Global winter wheat coverage inversion method and system based on corrected clearance rate model

Abstract

The invention belongs to the technical field of vegetation coverage, discloses a global winter wheat coverage inversion method based on a corrected clearance rate model, and provides PROSPECT-D and 4SAIL model parameters suitable for winter wheat by initializing model input parameters of a mechanism model and optimizing the parameters by utilizing sensitivity analysis. Based on the uniqueness that the average leaf tilt angle distribution of winter wheat is different from other coverage types, the setting of a projection function in a clearance rate model is corrected, a fixed value is changed into a dynamic value, a lookup table for projection function parameter values of winter wheat is constructed, LAI simulated by a PROSAIL model is converted into FVC, a simulation sample set is made, a random forest model is trained, and the winter wheat leaf tilt angle distribution model is constructed. The method is used for carrying out global high-precision winter wheat coverage inversion. According to the method, higher precision can be obtained when inversion is carried out by utilizing machine learning.

Description

Technical Field

[0001] The invention belongs to the technical field of vegetation coverage, and in particular relates to a global winter wheat coverage inversion method and system based on a modified gap rate model. Background Art

[0002] Current status of methods for estimating vegetation cover: There are three main methods for estimating vegetation cover using remote sensing data, including empirical methods, pixel unmixing models, and hybrid inverse methods based on physical models and machine learning.

[0003] The empirical method regresses the vegetation index and vegetation coverage calculated by a certain band reflectance of the remote sensing image, different band combinations, or the reflectance of each band, to construct an empirical model, and uses spatial extrapolation to obtain the vegetation coverage of a larger area. The pixel unmixing model assumes that each pixel is composed of multiple components, and the composition ratio of vegetation is regarded as the vegetation coverage of the pixel. The hybrid inversion method first simulates the physical relationship between vegetation canopy reflectance and vegetation coverage based on the radiation transfer model based on the physical model method. This method has clear physical meaning and is more suitable for coverage estimation under various vegetation conditions in large areas, but the complexity of the physical model makes direct inversion usually difficult. Therefore, it is often indirectly realized with the help of machine learning methods, and the physical model is trained through a pre-calculated reflectance database to simplify the inversion process.

[0004] The coverage inversion method used by the more mature coverage products in the world, such as GEOV3, belongs to a hybrid inversion model. It combines the mechanism model PROSAIL and the neural network model to achieve global coverage inversion. However, this framework is a coverage inversion solution for all global coverage types (including crops, forests, grasslands, wetlands, etc.). It does not focus on solving the high-precision inversion of coverage of a certain coverage type (such as crops), and it is even more impossible to achieve high-precision coverage inversion for a subclass of a certain coverage type (such as winter wheat in the crop coverage type). Therefore, the global winter wheat coverage cannot be accurately estimated at present, and the low-precision winter wheat coverage results greatly limit a series of applications including winter wheat growth monitoring, winter wheat disease and pest monitoring, and winter wheat yield estimation. Therefore, it is urgent to propose a set of high-precision global winter wheat coverage inversion technology.

[0005] In view of the problem that the existing hybrid inversion technology has low accuracy when applied to the inversion of global winter wheat coverage, the present invention considers improving the existing model from two aspects to solve the problem of low accuracy of the existing technology in the inversion of winter wheat coverage:

[0006] (1) Adjustment of the parameters of the mechanism model: The parameter input of the existing model has poor characterization ability for winter wheat. This method systematically surveyed more than 170 scientific research papers worldwide that used the PROSAIL model to invert the physiological and structural parameters of winter wheat, adjusted the initialization reference table of the mechanism model input parameters specifically for winter wheat, and combined with sensitivity analysis to determine the final input parameters of the model, in order to more accurately characterize the interaction process and characteristics between winter wheat and incident radiation.

