dielectric constant constraint mechanism, and embed it into the time-series inversion algorithm constructed in step S3 to obtain the high-precision soil moisture in the vegetation root zone; S5. Increase the observables by combining multi-band and multi-temporal observation data, expand the polarization target decomposition model in step S2, and construct a multi-band and multi-temporal polarization target decomposition model to invert the complex surface dielectric constant; S6. Use the complex surface dielectric constant inverted in step S5 to calculate the penetration depth of SAR in different bands in the vegetation root zone; S7. Determine the scale factor according to the measured data of the soil moisture in the surface profile, the soil moisture obtained in step S4, and the penetration depth calculated in step S6; S8. Based on steps S4, S6, and S7, calculate the inverted soil moisture, the surface penetration depth, and their scale factors of the profile SAR in different bands, determine the profile soil moisture analysis model in the vegetation root zone of the observation area, and solve the profile soil water content.

2. A method for retrieving soil moisture content in the vegetation root zone profile of multi-band fully polarized SAR, according to claim 1, wherein Step S1 includes: S11. Generate the coherence matrix T3 using the image data of the HH, VV, HV, or VH polarization modes of the polarimetric SAR remote sensing image; S12. Perform coherence filtering on T3 using the boxcar filtering algorithm to obtain the filtered coherence matrix T′3.

3. A method for retrieving soil moisture content in the vegetation root zone profile of multi-band fully polarized SAR, according to claim 1, characterized in that In step S2, the surface, even-order, and volume scattering models of the polarization target decomposition model respectively adopt the improved X-Bragg model, the improved double Fresnel scattering model, and the adaptive volume scattering model GVSM; Among them, the expression of the improved X-Bragg model is as follows: where T S_X-Bragg is the polarization coherence matrix of the improved X-Bragg model, f s is the surface scattering power coefficient, β is the Bragg scattering coefficient, and δ is the depolarization angle of the surface roughness; The expression of the improved double Fresnel scattering model is as follows: where, T XD_Fresnel is the improved double Fresnel scattering coherence matrix, f d is the even-order scattering power, and T XD11 , T XD22 , T XD12 , T XD21 , T XD33 are the elements of the even-order scattering polarization coherence matrix; The expression of the adaptive volume scattering model GVSM is as follows: where γ, ρ, and e are vegetation scattering model parameters, and e = 2<|S HV | 2 > / <|S VV | 2 >. Under the assumption of rotational symmetry in <S HH S VV * >, it can be obtained that f v is the power of the vegetation body scattering component.

4. A method for retrieving soil water content in the vegetation root zone profile of multi-band full-polarization SAR, according to claim 3, characterized in that Step S2 includes: S21. Construct the cost equations for the filtered coherent matrix T′3 and the theoretical scattering of T′3 using the improved double Fresnel scattering model and the adaptive volume scattering model GVSM. Through polarimetric target decomposition processing, solve the cost equations, eliminate the vegetation volume scattering and even scattering components, extract the coherent matrix of the surface scattering component, calculate the polarimetric coherent matrix of the improved X-Bragg model, and obtain the S′ of the surface scattering component. HH , S′ VV , S′ HV Backscattering coefficients are calculated to obtain the depolarization angle of the surface roughness. S22. Substitute the depolarization angle of the surface roughness obtained in step S21 into the polarization target decomposition model to perform the second-stage polarization target decomposition process, eliminate surface scattering and even-order scattering, extract the coherence matrix of the vegetation body scattering model, perform unitary matrix transformation, and calculate the corresponding covariance matrix C volume , and then according to C volume matrix elements, calculate the anisotropy and shape parameters of the vegetation body scattering model; S23. Substitute the depolarization angle of the surface roughness obtained in step S21 into the improved X-Bragg model, and substitute the anisotropy and shape parameters of the vegetation volume scattering model obtained in step S22 into the adaptive volume scattering model GVSM to construct the cost equation of the filtered coherence matrix T′3 and the T′3 theoretical scattering model. Through step-by-step iteration and least-squares solution, calculate the optimal co-polarization phase difference parameter; S24. Based on the polarimetric target decomposition model, according to the obtained surface roughness depolarization angle, co-polarization phase difference, vegetation anisotropy, and shape parameters, extract the HH and VV polarization backscattering coefficients of the surface scattering component and as follows: where "*" represents complex conjugate, T S_X-Bragg_11 , T S_X-Bragg_12 , T S_X-Bragg_22 are the respective elements of the surface scattering coherence matrix.

