Soil humidity inversion method, device, medium and equipment

Abstract

The invention discloses a soil humidity inversion method and device, a medium and equipment, and relates to the technical field of microwave remote sensing inversion. According to the method, the total backscattering coefficient, the normalized vegetation index, the leaf area index and the average leaf inclination angle of the electromagnetic waves of the target area pixel by pixel are obtained in combination with microwave remote sensing data and atmospheric apparent reflectance data of the target area; according to the normalized vegetation index, the leaf area index and the average leaf inclination angle, spatial clustering is carried out on vegetation, spatial heterogeneity of crop types is captured, and then based on the constraint that the surface roughness parameters and the vegetation parameters of the same vegetation type are the same and the soil humidity in the same surface space window is the same, the spatial heterogeneity of crop types is obtained. Through two-stage iteration inversion of the intra-class scale and the ground surface space window scale, the time dynamic state is adapted, the parameter calibration function of the in-situ soil humidity data is replaced, and the model parameters and the soil humidity can be determined without the in-situ soil humidity data. According to the invention, the accuracy and reliability of the soil humidity inversion method are improved.

Description

Technical Field

[0001] This invention relates to the field of microwave remote sensing inversion technology, and in particular to a method, apparatus, medium and equipment for soil moisture inversion. Background Technology

[0002] Currently, soil moisture, as a key regulator of crop root water absorption and transpiration metabolism in agricultural ecosystems, requires precise analysis of its spatiotemporal variability. This is of significant scientific importance for optimizing irrigation management strategies, alleviating regional water resource pressures, and even maintaining global food security. Remote sensing technology, with its macro-scale observation capabilities and periodic revisit characteristics, can acquire continuous surface information regionally and even globally, bringing revolutionary technological breakthroughs to agricultural soil moisture monitoring. Microwave remote sensing, based on the strong coupling relationship between soil dielectric constant and moisture content and its cloud penetration capability, demonstrates irreplaceable application potential in operational monitoring of farmland soil moisture.

[0003] In existing technologies, the standard procedure for microwave remote sensing soil moisture retrieval typically includes two key steps: first, constructing a mathematical mapping between radar backscattering and soil dielectric properties, a process that relies on various scattering models to characterize the electromagnetic interaction mechanism; and second, using a dielectric constant hybrid model to quantitatively estimate the surface soil moisture content. Research has developed a series of model systems to characterize the scattering properties of bare soil based on radiative transfer theory and multi-source observation data. Among them, the AIEM model, with its rigorous electromagnetic scattering theoretical foundation and the coupled characterization of soil moisture, temperature, texture (clay and sand content), and bulk density by the Dobson dielectric model, effectively solves the parameter adaptability problem for different soil types and exhibits excellent retrieval performance under complex surface conditions.

[0004] However, despite the significant achievements of the aforementioned models under bare soil conditions, their application in vegetated areas still faces the challenge of effectively decoupling surface scattering characteristics. To address this issue, researchers have proposed the Water-Cloud Model (WCM). This model decomposes the total backscattering coefficient of vegetated surfaces into two components: vegetation scattering and soil scattering after vegetation attenuation. Vegetation scattering is related to vegetation water content, while soil scattering is affected by vegetation attenuation, the degree of which depends on factors such as vegetation water content and canopy type. In this way, the Water-Cloud Model effectively decouples the scattering contributions of vegetation and soil, providing a feasible approach for soil moisture inversion in vegetated areas.

[0005] In practical inversion processes, the Advanced Integral Equation Model (AIEM) is currently the mainstream physical model for simulating the microwave scattering characteristics of exposed surfaces. However, AIEM models, along with water cloud models, face challenges in dynamic parameter updates and scale matching. For AIEM models, the surface roughness parameter (root mean square height) is crucial. s Relevant length l The backscattering coefficient (B) is the physical basis for characterizing the interaction between electromagnetic waves and the Earth's surface—its value directly determines the simulation accuracy of the backscattering coefficient, but this parameter exhibits significant spatiotemporal heterogeneity. Its temporal uncertainty stems from multi-scale dynamic variations driven by natural and anthropogenic processes such as precipitation compaction and farming activities. Spatial heterogeneity is determined by the non-uniform distribution at micro- and regional scales caused by topographic slope, soil texture, and farming systems. Vegetation parameters (volume scattering coefficient A, attenuation coefficient B) in the water cloud model also face determination challenges. In the spatial dimension, parameter heterogeneity across vegetation types leads to significant differentiation in the adaptability of inversion algorithms. In the temporal dimension, dynamic changes in parameters driven by crop phenology cause seasonal biases in inversion models based on static parameters. This spatiotemporal heterogeneity fundamentally limits the applicability of model calibration based on prior information in a limited spatiotemporal domain. Simultaneously, the ill-conditioned nature of the parameter inversion problem makes accurate model calibration at high spatiotemporal resolution scales virtually impossible.

[0006] To address this, some studies have introduced spatiotemporal constraints on model parameters to achieve more stable model calibration. Common assumptions include that model parameters remain stable within a time window and that vegetation types share a unified set of vegetation parameters. These assumptions essentially balance the contradiction between limited prior information and ill-conditioned parameter inversion. Based on this, the number of model parameters to be estimated is significantly reduced, and by combining multi-temporal remote sensing data with land cover type data, soil moisture inversion can be achieved at global or regional scales.

[0007] However, farmland areas have short crop growth cycles, frequent cultivation activities, and complex planting structures. Satellite revisit cycles are difficult to reconcile with the frequency of changes in crop parameters and surface roughness. Furthermore, the spatial scale and classification standards of land cover type data are relatively coarse, often failing to characterize the spatial variability of crop types. Additionally, current assumptions still rely on in-situ soil moisture data (soil moisture measured in the field, such as values ​​from soil moisture sensors buried in the soil, or soil moisture measured using the drying method from local soil samples) or calibration of existing soil moisture remote sensing data, which easily introduces errors in the soil moisture data. In summary, existing methods for retrieving farmland soil moisture have low accuracy and poor reliability. Summary of the Invention

[0008] Therefore, it is necessary to provide a method, apparatus, medium, and equipment for soil moisture inversion to address the aforementioned technical problems.

[0009] The present invention adopts the following technical solution: This invention provides a method for soil moisture retrieval. First, based on microwave remote sensing data of the target area, the total backscattering coefficient of electromagnetic waves in the target area is determined pixel-by-pixel. Then, based on atmospheric apparent reflectance data of the target area, the normalized vegetation index, leaf area index, and mean leaf tilt angle of the target area are determined pixel-by-pixel. Next, the normalized vegetation index, leaf area index, and mean leaf tilt angle of the target area are clustered to determine the vegetation types and corresponding regions. Then, by coupling a soil backscattering model, a soil dielectric constant model, and a water cloud model, the influence relationship between surface roughness parameters, vegetation parameters, and soil moisture on the total backscattering coefficient of electromagnetic waves is obtained. Thus, the method uses surface roughness parameters of the same vegetation type... Constrained by having the same vegetation parameters and the same soil moisture within the same surface space window, the soil moisture is initialized pixel by pixel. A system of simultaneous equations on the influence relationship at the intra-class scale is constructed for each vegetation type and solved to obtain the initial surface roughness parameters and vegetation parameters for each vegetation type. Each surface space window corresponds to multiple adjacent pixels. Then, for each surface space window in the pre-divided target area, a system of simultaneous equations on the influence relationship at the window scale is constructed, and the initial surface roughness parameters and vegetation parameters of the corresponding vegetation type are substituted to solve the problem to obtain the updated soil moisture for each surface space window. The surface roughness parameters and vegetation parameters of each vegetation type are iteratively updated and solved, and the soil moisture of each surface space window is iteratively updated and solved.

