Method and system for physical inversion of grassland aboveground biomass
By determining actual vegetation parameters through lookup tables and correcting remote sensing images, constructing a cost function and iteratively solving to optimize the stem-to-leaf ratio, the problem of low accuracy in the spatial distribution of aboveground biomass in existing technologies is solved, and high-precision estimation and management under different grassland types are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- AEROSPACE INFORMATION RES INST CAS
- Filing Date
- 2025-07-16
- Publication Date
- 2026-06-09
Smart Images
Figure CN120976736B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of ecological monitoring technology, and in particular to a method and system for physical inversion of grassland biomass. Background Technology
[0002] Aboveground biomass (AGB) refers to the total biomass of the aboveground parts of plants (including stems, leaves, etc.) per unit area. It is a key indicator for measuring vegetation growth and productivity, and its dynamic changes directly reflect the health status and resource potential of grassland ecosystems.
[0003] Current common methods for obtaining aboveground biomass primarily involve acquiring multi-source data (including remote sensing imagery, meteorological data, vegetation indices, etc.), preprocessing it, and then using correlation analysis and machine learning methods to select feature variables highly correlated with biomass. Finally, a regression model is used to construct an estimate of aboveground biomass, yielding the spatial distribution of aboveground biomass for the entire region. However, these methods mainly rely on regression modeling, lack explicit physical process support, have poor model universality, and are highly sensitive to the spatial distribution, size, and representativeness of the sample dataset. When the application scenario exceeds the sample coverage area, or when significant changes occur in the ecological environment characteristics, the accuracy of the spatial distribution of aboveground biomass for the entire region will decrease significantly.
[0004] Therefore, improving the accuracy of the spatial distribution of aboveground biomass in the overall region has become an urgent technical problem to be solved. Summary of the Invention
[0005] This invention provides a method, system, electronic device, storage medium, and computer program product for physical inversion of grassland aboveground biomass, which addresses the shortcomings of low accuracy in spatial distribution of aboveground biomass in the overall region in the prior art, thereby improving the accuracy of spatial distribution of aboveground biomass in the overall region.
[0006] This invention provides a method for physical inversion of grassland biomass, comprising the following steps:
[0007] Based on the lookup table and the corrected remote sensing image of the measurement area, the actual vegetation parameters of each measurement point in the measurement area are determined; wherein, the lookup table includes multiple preset vegetation parameters and preset vegetation spectral features corresponding to each preset vegetation parameter, and all the measurement points in the measurement area correspond to one grassland type;
[0008] Based on the actual vegetation parameters and the measured values of aboveground biomass at the corresponding measurement points, a cost function is constructed, and the optimal stem-to-leaf ratio corresponding to the grassland type is iteratively solved under a preset iteration termination condition.
[0009] Based on the optimized stem-to-leaf ratio and each of the actual vegetation parameters, the optimized aboveground biomass value for each of the measurement points is obtained;
[0010] Based on the optimized aboveground biomass values of all the measurement points and the grassland type distribution maps of all the measurement areas, a aboveground biomass distribution map of the overall region is generated.
[0011] According to the present invention, a method for physical inversion of grassland biomass is provided, wherein determining the actual vegetation parameters of each measurement point in the measurement area based on a lookup table and corrected remote sensing images of the measurement area includes:
[0012] Extract the actual vegetation spectral features corresponding to each pixel from the corrected remote sensing image, and determine the initial vegetation parameters corresponding to the actual vegetation spectral features according to the lookup table;
[0013] Based on the latitude and longitude information of each measurement point, the actual vegetation parameters at the corresponding locations of each measurement point are extracted from the spatial distribution results corresponding to the initial vegetation parameters.
