Regional soil moisture inversion method and system based on remote sensing data and water cloud model

Abstract

The invention discloses a regional soil moisture inversion method and system based on remote sensing data and a water cloud model, and relates to the technical field of microwave radar, and the method comprises the steps: obtaining satellite remote sensing image data of a research region; extracting reflectivity information of the satellite remote sensing image data, and acquiring a spectral index by using the reflectivity information; introducing the spectral index as a vegetation description parameter into a water cloud model, performing water cloud model parameter inversion, and selecting an optimal vegetation description parameter and an optimal soil moisture content inversion model; and establishing a lookup table according to the water cloud model under the condition that the vegetation coverage is lower than a threshold value, inputting the backscattering coefficient obtained in the lookup table into the optimal soil moisture content inversion model, and performing inversion to obtain the regional soil moisture content. According to the method, training inversion is carried out on the optimal soil moisture content inversion model through the posterior scattering coefficient of the water cloud model under the condition that the vegetation coverage is lower than the threshold value, and soil moisture inversion under the condition that different vegetation coverage is lower than the threshold value is achieved.

Description

Technical Field

[0001] The invention relates to the field of microwave radar technology, and in particular to a method and system for regional soil moisture inversion based on remote sensing data and a water cloud model. Background Art

[0002] Soil moisture is an important part of the water cycle in terrestrial ecosystems. Rapidly and accurately obtaining information on the spatiotemporal changes in soil moisture is of great significance to agricultural production, drought warning and water resources management.

[0003] Traditional soil moisture monitoring methods rely on direct measurements at ground stations, which is time-consuming and labor-intensive and difficult to achieve large-scale real-time monitoring. The Sentinel series of Earth observation satellites are an important part of the Global Environment and Security Monitoring System (i.e., the Copernicus Program) jointly initiated by the European Commission and the European Space Agency. By providing high-resolution, all-weather Earth observation data around the world, they support key areas such as global environmental monitoring, climate change research, natural disaster management, and emergency response.

[0004] In this field, high-resolution observation data can be obtained through earth observation satellite observations. In actual observations, vegetation coverage has a great influence on the radar backscatter coefficient. In the prior art, although some scholars have studied soil moisture in crop-covered areas, in the research, various inversion methods currently attempt to use a single inversion to invert areas in all cases. Due to the different data in the areas under different crop coverage conditions, the inversion results of the inversion model are different, which makes soil moisture inversion face challenges, resulting in reduced accuracy and reliability of soil moisture inversion. Summary of the invention

[0005] The purpose of the present invention is to address the deficiencies of the above-mentioned prior art and to provide a regional soil moisture inversion method and system based on remote sensing data and a water cloud model, so as to solve the problem in the prior art that various inversion methods currently attempt to use a single inversion to invert the region in all cases. Due to the different data of the region under different crop coverage conditions, the inversion results of the inversion model are different, which makes soil moisture inversion face challenges and leads to reduced accuracy and reliability of soil moisture inversion.

[0006] The present invention specifically provides the following technical solutions:

[0007] A regional soil moisture inversion method based on remote sensing data and water cloud model includes the following steps:

[0008] Obtain satellite remote sensing image data of the study area;

[0009] Extracting reflectivity information of the satellite remote sensing image data, calculating the spectral index using the reflectivity information, and performing supervised classification on the satellite remote sensing image data to obtain a vegetation classification result of the study area;

[0010] The spectral index is introduced as a vegetation description parameter into a multi-class water cloud model, and the parameters of the multi-class water cloud model are inverted. The optimal vegetation description parameter and the optimal soil moisture inversion model are selected according to the inversion results; the parameters include backscattering coefficient and soil moisture content;

[0011] Based on the vegetation classification results and the optimal vegetation description parameters, a lookup table is established using a water cloud model under the condition that vegetation coverage is below a threshold, and the backscattering coefficient obtained in the lookup table is input into the optimal soil moisture inversion model to invert and obtain the regional soil moisture content.

