Estimation method for earth surface biomass of typical grassland in temperate zone

Abstract

The invention relates to the field of unmanned aerial vehicle multispectral remote sensing image analysis and inversion model construction based on a machine learning algorithm, in particular to a temperate zone typical grassland surface biomass estimation method. A multi-rotor unmanned aerial vehicle is used for carrying a multi-spectral sensor, multi-band spectral reflection information of a pasture is obtained from a near low altitude, a physical spectral reflection model is constructed to calculate real and accurate reflectivity data of a same-band spectrum, vegetation coverage, soil attributes and pasture canopy information of a sampling area are recorded, and the earth surface biomass is measured by cutting a sample; according to the method, a plurality of vegetation indexes and a 3D space model are combined with a multi-modal input structure, data enhancement and small sample learning strategies are adopted, a biomass inversion model is constructed through an MANBA machine learning framework, the precision of grassland biomass estimation is improved, and effective technical support is provided for grassland ecological monitoring and management.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle multispectral remote sensing image analysis and inversion model construction based on machine learning algorithms, and in particular to a method for estimating the surface biomass of a typical temperate grassland. Background Art

[0002] Typical grasslands play a key role in climate regulation, water source protection and ecological balance maintenance, and have advantages in producing high-quality, pollution-free and low-cost herbivorous livestock and poultry products. However, grassland degradation, desertification and soil erosion threaten the ecological balance. Therefore, it is crucial to quickly survey grassland resources, achieve rational allocation of grassland resources, and then improve grassland management and ecological restoration capabilities.

[0003] The inventors found that although satellite remote sensing technology is widely used in grassland resource monitoring, it is difficult to provide detailed surface information due to resolution limitations, which affects the accurate assessment of grassland conditions. At the same time, although the traditional manual sampling method is accurate, it is time-consuming and labor-intensive, and it is difficult to cover large areas of grassland, resulting in incomplete data. In addition, the fitting model of a single spectral band is often inaccurate in reflecting the actual condition of the grassland and cannot capture complex ecological changes. These shortcomings limit the resource survey and management of grassland ecosystems. Summary of the invention

[0004] In order to address the shortcomings of the existing technology, the present invention proposes a multi-parameter fitting model based on UAV multispectral quantitative remote sensing, which combines UAV high-resolution multispectral data and advanced machine learning algorithms, and combines the strategy of small sample learning to improve the accuracy and efficiency of grassland biomass estimation, providing a new technical means for the rational allocation of grassland resources.

[0005] In order to achieve the above object, the present invention adopts the following technical solution: The first aspect of the present invention provides a method for estimating the surface biomass of a typical temperate grassland: Use a multi-rotor drone with a multispectral sensor to take aerial photos for mapping, obtain single-band spectral reflectance information of blue, green, red, red edge, and near infrared (Nir) in the pasture to be studied, use QGIS Desktop and ENVI 5.6 professional remote sensing software to process and analyze spectral images of multiple bands, extract remote sensing information of pasture grassland, construct 11 common vegetation index data, and use weighted fusion to obtain a leaf area index (LAI) with high accuracy; Manual sampling areas were selected in the pasture, and areas with balanced grass growth and rich species were selected as sampling areas as much as possible. The best sampling range was selected in such areas using a 1m*1m sample box, and the RTK coordinates of the center point of the sample box were recorded. The total percentage of grassland coverage, the individual percentage of various vegetation, the height of the vegetation canopy, the chlorophyll content of the canopy leaves, the distribution of leaf inclination, the regional soil firmness, and the soil moisture content information were recorded; All the grass in the sample box was cut evenly with scissors, and the fresh grass was put into a kraft paper bag for fresh weight weighing, and then dried in a DGF electric hot air drying oven at 75℃ for 48h until constant weight was measured, and the dry weight was weighed, and the difference between the two weights was recorded as the aboveground biomass (AGB); According to the RTK position information recorded in the sampling area, the corresponding mowing sampling area is found from the remote sensing image, and the leaf area index (LAI) corresponding to the 1m*1m range in the image is extracted; The relationship between independent variables and dependent variables was set for the extracted parameter information and the measured surface biomass data, and then imported into the machine learning framework of the MANBA (Multi-Agent Nonlinear Bayesian Algorithm) model. At the same time, the multimodal input structure of the model was used to import the RGB image information of the sample box and the 3D spatial model data of forage grass, and the small sample learning method was integrated to construct a typical grassland surface biomass inversion model.

