Soil moisture inversion method for heterogeneous covered farmland based on UAV microwave and multispectral

Abstract

A method for inverting soil moisture in heterogeneous covered farmland based on microwave and multi-spectral data from drones, the steps of which are as follows: the atmosphere is equivalent to a transparent medium, the vegetation layer is equivalent to a heterogeneous semi-transparent medium, and the ground surface is equivalent to a rough surface natural medium, so that the covered farmland under near-ground remote sensing conditions is physically abstracted as a heterogeneous multi-layer medium model; the radar echo transmission mechanism of the heterogeneous multi-layer medium model is studied by multi-physics field coupling simulation; the influence of non-homogeneous factors on soil moisture inversion is eliminated by collaboratively applying drone multi-spectral data and microwave remote sensing data, and soil moisture sensitivity parameters are extracted; the drone microwave inversion model of soil moisture in covered farmland is established by using the soil moisture sensitivity parameters. This invention will provide new technical methods and model tools for remote sensing monitoring of soil physiological and biochemical parameters of covered farmland, and at the same time provide a scientific basis for rapid perception of soil moisture and water and fertilizer regulation and management at the farmland scale with high precision and high efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of agricultural remote sensing, and in particular relates to a method for inverting soil moisture in farmland with heterogeneous coverage based on drone microwaves and multi-spectrum. Background Art

[0002] Timely and accurate acquisition and assessment of soil moisture information at the farmland scale is one of the prerequisites for achieving precision irrigation and smart agriculture with the goal of saving costs and increasing efficiency. Large-scale monitoring of soil moisture in covered farmland is a research hotspot and difficulty in the world's agricultural field.

[0003] At present, the acquisition of soil moisture information of large-scale covered farmland mainly relies on satellite remote sensing technology, using optical, thermal infrared, active and passive microwave, multi-source remote sensing and other methods, by obtaining the reflectivity and thermal radiation information of the ground and objects in different bands, and establishing the correspondence between soil moisture and reflection information, so as to achieve large-scale, rapid and long-term dynamic monitoring of farmland soil moisture information. Among them, microwave radar has been increasingly applied to the study of soil moisture inversion problems of covered farmland due to its advantages such as good range resolution, strong penetration and low power consumption. In addition, in order to further eliminate the influence of vegetation layer on the accuracy of soil moisture inversion of covered farmland, multi-source data remote sensing with comprehensive application of microwave and optical methods has become a hot topic of research. However, satellite remote sensing has the disadvantages of low temporal and spatial resolution, great influence by weather, and limitation by transit period, making it difficult to achieve real-time and accurate acquisition of spatial variability of soil moisture information at the farmland scale, and there are certain limitations in guiding irrigation and agricultural production in practice.

[0004] With its advantages of convenient transportation, high flexibility, short operation cycle and high data resolution, UAV remote sensing has become a new means to achieve rapid and accurate perception of soil moisture information at the farmland scale. Important research progress has been made in crop growth monitoring, water demand estimation, yield prediction, farmland greenhouse gas monitoring, etc. The application of UAV-borne microwave radar has also provided new ideas for the accurate perception of soil water information in covered farmland.

[0005] However, UAV remote sensing is more sensitive to the heterogeneity of farmland. The temporal and spatial resolution of the data, radar incidence angle, image width, surface roughness and coherence characteristics of radar wavelength are also quite different from satellite remote sensing. The interpretation and inversion methods of satellite remote sensing cannot be directly applied to UAV remote sensing technology. Therefore, it is urgent to study new methods for soil moisture inversion based on UAV multi-source data remote sensing in view of the prominent heterogeneity of farmland under near-ground remote sensing conditions. Summary of the invention

[0006] In view of the urgent need for high-precision, high-efficiency, and large-scale rapid perception technology of farmland soil moisture information in precision irrigation technology, as well as the insufficient monitoring capability of satellite remote sensing on the spatial variability of moisture in covered farmland, and the lack of soil moisture inversion model under near-ground conditions in UAV remote sensing, the purpose of the present invention is to provide a soil moisture inversion method for non-homogeneous covered farmland based on microwave and multi-spectral UAVs, abstract the research on electromagnetic transmission mechanism of non-homogeneous covered farmland into solving the scattering problem of multi-layer mixed model composed of non-homogeneous media under near-ground remote sensing conditions, and study the radar echo transmission mechanism of the model; synergistically apply optical and microwave remote sensing data to eliminate the influence of non-homogeneous factors on soil moisture inversion and extract soil moisture sensitivity parameters; establish a soil moisture inversion model suitable for UAVs to achieve high-precision and high-efficiency rapid perception of soil moisture information at farmland scale. This research can provide theoretical and technical basis for the rapid and accurate perception of spatial variability of soil moisture information at farmland scale, which has important scientific significance and broad application prospects.

