A regional soil moisture inversion method and system for inverting canopy parameters

Abstract

The present invention belongs to the field of agricultural remote sensing technology and proposes a regional soil moisture inversion method for inverting canopy parameters, comprising: S1, collecting measured data and dividing it into a training set and a test set; S2, performing polarization decomposition on SAR remote sensing images to obtain mechanistic polarization decomposition features; S3, screening polarization decomposition feature combinations that are strongly correlated with leaf area index (LAI) and plant water content (VWC); S4, training a canopy parameter inversion model based on the screened features to invert regional leaf area index (LAI) and regional plant water content (VWC); S5, calibrating water cloud model parameters; and S6, using the calibrated water cloud model to couple regional leaf area index (LAI) and regional plant water content (VWC) to invert soil moisture. The present invention overcomes the limitations of single-point measurement and optical inversion methods in obtaining regional crop canopy parameters, ensuring that the water cloud model input parameters have high spatial consistency and accuracy.

Description

Technical Field

[0001] The present invention belongs to the technical field of agricultural remote sensing, and in particular relates to a method and system for inverting canopy parameters of regional soil moisture. Background Art

[0002] Soil moisture, a crucial component of global ecosystems, serves as a bridge connecting surface water, the atmosphere, and groundwater storage, profoundly impacting ecosystem stability, functionality, and diversity. In agriculture, soil moisture in crop-covered areas is a crucial hydrological variable, directly influencing crop growth, development, and yield formation. Therefore, precise monitoring of soil moisture in crop-covered areas is crucial for agricultural monitoring, resource utilization, disaster prevention, and crop yield improvement.

[0003] Traditional soil moisture monitoring relies primarily on instruments or manual point-of-sight monitoring. While the data is highly accurate, it also comes at a high cost, making it inadequate for long-term, large-area soil moisture monitoring. While model simulations can provide information on the spatiotemporal distribution of soil moisture, these methods impose strict requirements on model input parameters and are not very practical. With the advancement of Earth observation technology, remote sensing has gradually been applied to soil moisture monitoring. Optical remote sensing offers advantages such as convenience, efficiency, high spatial resolution, and rich band information, effectively reflecting the growth status of surface crops. However, it is susceptible to weather influences and cannot provide high-quality monitoring in rainy and cloudy weather. Synthetic Aperture Radar (SAR), a type of active microwave remote sensing, utilizes a different imaging principle and penetration capability than optical remote sensing, potentially addressing the limitations of optical remote sensing. Changes in soil moisture content alter the dielectric properties of the soil, which in turn affects the SAR backscatter coefficient. Therefore, SAR remote sensing offers significant advantages in soil moisture retrieval.

[0004] However, in areas covered by crops, the SAR backscatter coefficient is not only related to its own polarization mode, incident angle, and soil moisture, but is also affected by factors such as crop moisture content. The SAR backscatter at this time includes the direct scattering contribution from the crop canopy, the ground scattering contribution that is twice attenuated by the crop canopy, and the scattering contribution from the complex interaction between the crop and the ground surface. How to reduce the impact of crops is the key and difficulty of soil moisture inversion in crop-covered areas. In addition, soil moisture varies rapidly over time, making it difficult to directly establish a relationship with SAR remote sensing features for inversion. In comparison, crop canopy parameters vary more slowly over time. Indirectly inverting soil moisture by coupling regional crop canopy parameters can help improve the robustness and robustness of the inversion model.

[0005] The water-cloud model is one of the main models for inferring soil moisture based on crop canopy parameters. This model assumes the crop canopy is a "water-cloud" model, focusing only on direct scattering between the crop and soil layers while ignoring multiple scattering between the crop and soil. Crop canopy parameters such as the Leaf Area Index (LAI) and Vegetation Water Content (VWC) are key input parameters. Currently, canopy parameters such as LAI and VWC are primarily obtained through spatial interpolation or optical image inversion based on single-point field measurements. Due to the influence of crop variety, soil properties, and field management practices, crop canopy parameters exhibit significant spatial variability, making spatial interpolation methods incapable of accurately obtaining the true value of crop canopy parameters for each pixel. In areas with heavy cloud cover and rain, optical image quality is poor, leading to large errors in regional crop canopy parameter inversion based on optical data.

[0006] In addition, the water cloud model's description of crop structure and stratification is too simple and its mechanism is weak. It is urgent to introduce new SAR remote sensing mechanism characteristics to improve the accuracy of soil moisture inversion. Summary of the Invention

[0007] In response to the problems in the background technology, the present invention proposes a regional soil moisture inversion method for inverting canopy parameters.