[0006] As a further limitation of the first aspect of the present invention, when collecting and processing remote sensing images, the following details are supplemented: First, Metashape is used to align and align aerial photos for mapping, correct spectral parameters, and use ground control points to correct the distortion of the stitched images and improve the image processing accuracy. QGIS Desktop and ENVI 5.6 were used to calculate and extract image data of 11 vegetation indices in all areas of the pasture, including the Normalized Difference Vegetation Index (NDVI), Optimized Soil Adjusted Vegetation Index (OSAVI), Enhanced Vegetation Index (EVI), Red Edge Chlorophyll Index (RECI), Difference Vegetation Index (DVI), Modified Soil Adjusted Vegetation Index (MSAVI), Green Normalized Difference Vegetation Index (GNDVI), Green Chlorophyll Index (GCI), Normalized Flavonoid Index (NFBDI), Ratio Vegetation Index (RVI), and Restricted Difference Vegetation Index (RDVI). The 11 vegetation indices were fused and analyzed, and the leaf area index (LAI) with higher accuracy was obtained through a multi-weighted method, while reducing the impact of soil background and atmosphere on spectral reflectance. As a further limitation of the first aspect of the present invention, when collecting grassland ground data, the following details are supplemented: When selecting sampling areas, areas with balanced grass growth and rich species should be used as the main sampling areas, and areas with sparse and dense grass should also be sampled so that the overall data presents a normal distribution; There are many types of vegetation in typical temperate grasslands. When recording the individual proportions of various vegetation in the sample box, only the proportions of the following 19 species were recorded: Stipa, Artemisia selengensis, Artemisia frigida, Caragana microphylla, Leymus chinensis, Artemisia selengensis-long flower, Potentilla bivariatea, Crane striata, Chenopodium album (green Chenopodium album), Saposhnikovia divaricata, Donkey's ear Saussurea, Cleistogenes squarrosa (Cleistogenes), Artemisia annua, Altai dogwa flower, Dauri Xinba, High two-lobed, Leontopodium lanceolata, and Huanghua lanceolata. When recording regional soil firmness and soil moisture information, data should be collected at the four corners and the middle of the sample box, and the probes of the corresponding equipment should be inserted about 20 cm underground to obtain data; Take stereoscopic photos of various vegetation canopies in the sampling area, and then calculate the vegetation canopy height using stereo vision technology; The damaged biological samples of 19 representative vegetation species in the study area were collected and packed into sealed bags. They were transported to the laboratory in a low-temperature environment. The relative chlorophyll content was extracted using a SPAD instrument and a spectrometer to obtain the chlorophyll content parameters of the canopy leaves of the 19 typical vegetation species. The LAI-2000 optical sensor was used to measure the intensity of diffuse light passing through the vegetation canopy in the sample box, and the leaf area index (LAI) was further calibrated and corrected based on this. The same instrument was used to measure the distribution information of leaf inclination angles in the vegetation canopy. As a further limitation of the first aspect of the present invention, when weighing the cut and dried grass, the following details are included: When mowing the grass flush with the ground, the grass should be cut strictly within the sample box range and the stubble height on the ground should not exceed 2 cm. Newly cut hay should be weighed within 1 hour to avoid evaporation of water in the kraft paper bag, which may lead to a decrease in weighing accuracy. Similarly, dry hay taken out of the drying box should be weighed within 30 minutes to avoid prolonged exposure to the air, which may cause it to absorb moisture again and result in a higher dry weight measurement result.

