Multi-source remote sensing collaborative method, apparatus, equipment and media for quantitative inversion of soil salinity

Abstract

This application discloses a method, apparatus, equipment, and medium for quantitative inversion of soil salinity using multi-source remote sensing collaboration, relating to the field of remote sensing inversion technology. The method involves preprocessing multi-source remote sensing images; calculating multiple band combinations from satellite multispectral data to construct a first inversion model; using spectral transformation to calculate multiple band combinations from UAV hyperspectral data to construct a second inversion model; selecting target first and second inversion models based on evaluation indicators; inverting the preprocessed multi-source remote sensing images using the target first inversion model to generate a first soil parameter distribution map; selecting areas to be inverted from the first soil parameter distribution map; and inverting the preprocessed multi-source remote sensing images under the areas to be inverted using the target second inversion model to generate a second soil parameter distribution map. This enables timely, efficient, and dynamic monitoring and assessment of soil salinization in large-scale tidal flats, improving the inversion capability for special surface environments of tidal flats.

Description

Technical Field

[0001] This invention relates to the field of remote sensing inversion technology, and in particular to a method, apparatus, equipment and medium for quantitative inversion of soil salinity using multi-source remote sensing synergy. Background Technology

[0002] Soil salinization refers to the accumulation of soluble salts in the soil, such as chlorides, sulfates, and carbonates of sodium, calcium, and magnesium, leading to a soil solution salt concentration exceeding normal levels—in other words, the soil becoming too "salty." This, in turn, inhibits or even destroys the growth of natural vegetation and crops. This is a global resource and ecological problem, especially in arid, semi-arid, and coastal regions. Coastal saline-alkali lands are an important component of China's saline-alkali land. Coastal tidal flat ecosystems, due to their low altitude, high groundwater levels and high mineralization, intense evaporation, and periodic seawater inundation or backflow, are severely affected by soil salinization. Soil salinization leads to land degradation, decreased biodiversity, and loss of agricultural production potential, seriously hindering the sustainable economic and social development of the region.

[0003] To effectively manage saline-alkali land and understand the true salinity of the soil, it is essential to obtain timely, accurate, and dynamic information on soil salinity, moisture, and organic matter. Traditional methods for monitoring soil salinization primarily rely on field surveys and sampling, followed by chemical analysis in the laboratory. While this method offers relatively high measurement accuracy, it suffers from fundamental drawbacks such as low efficiency, high cost, limited representativeness, and poor timeliness, requiring significant manpower and resources for site selection, sampling, and laboratory analysis. Due to the inherent spatial variability of soil, a limited number of sampling points cannot accurately reflect the continuous spatial distribution pattern of soil salinity over large areas. The long cycle from sampling to obtaining results fails to meet the needs of dynamic, real-time monitoring and cannot capture the dynamic processes of salinity changes with environmental factors such as seasons, tides, and rainfall. This monitoring method, which takes weeks or even months, is completely inadequate for meeting the requirements of dynamic and timely monitoring.

[0004] Existing monitoring methods are limited, relying on satellite remote sensing for high macroscopic efficiency but insufficient detail, while relying solely on ground surveys offers high accuracy but extremely low efficiency. Traditional satellite remote sensing is limited by spectral resolution and band range, resulting in low accuracy and poor reliability in retrieving parameters such as soil organic matter in complex tidal flat environments, thus limiting its application value. Many existing inversion models are empirical "black box models" built based on specific regions and data, with their optimal bands and model forms inconsistent across different data sources or regions, and unclear physical mechanisms, leading to poor model transferability and universality, thus limiting their promotional value. Tidal flat surfaces often exhibit complex conditions such as salt crusts, high humidity, and micro-topographical variations. Traditional remote sensing models are insensitive to or insufficiently correct for spectral interference caused by these special surface features, affecting the accuracy of inversion.

[0005] As can be seen from the above, how to solve the problems of insufficient accuracy of satellite multispectral remote sensing inversion, the inability of a single data source to balance monitoring range and accuracy, and poor adaptability in highly heterogeneous tidal flat areas, and how to achieve timely, efficient, and dynamic monitoring and assessment of the salinization status of large-scale tidal flat soils, and improve the sensitivity and inversion capability of the special surface environment of tidal flats, are problems that need to be solved in this field. Summary of the Invention

[0006] In view of this, the purpose of this invention is to provide a method, apparatus, equipment, and medium for quantitative inversion of soil salinity using multi-source remote sensing collaboration. This method addresses the problems of insufficient accuracy in satellite multispectral remote sensing inversion, the inability of a single data source to simultaneously cover both monitoring range and accuracy, and poor adaptability in highly heterogeneous tidal flat areas. It enables timely, efficient, and dynamic monitoring and assessment of soil salinization over large areas of tidal flats, enhancing sensitivity and inversion capabilities to the unique surface environment of tidal flats. The specific solution is as follows: In the first aspect, this application discloses a multi-source remote sensing collaborative method for quantitative inversion of soil salinity, including: Acquire multi-source remote sensing images of the target area and measured soil parameter data from ground sampling points; The multi-source remote sensing images are preprocessed, and spectral data corresponding to ground sampling points are extracted; the spectral data includes satellite multispectral data and UAV hyperspectral data. The first inversion model is constructed by calculating multiple band combinations in the satellite multispectral data, and the second inversion model is constructed by calculating multiple band combinations in the UAV hyperspectral data using the spectral transformation method. Evaluation indices are calculated for the first inversion model and the second inversion model based on the measured soil parameter data to obtain evaluation indices. Target first inversion model and target second inversion model are then selected from the first inversion model and the second inversion model, respectively, based on the evaluation indices. The preprocessed multi-source remote sensing image is inverted using the target first inversion model to generate a first soil parameter distribution map; the coverage area of ​​the first soil parameter distribution map is larger than a preset range. The regions to be inverted are selected from the regions of the first soil parameter distribution map; the regions to be inverted are those where the soil salinization degree is greater than a preset threshold. The preprocessed multi-source remote sensing image under the area to be inverted is inverted using the target second inversion model to generate a second soil parameter distribution map; the accuracy of the second soil parameter distribution map is greater than the preset accuracy.

[0007] Optionally, acquiring multi-source remote sensing images of the target area and measured soil parameter data from ground sampling points includes: Acquire multi-source remote sensing images of the target area; wherein, the multi-source remote sensing images include satellite multispectral remote sensing images and UAV hyperspectral remote sensing images; Obtain measured soil parameter data from ground sampling points; wherein the measured soil parameter data includes total soil salinity, soil organic matter content, and / or soil moisture content.

[0008] Optionally, the preprocessing of the multi-source remote sensing image includes: Geometric correction, radiometric calibration, and atmospheric correction are performed on the satellite multispectral remote sensing images in the multi-source remote sensing images; Geometric correction, radiometric calibration, atmospheric correction, image stitching, and effective band selection are performed on the UAV hyperspectral remote sensing images in the multi-source remote sensing images.

