Method and device for detecting spatial and temporal distribution of organic matter content of farmland soil

Abstract

The invention provides a farmland soil organic matter content spatial and temporal distribution detection method and device. The method comprises the following steps: acquiring hyperspectral remote sensing image data and farmland soil sample data; based on the hyperspectral remote sensing image data, obtaining first spectral curve data, and performing smooth filtering on the first spectral curve data to obtain second spectral curve data; based on the first spectrum curve data and the second spectrum curve data, screening out spectrum characteristic data related to the soil organic matter content; based on the soil organic matter content and the spectral feature data related to the soil organic matter content, applying to a preset model, constructing a target soil organic matter content remote sensing inversion model, and obtaining a soil organic matter content spatial distribution map based on the target soil organic matter content remote sensing inversion model; and performing mask treatment on the soil organic matter content spatial distribution diagram based on the spatial distribution of the farmland bare soil in the land coverage spatial distribution diagram to obtain a farmland soil organic matter content spatial distribution diagram.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method and device for detecting the spatiotemporal distribution of organic matter content in farmland soil. Background Art

[0002] Soil Organic Matter (SOM) is an important component of soil. It not only provides nutrients for plants and crops, but also plays an important role in increasing soil water content, forming soil aggregates, preventing soil erosion, and preventing soil degradation. Therefore, soil organic matter content is an important indicator for measuring soil fertility and sustainable land use, and has an important impact on soil quality. In addition, soil has a carbon fixation function. As the largest carbon pool in terrestrial ecosystems, the main form of soil carbon storage is organic matter and there is a quantitative relationship between the two. Therefore, the accurate determination of soil organic matter content is of great significance to the estimation of carbon storage in terrestrial ecosystems.

[0003] At present, the method for determining soil organic matter content is mainly to use chemical analysis methods such as potassium dichromate volumetric method on soil samples collected in the field in the laboratory, and then add spatial information interpolation to calculate the spatiotemporal distribution of organic matter content in the study area. The traditional method of determining the spatiotemporal distribution characteristics of soil organic matter has certain shortcomings. On the one hand, due to the different physical and chemical properties of the soil, the soil organic matter content varies with space, and it is difficult to effectively obtain accurate and continuous spatial distribution information by spatial information interpolation method; on the other hand, it is difficult to achieve continuous dynamic research of long time series due to the influence of large-scale on-site sampling and laboratory analysis. Summary of the invention

[0004] The present invention provides a method and device for detecting the spatiotemporal distribution of organic matter content in farmland soil, so as to solve the defects existing in the prior art.

[0005] The present invention provides a method for detecting the spatiotemporal distribution of organic matter content in farmland soil, comprising the following steps: Acquire hyperspectral remote sensing image data and farmland soil sample data of the area to be evaluated, wherein the farmland soil sample data includes soil organic matter content at multiple sampling points; Based on the hyperspectral remote sensing image data, first spectral curve data corresponding to each of the sampling points is obtained, and the first spectral curve data is smoothed and filtered to obtain second spectral curve data; Based on the first spectral curve data and the second spectral curve data corresponding to each of the sampling points, filter out spectral characteristic data related to the soil organic matter content corresponding to each of the sampling points; Based on the soil organic matter content corresponding to each of the sampling points and the spectral characteristic data related to the soil organic matter content, a preset model is applied to construct a remote sensing inversion model of the target soil organic matter content, and a spatial distribution map of the soil organic matter content in the area to be evaluated is obtained based on the remote sensing inversion model of the target soil organic matter content; The land cover spatial distribution map of the area to be evaluated is obtained, and the spatial distribution map of soil organic matter content of the area to be evaluated is masked based on the spatial distribution of bare farmland soil in the land cover spatial distribution map of the area to be evaluated to obtain the spatial distribution map of farmland soil organic matter content.

[0006] According to a method for detecting the spatiotemporal distribution of organic matter content in farmland soil provided by the present invention, the spectral characteristic data related to the soil organic matter content corresponding to each sampling point is screened out based on the first spectral curve data and the second spectral curve data corresponding to each sampling point, including: Performing first-order derivative processing on the second spectrum curve data to obtain third spectrum curve data; Performing three-point numerical derivation processing on the third spectrum curve data to obtain fourth spectrum curve data; Based on the correlation coefficient between the soil organic matter content corresponding to each sampling point and the reflectance of each band in each spectral curve data, the target band in each spectral curve data and the correlation coefficient corresponding to the target band are screened out as spectral feature data, and the spectral curve data includes the first spectral curve data, the second spectral curve data, the third spectral curve data and the fourth spectral curve data.