[0007] As a further limitation of the first aspect of the present invention, when the remote sensing image is deeply analyzed and aligned with the ground data, the following details are supplemented: When searching for the corresponding mowing area in the remote sensing image, all coordinate point information was first normalized and imported into the QGIS Desktop software by adding a delimited text layer function to determine the sample location, and the software's shape-based digital tool was used to select the range of the mowing sample, thereby extracting information related to the 11 vegetation indices in the region and generating the leaf area index (LAI); As a further limitation of the first aspect of the present invention, when constructing a biomass inversion model based on machine learning, the following details are supplemented: The leaf area index (LAI), leaf chlorophyll content, vegetation canopy height, and leaf inclination information fused from multiple vegetation indices were used as input parameters, and a radiation transmission model was constructed using the physical model method to simulate the multiple scattering and reflection routes and energy loss between leaves, stems, and the ground after light enters the vegetation canopy, and then calculate the reflection status information of light of different wavelengths; The PROSAIL model is used to analyze the reflection information of different wavelengths of light and the geometric structure of the canopy (SAIL), and finally the real and accurate reflectance data of different spectrum bands are obtained from the perspective of physical principles; The model input parameters include the real and accurate reflectance data of different spectral bands, the total grassland coverage in the sample frame, the proportion of 19 plant species and their canopy leaf chlorophyll content parameters, the soil firmness and soil moisture content information data of the sampling area; When using the MANBA framework to construct the surface biomass inversion model, the input data of the model is first preprocessed, including noise removal, standardization and normalization, to ensure the consistency and reliability of the data and improve the prediction accuracy of the model; All preprocessed input parameters were integrated into a multimodal dataset, and the input data were analyzed using a tree-based model and LASSO regression feature selection technique to identify the features most associated with aboveground biomass; When building the model, if the sample size is small, data enhancement techniques such as rotation, translation, and affine transformation can be used to derive relevant image training samples. Using a small sample learning framework, Siamese Networks can be used to calculate the similarity between samples for classification. Then, Generative Adversarial Networks (GANs) can be used to generate new samples to expand the data set. By fine-tuning existing key parameters, the model can be quickly adapted to new tasks, so that a small number of iterative updates are performed on small samples to improve performance on new samples. This method can effectively perform fast model training iterations when samples are scarce, thereby improving model performance. Under the MANBA model framework, the RGB image information of the sample plot will be used to extract edge and texture information through multi-layer convolution and pooling operations. The text information will be processed into an embedded vector through natural language processing to represent semantic information. Then, the image and text features will be fused into a unified feature space through a multi-head attention mechanism. The strategy of the multi-agent system is used to simulate the impact of different input features on biomass estimation. Each agent is responsible for analyzing a part of the feature vector in the feature space and optimizing the overall model through information sharing and collaboration. During the training process, the Monte Carlo cross-validation technique is used to ensure the stability and generalization ability of the model under different data partitions. The mean square error (MSE) and cross entropy are selected as loss functions for model training, and the RMSprop adaptive optimization algorithm is used to dynamically adjust the learning rate and optimization process to accelerate convergence. After the model training is completed, the validation set is used to evaluate the model performance and calculate the root mean square error (RMSE): evaluates the gap between the model prediction value and the actual value, the coefficient of determination (R²): reflects the model's ability to explain the variation in real data, precision, recall and F1-score: for classification tasks, evaluate the performance of the model in different categories, perform visual analysis on the model, draw scatter plots and residual plots of the predicted values ​​and actual values, and intuitively understand the prediction effect of the model. After confirming that the model performance meets expectations, it is applied to the actual grassland biomass estimation task, using multispectral remote sensing data collected by drones for real-time monitoring to obtain timely information on biomass changes.