[0007] (2) Modified gap ratio model: The direct input parameters of the PROSAIL model do not include coverage. The theoretical coverage needs to be obtained through a gap ratio model that simulates the vegetation canopy extinction process and characteristics. When the existing coverage inversion technology uses the gap ratio model to characterize the relationship between leaf area index and coverage, the projection function value is set to a fixed value of 0.5. In fact, the vegetation canopy structure is heterogeneous and has a great influence on the canopy extinction process. The winter wheat canopy structure can be inferred by using the prior knowledge of the average leaf inclination angle ALA for winter wheat, and the projection function and the modified gap ratio model can be adjusted based on this to achieve high-precision inversion of winter wheat coverage. Summary of the invention

[0008] In view of the problems existing in the prior art, the present invention provides a global winter wheat coverage inversion method based on a modified gap rate model.

[0009] The present invention is implemented in this way: a global winter wheat coverage inversion method based on a modified gap rate model comprises:

[0010] S1, parameterized model;

[0011] S2, generate simulation samples;

[0012] S3, generate a lookup table;

[0013] S4, Global winter wheat cover inversion.

[0014] Furthermore, the parameterized model:

[0015] Based on the survey of more than 170 scientific research papers worldwide that used the PROSAIL model to invert winter wheat physiological and structural parameters, we obtained prior knowledge of winter wheat mechanism models, including knowledge of crop growth periods and preliminary ranges of PROSAIL model input parameters;

[0016] Then, a sensitivity analysis was performed on each parameter to explore the impact of each parameter and the interaction between parameters on the model results, and the sensitivity of each parameter to the model and the sensitive band range were determined. For highly sensitive parameters, a smaller step size was set when simulating samples, and multiple situations were simulated. For parameters with low sensitivity, a larger step size was set or it was set to a fixed value based on references; finally, the PROSAIL model input parameter table was set for winter wheat.

[0017] Furthermore, the PROSAIL model includes PROSPECT-D and 4SAIL models.

[0018] Further, the generation of simulation samples:

[0019] The simulated samples are obtained by simulating the above winter wheat input parameter table through the PROSAIL model. The PROSAIL model is based on a flat plate model and takes into account the incident angle of light on the leaf surface and the Lambertian scattering inside the leaf to simulate the optical properties of the leaf in the wavelength range of 400nm to 2500nm. The PROSAIL model used is composed of PROSAIL-D and 4SAIL. The remote sensing image used to invert the FVC of winter wheat is Sentinel-2, and its spectral response function file is needed to resample the simulated samples from 400nm to 2500nm simulated by PROSAIL to the 13 bands of the Sentinel-2 image.

[0020] Further, the lookup table is generated:

[0021] The FVC calculated by PROSAIL is quantified based on the gap probability model, and the formula is defined as follows:

[0022]

[0023] Among them, G(θ) is the projection function, which represents the average projection area when the zenith angle is θ. According to the definition of FVC, FVC is calculated when θ is 0. This method is based on the Campbell leaf inclination distribution function and the possible values ​​of leaf inclination in the winter wheat parameter input table. The ratio of the horizontal and vertical axes of the canopy ellipsoid χ is calculated and brought into the calculation formula of the projection function G(θ). The projection function values ​​of different leaf inclination values ​​when the observed zenith angle is 0 are calculated, and the corresponding conversion formula and projection function table of FVC and LAI are obtained.

[0024] The simulated samples generated in step S2 are converted to construct a new lookup table; the lookup table consists of 13 columns and 48,000 rows. The first row is the column headers, which are B1, B2, B3, B4, B5, B6, B7, B8, B8A, B9, B11, B12, and FVC. The band values ​​of the B1-B12 bands are simulated samples, and the FVC value is calculated based on the gap rate model according to the LAI and ALA of each simulated sample.

[0025] Furthermore, the global winter wheat coverage inversion: the random forest model is trained using the lookup table in S3, the characteristic variables are the B3, B4, B5, B6, B7, B8, B8A, B11, and B12 columns in the lookup table, and the target variable is the FVC column; the trained random forest model can be used for global winter wheat coverage inversion.