[0007] According to a method for detecting the spatiotemporal distribution of farmland soil organic matter content provided by the present invention, the soil organic matter content corresponding to each sampling point and the spectral characteristic data related to the soil organic matter content are applied to a preset model to construct a remote sensing inversion model of the target soil organic matter content, including: Taking the soil organic matter content corresponding to the sampling point as the dependent variable, and taking the target band in each spectral curve data and the correlation coefficient corresponding to the target band as the independent variables, the preset model matched to each spectral curve data is applied to construct a candidate soil organic matter content remote sensing inversion model matched to each spectral curve data; Based on the mean absolute error and the determination coefficient of each candidate soil organic matter content remote sensing inversion model, the target soil organic matter content remote sensing inversion model is determined.

[0008] According to a method for detecting the spatiotemporal distribution of organic matter content in farmland soil provided by the present invention, the preset model matched with each spectral curve data is obtained in the following manner: Taking the soil organic matter content corresponding to the sampling point as the dependent variable and the reflectance of all bands in each spectral curve data as the independent variable, the multiple candidate models are applied to determine the average absolute error and the determination coefficient of each candidate model; Based on the mean absolute error and determination coefficient of each candidate model, the preset model that matches each spectral curve data is determined.

[0009] According to a method for detecting the spatiotemporal distribution of organic matter content in farmland soil provided by the present invention, the farmland soil sample data is obtained in the following manner: Acquire a farmland soil sample collected in the area to be evaluated, wherein the collection time of the farmland soil sample and the collection time of the hyperspectral remote sensing image data are within a preset time interval; The soil organic matter content of the farmland soil sample is tested by potassium dichromate method to obtain the soil sample collection data.

[0010] According to a method for detecting the spatiotemporal distribution of organic matter content in farmland soil provided by the present invention, a land cover spatial distribution map of the area to be evaluated is obtained, comprising: Obtaining a principal component transformation result of the second spectral curve data corresponding to the area to be evaluated; Obtaining the homogeneity, normalized vegetation index, and improved normalized water index of the area to be evaluated; Based on the principal component transformation result, the homogeneity, the normalized vegetation index, the improved normalized water index and DEM data corresponding to the area to be evaluated, a random forest classifier is applied to perform supervised classification to obtain a land cover spatial distribution map of the area to be evaluated.

[0011] The present invention also provides a device for detecting the spatiotemporal distribution of organic matter content in farmland soil, comprising the following modules: A first data acquisition module is used to acquire hyperspectral remote sensing image data and farmland soil sample data of the area to be evaluated, wherein the farmland soil sample data includes soil organic matter content at multiple sampling points; A second data acquisition module is used to acquire first spectral curve data corresponding to each sampling point based on the hyperspectral remote sensing image data, and perform smoothing filtering on the first spectral curve data to obtain second spectral curve data; A third data acquisition module, configured to filter out spectral characteristic data related to the soil organic matter content corresponding to each of the sampling points based on the first spectral curve data and the second spectral curve data corresponding to each of the sampling points; A first distribution detection module is used to apply a preset model to construct a remote sensing inversion model of target soil organic matter content based on the soil organic matter content corresponding to each sampling point and the spectral characteristic data related to the soil organic matter content, and obtain a spatial distribution map of the soil organic matter content in the area to be evaluated based on the remote sensing inversion model of the target soil organic matter content; The second distribution detection module is used to perform supervised classification on the area to be evaluated based on a random forest classifier, obtain a land cover spatial distribution map of the area to be evaluated, and mask the spatial distribution map of soil organic matter content in the area to be evaluated based on the spatial distribution of bare soil in the land cover spatial distribution map of the area to be evaluated, so as to obtain a spatial distribution map of soil organic matter content in the area to be evaluated.

[0012] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for detecting the spatiotemporal distribution of organic matter content in farmland soil as described above is implemented.

[0013] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described methods for detecting the spatiotemporal distribution of organic matter content in farmland soil.

[0014] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned methods for detecting the spatiotemporal distribution of organic matter content in farmland soil.

[0015] The method and device for detecting the spatiotemporal distribution of farmland soil organic matter content provided by the present invention adopt remote sensing detection, ground soil sample monitoring, machine learning and other means, take hyperspectral remote sensing image data as the main data source, supplemented by ground soil sample monitoring, to construct a high-precision soil organic matter content remote sensing inversion model, and realize the acquisition of the spatial distribution map of farmland soil organic matter content in the whole region. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 It is a schematic flow chart of the method for detecting the spatiotemporal distribution of organic matter content in farmland soil provided by the present invention.