[0009] Optionally, the step of calculating multiple band combinations in the satellite multispectral data to construct a first inversion model includes: The satellite multispectral data and the open format files of the spatial data of the ground sampling points are loaded into the remote sensing image processing platform to obtain the image spectral data of the ground sampling points; The image spectral data and the measured soil parameter data are imported into MATLAB for calculation, and outlier removal is performed to construct the first inversion model.

[0010] Optionally, the step of using the spectral transformation method to calculate multiple band combinations in the UAV hyperspectral data to construct a second inversion model includes: Using the spectral transformation method, the first and second derivative transformations of the UAV hyperspectral reflectance data in the UAV hyperspectral data are performed to obtain the transformed spectral data. A second inversion model is constructed based on the UAV hyperspectral data and multiple band combinations in the transformed spectral data.

[0011] Optionally, the step of calculating evaluation indicators for the first inversion model and the second inversion model based on the measured soil parameter data to obtain evaluation indicators, and then selecting target first inversion models and target second inversion models from the first inversion model and the second inversion model respectively according to the evaluation indicators, includes: Based on the measured soil parameter data, evaluation indicators are calculated for the first inversion model and the second inversion model respectively to obtain evaluation indicators; the evaluation indicators include the coefficient of determination and the root mean square error. The evaluation index corresponding to the maximum value of the coefficient of determination and the minimum value of the root mean square error is used as the target evaluation index; the target evaluation index includes a first target evaluation index and a second target evaluation index. The first inversion model corresponding to the first target evaluation index is used as the target first inversion model; The second inversion model corresponding to the second objective evaluation index is used as the objective second inversion model.

[0012] Optionally, the multi-source remote sensing collaborative method for quantitative inversion of soil salinity further includes: Determine the visualization method based on business needs; The first soil parameter distribution map and the second soil parameter distribution map are visualized according to the visualization method described above.

[0013] Secondly, this application discloses a multi-source remote sensing coordinated soil salinity quantitative inversion device, comprising: The image and data acquisition module is used to acquire multi-source remote sensing images of the target area and measured soil parameter data from ground sampling points. The preprocessing module is used to preprocess the multi-source remote sensing images and extract the spectral data corresponding to the ground sampling points; the spectral data includes satellite multispectral data and UAV hyperspectral data; The inversion model construction module is used to calculate multiple band combinations in the satellite multispectral data to construct a first inversion model, and to calculate multiple band combinations in the UAV hyperspectral data using the spectral transformation method to construct a second inversion model. The evaluation index calculation module is used to calculate the evaluation index of the first inversion model and the second inversion model based on the measured soil parameter data, obtain the evaluation index, and select the target first inversion model and the target second inversion model from the first inversion model and the second inversion model respectively according to the evaluation index. The first distribution map generation module is used to invert the preprocessed multi-source remote sensing image using the target first inversion model to generate a first soil parameter distribution map; the coverage area of ​​the first soil parameter distribution map is larger than a preset range. The region screening module is used to screen out the regions to be inverted from the regions of the first soil parameter distribution map; the regions to be inverted are regions where the soil salinization degree is greater than a preset threshold. The second distribution map generation module is used to invert the preprocessed multi-source remote sensing image under the area to be inverted using the target second inversion model to generate a second soil parameter distribution map; the accuracy of the second soil parameter distribution map is greater than the preset accuracy.

[0014] Thirdly, this application discloses an electronic device, including: Memory, used to store computer programs; A processor is used to execute the computer program to implement the aforementioned multi-source remote sensing collaborative method for quantitative inversion of soil salinity.

[0015] Fourthly, this application discloses a computer storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the steps of the aforementioned disclosed multi-source remote sensing collaborative soil salinity quantitative inversion method.

[0016] As can be seen, this application provides a multi-source remote sensing collaborative method for quantitative inversion of soil salinity, including acquiring multi-source remote sensing images of the target area and measured soil parameter data of ground sampling points; preprocessing the multi-source remote sensing images and extracting spectral data corresponding to the ground sampling points; the spectral data includes satellite multispectral data and UAV hyperspectral data; calculating multiple band combinations in the satellite multispectral data to construct a first inversion model, and using spectral transformation to calculate multiple band combinations in the UAV hyperspectral data to construct a second inversion model; calculating evaluation indicators for the first and second inversion models based on the measured soil parameter data to obtain evaluation indicators. According to the evaluation indicators, a target first inversion model and a target second inversion model are selected from the first inversion model and the second inversion model, respectively. The target first inversion model is used to invert the preprocessed multi-source remote sensing image to generate a first soil parameter distribution map. The coverage area of ​​the first soil parameter distribution map is larger than a preset range. An area to be inverted is selected from the area of ​​the first soil parameter distribution map. The area to be inverted is an area where the soil salinization degree is greater than a preset threshold. The target second inversion model is used to invert the preprocessed multi-source remote sensing image under the area to be inverted to generate a second soil parameter distribution map. The accuracy of the second soil parameter distribution map is greater than a preset accuracy.This application constructs a first inversion model by calculating multiple band combinations from satellite multispectral data, and a second inversion model by calculating multiple band combinations from UAV hyperspectral data using spectral transformation. This collaborative approach integrates satellite multispectral data and UAV hyperspectral data, with satellite multispectral data handling large-scale rapid screening and trend assessment, while UAV hyperspectral data performs precise verification of key areas. This forms a well-defined, mutually verifying collaborative observation and inversion process. Evaluation indicators are calculated for both the first and second inversion models based on measured soil parameter data. Target first and second inversion models are then selected from these models based on these evaluation indicators, achieving efficient and integrated monitoring from rapid regional screening to precise local diagnosis, overcoming the limitations of single data sources. This method significantly improves the accuracy and reliability of inversion for key parameters such as total salinity, organic matter, and moisture, meeting the data requirements for precise governance. It utilizes the first inversion model to invert preprocessed multi-source remote sensing images to generate a first soil parameter distribution map. This allows for rapid and low-cost understanding of the overall spatial pattern and trend of soil salinization, solving the problems of low efficiency and limited range associated with traditional field surveys or single UAV monitoring. From the regions in the first soil parameter distribution map, areas to be inverted are selected, and the second inversion model is used to invert preprocessed multi-source remote sensing images under these areas to generate a second soil parameter distribution map. This solves the core problem of insufficient detail and limited accuracy in inverting satellite data in complex tidal flat areas, enabling timely, efficient, and dynamic monitoring and assessment of soil salinization over large areas of tidal flats, and enhancing sensitivity and inversion capabilities for the unique surface environment of tidal flats. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0018] Figure 1 This is a flowchart of a multi-source remote sensing collaborative method for quantitative inversion of soil salinity disclosed in this application; Figure 2 This application discloses a satellite image and a soil quadrat distribution map. Figure 3 This application discloses a full-view hyperspectral image. Figure 4 This is a partial view of a hyperspectral image disclosed in this application; Figure 5This application discloses a study area and soil quadrat distribution map for surface salinization inversion based on UAV hyperspectral imagery. Figure 6 This is a spectral reflectance curve of saline-alkali soil disclosed in this application; Figure 7 This application discloses a spectral curve of soils with different salinity. Figure 8 This is a diagram showing the total salt content detection results of soil salinization in Hengsha East Beach as disclosed in this application; Figure 9 This is a graph showing the detection results of an organic matter disclosed in this application; Figure 10 This is a graph showing the results of a moisture content test as disclosed in this application; Figure 11 This is a diagram showing the inversion results of five multi-band models of total soil salinity in Hengsha East Beach based on Sentiel-2a imagery disclosed in this application. Figure 12 This is a diagram showing the results of a five-band model inversion of soil organic matter in Hengsha East Beach based on Sentiel-2a imagery, as disclosed in this application. Figure 13 This is a diagram showing the results of a five-band model inversion of soil moisture in Hengsha East Beach based on Sentiel-2a imagery disclosed in this application. Figure 14 This is a spatial distribution map of total salinity in a soil in the northeastern part of Hengsha East Beach, as disclosed in this application. Figure 15 This is a spatial distribution map of organic matter disclosed in this application; Figure 16 This application discloses a spatial distribution map of moisture content. Figure 17 This is a diagram showing the inversion results of a five-band model of total soil salinity based on UAV hyperspectral imagery disclosed in this application. Figure 18 This is a diagram showing the inversion results of a five-band multi-band model of soil organic matter based on UAV hyperspectral imagery disclosed in this application; Figure 19 This is a diagram showing the inversion results of a five-band model of total soil salinity based on UAV hyperspectral imagery disclosed in this application. Figure 20 This is a schematic diagram of the structure of a multi-source remote sensing collaborative soil salinity quantitative inversion device disclosed in this application; Figure 21 This application provides a structural diagram of an electronic device. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Soil salinization refers to the accumulation of soluble salts in the soil, such as chlorides, sulfates, and carbonates of sodium, calcium, and magnesium, leading to a soil solution salt concentration exceeding normal levels—in other words, the soil becoming too "salty." This, in turn, inhibits or even destroys the growth of natural vegetation and crops. This is a global resource and ecological problem, especially in arid, semi-arid, and coastal regions. Coastal saline-alkali lands are an important component of China's saline-alkali land. Coastal tidal flat ecosystems, due to their low altitude, high groundwater levels and high mineralization, intense evaporation, and periodic seawater inundation or backflow, are severely affected by soil salinization. Soil salinization leads to land degradation, decreased biodiversity, and loss of agricultural production potential, seriously hindering the sustainable economic and social development of the region.

