A soil water and salt dynamic monitoring and quality evaluation method based on multispectral remote sensing

Abstract

The application discloses a kind of soil water salt dynamic monitoring and quality evaluation method based on multispectral remote sensing, it is related to geological monitoring field, the method includes by satellite remote sensing image, unmanned aerial vehicle multispectral image, ground sample and meteorological driving data, scattering correction and roughness correction are carried out to observation spectrum, obtain soil intrinsic reflectivity, vegetation cover component and salt crust component;Water salt inversion model is established in combination with thermal infrared and red edge band information, and multi-time phase soil volume moisture content and surface salt index are inverted;By introducing meteorological conditions and vegetation dynamic characteristics, water salt coupling partial differential equation and graph time and space constraint model are constructed, and the continuous time and space distribution of soil moisture content and salt is obtained;Salinity, water, vegetation response and soil health index are extracted, and soil quality comprehensive evaluation model is established, and high-risk patch and management priority order are output.The method realizes the dynamic monitoring and quality classification evaluation of soil water salt, and is suitable for saline-alkali soil management and ecological restoration.

Description

Technical Field

[0001] This specification relates to the field of geological monitoring, and more specifically, this application relates to a method for dynamic monitoring and quality assessment of soil water and salt based on multispectral remote sensing. Background Technology

[0002] Existing soil water and salinity monitoring and quality assessment technologies mainly rely on ground sampling and experimental analysis methods, such as soil electrical conductivity measurement, volumetric water content detection, and chemical analysis. Although these methods have high accuracy, they suffer from typical "sampling limitations" and "time lag" problems: firstly, the number of sampling points is limited, making it difficult to reflect the spatial heterogeneity of water and salinity distribution at the regional scale; secondly, the experimental detection cycle is long, which cannot meet the need for continuous monitoring of dynamic changes in soil water and salinity.

[0003] Furthermore, while traditional remote sensing inversion methods can achieve large-scale estimations, they are typically based on single-temporal images or single spectral indices, neglecting the influence of temporal information and meteorological driving factors, resulting in poor model stability and insufficient accuracy. Simultaneously, existing methods often fail to consider spectral interference factors such as surface roughness and differences in incident angle, leading to systematic biases in the retrieved soil reflectance, which in turn affects the accuracy of water and salt estimations. Moreover, existing soil quality assessment methods mostly rely on single indicators (such as salinity concentration or water content) for classification and discrimination, lacking a comprehensive consideration of multidimensional ecological factors, making it difficult to reveal the spatial distribution and evolutionary trends of soil degradation risks.

[0004] Therefore, it is necessary to propose a method for dynamic monitoring and quality assessment of soil water and salt based on multispectral remote sensing to solve some of the above problems. Summary of the Invention

[0005] The summary section introduces a series of simplified concepts, which will be further explained in detail in the detailed description section. This summary section is not intended to limit the key and essential technical features of the claimed technical solution, nor is it intended to determine the scope of protection of the claimed technical solution.

[0006] Firstly, this application proposes a method for dynamic monitoring and quality assessment of soil water and salt based on multispectral remote sensing, including:

[0007] Acquire multi-source, multi-temporal, multispectral data, ground sample data, and meteorological driving data; wherein, the multi-source, multi-temporal, multispectral data includes satellite remote sensing imagery and UAV multispectral camera imagery, the ground sample data includes volumetric water content, surface and profile electrical conductivity and electrical conductivity ratio, soil texture and bulk density, and the meteorological driving data includes precipitation and evapotranspiration.

[0008] The above-mentioned multi-source, multi-temporal, multispectral data, ground sample data, and meteorological driving data are assimilated and registered on a spatiotemporal scale.

[0009] Based on the assimilated and registered multi-temporal multispectral data and ground sample data, a three-domain spectral component unmixing model for soil, vegetation and salt crust was constructed. Anisotropic scattering correction and roughness correction were performed on the observed spectra in the above multi-temporal multispectral data to obtain the soil intrinsic reflectance, vegetation cover component and salt crust component.

[0010] Based on the above-mentioned soil intrinsic reflectance, vegetation cover component, and salt crust component, and combined with the thermal infrared and red edge band information in the above-mentioned satellite remote sensing data, a water and salt inversion model is established to obtain multi-temporal soil water and salt inversion results.

[0011] Using the above multi-temporal water and salt inversion results as observation input, and combining the temporal variation of the above vegetation cover component with the assimilated and registered meteorological driving data, a water and salt coupled partial differential equation and a graph spatiotemporal canonical constraint model are constructed. Through joint optimization of the physical domain and the data domain, the continuous spatiotemporal distribution field of soil moisture and salinity is obtained.

[0012] Based on the above continuous spatiotemporal distribution field, salinity and moisture characteristic indicators are extracted, vegetation response indicators are extracted based on the temporal changes of the above vegetation cover components, and soil health indicators are determined based on the above intrinsic soil reflectance.

[0013] Based on the above-mentioned salinity, moisture, vegetation response, and soil health indicators, a comprehensive soil quality evaluation model is constructed to quantitatively score and classify the soil quality in different regions, and output high-risk patches, hotspot migration paths, and priority order for remediation.

[0014] In one feasible implementation, based on the assimilated and registered multi-temporal multispectral data and ground sample data, a three-domain spectral component unmixing model for soil, vegetation, and salt crust is constructed. Anisotropic scattering correction and roughness correction are then performed on the observed spectra in the multi-temporal multispectral data to obtain the soil intrinsic reflectance, vegetation cover component, and salt crust component, including:

[0015] Based on the multi-temporal multispectral data after assimilation and registration, the observation spectral reflectance information of each observation pixel is extracted, and combined with the typical soil, vegetation and salt crust sample spectra obtained from the above ground sample data, the standard spectral characteristics of the three types of endmembers are determined.

[0016] Using the above standard spectral features as input, a mixed spectral expression composed of soil endmembers, vegetation endmembers, and salt crust endmembers is established to represent the compositional relationship of the spectra of each pixel.

[0017] By decomposing and solving the above mixed spectral expression, the component weights of the soil endmember, vegetation endmember, and salt crust endmember in each pixel are obtained, thus obtaining the preliminary spectral component ratio.

[0018] To address the differences in reflection of the observed spectra under different incident and observation angles, anisotropic scattering correction calculations are performed to correct the reflection errors caused by the solar elevation angle and sensor viewing angle.

[0019] The surface roughness of the spectral data after anisotropic scattering correction is corrected to eliminate the influence of surface micromorphological differences on the spectrum, and the above-mentioned soil intrinsic reflectance, vegetation cover component and salt crust component are obtained.

[0020] In one feasible implementation, the component weights of the soil endmember, vegetation endmember, and salt crust endmember in each pixel are obtained by decomposing and solving the above-mentioned mixed spectral expression, thereby obtaining a preliminary spectral component ratio, including:

[0021] The above mixed spectrum expression is transformed into a matrix equation in matrix form, wherein the above matrix equation uses the above observed spectrum as the observation vector and the matrix composed of the basis vectors of each endmember spectrum as the coefficient matrix;

[0022] The least squares method is used to solve the above matrix equation and invert it to obtain the initial values ​​of the component weights of each end element.

[0023] The initial values ​​of the above component weights are subjected to non-negativity constraints and normalization to make the sum of the component weights of the soil end-member, the vegetation end-member, and the salt crust end-member equal to one.

[0024] The output is the component weight result after constraint optimization, which serves as the preliminary spectral component ratio.

[0025] In one feasible implementation, the anisotropic scattering correction operation is performed to correct the reflection error caused by the solar elevation angle and sensor viewing angle, based on the reflection differences of the observed spectrum under different incident and observation angles, including:

[0026] Based on the imaging geometric information of the above multi-temporal multispectral data, the geometric parameters at the corresponding time are obtained, including the solar incidence angle, azimuth angle and sensor observation angle parameters.

[0027] Based on the above geometric parameters, the relative observation geometric relationship of each pixel is calculated, and the reflection anisotropic response function is established. The original reflectance of the above observation spectrum is normalized by angle using the above anisotropic response function to obtain the corrected reflectance under the conditions of standard incident angle and observation angle.

[0028] The above-mentioned corrected reflectivity output is used as input data for subsequent surface roughness correction to correct the reflection error caused by the solar elevation angle and sensor viewing angle.

[0029] In one feasible implementation, the surface roughness correction is applied to the spectral data after anisotropic scattering correction to eliminate the influence of surface micromorphological differences on the spectrum, resulting in the soil intrinsic reflectance, vegetation cover component, and salt crust component, including:

[0030] Based on the above ground sample data and digital elevation model data, surface roughness characteristic parameters of each region are extracted. Among them, the above surface roughness parameters include slope, aspect and surface microstructure scale.