[0018] Figure 2This is the GF-5 hyperspectral remote sensing image provided by the present invention.

[0019] Figure 3 It is the original spectrum curve diagram provided by the present invention.

[0020] Figure 4 This is a spectrum curve diagram after Savitzky-Golay smoothing provided by the present invention.

[0021] Figure 5 It is the spatial distribution map of soil organic matter content provided by the present invention.

[0022] Figure 6 It is the land cover spatial distribution map provided by the present invention.

[0023] Figure 7 It is the spatial distribution map of farmland soil organic matter content provided by the present invention.

[0024] Figure 8 It is a first-order derivative spectrum curve graph provided by the present invention.

[0025] Fig. 9 It is a three-point numerical derivative spectrum curve diagram provided by the present invention.

[0026] Fig.10 It is a schematic diagram of the correlation coefficient change curve provided by the present invention.

[0027] Fig.11 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0029] Figure 1 This is one of the flow diagrams of the method for detecting the spatiotemporal distribution of organic matter content in farmland soil provided by the present invention, such as Figure 1 As shown, the method includes the following: Step 101: Acquire hyperspectral remote sensing image data and farmland soil sample data of the area to be evaluated, wherein the farmland soil sample data includes soil organic matter content at multiple sampling points; Hyperspectral remote sensing image data refers to image data with extremely high spectral resolution and spatial resolution, which can capture the subtle features of objects on the earth's surface in different spectral bands. Figure 2The hyperspectral remote sensing image data shown is obtained by the hyperspectral sensor based on the GF-5 satellite.

[0030] Here, after the hyperspectral remote sensing image data is collected, it can be preprocessed, such as removing invalid bands and stripe noise in the hyperspectral remote sensing image data, performing radiation calibration and atmospheric correction, orthorectification, image mosaicking and other preprocessing.

[0031] It should be understood that the sampling points in this embodiment are within the image coverage corresponding to the hyperspectral remote sensing image data.

[0032] Here, for each sampling point, a certain amount of farmland soil samples are collected and their organic matter content is measured. There are many methods for determining organic matter content, such as potassium dichromate volumetric method (external heating method), calcination method, etc.

[0033] In one example, the farmland soil sample data is obtained in the following manner: Acquire a farmland soil sample collected in the area to be evaluated, wherein the collection time of the farmland soil sample and the collection time of the hyperspectral remote sensing image data are within a preset time interval; The soil organic matter content of the farmland soil sample is tested by potassium dichromate method to obtain the soil sample collection data.

[0034] It should be noted that in order to ensure that the collected farmland soil sample data can be effectively combined with the hyperspectral remote sensing image data for subsequent analysis and evaluation, this embodiment stipulates that the collection time of the farmland soil samples and the collection time of the hyperspectral remote sensing image data are within the set preset time interval.

[0035] After obtaining farmland soil samples, the potassium dichromate method is used to test the soil organic matter content. For the soil samples collected at each sampling point, the test results of the organic matter content will be recorded to form soil sample collection data.

[0036] Step 102: based on the hyperspectral remote sensing image data, obtaining first spectral curve data corresponding to each sampling point, and performing smoothing filtering on the first spectral curve data to obtain second spectral curve data; Here, the first spectral curve data represents the reflectivity of each band at each sampling point determined based on the collected hyperspectral remote sensing image data.

[0037] Specifically, based on the latitude and longitude coordinate information of each sampling point, it is matched to the hyperspectral remote sensing image data to obtain the first spectral curve data of each sampling point. Figure 3 As shown, Figure 3Different color curves in the spectrum curve diagram shown represent first spectrum curve data at different sampling points.

[0038] Due to the difference in energy response of each band of the hyperspectral sensor, there is a lot of noise in the first spectrum curve data, especially in the near-infrared region, where the reflectivity changes dramatically and the signal-to-noise ratio is very low. Therefore, in this embodiment, the first spectrum curve data is smoothed and filtered to remove burr noise.

[0039] Specifically, the Savitzky-Golay filter can be sampled to perform smoothing filtering operations, as shown in the following formula:

[0040] In the formula, is the reflectivity sequence data after filtering, j is the band index, is the original reflectivity, C i is the filter coefficient, and N is the total number of data points in the filter window (i.e., 2m+1, where m is half the window size).

[0041] refer to Figure 3 and Figure 4 As shown, Figure 3 The first spectrum curve data of multiple sampling points in the first spectrum curve data are smoothed and filtered. After filtering, the high-frequency components (burrs) of the first spectrum curve data of each sampling point are basically eliminated, making the curve smooth, and the Figure 4 .