[0007] In order to achieve the above object, the technical solution adopted by the present invention is:

[0008] A method for inverting soil moisture in farmland with heterogeneous coverage based on drone microwave and multi-spectrum includes the following steps:

[0009] Step 1: The atmosphere (aerosol layer) is equivalent to a transparent medium, the vegetation layer is equivalent to a heterogeneous semi-transparent medium, and the surface is equivalent to a rough surface natural medium, so that the covered farmland under near-ground remote sensing conditions is physically abstracted into a heterogeneous multi-layer medium model;

[0010] Step 2, using multi-physics field coupling simulation to study the radar echo transmission mechanism of the heterogeneous multilayer medium model;

[0011] Step 3: Coordinated application of UAV multispectral data and microwave remote sensing data to eliminate the influence of non-homogeneous factors on soil moisture inversion and extract soil moisture sensitivity parameters;

[0012] Step 4: Using the soil moisture sensitivity parameters, establish a drone microwave inversion model for soil moisture in covered farmland.

[0013] Furthermore, in step 1, the actual farmland surface is represented by a two-dimensional random rough surface, the farmland surface without vegetation coverage is regarded as a rough surface composed of large-scale and small-scale undulating waves or multiple scale waves continuously distributed and superimposed, and the farmland surface covered with vegetation, i.e., the covered farmland, is equivalent to a non-homogeneous multi-layer medium model; wherein the vegetation layer is regarded as a porous substance composed of air wrapped in the crop canopy, and is approximated by a mixed medium layer.

[0014] Furthermore, in step 2, the transmission coefficient, wave velocity and backscattering coefficient of microwaves in the heterogeneous multilayer medium model are obtained, and by changing the parameters of each layer of the medium in the heterogeneous multilayer medium model, the influence of the corresponding parameters on the radar echo transmission characteristics is obtained.

[0015] Furthermore, in step 3, the spectral data acquired by the drone's onboard multispectral camera, i.e., the drone's multispectral data, is preprocessed, and vegetation layer growth characteristic parameters including crop planting structure, coverage, vegetation index, and vegetation moisture content are obtained through remote sensing interpretation and multispectral data inversion.

[0016] Furthermore, based on the characteristic parameters of vegetation layer growth, the electromagnetic scattering characteristic parameters of the heterogeneous layer including its optical thickness, equivalent density and equivalent dielectric constant are obtained, the influence of the characteristic parameters of vegetation layer growth on soil moisture inversion is clarified, and the key indicators sensitive to soil moisture information are extracted, namely crop canopy height, crop canopy coverage, crop canopy moisture content and leaf area index.

[0017] The heterogeneous factors include statistical parameters of surface roughness, equivalent optical thickness of vegetation layer, and equivalent dielectric constants of soil layer and vegetation layer. The characteristic parameters of actual surface roughness and vegetation layer are extracted, the relationship between the accuracy of characterization of heterogeneous multi-layer medium model and parameter complexity is balanced, and a physical model of electromagnetic scattering of farmland surface under heterogeneous conditions is established.

[0018] Compared with the prior art, the present invention has the following beneficial effects:

[0019] (1) Breaking through the previous problems of farmland soil moisture remote sensing inversion, based on the induction and summary of experimental data, a research idea of ​​the correlation between radar echo signals and soil moisture content is established, and the random two-dimensional rough surface physical model and the non-homogeneous multi-layer medium physical model are introduced into the research problem of electromagnetic scattering characteristics of farmland surface under near-ground remote sensing conditions. The transmission mechanism of radar echoes on the surface of covered farmland is deeply revealed from the mathematical and physical levels, laying a theoretical foundation for improving the accuracy and applicability of soil moisture inversion models.

[0020] (2) Coordinated application of UAV microwave radar data and multispectral data. Multispectral data can fully extract and invert crop growth and moisture information in large areas of farmland, maximize the impact of heterogeneous factors on the backscattering characteristics of soil moisture microwave remote sensing data, and improve the accuracy of UAV soil moisture inversion.