[0010] This invention provides a soil moisture inversion device, comprising: The preliminary determination module is used to determine the total backscattering coefficient of electromagnetic waves in the target area pixel by pixel based on microwave remote sensing data of the target area; and to determine the normalized vegetation index, leaf area index and mean leaf tilt angle of the target area pixel by pixel based on atmospheric apparent reflectance data of the target area. The clustering module is used to cluster the normalized vegetation index, leaf area index and mean leaf tilt angle of the target area to determine the vegetation types and corresponding areas of the target area. The coupling module is used to couple the soil backscattering model, soil dielectric constant model and water cloud model to obtain the influence relationship between surface roughness parameters, vegetation parameters and soil moisture on the total backscattering coefficient of electromagnetic waves. The intra-class scale solution module is used to initialize the soil moisture of each pixel under the constraints that the surface roughness parameters and vegetation parameters of the same vegetation type are the same and the soil moisture within the same surface space window is the same. It constructs and solves a system of simultaneous equations for the influence relationship of intra-class scale for each vegetation type to obtain the initial surface roughness parameters and vegetation parameters of each vegetation type. The surface space window corresponds to multiple adjacent pixels. The window scale solution module is used to construct a system of simultaneous equations on the influence relationship of each surface space window in the pre-divided target area, and substitute the initial surface roughness parameters and vegetation parameters of the corresponding vegetation type to solve for the updated soil moisture of each surface space window. This allows for iterative updating and solving of the surface roughness parameters and vegetation parameters of each vegetation type, as well as the soil moisture of each surface space window.

[0011] The present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described soil moisture inversion method.

[0012] The present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described soil moisture inversion method.

[0013] The above-mentioned at least one technical solution adopted in this invention can achieve the following beneficial effects: This invention combines microwave remote sensing data and atmospheric apparent reflectance data of the target area to obtain the total backscattering coefficient, normalized vegetation index, leaf area index, and mean leaf tilt angle of electromagnetic waves per pixel in the target area. Then, based on the normalized vegetation index, leaf area index, and mean leaf tilt angle, the vegetation is spatially clustered to capture the spatial heterogeneity of crop types. Subsequently, based on the constraints that the surface roughness parameters and vegetation parameters of the same vegetation type are the same, and the soil moisture is the same within the same surface spatial window, a two-stage iterative inversion at the intra-class scale and the surface spatial window scale is performed. This not only adapts to temporal dynamics but also replaces the parameter calibration function of in-situ soil moisture data, allowing the determination of model parameters and soil moisture without the need for in-situ soil moisture data. This invention improves the accuracy and reliability of the soil moisture inversion method by accurately characterizing the spatiotemporal variation of parameters while avoiding the errors introduced by in-situ soil moisture data. Attached Figure Description

[0014] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a schematic diagram of a soil moisture inversion method provided by the present invention; Figure 2 A schematic diagram showing the distribution of the research area and soil moisture observation points provided by this invention; Figure 3 A schematic diagram of a specific data processing flow for soil moisture inversion provided by the present invention; Figure 4A schematic diagram of experimental results provided by the present invention; Figure 5 This is a schematic diagram of the intra-class parameter inversion result provided by the present invention; Figure 6 A schematic diagram of soil moisture inversion results for a research area provided by this invention; Figure 7 A schematic diagram illustrating the influence of the k-value setting in cluster analysis provided by this invention on the soil moisture inversion method proposed in this study; Figure 8 This invention provides a schematic diagram illustrating the influence of window scale variation on the evolution of model parameters. Figure 9 This is a schematic diagram of a soil moisture inversion device provided by the present invention. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0016] Currently, researchers have developed a series of model systems to characterize the scattering properties of bare soil based on radiative transfer theory and multi-source observation data. These include models such as SPM, IEM, and AIEM, which describe the relationship between surface parameters and radar backscattering through electromagnetic scattering theory, as well as empirical / semi-empirical models such as the Dubois model, Oh model, and Shi model, which are built based on multi-band, multi-polarization, and multi-incident angle observation data. These models are all dedicated to establishing a quantitative correlation between parameters such as surface roughness and soil moisture and the backscattering coefficient. As the mainstream model for soil moisture inversion in bare soil areas, AIEM, with its rigorous electromagnetic scattering theoretical foundation and the coupled characterization of parameters such as soil moisture, temperature, texture (clay and sand content), and bulk density by the Dobson dielectric model, effectively solves the parameter adaptability problem for different soil types and exhibits excellent inversion performance under complex surface conditions.

[0017] The water cloud model decomposes the total backscattering coefficient of vegetated surfaces into two parts: scattering from vegetation and scattering from soil after vegetation attenuation. In actual inversion processes, both the AIEM and water cloud models face challenges in dynamic parameter updates and scale matching. Researchers have achieved more stable model calibration by introducing spatiotemporal constraints on model parameters. However, farmland areas have short crop growth cycles, frequent cultivation activities, and complex planting structures, making it difficult for satellite revisit cycles to accommodate the frequency of changes in crop parameters and surface roughness. The spatial scale and classification criteria of land cover type data are relatively coarse, often failing to characterize the spatial variability of crop types. Furthermore, while these assumptions minimize the number of parameters to be estimated, they still rely on in-situ soil moisture data or calibration using existing soil moisture remote sensing products. This limits the use of inversion models in data-scarce areas and introduces errors from soil moisture products.

[0018] To address the aforementioned issues, this invention constructs a parameter multi-scale constrained iterative inversion method that integrates multiple remote sensing radiative transfer models with high-resolution radar-optical remote sensing imagery. This method utilizes the high spatial resolution information provided by optical remote sensing data to characterize the spatial heterogeneity of crops, and applies joint constraints and inversion to model parameters and soil moisture at three scales: within vegetation classes, local windows, and global scope. The aim is to achieve stable inversion of model parameters and soil moisture at the farmland scale without relying on in-situ soil moisture data.

[0019] The technical solutions provided by the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0020] Figure 1 This is a schematic diagram of a soil moisture inversion method according to the present invention, which specifically includes the following steps: S101: Based on the atmospheric apparent reflectance data of the target area, determine the normalized vegetation index, leaf area index, average leaf tilt angle, and total backscattering coefficient of electromagnetic waves for the target area pixel by pixel.

[0021] S102: Cluster the normalized vegetation index, leaf area index and mean leaf tilt angle of the target area to determine the vegetation types and corresponding regions of the target area.

[0022] S103: Couple the soil backscattering model, soil dielectric constant model and water cloud model to obtain the influence relationship between surface roughness parameters, vegetation parameters and soil moisture on the total backscattering coefficient of electromagnetic waves.

[0023] S104: With the constraints that the surface roughness parameters and vegetation parameters of the same vegetation type are the same and the soil moisture is the same within the same surface space window, the soil moisture of each pixel is initialized. A system of simultaneous equations for the influence relationship at the intra-class scale is constructed for each vegetation type and solved to obtain the initial surface roughness parameters and vegetation parameters of each vegetation type.

[0024] S105: Construct a system of simultaneous equations to determine the influence relationship of each surface space window in the pre-divided target area, and substitute the initial surface roughness parameters and vegetation parameters of the corresponding vegetation type to solve the problem, thereby obtaining the updated soil moisture of each surface space window. Iteratively update and solve the surface roughness parameters and vegetation parameters of each vegetation type, and iteratively update and solve the soil moisture of each surface space window; coupled with leaf biochemical parameters, the surface space window corresponds to multiple adjacent pixels.

[0025] For ease of explanation, the following description focuses solely on the server as the executing entity. The server mentioned in this invention can be a server set up on a business platform, or a device such as a desktop computer or laptop computer capable of executing the solution of this invention.

[0026] This invention uses a specific research area as an example for illustration. This research area also serves as a test area for subsequent explanation of the invention's effectiveness. The research area has an altitude of 1032m–1050m and covers an area of ​​approximately 2.157 × 10⁵ hectares, including 1.543 × 10⁵ hectares of arable land. The soil is mainly silty loam, loam, and clay loam. Located in an arid, semi-arid, and semi-desert steppe zone, the area enjoys abundant sunshine, with an average annual sunshine duration of approximately 3180 hours and an average frost-free period of 130–168 days. The climate of this research area is predominantly temperate continental, with cold, snow-scarce winters and hot, dry summers. The average annual temperature is approximately 6.3℃–7.7℃, the average annual precipitation is 138.2 mm, and the average annual evaporation is 2096.4 mm. Agricultural production in the research area is highly dependent on irrigation. Irrigation in the area mainly relies on water from the Yellow River, with an annual water diversion volume reaching 1 billion–1.2 billion cubic meters.