[0031] Based on the above surface roughness characteristic parameters, a roughness influence function and a roughness correction coefficient are constructed;

[0032] By combining the spectral data after the anisotropic scattering correction and the roughness influence function, the spectral reflectance is roughened to compensate for the spectral scattering distortion caused by the surface micromorphology.

[0033] Output the roughness-corrected spectral data to obtain the above-mentioned soil intrinsic reflectance, the above-mentioned vegetation cover component, and the above-mentioned salt crust component.

[0034] In one feasible implementation, based on the aforementioned intrinsic soil reflectance, vegetation cover component, and salt crust component, and combined with thermal infrared and red-edge band information from the aforementioned satellite remote sensing data, a water-salt inversion model is established to obtain multi-temporal soil water-salt inversion results, including:

[0035] Based on the above-mentioned intrinsic soil reflectance, spectral characteristic parameters in the near-infrared and short-wave infrared bands were extracted.

[0036] By jointly analyzing the above spectral characteristic parameters and the thermal infrared band information in the above satellite remote sensing data, the surface temperature characteristic quantity is calculated.

[0037] The salt sensitivity index is extracted using the red-edge band information mentioned above, and the salt crust component is introduced as a constraint factor to correct the spectral aliasing error in salt inversion.

[0038] The above-mentioned spectral characteristic parameters, surface temperature characteristic quantities, and salinity sensitivity characteristic indices are input into the multi-index collaborative inversion model. Soil volumetric water content and surface salinity index are calculated through statistical regression to generate preliminary inversion results.

[0039] The above preliminary inversion results were subjected to time series consistency correction to maintain the physical continuity of water and salt change trends between adjacent time phases, thus obtaining the above multi-temporal soil water and salt inversion results.

[0040] In one feasible implementation, the aforementioned multi-temporal water-salt inversion results are used as observation inputs. Combined with the temporal variations of the vegetation cover component and the assimilated and registered meteorological driving data, a water-salt coupled partial differential equation and a graph-spatiotemporal canonical constraint model are constructed. Through joint optimization in the physical and data domains, a continuous spatiotemporal distribution field of soil moisture and salinity is obtained, including:

[0041] Using the above multi-temporal water and salt inversion results as observation inputs, combined with the temporal changes of the above vegetation cover components and the assimilated and registered meteorological driving data, the above precipitation is used as the boundary condition for infiltration and water replenishment, and the above evapotranspiration is used as the surface water loss term to construct a water-salt coupled partial differential equation and a graph-temporal regular constraint model.

[0042] By jointly optimizing the physical and data domains, a continuous spatiotemporal distribution field of soil moisture content and salinity driven by meteorological conditions is obtained.

[0043] In one feasible implementation, a comprehensive soil quality evaluation model is constructed based on the aforementioned salinity characteristic indicators, moisture characteristic indicators, vegetation response indicators, and soil health indicators. This model quantitatively scores and classifies the soil quality of different regions, outputting high-risk patches, hotspot migration paths, and remediation priorities, including:

[0044] The above-mentioned salinity and moisture characteristics were correlated with the vegetation response indicators obtained from the temporal changes of the above-mentioned vegetation cover components.

[0045] Based on the above soil health indicators, the above salinity characteristic indicators, the above water characteristic indicators, and the above vegetation response indicators were normalized, and a soil quality evaluation characteristic matrix was established.

[0046] Based on the soil quality evaluation feature matrix mentioned above, a weighted comprehensive evaluation algorithm is used to construct the soil quality comprehensive evaluation model and calculate the comprehensive soil quality score for each region.

[0047] Based on the comprehensive soil quality score, the soil was classified and spatially clustered to identify the high-risk patches, hotspot migration paths, and a priority order for remediation.

[0048] In one feasible implementation, the steps for determining the aforementioned high-risk plaques include:

[0049] Based on the above comprehensive soil quality score, the score distribution characteristics of each monitoring area were statistically analyzed, and the statistical results of the above scores were calculated. The above statistical results include the mean, standard deviation, and extreme value range.

[0050] Based on the above statistical results, a grading threshold range is set, the above comprehensive soil quality score is divided into multiple levels, and areas in the low score range are marked as potential high-risk areas.

[0051] The soil quality grade distribution layer is generated by spatially encoding and vectorizing the regions at each grade level.

[0052] The soil quality grade distribution layer is input into the spatial clustering analysis module, and the density-based clustering algorithm is used to identify adjacent high-risk pixel groups to form the aforementioned risk patches.

[0053] In one feasible implementation, the steps for determining the governance priority order include:

[0054] Based on the time-series changes of the aforementioned high-risk patches, the migration paths of these hotspots were extracted.

[0055] By combining the spatial clustering intensity of the high-risk patches, the migration rate of the hotspot migration paths, and the soil health score, the comprehensive risk weight of each region is calculated, and the priority order of remediation is determined based on the comprehensive risk weight.

[0056] In summary, by simultaneously acquiring satellite remote sensing imagery, UAV multispectral data, ground sample data, and meteorological driving data, this method achieves macro-micro and multi-temporal-multi-scale fusion monitoring. It unifies data from different sources and with different resolutions to the same spatial coordinates and time scale through spatiotemporal assimilation registration technology, thereby eliminating offset errors between multi-source data and significantly improving the comparability and continuity of monitoring results. This method proposes a three-domain spectral component unmixing model for soil, vegetation, and salt crust. Through anisotropic scattering correction and surface roughness correction, it effectively compensates for reflection errors caused by incident angle and topographic microstructure, obtaining soil intrinsic reflectance that more closely approximates the actual physical state. Compared to traditional linear mixing models, this method can more accurately distinguish the spectral characteristics of soil, vegetation, and salt crust, providing high-quality input for water and salt inversion. This method combines thermal infrared and red-edge band information to establish a water and salt co-inversion model. By introducing surface temperature characteristics and the salt sensitivity index (SSI), it achieves multi-parameter joint inversion of soil volumetric water content and surface salinity. Compared to traditional models based solely on single-band or single-index methods, this invention significantly improves the robustness and temporal consistency of soil water and salinity retrieval. By constructing coupled partial differential equations for water and salinity and introducing a graph-spatiotemporal canonical constraint model, this invention integrates physical mechanisms (precipitation, evapotranspiration, and infiltration) with remote sensing data into the optimization process, thereby obtaining a continuous spatiotemporal distribution field of soil moisture and salinity. This fusion method overcomes the limitation of traditional data-driven models lacking physical constraints, enabling dynamic tracking of the temporal evolution of water and salinity. In the soil quality assessment stage, this method comprehensively considers four indicators: salinity characteristics, water characteristics, vegetation response, and soil health. It uses weighted synthesis or principal component analysis to calculate soil quality scores and combines spatial clustering and temporal changes to identify high-risk patches and hotspot migration paths. By calculating comprehensive risk weights, this method can output a priority order for remediation, providing a quantitative decision-making basis for saline-alkali land remediation and ecological restoration. This method combines the interpretability of physical models with the generalization ability of machine learning, making it suitable for long-term dynamic monitoring of water and salt content and assessment of soil degradation in arid, irrigated, and coastal salinized areas. It can be widely applied to scenarios such as agricultural irrigation management, saline-alkali land improvement, ecological restoration, and regional sustainable land use planning. In summary, this method, by integrating multi-source, multi-temporal remote sensing data, constructing a joint model of physical constraints and data-driven approaches, and introducing a multi-index comprehensive evaluation system, not only addresses the shortcomings of traditional techniques in data continuity, inversion accuracy, and comprehensive evaluation, but also achieves a leap from "static monitoring" to "dynamic prediction."

[0057] Other advantages, objectives and features of this application will be apparent in part from the description which follows, and in part from what those skilled in the art will understand through study and practice of this application. Attached Figure Description

[0058] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit this specification. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0059] Figure 1 A schematic diagram of a method for dynamic monitoring and quality assessment of soil water and salt based on multispectral remote sensing, provided for an embodiment of this application;

[0060] Figure 2 This is a schematic flowchart of a method for performing anisotropic scattering correction and roughness correction according to an embodiment of this application;

[0061] Figure 3 A schematic flowchart of a method for obtaining preliminary spectral component ratios provided in an embodiment of this application;

[0062] Figure 4 A schematic flowchart of a method for performing anisotropic scattering correction calculation is provided in an embodiment of this application;

[0063] Figure 5 This is a schematic flowchart of a method for eliminating the influence of surface micromorphological differences on the spectrum, provided in an embodiment of this application.