[0008] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention combines the spectral information of multiple bands including blue, green, red, red edge and near infrared, the ground vegetation coverage information and constructs a physical model to simulate the light scattering and reflection conditions to generate true and accurate reflectance data of spectra in different bands, and integrates them into the construction of a typical temperate grassland surface biomass estimation model, which can greatly improve the ability of accurate measurement of grassland biomass and spatial distribution analysis.

[0009] 2. Through professional software such as QGIS Desktop and ENVI 5.6, a variety of vegetation indices are extracted from remote sensing images. At the same time, detailed data of ground sampling areas, including soil firmness, moisture content, vegetation type, canopy height, canopy leaf chlorophyll content, and leaf inclination distribution information, are combined to form a multi-dimensional and multi-level data integration and analysis system.

[0010] 3. The MANBA (Multi-Agent Nonlinear Bayesian Algorithm) model was introduced, and a precise biomass inversion model was constructed by using a multi-modal input structure and integrating multi-source data. The adaptability and prediction accuracy of the model were greatly improved through small sample learning and multi-agent system strategies. In the case of scarce samples, data enhancement technology and a small sample learning framework were used to expand the data set through rotation, translation, generative adversarial networks, etc., and the stability and generalization ability of the model were guaranteed through Monte Carlo cross-validation and adaptive optimization algorithms.

[0011] 4. The present invention not only focuses on the accurate estimation of biomass, but also forms a complete process system through refined data collection, in-depth analysis, and model construction. The comprehensive management method ensures the accurate and rapid investigation of grassland ecosystem resources.

[0012] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0014] Figure 1 The present invention provides a technical roadmap for the method for estimating the surface biomass of typical temperate grasslands based on unmanned aerial vehicle multi-spectral quantitative remote sensing. DETAILED DESCRIPTION

[0015] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0016] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0017] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0018] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0019] The technical solution of the embodiment of the present application will be described below in conjunction with the drawings in the embodiment of the present application.

[0020] A method for estimating the surface biomass of a typical temperate grassland comprises the following steps: S10. Use a multi-rotor drone with a multi-spectral sensor to take aerial photos for mapping, obtain the single-band spectral reflectance information of blue, green, red, red edge, and near infrared (Nir) in the pasture to be studied, use QGIS Desktop and ENVI 5.6 professional remote sensing software to process and analyze spectral images of multiple bands, extract remote sensing information of pasture grassland, construct 11 common vegetation index data, and use weighted fusion to obtain the leaf area index (LAI) with higher accuracy; S20. Manually select sampling areas in the pasture, try to select areas with balanced grass growth and rich species as sampling areas, and use a 1m*1m sample box to select the best sampling range in such areas, and record the RTK coordinates of the center point of the sample box, the total percentage of grassland coverage in the sample box, the individual percentage of various vegetation, the height of the vegetation canopy, the chlorophyll content of the canopy leaves, the distribution of leaf inclination, the regional soil firmness, and the soil moisture content information; S30. Cut all the grass in the sample box evenly with scissors, put the fresh grass into a kraft paper bag for fresh weight weighing, and then use a DGF electric heating blast drying oven to dry at 75°C for 48 hours until constant weight is obtained, then weigh the dry weight, and record the difference between the two weights as the aboveground biomass (AGB); S40, finding the corresponding mowing sampling area from the remote sensing image according to the RTK position information recorded in the sampling area, and extracting the leaf area index (LAI) corresponding to the range of 1m*1m in the image; S50. Set the relationship between independent variables and dependent variables for the extracted parameter information and the measured surface biomass data, import them into the machine learning framework of the MANBA (Multi-Agent Nonlinear Bayesian Algorithm) model, and use the multimodal input structure of the model to import the sample box RGB image information and the 3D spatial model data of the forage grass together, and jointly integrate the small sample learning method to construct a typical grassland surface biomass inversion model.