[0027] Located in the southern part of the North China Plain and the eastern foothills of the Taihang Mountains, this study area belongs to the temperate continental monsoon climate zone, specifically the semi-humid climate zone. The soils are mainly loam and sandy loam, with some localized clay loam, and are deep, suitable for agricultural production. The climate is characterized by four distinct seasons: cold, dry winters, hot, rainy summers, and short, variable spring and autumn seasons. The average annual sunshine duration is approximately 2300 hours, and the average frost-free period is 190-220 days, providing sufficient light and heat for crop growth. The average annual temperature is approximately 13.0℃, and the average annual precipitation is 600-700 mm, with 60%-70% concentrated in summer. The average annual evaporation is 1500-2000 mm, slightly higher than precipitation. The agricultural irrigation system in this study area is relatively complete, with water sources including groundwater, reservoir storage, and irrigation from the Yellow River. Inter-regional water transfer is achieved through water conservancy projects such as the Hongqi Canal, with an annual water diversion volume of hundreds of millions of cubic meters. Yellow River irrigation plays a role in some areas. Figure 2 This is a schematic diagram showing the distribution of the research area and soil moisture observation points in this invention. Figure 2 The image shows two study areas and the distribution of their corresponding observation points.

[0028] This invention also uses four types of datasets as examples for soil moisture retrieval, namely: (1) active microwave remote sensing data from the Sentinel-1 GRD IW model; (2) atmospheric apparent reflectance data from the Sentinel-2 L1C model; (3) global gridded soil information from SoilGrids; and (4) surface soil moisture data from in-situ observations. As shown in Table 1.

[0029] Table 1 Datasets used for soil moisture inversion

[0030] The Sentinel-1 GRD IW mode active microwave remote sensing data includes two satellites, Sentinel-1A and Sentinel-1B. The revisit period for a single satellite is 12 days, which can be reduced to 6 days through dual-satellite complementarity, enabling the acquisition of satellite imagery with a resolution of 5-40 meters under all-weather, day-and-night conditions. The Sentinel-1 satellite carries a C-band Synthetic Aperture Radar (SAR) sensor operating at 5.4 GHz, featuring multi-polarization imaging capabilities and four imaging modes. The Sentinel-1 imagery used in this example is down-orbit data in the interferometric wide-swath mode, corresponding to a ground resolution of 5m × 20m, a data level of Level-1, and a Ground Range Detected (GRD) format. The temporal resolution within the study area is 12 days. Dual polarization includes cross-polarization (VH, Vertical horizontal) and vertical polarization (VV, Vertical vertical). Considering that VV polarization is more suitable for inverting exposed soil moisture, and that AIEM can effectively simulate the backscattering of VV polarization, this invention example selects only VV polarization as the total backscattering coefficient of electromagnetic waves in the target area. Data is acquired from the Earth Engine platform, subjected to Refined Lee filtering, and exported with a spatial resolution of 20m.

[0031] Sentinel-2 L1C Atmospheric Apparent Reflectance Data: Like Sentinel-1, the Sentinel-2 satellite system consists of a binary constellation, Sentinel-2A and Sentinel-2B. The revisit period for a single satellite is 10 days, which can be reduced to 5 days through binary satellite complementarity, resulting in a revisit period of 2-3 days within the study area. Sentinel-2 carries a multispectral imager with 13 spectral bands and a swath width of 290 km. Ground resolution varies from 10 m to 60 m depending on the band. The Sentinel-2 L1C (S2L1C) product was selected, and atmospheric corrections were applied based on existing research to obtain the S2L2A product. To match the Sentinel-1 data, the Sentinel-2 date was chosen based on the closest possible Sentinel-1 overpass date. Temporal linear interpolation and SG filtering were used to fill local gaps in the images after cloud removal using the QA60 band. In this invention, 10 bands (B02-B8A, B11, and B12) covering the spectral region from visible light to shortwave infrared were selected at spatial resolutions of 10m and 20m, and all were resampled to a spatial resolution of 20m. Simultaneously, Normalized Difference Vegetation Index (NDVI) raster data were calculated and output.

[0032] SoilGrids global gridded soil information: The relationship between soil dielectric constant and soil moisture depends on properties such as clay proportion, sand proportion, and soil bulk density. At the regional scale, the spatial heterogeneity of soils cannot be ignored. SoilGrids, as a global digital soil mapping system, uses state-of-the-art machine learning methods to depict the spatial distribution of global soil properties. The predictive model is fitted using over 230,000 soil profile observations from the WoSIS database and a series of environmental covariates. The covariates are selected from over 400 environmental layers from Earth observation derivatives and other environmental information, including climate, land cover, and topography. SoilGrids outputs a global soil property map with a spatial resolution of 250 meters at six standard depth intervals (0-5cm, 5-15cm, 15-30cm, 30-60cm, 60-100cm, and 100-200cm). This helps to more accurately characterize the spatial variability of soil properties. In this invention, three features—clay_mean, sand_mean, and soil bulk density (bdod_mean)—at a depth of 0-5 cm in SoilGrids are resampled to 20 m and used as covariates to achieve the conversion between soil moisture and dielectric constant.

[0033] In-situ observational data of surface soil moisture: Multiple soil volumetric water content measurements were conducted at various observation points in the study area. For example, in... Figure 2 Soil volumetric moisture content was measured twice at 100 observation points in study area 2, with a measurement depth of 5 cm in both measurements. This data will be used to verify the accuracy of soil moisture inversion.

[0034] Soil moisture inversion in farmland areas can be broken down into four problems: 1) parameterization of crop canopy; 2) extraction of soil backscattering components; 3) calculation of soil dielectric constant; and 4) conversion of soil moisture. This invention addresses these four problems specifically using the PROSAIL model, water cloud model, AIEM model, and Dobson model.

[0035] Among them, the PROSAIL model, as a widely used international vegetation canopy radiative transfer model, was developed by merging the PROSPECT leaf optical property model and the SAIL canopy radiative transfer model (Verhoef & Baret, 2009). This model accurately simulates canopy directional reflectivity within the wavelength range of 400nm to 2500nm by coupling leaf biochemical parameters (such as chlorophyll content and leaf structure parameters) with the three-dimensional geometry of the canopy (such as leaf area index and leaf tilt angle distribution). In this invention example, Sentinel-2 imagery is used as observation data. The leaf area index (LAI) and average leaf tilt angle (ALA) of the target region per pixel are used as target inversion parameters to construct a search population for target inversion parameters and establish an iterative inversion process that includes forward model simulation and optimization algorithms. Canopy directional reflectance can be simulated by coupling leaf biochemical parameters with the three-dimensional geometry of the canopy using the PROSAIL model. Based on observational data, the fitness values ​​of each individual in the search population are determined. The optimization objective is to minimize the spectral residual between the simulated canopy directional reflectance based on the search population and the observed data. Parameter optimization is then performed on the target inversion parameters. Specifically, Particle Swarm Optimization (PSO) can be used to minimize the spectral residual between the model simulation values ​​and the image observation values, searching for the optimal parameter combination in the PROSAIL model parameter space, ultimately obtaining the LAI and ALA thematic raster data for the study area. Furthermore, canopy water content (CWC) is defined as the product of leaf area index (LAI) and leaf water content (LWC). Therefore, the canopy water content of the study area can be obtained further through the parameter search results.

