Farmland soil moisture inversion method and system based on ratio method

Abstract

The present invention discloses a method and system for inverting farmland soil moisture based on a ratio method. First, SAR images and optical images are obtained and preprocessed; then backscattering coefficients and vegetation description parameters are extracted; a soil moisture estimation data set is constructed, and a training set and a test set are determined; a ratio equation is constructed on the training set; soil moisture and estimation accuracy are estimated based on a minimum cost function strategy; the ratio equation constructed under the condition of an effective roughness parameter is the final soil moisture estimation model; and finally, accuracy verification is performed. The method provided by the present invention uses a ratio method to separate the contribution of vegetation scattering, which is simple and computationally efficient, reduces the dependence on measured roughness parameters, and uses vegetation description parameters extracted by SAR and vegetation parameters extracted by optical data to characterize the scattering characteristics of vegetation in the ratio equation, giving play to the advantage that optical data can better characterize vegetation growth conditions, and to a certain extent improves the inversion accuracy of soil moisture.

Description

Technical Field

[0001] The invention relates to the technical field of remote sensing image data processing, in particular to a method for inverting farmland soil moisture based on a ratio method. Background Art

[0002] Soil moisture is a key link in the ecological water cycle, controlling processes such as surface runoff, surface water infiltration, and surface water evaporation. In agricultural production, a decrease in soil moisture content can cause soil cracking, trigger agricultural drought, and then cause a large-scale reduction in crop yields, threatening social food security. Therefore, real-time and accurate acquisition of soil moisture information for large areas of farmland has important guiding significance for crop yield estimation, drought and flood prediction, and agricultural production management.

[0003] Traditional soil moisture relies on field measurements, which is time-consuming and labor-intensive and can only obtain information from a small number of limited points. The emergence of remote sensing technology provides an effective way to monitor soil moisture information in real time and accurately over a large area. Microwave remote sensing has strong penetration ability and the ability to monitor all day long. It is not affected by weather conditions such as clouds and fog. Compared with optical remote sensing, surface microwave scattered radiation is more sensitive to changes in soil moisture content. Therefore, microwave remote sensing is the main means of remote sensing monitoring of soil moisture. At present, many models have been proposed for soil moisture inversion based on active microwave remote sensing, especially SAR, and good inversion results have been achieved. However, most models are established for bare soil areas and cannot be directly applied to vegetation-covered areas. This is because in vegetation-covered areas, the microwave signals emitted by SAR sensors will produce complex interactions between vegetation and the surface, so that the backscatter received in the end contains multiple forms of scattering components: from vegetation, from the surface, or from the interaction between vegetation and the surface. The key to estimating the soil moisture on the surface of vegetation-covered areas such as farmland planting areas is to separate the scattering contribution from vegetation in the signal received by the sensor and obtain the backscatter from the surface.

[0004] The SAR-based soil moisture inversion model is affected by a variety of parameters, including soil moisture, vegetation canopy description parameters, sensor configuration parameters, surface roughness parameters, etc. However, in areas with dense vegetation, some parameters are difficult to measure and determine, such as roughness parameters and vegetation canopy description parameters. For areas where some measured parameters are missing, the use of the soil moisture inversion model will be limited. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a method for inverting farmland soil moisture based on the ratio method, which solves the above-mentioned problems that may exist in soil moisture inversion based on SAR, and uses the ratio method to separate the vegetation scattering component from the total SAR signal observed by the sensor, thereby eliminating the scattering contribution of vegetation and further estimating the soil moisture content.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] The method for inverting farmland soil moisture based on the ratio method provided by the present invention comprises the following steps:

[0008] Acquire SAR images and optical images and perform preprocessing, wherein the image data covers images of planted plants and soil;

[0009] Extract backscatter coefficients and vegetation description parameters from preprocessed SAR images and optical images;

[0010] Construct a soil moisture estimation dataset and determine the training set and test set;