[0021] The step S10 also includes: S11. First, Metashape is used to align and align the aerial photos for mapping, correct the spectral parameters, and use ground control points to correct the distortion of the stitched images and improve the image processing accuracy. S12. Use QGIS Desktop and ENVI 5.6 to calculate and extract image data of 11 vegetation indices in all areas of the pasture, including normalized difference vegetation index (NDVI), optimized soil adjusted vegetation index (OSAVI), enhanced vegetation index (EVI), red edge chlorophyll index (RECI), difference vegetation index (DVI), modified soil adjusted vegetation index (MSAVI), green normalized difference vegetation index (GNDVI), green chlorophyll index (GCI), normalized flavonoid index (NFBDI), ratio vegetation index (RVI), and restricted difference vegetation index (RDVI); S13. The 11 vegetation indices obtained were fused and analyzed, and a leaf area index (LAI) with higher accuracy was obtained through a multi-weighted approach, while reducing the impact of soil background and atmosphere on spectral reflectance.

[0022] The step S20 also includes: S21. When selecting sampling areas, areas with balanced grass growth and rich species should be used as the main sampling areas. At the same time, areas with sparse and dense grass should also be sampled so that the overall data presents a normal distribution. S22. There are many types of vegetation in typical temperate grasslands. When recording the individual proportions of various vegetation in the sample box, only the proportions of the following 19 species are recorded: Stipa, scutellaria, Artemisia frigida, Caragana microphylla, Leymus chinensis, scutellaria-long flower, Potentilla bivariatea, Crane scutellaria, Chenopodium album (green quinoa), Saposhnikovia divaricata, Donkey's ear Saussurea, Cleistogenes scutellariae (Cleistogenes), Artemisia annua, Altai dogwa flower, Dauri Xinba, High two-lobed, Leontopodium lanceolata, and Huanghua lanceolate; S23. When recording regional soil firmness and soil moisture information, data should be collected at the four corners and the middle of the sample box, and the probes of the corresponding equipment should be inserted about 20 cm underground to obtain data; S24, taking stereoscopic photos of various vegetation canopies in the sampling area, and then calculating the vegetation canopy height by stereo vision technology; S25. Collect damaged biological samples from 19 representative vegetation species in the study area, put them into sealed bags, and transport them to the laboratory in a low-temperature environment. Use a SPAD instrument and a spectrometer to extract the relative chlorophyll content and obtain the chlorophyll content parameters of the canopy leaves of the 19 typical vegetation species. S26. Use the LAI-2000 optical sensor to measure the diffuse light intensity passing through the vegetation canopy in the sample box, and use it to further calibrate the leaf area index (LAI). The instrument is also used to measure the inclination angle distribution information of the leaves in the vegetation canopy.

[0023] The step S30 also includes: S31. When mowing the grass flush with the ground, the grass should be mowed strictly within the scope of the sample box. And the stubble height on the ground should be ensured not to exceed 2cm; S32. Newly cut grass should be weighed within 1 hour to avoid evaporation of water in the kraft paper bag, which may lead to a decrease in weighing accuracy. Similarly, dry grass taken out of the drying box should be weighed within 30 minutes to avoid prolonged exposure to the air, which may cause it to absorb moisture again and result in a higher dry weight measurement result.

[0024] The step S40 also includes: S41. When searching for the corresponding mowing area in the remote sensing image, first normalize all coordinate point information, import it into QGIS Desktop software by adding a delimited text layer function, determine the sample plot location, and use the software's shape-based digital tools to select the mowing sample plot range, thereby extracting information related to the 11 vegetation indices in the region and generating the leaf area index (LAI).