[0064] Figure 6 This is a schematic flowchart of a method for obtaining multi-temporal soil water and salt inversion results provided in an embodiment of this application;

[0065] Figure 7 A schematic flowchart of a method for obtaining the continuous spatiotemporal distribution field of soil moisture content and salinity, provided for an embodiment of this application;

[0066] Figure 8 This is a schematic diagram of a method for quantitatively scoring and classifying soil quality in different regions, provided as an embodiment of this application. Detailed Implementation

[0067] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus. The technical solutions of the embodiments of this application will now be clearly and completely described in conjunction with the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them.

[0068] Please see Figure 1 This is a flowchart illustrating a method for monitoring and assessing soil water and salt dynamics based on multispectral remote sensing, provided in an embodiment of this application. Specifically, it may include:

[0069] Firstly, this application proposes a method for dynamic monitoring and quality assessment of soil water and salt based on multispectral remote sensing, including:

[0070] S110. Acquire multi-source, multi-temporal, multispectral data, ground sample data, and meteorological driving data; wherein, the multi-source, multi-temporal, multispectral data includes satellite remote sensing images and UAV multispectral camera images, the ground sample data includes volumetric water content, surface and profile electrical conductivity and electrical conductivity ratio, soil texture and bulk density, and the meteorological driving data includes precipitation and evapotranspiration.

[0071] S120. Assimilate and register the above-mentioned multi-source, multi-temporal, multispectral data, ground sample data, and meteorological driving data on a spatiotemporal scale.

[0072] S130. Based on the assimilated and registered multi-temporal multispectral data and ground sample data, a three-domain spectral component unmixing model for soil, vegetation and salt crust is constructed. Anisotropic scattering correction and roughness correction are performed on the observed spectra in the above multi-temporal multispectral data to obtain the soil intrinsic reflectance, vegetation cover component and salt crust component.

[0073] S140. Based on the above-mentioned soil intrinsic reflectance, vegetation cover component, and salt crust component, and combined with the thermal infrared and red edge band information in the above-mentioned satellite remote sensing data, a water and salt inversion model is established to obtain multi-temporal soil water and salt inversion results.

[0074] S150. Using the above multi-temporal water and salt inversion results as observation input, and combining the temporal changes of the above vegetation cover components and the assimilated and registered meteorological driving data, construct a water and salt coupled partial differential equation and a graph spatiotemporal regular constraint model. Through joint optimization of the physical domain and the data domain, obtain the continuous spatiotemporal distribution field of soil moisture content and salinity.

[0075] S160. Based on the above continuous spatiotemporal distribution field, salinity characteristic indicators and moisture characteristic indicators are extracted. Based on the temporal changes of the above vegetation cover components, vegetation response indicators are extracted. Soil health indicators are determined based on the above intrinsic soil reflectance.

[0076] S170. Based on the above salinity characteristic indicators, the above moisture characteristic indicators, the above vegetation response indicators, and the above soil health indicators, a comprehensive soil quality evaluation model is constructed to quantitatively score and classify the soil quality in different regions, and output high-risk patches, hotspot migration paths, and priority order for remediation.

[0077] For example, in step S110, multi-source, multi-temporal, multispectral data, ground sample data, and meteorological driving data are acquired to provide multi-dimensional information support for subsequent analysis. The multispectral data includes satellite remote sensing imagery and UAV multispectral camera imagery, which can simultaneously capture large-scale macroscopic features and local high-resolution details; the ground sample data provides real physical properties, such as volumetric water content, electrical conductivity ratio, and soil texture parameters; the precipitation and evapotranspiration in the meteorological driving data reflect the external water cycle environment and provide boundary conditions for dynamic modeling.

[0078] In step S120, spatiotemporal assimilation and registration are performed on the aforementioned multi-source heterogeneous data to align data from different times, spatial resolutions, and sources to a unified coordinate and time reference, thereby achieving data consistency and comparability. This process integrates geographic reference points and time series interpolation, enabling remote sensing observation results and ground-based measured data to be coordinated in the spatiotemporal domain.

[0079] In step S130, a three-domain spectral component unmixing model of "soil-vegetation-salt crust" is constructed based on the assimilated and registered multi-temporal multispectral data and ground sample data. By correcting the anisotropic scattering and roughness of the observed spectra, the interference of different incident angles and surface micromorphology on spectral reflectance is eliminated, thereby extracting the intrinsic reflectance of soil, vegetation cover component, and salt crust component, providing independent physical characteristics for subsequent water and salt inversion.

[0080] In step S140, a multi-index water and salt synergistic inversion model is established by combining the extracted spectral features with thermal infrared and red-edge band information from satellite remote sensing data. This model analyzes the changes in surface temperature in the thermal infrared band and the sensitive response of the red-edge band to salinity, inverts soil volumetric water content and surface salinity index, and achieves quantitative estimation of soil water and salt distribution at different time points.

[0081] In step S150, the multi-temporal water and salt inversion results, along with the temporal variation of vegetation cover components and meteorological driving data, are jointly input into the water and salt coupled partial differential equation model. By introducing precipitation and evapotranspiration boundary conditions in the physical domain and combining them with graph spatiotemporal regularization constraints, joint optimization is performed in the data domain to obtain the soil moisture and salinity distribution field at continuous spatiotemporal scales, revealing its temporal evolution law.

[0082] In step S160, salinity and moisture characteristics are extracted based on the above continuous spatiotemporal distribution field to quantitatively characterize soil salt accumulation and moisture change characteristics. At the same time, vegetation response indicators are extracted based on the temporal changes of vegetation cover components to reflect the adaptation or stress effect of vegetation growth to water and salt changes. Soil health indicators are then determined based on the soil intrinsic reflectance inversion results to evaluate long-term stable properties such as soil structure, texture, and organic matter content.

[0083] In step S170, a comprehensive soil quality evaluation model is constructed by integrating the aforementioned multiple indicators. This model calculates the comprehensive soil quality score for each region using a weighted comprehensive algorithm or principal component analysis, and classifies the regions according to grading threshold intervals, outputting high-risk patch areas. Combining time-series trends, hotspot migration paths are further extracted to reveal the spatiotemporal migration patterns of high-salinity and high-water risk areas. Finally, based on the comprehensive risk weights and spatial clustering characteristics, a governance priority order is generated, providing a scientific decision-making basis for soil improvement, water resource regulation, and ecological restoration.

[0084] In summary, this method achieves macro-micro and multi-temporal-multi-scale fusion monitoring by simultaneously acquiring satellite remote sensing imagery, UAV multispectral data, ground sample data, and meteorological driving data. This method unifies data from different sources and with different resolutions to the same spatial coordinates and time scale through spatiotemporal assimilation registration technology, thereby eliminating offset errors between multi-source data and significantly improving the comparability and continuity of monitoring results. This method proposes a three-domain spectral component unmixing model of "soil-vegetation-salt crust," which effectively compensates for reflection errors caused by incident angle and topographic microstructure through anisotropic scattering correction and surface roughness correction, obtaining soil intrinsic reflectance that more closely approximates the actual physical state. Compared with traditional linear mixing models, this method can more accurately distinguish the spectral characteristics of soil, vegetation, and salt crust, providing high-quality input for water and salt inversion. This method combines thermal infrared and red-edge band information to establish a water and salt co-inversion model, and by introducing surface temperature characteristics and the salt sensitivity index (SSI), achieves multi-parameter joint inversion of soil volumetric water content and surface salinity. Compared to traditional models based solely on single-band or single-index methods, this invention significantly improves the robustness and temporal consistency of soil water and salinity retrieval. By constructing coupled partial differential equations for water and salinity and introducing a graph-spatiotemporal canonical constraint model, this invention integrates physical mechanisms (precipitation, evapotranspiration, and infiltration) with remote sensing data into the optimization process, thereby obtaining a continuous spatiotemporal distribution field of soil moisture and salinity. This fusion method overcomes the limitation of traditional data-driven models lacking physical constraints, enabling dynamic tracking of the temporal evolution of water and salinity. In the soil quality assessment stage, this method comprehensively considers four indicators: salinity characteristics, water characteristics, vegetation response, and soil health. It uses weighted synthesis or principal component analysis to calculate soil quality scores and combines spatial clustering and temporal changes to identify high-risk patches and hotspot migration paths. By calculating comprehensive risk weights, this method can output a priority order for remediation, providing a quantitative decision-making basis for saline-alkali land remediation and ecological restoration. This method combines the interpretability of physical models with the generalization ability of machine learning, making it suitable for long-term dynamic monitoring of water and salt content and assessment of soil degradation in arid, irrigated, and coastal salinized areas. It can be widely applied to scenarios such as agricultural irrigation management, saline-alkali land improvement, ecological restoration, and regional sustainable land use planning. In summary, this method, by integrating multi-source, multi-temporal remote sensing data, constructing a joint model of physical constraints and data-driven approaches, and introducing a multi-index comprehensive evaluation system, not only addresses the shortcomings of traditional techniques in data continuity, inversion accuracy, and comprehensive evaluation, but also achieves a leap from "static monitoring" to "dynamic prediction."