[0042] Step 103: based on the first spectral curve data and the second spectral curve data corresponding to each of the sampling points, filter out spectral characteristic data related to the soil organic matter content corresponding to each of the sampling points; The first spectral curve data and the second spectral curve data are subjected to feature analysis, and main feature data related to soil organic matter content in the spectral curve data are extracted as spectral feature data.

[0043] For example, statistical methods (such as correlation coefficient, partial least squares method, etc.) can be used to analyze the correlation between the spectral curve data of each sampling point and the soil organic matter content to find out the spectral bands or characteristic data that are significantly correlated with the organic matter content.

[0044] Step 104: based on the soil organic matter content corresponding to each of the sampling points and the spectral characteristic data related to the soil organic matter content, a preset model is applied to construct a remote sensing inversion model of the target soil organic matter content, and a spatial distribution map of the soil organic matter content of the area to be evaluated is obtained based on the remote sensing inversion model of the target soil organic matter content; Here, the spatial distribution map of soil organic matter content refers to an image that can reflect the distribution of soil organic matter content in different geographical locations, such as Figure 5 As shown, Figure 5 The different soil organic matter contents are marked by the depth of color.

[0045] Specifically, the soil organic matter content corresponding to the sampling point can be used as the dependent variable, and the spectral feature data related to the soil organic matter content in each spectral curve data can be used as the independent variable, and the preset model for matching each type of spectral curve data can be applied to construct a candidate soil organic matter content remote sensing inversion model that matches each spectral curve data; finally, when the independent variable data gradually increases, the changing trend of the average absolute error and the determination coefficient of each candidate soil organic matter content remote sensing inversion model is judged, and finally the candidate soil organic matter content remote sensing inversion model with the most stable changing trend is selected as the target soil organic matter content remote sensing inversion model.

[0046] Step 105: Obtain a land cover spatial distribution map of the area to be evaluated, and mask the spatial distribution map of soil organic matter content of the area to be evaluated based on the spatial distribution of bare farmland soil in the land cover spatial distribution map of the area to be evaluated to obtain a spatial distribution map of farmland soil organic matter content.

[0047] The land cover spatial distribution map here refers to the spatial distribution map divided into different land cover types, such as reference Figure 6 The spatial distribution map of the seven land cover types shown in the figure is divided into bare farmland soil, natural bare land, vegetation cover, water bodies, impervious surfaces, snow cover and cloud areas.

[0048] In some embodiments, obtaining the land cover spatial distribution map of the area to be evaluated includes: Obtaining a principal component transformation result of the second spectral curve data corresponding to the area to be evaluated; Obtaining the homogeneity, normalized vegetation index, and improved normalized water index of the area to be evaluated; Based on the principal component transformation result, the homogeneity, the normalized vegetation index, the improved normalized water index and DEM data corresponding to the area to be evaluated, a random forest classifier is applied to perform supervised classification to obtain a land cover spatial distribution map of the area to be evaluated.

[0049] It should be understood that the second spectral curve data usually contains multiple bands, and there may be redundant information between these bands. In this embodiment, the second spectral curve data is subjected to PCA principal component transformation to convert these data into a set of new, mutually orthogonal variables (i.e., principal components), thereby reducing the dimension of the data and facilitating subsequent processing and analysis.

[0050] Here, homogeneity is an indicator that describes the uniformity of the distribution of pixel values ​​in an image or region. The Normalized Difference Vegetation Index (NDVI) reflects the growth status and coverage of vegetation by calculating the difference in reflectance between the red band and the near-infrared band. The Modified Normalized Difference Water Index (MNDWI) is an improved version of the Normalized Difference Water Index (NDWI). The MNDWI index uses the short-wave infrared band to replace the near-infrared band, which can significantly reduce the interference of shadows such as buildings on water extraction, thereby improving the accuracy of water extraction.

[0051] In one example, a random forest classifier can be trained using training samples of seven types of land features: bare soil in farmland, natural bare land, vegetation cover, water body, impervious surface, snow cover, and cloud area, which will not be described in detail here.

[0052] In this embodiment, after obtaining the land cover spatial distribution map of the area to be evaluated, a mask layer is created based on the farmland bare soil spatial distribution map. This mask layer is a binary image, in which the farmland bare soil area is assigned a value and other areas are assigned b. The soil organic matter content spatial distribution map is superimposed on the farmland bare soil mask layer, and only the soil organic matter content information of the area with a value in the mask layer is retained, which is the farmland soil organic matter content spatial distribution map. Figure 7 As shown, Figure 7 That is to pass Figure 6 Bare soil area in farmland Figure 5 The image obtained by masking.