[0025] The step S50 also includes: S51. Taking the leaf area index (LAI), leaf chlorophyll content, vegetation canopy height, and leaf inclination information fused from multiple vegetation indices as input parameters, a radiation transmission model is constructed using the physical model method to simulate the multiple scattering and reflection routes and energy loss between leaves, stems, and the ground surface after light enters the vegetation canopy, and then calculate the reflection status information of light of different wavelengths; S52. Use the PROSAIL model to analyze the reflection information of different wavelengths of light and the geometric structure of the canopy (SAIL), and finally obtain the true and accurate reflectance data of different bands of spectrum from the perspective of physical principles; S53. The model input parameters include the real and accurate reflectance data of different spectral bands, the total grassland coverage in the sample frame, the proportion of 19 plant species and their canopy leaf chlorophyll content parameters, the soil firmness and soil moisture content information data of the sampling area; S54. When using the MANBA framework to construct the surface biomass inversion model, the input data of the model should be preprocessed first, including denoising, standardization and normalization, to ensure the consistency and reliability of the data and improve the prediction accuracy of the model; S55. All preprocessed input parameters were integrated into a multimodal dataset, and the input data were analyzed using a tree-based model and LASSO regression feature selection technique to identify the features most relevant to surface biomass; S56. When building a model, if the sample size is small, data enhancement techniques such as rotation, translation, and affine transformation can be used to derive relevant image training samples. Using a small sample learning framework, Siamese Networks can be used to calculate the similarity between samples for classification. Then, Generative Adversarial Networks (GANs) can be used to generate new samples to expand the data set. By fine-tuning existing key parameters, the model can be quickly adapted to new tasks, so that a small number of iterative updates are performed on small samples to improve performance on new samples. This method can effectively perform fast model training iterations when samples are scarce, thereby improving model performance. S57. Under the MANBA model framework, the RGB image information of the sample plot will be extracted through multi-layer convolution and pooling operations to extract edge and texture information features, and the text information will be processed into an embedded vector through natural language processing to represent semantic information. Then, the image and text features will be fused into a unified feature space through a multi-head attention mechanism. The strategy of the multi-agent system is used to simulate the impact of different input features on biomass estimation. Each agent is responsible for analyzing a part of the feature vector in the feature space and optimizing the overall model through information sharing and collaboration. S58. During the training process, the Monte Carlo cross-validation technique is used to ensure the stability and generalization ability of the model under different data partitions. The mean square error (MSE) and cross entropy are selected as the loss function for model training, and the RMSprop adaptive optimization algorithm is used to dynamically adjust the learning rate and optimization process to accelerate convergence; S59. After the model training is completed, the validation set is used to evaluate the model performance. The root mean square error (RMSE) is calculated to evaluate the gap between the model prediction value and the actual value. The coefficient of determination (R²) reflects the model's ability to explain the variation in real data. Precision, recall and F1-score: for classification tasks, the performance of the model in different categories is evaluated. The model is visualized and analyzed. Scatter plots and residual plots of the predicted values ​​and actual values ​​are drawn to intuitively understand the prediction effect of the model. After confirming that the model performance meets expectations, it is applied to the actual grassland biomass estimation task. Multispectral remote sensing data collected by drones are used for real-time monitoring to obtain timely information on biomass changes.

