Soil water content estimation method fusing multi-source time sequence remote sensing data

Abstract

The invention discloses a soil water content estimation method fusing multi-source time sequence remote sensing data, and relates to the field of optical detection, and the method comprises the following steps: 1, obtaining an SAR remote sensing image of Sentinel-1 and an optical remote sensing image of Sentinel-2 in a designated area, carrying out the image fusion through a PCA algorithm, carrying out the feature set extraction of the fused images, and carrying out the feature set extraction of the feature set; a minimum analysis unit is generated through image segmentation, geometric shape features of the minimum analysis unit are extracted, a crop planting area recognition model is constructed through a convolutional neural network algorithm, and a crop coverage area is obtained; 2, introducing an MIMICS microwave scattering model, and constructing a crop multi-growth-period microwave scattering model; the SAR image and the optical image are fused through the PCA algorithm, the crop planting area recognition model based on multi-source remote sensing image fusion and the convolutional neural network algorithm is constructed, and the accuracy of crop planting area recognition is improved by utilizing complementary advantages of remote sensing images.

Description

Technical Field

[0001] The present invention relates to the field of optical detection technology, and in particular to a soil moisture estimation method by fusing multi-source time series remote sensing data. Background Art

[0002] Soil moisture directly affects the water and energy exchange between the earth's surface and the atmosphere. It is a key input parameter for hydrological models, climate models, and land surface process models. It plays an important role in the study of global and regional water cycles, energy balance, and climate change. Timely, accurate, and high-throughput SMC monitoring has important scientific value and practical significance for climate change, water resource distribution, agricultural soil moisture forecasting, and water use management.

[0003] Remote sensing monitoring methods for soil moisture content include optical remote sensing, thermal infrared remote sensing and microwave remote sensing, which are the main means of remote sensing detection. The dielectric properties of soil are closely related to soil moisture content, which is reflected in the changes of gray value and brightness temperature in remote sensing images. Imaging radar is used to obtain the total backscattering coefficient. After eliminating the vegetation scattering part, the relationship between the surface backscattering coefficient and the soil dielectric constant and soil moisture content is constructed.

[0004] However, the backscattering coefficient is not only closely related to soil moisture content, but also affected by multiple factors such as radar parameters, ground cover, and other surface parameters. Especially under vegetation cover conditions, the signal composition is very complex. The inversion accuracy of the two commonly used soil moisture inversion methods, whether based on the empirical model of statistical analysis or the inversion algorithm based on the scattering model, will be affected by the uncertain relationship between soil moisture content and backscattering coefficient.

[0005] With the involvement of more and more researchers, optical remote sensing, SAR remote sensing or microwave remote sensing technologies have been combined to achieve the fusion of multi-source parameters. How to remove the influence of uncertainties such as vegetation and soil parameters on the accuracy of soil moisture inversion, and how to establish a relationship model between soil moisture content and microwave backscattering coefficient are currently important research directions for multi-source remote sensing inversion of soil moisture. Summary of the invention

[0006] 1. Technical issues to be resolved

[0007] In view of the above-mentioned shortcomings of the prior art, the present invention provides a soil moisture estimation method by fusing multi-source time series remote sensing data, which can effectively solve the problems of the prior art.

[0008] (II) Technical solution

[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0010] The present invention discloses a soil moisture estimation method integrating multi-source time series remote sensing data, comprising the following steps:

[0011] Step 1: Obtain the SAR remote sensing image of Sentinel-1 and the optical remote sensing image of Sentinel-2 in the specified area, fuse the images using the PCA algorithm, extract feature sets from the fused images, generate the minimum analysis unit through image segmentation, extract its geometric shape features, build a crop planting area recognition model using the convolutional neural network algorithm, and obtain the crop coverage area;

[0012] Step 2: Introduce the MIMI CS microwave scattering model, obtain the growth parameters and soil parameters of crops in different growth stages, define the model components and basic parameters of model operation, and build a microwave scattering model for crops in multiple growth stages;

[0013] Step 3: Based on the total backscattering coefficient provided by Sentinel-1 and the crop moisture content provided by Sentinel-2, the contribution of the crop volume scattering and the secondary scattering between the crop and the ground in the total backscattering is eliminated through the microwave scattering model of the crop during multiple growth periods. A soil moisture estimation model between the soil moisture content and the soil backscattering coefficient under different polarization modes is established using a machine learning algorithm, and the model is verified based on the measured soil and crop parameters on the ground.