[0012] Preferably, the spectral index is introduced as a vegetation description parameter into a multi-class water cloud model, parameter inversion is performed on the multi-class water cloud model, and the optimal vegetation description parameter and the optimal soil moisture inversion model are selected according to the inversion results, including:

[0013] The vegetation coverage FVC is introduced into the water cloud model, and the total backscatter coefficient in the mixed pixel is split into the scattering contribution of the vegetation coverage area and the scattering contribution of the bare surface. The specific expression is:

[0014]

[0015] in, is the radar backscatter coefficient received by the SAR sensor in the vegetation coverage area, is the scattering coefficient of vegetation, τ 2 is the double-layer attenuation factor of the radar passing through vegetation; is the scattering coefficient of bare soil surface;

[0016] The normalized vegetation index NDVI, normalized difference water index NDWI and leaf area index LAI in the spectral index are introduced into the water cloud model as vegetation description parameters, and the soil moisture content Mv, incident angle θ, backscattering coefficient Vegetation description parameter m veg As input parameters, the nonlinear least square method is used to invert the water cloud model parameters;

[0017] The observed backscatter coefficients of different vegetation description parameter modeling sets and the backscatter coefficients obtained by inversion simulation of the water cloud model are established, and the optimal vegetation description parameters are selected by comparing the determination coefficient and root mean square error between the observed backscatter coefficients and the backscatter coefficients obtained by inversion simulation; among them, the water cloud model includes the conventional water cloud model WCM and the water cloud model MWCM under low vegetation coverage conditions;

[0018] Based on soil moisture content Mv, incident angle θ, backscattering coefficient Vegetation description parameter m veg The input parameters are established based on WCM and MWCM models respectively. Lookup table, invert soil moisture content with different algorithms, obtain the determination coefficient and root mean square error of the inversion results of different algorithms, and select the optimal soil moisture inversion model.

[0019] Preferably, the selecting the optimal soil moisture inversion model comprises:

[0020] When m is selected veg When used as a vegetation description parameter, the soil moisture expressions of the WCM and MWCM models are:

[0021]

[0022]

[0023] Among them, M v1 is the soil moisture content of the WCM model, M v2 is the soil moisture content of the MWCM model;

[0024] When LAI is selected as the vegetation description parameter, the soil moisture expressions of the WCM and MWCM models are:

[0025]

[0026]

[0027] The optimal soil moisture inversion model is selected based on the determination coefficient of soil moisture and the root mean square error of soil moisture in the inversion results of different algorithms.

[0028] Preferably, the obtaining of satellite remote sensing image data of the study area includes:

[0029] Obtain satellite remote sensing image data of the study area, preprocess the satellite remote sensing image data, and perform registration processing on the preprocessed satellite remote sensing image data;

[0030] The satellite remote sensing image data are all downloaded from the European Space Agency's Copernicus Sentinel Science Center.

[0031] Preferably, extracting reflectivity information of the satellite remote sensing image data and calculating the spectral index using the reflectivity information includes:

[0032] The reflectivity information of satellite remote sensing image data is extracted, and the spectral index including the normalized difference vegetation index NDVI, the normalized difference water index NDWI and the leaf area index LAI is obtained by using the reflectivity information; the calculation formulas of NDVI and NDWI are as follows:

[0033]

[0034]

[0035] Among them, B4, B8, and B11 are the reflectances of the 4th, 8th, and 11th bands of Sentinel-2 in satellite remote sensing images, respectively.

[0036] Preferably, the calculating of the spectral index using the reflectivity information further includes:

[0037] The normalized difference vegetation index NDVI is used to obtain the vegetation coverage FVC. The specific expression is:

[0038]

[0039] Among them, NDVI soil is the NDVI value of the pixel without vegetation cover, NDVI veg is the NDVI value of all pixels covered by vegetation.

[0040] The present invention provides a regional soil moisture inversion system based on remote sensing data and a water cloud model, comprising:

[0041] The acquisition module is used to obtain satellite remote sensing image data of the study area;

[0042] A data extraction module is used to extract the reflectivity information of the satellite remote sensing image data, calculate the spectral index using the reflectivity information, and perform supervised classification on the satellite remote sensing image data to obtain the vegetation classification result of the study area;

[0043] A model selection module is used to introduce the spectral index as a vegetation description parameter into the multi-class water cloud model, perform parameter inversion on the multi-class water cloud model, and select the optimal vegetation description parameter and the optimal soil moisture inversion model according to the inversion results; the parameters include backscattering coefficient and soil moisture content;

[0044] The inversion module is used to establish a lookup table based on the vegetation classification results and the optimal vegetation description parameters by a water cloud model under the condition that the vegetation coverage is lower than a threshold, input the backscattering coefficient obtained in the lookup table into the optimal soil moisture inversion model, and invert to obtain the regional soil moisture content.

[0045] The present invention provides a computer device, comprising a memory and a processor, wherein a program is stored in the memory, and when the program is executed by the processor, the processor executes the steps of the above-mentioned regional soil moisture inversion method based on remote sensing data and a water cloud model.

[0046] The present invention provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned regional soil moisture inversion method based on remote sensing data and a water cloud model are implemented.