Claims

1. A method for estimating the surface biomass of typical temperate grasslands, characterized in that: The following steps are involved: S10. Use a multi-rotor drone with a multispectral sensor to take aerial photos for mapping, obtain single-band spectral reflectance information of blue, green, red, red edge, and near infrared in the near-low altitude pasture to be studied, use QGIS Desktop and ENVI 5.6 professional remote sensing software to process and analyze spectral images of multiple bands, extract remote sensing information of pasture grassland, construct 11 common vegetation index data, and use weighted fusion to obtain leaf area index; S20. Manually select sampling areas in the pasture, select areas with balanced grass growth and rich species as sampling areas, and use a 1m*1m sample box to select the best sampling range in the sampling area, and record the RTK coordinates of the center point of the sample box, the total percentage of grassland coverage in the sample box, the individual percentage of various vegetation, the height of the vegetation canopy, the chlorophyll content of the canopy leaves, the distribution of leaf inclination, the regional soil firmness, and the soil moisture content information; S30, use scissors to cut all the grass in the sample box evenly, put the fresh grass into a kraft paper bag for fresh weight weighing, and then use a DGF electric heating blast drying oven to dry at 75°C for 48 hours until constant weight is obtained, then weigh the dry weight, and record the difference between the two weights as the surface biomass; S40, finding the corresponding mowing sampling area from the remote sensing image according to the RTK position information recorded in the sampling area, and extracting the leaf area index corresponding to the range of 1m*1m in the image; S50. The relationship between independent variables and dependent variables is set for the extracted parameter information and the measured surface biomass data, and the information is imported into the machine learning framework of the MANBA model. At the same time, the multimodal input structure of the model is used to import the RGB image information of the sample box and the 3D spatial model data of the forage grass, and the small sample learning method is integrated to construct a typical grassland surface biomass inversion model.

2. The biomass estimation method according to claim 1, characterized in that: The step S10 also includes: S11. First, Metashape is used to align and calibrate the spectral parameters of the aerial images for mapping, and ground control points are used to correct the distortion of the stitched images and improve the image processing accuracy. S12. Use QGIS Desktop and ENVI 5.6 to calculate and extract image data of 11 vegetation indices in all areas of the pasture, including normalized difference vegetation index, optimized soil-adjusted vegetation index, enhanced vegetation index, red edge chlorophyll index, difference vegetation index improved soil-adjusted vegetation index, green normalized difference vegetation index, green chlorophyll index, normalized flavonoid index, ratio vegetation index, and restricted difference vegetation index; S13. The 11 vegetation indices obtained are fused and analyzed, and a leaf area index with higher accuracy is obtained through a multi-weighted approach, while reducing the impact of soil background and atmosphere on spectral reflectance.

3. The biomass estimation method according to claim 2, characterized in that: The step S20 also includes: S21. When selecting sampling areas, areas with balanced grass growth and rich species should be used as the main sampling areas, while areas with sparse and dense grass should also be sampled so that the overall data presents a normal distribution; S22. There are many types of vegetation in the typical temperate grassland. When recording the individual proportions of various vegetation in the sample box, only the proportions of the following 19 species were recorded: Stipa, scutellaria, Artemisia frigida, Caragana microphylla, Leymus chinensis, scutellaria-long flower, Potentilla bivariate, Crane mites, Chenopodium album, Saposhnikovia divaricata, Cleistogenes squarrosa, Artemisia annua, Altai dogwa flower, Dawuli Xinba, High two-lobed, Leontopodium lanceolata, and Huanghua lanceolata. S23. When recording regional soil firmness and soil moisture information, data should be collected at the four corners and the middle of the sample box, and the probes of the corresponding equipment should be inserted about 20 cm underground to obtain data; S24, taking stereoscopic photos of various vegetation canopies in the sampling area, and then calculating the vegetation canopy height by stereo vision technology; S25. Collect damaged biological samples from 19 representative vegetation species in the study area, put them into sealed bags, and transport them to the laboratory in a low-temperature environment. Use a SPAD instrument and a spectrometer to extract the relative chlorophyll content and obtain the chlorophyll content parameters of the canopy leaves of the 19 typical vegetation species. S26. Use the LAI-2000 optical sensor to measure the diffuse light intensity passing through the vegetation canopy in the sample box, and use it to further calibrate the leaf area index, while measuring the inclination distribution information of the leaves in the vegetation canopy.

4. The biomass estimation method according to claim 3, characterized in that: The step S30 also includes: S31. When mowing the grass flush with the ground, strictly follow the range of the sample box and ensure that the stubble height on the ground does not exceed 2 cm; S32. Newly cut grass should be weighed within 1 hour to avoid evaporation of water in the kraft paper bag, which may lead to a decrease in weighing accuracy. Similarly, dry grass taken out of the drying box should be weighed within 30 minutes to avoid prolonged exposure to the air, which may cause it to absorb moisture again and result in a higher dry weight measurement result.