[0021] To effectively manage saline-alkali land and understand the true salinity of the soil, it is essential to obtain timely, accurate, and dynamic information on soil salinity, moisture, and organic matter. Traditional methods for monitoring soil salinization primarily rely on field surveys and sampling, followed by chemical analysis in the laboratory. While this method offers relatively high measurement accuracy, it suffers from fundamental drawbacks such as low efficiency, high cost, limited representativeness, and poor timeliness, requiring significant manpower and resources for site selection, sampling, and laboratory analysis. Due to the inherent spatial variability of soil, a limited number of sampling points cannot accurately reflect the continuous spatial distribution pattern of soil salinity over large areas. The long cycle from sampling to obtaining results fails to meet the needs of dynamic, real-time monitoring and cannot capture the dynamic processes of salinity changes with environmental factors such as seasons, tides, and rainfall. This monitoring method, which takes weeks or even months, is completely inadequate for meeting the requirements of dynamic and timely monitoring.

[0022] Existing monitoring methods are limited, relying on satellite remote sensing for high macroscopic efficiency but insufficient detail, while relying solely on ground surveys offers high accuracy but extremely low efficiency. Traditional satellite remote sensing is limited by spectral resolution and band range, resulting in low accuracy and poor reliability in retrieving parameters such as soil organic matter in complex tidal flat environments, thus limiting its application value. Many existing inversion models are empirical "black box models" built based on specific regions and data, with their optimal bands and model forms inconsistent across different data sources or regions, and unclear physical mechanisms, leading to poor model transferability and universality, thus limiting their promotional value. Tidal flat surfaces often exhibit complex conditions such as salt crusts, high humidity, and micro-topographical variations. Traditional remote sensing models are insensitive to or insufficiently correct for spectral interference caused by these special surface features, affecting the accuracy of inversion.

[0023] As can be seen from the above, how to solve the problems of insufficient accuracy of satellite multispectral remote sensing inversion, the inability of a single data source to balance monitoring range and accuracy, and poor adaptability in highly heterogeneous tidal flat areas, and how to achieve timely, efficient, and dynamic monitoring and assessment of the salinization status of large-scale tidal flat soils, and improve the sensitivity and inversion capability of the special surface environment of tidal flats, are problems that need to be solved in this field.

[0024] See Figure 1 As shown in the figure, this invention discloses a multi-source remote sensing collaborative method for quantitative inversion of soil salinity, which specifically includes: Step S11: Acquire multi-source remote sensing images of the target area and measured soil parameter data from ground sampling points.

[0025] In this embodiment, multi-source remote sensing images of the target area are acquired; wherein, the multi-source remote sensing images include satellite multispectral remote sensing images and UAV hyperspectral remote sensing images; measured soil parameter data of ground sampling points are acquired; wherein, the measured soil parameter data includes total soil salinity, soil organic matter content and / or soil moisture content.

[0026] Among them, the measured soil parameter data include at least the total soil salinity, as well as at least one of the soil organic matter content and soil moisture content.

[0027] Step S12: Preprocess the multi-source remote sensing image and extract the spectral data corresponding to the ground sampling points; the spectral data includes satellite multispectral data and UAV hyperspectral data.

[0028] In this embodiment, geometric correction, radiometric calibration, and atmospheric correction are performed on the satellite multispectral remote sensing images in the multi-source remote sensing imagery; geometric correction, radiometric calibration, atmospheric correction, image stitching, and effective band selection are performed on the UAV hyperspectral remote sensing images in the multi-source remote sensing imagery.

[0029] Preprocessing for satellite multispectral remote sensing images includes geometric correction, radiometric calibration, and atmospheric correction; preprocessing for UAV hyperspectral data includes image stitching and effective band selection.

[0030] Step S13: Calculate multiple band combinations in the satellite multispectral data to construct a first inversion model, and use the spectral transformation method to calculate multiple band combinations in the UAV hyperspectral data to construct a second inversion model.