[0008] Radar polarimetric eigenvalues ​​can provide information about the scattering properties of ground objects, which is correlated with soil moisture. The present invention first performs polarimetric decomposition of SAR remote sensing data based on polarimetric decomposition technology and selects polarimetric features that are mechanistically and physically meaningful and can best reflect crop plant characteristics. These selected polarimetric features are then input into a random forest algorithm to invert regional canopy parameters. The inverted crop canopy parameters are then substituted into a water-cloud model to indirectly invert regional soil moisture.

[0009] The invention provides a regional soil moisture inversion method for inverting canopy parameters, comprising: S1, collecting measured data and dividing the data into a training set and a test set, wherein the measured data include: SAR backscattering coefficient, leaf area index LAI and plant water content VWC of the measured sample points; S2, analyzing the microwave scattering characteristics of crops in the study area, and performing polarization decomposition on the SAR remote sensing image to obtain polarization decomposition features with mechanism: SAR backscattering coefficient components and radar polarization feature components; S3, random forest canopy parameter inversion: based on the random forest algorithm, combining ground sample point data, performing feature importance evaluation on the SAR backscattering coefficient components and the radar polarization feature components, screening the components related to leaf area index LAI, plant water content VWC, and crop microwave scattering characteristics; Polarization decomposition feature combination with strong correlation with plant water content VWC; S4, based on the screened polarization decomposition feature combination, based on the machine learning algorithm, training the canopy parameter inversion model to obtain the regional leaf area index LAI and the regional plant water content VWC; S5, water cloud model parameter calibration: the SAR backscattering coefficient, leaf area index LAI and plant water content VWC of the measured sample points in S1 are substituted into the water cloud model, and the parameters of the water cloud model are calibrated in combination with the particle swarm optimization algorithm and the trust region reflection algorithm; S6, using the calibrated water cloud model, coupling the regional leaf area index LAI obtained in S4 to invert soil moisture, and coupling the regional plant water content VWC to invert soil moisture.

[0010] The present invention also correspondingly proposes a regional soil moisture inversion system for inverting canopy parameters, comprising: a computer executable program, which completes the steps of the method described above when the program is executed.

[0011] The present invention has the following beneficial effects. On the one hand, due to the rapid temporal variations in soil moisture, soil moisture inversion models established using machine learning models that directly construct the relationship between soil moisture and SAR scattering components suffer from poor robustness. To address this issue, the present invention selects relatively stable crop canopy parameters as input parameters for a semi-empirical inversion model, the water cloud model. This method not only improves the model's robustness but also enhances its adaptability to varying soil conditions. Furthermore, from a mechanistic perspective, the relationship between crop canopy parameters and radar polarization characteristics is relatively clear, which can improve the mechanistic rationality of the semi-empirical soil moisture inversion model. On the other hand, obtaining the input parameters for the water cloud model on a regional scale is challenging. Currently, canopy parameters such as leaf area index (LAI) and plant water content (VWC) are primarily obtained through spatial interpolation or optical image inversion based on single-point field measurements. Crop canopy parameters exhibit significant spatial variability due to the influence of crop variety, soil properties, and field management practices. Spatial interpolation methods cannot accurately obtain the true values ​​of crop canopy parameters for each pixel. However, in areas covered by heavy cloud and rain, optical image quality is poor, leading to large errors in regional crop canopy parameter inversion based on optical data. To address this issue, the present invention utilizes a random forest model combined with radar polarimetric decomposition technology to effectively obtain regional crop canopy parameters such as leaf area index (LAI) and vegetation water content (VWC). This method overcomes the limitations of single-point measurement and optical inversion methods in obtaining regional crop canopy parameters, ensuring high spatial consistency and accuracy of water cloud model input parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to make the present invention more easily understood, the present invention will be described in more detail with reference to the specific embodiments shown in the accompanying drawings. These drawings only depict typical embodiments of the present invention and should not be considered as limiting the scope of protection of the present invention.

[0013] Figure 1 is a flow chart of the method of the present invention.

[0014] Figure 2 A technical roadmap for an embodiment of the method of the present invention.

[0015] Figure 3 This is a schematic diagram of the random forest model.

[0016] Figure 4 This is an overview of the study area to which the method of the present invention is applied.

[0017] Figure 5 Schematic diagram of the alpha characteristic component obtained after polarization decomposition of SAR remote sensing images.

[0018] Figure 6Schematic diagram of the Freeman_Vol characteristic component obtained after polarization decomposition of SAR remote sensing images.

[0019] Figure 7 This is the distribution map of ground sampling points.

[0020] Figure 8 The results of random forest feature importance evaluation for LAI are shown.

[0021] Figure 9 The results of random forest feature importance evaluation for VWC are shown.

[0022] Figure 10 The inversion accuracy of the training set for LAI inversion is shown.

[0023] Figure 11 The inversion accuracy of the test set for LAI inversion is shown.

[0024] Figure 12 The inversion accuracy of the training set for VWC inversion is shown.

[0025] Figure 13 The inversion accuracy of the test set for VWC inversion is shown.