[0047] Compared with the prior art, the present invention has the following significant advantages:

[0048] The present invention aims at the inversion of regional soil moisture, obtains spectral index through reflectivity information of satellite remote sensing image data of the study area, adopts spectral index as different vegetation description parameters to invert water cloud model, selects optimal vegetation description parameters and optimal soil moisture inversion model to realize soil moisture inversion under any conditions, and for soil moisture inversion when vegetation coverage is lower than threshold conditions, the present invention establishes a lookup table through water cloud model under vegetation coverage lower than threshold conditions, trains and inverts the optimal soil moisture inversion model through backscattering coefficient in the lookup table, and obtains regional soil moisture, thereby realizing soil moisture inversion under vegetation coverage lower than threshold conditions and under normal conditions, reducing the influence of vegetation coverage on radar backscattering coefficient, and reducing the difference produced by inversion results, effectively improving the accuracy, applicability and reliability of soil moisture inversion. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is the supervised classification result map of the image using the random forest algorithm;

[0050] Figure 2 is the modeling set Sentinel-1 backscattering coefficient observation value and model simulation value; where, Figure 2 (a)~ Figure 2 (c) are the MWCM model NDVI, NDWI and LAI maps of SAR satellite under VV conditions, Figure 2 (d)~ Figure 2 (f) are the NDVI, NDWI and LAI images of the WCM model of the SAR satellite under VV conditions, Figure 2 (g)~ Figure 2 (i) MWCM model NDVI, NDWI and LAI diagrams of SAR satellite under VH conditions, Figure 2 (j)~ Figure 2 (l) NDVI, NDWI and LAI maps of WCM model of SAR satellite under VH conditions;

[0051] Figure 3 The schematic diagram of the inversion accuracy of each model, in which V and W represent the m calculated by NDVI and NDWI respectively. veg is the model vegetation description parameter; L means LAI is used as the model vegetation description parameter; where, Figure 3 (a) is a schematic diagram of the inversion accuracy of each model when the coefficient is determined. Figure 3 (b) Schematic diagram of the inversion accuracy of each model when it is the root mean square error;

[0052] Figure 4 is the simulation accuracy of the MWCM model under different vegetation coverage; Figure 4 (a) is the simulation accuracy of the MWCM model for corn under VV conditions, Figure 4 (b) is the simulation accuracy of the MWCM model of corn under VH conditions, Figure 4 (c) is the simulation accuracy of the MWCM model for cotton under VV conditions, Figure 4 (d) The simulation accuracy of the MWCM model for cotton under VH conditions;

[0053] Figure 5 This is the inversion result diagram of soil moisture content under different vegetation coverage; Figure 5 (a) is the inversion result diagram of cotton soil moisture content. Figure 5 (b) is the inversion result diagram of corn soil moisture content;

[0054] Figure 6 Flowchart of the regional soil moisture retrieval method based on remote sensing data and water cloud model. DETAILED DESCRIPTION

[0055] The following is a clear and complete description of the technical solutions of the embodiments of the present invention in conjunction with the drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0056] The present embodiment is described in detail with reference to the accompanying drawings. Figure 1 to Figure 6 As shown, the regional soil moisture inversion method based on remote sensing data and water cloud model in this embodiment includes the following steps:

[0057] Step 1: Obtain satellite remote sensing image data of the study area and preprocess the data.

[0058] Obtain satellite remote sensing image data of the study area, preprocess the satellite remote sensing image data, and perform registration processing on the preprocessed satellite remote sensing image data.

[0059] The satellite remote sensing image data are all downloaded from the European Space Agency's Copernicus Sentinel Science Center.

[0060] SAR satellite data download: According to the requirements of the present invention, GRD image data of Level-1 products in IW imaging mode are selected. Satellite image data is downloaded from the Copernicus Sentinel Science Center of the European Space Agency.

[0061] Optical satellite data download: Select Level-2A products according to the needs of this invention. Level-2A products mainly contain atmospheric bottom reflectance data after atmospheric correction. Satellite image data is downloaded from the Copernicus Sentinel Science Center of the European Space Agency.

[0062] The imaging time difference between Sentinel-1 and Sentinel-2 image data should be less than 7 days, and the imaging time should be basically synchronized with the ground sampling time.

[0063] Sentinel-1 data preprocessing: SNAP software is used to preprocess Sentinel-1 image data. The processing steps of GRDH data include orbit correction, thermal noise removal, radiation calibration, speckle filtering and terrain correction.

[0064] Sentinel-2 data preprocessing: Sentinel-2 L2A data has been atmospherically corrected. Since the resolutions of Sentinel-2 bands are different, they need to be unified to the same resolution. The Sne2Res plug-in in SNAP is used to perform super-resolution synthesis on the 20m and 60m bands of Sentinel-2 data to reconstruct Sentinel-2 image data to a 10m resolution, ensuring the consistency of the spatial resolutions of Sentinel-1 and Sentinel-2 satellite images.