[0014] Step 4: Apply the soil moisture estimation model to the crop coverage area extracted by the crop planting area identification model, perform high-temporal and spatial resolution inversion of soil moisture content on the surface covered by crops at different growth stages, and output high-temporal and spatial resolution mapping of soil moisture distribution;

[0015] Step 5: Output the analysis and evaluation results of the potential of Sentinel-1 and Sentinel-2 image data in the soil moisture estimation model.

[0016] Furthermore, the image fusion process in step 1 is specifically as follows:

[0017] Step 11: spatially register the optical image with the SAR image and resample the optical image to the same size as the SAR image;

[0018] Step 12: Perform principal component transformation on the optical image to obtain the first principal component and other components; perform histogram matching between the SAR image and the first principal component obtained by PCA transformation;

[0019] Step 13: Replace the first principal component with the SAR image generated by histogram matching, and perform inverse principal component transformation on it together with the remaining components to obtain the fused image.

[0020] Furthermore, the growth parameters of the crop in step 2 include: leaf parameters, specifically leaf moisture content, leaf cover, leaf length, leaf width, leaf thickness and leaf density, stem parameters, specifically stem moisture content, stem height, ear parameters, specifically ear moisture content, ear cover, ear length, ear diameter and ear density, and the soil parameters include: soil moisture content and soil roughness parameters.

[0021] Furthermore, the construction process of the crop multi-growth period microwave scattering model in step 2 is divided into four growth stages, including: sowing period, sowing-jointing period, jointing-grain filling period and grain filling-maturity period.

[0022] Furthermore, the surface soil layer of the area constructed during the sowing period, the construction contents of the microwave scattering models of the sowing-jointing period, the jointing-grain filling period and the grain filling-maturity period are divided into the underlying soil layer and the vegetation layer.

[0023] Furthermore, the soil layer modeling is performed by random rough surface modeling through the AI ​​EM model, and the vegetation layer is modeled through vector radiation transmission.

[0024] Furthermore, in the process of verifying the model according to the ground-measured soil and crop parameters in step 3, the estimated soil moisture content is compared with the ground-measured data, error indicators such as root mean square error and determination coefficient are calculated, the sources of uncertainty in the model are analyzed, the performance of the model at different time and space scales is evaluated, and the influence coefficient of the time resolution and spatial resolution of the data is analyzed.

[0025] Furthermore, during the analysis and evaluation process in step 5, the estimated soil moisture content is compared with the ground measured data to analyze the impact of seasonal changes on the accuracy of soil moisture estimation, the impact of different crop types and growth periods on microwave scattering characteristics, and the sensitivity of soil moisture estimation, and evaluate the convenience and cost of data acquisition.

[0026] (III) Beneficial effects

[0027] Compared with the known prior art, the technical solution provided by the present invention has the following beneficial effects:

[0028] 1. The Sentinel-1 SAR image and the Sentinel-2 optical image are fused through the PCA algorithm to construct a crop planting area recognition model based on multi-source remote sensing image fusion and convolutional neural network algorithm. The convolutional neural network algorithm is combined to perform high-precision crop planting area recognition, which utilizes the complementary advantages of remote sensing images and improves the accuracy of crop planting area recognition.

[0029] 2. By considering the canopy characteristics of crops in different growth stages and the interaction with the ground surface, the MIMICS microwave scattering model is improved and simplified, and a microwave scattering model of crops is established for different growth stages of crops. This better describes the scattering characteristics of crops in different growth stages, improves the inversion accuracy, achieves a detailed simulation of the soil and vegetation layers, and enhances the accuracy of soil moisture estimation.

[0030] 3. By extracting the backscatter coefficient and crop parameters from the Sentinel-1 and Sentinel-2 image data and combining them with machine learning algorithms, a semi-empirical model between soil moisture content and soil backscatter coefficient is established. This model can estimate soil moisture content at high temporal and spatial resolution and is suitable for different crop growth periods.

[0031] 4. By applying the soil moisture estimation model proposed in the study to crop coverage areas and combining it with multi-temporal remote sensing data, high temporal and spatial resolution soil moisture inversion was achieved. This inversion and mapping method can effectively monitor the soil moisture conditions of crops in different growth stages. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0033] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments 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 making creative work are within the scope of protection of the present invention.

[0035] The present invention will be further described below in conjunction with the embodiments.