[0026] Figure 14 The soil moisture training set inversion accuracy of LAI using coupled radar polarimetric signatures is shown.

[0027] Figure 15 The soil moisture training set inversion accuracy of VWC using coupled radar polarimetric signature inversion is shown.

[0028] Figure 16 The regional soil moisture retrieval results of LAI coupled with radar polarimetric signature retrieval are shown.

[0029] Figure 17 The regional soil moisture inversion results of VWC coupled with radar polarimetric signature inversion are shown.

[0030] Figure 18 The soil moisture test set inversion accuracy of LAI coupled with radar polarimetric signature is shown.

[0031] Figure 19 The inversion accuracy of the soil moisture test set of VWC coupled radar polarimetric signature inversion is shown. DETAILED DESCRIPTION

[0032] The following describes the embodiments of the present invention with reference to the accompanying drawings so that those skilled in the art can better understand the present invention and implement it. However, the enumerated embodiments are not intended to limit the present invention. Unless there is a conflict, the following embodiments and the technical features in the embodiments can be combined with each other, wherein the same components are represented by the same figure marks.

[0033] Refer to the following Figure 1 An embodiment of the method of the present invention will be described.

[0034] S1, collect measured data, which are divided into training set and test set. The measured data include: SAR backscatter coefficient, leaf area index LAI and plant water content VWC of the measured sample points.

[0035] S2, analyze the microwave scattering characteristics of crops in the study area, and perform polarization decomposition of SAR remote sensing images to obtain mechanistic polarization decomposition characteristics: SAR backscattering coefficient component and radar polarization characteristic component.

[0036] S3, Random Forest Canopy Parameter Inversion: Based on the random forest algorithm and combined with ground sample data, the feature importance of the SAR backscatter coefficient component and the radar polarization feature component is evaluated to screen the polarization decomposition feature combination with the strongest correlation with leaf area index (LAI) and plant water content (VWC).

[0037] S4. Based on the selected polarization decomposition feature combinations and a machine learning algorithm, the measured sample dataset for each period is divided into a 3:5 ratio to serve as the training and validation sets for canopy parameter inversion. The canopy parameter inversion model is trained using the training set and verified using the validation set. The crop canopy parameter inversion results from the validation set are then fed into the water-cloud model for the next step, and the dataset is further divided into the training and test sets for the soil moisture inversion model. The regional leaf area index (LAI) and regional plant water content (VWC) are ultimately obtained.

[0038] S5, water cloud model parameter calibration: the SAR backscatter coefficient, leaf area index LAI and plant water content VWC of the measured sample points in S1 are substituted into the water cloud model, and the parameters of the water cloud model are calibrated by combining the particle swarm optimization algorithm and the trust region reflectance algorithm.

[0039] S6, using the calibrated water cloud model, coupled with the regional leaf area index LAI obtained in S4 to invert soil moisture, and coupled with the regional plant water content VWC to invert soil moisture.

[0040] Preferably, step S2 further includes pre-processing the fully polarimetric SAR remote sensing image: radiometric calibration, multi-look processing, speckle filtering, geocoding, and decibel conversion.

[0041] The polarization decomposition in step S2 in a preferred embodiment is described below.

[0042] Polarimetric decomposition essentially decomposes complex polarimetric scattering signals into basic scattering mechanisms. Polarimetric parameters extracted using different polarimetric decomposition methods can characterize the scattering characteristics of ground objects and reveal hidden ground object information in SAR data. Basic scattering mechanisms include odd-order scattering (surface scattering), even-order scattering (dihedral scattering), and volume scattering. Polarimetric decomposition, derived from the polarimetric target decomposition theorem, simplifies complex polarimetric radar echo signals into easily understood parameters by decomposing the target scattering matrix, thereby extracting characteristic information about the target ground object.

[0043] Polarization decomposition methods are mainly divided into two categories: one is coherent target decomposition based on the scattering matrix, such as Krogager decomposition, Pauli decomposition, and Cameron decomposition; the other is incoherent target decomposition based on the coherence matrix or covariance matrix, such as H / A / α decomposition, Freeman-Durden three-component scattering decomposition, Yamaguchi four-component scattering decomposition, etc. Considering that the crop-covered areas in the study area are distributed radar targets, the present invention preferably uses H / A / α decomposition and Freeman-Durden three-component scattering decomposition methods from incoherent target decomposition methods to perform polarization decomposition of SAR remote sensing images.

[0044] In the first embodiment, the polarization decomposition adopts H / A / α decomposition.

[0045] H / A / α decomposition is a commonly used polarimetric decomposition method. For preprocessed SAR remote sensing images, select H / A / Alpha Decomposition under Process in the PolSARpro software toolbar, select DecompositionParameters, and set the input and output paths in the dialog box. Select the desired polarimetric decomposition parameters. For Radarsat-2 fully polarimetric data, select Alpha, Entropy (H), and Anisotropy (A). Click Run to execute the H / A / α polarimetric decomposition program. The polarimetric decomposition results will be stored as an HDR file in the specified output path.