[0065] Sentinel-1&2 data registration processing: Due to the different imaging principles of SAR and multispectral satellite, there are significant differences between their images. For subsequent operations, the Sentinel-1&2 image data needs to be registered to ensure that the two images are completely consistent in geometric position. This processing is based on the Image Registration Workflow tool of the ENVI platform.

[0066] Step 2: Extract the reflectance information of satellite remote sensing image data, use the reflectance information to calculate the spectral index, and perform supervised classification on the satellite remote sensing image data (using the random forest method) to obtain the vegetation classification results of the study area.

[0067] Import the Sentinel-2 image matching the sampling date into the ENVI software (Sentinel-2 image here is a type of satellite remote sensing image data), extract the reflectance information of the satellite remote sensing image data, and use the extracted information to obtain spectral indices including the normalized difference vegetation index NDVI, the normalized difference water index ND WI and the leaf area index LAI to describe the vegetation coverage. The calculation formulas of NDVI and NDWI are as follows:

[0068]

[0069]

[0070] Where: B4, B8, B11 are the reflectivities of the 4th, 8th, and 11th bands of Sentinel-2 respectively.

[0071] LAI is calculated by the Biophysical Processor module of SNAP, which is an integrated module based on the PROSAIL model. It generates a comprehensive database of vegetation characteristics and TOC reflectance and trains a neural network to estimate canopy characteristics from TOC reflectance.

[0072] Fractional vegetation coverage (FVC) refers to the percentage of the vertical projection area of ​​vegetation (including leaves, stems, and branches) on the ground to the total area of ​​the statistical area. It is an important indicator for measuring the surface vegetation coverage of a region. The present invention uses the normalized vegetation index NDVI to obtain the fractional vegetation coverage FVC, and the specific expression is:

[0073]

[0074] Among them, NDVI soil It is the NDVI value of the pixel without vegetation cover. Usually, NDVI soil The value is closest to the NDVI cumulative 5% value; NDVI veg is the NDVI value of all pixels covered by vegetation. Usually, NDVI veg The value is closest to the cumulative 95% value of NDVI.

[0075] According to the planting structure recorded in the sampling, ENVI software was used to implement supervised classification of the image based on the Sentinel-2 image on August 16, 2023, and the random forest algorithm was used. The classification results are shown in the figure below. Figure 1 shown.

[0076] Supervised classification, also known as training classification, is a process of automatically identifying and classifying pixels of unknown categories in remote sensing images through machine learning algorithms based on training samples of known categories. It generally includes six steps, namely category definition / feature discrimination, sample selection, classifier selection, image classification, classification post-processing, and accuracy verification. The core of supervised classification is to use known training samples (i.e. labeled data) to train the classifier, and then apply the trained classifier to the entire image to identify and classify the category to which each pixel belongs.

[0077] Training sample selection: Select areas with typical features in the image. These areas should be able to represent the main features of each type of land object. Manually mark the categories of these areas. These marked areas are called training samples.

[0078] Training sample calibration: ENVI: Use the ROI tool of ENVI software to manually draw polygons or points and save them as ROI files. Divide the training samples into training sets and validation sets, and use the validation set to evaluate the performance of the classifier to ensure the generalization ability of the classifier. Use independent validation samples (samples that did not participate in training) to evaluate the accuracy of the classification results.

[0079] Classifier selection: Select a random forest classifier according to your needs (Random Forest (RF) is a powerful machine learning algorithm that performs well in processing high-dimensional data and preventing overfitting), and train the classifier using calibrated training sample data.

[0080] Precision verification: After generating the classification results, use the verification samples to evaluate the accuracy of the classification results. There are two methods for precision verification: confusion matrix and ROC curve. The confusion matrix is ​​selected according to the needs.

[0081] Evaluation indicators in the confusion matrix:

[0082] Overall Accuracy: refers to the proportion of all correctly classified instances to the total number of instances, which is one of the most intuitive evaluation indicators:

[0083]

[0084] In the formula, TP: the number of instances correctly predicted as positive by the model; TN: the number of instances correctly predicted as negative by the model; FP: the number of instances incorrectly predicted as positive by the model; FN: the number of instances incorrectly predicted as negative by the model. Kappa coefficient calculation formula:

[0085]

[0086] Where: p o is the observed consistency, i.e., the proportion of correct classifications by the classification model; p eis the expected consistency, that is, the probability that the classifier will correctly classify the object assuming that the classification is done randomly.