[0036] Example

[0037] A soil moisture estimation method for fusing multi-source time series remote sensing data in this embodiment is as follows: Figure 1 As shown, the following steps are included:

[0038] Step 1: Get the SAR remote sensing image of Sentinel-1 and the optical remote sensing image of Sentinel-2 in the designated area, fuse the images through the PCA algorithm, extract the feature set of the fused images, generate the minimum analysis unit through image segmentation, extract its geometric shape features, build a crop planting area recognition model through the convolutional neural network algorithm, and obtain the crop coverage area; fully utilize the advantages of the two remote sensing data. SAR images are not affected by weather and light, and provide surface structure information; optical images provide rich spectral information and can distinguish different types of land objects; PCA algorithm can effectively reduce data dimensions, extract main features, and improve the efficiency and accuracy of subsequent analysis. The crop planting area recognition model is built through CNN, which can accurately distinguish the crop coverage area. This process improves the accuracy of soil moisture estimation because it ensures that the estimation is concentrated in the actual crop growth area;

[0039] Step 2: Introduce the MIMCS microwave scattering model, obtain the growth parameters and soil parameters of crops in different growth stages, define the model components and basic parameters of the model operation, and construct a microwave scattering model for crops in multiple growth stages; the crop growth parameters include: leaf parameters, specifically leaf water content, leaf cover, leaf length, leaf width, leaf thickness and leaf density, stem parameters, specifically stem water content and stem height, ear parameters, specifically ear water content, ear cover, ear length, ear diameter and ear density, and the soil parameters include: soil water content and soil roughness parameters;

[0040] The construction process of the microwave scattering model of crops in multiple growth periods is divided into four growth stages, including sowing period, sowing-jointing period, jointing-grain filling period and grain filling-maturity period. The surface soil layer of the area is constructed in the sowing period, and the construction content of the microwave scattering model of the sowing-jointing period, jointing-grain filling period and grain filling-maturity period is divided into the underlying soil layer and the vegetation layer. The soil layer modeling is carried out by random rough surface modeling through the AI ​​EM model, and the vegetation layer is modeled by vector radiation transmission.

[0041] The model can simulate the scattering characteristics of crops at different growth stages, providing an important physical basis for estimating soil moisture content. It covers the modeling of the sowing period, jointing period, filling period and maturity period, enabling the model to dynamically reflect the changes in the crop growth process, making the estimation of soil moisture content more timely and accurate. The soil layer and vegetation layer are modeled respectively through the AI ​​EM model and the vector radiation transmission method, which refines the physical structure of the model and improves the ability to describe the microwave scattering characteristics of different growth stages.

[0042] Step 3: Based on the total backscattering coefficient provided by Sentinel-1 and the crop moisture content provided by Sentinel-2, the contribution of the crop volume scattering and the secondary scattering between the crop and the ground in the total backscattering is eliminated through the crop multi-growth period microwave scattering model, and a soil moisture estimation model between the soil moisture content and the soil backscattering coefficient under different polarization modes is established by using a machine learning algorithm, and the model is verified according to the ground measured soil and crop parameters; in the process of verifying the model according to the ground measured soil and crop parameters, the estimated soil moisture content is compared with the ground measured data, and error indicators such as the root mean square error and the determination coefficient are calculated, the sources of uncertainty in the model are analyzed, the performance of the model at different time and space scales is evaluated, and the influence coefficient of the time resolution and spatial resolution of the data is analyzed; it can handle complex nonlinear relationships and improve the estimation accuracy. The model is verified using measured data, and error indicators such as the root mean square error and the determination coefficient are calculated. The reliability of the model can be systematically evaluated and the sources of uncertainty can be identified. By analyzing the influence of the time and space resolution of the data on the model performance, the strategies of data acquisition and model application can be optimized to improve the accuracy and efficiency of estimation;

[0043] Step 4: Apply the soil moisture estimation model to the crop coverage area extracted by the crop planting area identification model, perform high-temporal and spatial resolution inversion of soil moisture content of the crop-covered surface at different growth stages, and output high-temporal and spatial resolution mapping of soil moisture distribution; Applying the estimation model to the identified crop coverage area can perform high-temporal and spatial resolution soil moisture inversion for different growth stages, thereby improving the spatial details and temporal dynamics of the estimation results. High-temporal and spatial resolution soil moisture mapping can provide accurate data support for agricultural management, such as irrigation planning and crop yield prediction;

[0044] Step 5: Output the results of analyzing and evaluating the potential of Sentinel-1 and Sentinel-2 image data in the soil moisture estimation model. During the analysis and evaluation process, the estimated soil moisture content is compared with the ground measured data, and the impact of seasonal changes on the accuracy of soil moisture estimation, the impact of different crop types and growth periods on microwave scattering characteristics, and the sensitivity of soil moisture estimation are analyzed. The convenience and cost of data acquisition are evaluated, which can provide a basis for the optimization and improvement of the soil moisture estimation model, help identify key influencing factors, and improve the robustness of the model. By analyzing the difficulty and cost of data acquisition, the practicality and promotion potential of the method can be evaluated, providing guidance for related applications.