[0036] Researchers have proposed a water cloud model to estimate soil moisture in agricultural areas by studying the backscattering characteristics of crop-covered surfaces. This model posits that total backscattering from vegetated surfaces consists of two parts: volume scattering directly reflected from the vegetation canopy and ground backscattering after double attenuation by the canopy. The water cloud model is based on the following assumptions: 1) it assumes the vegetation layer is a horizontally uniform cloud layer; 2) it considers only single scattering, neglecting multiple scattering between the vegetation layer and the surface; and 3) the only variables considered in the model are vegetation water content, soil moisture, and incident angle.

[0037] The expression for the water cloud model is: ; ; .

[0038] In the formula, This is the total backscattering coefficient of electromagnetic waves; The backscattering coefficient of direct vegetation; The soil backscattering coefficient; This refers to the water content of vegetation. The vegetation double-layer attenuation factor (transmittance); The angle of incidence of the electromagnetic wave; , The value depends on the vegetation type and the frequency of the incident electromagnetic wave.

[0039] The Advanced Integral Equation Model (AIEM) is currently the mainstream physical model for simulating the microwave scattering characteristics of exposed surfaces. It is an improvement on the integral equation model. By improving the power spectrum of surface roughness and the Fresnel reflection coefficient, it can adapt to different surface roughness conditions and accurately reproduce the backscattering of real surfaces. It is widely used in the simulation and analysis of microwave surface radiation and scattering.

[0040] The backscattering form of AIEM can be generalized as follows: .

[0041] In the formula, The azimuth angle for backscattering is fixed at 180°. For the relevant length; The root mean square height; The complex permittivity of soil; The frequency of the electromagnetic wave is 5.4 GHz (a fixed value in this invention).

[0042] Regarding correlation length, the surface is not a completely random undulation; the heights of adjacent points often exhibit a certain correlation (for example, near the top of a small hill, adjacent points are likely also located on the same sloping section of the hill, rather than suddenly becoming a depression). Correlation length is the maximum distance at which this correlation "effectively exists," calculated mathematically (such as the "autocorrelation function").

[0043] Root mean square height ( ) and related length ( These two factors collectively describe the surface roughness and both have a certain influence on backscattering. A combined roughness... It was proposed as a more concise way to describe surface roughness and achieved good results. Here A value of 3 is acceptable. In this case, the AIEM model can be represented as: .

[0044] The Dobson model is a classic soil dielectric constant model designed to quantitatively describe the coupling relationship between radar incident wave frequency, soil physical properties, and soil moisture content. It is a crucial foundation for soil moisture retrieval in the field of microwave remote sensing. Currently used soil dielectric models are improved versions, further expanding their applicability and effectively characterizing the variation of soil complex dielectric constant with soil volumetric water content, electromagnetic wave frequency, soil texture (sand / clay content), soil temperature, and bulk density.

[0045] The Dobson model can be generalized as: .

[0046] In the formula, Soil temperature (which has little impact on the simulation results; this invention fixes it at 25°C); Soil moisture; The clay ratio; The proportion of sandy soil; It is the bulk density of the soil.

[0047] When performing soil moisture inversion using the above model, the coupling of the model in this invention is divided into four steps: 1) coupling AIEM and Dobson models; 2) simplifying the AIEM_Dobson model; 3) cascading water cloud models; 4) using canopy moisture content to proxy vegetation moisture content.

[0048] First, by coupling the AIEM and Dobson models, under specific soil properties, the Dobson model can convert soil moisture into soil complex permittivity, providing input parameters for the AIEM model, and thus establishing a theoretical relationship between soil moisture and soil backscattering coefficient.

[0049] At this time, the soil backscattering coefficient It can be represented as: .

[0050] AIEM_Dobson Model Simplification. The high nonlinearity of both the AIEM and Dobson models poses challenges to parameter inversion, thus necessitating the simplification of the AIEM_Dobson model. Existing research indicates that when... , , as well as When the parameters are constant, the above model can be simplified to: .

[0051] In the formula, , , and These are empirical parameters, estimated using the AIEM_Dobson model. The specific steps are as follows: Fix the AIEM_Dobson model... , , as well as Parameters, and iterate through and A simulated dataset is generated, and finally, the coefficients are calculated using the least squares method. (This is from an example of the invention.) , , as well as The parameters are provided by the angular properties of Sentinel-1 and SoilGrids data, respectively, and the study area is generated pixel by pixel. , , and Raster data. For , , as well as The determination of the incident angle of the electromagnetic wave. The clay ratio can be determined using microwave remote sensing data of the target area. Sand ratio and soil bulk density It can be determined by using partial data of the target area corresponding to the global gridded soil information.

[0052] Then, the water cloud model is connected in series, and the water cloud model contains... The parameters are simulated by the AIEM_Dobson model, and the above formulas can be rearranged to obtain: .

[0053] This formula describes the relationship between soil moisture and backscattering on vegetated surfaces.

[0054] Next, canopy water content was used to proxy vegetation water content. Sentinel-2 imagery was used as observation data, and the canopy water content (CWC) raster data of the study area was obtained through inversion using the PROSAIL model. Here, it is assumed that the vegetation water content (VWC) of the same crop is a linear function of the canopy water content (CWC). .

[0055] Substituting into the formula, we get: , , .

[0056] Right now, .

[0057] and The specific values ​​still depend on the vegetation type and the frequency of the incident electromagnetic waves; therefore, further research is needed. and These are considered as the transformed vegetation parameters.

[0058] Finally, parameter inversion is performed for a specific pixel. of All are known, and the parameters to be inverted are: For the study area There are a total of 100 pixels. There are equations, but there are Solving for unknown parameters is difficult. Therefore, it is necessary to add constraints to achieve parameter inversion.

[0059] Considering parameters and Depending on the vegetation type, it can be hypothesized that there are [various vegetation types] within the study area. Crops of different types, and each crop type shares the same set of parameters. and At the same time, each type of crop also shares the same parameter. This is because the surface roughness of farmland areas is primarily influenced by tillage practices, and similar crops require similar tillage methods. Due to the large-scale nature of surface runoff and water inflow events, it is assumed that... The pixels within the window have the same .

[0060] This invention posits that the influence of vegetation type on backscattering primarily stems from differences in canopy structure among different vegetation types. Canopy structure can be easily characterized using leaf area index (LAI) and mean leaf tilt angle. Furthermore, NDVI is used to describe differences in vegetation cover. Therefore, based on NDVI calculated using Sentinel-2 and LAI and ALA clustering obtained through PROSAIL inversion, the vegetation categories of the study area are classified as follows: Based on the above assumptions, the number of unknown parameters is reduced to [number missing]. ,like and The value makes This preliminarily satisfies the conditions for parameter inversion. In the embodiment of the research area of ​​this invention, the following will be used: Set to 5, Set it to 3.

[0061] Parameter inversion can be divided into two parts: intra-class parameters Inversion and window parameters The inversion is performed by a method that iteratively calculates the inversion results. Since these two parameters are not on the same spatial scale and the inversion results are interdependent, this invention proposes such a method. The specific steps are as follows: First, assume that the soil moisture in the study area is... At this time, the initial window parameters Then substitute it into the equation; secondly, solve the system of equations within the same class (intra-class scale) to solve for the intra-class parameters. Perform the same operation on each class; third, set the class parameters... Substitute back into the equations, and solve the system of equations within the window (window scale) to find the new window parameters. First, iterate through each window; second, determine all windows. and Is the maximum difference less than an acceptable threshold? If the condition is not met, update. Continue iterating if the condition is met. Otherwise, stop iterating and save the current value. Based on this, the backscattering coefficients are reconstructed across the entire image (global scale), and the reconstruction accuracy is calculated. Finally, the process is iterated through... The optimal value of the backscattering coefficient is selected, and the corresponding value is output. The parameter inversion is complete. A special case is when the canopy water content of a certain type is close to 0; this type is considered bare soil, and the parameters are directly set... Only the intra-class parameters are solved. The remaining steps remain unchanged. Figure 3 This is a schematic diagram of a specific data processing flow for soil moisture inversion in this invention.