[0046] Open the HDR files corresponding to the polarization decomposition characteristic parameters in ENVI software and convert them into TIFF files. Import the TIFF files into ArcMap software and use the Extract Multivalued to Point tool in the Spatial Analysis Toolbox to match the polarization decomposition characteristic parameters with the measured sample points using their longitude and latitude coordinates. This will obtain the values ​​of the polarization decomposition characteristic parameters corresponding to each measured sample point, facilitating subsequent model training and parameter inversion.

[0047] The principle of H / A / α decomposition is shown in Equation (1). Cloude and Pottier analyzed the eigenvalues ​​and eigenvectors of the SAR coherence matrix T3 to obtain parameters describing different scattering processes and their respective scattering proportions, mainly including polarization information entropy (H), calculated using Equation (2), anisotropy (A), calculated using Equation (3), and polarization scattering angle (α), thereby understanding the scattering mechanism and physical characteristics of the target object.

[0048] (1),

[0049] (2),

[0050] (3),

[0051] Where, is the coherence matrix, is the characteristic matrix of rank 1 obtained by decomposing the coherence matrix, is the feature matrix The corresponding eigenvalue represents the scattering intensity corresponding to the corresponding eigenvector. is the feature matrix The corresponding eigenvector represents the polarization scattering characteristics corresponding to the corresponding eigenvalue. is the eigenvalue The pseudo probability obtained is, , N is the base of the logarithm, N is 3 for single-station radar, N is 4 for dual-station radar, and The eigenvalues ​​are obtained by polarization decomposition, and the eigenvalues ​​are sorted as follows: .

[0052] Through H / A / α decomposition, the polarization information entropy (H), anisotropy (A) and polarization scattering angle (α) are obtained.

[0053] Polarization information entropy (H) reflects the statistical randomness of various scattering mechanisms, and its value ranges from 0 to 1. A low polarization entropy (H < 0.3) indicates that the target's scattering mechanism is deterministic, and the dominant scattering mechanism can be used as the scattering mechanism for the target. When the polarization entropy H is close to 1, it indicates that the target's scattering mechanism is highly random, usually corresponding to a complex scattering model.

[0054] Anisotropy (A) describes the relationship between the two relatively weak scattering components obtained from polarization eigendecomposition and ranges from 0 to 1. Anisotropy A complements polarization information entropy H and is typically used to identify scattering mechanisms only when polarization entropy H is greater than 0.7. A values ​​close to 0 indicate that the scattering intensities of the target's second and third scattering mechanisms are similar; values ​​close to 1 indicate that the target's second scattering mechanism is overwhelmingly dominant over the third scattering mechanism.

[0055] The polarization scattering angle (α) is a key parameter for identifying the dominant scattering mechanism, and its value ranges from 0° to 90°. The scattering angle α corresponds to the variation in scattering mechanism: 0° indicates surface scattering, 45° indicates volume scattering, and 90° indicates even scattering.

[0056] Second embodiment - Freeman-Durden three-component decomposition.

[0057] The Freeman-Durden three-component decomposition is a polarimetric decomposition method for fully polarimetric data. For pre-processed SAR remote sensing images, click Process in the PolSARpro software toolbar, select the Polarimetric Decompositions tool, and run the FRE3: Freeman 3 Component Decomposition program. The polarimetric decomposition results are stored in the set output path as HDR files. Similar to the H / A / α decomposition, the polarimetric decomposition characteristic parameters are converted into TIFF files in ENVI software and matched with the measured sample points in ArcMap software. Freeman and Durden proposed the Freeman-Durden three-component decomposition method, which decomposes the polarimetric covariance matrix into three basic scattering mechanisms: the first-order Bragg surface scattering model describing rough surface scattering, the even-order scattering model simulating dihedral reflectors, and the volume scattering model. This model is based on the scattering symmetry assumption and requires that there is no correlation between the co-polarization and cross-polarization of the target object, that is, it satisfies Equation (4). The calculation method of the Freeman-Durden decomposition is shown in Equations (5) to (8).

[0058] (4),

[0059] (5),

[0060] (6),

[0061] (7),

[0062] (8),

[0063] Where, and is the SAR backscatter coefficient of the same polarization channel, and is the SAR backscatter coefficient of the cross-polarization channel, is the surface scattering weighting coefficient, is the surface scattering covariance matrix, is the even scattering weighting coefficient, is the even scattering covariance matrix, is the volume scattering weighting coefficient, is the volume scattering covariance matrix, is the even-order scattering parameter, is the surface scattering parameter.

[0064] The random forest algorithm of step S3 is described below.