[0011] On the training set, a surface scattering model with different roughness conditions was constructed, and the bare soil backscattering coefficient was simulated through the surface scattering model; a ratio equation was constructed by combining the observed total backscattering coefficient, the extracted vegetation description parameters, and the simulated bare soil backscattering coefficient; the bare soil backscattering coefficient estimated on the training set was calculated, and based on the minimum cost function strategy, the soil moisture was inverted from the lookup table constructed by the surface scattering model under the current roughness conditions; the root mean square error between the estimated soil moisture and the measured soil moisture was calculated as the estimation accuracy of the soil moisture in the training set;

[0012] Traverse the roughness parameters within the given roughness range, repeat the above steps, and take the roughness parameter with the highest estimation accuracy as the effective roughness parameter in the study area; the ratio equation constructed under the effective roughness parameter condition is the final soil moisture estimation model;

[0013] A surface scattering model lookup table is constructed based on the effective roughness parameters. Then, the soil moisture of the test set is inverted from the lookup table using the minimum cost function inversion strategy. The soil moisture is then compared with the measured soil moisture content of the test set for accuracy verification.

[0014] Further, the ratio equation is established according to the following formula:

[0015]

[0016] in, represents the backscattering coefficient of the bare soil surface, represents the total backscatter coefficient, pp represents the HH or VV polarization mode, and F represents the ratio function, which is as follows:

[0017] F=aV+bV c ;

[0018] Among them, V represents the vegetation description parameter, a, b, and c are the unknown coefficients of the ratio function;

[0019] The data of the training set is combined with the least squares method to find the unknown coefficients of the ratio function.

[0020] Further, the extraction of the backscatter coefficient is performed according to the following steps:

[0021] According to the latitude and longitude information of the sampling point, the backscatter coefficients of four polarization modes of the sampling point are extracted from the preprocessed SAR image, and the four polarization modes include HH / VV / HV / VH;

[0022] The extracted backscattering coefficient is normalized by the incident angle, and the calculation method is as follows:

[0023]

[0024] Where θ is the local angle of incidence, θ ref is the reference incident angle;

[0025] The reflectance information of the red light band and the near-infrared band is extracted from the multispectral data, and then the two vegetation indices of the sampling points are calculated. The RVI and NDVI are calculated according to the following formulas, and RVI and NDVI are used as vegetation description parameters in the ratio equation:

[0026]

[0027] in, and is the backscattering coefficient of different polarization modes;

[0028]

[0029] Among them, NIR and R represent the surface reflectivity in the near-infrared band and the red light band, respectively.

[0030] Furthermore, the training set and the test set are prepared in the following manner:

[0031] The extracted total backscatter coefficient, vegetation parameters and measured soil moisture are used together to construct a soil moisture estimation dataset, which is then divided into a training set and a test set according to a preset ratio.

[0032] Furthermore, the bare soil backscatter coefficient is calculated using the CIEM surface scattering model, and the empirical relationship between the correlation length and the root mean square height in the CIEM model is calculated according to the following formula:

[0033] L(s,θ,pp)=a(sinθ) b s (cθ+d) ;

[0034] Where L is the surface correlation length, s is the RMS height, pp represents the HH or VV polarization mode, θ represents the local incident angle, and a, b, c, and d are unknown coefficients.

[0035] Furthermore, the minimum cost function inversion strategy is calculated according to the following formula:

[0036]

[0037] in, and represent the estimated backscattering coefficients of bare soil surface under HH polarization and VV polarization respectively; and They respectively represent the backscattering coefficients of the bare soil surface under HH polarization and VV polarization modes simulated by the CIEM model.

[0038] Further, the soil moisture of the test set is calculated according to the following steps:

[0039] The estimated bare soil backscattering coefficient is calculated using the constructed final ratio equation. Then, based on the bare soil backscattering coefficient simulated by the CIEM model under the effective roughness parameters and combined with the minimum cost function strategy, the soil moisture on the test set is inverted and finally the accuracy evaluation is completed in combination with the measured soil moisture on the test set.

[0040] Furthermore, it also includes regional estimation and mapping of soil moisture content. The specific steps are as follows:

[0041] For each sampling date, the effective roughness parameter is used to construct the backscatter coefficient lookup table of the CIEM model. Then, all pixels in the planting area in the remote sensing data are traversed, and the soil moisture information of each pixel is calculated by the soil moisture estimation method. Finally, the soil moisture estimation is completed and the soil moisture regional mapping is carried out.