[0085] In one feasible implementation, such as Figure 2As shown, step S130 above, based on the assimilated and registered multi-temporal multispectral data and ground sample data, constructs a three-domain spectral component unmixing model for soil, vegetation, and salt crust. Anisotropic scattering correction and roughness correction are performed on the observed spectra in the multi-temporal multispectral data to obtain the soil intrinsic reflectance, vegetation cover component, and salt crust component, including:

[0086] S1301. Based on the multi-temporal multispectral data after assimilation and registration, extract the observation spectral reflectance information of each observation pixel, and combine it with the typical soil, vegetation and salt crust sample spectra obtained from the above ground sample data to determine the standard spectral characteristics of the three types of endmembers.

[0087] S1302. Using the above standard spectral characteristics as input, establish a mixed spectral expression composed of soil endmembers, vegetation endmembers and salt crust endmembers to represent the compositional relationship of each pixel spectrum.

[0088] S1303. By decomposing and solving the above mixed spectral expression, the component weights of the soil endmember, the vegetation endmember, and the salt crust endmember in each pixel are obtained, thereby obtaining the preliminary spectral component ratio.

[0089] S1304. To address the reflection differences of the observed spectra under different incident and observation angles, perform anisotropic scattering correction calculations to correct the reflection errors caused by the solar elevation angle and sensor viewing angle.

[0090] S1305. Surface roughness correction is performed on the spectral data after anisotropic scattering correction to eliminate the influence of surface micromorphological differences on the spectrum, and the above-mentioned soil intrinsic reflectance, the above-mentioned vegetation cover component and the above-mentioned salt crust component are obtained.

[0091] This embodiment provides a detailed description of a method for dynamic monitoring and quality assessment of soil water and salinity based on multispectral remote sensing. By combining multi-temporal multispectral data with ground sample data, and utilizing unmixing techniques, physical models, and data-driven analysis methods, this method can accurately assess soil water and salinity conditions and monitor changes in soil quality.

[0092] Using assimilated and registered multi-temporal multispectral data and ground sample data, a three-domain spectral composition unmixing model for soil, vegetation, and salt crust was constructed. This process accurately extracted the contribution ratios of soil, vegetation, and salt crust components through multi-step unmixing calculations.

[0093] The spectral reflectance information of each observed pixel is extracted based on the assimilated and registered multi-temporal multispectral data. The reflectance of each pixel is obtained from remote sensing imagery and can reflect the spectral characteristics of that location. By combining the spectra of typical soil, vegetation, and salt crust samples from ground sample data, the standard spectral characteristics of these three endmembers (i.e., soil, vegetation, and salt crust) can be determined.

[0094]

[0095] in: It is the observed spectral reflectance, which represents the spectrum of each pixel obtained from the remote sensing image; These are the standard spectral basis vectors of each endmember, obtained from ground sample data; It represents the spectral component weights of each endmember, indicating the contribution of different components to the observed spectrum; This is the error term, representing factors such as noise that were not considered.

[0096] The spectral characteristics of soil, vegetation, and salt crust are combined into a total observed spectrum using a weighted summation method. The weight of each endmember is... This reflects the contribution of the endmember to the spectrum of each pixel. Error term. Used to describe unconsidered noise and errors.

[0097] Based on the extracted endmember standard spectra, a mixed spectral expression is established to represent the compositional relationship of the spectra of each pixel:

[0098]

[0099] in: It is the observed spectrum; These are the standard spectra of each endmember; This represents the proportion of spectral components of the endmembers. The formula indicates that the observed spectrum is obtained by weighting the standard spectrum of each endmember (such as soil, vegetation, or salt crust) proportionally. The weight of each endmember (...) The value indicates the proportion of the endmember's contribution to the overall spectrum.

[0100] By decomposing the above mixed spectral expression, the component weights of the three types of endmembers—soil, vegetation, and salt crust—in each pixel are solved, thus obtaining the contribution ratio of each endmember in each pixel:

[0101]

[0102] in: The proportion of spectral components for each endmember; For observing the spectrum; Here are the standard spectra of each endmember. This formula indicates that by minimizing the difference between the observed and predicted spectra, the component weight of each endmember in each pixel can be derived. By using the least squares method or other optimization algorithms, the contribution ratio of each endmember to the observed spectrum is calculated, thus obtaining a preliminary spectral component ratio.

[0103] The observed spectrum of remote sensing data is affected by different incident and observation angles. To eliminate this effect, we perform anisotropic scattering correction calculations in this step to correct reflection errors caused by the solar altitude angle and sensor viewing angle.

[0104]

[0105] in: This is the corrected spectral reflectance; It is the observed spectrum; It is an anisotropic response function that depends on the solar incidence angle. and sensor observation angle .

[0106] By correcting the observed spectrum using anisotropic response functions, reflection errors caused by differences in solar altitude angle and sensor viewing angle are eliminated. This ensures the comparability of spectral data from different observation angles, making the spectral information of remote sensing data more accurate.

[0107] Surface micromorphological differences (such as slope, cracks, etc.) can affect the spectrum, requiring roughness correction of spectral data to eliminate the effects of these differences.

[0108]

[0109] in: This is the final corrected spectrum; It is the spectrum after scattering correction; This is the roughness correction factor; roughness is the surface roughness viscosity. This formula represents the roughness viscosity correction factor. Spectral data are corrected to compensate for spectral differences caused by surface micromorphology (such as slope, cracks, etc.). Roughness parameters are calculated using a digital elevation model and reflect the morphological characteristics of the surface.

[0110] This embodiment combines multi-temporal, multispectral remote sensing data with ground sample data, utilizing advanced spectral unmixing models, scattering correction, and coarse viscosity correction techniques to accurately extract the spectral components of soil, vegetation, and salt crust. Each step processes the observed spectra using mathematical models and optimization algorithms, ensuring the accuracy of spectral component decomposition and providing high-precision data support for subsequent water-salt inversion and soil quality assessment.

[0111] In one feasible implementation, such as Figure 3 As shown, step S1303 above decomposes and solves the above mixed spectral expression to obtain the component weights of the soil endmember, the vegetation endmember, and the salt crust endmember in each pixel, thereby obtaining the preliminary spectral component ratio, including:

[0112] S13031. Transform the above mixed spectrum expression into a matrix equation in matrix form, wherein the above matrix equation uses the above observed spectrum as the observation vector and the matrix composed of the basis vectors of each endmember spectrum as the coefficient matrix.

[0113] S13032. The least squares method is used to solve the above matrix equation and invert it to obtain the initial values ​​of the component weights of each end element.

[0114] S13033. Perform non-negative constraint and normalization processing on the initial values ​​of the above component weights so that the sum of the component weights of the above soil end-member, the above vegetation end-member, and the above salt crust end-member is equal to one.

[0115] S13034. Output the component weights after constraint optimization as the preliminary spectral component ratios.

[0116] For example, this embodiment combines multi-temporal multispectral data with ground sample data and uses unmixing technology to extract three types of end-member spectral components: soil, vegetation, and salt crust, thereby enabling the monitoring and quality assessment of soil water and salt status.

[0117] By decomposing the mixed spectral expression, the component weights of each endmember in each pixel are obtained, thus yielding a preliminary proportion of spectral components. This is one of the core steps in remote sensing data demixing. The mixed spectral expression is then transformed into a matrix equation in matrix form. The goal of this equation is to establish a mathematical relationship between the observed spectrum and the endmember spectrum in matrix form, facilitating subsequent calculations.

[0118]

[0119] in: To observe spectral reflectance, we represent the spectral information of each pixel obtained from remote sensing images; A is a matrix composed of the spectral basis vectors of each endmember (such as soil, vegetation, salt crust), and the columns of the matrix represent the spectra of different endmembers; is the component weight vector of the endmembers, representing the proportion of each endmember in the observed spectrum; This is the error term, which takes into account measurement errors, noise, or other factors that were not considered. It is the observed spectral reflectance, obtained from remote sensing images, representing the reflectance of the pixel in different spectral bands. It is an endmember spectral matrix, which contains the standard spectra of different endmembers (such as soil, vegetation, and salt crust) in various bands. It is the weighting coefficient of each endmember, representing the proportion of that component's contribution to the overall observed spectrum. This is the error term, used to represent the portion that cannot be explained by the known endmembers, such as noise. This equation transforms the mixed spectral model into a linear equation, representing the relationship between the observed spectrum and the endmember spectrum in a matrix manner. The goal is to solve for the weight of each endmember in the observed spectrum.