[0014] According to the present invention, a method for physical inversion of grassland aboveground biomass is provided, wherein a cost function is constructed based on the actual vegetation parameters and the measured aboveground biomass values at corresponding measurement points, and the optimal stem-to-leaf ratio corresponding to the grassland type is iteratively solved under a preset iteration termination condition, including:
[0015] Set the initial stem-to-leaf ratio corresponding to the grassland type;
[0016] Based on the actual vegetation parameters and the initial stem-to-leaf ratio, the estimated aboveground biomass at each measurement point is calculated.
[0017] By comparing each of the estimated aboveground biomass values with the corresponding measured aboveground biomass values, a cost function corresponding to the grassland type is constructed.
[0018] The initial stem-to-leaf ratio is iteratively optimized according to the cost function. The iteration is terminated when the number of iterations is greater than the first threshold or the error of the cost function is less than the second threshold, and the optimized stem-to-leaf ratio is output.
[0019] According to the present invention, a method for physical inversion of grassland aboveground biomass includes comparing each estimated aboveground biomass value with its corresponding measured aboveground biomass value to construct a cost function corresponding to the grassland type, comprising:
[0020] The cost function is expressed by the following formula:
[0021]
[0022] In the formula, n is the number of measurement points in the measurement area, i is a positive integer between 1 and n, and j is the grassland type. Let j be the cost function corresponding to the j-th grassland type. This represents the estimated aboveground biomass at the i-th measurement point for the j-th grassland type. This represents the measured aboveground biomass at the i-th measurement point for the j-th grassland type.
[0023] The method for physical inversion of grassland biomass provided by the present invention further includes:
[0024] Obtain multiple vegetation parameter samples;
[0025] The vegetation parameter samples are input into the radiative transfer model to obtain the vegetation spectral feature samples output by the radiative transfer model corresponding to each of the vegetation parameter samples.
[0026] The lookup table is constructed based on each of the vegetation parameter samples and their corresponding vegetation spectral feature samples.
[0027] The method for physical inversion of grassland biomass provided by the present invention further includes:
[0028] Acquire the first initial remote sensing image of the measurement area;
[0029] The grayscale values in the first initial remote sensing image are converted into radiance values to obtain the second initial remote sensing image;
[0030] The second initial remote sensing image is subjected to atmospheric correction processing using a preset atmospheric correction model, and simultaneously subjected to geometric correction processing using an image conversion method to obtain the corrected remote sensing image.
[0031] This invention also provides a grassland aboveground biomass physical inversion system, comprising the following modules:
[0032] The first processing module is used to determine the actual vegetation parameters of each measurement point in the measurement area based on a lookup table and the corrected remote sensing image of the measurement area; wherein, the lookup table includes multiple preset vegetation parameters and preset vegetation spectral features corresponding to each preset vegetation parameter, and all the measurement points in the measurement area correspond to one type of grassland;
[0033] The second processing module is used to construct a cost function based on the actual vegetation parameters and the measured values of aboveground biomass at the corresponding measurement points, and to iteratively solve the optimized stem-to-leaf ratio corresponding to the grassland type under a preset iteration termination condition.
[0034] The third processing module is used to obtain the optimized aboveground biomass value for each measurement point based on the optimized stem-leaf ratio and each of the actual vegetation parameters.
[0035] The fourth processing module is used to generate a ground biomass distribution map of the entire region based on the optimized ground biomass values of all the measurement points and the grassland type distribution maps of all the measurement areas.
[0036] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described method for physical inversion of grassland biomass.
[0037] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method for physical inversion of grassland biomass.
[0038] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method for physical inversion of grassland biomass.