[0042] The farmland soil moisture inversion system based on the ratio method provided by the present invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the steps of the above method are implemented when the processor executes the program.

[0043] The beneficial effects of the present invention are:

[0044] The soil moisture content method based on SAR and ratio method provided by the present invention uses ratio method to separate vegetation scattering contribution, which is simple and efficient in calculation. The effective roughness parameter is used to parameterize the surface scattering model, reducing the dependence on the measured roughness parameter. In addition to using vegetation description parameters extracted by SAR, vegetation parameters are also extracted with the help of optical data to characterize the scattering characteristics of vegetation in the ratio equation, giving play to the advantage that optical data can better characterize vegetation growth conditions, and improving the inversion accuracy of soil moisture to a certain extent.

[0045] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to make the purpose, technical solution and beneficial effects of the present invention clearer, the present invention provides the following drawings for illustration:

[0047] Figure 1 This is a schematic diagram of a wheat research area in a specific embodiment of the present invention. The selected research area is a wheat growing area in southeastern Ontario, Canada.

[0048] Figure 2 It is a flow chart of the overall solution of the present invention.

[0049] Figure 3 In a specific embodiment of the present invention, two different vegetation parameters are used to parameterize the estimation accuracy of the ratio equation on the test set.

[0050] Figure 4 Regional mapping of soil moisture content in wheat-growing areas at different sampling dates obtained in a specific embodiment of the present invention. DETAILED DESCRIPTION

[0051] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

[0052] Example 1

[0053] like Figure 1 As shown, the farmland soil moisture inversion method based on the ratio method provided in this embodiment includes the following steps:

[0054] SAR images and optical images are acquired and preprocessed, wherein the image data covers images of planted plants and soil.

[0055] The backscatter coefficients (VV, HH, HV and VH) and vegetation description parameters (RVI and NDVI) were extracted from the preprocessed SAR images and optical images, where RVI was obtained from the SAR images and NDVI was obtained from the optical images.

[0056] Construct a soil moisture estimation dataset and determine the training set and test set.

[0057] On the training set, a surface scattering model with different roughness conditions is constructed, and the bare soil backscattering coefficient is simulated by the surface scattering model. The ratio equation is constructed by combining the observed total backscattering coefficient, the extracted vegetation description parameters, and the simulated bare soil backscattering coefficient, and the undetermined coefficients of the ratio equation are obtained using the least squares method. After the undetermined coefficients of the ratio equation are obtained, the bare soil backscattering coefficient estimated on the training set is calculated, and the soil moisture is inverted from the lookup table constructed by the surface scattering model under the current roughness conditions based on the minimum cost function strategy. The root mean square error (RMSE) between the estimated soil moisture and the measured soil moisture is calculated as the estimation accuracy of the soil moisture in the training set.

[0058] Traverse the roughness parameters within the given roughness range, repeat the above steps, and take the roughness parameter with the highest estimation accuracy as the effective roughness parameter in the study area. The ratio equation constructed under the effective roughness parameter condition is the final soil moisture estimation model.

[0059] A surface scattering model lookup table is constructed based on the effective roughness parameters. Then, the soil moisture of the test set is inverted from the lookup table using the minimum cost function inversion strategy. The soil moisture is then compared with the measured soil moisture content of the test set for accuracy verification.

[0060] Example 2

[0061] The soil moisture estimation method based on SAR and ratio method provided by the present invention is further described below in conjunction with specific embodiments and drawings of the specification:

[0062] Data acquisition: The remote sensing data used in this method are RADARSAT-2 fully polarized SAR data and Sentinel-2 multispectral data. The measured data include soil moisture information of some sampling points in the wheat planting area. Figure 1 The SAR data and the measured data were obtained on the same date, covering eight dates during the wheat growth period. The specific dates are as follows: Figure 4 The optical data acquisition date was as close to the sampling date as possible, and there was no cloud coverage in the study area within the image. Finally, there were 6 sampling dates with optical images that met the requirements (excluding May 9 and May 20).