[0026] Another object of the present invention is to provide a global winter wheat coverage inversion system based on a modified gap rate model, comprising:

[0027] The parameterization module is used to obtain the prior knowledge of winter wheat mechanism model through parameterization model based on a large amount of literature research, including knowledge of crop growth period and preliminary range of input parameters of PROSAIL model;

[0028] A simulation sample generation module, used for generating simulation samples;

[0029] A lookup table generation module, used for generating a lookup table;

[0030] Inversion module, used for global winter wheat coverage inversion.

[0031] Another object of the present invention is to provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the global winter wheat coverage inversion method based on the modified gap ratio model.

[0032] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to execute the steps of the global winter wheat coverage inversion method based on the modified gap ratio model.

[0033] Another object of the present invention is to provide an information data processing terminal, which is used to implement the global winter wheat coverage inversion system based on the modified gap rate model.

[0034] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0035] First, the present invention first sets reasonable mechanism model input parameters for winter wheat based on prior knowledge of winter wheat, in order to more accurately characterize the interaction process and characteristics between winter wheat and incident radiation; then, the winter wheat canopy structure can be inferred using prior knowledge of average leaf inclination angle ALA for winter wheat, and the projection function and the modified gap rate model can be adjusted based on this. The above two steps generally optimize the PROSAIL simulation data set, making the samples more representative, so that higher accuracy can be obtained when using machine learning to invert winter wheat coverage.

[0036] This technology intends to perform high-precision coverage inversion for global winter wheat based on prior knowledge of winter wheat and a systematic literature review of a large number of historical references. The main optimization technical points are:

[0037] 1. Based on the understanding of the unique physiological parameters and canopy structure of winter wheat, the model input parameters of the mechanism model were initialized using prior knowledge, and each parameter was optimized using sensitivity analysis. The PROSPECT-D and 4SAIL model parameters suitable for winter wheat were proposed.

[0038] 2. Based on the uniqueness of the average leaf inclination distribution of winter wheat compared with other cover types, the setting of the projection function in the gap ratio model was revised, the fixed value was changed to a dynamic value, and a lookup table of the projection function parameter values ​​for winter wheat was constructed. The LAI simulated by the PROSAIL model was converted into FVC, a simulation sample set was made, and a random forest model was trained to perform high-precision winter wheat cover inversion on a global scale.

[0039] Second, the technical solution of the present invention fills the technical gap in the industry at home and abroad: winter wheat is one of the most important food crops in my country and even the world. However, the yield of winter wheat is susceptible to climate and pests and diseases. According to statistics, more than 50% of the wheat planting area in the world is affected by drought every year, and the four winter wheat stripe rust pandemics in my country have caused a 20%-30% reduction in wheat production. Therefore, the use of remote sensing technology to quickly monitor the growth of winter wheat in each growth period is crucial for identifying climate and pest and disease stress, assisting in regulating the intelligent management of winter wheat fields, achieving stable production of winter wheat and even national food security. Winter wheat coverage is closely related to the absorption of photosynthetically active radiation by winter wheat. The size of winter wheat coverage is controlled by key processes such as photosynthesis and transpiration crops during its growth process, and is an important parameter to characterize its growth status. Obtaining high-precision winter wheat coverage information is conducive to better monitoring the growth of winter wheat, guiding water and fertilizer management and pest and disease control, and is very important for predicting winter wheat yields.

[0040] However, there is currently a lack of a global universal coverage inversion model for winter wheat. The model used by the existing global coverage inversion product is a coverage inversion solution for all coverage types (including crops, forests, grasslands, wetlands, etc.). It does not focus on solving the high-precision inversion of coverage of a certain coverage type (such as crops), and it is even more unable to achieve high-precision coverage inversion for a sub-type of a certain coverage type (such as winter wheat in the crop coverage type). Verification results show that its inversion accuracy for winter wheat coverage is low ( Figure 4 b). Therefore, the global winter wheat coverage cannot be accurately estimated at present, which restricts the high-precision real-time dynamic monitoring of winter wheat growth at home and abroad, and further affects the high-precision yield estimation of global winter wheat. The present invention provides a global winter wheat coverage inversion method and system based on a modified gap rate model, which fills the gap in the high-precision inversion of global large-scale winter wheat coverage.