[0039] By determining the actual vegetation parameters at each measurement point using a lookup table and calibrated remote sensing imagery of the measurement area, an effective correspondence between remote sensing data and vegetation structure parameters is achieved. This allows for the acquisition of actual vegetation parameters without relying on extensive field surveys, improving the efficiency and spatial coverage of parameter acquisition. By constructing a cost function based on actual vegetation parameters and measured aboveground biomass values, and optimizing the stem-to-leaf ratio under preset iteration termination conditions, structural parameters can be automatically adjusted based on measured data, improving the accuracy and adaptability of aboveground biomass estimation for different grassland types. By calculating optimized aboveground biomass values at each measurement point based on the optimized stem-to-leaf ratio and actual vegetation parameters, the estimation results fully reflect the true vegetation structure characteristics of each measurement point, making the aboveground biomass calculation at each point more accurate. By combining the optimized aboveground biomass values of all measurement points with grassland type distribution maps to generate an overall aboveground biomass distribution map of the region, a precise conversion from aboveground biomass in a single measurement area to aboveground biomass in the overall region is achieved. This improves the accuracy of the spatial distribution of aboveground biomass in the overall region, enabling a more accurate measurement of vegetation growth status and productivity. Attached Figure Description
[0040] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0041] Figure 1 This is one of the flowcharts of the grassland biomass physical inversion method provided by the present invention.
[0042] Figure 2 This is the second flowchart of the grassland biomass physical inversion method provided by the present invention.
[0043] Figure 3 This is the third flowchart of the grassland biomass physical inversion method provided by the present invention.
[0044] Figure 4 This is the fourth flowchart of the grassland biomass physical inversion method provided by the present invention.
[0045] Figure 5 This is the fifth flowchart of the grassland biomass physical inversion method provided by the present invention.
[0046] Figure 6 This is a schematic diagram of the structure of the grassland aboveground biomass physical inversion system provided by the present invention.
[0047] Figure 7 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation
[0048] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.
[0049] The following is combined with Figures 1 to 7 The present invention describes a method, system, electronic device, storage medium, and computer program product for physical inversion of grassland aboveground biomass.
[0050] Figure 1 This is one of the flowcharts illustrating the physical inversion method for grassland biomass provided by this invention, such as... Figure 1 As shown, the method includes the following steps:
[0051] Step 101: Determine the actual vegetation parameters of each measurement point in the measurement area based on the lookup table and the corrected remote sensing image of the measurement area; wherein, the lookup table includes multiple preset vegetation parameters and preset vegetation spectral characteristics corresponding to each preset vegetation parameter, and all measurement points in the measurement area correspond to one grassland type.
[0052] In this embodiment, the goal of step 101 is to determine the actual vegetation parameters at each measurement point in the measurement area based on a lookup table and calibrated remote sensing images of the measurement area. This step is introduced because the estimation of grassland aboveground biomass relies on accurate characterization of vegetation structure and spectral physiological parameters, especially key parameters such as leaf area index (LAI) and leaf dry matter content (LDMC). These parameters cannot be directly obtained from remote sensing images and require the use of physical models to invert remote sensing observation information into biologically meaningful parameters. Therefore, step 101, as the core step in parameter acquisition for the entire method, is crucial for subsequent aboveground biomass calculation and optimization.
[0053] The lookup table in step 101 refers to a mapping structure built based on the radiative transfer model. This structure takes a combination of multiple preset vegetation parameters (such as LAI and LDMC) as input and outputs the preset vegetation spectral characteristics (such as reflectance in a specific band or vegetation index value) corresponding to each parameter combination, thereby establishing a correspondence between remote sensing observation information and vegetation parameters. This lookup table is generally generated through offline simulation to ensure broad coverage of different parameter combinations. On the other hand, calibrated remote sensing imagery refers to remote sensing observation data that has undergone radiometric calibration, atmospheric correction, and geometric correction. It possesses the true meaning of surface reflectance and spatial geometric consistency, and is a necessary prerequisite for parameter inversion.
[0054] The preset vegetation spectral characteristics can be obtained by multi-band vegetation reflectance data simulated by the radiative transfer model PROSAIL (PROSPECT model and SAIL model, namely Properties Spectra and Scattering by Arbitrarily Inclined Leaves), or they can be replaced by characteristic index forms constructed by band combination, such as Normalized Difference Vegetation Index (NDVI), Modified Normalized Difference Vegetation Index (MNDVI), SimpleRatio (SR), Perpendiculars Vegetation Index (PVI), etc.