[0063] like Figure 2As shown, the method for estimating soil moisture content comprises the following steps:

[0064] Step 1: Preprocessing of remote sensing images, mainly including radiation calibration, speckle filtering, terrain correction and projection conversion.

[0065] Step 2: According to the latitude and longitude information of the sampling point, the backscatter coefficients of the four polarization modes (HH / VV / HV / VH) of the sampling point are extracted from the preprocessed SAR image.

[0066] Among them, HH means horizontal transmission and horizontal reception of electromagnetic waves; VV means vertical transmission and vertical reception of electromagnetic waves; HV means horizontal transmission and vertical reception of electromagnetic waves; VH means vertical transmission and horizontal reception of electromagnetic waves.

[0067] The local incident angle of SAR usually affects the surface backscatter coefficient, which cannot be ignored in soil moisture estimation. In order to limit the influence of different local incident angles, the extracted backscatter coefficient is normalized by the incident angle. The calculation method is as follows:

[0068]

[0069] Where θ is the local angle of incidence, θ ref is the reference incident angle. ref Set it to 30°, and fix the incident angle in the surface scattering model to 30°.

[0070] The reflectance information of the red band (Red) and the near infrared band (NIR) is extracted from the Sentinel-2 image, and then two vegetation indices of the sampling point are calculated, RVI and NDVI as vegetation description parameters in the ratio equation. The calculation method of RVI and NDVI is as follows:

[0071]

[0072] in, and is the backscattering coefficient of different polarization modes.

[0073]

[0074] Among them, NIR and R represent the surface reflectivity in the near-infrared band and the red light band, respectively.

[0075] The extracted total backscatter coefficient, vegetation parameters and measured soil moisture are used together to construct a soil moisture estimation dataset, which is then divided into a training set and a test set in a ratio of 7:3.

[0076] Step 3: Use the CIEM surface scattering model to simulate the backscattering coefficient of bare soil. The IEM model constructs the functional relationship between the co-polarization backscattering coefficient and the surface roughness parameters (root mean square height, correlation length), incident angle, incident wavelength, and soil dielectric constant, which can be expressed as Formula 4:

[0077]

[0078] in:

[0079]

[0080]

[0081]

[0082] Where s represents the root mean square height of the bare soil surface; k represents the radar wave number, which is calculated as k = 2π / λ, where λ is the incident wavelength; θ represents the incident angle; represents the simulated backscattering coefficient; f pp and F pp represent Kirchhoff coefficient and compensation field coefficient respectively, pp represents VV polarization or HH polarization; R ⊥ and R || represents the Fresnel reflection coefficient; μ represents the magnetic permeability, ε represents the soil dielectric constant; Re represents the real part of the complex number, and "*" represents the complex conjugate form. W n is the nth-order roughness spectrum of the surface autocorrelation function, which is a function related to the correlation length L.

[0083] Baghdadi et al. constructed a functional expression between the root mean square height and the correlation length based on a large amount of roughness data and C-band SAR data from field surveys, combined with the simulation values ​​of the IEM model, and proposed the CIEM model, which reduces the dependence on the correlation length or root mean square height. The empirical relationship between the correlation length and the root mean square height in the CIEM model established for C-band SAR can be expressed as:

[0084] L(s,θ,pp)=a(sinθ) b s (cθ+d) (8)

[0085] Where L is the surface correlation length, θ is the incident angle, s is the root mean square height, pp represents the HH or VV polarization mode, a, b, c, d are unknown coefficients, and their values ​​are as follows depending on the polarization mode:

[0086] a hh =4.026,d hh =1.551,a vv =3.289,d vv=1.222

[0087] b hh =b vv =-1.774,c hh =c vv =-0.0025 (9)

[0088] Among them, a hh and a vv They represent the value of a in formula (8) when the polarization mode is HH and VV respectively; b hh and b vv They represent the value of b in formula (8) when the polarization mode is HH and VV respectively; c hh and c vv They represent the value of c in formula (8) when the polarization mode is HH and VV respectively; d hh and d vv They represent the value of d in formula (8) when the polarization modes are HH and VV respectively.