[0045] In this embodiment, the crop parameters and soil parameters are measured by:

[0046] The size parameters of leaves, stems and ears in the crop parameters were measured with a tape measure. The density of leaves and ears was calculated by the number of leaves / number of ears in a specific area. The sampled plants were brought back to the laboratory to separate the leaves, stems and ears. After natural air drying, the moisture content parameter was obtained by subtracting the dry weight from the wet weight.

[0047] Soil parameters include soil water content and soil roughness parameters;

[0048] The soil moisture content was measured using a portable TDR350 soil moisture meter produced by Spectrum. A 7.5 cm probe was used to measure the soil moisture content of five points in a "cross" shape at each sampling point. The average soil moisture content of the five points was used as the final measured soil moisture content of the sampling point to reduce the impact of the temporal and spatial heterogeneity of soil moisture content.

[0049] The surface roughness is obtained in two ways:

[0050] Method 1: Use a scale plate with a length of 0.9 m and a height of 0.6 m to obtain the root mean square height s and the related length l.

[0051] Method 2: Use a camera to take photos containing soil surface edge information to obtain surface roughness information.

[0052] The specific method is as follows: Place a 0.9m long and 0.6m wide reference whiteboard vertically in a pre-dug trench, ensuring that the lower edge of the whiteboard is completely within the trench. Set a 10cm×10cm square reference frame in the middle of the upper half of the whiteboard, and use the reference frame instead of uniform scale information as a quantitative scale to reduce the complexity of scale drawing and subsequent scale initial reading. When shooting, use an ordinary digital camera to take a photo of the boundary between the whiteboard and the soil, and place the camera near the center of the whiteboard.

[0053] In other aspects, this embodiment also provides an image fusion process that can perform high-precision crop planting area identification, specifically:

[0054] Step 11: spatially register the optical image with the SAR image and resample the optical image to the same size as the SAR image;

[0055] Step 12: Perform principal component transformation on the optical image to obtain the first principal component and other components; perform histogram matching between the SAR image and the first principal component obtained by PCA transformation;

[0056] Step 13: Replace the first principal component with the SAR image generated by histogram matching, and perform inverse principal component transformation on it together with the remaining components to obtain the fused image.

[0057] As a preferred implementation method of this embodiment, remote sensing images from multi-source sensors are processed at multiple levels, including fusion and combination, with the purpose of improving the remote sensing image interpretation capability, realizing the complementary advantages of multi-source remote sensing images, reducing redundancy, and making maximum use of multi-source remote sensing image information, thereby improving the accuracy of ground object interpretation.

[0058] In summary, the present invention utilizes the SAR remote sensing images of Sentinel-1 and the optical remote sensing images of Sentinel-2, and performs image fusion through the PCA algorithm, thereby improving the richness of the spatial and spectral information of the images, capturing the surface information more comprehensively, and thus improving the accuracy of soil moisture estimation. The crop planting area recognition model constructed by the convolutional neural network algorithm realizes accurate crop coverage area recognition, which not only improves the accuracy of crop area recognition, but also provides a reliable data basis for subsequent soil moisture estimation.

[0059] The MIM ICS microwave scattering model was introduced, and the growth parameters and soil parameters of crops in different growth periods were combined to construct a microwave scattering model for crops with multiple growth periods. The dynamic adaptability and accuracy of soil moisture estimation were improved by analyzing the impact of crop growth period changes on microwave scattering. The contribution of crop volume scattering and secondary scattering between crops and the ground in total backscattering was eliminated through the microwave scattering model for crops with multiple growth periods, which further improved the accuracy of soil moisture estimation. The relationship model between soil moisture content and soil backscattering coefficient under different polarization modes was established using a machine learning algorithm, and the model was verified using ground measured data to ensure the reliability of the model.

[0060] The soil moisture estimation model is applied to the crop coverage area extracted by the crop planting area identification model, and high spatiotemporal resolution inversion of soil moisture content of crop-covered surfaces in different growth periods is performed. High spatiotemporal resolution mapping of soil moisture distribution is output, thereby providing more detailed and real-time soil moisture information, which is convenient for crop growth monitoring and water resources management. The potential of Sentinel-1 and Sentinel-2 image data in the soil moisture estimation model is analyzed and evaluated, providing a scientific basis for further optimization and improvement of the model.