[0055] All measurement points in the measurement area correspond to a single grassland type. This setting is introduced to ensure that all data involved in the optimization have structural and ecological consistency when optimizing the stem-to-leaf ratio for that grassland type, thus avoiding systematic errors introduced by mixed grassland types.
[0056] In the specific implementation process, firstly, by using the matching relationship between spectral features and vegetation parameters provided by the lookup table, a reverse lookup is performed on the spectral features of each pixel in the corrected remote sensing image to determine its corresponding actual vegetation parameters. Then, combined with the geographical location information of the measurement points, the vegetation parameters retrieved from the corresponding pixels in the remote sensing image are assigned to the corresponding measurement points, thereby determining the actual vegetation parameters of each measurement point in the measurement area. It is also ensured that these measurement points are located within the distribution area of a known grassland type, thus guaranteeing the consistency of the stem-to-leaf ratio across all measurement points under that grassland type.
[0057] By employing the aforementioned scheme based on lookup tables and calibrated remote sensing image collaborative inversion, actual vegetation parameters at multiple measurement points within a measurement area can be rapidly and in batches obtained without direct ground measurements, enabling efficient acquisition of large-scale grassland structure information. This implementation significantly reduces field sampling costs and, relying on the spectral-parameter correspondence established by the physical model, possesses good cross-regional adaptability and scalability, providing a reliable foundation for subsequent accurate estimation of aboveground biomass.
[0058] Step 102: Based on the actual vegetation parameters and the measured values of aboveground biomass at the corresponding measurement points, construct a cost function and iteratively solve the optimal stem-to-leaf ratio corresponding to the grassland type under the preset iteration termination condition.
[0059] In this embodiment, step 102 aims to construct a cost function corresponding to the grassland type based on the actual vegetation parameters and corresponding measured aboveground biomass values at each measurement point in the measurement area, and optimize the stem-to-leaf ratio of the grassland type under a preset iteration termination condition. This step occupies a crucial intermediate position between parameter inversion and aboveground biomass estimation in the overall method. Its core purpose is to improve estimation accuracy and regional adaptability by introducing an optimization mechanism that enables the biomass estimation model to adaptively adjust its structural parameters under different grassland types.
[0060] In grassland ecosystems, vegetation structures differ significantly among different grassland types, particularly exhibiting type dependence in the mass distribution relationship between leaves and stems. The stem-to-leaf ratio, a core parameter characterizing this structural relationship, directly impacts the output of biomass estimation models. Using a fixed stem-to-leaf ratio parameter often fails to adapt to the diverse growth patterns of different grassland types, easily leading to systematic errors. Therefore, this embodiment constructs a cost function and introduces an iterative optimization process to achieve dynamic adaptation between grassland type and stem-to-leaf ratio, thereby enhancing the model's structural rationality and predictive ability.
[0061] In this step, the actual vegetation parameters refer to the vegetation structure parameters determined in step 101 above, specifically including parameters such as leaf area index (LAI) and leaf dry matter content (LDMC). These parameters reflect the potential information of grassland yield and are the core inputs of the biomass estimation model. The measured aboveground biomass values come from ground plot surveys, providing a reference for the model output. The cost function is constructed to measure the deviation between the estimated and measured biomass values, and then serves as the objective function in the optimization process to constrain the adjustment direction of the structure parameters.
[0062] Specifically, by substituting the actual vegetation parameters and measured aboveground biomass values of all measurement points under the same grassland type into the constructed cost function, an overall error metric for that grassland type under the current structural parameter settings can be established. Then, using this cost function as the optimization objective, the stem-to-leaf ratio parameter is iteratively updated to make the estimated value generally approximate the measured value. After satisfying preset termination conditions (such as a maximum iteration limit or an error threshold), the optimized stem-to-leaf ratio parameter in the converged state is output and used as the dedicated input for subsequent steps in optimizing and estimating aboveground biomass.