[0065] In S3, when selecting the polarization decomposition feature combinations with the strongest correlations with leaf area index (LAI) and plant water content (VWC), the feature importance scores of the inputs were calculated using the feature importance assessment method in the random forest model, namely, reducing Gini impurity or reducing mean square error. Based on an empirical partitioning method, the feature importance score threshold was set at 0.01. Eigenvalues ​​with feature importance scores of 0.01 or above in the random forest feature importance assessment were selected as the eigenvalues ​​for subsequent canopy parameter inversion.

[0066] The random forest model is an ensemble learning method based on the combination of multiple decision trees. It improves the robustness and accuracy of the random forest model regression by integrating the prediction results of multiple decision trees. It also has great advantages in processing high-dimensional data, evaluating feature importance, and reducing overfitting.

[0067] The random forest model structure is as follows Figure 3As shown in the figure, the random forest model divides the measured sample dataset into a training set and a test set for model training and evaluation. The random forest model uses the bootstrap aggregating method (bagging) to reduce generalization error and improve the model's regression accuracy. The random forest model first uses the bootstrap sampling method to generate decision trees. This method involves uniformly sampling samples with replacement from the original dataset to form multiple training sets, with a separate decision tree constructed for each training set. The nodes of the decision tree represent subsets of the dataset. During the decision tree construction process, the model randomly selects some features rather than all features to find the optimal split point at each node. This increases the diversity of the decision tree, avoids the model's dependence on a single feature and the risk of overfitting, and reduces the model's sensitivity to noise and outliers, thereby improving the model's generalization and robustness. Target parameter inversion is a regression problem. The random forest model trains on randomly selected feature subsets and data samples separately, performs regression analysis, and then uses a combiner to average the predictions of all decision trees to output the final predicted value of the target parameter. In addition, the model is verified and optimized by combining cross-validation, accuracy assessment, model optimization and other steps to achieve accurate and efficient inversion of soil moisture.

[0068] The process of inverting the target parameters using the random forest model is to use the random forest to establish the mapping relationship between the target parameters and the input covariates, which can be expressed as Equations (9) to (12).

[0069] (9),

[0070] (10),

[0071] (11),

[0072] (12),

[0073] Where, is the target inversion parameter, is a function that uses the random forest model to invert canopy parameters, is the input feature dataset of the random forest model (polarization decomposition features obtained in step S2), is the inversion error term, which represents the random error or noise of the model, To include The feature vector of the feature, for dimensional real vector space, It is a feature At the new data point The final weight on It is Features in a tree For data points The weight of The first The randomized feature selection used by the trees, is the number of trees.

[0074] In one embodiment, step S3 includes: S31, dividing the measured sample points of each period into a training sample set and a validation sample set for target parameter inversion. S32, for the training sample set data, inputting the measured data and the corresponding eigenvalues ​​into a random forest model to perform feature importance evaluation. S33, based on the evaluation results, eigenvalues ​​with a feature importance score greater than or equal to 0.01 are regarded as important eigenvalues ​​and retained. S34, training and optimizing a random forest regression model through grid search and 10-fold cross validation, and applying the optimal model to the validation set to perform regional inversion and accuracy verification of the target parameters.

[0075] In step S4, the random forest model is selected to invert the crop canopy parameters LAI and VWC. According to the result of feature optimization in step S3, the measured sample data of each period is divided into a training data set and a validation data set for canopy parameter inversion according to a ratio of 3:5. For the training set, the method of ten-fold cross-validation combined with grid search is used to determine the number of decision trees, the maximum depth of the tree, the minimum number of samples required for internal node splitting, and the minimum number of samples for leaf nodes. The random forest inversion model is trained and optimized, and then applied to the validation set for the inversion of LAI and VWC to obtain the single-point inversion results of the validation set of LAI and VWC. The random forest inversion model constructed by the training set is applied to the regional LAI and VWC inversion to obtain the regional LAI and VWC inversion results.

[0076] The present invention adopts a water cloud model. In step S5, the parameters of the water cloud model are calibrated, and the calibrated water cloud model is used in S6. The water cloud model assumes that the crop is a "cloud" composed of a large number of uniformly distributed identical water particles. It only considers the volume scattering from the crop reflection and the backscattering of the ground after the double attenuation of the crop layer, and ignores other forms of scattering in the crop and soil. The variables in the model only include the radar incident angle, crop water content and soil moisture. The parameters of the water cloud model are easy to obtain and the form is simple. It can effectively separate the backscattering coefficients of the crop layer and the soil layer. The specific form is shown in Equation (13) to Equation (15). The backscattering coefficient of the soil layer and soil moisture content There is a simple linear relationship between them, see Equation (16). Therefore, the soil moisture inversion in the crop-covered area based on the water cloud model can be expressed by Equation (17).