[0062] Taking the aforementioned study area as an example, this invention simplifies the model pixel-by-pixel by fusing soil properties from SoilGrids and incident angles from Sentinel-1, achieving satisfactory results. The coefficient of determination between the simplified empirical model and the AIEM_Dobson model simulation results is almost above 0.8. Figure 4 As shown, Figure 4 This is a schematic diagram of an experimental result in this invention. Figure 4 In the table, (a) represents the known input parameters of the AIEM_Dobson model, (b) represents the simplified semi-empirical model coefficients, (c) represents the simplification effect of the semi-empirical model on the AIEM_Dobson model within the study area, and (d) represents the correlation between the input parameters and the semi-empirical coefficients.

[0063] observe Figure 4 In part (d), it can be observed that the coefficient 'a' of the semi-empirical model is mainly affected by... , as well as The coefficient b is mainly affected by parameters such as [parameter 'b'], and all of them are negatively correlated. and The impact, and its relationship Positive correlation with Negative correlation. Coefficient d also exhibits a similar response pattern to coefficient b, but coefficient b is more dependent on... The coefficient d depends more on The coefficient c is mainly affected by the parameter , and The impact, and its relationship Positive correlation with and Negative correlation. At the regional scale, all input parameters of the model exhibit spatial heterogeneity, which in turn leads to a certain degree of spatial heterogeneity in the simplified semi-empirical model coefficients. Although this characteristic is not obvious, the pixel-by-pixel operation avoids as much as possible the negative impact of using fixed model coefficients when performing soil moisture inversion at the regional scale.

[0064] Figure 5 This is a schematic diagram of an intraclass parameter inversion result in this invention. Figure 5 (ad) shows the clustering results for study area (1) on July 9, 2018, August 3, 2018, and September 3, 2019, and for study area (2) on August 29, 2024, as well as the inversion results for each category A', B', and Z. (e) shows the relationship between parameters A' and Z and NDVI, LAI, and ALA. The clustering results are only used as labels within a specific spatiotemporal range and are not universally applicable to different regions or different periods.

[0065] In this invention example, cluster analysis was used to divide the vegetation within a specific study area at a specific time into five categories. It can be observed that the distribution of each vegetation category within the study area exhibits a general clustering phenomenon, meaning that a certain type of vegetation is mainly concentrated within a specific area. This generally aligns with the characteristics of land types and farmers' planting preferences within the study area.

[0066] On July 10, 2018, the parameters A' of various vegetation types in study area (1) were -20.764, -19.126, -18.995 and -14.825, respectively. The parameters B' were 0.105, 0.210, 0.138 and 0.137, respectively. The parameters Z were 0.112, 0.066, 0.063, 0.033 and 0.207, respectively. Among them, category 5 was considered to have no vegetation cover due to the overall low canopy water content, and the parameters A' and B' were directly set to 0. On August 3, 2018, the parameters A' of various vegetation types in study area (1) were -23.746, -21.034, -20.712 and -15.255, respectively. The parameters B' were 0.099, 0.109, 0.101 and 0.074, respectively. The parameters Z were 0.505, 0.822, 0.366, 0.735 and 0.828, respectively. As on July 9, 2018, the parameters A' and B' of category 5 were directly set to 0. On September 3, 2019, the parameters A' of various vegetation types in the study area (1) were -21.309, -19.282, -16.289 and -15.502, respectively. The parameters B' were 0.031, 0.154, 0.075 and 0.176, respectively. The parameters Z were 0.027, 0.007, 0.009, 0.005 and 0.023, respectively. The parameters A' and B' of category 5 were directly set to 0. On August 29, 2024, the parameters A' of various vegetation types in study area (2) were -18.188, -15.679, -14.944, -12.615 and -11.274, respectively. The parameters B' were 0.110, 0.130, 0.136, 0.112 and 0.091, respectively. The parameters Z were 0.538, 0.299, 0.197, 0.180 and 0.332, respectively.

[0067] From the study area (1) July 10, 2018 ( Figure 5 (a) , August 3, 2018 ( Figure 5 (b) and September 3, 2019 ( Figure 4 From the clustering results and parameter inversion of (c) in the study area, the distribution of each category and the parameter values ​​differ between months and years. This reflects that the change in vegetation structure and coverage over time, coupled with agricultural activities, affects the parameters of the water cloud model, demonstrating the dynamic response of the model parameters to the temporal changes in farmland surface characteristics. The parameter inversion results of the study area (2) are as follows: Figure 4 The results in the middle (d) and the study area (1) at different time phases are significantly different. This is due to the differences in vegetation type and structure, soil properties, and farming patterns between the two areas, indicating that the parameters of the water cloud model are regionally specific and constrained by the comprehensive characteristics of the underlying surface. In summary, the same parameter varies at different times, locations, and vegetation types. Ignoring the spatiotemporal heterogeneity of the parameters may have a negative effect on the inversion of soil moisture.

[0068] This invention also calculates the mean of three canopy features (NDVI, LAI, and ALA) from the cluster analysis, attempting to explore their impact on parameters A' and B'. Figure 5 As shown in equation (d), parameter A' generally increases with increasing NDVI and LAI, but its relationship with ALA is not significant. Parameter B' generally increases with increasing NDVI and LAI and then tends to plateau. When LAI is the same, the smaller ALA is, the larger parameter B' is. However, when NDVI is approximately greater than 0.5, the effect of ALA on parameter B' is no longer significant. According to the equation, parameter A' is related to vegetation backscattering. NDVI and LAI represent vegetation density; the increase of parameter A' along with NDVI and LAI means that the greater the vegetation density, the greater the contribution of vegetation to radar wave backscattering. Parameter B' mainly reflects the attenuation of radar waves propagating in the vegetation layer. The larger the value of B', the more severe the attenuation of radar waves propagating in the vegetation layer, that is, the stronger the shielding effect of vegetation on radar waves. B' initially increases and then stabilizes with increasing NDVI and LAI, essentially due to the combined effect of the vegetation canopy's microwave attenuation mechanism and the canopy structure's saturation effect: When LAI is small, the canopy leaves are sparse, and microwave attenuation is mainly due to single scattering. As LAI increases, the absorption and scattering paths of microwaves by the leaves lengthen, leading to an increase in the B' value. When LAI exceeds a certain threshold, the overlapping canopy leaves form a closed layer, and microwaves are largely absorbed or scattered at the top of the canopy. The excess leaves only increase internal multiple scattering, weakening the overall attenuation efficiency improvement, and the B' value tends to stabilize. Parameter Z is mainly affected by cultivation practices.

[0069] Soil moisture inversion results: To verify the reliability of the method proposed in this invention, the inversion results were verified using in-situ observation data from study areas (1) and (2). At the same time, the method of this invention was compared across regions with the model calibrated using partial data from study area (2).

[0070] Figure 6This is a schematic diagram of soil moisture inversion results in a study area according to the present invention. (a) shows the spatial distribution of soil moisture at different times and locations obtained based on this method. (b) shows the relationship between the vegetation backscattering coefficient and the actual scattering coefficient in the forward simulation based on the soil moisture inversion results. (c) shows the soil moisture inversion accuracy based on multi-scale parameter optimization and unified parameter calibration. For the unified parameter calibration method, 50 measured data from study area (2) were used to calibrate the vegetation parameters. The remaining 50 data from study area (2) and all in-situ soil moisture data from study area (1) were used to verify the inversion accuracy of the two methods.

[0071] The surface soil moisture values ​​in the study areas at different times and locations were consistently distributed in the range of 0 to 0.4. Due to the influence of different crop irrigation systems, the irrigation time showed significant spatial heterogeneity, resulting in a strong correlation between soil moisture distribution patterns and vegetation types. Taking the study area (1) on July 9, 2018 as an example, the soil moisture in its northeastern region was relatively higher than that in other regions, forming a certain concentrated area of ​​soil moisture.

[0072] Based on the soil moisture and intraclass parameter inversion results, this study constructed a model to simulate the soil backscattering coefficient. The results show that, across the entire spatial and temporal scale, the model's coefficient of determination R² does not differ significantly and remains above 0.88, indicating that the model can effectively characterize soil backscattering characteristics when facing the spatial heterogeneity of different vegetation structures, and effectively reflects the influence mechanism of soil moisture and vegetation canopy characteristics on the backscattering coefficient.

