An improved soil moisture inversion method based on water cloud model and spaceborne SAR data

By introducing the homopolarization purity parameter mc in the water cloud model, using satellite-borne SAR data to optimize soil moisture inversion, the problems of low accuracy and complex data processing in soil moisture monitoring are solved, and real-time accurate inversion of soil moisture and Efficient monitoring.

CN118688226BActive Publication Date: 2024-11-15NANJING UNIV OF AERONAUTICS & ASTRONAUTICS
View PDF 9 Cites 0 Cited by

Patent Information

Application Number
CN202411170113.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-26
Publication Date
2024-11-15
Estimated Expiration
2044-08-26

AI Technical Summary

Technical Problem

The existing soil moisture monitoring methods have problems such as low accuracy, long time consumption, high cost and complex multi-source remote sensing data processing. In particular, the traditional water cloud model fails to fully consider the proportion of vegetation and soil scattering contributions, which affects the inversion accuracy.

Method used

Based on the satellite-borne SAR data, the homopolarization purity parameter mc is introduced to optimize the water cloud model, and only Sentinel-1 GRD radar data is used for real-time accurate inversion of soil moisture, simplifying the data processing process, and improving inversion efficiency and accuracy.

Benefits of technology

It realizes large-scale real-time accurate inversion of soil moisture, overcomes the problem of the susceptibility of multi-source remote sensing inversion data quality and the traditional model being too simple, and improves the inversion accuracy and efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118688226B_ABST
    Figure CN118688226B_ABST
Patent Text Reader

Abstract

The invention discloses an improved water cloud model soil moisture inversion method based on satellite-borne SAR data, which relates to the technical field of soil moisture content inversion. A plurality of random sample points are selected, and the humidity and longitude and latitude information of the target area are measured by using a soil moisture sensor; the Sentinel-1 GRD data of the target area are downloaded according to the measured longitude and latitude information of the point; the acquired Sentinel-1 GRD data are preprocessed to obtain radar polarization data and incident angle data, and the co-polarization purity parameter m is calculated by using the radar polarization data. c The water cloud model is improved to obtain an improved soil moisture inversion model; soil moisture is inverted using the improved water cloud model and the original water cloud model to obtain soil moisture inversion data, the inversion accuracy is evaluated based on the measured soil moisture data and the inversion accuracy of the improved water cloud model and the original water cloud model is compared; the improved water cloud model is used to predict the soil moisture in the target area, and the spatial distribution map of soil moisture in the image area can be obtained.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of soil moisture content inversion, and in particular to an improved water cloud model soil moisture inversion method based on spaceborne SAR data. Background Art

[0002] Soil moisture monitoring is of great significance for guiding scientific fertilization and reasonable irrigation, especially in arid areas, where the rational use of soil moisture is the key to improving agricultural productivity. In addition, soil moisture monitoring helps to scientifically evaluate the status of groundwater and surface water resources, guide the rational development and protection of water resources, and achieve sustainable use of water resources.

[0003] At present, the measurement methods of soil moisture mainly include direct method and indirect method. Direct method, such as drying and weighing method, measures soil moisture content by sampling soil samples in the field and drying and weighing them. It has high accuracy but is time-consuming and can only be measured at a single point. Indirect methods, including resistance method, seed method and negative pressure meter method, can obtain soil moisture information at limited points, but there are problems such as time-consuming, labor-intensive and high cost.

[0004] With the development of remote sensing technology, especially the application of radar signals, a new way to measure soil moisture has been provided. Radar signals have the ability to work all day and all weather, and can penetrate vegetation, making it possible to continuously monitor soil moisture.