[0031] In this embodiment, satellite multispectral data and open-format spatial data files from ground sampling points are loaded into a remote sensing image processing platform to obtain image spectral data of the ground sampling points. The image spectral data and the measured soil parameter data are imported into MATLAB for calculation and outlier removal to construct a first inversion model. Using the spectral transformation method, the first and second derivatives of the UAV hyperspectral reflectance data in the UAV hyperspectral data are transformed to obtain the transformed spectral data. Based on the UAV hyperspectral data and multiple band combinations in the transformed spectral data, a second inversion model is constructed.

[0032] In this step, the first and second derivative transformations are performed on the hyperspectral reflectance data of the UAV; multiple spectral indices are constructed from the UAV hyperspectral data and the transformed spectral data by using band combinations to form a second inversion model.

[0033] Step S14: Calculate evaluation indicators for the first inversion model and the second inversion model based on the measured soil parameter data to obtain evaluation indicators, and select the target first inversion model and the target second inversion model from the first inversion model and the second inversion model respectively according to the evaluation indicators.

[0034] In this embodiment, evaluation indices are calculated for the first inversion model and the second inversion model based on the measured soil parameter data to obtain evaluation indices. The evaluation indices include the coefficient of determination and the root mean square error. The evaluation indices corresponding to the maximum value of the coefficient of determination and the minimum value of the root mean square error are taken as target evaluation indices. The target evaluation indices include a first target evaluation index and a second target evaluation index. The first inversion model corresponding to the first target evaluation index is taken as the target first inversion model. The second inversion model corresponding to the second target evaluation index is taken as the target second inversion model.

[0035] This application selects the optimal models for retrieving target soil parameters from satellite and UAV data sources using evaluation indicators, respectively; that is, it selects a first target inversion model and a second target inversion model for each soil parameter. The evaluation indicators include... (Determination coefficient) and RMSE (Root Mean Squared Error); maximum value and RMSE The evaluation index corresponding to the minimum value is used as the target evaluation index.

[0036] Step S15: Use the target first inversion model to invert the preprocessed multi-source remote sensing image to generate a first soil parameter distribution map; the coverage area of ​​the first soil parameter distribution map is larger than a preset range.

[0037] Step S16: Select the region to be inverted from the region of the first soil parameter distribution map; the region to be inverted is the region where the soil salinization degree is greater than a preset threshold.

[0038] Step S17: Use the target second inversion model to invert the preprocessed multi-source remote sensing image in the area to be inverted to generate a second soil parameter distribution map; the accuracy of the second soil parameter distribution map is greater than the preset accuracy.

[0039] In this embodiment, the multi-source remote sensing collaborative method for quantitative inversion of soil salinity further includes: determining the visualization method according to business needs; and visualizing the first soil parameter distribution map and the second soil parameter distribution map according to the visualization method.

[0040] Taking Hengsha East Beach as an example, the surface salinization level of Hengsha East Beach was inverted using Sentinel-2A satellite imagery from September 21, 2021. The specific steps are as follows: 1. Data source: From west to east, along different temporal gradients of the enclosure, a total of 20 quadrats were set up. Satellite imagery and soil quadrat distribution maps are shown below. Figure 2 As shown, the reflectance differences within the soil plots in the northern part of Hengsha East Beach are quite significant, with the bright white area in the lower left corner exhibiting a higher level of soil salinization. Based on the satellite image pixel size, each quadrat was 10m × 10m in size. Soil samples were collected from the topsoil (above 5cm) according to HJ / T166-2004 "Technical Specification for Soil Environmental Monitoring". Approximately 1kg of mixed sample was collected using the diagonal method (a total of 5 sampling points were set at the four corners and the center of each quadrat, and the samples were mixed thoroughly to form a mixed sample). The samples were sealed and placed in a 0℃ sample storage box before being brought back to the laboratory. The GPS (Global Positioning System) coordinates, moisture content, electrical conductivity, and temperature of the quadrat were recorded on-site. After being brought back to the laboratory, the total salt content, organic matter, and moisture content were measured. 2. A full-scale hyperspectral image was acquired using a UAV (Unmanned Aerial Vehicle) platform, as shown below. Figure 3 As shown, a partial view of the hyperspectral image is as follows. Figure 4 As shown in the imagery from UAVs, in areas with high soil salinity levels observed during field surveys, the soil surface has a salt crust or skin, is smooth, firm, and whitish, with a higher spectral reflectance than other soils, appearing bright white. The remote sensing analysis process includes: geographic correction, radiometric calibration, atmospheric correction, construction of a quantitative inversion model, and accuracy evaluation. (1) Geometric correction: Select 5-10 GCPs (Ground Control Points), use sub-meter differential GPS to measure their geographic coordinates on site, and use the Polynomial second-order correction model to complete the geometric fine correction. It is expected that the error of the final corrected image data can be controlled within 0.3 pixels, and the error of the CCD (Charge-Coupled Device) data can be controlled within 1 pixel. (2) Radiometric calibration: Radiometric calibration is the establishment of a quantitative relationship between the radiation intensity of a remote sensing sensor and the radiance value in its corresponding field of view. Since the absolute radiometric calibration model and calibration coefficients differ for each sensor, this application uses Sentinel-2A and the dedicated tool SNAP V9.0.0 provided by ESA (European Space Agency) for radiometric calibration; (3) Atmospheric correction: Atmospheric correction is particularly important in quantitative remote sensing inversion and is one of the key steps in quantitative remote sensing. This application uses the 6S (Second Simulation of the Satellite Signal in the Solar Spectrum) model to carry out atmospheric correction; (4) Constructing a quantitative inversion model: Five different multi-band models were selected and constructed in ENVI (The Environment for Visualizing Images) and MATLAB environments. The steps are as follows: satellite images and shp (ESRI Shapefile) files were loaded into ENVI and the spectral data of the measured points were exported; the spectral data and the measured surface data were imported into MATLAB; the optimal band combination was calculated; and the fitting was completed after removing outliers.

[0041] UAV Image Processing Method: This application combines empirical and semi-empirical methods, using band combination parameters as intermediate coefficients. Each intermediate coefficient is then fitted to the results, and the method with the best-fitting intermediate coefficients is finally selected. Some of the intermediate parameters used are simple combinations of band reflectance values, while others contain certain optical properties. After calculating the model equations, the raster calculator module of ArcGIS (Ver. 10.2) is used to invert various soil parameters.

[0042] To identify the main characteristics of salinization, hyperspectral image inversion was conducted to determine total salt content, organic matter, and moisture. The hyperspectral image analysis workflow included: screening effective bands, hyperspectral image stitching, constructing a quantitative inversion model, and accuracy evaluation. (1) Filtering effective bands: The images captured by the S185 hyperspectral camera contain 147 bands. The spectral resolution of the hyperspectral data is 10nm. According to the camera user manual, the effective bands 450-950nm were exported using the CurtUtilsTouch software. The exported files were renamed and placed in a new folder.