[0073] The soil moisture inversion method constructed in this invention exhibits robust applicability at the global scale, with coefficients of determination (R²) of 0.265, 0.363, 0.367, and 0.379 in four scenarios, respectively. Root mean square errors (RMSE) are 0.083, 0.097, 0.087, and 0.127, respectively. In contrast, the unified parameter inversion model calibrated based on in-situ data from study area (2) achieves higher accuracy locally (R²=0.548, RMSE=0.096). However, its spatial generalization ability is significantly insufficient, with coefficients of determination of 0.105, 0.202, and 0.197 in three scenarios of study area (1), respectively. Root mean square errors are 0.120, 0.138, and 0.113, respectively, and the inversion accuracy is lower than that of the method proposed in this invention. This indicates that the spatiotemporal mobility of a single parameter calibrated based on a specific spatiotemporal period has significant limitations. The above phenomena reveal the limitations of relying on data from specific regions to calibrate parameters: the unified parameter system is highly dependent on the observation data of the study area (2), making it difficult to directly adapt to other geographical environments. Considering the high human and economic costs of large-scale in-situ data acquisition and the difficulty of implementation in areas with scarce ground observation data, the application scenarios of such calibration methods are significantly limited. In contrast, the inversion method proposed in this study breaks through the data dependence bottleneck and can still maintain stable performance under the condition of limited data acquisition, providing a more universal technical path for soil moisture inversion.

[0074] The rapid phenological processes of crops in farmland areas, coupled with dramatic changes in canopy structure across growth stages and significant differentiation in canopy characteristics among plots due to variations in farming systems, necessitate dynamic adjustments to model parameters characterizing vegetation scattering and attenuation properties. Simultaneously, frequent human activities such as tillage, irrigation, and harvesting alter surface microtopography, while natural processes like precipitation compaction cause short-term fluctuations in surface roughness. Differences in field-scale management practices further exacerbate spatial heterogeneity. These factors combined result in strong spatiotemporal variability in microwave scattering model parameters for farmland areas. Parameters calibrated based on data from specific regions struggle to adapt to dynamic changes under different crop types, growth stages, and management practices, leading to a significant decrease in inversion accuracy when applying models across regions or over long time series. This invention couples multiple remote sensing radiative transfer models and proposes multi-scale parameter constraints based on assumptions of parameter sharing within vegetation types and consistent soil moisture at the window scale, reducing the dimensionality of unknown parameters. By using high-resolution remote sensing data to finely characterize the spatial heterogeneity of crops, and combining iterative inversion mechanisms to achieve coupled solution of model parameters and soil moisture, the limitations of calibrating model parameters based on in-situ observation data are avoided, thereby improving the applicability of the model in farmland areas and the stability of parameter inversion.

[0075] To more clearly illustrate the systematic errors and applicability of the proposed method, we take the irrigation area (1) of the study area on July 10, 2018 as an example and explore it from three aspects: First, we analyze the sensitivity of the number of clusters (k value) to the accuracy of soil moisture inversion, and reveal the intrinsic relationship between vegetation classification granularity and model adaptability; Second, we explore the influence of window scale change on the evolution law of model parameters, and reveal the coupling response mechanism between the spatial heterogeneity assumption and the scattering process modeling.

[0076] Impact of Cluster Size: The algorithm proposed in this invention characterizes the differentiation of crop types in the study area by clustering LAI, ALA, and NDVI. However, the setting of the k value is subjective, and it is necessary to clarify the impact of the k value on soil moisture retrieval. Figure 7 This diagram illustrates the influence of the k-value setting in a cluster analysis method on the soil moisture retrieval method proposed in this study. (a) shows the coefficient of variation of soil moisture retrieved based on different k-values ​​within the study area. (b) shows the retrieval accuracy of soil moisture based on different k-values. (c) shows the model parameters retrieved when k is 10. (d) shows the variation of the coefficient of variation with leaf area index.

[0077] Figure 7 Figure (a) shows the coefficient of variation of soil moisture retrieved based on different k values ​​within the study area. The coefficient of variation ranges from 0 to 0.04, with relatively high values ​​concentrated mainly in the northeast corner. To further illustrate this phenomenon, Figure 7 Figure (d) shows the trend of the coefficient of variation with LAI. As LAI gradually increases, the coefficient of variation also increases. This is because when LAI approaches 0, radar backscattering at the surface is mainly caused by direct backscattering from bare soil, with vegetation playing a relatively minor role. The classification granularity of vegetation does not affect the applicability of the model. As LAI gradually increases, the weight of vegetation in backscattering becomes larger. Accurately characterizing the heterogeneity of vegetation parameters (A', B') in the model helps the model to reasonably reproduce the radar backscattering process. Therefore, the coefficient of variation of soil moisture retrieved based on different k values ​​increases with increasing LAI.

[0078] Figure 7Figure (b) shows the model's accuracy in determining soil moisture as a function of the k value. It can be observed that as the k value increases, the accuracy of soil moisture retrieval initially increases and then plateaus (R² initially increases and then plateaus, RMSE initially decreases and then plateaus). This indicates that the model performs stably when faced with the uncertainty of k value selection and does not require overly detailed differentiation of vegetation within the study area. This is mainly because when k is less than a certain threshold, the differences in canopy characteristics between different vegetation types are still significant. Increasing the k value at this point helps the algorithm analyze the spatial heterogeneity of the model parameters. However, when k exceeds a certain threshold, the differences in canopy characteristics between some vegetation types become less significant, and the corresponding heterogeneity of the model parameters also becomes insignificant. Figure 7 As shown in (c), when k is set to 10, the model parameters for categories 3 and 4 are very similar, as are those for categories 5 and 6. At this point, increasing the k value gradually reduces the benefit to model accuracy and increases computational resource consumption. In this study, Figure 7 The soil moisture variation shown in (a) is also mainly attributed to the change in k value between 1 and 5. In practical applications, a larger k value can be conservatively selected first to avoid insufficient characterization of parameter heterogeneity, and then the value of k can be appropriately adjusted according to the planting structure in the region, thereby balancing the accuracy of the model and the computational cost.

[0079] Impact of Window Size: The algorithm proposed in this invention assumes consistent soil moisture within the same window, thus reducing the number of model parameters. This makes the solution of the model equations more stable, the parameters more easily converge, and improves the mathematical recognizability of the inversion. However, as the window size increases, this assumption gradually deviates from the actual spatial distribution characteristics of soil moisture. Actual surface soil moisture exhibits spatial heterogeneity, and this difference cannot be accurately expressed within a large window range using fixed SM parameters, limiting the model's ability to interpret backscattering. Therefore, it is necessary to investigate the impact of window size on model parameter inversion. This invention sets window sizes from 2 to 7 with a step size of 1 for parameter inversion. The results show that during parameter inversion, to reduce the structural error caused by the deviation between the above assumptions and reality, the model parameters exhibit a "statistical compensation mechanism."

[0080] Figure 8 This diagram illustrates the influence of window scale variation on the evolution of model parameters in this invention. (a) shows the vegetation scattering coefficient A' retrieved based on different window scales. (b) shows the vegetation attenuation coefficient B' retrieved based on different window scales. (c) shows the surface roughness Z retrieved based on different window scales. (d) shows the variation of the accuracy of the model reconstruction backscattering coefficients with window scale.

[0081] Specifically, the water cloud model parameters A' (vegetation scattering coefficient) and B' (vegetation attenuation coefficient) exhibit a monotonically decreasing and stabilizing characteristic as the window size increases. Figure 8 (a, b)). This trend reflects that the model reduces the weight of the vegetation term in the total scattering, thereby decreasing its sensitivity to spatial variations and avoiding the erroneous attribution of spatial heterogeneity of soil moisture to vegetation factors. This does not mean that the canopy physical processes themselves are weakened, but rather that in statistical fitting, the model chooses to suppress the canopy parameter as a compensatory strategy to fit the simplified SM assumptions. As the window continues to increase, the model gradually reaches the optimal balance between the contribution of the vegetation term and error control, thus both A' and B' exhibit convergence characteristics.