[0005] Remote sensing inversion of soil moisture includes two methods: optical and microwave remote sensing. Optical inversion is often restricted by meteorological factors such as weather and clouds, while microwave remote sensing is not easily disturbed by weather conditions. Therefore, microwave remote sensing is often used for soil moisture inversion. Microwaves are divided into active microwaves and passive microwaves. Synthetic aperture radar (SAR) technology is an active microwave method, which means that the transmitter actively transmits microwave signals, and then the receiver receives the reflected signals. This method can penetrate clouds and rain and fog, and has the characteristics of all-weather and all-day; passive microwaves only receive microwave signals radiated from the object itself, without transmitting any signals. This method is less affected by weather conditions, but the signal strength is low. In order to eliminate interference from other factors and improve the inversion accuracy, most SAR technology is used for soil moisture inversion. The principle of SAR technology inverting soil moisture lies in the linear relationship between the radar backscattering coefficient and the soil dielectric constant, but the radar backscattering coefficient includes not only the contribution of the soil part, but also the contribution of soil cover (vegetation). Therefore, it is often necessary to use physical models (such as water cloud models) to achieve higher-precision inversion. The commonly used water cloud model is based on the radiation transfer equation and takes into account the scattering characteristics of vegetation and soil. However, the model is relatively simplified and there are challenges in acquiring and processing multi-source data such as optical and microwave remote sensing. Summary of the invention

[0006] In order to improve the inversion accuracy, the present invention is based on the water cloud model and introduces the co-polarization purity parameter m c , optimize the model parameters, and use only Sentinel-1 GRD radar data to achieve real-time and accurate inversion of soil moisture. This method simplifies the data processing process, improves the efficiency and accuracy of inversion, and provides a new technical approach for soil moisture monitoring.

[0007] To this end, the present invention provides an improved water cloud model soil moisture inversion method based on spaceborne SAR data to solve the problems raised in the background technology.

[0008] In order to achieve the above-mentioned object, the present invention provides the following technical solutions: an improved water cloud model soil moisture inversion method based on spaceborne SAR data, selecting a number of random sample points, using a soil moisture sensor to measure the humidity and longitude and latitude information of the target area; downloading the Sentinel-1 GRD data of the target area according to the measured longitude and latitude information of the point; preprocessing the acquired Sentinel-1 GRD data to obtain radar polarization data and incident angle data, including the following steps:

[0009] Step 1: Calculate the co-polarization purity parameter m using radar polarization data c Improve the water cloud model and obtain an improved soil moisture inversion model;

[0010] Step 2: Use the improved water cloud model and the original water cloud model to invert soil moisture to obtain soil moisture inversion data, evaluate the inversion accuracy based on the measured soil moisture data, and compare the inversion accuracy of the improved water cloud model and the original water cloud model;

[0011] Step 3: Use the improved water cloud model to predict the soil moisture in the target area, and the output result can be the spatial distribution map of soil moisture in the image area.

[0012] Preferably, data preprocessing includes orbit correction, thermal noise removal, radiation calibration, speckle filtering and terrain correction of Sentinel-1 GRD data, all of which are implemented by ESA Snap software, and finally the backscatter coefficient information of VH and VV polarizations and the incident angle information of the target area image are obtained;

[0013] The pre-processing operation is as follows:

[0014] Sentinel-1 GRD data orbit correction: Automatically download accurate orbit files and update the orbit status information of Sentinel-1 satellites in the metadata file (.xml) in the Sentinel-1 satellite data. The orbit status data in the default metadata file is not accurate enough.

[0015] Thermal noise removal: The acquired synthetic aperture radar (SAR) images should contain thermal noise, which will affect the accuracy of the estimated radar backscatter signal. The data should be denoised to improve the quality and clarity of the data and reduce the interference of subsequent processing;

[0016] Radiometric calibration: Radiometric calibration is the conversion of the received backscattered signal (energy) into a physical quantity with units (some are proportional values ​​without units), such as the backscatter coefficient here. For SAR data, since radar waves can penetrate clouds, only radiometric calibration is required, and atmospheric correction of optical images is not required;

[0017] Speckle filtering: Speckle filtering is a commonly used method in SAR image processing, which is used to reduce or eliminate speckles in images, thereby improving image quality and visualization. Speckle is a type of noise caused by the coherence of the SAR system, making the image look rough and messy. The basic idea of ​​speckle filtering is to process the image through some filtering techniques to reduce the impact of speckles while retaining the details and features of the image as much as possible;

[0018] Terrain correction: SAR image terrain correction is a correction process to eliminate image distortion caused by terrain changes. Image distortion caused by terrain changes mainly includes slope distortion and shadow effect. In addition to geocoding (giving the image actual coordinate information), terrain correction also performs terrain radiation correction.