[0043] We plan to select ground control points and use sub-meter differential GPS to measure their geographic coordinates on-site. We will then use a polynomial second-order correction model to complete the geometric fine correction. It is expected that the error of the final corrected image data will be controlled within 0.3 pixels, and the error of the CCD data will be controlled within 1 pixel.

[0044] (2) Hyperspectral image stitching: Import the exported data into Photoscan, import the GPS information of each image into the software and match them one by one to align the latitude and longitude between the images. Generate a dense point cloud dataset, stitch the images together, and export the hyperspectral orthophoto data; (3) Constructing a quantitative inversion model. The study area and soil sample plot distribution for surface salinization inversion based on UAV hyperspectral imagery are shown below. Figure 5 As shown.

[0045] To support the research, spectroscopic measurements were performed on September 24, 2021, using an ASD Fieldspec3 spectrometer. The spectrometer had sampling intervals of 1.4 nm (in the 350 nm–1000 nm band) and 2 nm (in the 1000 nm–2500 nm band), with spectral resolutions of 3.5 nm (350 nm–1000 nm) and 10 nm (1000 nm–2500 nm), respectively. The resampling interval was 1 nm, and a total of 2151 output bands were obtained.

[0046] For field measurements, choose clear, low-cloudy weather with low wind speeds, between 10:00 and 14:00 to ensure sufficient solar altitude angle. The probe's field of view is 25°. During measurement, the probe is perpendicular to the ground, approximately 50cm above the surface, using an ASDFieldspec3 spectrometer with a matching diffuse reflectance standard white board as a reference. Before data acquisition, the white board is calibrated, and then the ground spectral data is measured. The measurements are averaged 10 times to obtain the average spectral reflectance of the saline-alkali land.

[0047] To eliminate the noise influence of soil sample hyperspectral data, a moving average method was used for noise reduction in the visible to near-infrared (350nm-1300nm) band. This involved selecting the average value of the spectral curves over a certain range before and after a specific point in the sample, and using this average as the value for that point. Due to differences in the instrument's internal detection elements, a 9-point weighted sampling smoothing method was used for noise reduction in the short-wave infrared (1301nm-2500nm) band. This process effectively removed noise while preserving the overall spectral trend, thus better reflecting the absorption characteristics of the spectrum.

[0048] Spectral measurement results: The spectral reflectance curve of saline-alkali soil is as follows Figure 6 As shown, all spectral curves exhibit similar overall trends, with some fluctuation in their overall shape. Spectral reflectance is low in the visible light region but increases in the near-infrared and short-wave infrared bands. The spectral shapes are very similar, and there is a good positive correlation between different bands (e.g., when soil reflectance is low in the visible light region, it is also low in the short-wave infrared region). Within the 350nm-2500nm spectral range, there are two distinct water vapor absorption peaks (around 1400nm and 1900nm). The spectral reflectance curves across the entire wavelength range can be roughly divided into three intervals: 350nm-1400nm, 1400nm-1900nm, and 1900nm-2500nm. The spectral reflectance values ​​are relatively low between 350nm and 600nm, and gradually increase with increasing wavelength, exhibiting a steep slope. The slope reaches its maximum near 600nm. Between 600nm and 1400nm, reflectance continues to increase, but the rate of increase decreases. Two water absorption bands are located near 1400nm and 1900nm; between these bands, the spectral curve changes gradually but still shows an upward trend. After reaching the reflectance trough near 1900nm, the spectral reflectance curve continues to rise. Around 2200nm, the spectral reflectance begins to decrease, with minor increases during this period, but the overall trend of the spectral curve is decreasing.

[0049] Soil spectral curves with different salinity levels, such as Figure 7As shown, the spectral curves of different salinization degrees basically conform to the above pattern: with the increase of salt content, the spectral reflectance gradually decreases. The spectral curves of soils with different salinization degrees are similar, with consistent overall shape and trend. The spectral reflectance is not high in the visible light region, showing an increasing trend, while it continues to increase in the near-infrared band. At 1400nm and 1900nm, there are obvious water vapor absorption peaks due to the influence of water vapor, representing reflection valleys in the spectral reflectance. The spectral data of soils with different salinization degrees are not significantly different. Overall, with the increase of soil salinization degree, the spectral reflectance shows a decreasing trend; the difference in spectral reflectance between different salinization degrees is small in the visible light band; the difference in spectral reflectance between the near-infrared and short-wave infrared bands increases. There are also significant differences in the water vapor absorption peaks at 1400nm and 1900nm.

[0050] Based on satellite remote sensing inversion results: the detection results of soil salinization-related indicators at 20 locations in Hengsha East Beach are shown in Table 1: Table 1 Soil index test results

[0051] The total salinity test results of soil salinization in Hengsha East Beach are as follows: Figure 8 As shown, the detection results of organic matter are as follows: Figure 9 As shown, the moisture content test results are as follows: Figure 10 As shown.

[0052] Using the band combination method, five different band combination methods were employed, and the following five multi-band models were constructed using an exponential model, as shown in Tables 2, 3, and 4: Table 2 Total Salt Content Inversion Model

[0053] Table 3 Organic matter inversion model

[0054] Table 4 Moisture content inversion model

[0055] The results of the five-band model inversion of total soil salinity in Hengsha East Beach based on Sentiel-2a imagery are as follows: Figure 11 As shown, the inversion results of five multi-band models of soil organic matter in Hengsha East Beach based on Sentiel-2a imagery are as follows: Figure 12 As shown, the inversion results of the five multi-band model of soil moisture in Hengsha East Beach based on Sentiel-2a imagery are as follows: Figure 13 As shown.

[0056] The statistical parameters of the total salinity inversion model are shown in Table 5, the statistical parameters of the organic matter inversion model are shown in Table 6, and the statistical parameters of the moisture content inversion model are shown in Table 7. Table 5 Statistical parameters of the total salt content inversion model

[0057] Table 6 Statistical parameters of the organic matter inversion model

[0058] Table 7 Statistical parameters of the moisture content inversion model

[0059] The results show that Model 3 (5-band model) has the highest accuracy in retrieving both total salinity and water content. Among them, the total salinity retrieving model... It is 0.78. RMSE The value is 0.48; Moisture content inversion model It is 0.67. RMSE The value is 0.52. The accuracy of the organic matter inversion model is relatively low. The highest is 0.43. RMSE It is 1.49.

[0060] The soil parameter test results at 20 locations in the Hengsha East Beach saline-alkali land are shown in Table 8: Table 8 Soil parameter test results

[0061] The spatial distribution of total salinity in the soil of the northeastern part of Hengsha East Beach is as follows: Figure 14 As shown, the spatial distribution of organic matter is as follows: Figure 15 As shown, the spatial distribution of moisture content is as follows: Figure 16 As shown.