[0053] In this embodiment, by adopting remote sensing detection, ground soil sample monitoring, machine learning and other means, hyperspectral remote sensing image data is used as the main data source, supplemented by ground soil sample monitoring, to construct a high-precision soil organic matter content remote sensing inversion model, so as to obtain the spatial distribution map of farmland soil organic matter content in the entire region.

[0054] In some embodiments, the filtering out of spectral characteristic data related to the soil organic matter content corresponding to each sampling point based on the first spectral curve data and the second spectral curve data corresponding to each sampling point includes: Performing first-order derivative processing on the second spectrum curve data to obtain third spectrum curve data; Performing three-point numerical derivation processing on the third spectrum curve data to obtain fourth spectrum curve data; Based on the correlation coefficient between the soil organic matter content corresponding to each sampling point and the reflectance of each band in each spectral curve data, the target band in each spectral curve data and the correlation coefficient corresponding to the target band are screened out as spectral feature data, and the spectral curve data includes the first spectral curve data, the second spectral curve data, the third spectral curve data and the fourth spectral curve data.

[0055] In this embodiment, the first-order derivative processing is first performed on the second spectrum curve data to obtain the corresponding third spectrum curve data (such as Figure 8 ).

[0056] Specifically, the first-order derivative calculation formula is as follows: ; Next, the third spectrum curve data is numerically derived at three points to obtain the fourth spectrum curve data (such as Fig. 9 The three-point numerical derivation process will not be described in detail here.

[0057] After obtaining four different sets of spectral curve data, the correlation coefficients between the soil organic matter content corresponding to all sampling points and the reflectance of each band represented by each set of spectral curve data were calculated respectively. Fig.10 As shown. Then, according to the correlation analysis results of each group of spectral curve data, multiple bands with high correlation and corresponding correlation coefficients in the correlation analysis results of each group of spectral curve data are determined. For example, for the correlation analysis results of each group of spectral curve data, the top three bands with the top three correlation coefficients are selected as target bands. The top three bands and their corresponding correlation coefficients are the required spectral feature data.

[0058] Furthermore, the soil organic matter content corresponding to each sampling point and the spectral characteristic data related to the soil organic matter content are applied to a preset model to construct a remote sensing inversion model of the target soil organic matter content, including: Taking the soil organic matter content corresponding to the sampling point as the dependent variable, and taking the target band in each spectral curve data and the correlation coefficient corresponding to the target band as the independent variables, the preset model matched to each spectral curve data is applied to construct a candidate soil organic matter content remote sensing inversion model matched to each spectral curve data; Based on the mean absolute error and the determination coefficient of each candidate soil organic matter content remote sensing inversion model, the target soil organic matter content remote sensing inversion model is determined.

[0059] Specifically, the soil organic matter content corresponding to the sampling point is taken as the dependent variable, and the target band related to the soil organic matter content in each spectral curve data and the correlation coefficient corresponding to the target band are taken as independent variables. The preset model for matching each type of spectral curve data is applied to construct a candidate soil organic matter content remote sensing inversion model matching each spectral curve data.

[0060] Next, based on the candidate soil organic matter content remote sensing inversion model matched with each spectral curve data, all remaining bands and corresponding correlation coefficients except the target band in each spectral curve data are traversed, and the model accuracy is tested by cross-validation. The band with the highest accuracy is retained as the independent variable, and then the cycle is repeated again to traverse all remaining bands. Finally, the change trend of the mean absolute error and determination coefficient of each candidate soil organic matter content remote sensing inversion model is judged when the input of the independent variable gradually increases, and finally the candidate soil organic matter content remote sensing inversion model with the most stable change trend is selected as the target soil organic matter content remote sensing inversion model.

[0061] Specifically, the preset model matched with each spectral curve data is obtained in the following manner: Taking the soil organic matter content corresponding to the sampling point as the dependent variable and the reflectance of all bands in each spectral curve data as the independent variable, the multiple candidate models are applied to determine the average absolute error and the determination coefficient of each candidate model; Based on the mean absolute error and determination coefficient of each candidate model, the preset model that matches each spectral curve data is determined.

[0062] In one example, the multiple candidate models are Support Vector Machine (SVM), Random Forest (RF), XGBoost and Cubist algorithms.