[0019] Preferably, the soil measurement points should not be too dense and should be distributed as evenly as possible. The soil moisture measurement data includes the point WGS84 coordinate information and 5 cm soil moisture information.

[0020] Preferably, the required remote sensing satellite data is only Sentinel-1 GRD data, and no other optical remote sensing data is required, thereby reducing the amount of data and the data processing flow.

[0021] Preferably, the original water cloud model in step 1 is:

[0022] ;

[0023] ;

[0024] ;

[0025] ;

[0026] in, is the total backscatter of the vegetation-covered surface, in dB; is the backscatter of the vegetation layer, in dB; is the underlying soil backscatter, in dB; is the two-way attenuation factor of radar waves passing through the vegetation layer; dB; is the incident angle of radar wave, in degrees; is the optical thickness of vegetation; the empirical constants of A and B depend on the vegetation type and the frequency of the incident electromagnetic wave; and Represents the vegetation canopy descriptor, usually related to the leaf area index LAI;

[0027] when When , the linear equation of the vegetation index (which can be a vegetation descriptor closely related to the vegetation canopy, such as NDVI, EVI, etc.) is expressed as:

[0028] ;

[0029] Where C and D are fitting constants, is the vegetation index.

[0030] The water cloud model has been improved as follows:

[0031] ;

[0032] in is the co-polarization purity parameter, which indicates the purity of the co-polarization signal in a pixel. Specifically, it is the ratio of VV polarization to VH polarization in a pixel in the Sentinel-1 image. The expression formula is as follows:

[0033] ;

[0034] ;

[0035] in, Represents the radar polarization ratio, which is the ratio of cross-polarization VH to co-polarization VV in GRD data.

[0036] In VV polarization mode, the radar signal is reflected along the ground surface and is more sensitive to changes in soil moisture. VV polarization is often used for soil moisture estimation because it is more sensitive to reflections from horizontal surfaces and can provide information on surface soil moisture.

[0037] In VH polarization mode, the radar signal propagates in a vertical direction and reflects along the ground surface. VH polarization is generally more sensitive to factors such as vegetation cover and surface roughness.

[0038] In low vegetation conditions, the co-polarization returns will be high, with the cross-polarization returns being negligible (q→0).

[0039] It can be seen from the above formula that under bare ground conditions The higher the value, the higher the vegetation density and the terrain advances. The value gradually decreases, which is inversely proportional to the vegetation coverage. The parameter can infer the mixture of the two polarization intensities in a pixel, thus indicating the purity of the co-polarization component in the same pixel. Co-polarization inversion of soil moisture is better, while cross-polarization is more sensitive to vegetation cover, so is considered as the soil contribution to the total backscattering coefficient in the water cloud model, It is considered as the contribution of the vegetation layer. Previous studies have shown that the mechanism of water cloud model inversion of soil moisture lies in the radar backscattering coefficient of bare soil. With soil moisture There is a linear relationship, so we can Expressed as a linear expression of the measured soil moisture value, the improved water cloud model is obtained:

[0040] ;

[0041] ;

[0042] ;

[0043] ;

[0044] ;

[0045] ;

[0046] Among them, A, B, C, D, E, and F are all constant terms, and the optimal solution can be found by the least squares method.

[0047] Preferably, the method for evaluating the inversion accuracy in step 2 is:

[0048] The root mean square error (RMSE) and the coefficient of determination R 2 As an accuracy evaluation indicator, the formula is as follows:

[0049] The root mean square error (RMSE) is calculated as follows:

[0050] ;

[0051] In the formula, n is the total number of samples, is the model prediction value, is the true value; the range of root mean square error and mean absolute percentage error is , when the predicted value is exactly the same as the true value, it is equal to 0, which is a perfect prediction.