[0041] Third, the present invention solves several technical problems existing in the prior art in the inversion of global winter wheat coverage. Existing methods usually rely on simple statistical models or estimates based on empirical formulas, and lack accurate biophysical models to describe the complex changes in the crop growth process. This leads to insufficient accuracy and adaptability of the inversion results in different regions or different growth stages, affecting the effect of agricultural production monitoring.

[0042] By introducing the modified gap rate model and combining the PROSAIL model to accurately parameterize winter wheat, the present invention can more accurately simulate the growth characteristics of winter wheat and its relationship with spectral data. Through parameter sensitivity analysis, the setting of each parameter can be finely adjusted to improve the model's adaptability to actual conditions, overcoming the limitations of traditional methods that rely heavily on parameters and lack systematic adjustment.

[0043] In addition, the present invention realizes an efficient coverage inversion process by generating simulated samples and constructing a lookup table. Traditional inversion methods often require heavy computing resources and time, while the lookup table technology of the present invention can quickly and accurately invert the coverage of winter wheat after remote sensing data acquisition, greatly improving the inversion efficiency and accuracy, and providing more reliable data support for real-time agricultural monitoring and decision-making.

[0044] In terms of industrial application, the technical progress of this invention is that it can accurately estimate the coverage of winter wheat on a large scale around the world. This not only provides a new technical path for agricultural management, climate change research and crop production prediction, but also effectively supports the development of precision agriculture and promotes the realization of agricultural automation and intelligent management. It has broad practical application prospects and far-reaching industry impact. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1It is a flow chart of a global winter wheat coverage inversion method based on a modified gap rate model provided in an embodiment of the present invention.

[0046] Figure 2 It is a structural block diagram of a global winter wheat coverage inversion system based on a modified gap rate model provided in an embodiment of the present invention.

[0047] Figure 3 It is a technical roadmap provided by the embodiments of the present invention.

[0048] Figure 4 The verification accuracy provided by the embodiments of the present invention is: (a) the verification accuracy of the coverage inverted by this method, and (b) the verification accuracy diagram of the GEOV3 coverage product.

[0049] Figure 5 The embodiment of the present invention provides a method of taking the third sampling result to produce a scatter plot. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0051] like Figure 1 As shown, a global winter wheat coverage inversion method based on a modified gap rate model provided by an embodiment of the present invention comprises the following steps:

[0052] S1, parameterized model;

[0053] S2, generate simulation samples;

[0054] S3, generate a lookup table;

[0055] S4, Global winter wheat cover inversion.

[0056] Detailed working principle of the global winter wheat coverage inversion method based on the modified gap rate model:

[0057] Parameterized model: In the first step, we first obtained prior knowledge related to winter wheat growth through extensive literature research, including its growth period characteristics and key parameters related to crop reflectance spectra. Then, based on this prior knowledge, we constructed a modified gap rate model and modeled different growth stages of crops. In this process, the PROSAIL model was used as the basic model to set the initial range of input parameters, such as chlorophyll content, leaf area index, etc. Through a deep understanding of each parameter and their interrelationships, we provide a basis for the subsequent steps of the model.

[0058] Generate simulation samples: After determining the preliminary parameter range, perform parameter sensitivity analysis. This analysis helps identify which parameters are most sensitive to the model's output. Based on the analysis results, for parameters with high sensitivity, select a smaller step size during the simulation process to perform detailed simulations to ensure accuracy; for parameters with low sensitivity, select a larger step size or even set it to a fixed value. In this process, a sample set containing different parameter combinations is generated through multiple simulations, providing a data basis for the subsequent generation of the lookup table.

[0059] Generate a lookup table: Based on the simulation samples generated in the previous step, a lookup table is constructed. The lookup table maps the model input parameters (such as chlorophyll content, spectral data, etc.) to the output results (i.e., the coverage of winter wheat) one by one. When generating the lookup table, considering the different effects of different parameter combinations on the model results, the design of the lookup table must be able to quickly and accurately map the coverage values ​​under different circumstances. With an efficient lookup table, the estimated value of winter wheat coverage can be quickly obtained in the subsequent steps, thereby improving the inversion accuracy and efficiency.