[0077] (13),

[0078] (14),

[0079] (15),

[0080] (16),

[0081] (17),

[0082] Where, is the total backscatter coefficient of the vegetation canopy at any polarization, 、 are the transmit and receive polarization modes, is the backscattering coefficient of the crop layer, is the backscattering coefficient of the soil layer, is the crop canopy parameter. The plant water content VWC and leaf area index LAI were taken respectively. is the double-layer attenuation factor of microwave penetration into the crop layer, is the radar incident angle, is the soil moisture content in the crop-covered area, and A, B, C, and D are the water cloud model parameters that need to be calibrated.

[0083] During the calibration of the water cloud model parameters, the function of the water cloud model is defined using Equation (17). First, the canopy parameter validation set obtained by random forest inversion is divided into a training set and a test set. The training set is then divided into 10 folds using KFold. For each fold of the training set data, the particle swarm optimization algorithm (PSO) is first used to find the initial parameter combination of the water cloud model parameters. Then, the trust region reflective algorithm (TRF) is used to finely solve the initial parameter combination of the water cloud model in the local area to determine the precise parameter combination of the water cloud model. Furthermore, the water cloud model is calibrated 10 times using the training set data. The validation set error in cross-validation is compared, and the water cloud model parameter combination with the smallest error is selected as the optimal parameter combination. Regional inversion of the water cloud model and accuracy verification of the test set are then performed.

[0084] The following describes an application example of the present invention. The method of the present invention has been verified. The research area selected City A, the main rapeseed producing area in China, as the research area to verify the feasibility and applicability of the regional soil moisture inversion technology of canopy parameters using coupled radar polarization characteristics. City A has a subtropical monsoon climate with distinct four seasons and abundant rainfall. The main crop grown is rapeseed. An overview of the research area is as follows: Figure 4 shown.

[0085] The research data used are four phases of RADARSAT-2 fully polarimetric SAR remote sensing imagery covering the study area, acquired on November 30, 2023, January 17, 2024, March 5, 2024, and March 29, 2024. Launched in 2007, RADARSAT-2 is a Canadian high-resolution SAR satellite equipped with a C-band sensor. It has a 24-day revisit period and a maximum resolution of 1 meter. It includes four polarization modes: HH, HV, VH, and VV. It features a short revisit period, high spatial resolution, robust data storage capacity, and precise satellite positioning. It is currently widely used in agriculture, surveying and mapping, environmental resources, and other fields.

[0086] In this embodiment, SNAP 9.0 software is used to preprocess RADARSAT-2 fully polarimetric SAR remote sensing images. The main operation steps are as follows. (1) Radiometric calibration. The Calibrate tool in the SNAP software is used to perform radiometric calibration of the RADARSAT-2 SAR remote sensing image, and the DN value of the original image is converted into a backscattering coefficient with a clear physical meaning. (2) Multi-looking processing. The Multilooking tool in the SNAP software is used to perform multi-looking processing of the SAR remote sensing image. The number of rows and columns in the multi-looking processing is set to 2×3, that is, 2 in the range direction and 3 in the azimuth direction. Multi-looking processing helps to reduce the coherent speckle noise of the SAR remote sensing image, improve the signal-to-noise ratio, and enhance the data quality. (3) Coherent speckle filtering. The Single Product SpeckleFilter tool in the SNAP software is used to perform coherent speckle filtering. The spatial filtering of the Refined Lee adaptive window is selected to smooth the coherent speckle noise in the image. (4) Geocoding. The Range Doppler Terrain Correction tool in the SNAP software is used for geocoding. The DEM elevation data is SRTM 1 Sec HGT. Terrain correction converts SAR remote sensing images to a geographic coordinate system, allowing them to be aligned and analyzed with other geographic data. (5) Decibelization. Use the Converts bands to / from dB tool in the SNAP software to decibelize SAR remote sensing images. Decibelization converts the backscatter coefficient from a linear scale to a logarithmic scale (decibels), facilitating the subsequent application of SAR backscatter coefficient parameters.

[0087] After obtaining the backscatter coefficients of each polarization channel of the SAR remote sensing image, the PolSARpro 6.0 software is used to obtain the polarization characteristic components of the SAR remote sensing image through the HA-α decomposition and Freeman-Durden polarization decomposition methods, as shown in the schematic diagram. Figure 5 and Figure 6 As shown, Figure 5 Represents a schematic diagram of the alpha feature component, Figure 6Schematic diagram showing the Freeman_Vol characteristic components.