[0052] Coefficient of determination R2 The calculation formula is as follows:

[0053] ;

[0054] in:

[0055] is the residual sum of squares, which represents the sum of differences between the model predictions and the actual observed values.

[0056] is the total sum of squares, which represents the sum of the differences between the actual observations and the observed mean.

[0057] R 2 Represents the data variance ratio of the model; when R 2 When it is close to 1, it means that the model can fit the data well; when R 2 When it is close to 0, it means that the model does not fit the data well.

[0058] The present invention has the following advantages:

[0059] The present invention provides a method for soil moisture inversion based on an improved water cloud model based on Sentinel-1 GRD data. Based on the water cloud model optimization parameters, only Sentinel-1 GRD radar data is used, and a co-polarization purity parameter mc is introduced to improve the water cloud model. The method effectively overcomes the problems that the quality of multi-source remote sensing inverted soil moisture data is easily disturbed, the real-time performance is poor, and the traditional water cloud model is too simple and does not fully consider the proportion of vegetation and soil scattering contribution, thereby realizing large-scale real-time and accurate inversion of soil moisture. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 is a flow chart of the inversion method described in this embodiment;

[0061] Figure 2 A schematic diagram of the measured positions of the sample points used in this embodiment;

[0062] Figure 3 This is the training accuracy map of the original water cloud model based on this embodiment;

[0063] Figure 4 This is a training accuracy diagram based on the improved water cloud model in this embodiment. DETAILED DESCRIPTION

[0064] The following is a description of the implementation of the present invention by specific embodiments. People familiar with the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. 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 are within the scope of protection of the present invention.

[0065] This embodiment provides a soil moisture estimation method based on multi-source remote sensing data. For detailed process, please refer to Figure 1 , including the following specific steps:

[0066] Step 1: Select a number of random sample points and use the soil moisture sensor to measure the soil moisture 5 cm below the surface of the target area and the WGS84 coordinate longitude and latitude information of the point;

[0067] This embodiment uses the 2016 SMAP soil moisture satellite verification experiment data of City M in Country A as the measured data. These data include soil moisture, soil temperature and precipitation measurement data recorded by the permanent soil moisture network of City SMAPVEX16-M from May 1 to September 30, 2016. These locations are distributed in an experimental area of ​​about 30 kilometers north of City N and about 30 kilometers by 40 kilometers. The experimental area is an agricultural ecosystem with low vegetation coverage, which is suitable for water cloud models. The present invention selects 15 locations with 5 cm underground soil moisture measurement data at 12 noon on August 5 as the measured data. The sample points are distributed as follows: Figure 2 shown.

[0068] Step 2: Download the Sentinel-1 GRD data of the target area based on the latitude and longitude information of the measured points;

[0069] According to the description of the surface microwave scattering mechanism in low vegetation coverage areas in the water cloud model, the overall backscattering in low vegetation coverage areas is composed of vegetation effects, soil effects weakened by vegetation, and multiple soil-vegetation scattering effects. Specifically, Sentinel-1SAR GRDH format data is selected as the data source of microwave scattering signal parameters;

[0070] Step 3: Preprocess the acquired Sentinel-1 GRD data to obtain radar polarization data and incident angle data;

[0071] The GRD data preprocessing operations involved in this embodiment include: orbit correction, thermal noise removal, radiation calibration, coherent speckle filtering and terrain correction, and the above operations are all processed by the ESA official data processing software Snap. After the processing is completed, the backscattering information and radar incident angle information under the two polarization modes of VV and VH are extracted;

[0072] Step 4: Calculate the co-polarization purity parameter m using radar polarization data c Improve the water cloud model and obtain an improved soil moisture inversion model;

[0073] The radar polarization ratio q is calculated based on the cross-polarization coefficient VH and co-polarization coefficient VV obtained after preprocessing the Sentinel-1 data. The basic formula is as follows:

[0074] ;

[0075] The radar vegetation index RVI is calculated based on q. The basic formula is as follows:

[0076] ;

[0077] The radar vegetation index DpRVIc is calculated according to the radar polarization ratio q. The calculation formula is as follows:

[0078] ;

[0079] The co-polarization purity parameter is calculated based on the radar polarization ratio q :

[0080] ;

[0081] The calculated q, RVI, DpRVIc, The soil moisture measured in step 1 and the VV, VH backscatter coefficients extracted in step 3, radar incident angle information and sample point WGS84 coordinate information are compiled into a table, as shown in Table 1;

[0082] Table 1

[0083]

[0084] The original water cloud model is as follows:

[0085] ;

[0086] ;

[0087] ;

[0088] ;

[0089] In the above formula, is the total backscatter of the vegetation-covered surface, in dB; is the backscatter of the vegetation layer, in dB; is the underlying soil backscatter, in dB; is the two-way attenuation factor of radar waves passing through the vegetation layer, dB; is the incident angle of radar wave, in degrees; is the optical thickness of vegetation. The empirical constants of A and B depend on the vegetation type and the frequency of the incident electromagnetic wave; V1 and V2 represent vegetation canopy descriptors, which are usually related to the leaf area index LAI. When V1=V2, it can be expressed as a linear equation of vegetation index (which can be NDVI, EVI and other vegetation descriptors closely related to vegetation canopy):

[0090]

[0091] Where C and D are fitting constants, and VI is the vegetation index.

[0092] The water cloud model has been improved as follows:

[0093]

[0094]

[0095]

[0096]

[0097]

[0098]

[0099] Among them, A, B, C, D, E, and F are all constant terms, and the optimal solution can be obtained by the least squares method.

[0100] The 15 soil moisture sampling points were randomly divided into a modeling group and a validation group. The data of 8 soil moisture sampling points were used to solve the coefficients and establish a microwave soil moisture inversion semi-empirical model. The other 7 soil moisture sampling points were used for model verification. The backscattering coefficients (VV polarization and VH polarization), radar vegetation index (q, DpRVIc, RVI), co-polarization purity parameter (mc), radar incident angle information and 5cm soil moisture of the sampling points in the modeling group were substituted into the improved water cloud model for modeling and solving, and the best model parameter combination was selected. The sampling point parameter information is shown in Table 1.

[0101] Through the parameter combination solution, it can be seen that the improved inversion model obtained by selecting the combination of VV polarization and DpRVIc radar vegetation index has the best effect, so it is used as the final improved water cloud model.

[0102] Step 5: Use the improved water cloud model and the original water cloud model to invert soil moisture to obtain soil moisture inversion data, evaluate the inversion accuracy based on the measured soil moisture data, and compare the inversion accuracy of the improved water cloud model and the original water cloud model;

[0103] The measured soil moisture data of the validation group were substituted into the improved water cloud model and the original water cloud model obtained by the modeling group to obtain the predicted soil moisture value, and the root mean square error RMSE and determination coefficient R were calculated. 2 The improved water cloud model and the original water cloud model were evaluated, and the RMSE and R of the improved water cloud model were 0.03 and 0.1, respectively. 2 is 0.88; the RMSE of the original water cloud model is 0.04, R 2 is 0.78, see Figure 3 and Figure 4 Comprehensive analysis shows that the inversion effect of the improved water cloud model is better than that of the original water cloud model.

[0104] Step 6: Use the improved water cloud model to predict the soil moisture in the target area. The output result can be the spatial distribution map of soil moisture in the image area.

[0105] Although the present invention has been described in detail above by general description and specific embodiments, it is obvious to those skilled in the art that some modifications or improvements can be made to the present invention. Therefore, these modifications or improvements made without departing from the spirit of the present invention all belong to the scope of protection claimed by the present invention.