[0021] (3) Clarify the correlation between vegetation layer growth parameters and electromagnetic scattering characteristics of covered farmland under near-ground remote sensing conditions, and expand the current soil moisture remote sensing monitoring technology and theory to the farmland scale. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a flow chart of the present invention.

[0023] Figure 2 It is a schematic diagram of a multi-layer heterogeneous mixed model under near-earth remote sensing conditions established by the present invention.

[0024] Figure 3 This is a farmland surface element electromagnetic scattering physical model established by the present invention. Figure (a) is a model without vegetation coverage, and Figure (b) is a model with vegetation coverage.

[0025] Figure 4 It is a two-dimensional random rough surface model established by the present invention, wherein Figure (a) is a model without vegetation coverage, and Figure (b) is a model with vegetation coverage. DETAILED DESCRIPTION

[0026] The embodiments of the present invention are described in detail below with reference to the accompanying drawings and examples.

[0027] like Figure 1 As shown, the present invention is a method for inverting soil moisture in farmland with heterogeneous coverage based on drone microwaves and multi-spectral, comprising the following steps:

[0028] (1) Establishing a heterogeneous multilayer medium model

[0029] The atmosphere is equivalent to a transparent medium layer, the vegetation layer is equivalent to a non-homogeneous semi-transparent medium layer, and the surface is equivalent to a rough surface natural medium layer. Thus, the covered farmland under near-ground remote sensing conditions is physically abstracted into a non-homogeneous multi-layer medium model, such as Figure 2 shown.

[0030] Specifically, the present invention uses a two-dimensional random rough surface to represent the actual farmland surface, and uses statistical parameters such as root mean square height, correlation function, power spectrum density, curvature radius, and root mean square slope to measure the roughness of the two-dimensional random rough surface. The farmland surface without vegetation coverage is regarded as a rough surface composed of large-scale and small-scale undulating waves or a continuous distribution of multiple scale waves. Figure 3 As shown in (a), the surface of the farmland covered with vegetation (i.e., covered farmland) is equivalent to a non-homogeneous multi-layer medium model, and the schematic diagram of its surface element electromagnetic scattering physical model is as follows: Figure 3 As shown in (b), the vegetation layer is regarded as a porous substance composed of air wrapped by the crop canopy, which is approximated by a non-homogeneous semi-transparent medium layer.

[0031] Among them, under near-ground remote sensing conditions, the atmospheric layer has little influence, which can be eliminated by atmospheric calibration. Under near-ground remote sensing conditions, the electromagnetic scattering characteristics are mainly affected by the vegetation layer and the rough surface. The generated random two-dimensional rough surface model is as follows Figure 4 As shown, it can be seen Figure 4(a) In the model without vegetation cover, soil roughness has a great influence, and the influence of soil layer roughness cannot be ignored in the multi-physics field coupling simulation calculation; Figure 4 (b) In the two-dimensional rough surface model with vegetation coverage, the influence of soil layer roughness is much smaller than that of vegetation layer thickness and moisture content and can be ignored.

[0032] (2) Using multi-physics field coupling simulation, the radar echo transmission mechanism of the inhomogeneous multi-layer medium model is studied.

[0033] In this embodiment, the radar bandwidth of the ultra-wideband microwave radar is 3.1G-4.8GHz, and the center frequency is 4.3GHz. The specific implementation method is as follows:

[0034] In the first step, in the model calculation, multi-physics field coupling analysis software is used to solve the amplitude, phase angle, and wave velocity of the incident wave on the upper and lower surfaces of different dielectric layers, so as to obtain the transmission coefficient, scattering coefficient, and wave velocity variation law of the electromagnetic wave in different media.

[0035] The second step is to obtain the influence of the corresponding parameters on the transmission characteristics of radar echoes by changing the parameters of each layer of the heterogeneous multilayer medium model. The parameters of the natural medium with rough surface mainly include roughness and equivalent dielectric constant of the soil layer. For the soil layer, when the crop coverage reaches a certain level, the influence of roughness can be ignored; the equivalent dielectric constant of the soil layer is generally considered to be directly related to the moisture content. By changing the equivalent dielectric constant of the soil layer, the influence of moisture content on the electromagnetic wave transmission mechanism can be obtained.

[0036] The electromagnetic scattering characteristic parameters of inhomogeneous semitransparent media include equivalent optical thickness, duty cycle, and equivalent dielectric constant. Corresponding to different growth periods, the height, coverage, leaf area index, and crop canopy moisture content of the vegetation layer change. According to the vegetation layer growth parameters, the approximate equivalent optical thickness, duty cycle, and equivalent dielectric constant value range can be obtained from the previous calculation formulas and experimental conclusions. These parameters can be set by themselves in the multi-physics field coupling analysis software. By changing the parameters, the influence of parameter changes on the electromagnetic wave transmission mechanism can be analyzed.