[0060] Global winter wheat coverage inversion: Finally, the global winter wheat coverage is inverted using the modified gap rate model and lookup table. By analyzing the remote sensing image data, the corresponding spectral features are extracted and compared with the data in the lookup table, and finally the global winter wheat coverage is inverted. This process can provide accurate data support for agricultural monitoring, climate prediction and crop management. By continuously optimizing the model and lookup table, more accurate winter wheat coverage estimation can be achieved in different regions and different growth stages.

[0061] The combination of these steps enables the global winter wheat cover inversion method based on the modified gap rate model to efficiently and accurately reflect the growth conditions of winter wheat worldwide, and has strong adaptability and universality.

[0062] The parameterized model provided by the embodiment of the present invention:

[0063] Based on extensive literature research, we obtained prior knowledge of winter wheat mechanism models, including knowledge of crop growth periods and preliminary ranges of PROSAIL model input parameters;

[0064] Then, a sensitivity analysis was performed on each parameter to explore the impact of each parameter and the interaction between parameters on the model results, and the sensitivity of each parameter to the model and the sensitive band range were determined. For highly sensitive parameters, a smaller step size was set when simulating samples, and multiple situations were simulated. For parameters with low sensitivity, a larger step size was set, and the references set them as fixed values; finally, the PROSAIL model input parameter table was set for winter wheat.

[0065] The PROSAIL model provided by the embodiment of the present invention includes PROSPECT-D and 4SAIL models.

[0066] The generation simulation sample provided by the embodiment of the present invention:

[0067] The simulated samples are obtained by simulating the above winter wheat input parameter table through the PROSAIL model. The PROSAIL model is based on a flat plate model and takes into account the incident angle of light on the leaf surface and the Lambertian scattering inside the leaf to simulate the optical properties of the leaf in the wavelength range of 400nm to 2500nm. The PROSAIL model used is composed of PROSAIL-D and 4SAIL. The remote sensing image used to invert the FVC of winter wheat is Sentinel-2, and its spectral response function file is needed to resample the simulated samples from 400nm to 2500nm simulated by PROSAIL to the 13 bands of the Sentinel-2 image.

[0068] The generation lookup table provided in the embodiment of the present invention is as follows:

[0069] The FVC calculated by PROSAIL is quantified based on the gap probability model, and the formula is defined as follows:

[0070]

[0071] Among them, G(θ) is the projection function, which represents the average projection area when the zenith angle is θ. According to the definition of FVC, FVC is calculated when θ is 0. This method is based on the Campbell leaf inclination distribution function and the possible values ​​of leaf inclination in the winter wheat parameter input table. The ratio of the horizontal and vertical axes of the canopy ellipsoid χ is calculated and brought into the calculation formula of the projection function G(θ). The projection function values ​​of different leaf inclination values ​​when the observed zenith angle is 0 are calculated, and the corresponding conversion formula and projection function table of FVC and LAI are obtained.

[0072] The simulated samples generated in step S2 are converted to construct a new lookup table; the lookup table consists of 13 columns and 48,000 rows. The first row is the column headers, which are B1, B2, B3, B4, B5, B6, B7, B8, B8A, B9, B11, B12, and FVC. The band values ​​of the B1-B12 bands are simulated samples, and the FVC value is calculated based on the gap rate model according to the LAI and ALA of each simulated sample.

[0073] The global winter wheat coverage inversion provided by an embodiment of the present invention: the random forest model is trained using the lookup table in S3, the characteristic variables are the B3, B4, B5, B6, B7, B8, B8A, B11, and B12 columns in the lookup table, the target variable is the FVC column, the number of random trees is set to 100, and the number of seeds is set to 42; the trained random forest model can be used for global winter wheat coverage inversion.