[0087] The overall classification accuracy and Kappa coefficient based on the confusion matrix were calculated to be 93.46% and 0.887, respectively, which indicates that the classification results of the model have high credibility. From the classification result diagram, it can be seen that the main crops in the study area are cotton and corn. The sample data with FVC>0.1 were classified according to different vegetation types, and the results are shown in Table 1.

[0088] Table 1 Statistics of sample quantity and soil moisture characteristics under different vegetation coverage

[0089] Vegetation Type Sample size Minimum value / % Maximum value / % Mean / % Standard Deviation / % Coefficient of variation / % corn 48 8.86 36.27 18.86 6.70 35.50 cotton 212 7.14 40.29 15.99 5.20 32.52 wheat 2 17.09 21.47 19.28 3.10 16.08 tomato 9 9.16 21.77 15.19 4.27 28.12 Edamame 3 8.08 12.32 10.65 2.26 21.20

[0090] Since the number of wheat, tomato, and edamame samples was small, and the main crops grown in the study area were corn and cotton, the corn and cotton sample data were used for modeling.

[0091] Step 3: Introduce the spectral index as a vegetation description parameter into the multi-class water cloud model, perform parameter inversion on the multi-class water cloud model, and select the optimal vegetation description parameter and the optimal soil moisture inversion model based on the inversion results; the parameters include backscattering coefficient and soil moisture content.

[0092] 1. Introducing FVC to improve the water cloud model:

[0093] The traditional water cloud model expression is:

[0094]

[0095]

[0096]

[0097] Where: is the radar backscatter coefficient received by the SAR sensor in the vegetation coverage area; is the scattering coefficient of vegetation; V 1 、V 2 is the vegetation description parameter; τ 2 is the double-layer attenuation factor of the radar passing through vegetation; is the scattering coefficient of the bare soil surface; θ is the radar incident angle. The incident angle of the Sentinel-1 image used in the present invention is about 39°07′, so the incident angle is taken as 39°07′; A and B are the calibration coefficients of the model.

[0098] The vegetation coverage FVC is introduced into the water cloud model, and the total backscatter coefficient in the mixed pixel is split into the scattering contribution of the vegetation coverage area and the scattering contribution of the bare surface. The specific expression is:

[0099]

[0100] in is the radar backscatter coefficient received by the SAR sensor in the vegetation coverage area, is the scattering coefficient of vegetation, τ 2 is the double-layer attenuation factor of the radar passing through vegetation; is the scattering coefficient of bare soil surface.

[0101] The scattering coefficient of bare soil surface is expressed as a linear function of soil moisture content, and its expression is:

[0102]

[0103] Where: M v is the soil moisture content; C is a constant; D is a coefficient related to soil roughness.

[0104] The double-layer attenuation factor is in exponential form and is expanded according to the Maclaurin series formula:

[0105]

[0106] 2. The normalized vegetation index NDVI, normalized difference water index NDWI and leaf area index LAI are introduced into the water cloud model as vegetation description parameters.

[0107] The NDVI, NDWI and LAI in the spectral index are introduced into the water cloud model as vegetation description parameters, with soil moisture content Mv, incident angle θ, backscattering coefficient Vegetation description parameter m veg The nonlinear least squares method is used to invert the water cloud model parameters.

[0108] Among them, the vegetation water content m veg When used as a vegetation description parameter (in practical applications, due to the field measurement m veg It is more difficult, so NDVI or NDWI is usually used to estimate m veg ), the model assumes that the canopy is composed of uniformly sized, homogeneous water clouds. In this case, the vegetation canopy factor can be considered as:

[0109] V 1 =1, V 2 =m veg (12);

[0110] Use NDVI or NDWI to estimate m veg The formula is as follows:

[0111] m veg (NDVI) = 7.63NDVI 4-11.41 NDVI 3 +6.87NDVI 2 -1.24NDVI+0.13(13);

[0112] m veg (NDWI) = 1.44NDWI 2 +1.36NDWI+0.34 (14);

[0113] When LAI is selected as the vegetation description parameter, the model uses LAI to describe the vegetation canopy characteristics:

[0114]

[0115] Where: E and F are the calibration coefficients of the model.

[0116] When m is selected veg When used as vegetation description parameters, the expressions of total backscatter coefficients of WCM and MWCM are:

[0117]

[0118]

[0119] When LAI is selected as the vegetation description parameter, the total backscatter coefficient expressions of WCM and MWCM are:

[0120]

[0121]

[0122] Table 2 Model parameter inversion results

[0123]

[0124] The soil moisture was calculated from the average soil moisture of 0-10cm and 10-20cm. 50% of the data were randomly selected to form the modeling set, and the remaining 50% were used as the validation set. v ), incident angle (θ), backscattering coefficient Vegetation description parameters (m veg , LAI) as input parameters, nonlinear least squares method was used to invert model parameters, and the validation set was used for model validation. Table 2 shows the model parameter inversion results using different vegetation description parameters.