5. The biomass estimation method according to claim 4, characterized in that: The step S40 also includes: S41. When searching for the corresponding mowing area in the remote sensing image, first normalize all coordinate point information, import it into QGIS Desktop software by adding a delimited text layer function, determine the sample plot location, and use the software's shape-based digital tool to select the mowing sample plot range, thereby realizing the extraction of information related to the 11 vegetation indices in the region and the generation of the leaf area index.

6. The biomass estimation method according to claim 5, characterized in that: The step S50 also includes: S51. Using the leaf area index, leaf chlorophyll content, vegetation canopy height, and leaf inclination information fused from multiple vegetation indices as input parameters, a radiation transmission model is constructed using a physical model method to simulate the multiple scattering and reflection routes and energy loss of light between leaves, stems, and the ground after the light enters the vegetation canopy, and then calculate the reflection status information of light of different wavelengths; S52. Use the PROSAIL model to analyze the reflection information of different wavelengths of light and the geometric structure of the canopy, and finally obtain the true and accurate reflectance data of different bands of spectrum from the perspective of physical principles; S53. The model input parameters include the real and accurate reflectance data of different spectral bands, the total grassland coverage in the sample frame, the proportion of 19 plant species and their canopy leaf chlorophyll content parameters, the soil firmness and soil moisture content information data of the sampling area; S52. When using the MANBA framework to construct the surface biomass inversion model, the input data of the model should first be preprocessed, including denoising, standardization and normalization, to ensure the consistency and reliability of the data and improve the prediction accuracy of the model; S53, integrating all preprocessed input parameters into a multimodal dataset, analyzing the input data using a tree-based model and LASSO regression feature selection technique to identify the features most relevant to surface biomass; S54. When constructing a model, if the sample size is small, data enhancement techniques such as rotation, translation, and affine transformation can be used to derive relevant image training samples. Using a small sample learning framework, the twin network is used to calculate the similarity between samples for classification, and then a generative adversarial network is used to generate new samples to expand the data set. By fine-tuning existing key parameters, the model can quickly adapt to new tasks, so that a small number of iterative updates are performed on small samples to improve performance on new samples. This method can effectively perform fast model training iterations when samples are scarce, thereby improving model performance. S55. Under the MANBA model framework, the RGB image information of the sample plot will be extracted through multi-layer convolution and pooling operations to extract edge and texture information features, and the text information will be processed into an embedded vector through natural language processing to represent semantic information. Then, the image and text features will be fused into a unified feature space through a multi-head attention mechanism. The strategy of the multi-agent system is used to simulate the impact of different input features on biomass estimation. Each agent is responsible for analyzing a part of the feature vector in the feature space and optimizing the overall model through information sharing and collaboration. S57. During the training process, the Monte Carlo cross-validation technique is used to ensure the stability and generalization ability of the model under different data partitions. The mean square error and cross entropy are selected as the loss function for model training, and the RMSprop adaptive optimization algorithm is used to dynamically adjust the learning rate and optimization process to accelerate convergence. S58. After the model training is completed, the validation set is used to evaluate the model performance, and the root mean square error is calculated to evaluate the gap between the model prediction value and the actual value. The coefficient of determination reflects the model's ability to explain the variation in real data. Precision, recall and F1-score: for classification tasks, the performance of the model in different categories is evaluated. The model is visualized and analyzed, and scatter plots and residual plots of the predicted values ​​and actual values ​​are drawn to intuitively understand the prediction effect of the model. After confirming that the model performance meets expectations, it is applied to the actual grassland biomass estimation task, and multispectral remote sensing data collected by drones are used for real-time monitoring to obtain timely information on biomass changes.

Images