[0089] Given a roughness parameter range, for this example, the range of the root mean square height s is set to 1mm-30mm, with a step size of 0.1mm. Then, the surface scattering model is constructed using the roughness parameters within the given range to simulate the backscattering coefficient of bare soil in the wheat planting area. Then, the ratio equation is constructed by combining the observed total backscattering coefficient and the vegetation description parameters extracted from the remote sensing image. The ratio equation is expressed as follows:

[0090]

[0091] Where pp represents HH or VV polarization mode, represents the backscattering coefficient of the bare soil surface, Represents the total backscatter coefficient. The ratio function F on the right side of the equation uses the combination of power function and linear function proposed by Bai et al.:

[0092] F=aV+bV c (11)

[0093] Among them, V represents the vegetation description parameter. In this example, V is parameterized using RVI and NDVI, and parameters such as a, b, and c are unknown coefficients of the ratio function.

[0094] The data of the training set is combined with the least squares method to find the unknown coefficient of the ratio function. Then the bare soil backscattering coefficient estimated on the training set is obtained. Then, combined with the lookup table constructed by the CIEM model under the current roughness conditions, the estimated soil moisture on the training set is inverted based on the minimum cost function strategy (Formula 12). It is further compared with the measured soil moisture on the training set, and the RMSE of the estimated value and the measured value is calculated as the estimation accuracy on the training set.

[0095]

[0096] in, and represent the estimated backscattering coefficients of bare soil surface under HH polarization and VV polarization respectively; and They respectively represent the backscattering coefficients of the bare soil surface under HH polarization and VV polarization modes simulated by the CIEM model.

[0097] Step 4: Traverse all RMS heights (s) within the given roughness parameter interval and repeat step 3 until s reaches the maximum value in the given range. Compare the estimation accuracy of soil moisture in the training set under different s, and take the s with the highest estimation accuracy as the effective roughness parameter of the study area at this time phase. The ratio equation constructed under the effective roughness parameter is used as the final soil moisture inversion model.

[0098] Step 5: On the test set, the estimated bare soil backscattering coefficient is calculated using the constructed final ratio equation. Then, based on the bare soil backscattering coefficient simulated by the CIEM model under the effective roughness parameter, combined with the minimum cost function strategy, the soil moisture on the test set is inverted. Finally, the accuracy evaluation is completed in combination with the measured soil moisture on the test set (such as Figure 3 shown).

[0099] Step 6: Regional estimation and mapping of soil moisture content in wheat farmland. For each sampling date, the backscatter coefficient lookup table of the CIEM model is constructed using the effective roughness parameter, and then all pixels in the wheat planting area in the RADARSAT-2 image are traversed. The soil moisture estimation method in the above steps is used to estimate the soil moisture information of each pixel, and finally the soil moisture content estimation of the wheat farmland area is completed and the soil moisture regional mapping is performed. The soil moisture inversion mapping of the study area based on RVI is as follows: Figure 4 shown.

[0100] The above-described embodiments are only preferred embodiments for fully illustrating the present invention, and the protection scope of the present invention is not limited thereto. Equivalent substitutions or changes made by those skilled in the art based on the present invention are within the protection scope of the present invention. The protection scope of the present invention shall be subject to the claims.

Claims

1. The method for inverting farmland soil moisture based on the ratio method is characterized by: The following steps are involved: Acquire SAR images and optical images and perform preprocessing, wherein the image data covers images of planted plants and soil; Extract backscatter coefficients and vegetation description parameters from preprocessed SAR images and optical images; Construct a soil moisture estimation dataset and determine the training set and test set; On the training set, a surface scattering model with different roughness conditions is constructed, and the bare soil backscattering coefficient is simulated by the surface scattering model; The ratio equation is constructed by combining the observed total backscatter coefficient, the extracted vegetation description parameters, and the simulated bare soil backscatter coefficient; The bare soil backscatter coefficient estimated on the training set is calculated, and the soil moisture is inverted from the lookup table constructed by the surface scattering model under the current roughness conditions based on the minimum cost function strategy; The root mean square error between the estimated soil moisture and the measured soil moisture was calculated as the estimation accuracy of the soil moisture in the training set; Traverse the roughness parameters within the given roughness range, repeat the above steps, and take the roughness parameter with the highest estimation accuracy as the effective roughness parameter in the study area; the ratio equation constructed under the effective roughness parameter condition is the final soil moisture estimation model; A surface scattering model lookup table is constructed based on the effective roughness parameters. Then, the soil moisture of the test set is inverted from the lookup table using the minimum cost function inversion strategy. The soil moisture is then compared with the measured soil moisture content of the test set for accuracy verification.