[0074] like Figure 2 As shown, a global winter wheat coverage inversion system based on a modified gap rate model provided by an embodiment of the present invention includes:

[0075] The parameterization module is used to obtain the prior knowledge of winter wheat mechanism model through parameterization model based on a large amount of literature research, including knowledge of crop growth period and preliminary range of input parameters of PROSAIL model;

[0076] A simulation sample generation module, used for generating simulation samples;

[0077] A lookup table generation module, used for generating a lookup table;

[0078] Inversion module, used for global winter wheat coverage inversion.

[0079] Another object of the present invention is to provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the global winter wheat coverage inversion method based on the modified gap ratio model.

[0080] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to execute the steps of the global winter wheat coverage inversion method based on the modified gap ratio model.

[0081] Another object of the present invention is to provide an information data processing terminal, which is used to implement the global winter wheat coverage inversion system based on the modified gap rate model.

[0082] The present invention is specifically implemented:

[0083] Technical route such as Figure 3 As shown, it mainly consists of four parts: crop model parameterization (S1), simulation sample generation (S2), lookup table generation (S3), and wheat FVC product production (S4).

[0084] The following is a detailed description of each part:

[0085] S1 parameterized model: Based on a large number of literature surveys, we obtained prior knowledge of winter wheat mechanism models, including knowledge of crop growth periods and the preliminary range of PROSAIL model input parameters. Then, we conducted sensitivity analysis on each parameter to explore the impact of each parameter and the interaction between parameters on the model results, and determined the sensitivity of each parameter to the model and the sensitive band range. For highly sensitive parameters, we can set a smaller step size when simulating samples to simulate as many situations as possible. For parameters with low sensitivity, we can set a larger step size, or we can set it to a fixed value by referring to the literature. Finally, we set the input parameter table of the PROSAIL model (including PROSPECT-D and 4SAIL models) for winter wheat, as shown in Table 1.

[0086] Table 1 Input parameters of winter wheat PROSAIL model

[0087]

[0088] S2 Generate simulated samples: The simulated samples are obtained by simulating the above winter wheat input parameter table through the PROSAIL model. The PROSAIL model is based on a flat plate model, taking into account the incident angle of light on the leaf surface and the Lambertian scattering inside the leaf, and simulating the optical properties of the leaf in the wavelength range of 400nm to 2500nm. The PROSAIL model used in this method consists of PROSAIL-D and 4SAIL. The remote sensing image used to invert the FVC of winter wheat is Sentinel-2, and its spectral response function file needs to be used to resample the simulated samples from 400nm to 2500nm simulated by PROSAIL to the 13 bands of the Sentinel-2 image.

[0089] S3 generates a lookup table: based on the gap probability model to quantify the FVC calculated by PROSAIL, the formula is defined as follows:

[0090]

[0091] Among them, G(θ) is the projection function, which represents the average projection area when the zenith angle is θ. According to the definition of FVC, FVC is calculated when θ is 0. This method is based on the Campbell leaf inclination distribution function, the possible values ​​of leaf inclination according to the winter wheat parameter input table, calculates the ratio of the horizontal and vertical axes of the canopy ellipsoid χ, and brings it into the calculation formula of the projection function G(θ). The projection function values ​​of different leaf inclination values ​​when the observed zenith angle is 0 are calculated, and the corresponding conversion formulas of FVC and LAI and the projection function table are shown in the following table.

[0092] Table 2 Correspondence table of average leaf inclination angle and projection function

[0093]

[0094] The simulated samples generated in step S2 are converted to construct a new lookup table. The lookup table consists of 13 columns and 48,000 rows. The first row is the column headers, which are B1, B2, B3, B4, B5, B6, B7, B8, B8A, B9, B11, B12, and FVC. The band values ​​of the B1-B12 bands are simulated samples, and the FVC value is calculated based on the gap rate model according to the LAI and ALA of each simulated sample.

[0095] S4: Global winter wheat coverage inversion: Use the lookup table in S3 to train a random forest model, with the feature variables being the B3, B4, B5, B6, B7, B8, B8A, B11, and B12 columns in the lookup table, the target variable being the FVC column, the number of random trees being set to 100, and the number of seeds being set to 42. The trained random forest model can be used for global winter wheat coverage inversion.