[0120] The matrix equation above is solved using the least squares method to obtain the initial values ​​of the component weights for each endmember. The least squares method yields the optimal component weights by minimizing the error between the observed and calculated spectra.

[0121]

[0122] in: The endmember weights are obtained using the least squares method. For observing the spectrum; For the endmember spectral matrix, It is a coefficient matrix containing the spectra of each endmember, representing the contribution of the endmember spectra to the observed spectrum. This is the component weight vector of the endmembers, representing the proportion of each endmember. The purpose of the least squares method is to minimize the difference between the observed spectrum and the spectrum calculated from the known endmember spectra and component weights. In this way, we can deduce the weight of each endmember in that pixel, i.e., its contribution to the overall spectrum.

[0123] Since the endmember component weights must be non-negative and their sum must equal 1 (i.e., the spectral components of each pixel are positive and their sum is 1), The initial values ​​are subjected to non-negativity constraints and normalization to ensure physical rationality.

[0124]

[0125] in: This indicates the proportion of spectral components for each endmember. (Constraints) Ensure that the proportion of each endmember is non-negative, which is consistent with physical meaning. Normalization constraint. This ensures that the proportions of all endmembers sum to 1. This step ensures that the contribution of each endmember is reasonable by applying nonnegativity constraints and normalization conditions.

[0126] The output endmember component weights, optimized by nonnegativity constraints and normalization, serve as the initial spectral component proportions.

[0127]

[0128] in: It is the optimized endmember component weight, representing the contribution ratio of each endmember. The optimized endmember weights conform to nonnegativity and normalization constraints. This step ensures that the output endmember weights meet physical conditions and serve as input for subsequent analyses. These results can be used for subsequent water-salinity retrieval, quality assessment, and remediation prioritization.

[0129] In this embodiment, the proportions of spectral components of soil, vegetation and salt crust are accurately extracted from multi-temporal multispectral data. The scientific and physical rationality of spectral unmixing are ensured by techniques such as least squares method, non-negative constraints and normalization processing.

[0130] In one feasible implementation, such as Figure 4 As shown, step S1304 above performs anisotropic scattering correction calculation to correct the reflection error caused by the solar altitude angle and sensor viewing angle for the reflection differences of the observed spectrum under different incident angles and observation angles, including:

[0131] S13041. Based on the imaging geometric information of the above multi-temporal multispectral data, obtain the geometric parameters at the corresponding time, wherein the above geometric parameters include solar incidence angle, azimuth angle and sensor observation angle parameters;

[0132] S13042. Calculate the relative observation geometric relationship of each pixel based on the above geometric parameters, and establish the reflection anisotropic response function; use the above anisotropic response function to perform angle normalization processing on the original reflectance of the above observation spectrum to obtain the corrected reflectance under the conditions of standard incident angle and observation angle.

[0133] S13043. Use the above-mentioned corrected reflectivity output as input data for subsequent surface roughness correction to correct the reflection error caused by the solar altitude angle and sensor viewing angle.

[0134] For example, in remote sensing data processing, the reflectance of the observed spectrum is affected by the incident angle and the observation angle. Therefore, in this step, anisotropic scattering correction is performed to correct reflection errors caused by the solar altitude angle and the sensor's viewing angle. Specifically, the following steps use mathematical methods to perform angle normalization correction on the spectral data to ensure data comparability under different viewing angles.

[0135] First, geometric parameters for the corresponding time moments are obtained using imaging geometric information from multi-temporal and multispectral data. These geometric parameters include the solar incidence angle, azimuth angle, and sensor observation angle, which directly affect the reflectance spectrum. ,in: The angle of incidence of the sun represents the angle at which solar radiation reaches the Earth's surface. The solar azimuth angle represents the angle between the sun's rays and the horizontal plane of the Earth's surface. The sensor observation angle represents the angle at which the remote sensing sensor receives reflected light. These geometric parameters reflect the angular relationship between solar radiation and the reflected light received by the sensor, and are the basis for performing anisotropic scattering correction calculations. These geometric parameters are obtained through...

[0136] Based on the above geometric parameters, we calculate the relative observation geometry of each pixel, and then establish a reflectance anisotropic response function. This response function describes how spectral reflectance changes with angle. Using this response function, we perform angle normalization on the original reflectance spectrum to obtain the corrected reflectance under standard incident and observation angle conditions:

[0137]

[0138] in: This is the corrected spectral reflectance; To observe spectral reflectance; The anisotropic response function describes how spectral reflectance changes with angle.

[0139] The corrected spectral data after anisotropic scattering correction is output and used as input data for subsequent surface roughness correction to further eliminate the influence of surface micromorphological differences on the spectrum.

[0140]

[0141] in, This is the final corrected spectral reflectance; The spectral reflectance after scattering correction; is the coarse viscosity correction factor; roughness is the surface coarse viscosity parameter. The spectral data after scattering correction will undergo further coarse viscosity correction to compensate for the influence of surface micromorphology (such as slope, cracks, etc.) on spectral reflectance. The roughness parameter is calculated based on ground sample data and a digital elevation model (DEM), which can effectively eliminate errors caused by surface micromorphology.

[0142] This embodiment corrects for reflection errors caused by different incident and observation angles, and combines this with surface coarse viscosity information to obtain more standardized and accurate spectral data. This corrected spectral data provides accurate foundational data for dynamic monitoring and quality assessment of soil water and salinity.

[0143] In one feasible implementation, such as Figure 5 As shown, step S1305 above performs surface roughness correction on the spectral data after anisotropic scattering correction to eliminate the influence of surface micromorphological differences on the spectrum, obtaining the above-mentioned soil intrinsic reflectance, the above-mentioned vegetation cover component, and the above-mentioned salt crust component, including:

[0144] S13051. Based on the above ground sample data and digital elevation model data, extract the surface roughness characteristic parameters of each region, wherein the above surface roughness parameters include slope, aspect and surface microstructure scale.

[0145] S13052. Based on the above surface roughness characteristic parameters, construct the roughness influence function and roughness correction coefficient;

[0146] S13053. Combine the spectral data after the above anisotropic scattering correction with the above roughness influence function to perform roughness correction on the above spectral reflectance to compensate for the spectral scattering distortion caused by the surface micromorphology.

[0147] S13054. Output the roughness-corrected spectral data to obtain the above-mentioned soil intrinsic reflectance, the above-mentioned vegetation cover component, and the above-mentioned salt crust component.

[0148] For example, this embodiment combines multi-temporal multispectral data with ground sample data, and uses advanced physical and mathematical methods to process the spectral data, thereby achieving dynamic monitoring of soil, vegetation, and salt crust. Surface roughness correction is performed on the spectral data that has already been corrected by anisotropic scattering. Roughness correction aims to eliminate the influence of surface micromorphological differences (such as slope, cracks, etc.) on the spectral data, thereby improving the accuracy of the spectral data.

[0149] Based on ground sample data and digital elevation model (DEM) data, we extracted surface roughness characteristic parameters for each region. Surface roughness parameters typically include slope, aspect, and surface microstructure scale. By calculating these parameters, we can accurately describe the micromorphological characteristics of the surface:

[0150] roughness Slope,Aspect,Surface micro structure scale

[0151] Here, roughness is a surface coarse viscosity parameter, representing the micromorphological characteristics of the surface, such as slope, aspect, and surface microstructure scale; slope is the slope angle, representing the inclination angle of the surface; aspect is the aspect angle, representing the angle of the slope; and surface microstructure scale is the surface microstructure scale, representing the details of surface texture and morphology. Slope and aspect are extracted from ground data using a digital elevation model (DEM). Surface microstructure scale is obtained from ground sample data, typically estimated by analyzing high-frequency variations related to surface roughness.

[0152] Based on the extracted surface coarse viscosity characteristic parameters, a coarse viscosity influence function is constructed ( ) and roughness correction factor ( These functions and coefficients will be used to describe and compensate for the effects of surface micromorphological differences on the spectrum:

[0153]

[0154] in: (roughness) is the roughness influence function, representing the effect of surface roughness on spectral reflectance; is the roughness correction coefficient, used to adjust the influence of coarse viscosity on the spectrum; roughness is the surface roughness parameter, representing the micromorphological characteristics of the surface. Using the roughness correction coefficient and the roughness influence function, the spectral data after scattering correction can be further refined.

[0155] The roughness-corrected spectral data is output, yielding soil intrinsic reflectance, vegetation cover component, and salt crust component. The corrected spectral data will serve as input for subsequent water-salt inversion and soil quality assessment.

[0156]

[0157] The final spectral data were obtained after anisotropic scattering correction and roughness correction. These data, having eliminated the influence of solar angle, sensor perspective, and surface micromorphology, provide more accurate information on soil, vegetation, and salt crust composition.