[0063] In this embodiment, with respect to the selection of the preset model corresponding to each spectral curve data, the reflectance of all bands in each spectral curve data is used as the independent variable and input into each candidate model, and the soil organic matter content corresponding to the sampling point is used as the dependent variable. The determination coefficient and the mean absolute error corresponding to each candidate model are calculated through cross-validation. Finally, based on the determination coefficient and the mean absolute error, the model with the highest algorithm accuracy (i.e., the larger the determination coefficient, the smaller the mean absolute error) is selected from multiple candidate models as the preset model for matching this spectral curve data.

[0064] In this embodiment, a preset model matching each group of spectral curve data is first selected from multiple candidate models, and then the preset model is trained according to the target band screened out by the features and the correlation coefficient corresponding to the target band to construct a candidate soil organic matter content remote sensing inversion model, and finally the target soil organic matter content remote sensing inversion model with the highest accuracy is screened out from multiple candidate soil organic matter content remote sensing inversion models, which effectively solves the deficiency in ordinary linear models that the influencing factors of soil organic matter content are complex and there is no strict linear statistical law between it and the soil spectrum, resulting in reduced estimation accuracy.

[0065] The device for detecting the spatiotemporal distribution of organic matter content in farmland soil provided by the present invention is described below. The device for detecting the spatiotemporal distribution of organic matter content in farmland soil described below and the method for detecting the spatiotemporal distribution of organic matter content in farmland soil described above can be referred to each other.

[0066] In this embodiment, the device for detecting the spatiotemporal distribution of organic matter content in farmland soil includes: A first data acquisition module is used to acquire hyperspectral remote sensing image data and farmland soil sample data of the area to be evaluated, wherein the farmland soil sample data includes soil organic matter content at multiple sampling points; A second data acquisition module is used to acquire first spectral curve data corresponding to each sampling point based on the hyperspectral remote sensing image data, and perform smoothing filtering on the first spectral curve data to obtain second spectral curve data; A third data acquisition module, configured to filter out spectral characteristic data related to the soil organic matter content corresponding to each of the sampling points based on the first spectral curve data and the second spectral curve data corresponding to each of the sampling points; A first distribution detection module is used to apply a preset model to construct a remote sensing inversion model of target soil organic matter content based on the soil organic matter content corresponding to each sampling point and the spectral characteristic data related to the soil organic matter content, and obtain a spatial distribution map of the soil organic matter content in the area to be evaluated based on the remote sensing inversion model of the target soil organic matter content; The second distribution detection module is used to perform supervised classification on the area to be evaluated based on a random forest classifier, obtain a land cover spatial distribution map of the area to be evaluated, and mask the spatial distribution map of soil organic matter content in the area to be evaluated based on the spatial distribution of bare soil in the land cover spatial distribution map of the area to be evaluated, so as to obtain a spatial distribution map of soil organic matter content in the area to be evaluated.

[0067] The device for detecting the spatiotemporal distribution of organic matter content in farmland soil provided in this embodiment adopts a variety of means such as remote sensing detection, ground soil sample monitoring, and machine learning, takes hyperspectral remote sensing image data as the main data source, supplemented by ground soil sample monitoring, to construct a high-precision remote sensing inversion model for soil organic matter content, and realizes the acquisition of the spatial distribution map of the organic matter content in farmland soil over the entire region.

[0068] Fig.11 An example of a physical structure diagram of an electronic device is shown in FIG. Fig.11 As shown, the electronic device may include: a processor 1110, a communication interface 1120, a memory 1130 and a communication bus 1140, wherein the processor 1110, the communication interface 1120 and the memory 1130 communicate with each other through the communication bus 1140. The processor 1110 may call the logic instructions in the memory 1130 to execute the method for detecting the spatiotemporal distribution of organic matter content in farmland soil, and the method includes: Acquire hyperspectral remote sensing image data and farmland soil sample data of the area to be evaluated, wherein the farmland soil sample data includes soil organic matter content at multiple sampling points; Based on the hyperspectral remote sensing image data, first spectral curve data corresponding to each of the sampling points is obtained, and the first spectral curve data is smoothed and filtered to obtain second spectral curve data; Based on the first spectral curve data and the second spectral curve data corresponding to each of the sampling points, filter out spectral characteristic data related to the soil organic matter content corresponding to each of the sampling points; Based on the soil organic matter content corresponding to each of the sampling points and the spectral characteristic data related to the soil organic matter content, a preset model is applied to construct a remote sensing inversion model of the target soil organic matter content, and a spatial distribution map of the soil organic matter content in the area to be evaluated is obtained based on the remote sensing inversion model of the target soil organic matter content; The land cover spatial distribution map of the area to be evaluated is obtained, and the spatial distribution map of soil organic matter content of the area to be evaluated is masked based on the spatial distribution of bare farmland soil in the land cover spatial distribution map of the area to be evaluated to obtain the spatial distribution map of farmland soil organic matter content.