[0096] (2) Technical Effect

[0097] The advantages and effects of the present invention compared with the prior art should be given. They should be described in combination with the technical solution and explained in a reasoning way to be reasonable and well-founded. Do not just draw conclusions without analysis.

[0098] 1. Comparison between the coverage product inverted by the present method and the existing GEOV3 coverage product

[0099] The FVC calculated from drone images of 30 plots (400m×400m) sampled in the winter wheat coverage area of ​​Henan Province in 2024 was used as the validation data set. The 10m scale FVC calculated by the present invention was aggregated to the same 300m scale as GEOV3. The two sets of results were verified and compared at the 300×300m scale. The accuracy evaluation index was evaluated using R 2 , RMSE, NRMSE. The results are as follows Figure 4 As shown in Figure 2, the performance of the method used in the present invention is generally better than that of the GEOV3 product in 30 plots. 2 The GEOV3 accuracy shows R 2 The average value of the regression line is 0.295, and the NRMSE is 39.8%.

[0100] 2. The impact of the improvement of the gap ratio model in the present invention on the accuracy results at the 10m scale

[0101] In order to illustrate the feasibility of the modified gap rate model, a comparative experiment was conducted using a set of 48,000 simulated winter wheat samples generated by PROSAIL based on steps S1 and S2. Experiment (a) used the gap rate corrected by S3 and step S4 for inversion. The projection function of the comparative experiment (b) took a fixed value of 0.5, and the lookup table was generated using the fixed gap rate conversion formula and then combined with step S4 for inversion. The 10m-scale FVC calculated from drone images of 30 plots sampled in the winter wheat coverage area of ​​Henan Province in 2024 was used as the validation data set. Stratified (10 layers) random sampling was used in the winter wheat sample library to obtain 150 samples to verify the accuracy results. After ten random samplings under the above rules, the R of the ten groups of samples was calculated. 2 The accuracy difference of the two coverage inversion techniques is compared with the average value of NRMSE. The results show that the method used in the present invention (experiment a) is better than the inversion result without correcting the gap ratio (experiment b) in ten samplings.

[0102] Table 3 Comparative experiment on the sampling verification accuracy of 10m scale

[0103]

[0104] Take the third sampling result to make scattered points Figure 5 As follows, R of Experiment a (the present invention) 2 is 0.842, NRMSE is 16.7%, RMSE is 0.139, and R 2 The inversion accuracy after the gap ratio correction is higher than that without gap ratio correction.

[0105] The present invention is specifically used for the production of global winter wheat coverage products. Based on Sentinel-2 images, it can produce global winter wheat coverage with a resolution of 10m every 10 days, providing important data support for a series of agricultural practices such as winter wheat growth monitoring, pest and disease monitoring, and yield estimation. At present, there is no relevant global coverage product specifically for winter wheat. The present invention can be used for the production of global winter wheat coverage.

[0106] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. It can be understood by a person of ordinary skill in the art that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0107] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A global winter wheat coverage inversion method based on a modified gap rate model, characterized in that: The following steps are involved: S1, parameterized model: prior knowledge related to winter wheat growth was obtained through literature research, including growth period characteristics and key parameters of reflectance spectrum; Based on prior knowledge, a modified gap ratio model was constructed and different growth stages were modeled, using the PROSAIL model as a basis to set the preliminary range of input parameters; S2, generate simulation samples: according to the preliminary parameter range, perform sensitivity analysis, identify sensitive parameters and adjust the step size for detailed simulation, and generate sample sets containing different parameter combinations through multiple simulations; S3, generate a lookup table: construct a lookup table based on the generated simulation samples, and correspond the input parameters to the coverage of winter wheat one by one, so as to quickly obtain the estimated coverage value later; S4, global winter wheat coverage inversion: Through remote sensing image data, spectral features are extracted and compared with the data in the lookup table to invert the global winter wheat coverage.