[0062] The basic principle of the multi-band combination index method is to construct a spectral index by combining arbitrary bands and performing linear correlation analysis with measured data. The optimal spectral index is obtained by comparing the correlation coefficients. This application further performs spectral transformation on the original spectral data to enhance its sensitivity to measured data and uses the band combination index method to screen the band combinations after each spectral transformation. The spectral indices selected by this method have richer spectral information and higher sensitivity to measured data on tidal flat salinity, thus estimating the data more accurately. The optimized spectral indices have higher stability, improving the reliability of the prediction model.

[0063] In this application, five spectral indices are constructed based on the original spectral data, the first derivative spectral data, and the second derivative spectral data, respectively. These constructed original spectral indices, the first derivative spectral indices, and the second derivative spectral indices are then used as input variables to predict data. These novel multi-band spectral indices capture vegetation information that traditional vegetation indices cannot cover, thus improving the sensitivity to measured data.

[0064] The five constructed multi-band models are shown in Tables 9, 10, and 11: Table 9 Total Salt Content Inversion Model

[0065] Table 10 Organic Matter Inversion Model

[0066] Table 11 Moisture content inversion model

[0067] The inversion results of five multi-band models of soil total salinity based on UAV hyperspectral imagery are as follows: Figure 17 As shown, the inversion results of five multi-band models of soil organic matter based on UAV hyperspectral imagery are as follows: Figure 18 As shown, the inversion results of five multi-band models of soil total salinity based on UAV hyperspectral imagery are as follows: Figure 19 As shown. Because there is a body of water in the middle of the study area, data from two points in that area were removed during the inversion process.

[0068] The inversion accuracies based on the five multi-band models are shown in Tables 12, 13, and 14. Among them, Model 2 has the highest accuracy for total salinity inversion. It is 0.97. RMSE The accuracy was 0.25; the highest accuracy of the organic matter inversion model was Model 2. It is 0.85. RMSE The accuracy is 0.06; the highest accuracy for moisture content inversion is achieved by model 5, which is the moisture content inversion model. It is 0.95. RMSE It is 0.28.

[0069] Table 12 Statistical parameters of the total salt content inversion model

[0070] Table 13 Statistical parameters of the organic matter inversion model

[0071] Table 14 Statistical parameters of the moisture content inversion model

[0072] Existing technologies often rely on a single data source, making it difficult to balance the efficiency of large-scale monitoring with the accuracy of inversion in key areas. This invention coordinates satellite multispectral imagery and UAV hyperspectral imagery. Satellite data handles rapid screening and trend assessment over a large area, while UAV hyperspectral data performs precise verification in key areas, forming a collaborative observation and inversion process with clear division of labor and mutual verification. In a specific implementation example, for total salinity inversion, the optimal satellite model... It is 0.78. RMSE The value is 0.48; while the optimal model for drones... Up to 0.97, RMSE The value is 0.25. This indicates that UAV data can achieve near-ground-based measurement accuracy in local areas, while satellite data can provide a reliable overview of regional distribution.

[0073] The spectra of tidal flat soils are complexly influenced by multiple factors, including salinity, moisture, organic matter, and salt crust surface characteristics, making the use of fixed, single indices or models limited in effectiveness. This invention designs a systematic strategy for screening the optimal inversion model from a vast array of potential band combinations specifically for tidal flat environments. Multiple multi-band combination models are selected, and first-order and second-order spectral transformations are performed on the original spectral data to enhance sensitivity to salinization information. The optimal model is then selected through goodness-of-fit comparison. In a specific implementation case, five models were constructed for total salinity, organic matter, and moisture, respectively, and their accuracy was compared. The results show that the optimal model differs for different parameters, and the band combinations screened through spectral transformations exhibit significantly higher model accuracy than traditional empirical methods. This demonstrates that this screening strategy can effectively uncover the spectral characteristics of tidal flat soils.

[0074] High-precision quantitative inversion relies on rigorously registered ground truth values. This invention proposes a ground survey method closely coordinated with remote sensing data acquisition. Differentiated quadrats are designed based on satellite pixel size and the high resolution characteristics of UAVs; the sampling time is close to the remote sensing transit time; and red balloon markers are used for UAV sampling points to ensure accurate positioning on the image. This set of specifications ensures maximum spatial, temporal, and physical matching between "ground truth values" and "remote sensing image pixels," reducing inversion errors at the source.

[0075] The core technology of this invention is based on direct quantitative inversion of satellite and UAV remote sensing imagery. Only a small number of validation plots need to be deployed in representative key areas of the tidal flats for model calibration, enabling contactless and rapid monitoring of large-scale tidal flat areas. This makes long-term, dynamic monitoring of tidal flat salinization distribution economical, safe, and feasible, truly leveraging the macroscopic advantages of remote sensing technology and solving the problem of existing methods being ineffective in tidal flat scenarios.

[0076] Furthermore, this invention employs more advanced remote sensing data sources and a more direct inversion path. By selecting the optimal bands or band combinations closely related to the salinity of tidal flat soils, a direct quantitative model of spectral reflectance and salinity parameters is established. Implementation examples show that in Hengsha East Beach, the optimal total salinity inversion model constructed based on UAV hyperspectral data achieves a coefficient of determination of 0.97, demonstrating the high effectiveness of this direct modeling approach in capturing subtle variations in tidal flat salinity. Compared to existing technologies that rely on indirect conductivity correlation models, which may not be robust in tidal flats, this invention provides a high-precision solution that more closely approximates the surface spectral response mechanism and is more suitable for the complex underlying surface of tidal flats.

[0077] In addition, a step-by-step, data-guided asynchronous acquisition strategy can be adopted. For example, historical or recent satellite imagery can be acquired first for large-scale preliminary inversion. Based on this result, suspected high-salinity key areas can be identified, and then UAVs can be deployed in these key areas to collect hyperspectral data and perform ground verification sampling. Alternatively, a wider range of different types of equipment can be used for data acquisition: satellite data sources can cover any multispectral / hyperspectral satellite providing visible light to shortwave infrared bands. UAV platforms can include various types of UAVs such as fixed-wing and multi-rotor UAVs, and payloads can include hyperspectral, multispectral, and even thermal infrared sensors for inverting surface temperature information related to salt stress.