[0063] Step 103: Based on the optimized stem-to-leaf ratio and various actual vegetation parameters, obtain the optimized aboveground biomass value for each measurement point.
[0064] In this embodiment, the grassland aboveground biomass estimation model is highly sensitive to the stem-to-leaf ratio. Since different types of grassland have different growth structures, relying solely on general or empirical stem-to-leaf ratio parameters will lead to low estimation accuracy at the point scale. Therefore, after completing the cost function-based optimization process in the preceding steps, the optimized stem-to-leaf ratio parameters need to be reintroduced in step 103 and combined with the actual vegetation parameters of the measurement point to recalculate the aboveground biomass estimate. This yields a set of highly accurate optimized values after structural parameter adaptation, i.e., optimized aboveground biomass values, which accurately reflect the grass production capacity of each measurement point under different grassland types in the measurement area.
[0065] The specific calculation formula is as follows:
[0066]
[0067] In the formula, This represents the optimized aboveground biomass value for the i-th measurement point of the j-th grassland type. To optimize the stem-to-leaf ratio, Let be the leaf area index of the i-th measurement point for the j-th grassland type. Let represent the leaf dry matter content at the i-th measurement point for the j-th grassland type.
[0068] Step 104: Generate a ground biomass distribution map of the entire region based on the optimized aboveground biomass values of all measurement points and the grassland type distribution map of all measurement areas.
[0069] The preceding steps only completed the optimized estimation at the measurement points, and have not yet generated continuous distribution results at the regional scale. However, aboveground biomass, as a typical spatial variable, has practical value in forming a spatially continuous and logically consistent estimation layer, facilitating applications such as ecological assessment, productivity monitoring, and grassland management. Therefore, it is necessary to combine the optimized estimation point data and spatial classification information to generate an aboveground biomass distribution map of the entire measurement area through mapping extension.
[0070] The grassland type distribution map in step 104 refers to spatial vector or raster data constructed based on remote sensing image classification results or existing grassland resource survey results, used to label the grassland type to which each pixel belongs in the overall area; this distribution map not only provides geographic zoning information, but also directly corresponds to the type labels distinguished in the aforementioned stem-to-leaf ratio optimization process.
[0071] In a specific implementation, all measurement points can first be grouped according to grassland type, and measurement points of the same type can be used as control samples. Combining the known distribution of actual vegetation parameters with the spatial location relationship, spatial statistical methods such as multiple regression or kriging interpolation are used to generate pixel-level biomass estimation surfaces within the area covered by each grassland type. Subsequently, the estimation surfaces generated by different types are stitched together along the grassland type boundaries to finally form a ground biomass distribution map with the same resolution as the corrected remote sensing image and covering the entire area. To ensure spatial continuity and boundary consistency, buffer fusion or weighted smoothing strategies can be applied to the pixels at the boundary between adjacent types during the stitching process to eliminate abrupt changes introduced by type switching.
[0072] The above methods enable precise conversion of aboveground biomass from a single measurement area to the aboveground biomass of the entire area, thereby allowing for more accurate measurement of vegetation growth and productivity, and more accurate reflection of the health status and resource potential of grassland ecosystems. This, in turn, enables efficient utilization and refined management of grassland resources.
[0073] Before step 101, it is also necessary to acquire calibrated remote sensing images and a lookup table. Therefore, refer to... Figure 2 , Figure 2 This is a second schematic flowchart of the grassland biomass physical inversion method provided by the present invention. In one possible implementation, the method further includes steps 201 to 203:
[0074] Step 201: Acquire the first initial remote sensing image of the measurement area.
[0075] Step 202: Convert the grayscale values in the first initial remote sensing image into radiance values to obtain the second initial remote sensing image.
[0076] Step 203: Perform atmospheric correction processing on the second initial remote sensing image using a preset atmospheric correction model, and simultaneously perform geometric correction processing on the second initial remote sensing image through image conversion to obtain a corrected remote sensing image.