[0158] Through the above steps, this embodiment utilizes ground sample data and a digital elevation model for coarse viscosity correction in a multispectral remote sensing-based method for dynamic monitoring and quality assessment of soil water and salt. By precisely adjusting the influence of surface coarse viscosity on the spectrum, it ensures that the final spectral data accurately reflects the true condition of the surface soil, vegetation, and salt crust.

[0159] In one feasible implementation, such as Figure 6 As shown, step S140 above, based on the soil intrinsic reflectance, vegetation cover component, and salt crust component, and combined with the thermal infrared and red-edge band information from the satellite remote sensing data, establishes a water and salt inversion model to obtain multi-temporal soil water and salt inversion results, including:

[0160] S1401. Extract spectral characteristic parameters of near-infrared and short-wave infrared bands based on the above-mentioned intrinsic soil reflectance.

[0161] S1402. Combine the above-mentioned spectral characteristic parameters with the thermal infrared band information in the above-mentioned satellite remote sensing data to calculate the surface temperature characteristic quantity.

[0162] S1403. The salt sensitivity index is extracted using the red edge band information mentioned above, and the salt crust component is introduced as a constraint factor to correct the spectral aliasing error in salt inversion.

[0163] S1404. Input the above spectral characteristic parameters, the above surface temperature characteristic quantities and the above salinity sensitivity characteristic index into the multi-index collaborative inversion model, and calculate the soil volumetric water content and surface salinity index through statistical regression to generate preliminary inversion results.

[0164] S1405. Perform time series consistency correction on the above preliminary inversion results to maintain the physical continuity of water and salt change trends between adjacent time phases, thereby obtaining the above multi-temporal soil water and salt inversion results.

[0165] For example, a water and salt retrieval model is established using soil intrinsic reflectance, vegetation cover component, and salt crust component, combined with thermal infrared and red-edge band information from satellite remote sensing data, to obtain multi-temporal inversion results of soil water and salt. This process includes several sub-steps, detailing how to accurately obtain soil water and salt status by combining spectral features and the model.

[0166] Spectral characteristic parameters in the near-infrared (NIR) and short-wave infrared (SWIR) bands were extracted from the intrinsic reflectance of the soil. These bands play an important role in soil water and salt retrieval because they are sensitive to soil moisture content and salinity.

[0167]

[0168] in: The reflectivity is in the near-infrared band; The reflectivity is in the shortwave infrared band; The angle of incidence of the sun; The sensor's observation angle; These are the wavelengths of the near-infrared and short-wave infrared bands, respectively.

[0169] By combining extracted spectral characteristic parameters (such as reflectance in the near-infrared and short-wave infrared bands) with thermal infrared information, surface temperature characteristics are calculated. Surface temperature has a significant impact on changes in soil moisture and salinity, and can effectively assist in water and salt inversion.

[0170]

[0171] in: It refers to the surface temperature; This refers to the reflectance in the thermal infrared band. Land surface temperature is one of the key parameters for water and salt retrieval. By combining information from the near-infrared, short-wave infrared, and thermal infrared bands, the accuracy of the retrieval can be improved. The thermal infrared band is particularly suitable for monitoring land surface temperature and can be analyzed in conjunction with data from other bands to provide more accurate soil moisture information.

[0172] Salt Sensitivity Index (SSI) is extracted using red-edge band information. The red-edge band is highly sensitive to changes in salinity, especially in soil salt crust regions, thus playing a crucial role in salinity retrieval.

[0173]

[0174] Wherein: SSI is the salt sensitivity characteristic index; The reflectivity is in the red-edge band; This represents the reflectance in the shortwave infrared band. A significant relationship exists between the reflectance in the red-edge band and soil salinity. Calculating the Salt Sensitivity Index (SSI) allows for better identification of salt content, especially in areas where salt crusts have formed.

[0175] The extracted spectral characteristic parameters, surface temperature characteristic quantities, and salinity sensitivity characteristic indices are input into a multi-index collaborative inversion model. Soil volumetric water content (VWC) and surface salinity index (ECe) are calculated through statistical regression to generate preliminary inversion results.

[0176]

[0177] in: This refers to the soil volumetric water content. The surface salinity index; These are the regression coefficients of the soil moisture inversion model; These are the regression coefficients of the salt inversion model; This is the error term.

[0178] By establishing a multi-index synergistic inversion model, combining multiple spectral features and surface temperature information, statistical regression analysis is used to invert soil volumetric water content and surface salinity index.

[0179] After obtaining the preliminary inversion results, time series consistency correction is performed on these results to ensure that the water and salt variation trends between adjacent time phases maintain physical continuity.

[0180]

[0181] in: The corrected change in soil volumetric water content; This represents the change in the original soil volumetric water content; This is a consistency correction factor. Time series consistency correction ensures the consistency of soil water and salt inversion results across different time points, thereby guaranteeing the physical continuity of the inversion results and avoiding data jumps or unreasonable changes.

[0182] Through the above steps, a water and salt retrieval model is established based on multispectral remote sensing data and various spectral characteristic parameters (such as near-infrared, short-wave infrared, and red-edge bands). In this process, sensitive features of soil moisture and salinity are extracted, and the accuracy and physical rationality of the water and salt retrieval results are ensured through multi-index collaborative retrieval and time-series consistency correction.

[0183] In one feasible implementation, such as Figure 7 As shown, step S150 above uses the multi-temporal water-salt inversion results as observation input, combines the temporal variation of the vegetation cover component with the assimilated and registered meteorological driving data, constructs a water-salt coupled partial differential equation and a graph spatiotemporal canonical constraint model, and obtains the continuous spatiotemporal distribution field of soil moisture and salinity through joint optimization of the physical domain and the data domain, including:

[0184] S1501. Using the above multi-temporal water and salt inversion results as observation input, combined with the temporal variation of the above vegetation cover component and the assimilated and registered meteorological driving data, the above precipitation is used as the boundary condition for infiltration and water replenishment, and the above evapotranspiration is used as the surface water loss term. A water-salt coupled partial differential equation and a graph-temporal regular constraint model are constructed.

[0185] S1502. Through joint optimization of the physical domain and the data domain, a continuous spatiotemporal distribution field of soil moisture content and salinity driven by meteorological conditions is obtained.

[0186] For example, by combining multi-temporal water and salt inversion results with temporal variations of vegetation cover components and meteorological driving data, a water-salt coupled partial differential equation (PDE) and a graph spatiotemporal canonical constraint model are constructed. Through joint optimization of the physical model and the data domain, a continuous spatiotemporal distribution field of soil moisture content and salinity is finally obtained.

[0187] A water-salt coupled partial differential equation (PDE) and a graph spatiotemporal canonical constraint model were constructed, and multiple input data (such as multi-temporal water-salt inversion results, temporal variations of vegetation cover components, and meteorological driving data) were incorporated into the model. The model needs to consider meteorological conditions such as precipitation and evapotranspiration.

[0188]

[0189] in: VWC represents the water content per unit volume of soil. Salt content (ECe) indicates the salt concentration in the soil. It is a time variable. is the water diffusion coefficient, which describes the diffusion process of water in the soil. is the salt diffusion coefficient, which describes the diffusion process of salt in the soil. This represents the Laplace operator, which describes the spatial variation of water and salt content. The efficiency of soil moisture evaporation depends on soil moisture content and surface temperature. The effective migration rate of salt depends on soil moisture and salt concentration. Precipitation is one of the meteorological driving factors, and infiltration and recharge boundary conditions are represented. Evapotranspiration is a meteorological driver, and surface water loss is represented by the equations. These two equations depict the spatial and temporal evolution of water and salt in the soil. The first equation describes the changes in soil moisture content, including diffusion, evaporation, and precipitation replenishment; the second equation describes the diffusion and changes in salt in the soil, considering the coupling effect of water and salt. By solving these two equations, the spatiotemporal distribution of soil moisture and salt can be obtained.

[0190] By jointly optimizing the physical and data domains, and combining observational results obtained from remote sensing and ground data with theoretical models, the spatial distribution of soil moisture content and salinity is further optimized. The goal of the joint optimization is to minimize the error between the model predictions and the observed data, ensuring consistency between the physical model and the actual data.

[0191]

[0192] in: Let be the loss function, representing the optimization objective of the model. For the first Predicted soil moisture at various spatiotemporal locations. For the first Soil moisture was observed at various spatiotemporal locations. For the first Predicted salinity at each spatiotemporal location. For the first Salt content observed at various spatiotemporal locations. This is the regularization coefficient, which controls the smoothness of the model. To smooth the predicted results and ensure their physical continuity, the loss function L is minimized to achieve joint optimization between the physical model and actual observation data. This ensures that the resulting spatiotemporal distribution fields of soil moisture and salinity not only conform to physical laws but also are as consistent as possible with the actual observation data.