[0078] In this embodiment, multi-source remote sensing images of the target area and measured soil parameter data from ground sampling points are acquired; the multi-source remote sensing images are preprocessed, and spectral data corresponding to the ground sampling points are extracted; the spectral data includes satellite multispectral data and UAV hyperspectral data; multiple band combinations in the satellite multispectral data are calculated to construct a first inversion model, and multiple band combinations in the UAV hyperspectral data are calculated using spectral transformation to construct a second inversion model; evaluation indicators are calculated for the first and second inversion models based on the measured soil parameter data to obtain evaluation indicators, and the soil parameters are then used to construct the first inversion model. Target first inversion model and target second inversion model are selected from the first inversion model and the second inversion model, respectively. The target first inversion model is used to invert the preprocessed multi-source remote sensing image to generate a first soil parameter distribution map. The coverage area of ​​the first soil parameter distribution map is larger than a preset range. The area to be inverted is selected from the area of ​​the first soil parameter distribution map. The area to be inverted is the area where the soil salinization degree is greater than a preset threshold. The target second inversion model is used to invert the preprocessed multi-source remote sensing image under the area to be inverted to generate a second soil parameter distribution map. The accuracy of the second soil parameter distribution map is greater than a preset accuracy.This application constructs a first inversion model by calculating multiple band combinations from satellite multispectral data, and a second inversion model by calculating multiple band combinations from UAV hyperspectral data using spectral transformation. This collaborative approach integrates satellite multispectral data and UAV hyperspectral data, with satellite multispectral data handling large-scale rapid screening and trend assessment, while UAV hyperspectral data performs precise verification of key areas. This forms a well-defined, mutually verifying collaborative observation and inversion process. Evaluation indicators are calculated for both the first and second inversion models based on measured soil parameter data. Target first and second inversion models are then selected from these models based on these evaluation indicators, achieving efficient and integrated monitoring from rapid regional screening to precise local diagnosis, overcoming the limitations of single data sources. This method significantly improves the accuracy and reliability of inversion for key parameters such as total salinity, organic matter, and moisture, meeting the data requirements for precise governance. It utilizes the first inversion model to invert preprocessed multi-source remote sensing images to generate a first soil parameter distribution map. This allows for rapid and low-cost understanding of the overall spatial pattern and trend of soil salinization, solving the problems of low efficiency and limited range associated with traditional field surveys or single UAV monitoring. From the regions in the first soil parameter distribution map, areas to be inverted are selected, and the second inversion model is used to invert preprocessed multi-source remote sensing images under these areas to generate a second soil parameter distribution map. This solves the core problem of insufficient detail and limited accuracy in inverting satellite data in complex tidal flat areas, enabling timely, efficient, and dynamic monitoring and assessment of soil salinization over large areas of tidal flats, and enhancing sensitivity and inversion capabilities for the unique surface environment of tidal flats.

[0079] See Figure 20 As shown in the figure, this invention discloses a multi-source remote sensing synergistic soil salinity quantitative inversion device, which specifically includes: Image and data acquisition module 11 is used to acquire multi-source remote sensing images of the target area and measured soil parameter data of ground sampling points; The preprocessing module 12 is used to preprocess the multi-source remote sensing image and extract the spectral data corresponding to the ground sampling points; the spectral data includes satellite multispectral data and UAV hyperspectral data; The inversion model construction module 13 is used to calculate multiple band combinations in the satellite multispectral data to construct a first inversion model, and to calculate multiple band combinations in the UAV hyperspectral data using the spectral transformation method to construct a second inversion model. The evaluation index calculation module 14 is used to calculate the evaluation index of the first inversion model and the second inversion model based on the measured soil parameter data, obtain the evaluation index, and select the target first inversion model and the target second inversion model from the first inversion model and the second inversion model respectively according to the evaluation index. The first distribution map generation module 15 is used to invert the preprocessed multi-source remote sensing image using the target first inversion model to generate a first soil parameter distribution map; the coverage area of ​​the first soil parameter distribution map is greater than a preset range. The region screening module 16 is used to screen out the regions to be inverted from the regions of the first soil parameter distribution map; the regions to be inverted are regions where the soil salinization degree is greater than a preset threshold. The second distribution map generation module 17 is used to invert the preprocessed multi-source remote sensing image under the area to be inverted using the target second inversion model to generate a second soil parameter distribution map; the accuracy of the second soil parameter distribution map is greater than the preset accuracy.

[0080] In some specific embodiments, the image and data acquisition module 11 may specifically include: The image acquisition module is used to acquire multi-source remote sensing images of the target area; wherein, the multi-source remote sensing images include satellite multispectral remote sensing images and UAV hyperspectral remote sensing images; The measured soil parameter data acquisition module is used to acquire measured soil parameter data from ground sampling points; wherein, the measured soil parameter data includes total soil salinity, soil organic matter content and / or soil moisture content.

[0081] In some specific embodiments, the preprocessing module 12 may specifically include: The first correction module is used to perform geometric correction, radiometric calibration, and atmospheric correction on the satellite multispectral remote sensing images in the multi-source remote sensing images. The second correction module is used to perform geometric correction, radiometric calibration, atmospheric correction, image stitching, and effective band selection on the UAV hyperspectral remote sensing images in the multi-source remote sensing images.

[0082] In some specific embodiments, the inversion model construction module 13 may specifically include: The loading module is used to load satellite multispectral data and open format files of spatial data from ground sampling points into the remote sensing image processing platform to obtain image spectral data of ground sampling points. The outlier removal module is used to import the image spectral data and the measured soil parameter data into MATLAB for calculation and outlier removal to construct the first inversion model.

[0083] In some specific embodiments, the inversion model construction module 13 may specifically include: The transformation module is used to perform first-order and second-order derivative transformations on the UAV hyperspectral reflectance data in the UAV hyperspectral data using the spectral transformation method to obtain the transformed spectral data. The second inversion model construction module is used to construct a second inversion model based on the UAV hyperspectral data and multiple band combinations in the transformed spectral data.

[0084] In some specific embodiments, the evaluation index calculation module 14 may specifically include: The evaluation index calculation module is used to calculate the evaluation index for the first inversion model and the second inversion model based on the measured soil parameter data, respectively, to obtain the evaluation index; the evaluation index includes the coefficient of determination and the root mean square error; The target evaluation index determination module is used to select the evaluation index corresponding to the maximum value of the coefficient of determination and the minimum value of the root mean square error as the target evaluation index; the target evaluation index includes a first target evaluation index and a second target evaluation index. The first inversion model determination module is used to determine the first inversion model corresponding to the first target evaluation index as the first inversion model of the target; The second inversion model determination module is used to determine the second inversion model corresponding to the second target evaluation index as the second inversion model of the target.

[0085] In some specific embodiments, the multi-source remote sensing collaborative soil salinity quantitative inversion device may further include: The visualization method determination module is used to determine the visualization method based on business requirements. The visualization module is used to visualize the first soil parameter distribution map and the second soil parameter distribution map according to the visualization method.

[0086] Figure 21 This is a schematic diagram of an electronic device provided in an embodiment of this application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the multi-source remote sensing coordinated soil salinity quantitative inversion method disclosed in any of the foregoing embodiments.

[0087] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0088] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored on it include operating system 221, computer program 222 and data 223, etc., and the storage method can be temporary storage or permanent storage.