[0125] 3. Select the optimal vegetation description parameters:

[0126] The observed backscatter coefficients of VV and VH of different vegetation description parameter modeling sets and the scatter plots of VV and VH backscatter coefficients obtained by inversion simulation of water cloud models MWCM and WCM are established, as shown in Figure 2 By comparing the determination coefficient (R 2 ) and root mean square error (RMSE) reflect the fitting effect of the model, and the optimal vegetation description parameters are selected; among them, the water cloud model includes the conventional water cloud model WCM and the water cloud model MWCM under low vegetation coverage conditions.

[0127] R 2 The closer it is to 1, the smaller the RMSE is, indicating that the model prediction effect is better and the error between the predicted value and the measured value is smaller. 2 The calculation formula of RMSE is as follows:

[0128]

[0129]

[0130] Where: y i represents the measured value, represents the predicted value, represents the average of the measured values, and n represents the number of samples.

[0131] Combination Figure 2 It can be seen that under the same polarization mode, the fitting effect of the MWCM model is better than that of the WCM model.

[0132] When LAI is used as the vegetation description parameter under VV and VH polarizations and the MWCM model is used for simulation, R 2 Highest, VV polarization R 2 is 0.763, VH polarization R 2 It is 0.757.

[0133] Based on soil moisture content Mv, incident angle θ, backscattering coefficient Vegetation description parameter m veg The input parameters are established based on WCM and MWCM models respectively. Lookup table, invert soil moisture content with different algorithms, obtain the determination coefficient and root mean square error of the inversion results of different algorithms, and select the optimal soil moisture inversion model.

[0134] 4. Select the optimal soil moisture inversion model, including:

[0135] When m is selected veg When used as a vegetation description parameter, the soil moisture expressions of the WCM and MWCM models are:

[0136]

[0137]

[0138] Among them, M v1 is the soil moisture content of the WCM model, M v2 is the soil moisture content of the MWCM model;

[0139] When LAI is selected as the vegetation description parameter, the soil moisture expressions of the WCM and MWCM models are:

[0140]

[0141]

[0142] The soil moisture content (M v ), incident angle (θ), backscattering coefficient Vegetation description parameters (m veg , LAI) are input parameters and are established based on WCM and MWCM models respectively. Lookup table (LUT). Based on the Python platform, LUT, IO and SVR algorithms are constructed to invert soil moisture content, and the R of the inversion results of different algorithms is calculated. 2 , RMSE, draw a radar chart such as Figure 3 As shown in the figure, the optimal soil moisture inversion model is selected.

[0143] The optimal soil moisture inversion model is selected based on the determination coefficient of soil moisture and the root mean square error of soil moisture in the inversion results of different algorithms.

[0144] Step 5: Based on the vegetation classification results and the optimal vegetation description parameters, a lookup table is established using the water cloud model when the vegetation coverage is below the threshold. The backscatter coefficient obtained in the lookup table is input into the optimal soil moisture inversion model to obtain the regional soil moisture content.

[0145] Step 5.1: Inversion of model parameters under different vegetation coverage:

[0146] Based on the classification results of step 3 and the selection of the optimal vegetation description parameters, the MWCM model with LAI as the vegetation description parameter was used to simulate the corn and cotton sample data. 50% of the data were randomly selected to form the modeling set, and the remaining 50% were used as the validation set. v ), incident angle (θ), backscattering coefficient The vegetation description parameter (LAI) is the input parameter, and the nonlinear least square method is used to invert the model coefficients. Table 3 shows the inversion results of model parameters under different vegetation coverage.

[0147] Table 3 Inversion results of model parameters under different vegetation coverage

[0148]

[0149] Step 5.2: Scatter plot analysis of MWCM model under different vegetation cover and R 2 and RMSE evaluation;

[0150] The scatter plots of the observed VV and VH backscatter coefficients of different vegetation cover modeling sets and the VV and VH backscatter coefficients obtained by MWCM simulation were established, as shown in Figure 4 By calculating the R between the observed and predicted values 2 and RMSE reflect the fitting effect of the model.

[0151] By comparing LAI as a vegetation description parameter, the R 2 From the RMSE, we can see that after the vegetation types are divided, the R 2 Both increase, RMSE decreases, and the goodness of fit between the observed backscattering coefficient and the simulated backscattering coefficient of the model is improved, which is more conducive to the inversion of soil moisture content.