[0193] In this embodiment, remote sensing data, ground sample data, and meteorological driving data were combined to establish a water-salt coupled partial differential equation and a graph spatiotemporal regular constraint model. This method can accurately simulate the dynamic changes in soil moisture and salinity, obtain their spatiotemporal distribution, and ensure the physical rationality and data consistency of the results through a joint optimization method.

[0194] In one feasible implementation, such as Figure 8 As shown, step S170 above, based on the salinity characteristic index, the moisture characteristic index, the vegetation response index, and the soil health index, constructs a comprehensive soil quality evaluation model to quantitatively score and classify the soil quality of different regions, and outputs high-risk patches, hotspot migration paths, and remediation priorities, including:

[0195] S1701. Perform correlation analysis between the above-mentioned salinity characteristic index and the above-mentioned moisture characteristic index and the above-mentioned vegetation response index obtained from the temporal changes of the above-mentioned vegetation cover component.

[0196] S1702. Based on the above soil health indicators, the above salinity characteristic indicators, the above water characteristic indicators and the above vegetation response indicators are normalized, and a soil quality evaluation characteristic matrix is ​​established.

[0197] S1703. Based on the above soil quality evaluation feature matrix, the above soil quality comprehensive evaluation model is constructed using a weighted comprehensive evaluation algorithm, and the comprehensive soil quality score of each region is calculated.

[0198] S1704. Based on the above comprehensive soil quality score, perform grade classification and spatial cluster analysis, output the above high-risk patches, identify the above hotspot migration paths, and generate the above remediation priority order.

[0199] In one feasible implementation, the steps for determining the aforementioned high-risk plaques include:

[0200] Based on the above comprehensive soil quality score, the score distribution characteristics of each monitoring area were statistically analyzed, and the statistical results of the above scores were calculated. The above statistical results include the mean, standard deviation, and extreme value range.

[0201] Based on the above statistical results, a grading threshold range is set, the above comprehensive soil quality score is divided into multiple levels, and areas in the low score range are marked as potential high-risk areas.

[0202] The soil quality grade distribution layer is generated by spatially encoding and vectorizing the regions at each grade level.

[0203] The soil quality grade distribution layer is input into the spatial clustering analysis module, and the density-based clustering algorithm is used to identify adjacent high-risk pixel groups to form the aforementioned risk patches.

[0204] In one feasible implementation, the steps for determining the governance priority order include:

[0205] Based on the time-series changes of the aforementioned high-risk patches, the migration paths of these hotspots were extracted.

[0206] By combining the spatial clustering intensity of the high-risk patches, the migration rate of the hotspot migration paths, and the soil health score, the comprehensive risk weight of each region is calculated, and the priority order of remediation is determined based on the comprehensive risk weight.

[0207] For example, a comprehensive soil quality evaluation model is constructed based on salinity, moisture, vegetation response, and soil health indicators. This model scores and classifies the soil quality in different regions and outputs high-risk patches, hotspot migration paths, and remediation priorities. The following is a detailed description of each sub-step.

[0208] A correlation analysis was conducted between salinity characteristics, moisture characteristics, and vegetation response indicators (derived from the temporal changes in vegetation cover components). This analysis helps us understand the correlations among these indicators, providing a foundation for subsequent soil quality assessment.

[0209]

[0210] in: The correlation between salinity characteristic indicators and vegetation response indicators; Salinity is a characteristic indicator; This serves as a vegetation response index. By calculating the correlation between salinity characteristics and vegetation response, we can understand how vegetation status affects salt distribution, thereby providing a more accurate model for soil quality assessment.

[0211] Based on soil health indicators, salinity, moisture, and vegetation response indicators were normalized. The normalized indicator data will form a soil quality assessment feature matrix for subsequent comprehensive evaluation.

[0212]

[0213] in: These are the normalized eigenvalues; These are the original feature values ​​(salinity, moisture, vegetation response, etc.). The mean of the eigenvalues; denoted as the standard deviation of the eigenvalues. These represent different characteristic values, such as salinity, moisture, vegetation response, or other relevant indicators. and These are the mean and standard deviation of the eigenvalues, used to standardize the data and ensure that each feature has equal influence in the evaluation model. Normalization helps eliminate dimensional differences between different eigenvalues, making the indicators consistent in the comprehensive evaluation and providing standardized data input for subsequent weighted comprehensive evaluation.

[0214] Based on the normalized soil quality evaluation feature matrix, a weighted comprehensive evaluation algorithm or principal component analysis (PCA) algorithm is used to construct a comprehensive soil quality evaluation model and calculate the comprehensive soil quality score for each region.

[0215]

[0216] in: A comprehensive score for soil quality; These are the normalized feature values ​​(salinity features, water features, vegetation response, etc.). The weights are for the corresponding features; The number of features is denoted by . Using weighted comprehensive evaluation or principal component analysis, we sum all normalized features (such as salinity, moisture, and vegetation response) to calculate the comprehensive soil quality score for each region.

[0217] Based on the comprehensive soil quality score, we performed grading and spatial cluster analysis to identify high-risk patches and hotspot migration paths, and generated a priority order for remediation.

[0218]

[0219] in: The grade of soil quality; The soil quality comprehensive score is used; `cluster()` is a spatial clustering algorithm that performs classification and hotspot identification based on density or other clustering methods. Through spatial clustering analysis of the soil quality comprehensive score, we divide the region into different risk levels. Low-scoring areas are identified as potentially high-risk areas, and high-risk patches and hotspot migration paths are identified based on soil health status.

[0220] The identification of high-risk patches involves the distribution characteristics of the statistical soil quality comprehensive score and the calculation of the mean, standard deviation and extreme value range.

[0221]

[0222] in: This represents the average of the comprehensive soil quality score. The standard deviation of the overall soil quality score; This represents the number of regions. By calculating the mean and standard deviation of the scores, I can understand the overall distribution of soil quality scores.

[0223] High-risk patches are formed by identifying adjacent high-risk pixel clusters through spatial clustering algorithms (such as density-based clustering methods).

[0224]

[0225] Among them: cluster For the first One high-risk plaque; Soil quality was scored for each region. Cluster analysis identified several high-risk patches that require priority remediation.

[0226] Based on the temporal changes of high-risk patches, hotspot migration paths are extracted, and a governance priority order is generated.

[0227]

[0228] Where: path This represents the hotspot migration path; "change in risk" refers to the changes in high-risk areas over time. By analyzing the changing trends of high-risk areas over time, hotspot migration paths can be identified.

[0229] Through the above steps, a complete comprehensive soil quality assessment model was constructed, incorporating considerations of salinity, moisture, vegetation response, and soil health. A weighted comprehensive evaluation algorithm was used to calculate the soil quality score, and spatial clustering and temporal series analysis were employed to identify high-risk patches and hotspot migration pathways. This model not only enables quantitative scoring and classification of soil quality but also identifies areas requiring priority remediation, providing a scientific basis for soil management.