[0061] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A soil moisture estimation method integrating multi-source time series remote sensing data, characterized in that: The following steps are involved: Step 1: Acquire the SAR remote sensing image of Sentinel-1 and the optical remote sensing image of Sentinel-2 of the specified area, fuse the images using the PCA algorithm, extract the feature set of the fused images, generate the minimum analysis unit through image segmentation, extract its geometric shape features, build a crop planting area recognition model through the convolutional neural network algorithm, and obtain the crop coverage area; Step 2: Introduce the MIMICS microwave scattering model, obtain the growth parameters and soil parameters of crops in different growth stages, define the model components and basic parameters of model operation, and build a microwave scattering model for crops in multiple growth stages; Step 3: Based on the total backscatter coefficient provided by Sentinel-1 and the crop moisture content provided by Sentinel-2, the contribution of crop volume scattering and secondary scattering between crops and the ground in the total backscattering is eliminated through the microwave scattering model of crops in multiple growth periods. A soil moisture estimation model between soil moisture content and soil backscatter coefficient under different polarization modes is established using a machine learning algorithm, and the model is verified based on the measured soil and crop parameters on the ground. Step 4: Apply the soil moisture estimation model to the crop coverage area extracted by the crop planting area identification model, perform high-temporal and spatial resolution inversion of soil moisture content on the surface covered by crops at different growth stages, and output high-temporal and spatial resolution mapping of soil moisture distribution; Step 5: Output of analysis and evaluation results on the potential of Sentinel-1 and Sentinel-2 image data in soil moisture estimation model.

2. The soil moisture estimation method based on multi-source time series remote sensing data according to claim 1 is characterized in that: The image fusion process in step 1 is specifically as follows: Step 11: spatially register the optical image with the SAR image and resample the optical image to the same size as the SAR image; Step 12: Perform principal component transformation on the optical image to obtain the first principal component and other components; perform histogram matching between the SAR image and the first principal component obtained by PCA transformation; Step 13: Replace the first principal component with the SAR image generated by histogram matching, and perform inverse principal component transformation on it together with the remaining components to obtain the fused image.

3. The soil moisture estimation method based on multi-source time series remote sensing data according to claim 1 is characterized in that: The crop growth parameters in step 2 include: leaf parameters, specifically leaf moisture content, leaf cover, leaf length, leaf width, leaf thickness and leaf density, stem parameters, specifically stem moisture content and stem height, ear parameters, specifically ear moisture content, ear cover, ear length, ear diameter and ear density, and the soil parameters include: soil moisture content and soil roughness parameters.

4. The soil moisture estimation method based on multi-source time series remote sensing data according to claim 1 is characterized in that: The construction process of the crop multi-growth period microwave scattering model in step 2 is divided into four growth stages, including: sowing period, sowing-jointing period, jointing-grain filling period and grain filling-maturity period.

5. The soil moisture estimation method based on multi-source time series remote sensing data according to claim 4 is characterized in that: The surface soil layer of the sowing period is constructed, and the construction contents of the microwave scattering models of the sowing-jointing period, the jointing-grain filling period and the grain filling-maturity period are divided into the underlying soil layer and the vegetation layer.

6. The soil moisture estimation method based on multi-source time series remote sensing data according to claim 5 is characterized in that: The soil layer modeling is performed by random rough surface modeling through the AIEM model, and the vegetation layer is modeled through vector radiation transmission.

7. The soil moisture estimation method based on multi-source time series remote sensing data according to claim 1 is characterized in that: In the process of verifying the model according to the ground-measured soil and crop parameters in step 3, the estimated soil moisture content is compared with the ground-measured data, error indicators such as root mean square error and determination coefficient are calculated, the sources of uncertainty in the model are analyzed, the performance of the model at different time and space scales is evaluated, and the influence coefficient of the time resolution and spatial resolution of the data is analyzed.

8. The soil moisture estimation method based on fusion of multi-source time series remote sensing data according to claim 1 is characterized in that: During the analysis and evaluation process in step 5, the estimated soil moisture content is compared with the ground measured data, the impact of seasonal changes on the accuracy of soil moisture estimation, the impact of different crop types and growth periods on microwave scattering characteristics, and the sensitivity of soil moisture estimation are analyzed, and the convenience and cost of data acquisition are evaluated.

Images