[0152] Step 5.3: Inversion of soil moisture content under different vegetation covers;

[0153] Based on the classification results of step 2 and the selection of the optimal soil moisture inversion model in step 3, it can be seen that the soil moisture inversion model constructed by the SVR algorithm has the highest accuracy. Therefore, this step establishes a lookup table based on the MWCM model, trains the model with the backscattering coefficient obtained in the lookup table, and uses the soil moisture as the output variable to calculate R. 2 and RMSE reflect the fitting effect of the model. The inversion results are as follows: Figure 5 shown.

[0154] Step 6: Comparative analysis of the accuracy of the soil moisture inversion model before and after vegetation type division.

[0155] Combination Figure 5 By comparing the R when the soil moisture model inversion effect is best before and after vegetation type division 2 and RMSE, it can be seen that after the vegetation type is divided, the inversion accuracy of the model is improved. When the surface crop is cotton, R 2 When the ground crop is corn, R 2 From 0.690 to 0.763 (10.58%), RMSE decreased from 0.031 to 0.030 (3.23%); when the soil moisture was greater than 0.35cm 3 / cm3 When the predicted value is 0.04, the deviation between the measured value and the predicted value is large, and there is an underestimation phenomenon. The inversion model R of cotton soil moisture content after vegetation type classification 2 Although the improvement is not significant, the RMSE is significantly reduced.

[0156] Based on the above method, the present invention provides a regional soil moisture inversion system based on remote sensing data and a water cloud model, comprising: an acquisition module, a data extraction module, a model selection module and an inversion module.

[0157] Among them, the acquisition module is used to obtain satellite remote sensing image data of the study area; the data extraction module is used to extract the reflectance information of satellite remote sensing image data, use the reflectance information to obtain the spectral index, and classify the vegetation in the study area; the model selection module is used to introduce the spectral index as a vegetation description parameter into the water cloud model, perform water cloud model parameter inversion, and select the optimal vegetation description parameters and the optimal soil moisture inversion model through the inversion results; the inversion module is used to establish a lookup table based on the vegetation classification results and the optimal vegetation description parameters, and the water cloud model under the condition that the vegetation coverage is below the threshold, and the backscattering coefficient obtained in the lookup table is input into the optimal soil moisture inversion model to invert the regional soil moisture content.

[0158] The present invention also provides a computer device, including a memory and a processor. The memory stores a program. When the program is executed by the processor, the processor executes the steps of a regional soil moisture inversion method based on remote sensing data and a water cloud model.

[0159] In accordance with the disclosed embodiments, a computing device may communicate with one or more external devices (e.g., keyboards, pointing devices, Bluetooth communications, etc.), or with any device (e.g., routers, modems, etc.) that enables a computing device to communicate with one or more other computing devices.

[0160] The present invention also provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of a regional soil moisture inversion method based on remote sensing data and a water cloud model are implemented.

[0161] According to the disclosed embodiments, the storage medium may be a non-volatile computer-readable storage medium, such as but not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, the storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0162] The above content is a further detailed description of the present invention in combination with a specific preferred embodiment. For technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as belonging to the protection scope of the present invention.

Claims

1. A regional soil moisture inversion method based on remote sensing data and water cloud model, characterized in that: The steps include: Obtain satellite remote sensing image data of the study area; Extracting reflectivity information of the satellite remote sensing image data, calculating the spectral index using the reflectivity information, and performing supervised classification on the satellite remote sensing image data to obtain a vegetation classification result of the study area; The spectral index is introduced as a vegetation description parameter into a multi-class water cloud model, and the parameters of the multi-class water cloud model are inverted. The optimal vegetation description parameter and the optimal soil moisture inversion model are selected according to the inversion results; the parameters include backscattering coefficient and soil moisture content; Based on the vegetation classification results and the optimal vegetation description parameters, a lookup table is established using a water cloud model under the condition that vegetation coverage is below a threshold, and the backscattering coefficient obtained in the lookup table is input into the optimal soil moisture inversion model to invert and obtain the regional soil moisture content.