[0069] In addition, the logic instructions in the above-mentioned memory 1130 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0070] On the other hand, the present invention also provides a computer program product, the computer program product includes a computer program, the computer program can be stored in a non-transitory computer-readable storage medium, when the computer program is executed by a processor, the computer can execute the method for detecting the spatiotemporal distribution of organic matter content in farmland soil provided by the above methods, the method comprising: Acquire hyperspectral remote sensing image data and farmland soil sample data of the area to be evaluated, wherein the farmland soil sample data includes soil organic matter content at multiple sampling points; Based on the hyperspectral remote sensing image data, first spectral curve data corresponding to each of the sampling points is obtained, and the first spectral curve data is smoothed and filtered to obtain second spectral curve data; Based on the first spectral curve data and the second spectral curve data corresponding to each of the sampling points, filter out spectral characteristic data related to the soil organic matter content corresponding to each of the sampling points; Based on the soil organic matter content corresponding to each of the sampling points and the spectral characteristic data related to the soil organic matter content, a preset model is applied to construct a remote sensing inversion model of the target soil organic matter content, and a spatial distribution map of the soil organic matter content in the area to be evaluated is obtained based on the remote sensing inversion model of the target soil organic matter content; The land cover spatial distribution map of the area to be evaluated is obtained, and the spatial distribution map of soil organic matter content of the area to be evaluated is masked based on the spatial distribution of bare farmland soil in the land cover spatial distribution map of the area to be evaluated to obtain the spatial distribution map of farmland soil organic matter content.

[0071] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the computer program is executed by a processor to execute the method for detecting the spatiotemporal distribution of organic matter content in farmland soil provided by the above methods, the method comprising: Acquire hyperspectral remote sensing image data and farmland soil sample data of the area to be evaluated, wherein the farmland soil sample data includes soil organic matter content at multiple sampling points; Based on the hyperspectral remote sensing image data, first spectral curve data corresponding to each of the sampling points is obtained, and the first spectral curve data is smoothed and filtered to obtain second spectral curve data; Based on the first spectral curve data and the second spectral curve data corresponding to each of the sampling points, filter out spectral characteristic data related to the soil organic matter content corresponding to each of the sampling points; Based on the soil organic matter content corresponding to each of the sampling points and the spectral characteristic data related to the soil organic matter content, a preset model is applied to construct a remote sensing inversion model of the target soil organic matter content, and a spatial distribution map of the soil organic matter content in the area to be evaluated is obtained based on the remote sensing inversion model of the target soil organic matter content; The land cover spatial distribution map of the area to be evaluated is obtained, and the spatial distribution map of soil organic matter content of the area to be evaluated is masked based on the spatial distribution of bare farmland soil in the land cover spatial distribution map of the area to be evaluated to obtain the spatial distribution map of farmland soil organic matter content.

[0072] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0073] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting the spatiotemporal distribution of organic matter content in farmland soil, characterized in that: include: Acquire hyperspectral remote sensing image data and farmland soil sample data of the area to be evaluated, wherein the farmland soil sample data includes soil organic matter content at multiple sampling points; Based on the hyperspectral remote sensing image data, first spectral curve data corresponding to each of the sampling points is obtained, and the first spectral curve data is smoothed and filtered to obtain second spectral curve data; Based on the first spectral curve data and the second spectral curve data corresponding to each of the sampling points, filter out spectral characteristic data related to the soil organic matter content corresponding to each of the sampling points; Based on the soil organic matter content corresponding to each of the sampling points and the spectral characteristic data related to the soil organic matter content, a preset model is applied to construct a remote sensing inversion model of the target soil organic matter content, and a spatial distribution map of the soil organic matter content in the area to be evaluated is obtained based on the remote sensing inversion model of the target soil organic matter content; The land cover spatial distribution map of the area to be evaluated is obtained, and the spatial distribution map of soil organic matter content of the area to be evaluated is masked based on the spatial distribution of bare farmland soil in the land cover spatial distribution map of the area to be evaluated to obtain the spatial distribution map of farmland soil organic matter content.