[0088] Obtaining ground observation data: Based on the distribution of rapeseed in the study area, 32 ground measurement sampling points were deployed in the study area. Field observation experiments were conducted from December 1 to 3, 2023, January 13 to 15, 2024, March 2 to 4, 2024, and March 29 to 31, 2024. Data from 128 ground-based samples were collected for four stages: the seedling stage, the bud stage, the flowering stage, and the silique stage. The leaf area index (LAI) was measured using the LAI-2200 C measuring instrument. The LAI-2200 C uses a "fisheye" optical sensor with five different inclination angles to measure transmitted light at five angles above and below the crop canopy. The radiation transfer model was used to calculate the leaf area index, average inclination angle, and canopy porosity of the crop canopy. The measurements were repeated multiple times to obtain the average LAI value at the sampling points. Soil moisture was measured using the TDR 350, a portable rapid three-parameter instrument for measuring soil moisture, temperature, and conductivity. Using time-domain reflectometry, the instrument measures the volumetric moisture content of the surface soil. Three measurements were repeated at each sampling point, and the average value was taken as the soil moisture value. The fresh and dry weights of the crop were used to calculate the plant volumetric moisture content. The difference between the fresh and dry weights of the crop was calculated and then divided by the fresh weight to obtain the volumetric moisture content.

[0089] Determine the training set: set the 10 sample points in each period of measured sample points that are multiples of 3 as the soil moisture inversion test set sample points, and the remaining 22 sample points as the training set sample points. In the training set, the ground sampling points are distributed as follows: Figure 7 shown.

[0090] Random forest feature importance assessment: The backscatter coefficient obtained after SAR remote sensing image preprocessing and the polarization feature component obtained after polarization decomposition are input into the random forest model as feature vectors. Feature importance assessment is performed for the inversion of LAI and VWC respectively. The feature importance score of not less than 0.01 is regarded as the feature value with high feature importance, and the subsequent random forest target parameter inversion is performed. Taking the first data on November 30, 2023 as an example, the feature importance assessment results are as follows: Figure 8 and Figure 9 As shown. Figure 8 shows the importance evaluation results of random forest features of LAI, Figure 9 The random forest feature importance evaluation results of VWC are shown. From the feature importance evaluation results, it can be concluded that the feature importance scores of all eigenvalues ​​of LAI and VWC in each period are above 0.01. After passing the feature importance test, all eigenvalues ​​are retained for subsequent crop canopy parameter inversion.

[0091] Crop canopy parameter inversion modeling and accuracy verification based on random forest:

[0092] (1) LAI inversion: Based on the feature importance evaluation results of LAI inversion, the feature values ​​with a feature importance score of not less than 0.01 are input into the random forest model. The 12 sample points with sequence numbers 1 and 3n-1 in the training set are used as the training set for LAI inversion, and the remaining 10 sample points and the 10 test sets of soil moisture inversion are used as the validation set for LAI inversion (a total of 20 sample points). Combined with 10-fold cross-validation and grid search, the inversion modeling and accuracy verification of the LAI random forest regression model are carried out. Figure 10 and Figure 11 As shown in Figure 2, the LAI inversion accuracy based on random forest is higher. Figure 10 The inversion accuracy of the training set of LAI inversion is shown. Figure 11 The inversion accuracy of the validation set for LAI inversion is shown.

[0093] (2) VWC inversion: Based on the feature importance evaluation results of VWC inversion, the feature values ​​with a feature importance score of not less than 0.01 are input into the random forest model. The training set and validation set of VWC inversion are divided in the same way as LAI inversion. Combined with 10-fold cross validation and grid search, the inversion modeling and accuracy verification of the VWC random forest regression model are carried out. Figure 12 and Figure 13 As shown in Figure 2, the VWC inversion accuracy based on random forest is higher. Figure 12 The random forest single-point inversion results of soil moisture for the training set of VWC inversion are shown. Figure 13 The random forest single-point inversion soil moisture results of the test set of VWC inversion are shown.

[0094] Water-cloud model parameter calibration: Ten sample points numbered multiples of 3 from the canopy parameter validation set retrieved from each phase of random forest inversion were used as the test set, and the remaining 10 sample points were used as the training set. The radar characteristic values ​​from the training set, the canopy parameter inversion results, and the measured soil moisture content were input into the water-cloud model. The water-cloud model parameters were calibrated using a combination of 10-fold cross-validation, particle swarm optimization, and a trust region reflective algorithm to identify the optimal parameter combination. The water-cloud model parameter calibration results are shown in Table 1.

[0095] Table 1 Parameter calibration results of water cloud model

[0096]

[0097] Regional soil moisture inversion of canopy parameters using coupled radar polarimetric characteristics: The LAI and VWC regional results from the random forest inversion were substituted into the water-cloud model after parameter calibration to obtain the regional results of soil moisture inversion from the water-cloud model. In the test set, R² and nRMSE were used as accuracy evaluation indicators for soil moisture inversion from the water-cloud model to verify the inversion accuracy. Figures 14-19 Taking the VH polarization method as an example, the regional soil moisture inversion results of canopy parameters inverted by coupling radar polarization characteristics are shown. Figures 14-19 It can be seen that the regional soil moisture inversion accuracy of VWC coupled with radar polarization characteristics inversion is relatively high, with R² reaching 0.62, and the regional soil moisture inversion accuracy of LAI coupled with radar polarization characteristics inversion is R² reaching 0.46, which improves the robustness of the inversion model compared with the random forest direct inversion of soil moisture. Figure 14-15 They represent the soil moisture inversion training set accuracy of LAI and VWC inversion of coupled radar polarization characteristics, Figure 16-17 The results of soil moisture regional inversion under VH polarization are shown when LAI and VWC are used as canopy parameters for inversion. Figure 18-19 The soil moisture inversion test set accuracy of LAI and VWC using coupled radar polarimetric signature inversion is shown respectively.