2. A method for regional soil moisture inversion based on remote sensing data and water cloud model as claimed in claim 1, characterized in that: The spectral index is introduced as a vegetation description parameter into the multi-class water cloud model, the parameters of the multi-class water cloud model are inverted, and the optimal vegetation description parameter and the optimal soil moisture inversion model are selected according to the inversion results, including: The vegetation coverage FVC is introduced into the water cloud model, and the total backscatter coefficient in the mixed pixel is split into the scattering contribution of the vegetation coverage area and the scattering contribution of the bare surface. The specific expression is: in, is the radar backscatter coefficient received by the SAR sensor in the vegetation coverage area, is the scattering coefficient of vegetation, τ 2 is the double-layer attenuation factor of the radar passing through vegetation; is the scattering coefficient of bare soil surface; The normalized vegetation index NDVI, normalized difference water index NDWI and leaf area index LAI in the spectral index are introduced into the water cloud model as vegetation description parameters, and the soil moisture content Mv, incident angle θ, backscattering coefficient Vegetation description parameter m veg As input parameters, the nonlinear least squares method is used to invert the water cloud model parameters; The observed backscatter coefficients of different vegetation description parameter modeling sets and the backscatter coefficients obtained by inversion simulation of the water cloud model are established, and the optimal vegetation description parameters are selected by comparing the determination coefficient and root mean square error between the observed backscatter coefficients and the backscatter coefficients obtained by inversion simulation; among them, the water cloud model includes the conventional water cloud model WCM and the water cloud model MWCM under low vegetation coverage conditions; Based on soil moisture content Mv, incident angle θ, backscattering coefficient Vegetation description parameter m veg The input parameters are established based on WCM and MWCM models respectively. Lookup table, invert soil moisture content with different algorithms, obtain the determination coefficient and root mean square error of the inversion results of different algorithms, and select the optimal soil moisture inversion model.

3. A method for regional soil moisture inversion based on remote sensing data and water cloud model as claimed in claim 2, characterized in that: The selecting of the optimal soil moisture inversion model comprises: When m is selected veg When used as a vegetation description parameter, the soil moisture expressions of the WCM and MWCM models are: Among them, M v1 is the soil moisture content of the WCM model, M v2 is the soil moisture content of the MWCM model; When LAI is selected as the vegetation description parameter, the soil moisture expressions of the WCM and MWCM models are: The optimal soil moisture inversion model is selected based on the determination coefficient of soil moisture and the root mean square error of soil moisture in the inversion results of different algorithms.

4. The method for regional soil moisture inversion based on remote sensing data and water cloud model according to claim 1, characterized in that: The obtaining of satellite remote sensing image data of the study area includes: Obtain satellite remote sensing image data of the study area, preprocess the satellite remote sensing image data, and perform registration processing on the preprocessed satellite remote sensing image data; The satellite remote sensing image data are all downloaded from the European Space Agency's Copernicus Sentinel Science Center.

5. The method for regional soil moisture inversion based on remote sensing data and water cloud model according to claim 1, characterized in that: Extracting reflectivity information of the satellite remote sensing image data and calculating the spectral index using the reflectivity information include: The reflectivity information of satellite remote sensing image data is extracted, and the spectral index including the normalized difference vegetation index NDVI, the normalized difference water index NDWI and the leaf area index LAI is obtained by using the reflectivity information; the calculation formulas of NDVI and NDWI are as follows: Among them, B4, B8, and B11 are the reflectances of the 4th, 8th, and 11th bands of Sentinel-2 in satellite remote sensing images, respectively.

6. A method for regional soil moisture inversion based on remote sensing data and water cloud model as claimed in claim 5, characterized in that: The method of calculating the spectral index using the reflectivity information further includes: The normalized difference vegetation index NDVI is used to obtain the vegetation coverage FVC. The specific expression is: Among them, NDVI soil is the NDVI value of the pixel without vegetation cover, NDVI veg is the NDVI value of all pixels covered by vegetation.

7. A regional soil moisture inversion system based on remote sensing data and water cloud model, characterized in that: include: The acquisition module is used to obtain satellite remote sensing image data of the study area; A data extraction module is used to extract the reflectivity information of the satellite remote sensing image data, calculate the spectral index using the reflectivity information, and perform supervised classification on the satellite remote sensing image data to obtain the vegetation classification result of the study area; A model selection module is used to introduce the spectral index as a vegetation description parameter into the multi-class water cloud model, perform parameter inversion on the multi-class water cloud model, and select the optimal vegetation description parameter and the optimal soil moisture inversion model according to the inversion results; the parameters include backscattering coefficient and soil moisture content; The inversion module is used to establish a lookup table based on the vegetation classification results and the optimal vegetation description parameters by a water cloud model under the condition that the vegetation coverage is lower than a threshold, input the backscattering coefficient obtained in the lookup table into the optimal soil moisture inversion model, and invert to obtain the regional soil moisture content.

8. A computer device, characterized in that: It comprises a memory and a processor, wherein a program is stored in the memory, and when the program is executed by the processor, the processor executes the steps of a regional soil moisture inversion method based on remote sensing data and a water cloud model as described in any one of claims 1 to 6.

9. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a method for regional soil moisture inversion based on remote sensing data and a water cloud model according to any one of claims 1 to 6 are implemented.

Images