Claims

1. An improved water cloud model soil moisture inversion method based on spaceborne SAR data, randomly selecting a number of sample points, using soil moisture sensors to measure the humidity and longitude and latitude information of the target area; downloading the Sentinel-1 GRD data of the target area according to the measured longitude and latitude information of the point; preprocessing the acquired Sentinel-1 GRD data to obtain radar polarization data and incident angle data, characterized in that: The steps include: Step 1: Calculate the co-polarization purity parameter m using radar polarization data c Improve the water cloud model and obtain an improved soil moisture inversion model; The original water cloud model is: ; ; ; ; in, is the total backscatter of the vegetation-covered surface, in dB; is the backscatter of the vegetation layer, in dB; is the underlying soil backscatter, in dB; is the two-way attenuation factor of the radar wave passing through the vegetation layer, in dB; is the incident angle of radar wave, in degrees; is the optical thickness of vegetation; the empirical constants of A and B depend on the vegetation type and the frequency of the incident electromagnetic wave; and They represent vegetation canopy descriptors respectively; when When , the linear equation of vegetation index is expressed as: ; Where C and D are fitting constants, is the vegetation index; The water cloud model has been improved as follows: ; in is the co-polarization purity parameter, which indicates the purity of the co-polarization signal in a pixel. Specifically, it is the ratio of VV polarization to VH polarization in a pixel in the Sentinel-1 image. The expression formula is as follows: ; ; in, Represents the radar polarization ratio, which is the ratio of cross-polarization VH to co-polarization VV in GRD data; Will Expressed as a linear expression of the measured soil moisture value, the improved water cloud model is obtained: ; ; ; ; ; ; Among them, A, B, C, D, E, and F are all constant terms. is the soil moisture, and the optimal solution can be obtained by the least square method; Step 2: Use the improved water cloud model and the original water cloud model to invert soil moisture to obtain soil moisture inversion data, evaluate the inversion accuracy based on the measured soil moisture data, and compare the inversion accuracy of the improved water cloud model and the original water cloud model; Step 3: Use the improved water cloud model to predict the soil moisture in the target area, and the output result can be the spatial distribution map of soil moisture in the image area.

2. The improved water cloud model soil moisture inversion method based on spaceborne SAR data according to claim 1 is characterized by: Data preprocessing includes orbit correction, thermal noise removal, radiation calibration, speckle filtering and terrain correction of Sentinel-1 GRD data.

3. The improved water cloud model soil moisture inversion method based on spaceborne SAR data according to claim 1 is characterized by: The measured soil moisture data includes the point WGS84 coordinate information and 5cm soil moisture information.

4. The improved water cloud model soil moisture inversion method based on spaceborne SAR data according to claim 1 is characterized by: The method for evaluating the inversion accuracy in step 2 is: The root mean square error and the coefficient of determination R 2 As an accuracy evaluation indicator, the formula is as follows: The root mean square error is calculated as follows: ; In the formula, n is the total number of samples, is the model prediction value, is the true value; the range of root mean square error and mean absolute percentage error is , when the predicted value is exactly the same as the true value, it is equal to 0, that is, perfect prediction; Coefficient of determination R 2 The calculation formula is as follows: ; in: is the residual sum of squares, which represents the sum of the differences between the model predictions and the actual observed values; is the total sum of squares, which represents the sum of the differences between the actual observations and the observed mean; R 2 Represents the data variance ratio of the model; when R 2 When it is close to 1, it means that the model can fit the data well; when R 2 When it is close to 0, it means that the model does not fit the data well.

Citation Information

Patent Citations

  • PolSAR soil water content inversion method based on two-component decomposition model

    CN109521182A

  • Soil moisture inversion method and device, computer readable medium and electronic equipment

    CN111860325A

  • Vegetation leaf water content remote sensing inversion method based on polarimetric SAR data

    CN114994087A

  • Remote sensing, light energy utilization rate model and weather integrated crop mature period prediction method

    CN115375036A

  • Earth surface thermal infrared emissivity inversion method based on MODIS data and Transform network

    CN115422703A