[0230] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for dynamic monitoring and quality assessment of soil water and salt based on multispectral remote sensing, characterized in that, include: Acquire multi-source, multi-temporal, multispectral data, ground sample data, and meteorological driving data; wherein, the multi-source, multi-temporal, multispectral data includes satellite remote sensing imagery and UAV multispectral camera imagery, the ground sample data includes volumetric water content, surface and profile electrical conductivity and electrical conductivity ratio, soil texture and bulk density, and the meteorological driving data includes precipitation and evapotranspiration; The multi-source, multi-temporal, multispectral data, the ground sample data, and the meteorological driving data are assimilated and registered on a spatiotemporal scale. Based on the assimilated and registered multi-temporal multispectral data and ground sample data, a three-domain spectral component unmixing model for soil, vegetation and salt crust was constructed. Anisotropic scattering correction and roughness correction were performed on the observed spectra in the multi-temporal multispectral data to obtain the soil intrinsic reflectance, vegetation cover component and salt crust component. Based on the intrinsic soil reflectance, the vegetation cover component, and the salt crust component, and combined with thermal infrared and red edge band information from satellite remote sensing data, a water and salt inversion model is established to obtain multi-temporal soil water and salt inversion results. Using the multi-temporal soil water and salt inversion results as observation input, and combining the temporal variation of the vegetation cover component with the assimilated and registered meteorological driving data, a water and salt coupled partial differential equation and a graph spatiotemporal regular constraint model are constructed. Through joint optimization of the physical domain and the data domain, the continuous spatiotemporal distribution field of soil moisture and salinity is obtained. Based on the continuous spatiotemporal distribution field, salinity and moisture characteristic indicators are extracted, vegetation response indicators are extracted based on the temporal changes of the vegetation cover component, and soil health indicators are determined based on the intrinsic soil reflectance. Based on the salinity characteristic index, the moisture characteristic index, the vegetation response index, and the soil health index, a comprehensive soil quality evaluation model is constructed to quantitatively score and classify the soil quality in different regions, and output high-risk patches, hotspot migration paths, and priority order for remediation. Based on the intrinsic soil reflectance, the vegetation cover component, and the salt crust component, and combined with the thermal infrared and red-edge band information from the satellite remote sensing data, a water and salt inversion model is established to obtain multi-temporal soil water and salt inversion results, including: Based on the intrinsic reflectance of the soil, spectral characteristic parameters in the near-infrared and short-wave infrared bands are extracted. The surface temperature characteristic quantity is calculated by jointly analyzing the spectral characteristic parameters and the thermal infrared band information in the satellite remote sensing data. The salt sensitivity index is extracted using the red-edge band information, and the salt crust component is introduced as a constraint factor to correct the spectral aliasing error in salt inversion. The spectral characteristic parameters, the surface temperature characteristic quantities, and the salinity sensitivity characteristic index are input into a multi-index collaborative inversion model. Soil volumetric water content and surface salinity index are calculated through statistical regression to generate preliminary inversion results. The preliminary inversion results are subjected to time series consistency correction to maintain the physical continuity of water and salt change trends between adjacent time phases, thereby obtaining the multi-temporal soil water and salt inversion results; The method uses the multi-temporal soil water and salt inversion results as observation input, combines the temporal variation of the vegetation cover component with the assimilated and registered meteorological driving data, constructs a water and salt coupled partial differential equation and a graph spatiotemporal canonical constraint model, and obtains the continuous spatiotemporal distribution field of soil moisture and salinity through joint optimization of the physical domain and the data domain, including: Using the multi-temporal soil water and salt inversion results as observation input, combined with the temporal variation of the vegetation cover component and the assimilated and registered meteorological driving data, the precipitation is used as the boundary condition for infiltration and recharge, and the evapotranspiration is used as the surface water loss term to construct a water-salt coupled partial differential equation and a graph spatiotemporal regular constraint model. By jointly optimizing the physical and data domains, a continuous spatiotemporal distribution field of soil moisture content and salinity driven by meteorological conditions is obtained.

2. The method for dynamic monitoring and quality assessment of soil water and salt based on multispectral remote sensing according to claim 1, characterized in that, Based on the assimilated and registered multi-temporal multispectral data and ground sample data, a three-domain spectral component unmixing model for soil, vegetation, and salt crust is constructed. Anisotropic scattering correction and roughness correction are performed on the observed spectra in the multi-temporal multispectral data to obtain the soil intrinsic reflectance, vegetation cover component, and salt crust component, including: Based on the assimilated and registered multi-temporal multispectral data, the observed spectral reflectance information of each observed pixel is extracted, and combined with the typical soil, vegetation and salt crust sample spectra obtained from the ground sample data, the standard spectral characteristics of the three types of endmembers are determined. Using the standard spectral features as input, a mixed spectral expression composed of soil endmembers, vegetation endmembers, and salt crust endmembers is established to represent the compositional relationship of the spectra of each pixel. By decomposing and solving the mixed spectral expression, the component weights of the soil endmember, the vegetation endmember, and the salt crust endmember in each pixel are obtained, thereby obtaining the preliminary spectral component ratio. To address the reflection differences of the observed spectrum under different incident and observation angles, anisotropic scattering correction calculations are performed to correct the reflection errors caused by the solar elevation angle and sensor viewing angle. The spectral data after anisotropic scattering correction are subjected to surface roughness correction to eliminate the influence of surface micromorphological differences on the spectrum, thereby obtaining the soil intrinsic reflectance, the vegetation cover component, and the salt crust component.

3. The method for dynamic monitoring and quality assessment of soil water and salt based on multispectral remote sensing according to claim 2, characterized in that, The process involves decomposing and solving the mixed spectral expression to obtain the component weights of the soil endmember, the vegetation endmember, and the salt crust endmember in each pixel, thereby obtaining a preliminary spectral component ratio, including: The mixed spectral expression is transformed into a matrix equation in matrix form, wherein the matrix equation uses the observed spectrum as the observation vector and the matrix composed of the basis vectors of each endmember spectrum as the coefficient matrix; The least squares method is used to solve the matrix equation and invert it to obtain the initial values ​​of the component weights of each end element. The initial values ​​of the component weights are subjected to non-negativity constraints and normalization to make the sum of the component weights of the soil endmember, the vegetation endmember, and the salt crust endsource equal to one. The output is the component weight result after constraint optimization, which serves as the preliminary spectral component ratio.

4. The method for dynamic monitoring and quality assessment of soil water and salt based on multispectral remote sensing according to claim 2, characterized in that, The anisotropic scattering correction operation is performed to correct the reflection error caused by the solar elevation angle and sensor viewing angle, based on the reflection differences of the observed spectrum under different incident and observation angles. This includes: Based on the imaging geometric information of the multi-temporal multispectral data, the geometric parameters at the corresponding time are obtained, wherein the geometric parameters include the solar incidence angle, azimuth angle and sensor observation angle parameters; The relative observation geometric relationship of each pixel is calculated based on the geometric parameters, and a reflection anisotropic response function is established. The original reflectance of the observed spectrum is normalized by angle using the anisotropic response function to obtain the corrected reflectance under the conditions of standard incident angle and observation angle. The corrected reflectivity output is used as input data for subsequent surface roughness correction to correct reflection errors caused by solar elevation angle and sensor viewing angle.

5. The method for dynamic monitoring and quality assessment of soil water and salt based on multispectral remote sensing according to claim 2, characterized in that, The process of performing surface roughness correction on the spectral data after anisotropic scattering correction to eliminate the influence of surface micromorphological differences on the spectrum, and obtaining the soil intrinsic reflectance, the vegetation cover component, and the salt crust component, includes: Based on the ground sample data and digital elevation model data, surface roughness characteristic parameters of each region are extracted, wherein the surface roughness parameters include slope, aspect and surface microstructure scale; Based on the surface roughness characteristic parameters, a roughness influence function and a roughness correction coefficient are constructed; The spectral data after anisotropic scattering correction is combined with the roughness influence function to perform roughness correction on the spectral reflectance, compensating for spectral scattering distortion caused by surface micromorphology. Output the roughness-corrected spectral data to obtain the soil intrinsic reflectance, the vegetation cover component, and the salt crust component.

6. The method for dynamic monitoring and quality assessment of soil water and salt based on multispectral remote sensing according to claim 1, characterized in that, Based on the salinity characteristic index, the moisture characteristic index, the vegetation response index, and the soil health index, a comprehensive soil quality evaluation model is constructed to quantitatively score and classify the soil quality of different regions, and output high-risk patches, hotspot migration paths, and remediation priority orders, including: The salinity characteristic index and the moisture characteristic index are correlated with the vegetation response index obtained from the temporal changes of the vegetation cover component; Based on the soil health indicators, the salinity characteristic indicators, the moisture characteristic indicators, and the vegetation response indicators are normalized, and a soil quality evaluation feature matrix is ​​established. Based on the soil quality evaluation feature matrix, a weighted comprehensive evaluation algorithm is used to construct the soil quality comprehensive evaluation model and calculate the comprehensive soil quality score for each region. Based on the comprehensive soil quality score, a classification and spatial clustering analysis are performed to output the high-risk patches, identify the hotspot migration paths, and generate the remediation priority order.

7. The method for dynamic monitoring and quality assessment of soil water and salt based on multispectral remote sensing according to claim 6, characterized in that, The steps for identifying the high-risk plaques include: Based on the comprehensive soil quality score, the score distribution characteristics of each monitoring area are statistically analyzed, and the statistical results of the score are calculated, wherein the statistical results include the mean, standard deviation, and extreme value range. Based on the statistical results, a grading threshold range is set, the comprehensive soil quality score is divided into multiple levels, and areas in the low score range are marked as potential high-risk areas. The soil quality grade distribution layer is generated by spatially encoding and vectorizing the regions at each grade level. The soil quality grade distribution layer is input into the spatial clustering analysis module, and the adjacent high-risk pixel groups are identified using a density-based clustering algorithm to form the risk patches.

8. The method for dynamic monitoring and quality assessment of soil water and salt based on multispectral remote sensing according to claim 7, characterized in that, The steps for determining the governance priority include: Based on the time-series changes of the high-risk patches, the hotspot migration paths are extracted; By combining the spatial clustering intensity of the high-risk patches, the migration rate of the hotspot migration paths, and the soil health score, the comprehensive risk weight of each region is calculated, and the priority order of remediation is determined based on the comprehensive risk weight.

Images