[0098] The embodiments described above are merely preferred embodiments of the present invention. The phrases "in one embodiment," "in another embodiment," "in yet another embodiment," or "in other embodiments" used in this specification may refer to one or more of the same or different embodiments of the present disclosure. Any common changes and substitutions made by those skilled in the art within the scope of the present invention are intended to be encompassed within the scope of protection of the present invention.

Claims

1. A regional soil moisture inversion method for inverting canopy parameters, characterized in that: include: S1, for the soil in the crop coverage area throughout the entire growth period, collect measured data, which are divided into training set and test set. The measured data include: SAR backscatter coefficient, surface soil moisture content, leaf area index LAI and plant water content VWC at the measured sample points; S2, analyze the microwave scattering characteristics of crops in the study area, and perform polarization decomposition of SAR remote sensing images through H / A / α decomposition or Freeman-Durden three-component decomposition to obtain mechanistic polarization decomposition characteristics: SAR backscatter coefficient component and radar polarization characteristic component; S3, Random Forest Canopy Parameter Inversion: S31, divide each period of measured sample points into a training sample set and a validation sample set for target parameter inversion; S32, for the training sample set data, input the measured data and the corresponding eigenvalues ​​into the random forest model to evaluate feature importance; S33, based on the evaluation results, eigenvalues ​​with feature importance scores greater than or equal to 0.01 are considered important and retained; S34, train and optimize the random forest regression model through grid search and 10-fold cross-validation, and apply the optimal model to the validation set for regional inversion and accuracy verification of the target parameters; S4, based on the screened polarization decomposition feature combination and a machine learning algorithm, training a canopy parameter inversion model to obtain the regional leaf area index LAI and the regional plant water content VWC; S5, Water cloud model parameter calibration: S51, divide the canopy parameter validation set obtained by random forest inversion into a training set and a test set, and divide the training set into 10 folds; S52, for each fold of the training set data, use the particle swarm optimization algorithm to find the initial parameter combination of the water cloud model parameters, and then use the trust region reflective algorithm to finely solve the water cloud model parameter combination in the local area based on the initial parameter combination to determine the precise parameter combination of the water cloud model; S53, use the training set data to calibrate the water cloud model 10 times, compare the validation set error in cross-validation, and select the water cloud model parameter combination with the smallest error as the optimal parameter combination; S6 uses the calibrated water cloud model to couple the regional leaf area index LAI obtained in S4 to invert soil moisture, and also couples the regional plant water content VWC to invert soil moisture.

2. The method according to claim 1, characterized in that In step S3, When screening the polarization decomposition feature combination with the strongest correlation with leaf area index LAI and plant water content VWC, the feature importance evaluation method in the random forest model was used to calculate the input feature importance score.

3. The method according to claim 2, characterized in that The eigenvalues ​​with feature importance scores of 0.01 and above in the random forest feature importance assessment were used as the eigenvalues ​​for subsequent canopy parameter inversion.

4. The method according to claim 1, wherein The water cloud model is: , Where, is the total backscattering coefficient of the vegetation canopy at any polarization, is plant water content VWC and leaf area index LAI, is the radar incident angle, is the soil moisture content in the crop-covered area, and A, B, C, and D are the water cloud model parameters that need to be calibrated.

5. The method according to claim 4, characterized in that Step S2 also includes: pre-processing the fully polarimetric SAR remote sensing image, including: radiometric calibration, multi-look processing, speckle filtering, geocoding and decibel conversion.

6. The method according to claim 1, characterized in that In step S4, According to the result of feature optimization in step S3, the measured sample data of each period is divided into a training data set and a validation data set for canopy parameter inversion in a ratio of 3:5; For the training set, a ten-fold cross-validation combined with grid search method was used to determine the number of decision trees, the maximum depth of the tree, the minimum number of samples required for internal node splitting, and the minimum number of samples for leaf nodes, and to train and optimize the random forest inversion model. It is then applied to the validation set to invert LAI and VWC, and the single-point inversion results of the validation set of LAI and VWC are obtained.

7. A regional soil moisture inversion system for inverting canopy parameters, characterized in that: The method comprises a computer executable program, and when the program is run, the method according to any one of claims 1 to 6 is implemented.

Images