2. The global winter wheat coverage inversion method based on the modified gap rate model as claimed in claim 1, characterized in that: The parameterized model: Based on extensive literature research, we obtained prior knowledge of winter wheat mechanism models, including knowledge of crop growth periods and preliminary ranges of PROSAIL model input parameters; Then, a sensitivity analysis was performed on each parameter to explore the impact of each parameter and the interaction between parameters on the model results, and the sensitivity of each parameter to the model and the sensitive band range were determined. For highly sensitive parameters, a smaller step size was set when simulating samples, and multiple situations were simulated. For parameters with low sensitivity, a larger step size was set, and the references set them as fixed values; finally, the PROSAIL model input parameter table was set for winter wheat.

3. The global winter wheat coverage inversion method based on the modified gap rate model as claimed in claim 2, characterized in that: The PROSAIL model includes PROSPECT-D and 4SAIL models.

4. The global winter wheat coverage inversion method based on the modified gap rate model as claimed in claim 1, characterized in that: The generated simulation sample: The simulated samples are obtained by simulating the above winter wheat input parameter table through the PROSAIL model. The PROSAIL model is based on a flat plate model and takes into account the incident angle of light on the leaf surface and the Lambertian scattering inside the leaf to simulate the optical properties of the leaf in the wavelength range of 400nm to 2500nm. The PROSAIL model used is composed of PROSAIL-D and 4SAIL. The remote sensing image used to invert the FVC of winter wheat is Sentinel-2, and its spectral response function file is needed to resample the simulated samples from 400nm to 2500nm simulated by PROSAIL to the 13 bands of the Sentinel-2 image.

5. The global winter wheat coverage inversion method based on the modified gap rate model as claimed in claim 1, characterized in that: The generated lookup table: The FVC calculated by PROSAIL is quantified based on the gap probability model, and the formula is defined as follows: Among them, G(θ) is the projection function, which represents the average projection area when the zenith angle is θ. According to the definition of FVC, FVC is calculated when θ is 0. This method is based on the Campbell leaf inclination distribution function and the possible values ​​of leaf inclination in the winter wheat parameter input table. The ratio of the horizontal and vertical axes of the canopy ellipsoid χ is calculated and brought into the calculation formula of the projection function G(θ). The projection function values ​​of different leaf inclination values ​​when the observed zenith angle is 0 are calculated, and the corresponding conversion formula and projection function table of FVC and LAI are obtained. The simulated samples generated in step S2 are converted to construct a new lookup table; the lookup table consists of 13 columns and 48,000 rows. The first row is the column headers, which are B1, B2, B3, B4, B5, B6, B7, B8, B8A, B9, B11, B12, and FVC. The band values ​​of the B1-B12 bands are simulated samples, and the FVC value is calculated based on the gap rate model according to the LAI and ALA of each simulated sample.

6. The global winter wheat coverage inversion method based on the modified gap rate model according to claim 1, characterized in that: The global winter wheat coverage inversion: use the lookup table in S3 to train the random forest model, the characteristic variables are the B3, B4, B5, B6, B7, B8, B8A, B11, and B12 columns in the lookup table, and the target variable is the FVC column; the trained random forest model can be used for global winter wheat coverage inversion.

7. A global winter wheat coverage inversion system based on a modified gap rate model that implements the global winter wheat coverage inversion method based on a modified gap rate model as described in any one of claims 1 to 6, characterized in that: The global winter wheat coverage inversion system based on the modified gap rate model includes: The parameterization module is used to obtain the prior knowledge of winter wheat mechanism model through parameterization model based on a large amount of literature research, including knowledge of crop growth period and preliminary range of input parameters of PROSAIL model; A simulation sample generation module, used for generating simulation samples; A lookup table generation module, used for generating a lookup table; Inversion module, used for global winter wheat coverage inversion.

8. A computer device, characterized in that: The computer device includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the global winter wheat coverage inversion method based on the modified gap ratio model as described in any one of claims 1-6.

9. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the global winter wheat coverage inversion method based on a modified gap ratio model as described in any one of claims 1 to 6.

10. An information data processing terminal, characterized in that: The information data processing terminal is used to implement the global winter wheat coverage inversion system based on the modified gap rate model as described in claim 7.

Images