5. A method for retrieving soil water content in the vegetation root zone profile of multi-band fully polarized SAR, characterized in that, In step S3, the alpha approximation model eliminates the scattering contributions of vegetation cover and rough surface using the ratio of backscattering coefficients of adjacent time phases, and by jointly using multi-band observation data and different polarization observation information, it increases the observables, expands the observation matrix, and realizes the positive definite solution of unknown parameters.

6. A method for retrieving soil water content in the vegetation root zone profile of multi-band fully polarimetric SAR, according to claim 1 or claim 5, characterized in that The multi-band and multi-polarization SAR observation joint time series inversion algorithm includes obtaining the surface dielectric constant of each time phase using the least squares method or a look-up table, and the expression is as follows: Wherein, M HH_VV is the time series observation matrix, A HH_VV is the time series alpha coefficient vector, λ1 and λ2 represent different bands and λ1 > λ2, N represents the number of time phases of the time series observation, ε soil is the soil dielectric constant, θ is the incident angle, α HH , α VV respectively represent the horizontal polarization and vertical polarization Fresnel scattering coefficients.

7. A method for retrieving soil water content in the vegetation root zone profile of multi-band fully polarized SAR, according to claim 1, wherein Step S4 includes: S41. Collect passive microwave data within the time and area range corresponding to step S1, calculate the minimum and maximum values of the soil moisture content data product within the selected time and area range, and back-calculate the minimum and maximum values of the corresponding surface dielectric constant; S42. Substitute the minimum and maximum values of the surface dielectric constant calculated in step S41 as the solution constraints of the unknown parameters of the surface dielectric constant into the multi-band and multi-polarization SAR observation joint time series inversion algorithm constructed in step S3, solve the surface dielectric constant under high-resolution observation, and convert it into soil moisture to obtain a high-precision soil moisture inversion result.

8. A method for retrieving soil water content in the vegetation root zone profile of multi-band fully polarized SAR, according to claim 1, characterized in that In step S5, the polarization target decomposition model of multi-band and multi-time phase observation joint is as follows: Among them, [T] O1 and [T] O2 are the polarization coherence matrices observed by polarimetric SAR in two different bands or different time phases respectively. [T S_X-Bragg , [T XD_Fresnel ] , [T GV GVSM are the corresponding surface scattering, even-order scattering and volume scattering models respectively. The superscripts 1 and 2 represent the corresponding bands or time phases.​ 9. A method for retrieving soil water content in the vegetation root zone profile of multi-band fully polarized SAR, according to claim 1, characterized in that In step S7, under observations of different wavebands, the proportionality factor K SAR is calculated as follows: where K C , K L , K P are the scale factors for the C, L, and P bands respectively, and M msr_fit is the relationship between the actual depth and the measured soil moisture content.

10. A method for retrieving soil moisture content in the vegetation root zone profile of multi-band fully polarimetric SAR, characterized in that, In step S8, the profile soil moisture analysis model is: M (i,j) = a (i,j) ξ 2 + b (i,j) ξ + d (i,j) ; where, M (i,j) is the soil moisture at pixel (i, j), ξ is the penetration depth; a (i,j) , b (i,j) , d (i,j) are polynomial coefficients determined by a multivariate observation equation constructed from soil moisture, penetration depth, and scale factor retrieved from SAR in different bands.

Images