[0037] (3) Coordinated application of UAV multispectral data and microwave remote sensing data to eliminate the influence of non-homogeneous factors on soil moisture inversion and extract soil moisture sensitivity parameters.

[0038] The specific implementation methods are as follows:

[0039] The first step is to intercept the band containing the covered farmland information in the spectrum based on the known frequency, amplitude and incident angle of the electromagnetic wave emitted by the radar, and the radar technical parameters such as the backscattered echo spectrum received by the radar antenna, as well as the electromagnetic wave transmission distance, radar irradiation area, etc., and perform time-frequency analysis on it to obtain characteristic values ​​such as the backscattering coefficient related to the soil moisture information of the covered farmland.

[0040] The second step is to obtain the spectral data obtained by the multispectral camera on the drone (i.e., multispectral data of drones), and select the spectral data of five bands: red, green, blue, infrared, and near-infrared. Using remote sensing data interpretation software, the multispectral data of drones are first preprocessed through image stitching, orthorectification, radiation correction, geographic information registration, etc. Then, according to the color, texture, shape, spatial relationship, spectral reflectance of each band and other characteristics of the preprocessed image, the vegetation index and spectral characteristics are extracted, and based on this, the vegetation layer growth characteristic parameters including planting structure, vegetation layer coverage, vegetation index, and vegetation moisture content are inverted.

[0041] The calculation formulas for each vegetation index are as follows: Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), and Normalized Difference Water Index (NDWI):

[0042]

[0043]

[0044]

[0045] Where: R NIR is the reflectivity in the near-infrared band; R RED is the reflectivity of red light band; R BLUE is the reflectivity of blue light band, R SWIR is the reflectivity in the shortwave infrared band.

[0046] From this, we can get the quadratic vegetation water content:

[0047] M V =aI 2 +bI+c (4)

[0048] Among them: I is the vegetation index, including NDVI, EVI and NDWI.

[0049] The third step is to obtain the equivalent dielectric constant, equivalent optical thickness, and the ratio of vegetation cover to bare soil (duty cycle) of the vegetation layer based on the growth characteristic parameters of the vegetation layer and the multispectral images of the UAV.

[0050] The present invention adopts the Maxwell-Garnett mixed medium model to calculate the equivalent dielectric constant of the vegetation layer:

[0051]

[0052] Where: 0 Represents the dielectric constant of air; ε r represents the complex relative dielectric constant of the blade; ε′ r represents the real part of the complex relative dielectric constant of leaves; V represents the duty cycle of the crop canopy.

[0053] The fourth step is to eliminate the influence of the parameters of each layer of the medium in the heterogeneous multilayer medium model obtained in step (2) on the radar echo transmission characteristics, and extract soil moisture sensitivity parameters, including crop canopy height, crop canopy coverage, crop canopy moisture content, leaf area index, etc., by eliminating the influence of the non-homogeneous semi-transparent layer on the backscattering coefficient related to the soil moisture information of the covered farmland obtained by the microwave radar.

[0054] (4) Using the soil moisture sensitivity parameters, a drone microwave inversion model for soil moisture in covered farmland is established to achieve high-precision and high-efficiency rapid perception of soil moisture information at the farmland scale.

[0055] The specific implementation methods are as follows:

[0056] In the first step, based on the electromagnetic transmission mechanism of heterogeneous farmland surface and the extraction of sensitive characteristic parameters of soil moisture inversion from multi-source data of drones, a multi-index drone soil moisture inversion model under vegetation coverage conditions was established using linear regression analysis, energy balance method and other methods.

[0057] The second step is to conduct farmland experiments, simultaneously collect soil and vegetation information on the ground, select appropriate sampling points, and invert the soil moisture in the area according to the model established by the present invention. The difference analysis is performed with the ground data, the accuracy and applicability of the model are studied, and the model parameters are repeatedly adjusted and optimized to finally obtain a high-precision inhomogeneous farmland soil moisture inversion model that meets actual needs.