2. The method for inverting farmland soil moisture based on the ratio method according to claim 1, characterized in that: The ratio equation is established according to the following formula: in, represents the backscattering coefficient of the bare soil surface, represents the total backscatter coefficient, pp represents the HH or VV polarization mode, and F represents the ratio function, which is as follows: F=aV+bV c ; Among them, V represents the vegetation description parameter, a, b, and c are the unknown coefficients of the ratio function; The data of the training set is combined with the least squares method to find the unknown coefficients of the ratio function.

3. The method for inverting farmland soil moisture based on the ratio method according to claim 1, characterized in that: The extraction of the backscatter coefficient is carried out according to the following steps: According to the latitude and longitude information of the sampling point, the backscatter coefficients of four polarization modes of the sampling point are extracted from the preprocessed SAR image, and the four polarization modes include HH / VV / HV / VH; The extracted backscattering coefficient is normalized by the incident angle, and the calculation method is as follows: Where θ is the local angle of incidence, θ ref is the reference incident angle; The reflectance information of the red light band and the near-infrared band is extracted from the multispectral data, and then the two vegetation indices of the sampling points are calculated. The RVI and NDVI are calculated according to the following formulas, and RVI and NDVI are used as vegetation description parameters in the ratio equation: in, and is the backscattering coefficient of different polarization modes; Among them, NIR and R represent the surface reflectivity in the near-infrared band and the red light band, respectively.

4. The method for inverting farmland soil moisture based on the ratio method according to claim 1, characterized in that: The training set and test set are prepared in the following manner: The extracted total backscatter coefficient, vegetation parameters and measured soil moisture are used together to construct a soil moisture estimation dataset, which is then divided into a training set and a test set according to a preset ratio.

5. The method for inverting farmland soil moisture based on the ratio method according to claim 1, characterized in that: The bare soil backscatter coefficient is calculated using the CIEM surface scattering model, and the empirical relationship between the correlation length and the root mean square height in the CIEM model is calculated according to the following formula: L(s,θ,pp)=a(sinθ) b s (cθ+d) ; Where L is the surface correlation length, s is the RMS height, pp represents the HH or VV polarization mode, θ represents the local incident angle, and a, b, c, and d are unknown coefficients.

6. The method for inverting farmland soil moisture based on the ratio method according to claim 1, characterized in that: The minimum cost function inversion strategy is calculated according to the following formula: in, and represent the estimated backscattering coefficients of bare soil surface under HH polarization and VV polarization respectively; and They respectively represent the backscattering coefficients of the bare soil surface under HH polarization and VV polarization modes simulated by the CIEM model.

7. The method for inverting farmland soil moisture based on the ratio method according to claim 1, characterized in that: The calculation of soil moisture of the test set is carried out according to the following steps: The estimated bare soil backscattering coefficient is calculated using the constructed final ratio equation. Then, based on the bare soil backscattering coefficient simulated by the CIEM model under the effective roughness parameters and combined with the minimum cost function strategy, the soil moisture on the test set is inverted and finally the accuracy evaluation is completed in combination with the measured soil moisture on the test set.

8. The method for inverting farmland soil moisture based on the ratio method according to claim 1, characterized in that: It also includes regional estimation and mapping of soil moisture content, and the specific steps are as follows: For each sampling date, the effective roughness parameter is used to construct the backscatter coefficient lookup table of the CIEM model. Then, all pixels in the planting area in the remote sensing data are traversed, and the soil moisture information of each pixel is calculated by the soil moisture estimation method. Finally, the soil moisture estimation is completed and the soil moisture regional mapping is carried out.

9. A farmland soil moisture inversion system based on a ratio method, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the method according to any one of claims 1 to 8 are implemented.

Images