[0082] Meanwhile, the surface roughness parameter Z is also affected by the accuracy of the SM assumption, and its value shows an increasing trend in the early stage of window expansion for most surface types. Figure 8 (c) Since Z and SM are modeled together as interaction terms, this phenomenon can be understood as the model enhancing its ability to explain the spatial variation of SM by improving Z, thereby compensating to some extent for the model residuals caused by the decrease in soil moisture accuracy. However, as the window continues to expand, the actual expressive power of SM-related terms gradually decreases, the "compensation space" of Z tends to saturate, the parameter estimation enters a stationary state, and thus tends to stabilize at large scales.

[0083] However, for bare soil, Z shows a decreasing trend as the window size increases, revealing the difference in model error compensation mechanisms across different surface types. In bare soil areas, backscattering is mainly controlled directly by surface roughness and soil moisture, with vegetation having a very weak or even negligible impact. With a larger window, the contribution of the SM assumption error to total scattering becomes more sensitive on this type of surface. However, due to the lack of a vegetation term to adjust for bare soil surfaces, the model must suppress Z to reduce the risk of overfitting the roughness term in the SM error interpretation, thus avoiding an overestimation of scattering changes caused by surface structure. In other words, in the absence of redundant variables for error absorption, the model maintains overall fitting stability by weakening the effect of the roughness scattering term.

[0084] During soil moisture inversion with different window sizes, the accuracy of the model reconstructed backscattering coefficient showed a typical trend of "first increasing and then decreasing". Figure 8(d) This phenomenon intuitively reflects the dynamic trade-off between parameter estimation stability and model structural error. In the early stage of window expansion, parameter estimation gradually stabilizes and the error absorption mechanism is enhanced, leading to an improvement in model fitting accuracy. However, as the window continues to expand, the heterogeneity of the SM space can no longer be offset by the simple "parameter compensation" mechanism, and the accumulation of structural error leads to a decrease in reconstruction accuracy. The window size at which the model accuracy reaches its peak coincides with the inflection point where the changes in parameters (especially A', B', and Z) tend to stabilize, suggesting that in this study region, this scale may be the optimal window size for balancing physical consistency and statistical fit optimality.

[0085] In summary, the impact of window size variation on model parameter estimation reflects the spatial heterogeneity of soil moisture, the compensation effect in the parameter estimation process, and the complex relationship of coupling different scattering mechanisms in statistical modeling. This phenomenon highlights the necessity of appropriately selecting the window size in microwave remote sensing soil moisture inversion to improve the physical reliability and robustness of the model.

[0086] based on Figure 1 The soil moisture retrieval method presented in this invention combines microwave remote sensing data and atmospheric apparent reflectance data of the target area to obtain the total backscattering coefficient, normalized vegetation index, leaf area index, and mean leaf tilt angle of electromagnetic waves per pixel in the target area. Then, based on the normalized vegetation index, leaf area index, and mean leaf tilt angle, the vegetation is spatially clustered to capture the spatial heterogeneity of crop types. Subsequently, based on the constraints that the surface roughness parameters and vegetation parameters of the same vegetation type are the same, and the soil moisture is the same within the same surface spatial window, a two-stage iterative retrieval at the intra-class scale and the surface spatial window scale is performed. This method adapts to temporal dynamics and replaces the parameter calibration function of in-situ soil moisture data, allowing the determination of model parameters and soil moisture without the need for in-situ soil moisture data. This invention improves the accuracy and reliability of the soil moisture retrieval method by accurately characterizing the spatiotemporal variation of parameters while avoiding the errors introduced by in-situ soil moisture data.

[0087] Existing technologies for microwave remote sensing inversion of farmland soil moisture face three major technical bottlenecks. This solution addresses these bottlenecks through a layered design of "model coupling - multi-scale constraints - iterative inversion," with the specific solution logic as follows: Technical bottleneck 1. Decoupling of scattering signals in vegetation-covered areas requires in-situ data to calibrate model parameters (such as vegetation parameters and surface roughness in water cloud models). However, the prior information of limited in-situ data cannot adapt to the dynamic changes in farmland, leading to the failure of inversion in areas with scarce data.

[0088] The solution logic of this invention is as follows: by using the triple scale constraint of "vegetation clustering - local window - global verification", the parameter calibration function of in-situ soil moisture data is replaced, and the model parameters and soil moisture (SM) can be determined without in-situ soil moisture data.

[0089] Technical bottleneck 2. It is assumed that the model parameters (vegetation parameters and surface roughness) are stable in time and space (e.g., uniform vegetation parameters and fixed roughness). However, the vegetation parameters in farmland change dynamically with crop phenology (NDVI / LAI changes) and the surface roughness varies in time and space with cultivation activities (e.g., irrigation and tillage), which leads to the inversion results deviating from reality.

[0090] The solution logic of this invention is: to accurately characterize the spatiotemporal variation features of parameters by "clustering and differentiation to capture spatial heterogeneity + iterative inversion to adapt to temporal dynamics".

[0091] Technical bottleneck 3. Parameter inversion is ill-conditioned (unknowns > number of equations), making it difficult for the model to converge.

[0092] The solution logic of this invention is to achieve stable convergence of the model by "reducing dimensionality through parameter constraints and reducing nonlinearity through model simplification".

[0093] This invention couples multiple remote sensing radiative transfer models and integrates high-resolution radar and optical remote sensing imagery to construct an iterative inversion framework based on multi-scale parameter constraints. This method utilizes optical remote sensing data to characterize the spatial heterogeneity of crops at a fine scale and jointly solves for model parameters and soil moisture at three scales: vegetation type, local window, and global scope. Stable inversion of soil moisture and model parameters is achieved without relying on in-situ observations.

[0094] The method proposed in this invention performs better in soil moisture inversion in areas with scarce data: compared with the unified parameter model that relies on in-situ data calibration, its inversion results in study areas (1) and (2) (R² is 0.265-0.379, RMSE is 0.083-0.127) are not only more accurate, but also show stronger spatial generalization ability.

[0095] The water cloud model parameters (A', B') and the surface roughness parameter (Z) exhibit significant spatiotemporal variations. Parameter A' increases with increasing NDVI and LAI, while parameter B' initially increases and then stabilizes with increasing NDVI and LAI.

[0096] The number of clusters and the window size have a significant impact on the inversion accuracy: the accuracy first increases and then stabilizes as the k value increases; there is an optimal window size that balances the assumption of spatial heterogeneity of soil moisture with the stability of parameter estimation, at which the accuracy of the model reconstructing backscattering coefficients is the highest.

[0097] The above describes a soil moisture inversion method provided by one or more embodiments of the present invention. Based on the same idea, the present invention also provides a corresponding soil moisture inversion device, such as... Figure 9 As shown.

[0098] Figure 9 A schematic diagram of a soil moisture inversion device provided by the present invention includes: The preliminary determination module 201 is used to determine the total backscattering coefficient of electromagnetic waves in the target area pixel by pixel based on microwave remote sensing data of the target area; and to determine the normalized vegetation index, leaf area index and mean leaf tilt angle of the target area pixel by pixel based on atmospheric apparent reflectance data of the target area. Clustering module 202 is used to cluster the normalized vegetation index, leaf area index and mean leaf tilt angle of the target area to determine the vegetation types and corresponding areas of the target area. The coupling module 203 is used to couple the soil backscattering model, the soil dielectric constant model and the water cloud model to obtain the influence relationship between the surface roughness parameters, vegetation parameters and soil moisture of the vegetation-covered surface and the total backscattering coefficient of electromagnetic waves. The intra-class scale solution module 204 is used to initialize the soil moisture of each pixel under the constraints that the surface roughness parameters and vegetation parameters of the same vegetation type are the same and the soil moisture within the same surface space window is the same. It constructs and solves a system of simultaneous equations for the influence relationship of intra-class scale for each vegetation type to obtain the initial surface roughness parameters and vegetation parameters of each vegetation type. The surface space window corresponds to multiple adjacent pixels. The window scale solution module 205 is used to construct a set of simultaneous equations on the influence relationship of each surface space window in the pre-divided target area, and substitute the initial surface roughness parameters and vegetation parameters of the corresponding vegetation type to solve for the updated soil moisture of each surface space window. This allows for iterative updating and solving of the surface roughness parameters and vegetation parameters of each vegetation type, as well as iterative updating and solving of the soil moisture of each surface space window.