2. The method for detecting the spatiotemporal distribution of organic matter content in farmland soil according to claim 1, characterized in that: The filtering out spectral characteristic data related to the soil organic matter content corresponding to each sampling point based on the first spectral curve data and the second spectral curve data corresponding to each sampling point includes: Performing first-order derivative processing on the second spectrum curve data to obtain third spectrum curve data; Performing three-point numerical derivation processing on the third spectrum curve data to obtain fourth spectrum curve data; Based on the correlation coefficient between the soil organic matter content corresponding to each sampling point and the reflectance of each band in each spectral curve data, the target band in each spectral curve data and the correlation coefficient corresponding to the target band are screened out as spectral feature data, and the spectral curve data includes the first spectral curve data, the second spectral curve data, the third spectral curve data and the fourth spectral curve data.

3. The method for detecting the spatiotemporal distribution of organic matter content in farmland soil according to claim 2, characterized in that: The soil organic matter content corresponding to each sampling point and the spectral characteristic data related to the soil organic matter content are applied to a preset model to construct a remote sensing inversion model of the target soil organic matter content, including: Taking the soil organic matter content corresponding to the sampling point as the dependent variable, and taking the target band in each spectral curve data and the correlation coefficient corresponding to the target band as the independent variables, the preset model matched to each spectral curve data is applied to construct a candidate soil organic matter content remote sensing inversion model matched to each spectral curve data; Based on the mean absolute error and the determination coefficient of each candidate soil organic matter content remote sensing inversion model, the target soil organic matter content remote sensing inversion model is determined.

4. The method for detecting the spatiotemporal distribution of organic matter content in farmland soil according to claim 3, characterized in that: The preset model for matching each spectral curve data is obtained in the following manner: Taking the soil organic matter content corresponding to the sampling point as the dependent variable and the reflectance of all bands in each spectral curve data as the independent variable, the multiple candidate models are applied to determine the average absolute error and the determination coefficient of each candidate model; Based on the mean absolute error and determination coefficient of each candidate model, the preset model that matches each spectral curve data is determined.

5. The method for detecting the spatiotemporal distribution of organic matter content in farmland soil according to claim 1, characterized in that: The farmland soil sample data is obtained in the following way: Acquire a farmland soil sample collected in the area to be evaluated, wherein the collection time of the farmland soil sample and the collection time of the hyperspectral remote sensing image data are within a preset time interval; The soil organic matter content of the farmland soil sample is tested by potassium dichromate method to obtain the soil sample collection data.

6. The method for detecting the spatiotemporal distribution of organic matter content in farmland soil according to claim 1, characterized in that: Obtaining a land cover spatial distribution map of the area to be evaluated, including: Obtaining a principal component transformation result of the second spectral curve data corresponding to the area to be evaluated; Obtaining the homogeneity, normalized vegetation index, and improved normalized water index of the area to be evaluated; Based on the principal component transformation result, the homogeneity, the normalized difference vegetation index, the improved normalized difference vegetation index and DEM data corresponding to the area to be evaluated, a random forest classifier is applied to perform supervised classification to obtain a land cover spatial distribution map of the area to be evaluated.

7. A device for detecting the spatiotemporal distribution of organic matter content in farmland soil, characterized in that: include: A first data acquisition module is used to acquire hyperspectral remote sensing image data and farmland soil sample data of the area to be evaluated, wherein the farmland soil sample data includes soil organic matter content at multiple sampling points; A second data acquisition module is used to acquire first spectral curve data corresponding to each sampling point based on the hyperspectral remote sensing image data, and perform smoothing filtering on the first spectral curve data to obtain second spectral curve data; A third data acquisition module, configured to filter out spectral characteristic data related to the soil organic matter content corresponding to each of the sampling points based on the first spectral curve data and the second spectral curve data corresponding to each of the sampling points; A first distribution detection module is used to apply a preset model to construct a remote sensing inversion model of target soil organic matter content based on the soil organic matter content corresponding to each sampling point and the spectral characteristic data related to the soil organic matter content, and obtain a spatial distribution map of the soil organic matter content in the area to be evaluated based on the remote sensing inversion model of the target soil organic matter content; The second distribution detection module is used to perform supervised classification on the area to be evaluated based on a random forest classifier, obtain a land cover spatial distribution map of the area to be evaluated, and mask the spatial distribution map of soil organic matter content in the area to be evaluated based on the spatial distribution of bare soil in the land cover spatial distribution map of the area to be evaluated, so as to obtain a spatial distribution map of soil organic matter content in the area to be evaluated.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method for detecting the spatiotemporal distribution of organic matter content in farmland soil as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for detecting the spatiotemporal distribution of organic matter content in farmland soil as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for detecting the spatiotemporal distribution of organic matter content in farmland soil as described in any one of claims 1 to 6 is implemented.

Images