[0089] The operating system 221 manages and controls the various hardware devices on the electronic device 20 and the computer program 222 to enable the processor 21 to perform calculations and processing on the data 223 in the memory 22. It can be Windows, Unix, Linux, etc. The computer program 222, in addition to including a computer program capable of performing the multi-source remote sensing coordinated soil salinity quantitative inversion method executed by the electronic device 20 as disclosed in any of the foregoing embodiments, may further include computer programs capable of performing other specific tasks. The data 223 may include data received by the multi-source remote sensing coordinated soil salinity quantitative inversion device from external devices, and may also include data collected by its own input / output interface 25.

[0090] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0091] Furthermore, this application also discloses a computer-readable storage medium storing a computer program. When the computer program is loaded and executed by a processor, it implements the steps of the multi-source remote sensing collaborative soil salinity quantitative inversion method disclosed in any of the foregoing embodiments.

[0092] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0093] The present invention provides a detailed description of a multi-source remote sensing collaborative method, apparatus, equipment, and storage medium for quantitative inversion of soil salinity. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for quantitative inversion of soil salinity using multi-source remote sensing, characterized in that, include: Acquire multi-source remote sensing images of the target area and measured soil parameter data from ground sampling points; The multi-source remote sensing images are preprocessed, and spectral data corresponding to ground sampling points are extracted; the spectral data includes satellite multispectral data and UAV hyperspectral data. The first inversion model is constructed by calculating multiple band combinations in the satellite multispectral data, and the second inversion model is constructed by calculating multiple band combinations in the UAV hyperspectral data using the spectral transformation method. Evaluation indices are calculated for the first inversion model and the second inversion model based on the measured soil parameter data to obtain evaluation indices. Target first inversion model and target second inversion model are then selected from the first inversion model and the second inversion model, respectively, based on the evaluation indices. The preprocessed multi-source remote sensing image is inverted using the target first inversion model to generate a first soil parameter distribution map; The coverage area of ​​the first soil parameter distribution map is larger than the preset range; The regions to be inverted are selected from the regions of the first soil parameter distribution map; the regions to be inverted are those where the soil salinization degree is greater than a preset threshold. The preprocessed multi-source remote sensing image under the area to be inverted is inverted using the target second inversion model to generate a second soil parameter distribution map; the accuracy of the second soil parameter distribution map is greater than the preset accuracy.

2. The method for quantitative inversion of soil salinity using multi-source remote sensing synergy according to claim 1, characterized in that, The acquisition of multi-source remote sensing images of the target area and measured soil parameter data from ground sampling points includes: Acquire multi-source remote sensing images of the target area; wherein, the multi-source remote sensing images include satellite multispectral remote sensing images and UAV hyperspectral remote sensing images; Obtain measured soil parameter data from ground sampling points; wherein the measured soil parameter data includes total soil salinity, soil organic matter content, and / or soil moisture content.

3. The method for quantitative inversion of soil salinity using multi-source remote sensing synergy according to claim 2, characterized in that, The preprocessing of the multi-source remote sensing images includes: Geometric correction, radiometric calibration, and atmospheric correction are performed on the satellite multispectral remote sensing images in the multi-source remote sensing images; Geometric correction, radiometric calibration, atmospheric correction, image stitching, and effective band selection are performed on the UAV hyperspectral remote sensing images in the multi-source remote sensing images.

4. The method for quantitative inversion of soil salinity using multi-source remote sensing synergy according to claim 1, characterized in that, The calculation of multiple band combinations in the satellite multispectral data to construct the first inversion model includes: The satellite multispectral data and the open format files of the spatial data of the ground sampling points are loaded into the remote sensing image processing platform to obtain the image spectral data of the ground sampling points; The image spectral data and the measured soil parameter data are imported into MATLAB for calculation, and outlier removal is performed to construct the first inversion model.

5. The method for quantitative inversion of soil salinity using multi-source remote sensing synergy according to claim 1, characterized in that, The step of using spectral transformation to calculate multiple band combinations in the UAV hyperspectral data to construct a second inversion model includes: Using the spectral transformation method, the first and second derivative transformations of the UAV hyperspectral reflectance data in the UAV hyperspectral data are performed to obtain the transformed spectral data. A second inversion model is constructed based on the UAV hyperspectral data and multiple band combinations in the transformed spectral data.

6. The method for quantitative inversion of soil salinity using multi-source remote sensing synergy according to claim 1, characterized in that, The step of calculating evaluation indicators for the first and second inversion models based on the measured soil parameter data to obtain evaluation indicators, and then selecting target first and target second inversion models from the first and second inversion models respectively based on the evaluation indicators, includes: Based on the measured soil parameter data, evaluation indicators are calculated for the first inversion model and the second inversion model respectively to obtain evaluation indicators; the evaluation indicators include the coefficient of determination and the root mean square error. The evaluation index corresponding to the maximum value of the coefficient of determination and the minimum value of the root mean square error is used as the target evaluation index; the target evaluation index includes a first target evaluation index and a second target evaluation index. The first inversion model corresponding to the first target evaluation index is used as the target first inversion model; The second inversion model corresponding to the second objective evaluation index is used as the objective second inversion model.

7. The method for quantitative inversion of soil salinity using multi-source remote sensing synergy according to any one of claims 1 to 6, characterized in that, Also includes: Determine the visualization method based on business needs; The first soil parameter distribution map and the second soil parameter distribution map are visualized according to the visualization method described above.

8. A multi-source remote sensing collaborative soil salinity quantitative inversion device, characterized in that, include: The image and data acquisition module is used to acquire multi-source remote sensing images of the target area and measured soil parameter data from ground sampling points. The preprocessing module is used to preprocess the multi-source remote sensing images and extract the spectral data corresponding to the ground sampling points; the spectral data includes satellite multispectral data and UAV hyperspectral data; The inversion model construction module is used to calculate multiple band combinations in the satellite multispectral data to construct a first inversion model, and to calculate multiple band combinations in the UAV hyperspectral data using the spectral transformation method to construct a second inversion model. The evaluation index calculation module is used to calculate the evaluation index of the first inversion model and the second inversion model based on the measured soil parameter data, obtain the evaluation index, and select the target first inversion model and the target second inversion model from the first inversion model and the second inversion model respectively according to the evaluation index. The first distribution map generation module is used to invert the preprocessed multi-source remote sensing image using the target first inversion model to generate a first soil parameter distribution map. The coverage area of ​​the first soil parameter distribution map is larger than the preset range; The region screening module is used to screen out the regions to be inverted from the regions of the first soil parameter distribution map; the regions to be inverted are regions where the soil salinization degree is greater than a preset threshold. The second distribution map generation module is used to invert the preprocessed multi-source remote sensing image under the area to be inverted using the target second inversion model to generate a second soil parameter distribution map; the accuracy of the second soil parameter distribution map is greater than the preset accuracy.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the multi-source remote sensing collaborative method for quantitative inversion of soil salinity as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Used to store computer programs; wherein, when the computer programs are executed by a processor, they implement the multi-source remote sensing collaborative soil salinity quantitative inversion method as described in any one of claims 1 to 7.

Images