[0099] Specific limitations regarding the soil moisture inversion device can be found in the limitations of the soil moisture inversion method described above, and will not be repeated here. Each module in the aforementioned soil moisture inversion device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0100] The present invention also provides a computer-readable storage medium storing a computer program that can be used to execute the above-described... Figure 1 The provided method for soil moisture inversion.

[0101] This invention also provides a computer device. At the hardware level, the computer device includes a processor, an internal bus, a network interface, memory, and non-volatile memory, and may also include other hardware required for various operations. The processor reads the corresponding computer program from the non-volatile memory into memory and then executes it to achieve the above-mentioned functions. Figure 1 The provided method for soil moisture inversion.

[0102] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory may include random access memory (RAM) or external cache memory.

[0103] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this invention.

Claims

1. A method for soil moisture inversion, characterized in that, include: Based on microwave remote sensing data of the target area, the total backscattering coefficient of electromagnetic waves in the target area is determined pixel by pixel; based on atmospheric apparent reflectance data of the target area, the normalized vegetation index, leaf area index and mean leaf tilt angle of the target area are determined pixel by pixel. Clustering was performed on the normalized vegetation index, leaf area index and mean leaf tilt angle of the target area to determine the vegetation types and corresponding regions of the target area. By coupling the soil backscattering model, soil dielectric constant model and water cloud model, the influence relationship between surface roughness parameters, vegetation parameters and soil moisture on the total backscattering coefficient of electromagnetic waves on vegetated surfaces is obtained. With the constraints that the surface roughness parameters and vegetation parameters of the same vegetation type are the same and the soil moisture is the same within the same surface space window, the soil moisture of each pixel is initialized. A system of simultaneous equations on the influence relationship at the intra-class scale is constructed for each vegetation type and solved to obtain the initial surface roughness parameters and vegetation parameters of each vegetation type; the surface space window corresponds to multiple adjacent pixels. For each surface space window in the pre-defined target area, a system of simultaneous equations is constructed to determine the influence relationship of the window scale. The initial surface roughness parameters and vegetation parameters of the corresponding vegetation types are substituted into the equations to obtain the updated soil moisture of each surface space window. The surface roughness parameters and vegetation parameters of each vegetation type are then iteratively updated and solved, and the soil moisture of each surface space window is also iteratively updated and solved.

2. The soil moisture inversion method as described in claim 1, characterized in that, The step of determining the leaf area index and average leaf tilt angle of the target area pixel by pixel based on the atmospheric apparent reflectance data of the target area specifically includes: The atmospheric apparent reflectance data of the target area is used as the observation data, and the leaf area index and average leaf tilt angle of the target area per pixel are used as the target inversion parameters to construct a search population of target inversion parameters. The canopy directional reflectance is simulated by coupling leaf biochemical parameters with the three-dimensional geometry of the canopy using the PROSAIL model. The fitness value of each individual in the search population is determined based on the observation data. The optimization objective is to minimize the spectral residual between the simulated canopy directional reflectance based on the search population and the observed data, and the target inversion parameters are optimized.

3. The soil moisture inversion method as described in claim 1, characterized in that, The coupled soil backscattering model, soil dielectric constant model, and water cloud model yield the influence relationship between surface roughness parameters, vegetation parameters, and soil moisture on the total backscattering coefficient of electromagnetic waves, specifically including: A backscattering model of the soil is constructed based on an advanced integral equation model using the following formula: ； The soil dielectric constant model is constructed using the following formula: ； The water cloud model is constructed using the following formula: ， ， ； When the angle of incidence of electromagnetic waves , clay ratio Sand ratio and soil bulk density When determined, the soil backscattering model and the soil dielectric constant model are coupled and simplified by the following equation: ， ； By connecting the simplified coupled model to the water cloud model using the following formula, the influence relationship between surface roughness parameters, vegetation parameters, and soil moisture on the total backscattering coefficient of electromagnetic waves on vegetated surfaces is obtained: ； In the formula, The soil backscattering coefficient, The angle of incidence of the electromagnetic wave. The azimuth angle of the scattering. For the relevant length, The root mean square height, The complex permittivity of soil, The frequency of electromagnetic waves, Represents a high-level integral equation model; For soil temperature, For soil moisture, The clay ratio, The proportion of sandy soil, For soil bulk density, express Dobson Model; The total backscattering coefficient of electromagnetic waves is denoted as . The backscattering coefficient of direct vegetation. For vegetation water content, As a vegetation double-layer attenuation factor, The volume scattering coefficient, The attenuation coefficient; , , and These are empirical parameters.

4. The soil moisture inversion method as described in claim 3, characterized in that, The moisture content of the vegetation is expressed by the following formula: ;in, For vegetation water content, Canopy water content, The coefficients are linear. Based on the linear relationship between vegetation water content and canopy water content, the influence relationship is expressed as follows: ， ， ； The method further includes: Based on the atmospheric apparent reflectance data of the target area, the canopy water content of the target area is determined pixel by pixel.

5. The soil moisture inversion method as described in claim 3, characterized in that, Determine the empirical parameters , , and Specifically, it includes: The coupled model is obtained by coupling the soil backscattering model and the soil dielectric constant model using the following equation: ； According to the pre-determined coupling model per pixel , , as well as Parameters, iterating through parameters and The value obtained and and The corresponding simulated dataset is fitted to the coupled and simplified model using the least squares method to obtain the empirical parameters per pixel. , , and .

6. The soil moisture inversion method as described in claim 5, characterized in that, The electromagnetic wave incident angle Determined using microwave remote sensing data of the target area; The clay ratio Sand ratio and soil bulk density The target area was determined using partial data from globally gridded soil information.

7. A soil moisture inversion device, characterized in that, include: The preliminary determination module is used to determine the total backscattering coefficient of electromagnetic waves in the target area pixel by pixel based on microwave remote sensing data of the target area; and to determine the normalized vegetation index, leaf area index and mean leaf tilt angle of the target area pixel by pixel based on atmospheric apparent reflectance data of the target area. The clustering module is used to cluster the normalized vegetation index, leaf area index and mean leaf tilt angle of the target area to determine the vegetation types and corresponding areas of the target area. The coupling module is used to couple the soil backscattering model, soil dielectric constant model and water cloud model to obtain the influence relationship between surface roughness parameters, vegetation parameters and soil moisture on the total backscattering coefficient of electromagnetic waves. The intra-class scale solution module is used to initialize the soil moisture of each pixel under the constraints that the surface roughness parameters and vegetation parameters of the same vegetation type are the same and the soil moisture within the same surface space window is the same. It constructs and solves a system of simultaneous equations for the influence relationship of intra-class scale for each vegetation type to obtain the initial surface roughness parameters and vegetation parameters of each vegetation type. The surface space window corresponds to multiple adjacent pixels. The window scale solution module is used to construct a system of simultaneous equations on the influence relationship of each surface space window in the pre-divided target area, and substitute the initial surface roughness parameters and vegetation parameters of the corresponding vegetation type to solve for the updated soil moisture of each surface space window. This allows for iterative updating and solving of the surface roughness parameters and vegetation parameters of each vegetation type, as well as the soil moisture of each surface space window.

8. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the method as described in any one of claims 1 to 6.

9. A computer device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method as described in any one of claims 1 to 6.

Images