Claims

1. A method for soil moisture inversion of heterogeneous covered farmland based on UAV microwave and multi-spectral. It is characterized in that The steps include: Step 1: The atmosphere is equivalent to a transparent medium, the vegetation layer is equivalent to a non-homogeneous semi-transparent medium, and the surface is equivalent to a rough surface natural medium, so that the covered farmland under near-ground remote sensing conditions is physically abstracted into a non-homogeneous multi-layer medium model; The actual farmland surface is represented by a two-dimensional random rough surface, and the farmland surface without vegetation coverage is regarded as a rough surface composed of large-scale and small-scale undulating waves or multiple scale waves continuously distributed and superimposed, so that the farmland surface covered with vegetation, i.e. covered farmland, is equivalent to a non-homogeneous multi-layer medium model; the vegetation layer is regarded as a porous substance composed of crop canopy wrapped in air, and is approximately represented by a mixed medium layer; Step 2, using multi-physics field coupling simulation to study the radar echo transmission mechanism of the heterogeneous multilayer medium model, the method is as follows: The multi-physics field coupling simulation method is used to solve the amplitude, phase angle, and wave velocity of the incident wave on the upper and lower surfaces of different dielectric layers, thereby obtaining the transmission coefficient, scattering coefficient, and wave velocity variation law of electromagnetic waves in different media. By changing the parameters of each layer of the heterogeneous multilayer dielectric model, the influence of the corresponding parameter changes on the radar echo transmission characteristics is obtained. The method for obtaining the transmission coefficient, wave velocity and backscattering coefficient of the microwave in the heterogeneous multilayer medium model is: Under the premise of knowing the frequency, amplitude and incident angle of the electromagnetic wave emitted by the radar, the backscatter echo received by the radar antenna is intercepted, the band containing the covered farmland information is analyzed in time and frequency, and the backscatter coefficient related to the soil moisture information of the covered farmland is obtained; The parameters of the rough surface natural medium include roughness and soil equivalent dielectric constant. When the crop-covered soil layer reaches a set level, the influence of roughness is ignored. The soil equivalent dielectric constant is directly related to the moisture content. In the process of model solution, the soil equivalent dielectric constant is changed to obtain the influence of moisture content on the electromagnetic wave transmission mechanism. In actual measurement, the soil moisture content is inverted through the electromagnetic wave transmission law. The parameters in the heterogeneous semitransparent medium include equivalent optical thickness, duty cycle and equivalent dielectric constant; corresponding to different growth periods, the vegetation layer height, coverage, leaf area index and crop canopy moisture content change. By changing the parameters in the multi-physics field coupling analysis software, the influence of the parameter changes on the electromagnetic wave transmission mechanism can be analyzed; Step 3: Coordinated application of UAV multispectral data and microwave remote sensing data to eliminate the influence of non-homogeneous factors on soil moisture inversion and extract soil moisture sensitivity parameters; Step 4: Using the soil moisture sensitivity parameters, establish a drone microwave inversion model for soil moisture in covered farmland.

2. According to the method for inversion of soil moisture in heterogeneously covered farmland based on drone microwave and multi-spectrum according to claim 1, It is characterized in that In step 3, the electromagnetic scattering characteristic parameters of the vegetation layer are obtained using the drone multispectral data, and the spectral data obtained by the drone-mounted multispectral camera, i.e., the drone multispectral data, is preprocessed. Through remote sensing interpretation and multispectral data inversion, the vegetation layer growth characteristic parameters including crop planting structure, coverage, vegetation index, and vegetation moisture content are obtained.

3. According to the method for inversion of soil moisture in heterogeneously covered farmland based on drone microwave and multi-spectrum as described in claim 2, It is characterized in that Based on the characteristic parameters of vegetation layer growth, the electromagnetic scattering characteristic parameters of the heterogeneous layer including its optical thickness, equivalent density and equivalent dielectric constant are obtained, the influence of the characteristic parameters of vegetation layer growth on soil moisture inversion is clarified, and the key indicators sensitive to soil moisture information are extracted, namely crop canopy height, crop canopy coverage, crop canopy moisture content and leaf area index.

4. According to the method for inversion of soil moisture in heterogeneously covered farmland based on drone microwave and multi-spectrum as claimed in claim 3, It is characterized in that The heterogeneous factors include statistical parameters of surface roughness, equivalent optical thickness of vegetation layer, and equivalent dielectric constants of soil layer and vegetation layer. The characteristic parameters of actual surface roughness and vegetation layer are extracted, the relationship between the accuracy of characterization of heterogeneous multi-layer medium model and parameter complexity is balanced, and a physical model of electromagnetic scattering of farmland surface under heterogeneous conditions is established.

Images