[0077] In this embodiment, steps 201 to 203 are used to perform radiometric, geometric, and atmospheric consistency processing on the initial remote sensing image of the measurement area. The purpose of this is to lay a unified and reliable data foundation for subsequent vegetation spectral feature extraction and lookup table inversion based on the remote sensing image. Since the original remote sensing image only records digital quantities (grayscale values) received by the sensor, these grayscale values are simultaneously affected by various factors such as atmospheric scattering, changes in solar altitude angle, topographic relief, and sensor geometric distortion. If used directly for vegetation parameter identification without sufficient correction, it often leads to spectral distortion and spatial misalignment, severely affecting the accuracy of subsequent biomass estimation. Therefore, this step must complete a series of physical corrections to the remote sensing image to obtain image input with true physical meaning and geometric consistency.
[0078] In step 201, the first initial remote sensing image of the measurement area is acquired. This image is a raw image product from a satellite sensor, recording multispectral data in grayscale format, covering all measurement points and their surrounding ecological areas within the measurement area. The timing of the remote sensing image of this area is selected to be synchronized with or close to the time of the ground plot survey to enhance the consistency of vegetation status and reduce the structural shift introduced by the observation time difference.
[0079] In step 202, to convert the grayscale image into a data form with physical energy meaning, the first initial remote sensing image needs to be radiometrically calibrated based on the sensor calibration parameters (such as gain and offset values) attached to the remote sensing image. This converts the grayscale values into radiance values, generating the second initial remote sensing image. This process can be completed according to a linear transformation formula and performed separately for each spectral band, thereby ensuring that the energy expression between channels remains consistent. Radiance, as the first step in the remote sensing physical processing chain, provides a fundamental value that can be further converted into ground reflectance.
[0080] In step 203, atmospheric correction is first performed to eliminate radiation interference caused by atmospheric scattering and absorption. This process can be implemented in various ways. For example, atmospheric radiative transfer simulation technology based on the MODTRAN model can be used to construct a lookup table between input parameters (such as atmospheric type, water vapor content, aerosol concentration, etc.) and surface reflectivity, and atmospheric correction can be achieved through reverse lookup. Alternatively, modular atmospheric correction tools such as FLAASH can be used to directly solve for surface reflectivity based on known radiance input and atmospheric state parameters.
[0081] Meanwhile, to ensure strict alignment between remote sensing imagery and geospatial information, geometric correction processing is also required for the second initial remote sensing imagery. Geometric correction can be achieved using a ground control point (GCP)-based method. This involves combining the image-to-map module in remote sensing processing software (such as ENVI) to establish a transformation model based on the registration relationship between control points and known map coordinates. Image resampling is then used to reproject the image onto a unified geographic reference system, thereby eliminating geometric offsets caused by attitude errors, terrain undulations, and other factors.
[0082] The remote sensing image correction process based on radiation conversion, atmospheric inversion, and geometric registration significantly improves the spectral physical meaning and spatial accuracy of remote sensing data. It effectively suppresses error factors introduced by environmental interference and sensor limitations in the original remote sensing images, provides accurate and reliable spectral input data for lookup table inversion, and establishes a standardized remote sensing data preprocessing foundation for the physical inversion method of biomass on the entire grassland.
[0083] In one possible implementation, refer to Figure 3 , Figure 3 This is the third flowchart of the grassland biomass physical inversion method provided by the present invention, which also includes steps 301 to 303:
[0084] Step 301: Obtain multiple vegetation parameter samples.
[0085] Step 302: Input the vegetation parameter samples into the radiative transfer model to obtain the vegetation spectral feature samples corresponding to each vegetation parameter sample output by the radiative transfer model.
[0086] Step 303: Construct a lookup table based on each vegetation parameter sample and its corresponding vegetation spectral feature sample.
[0087] Since the spectral features acquired from remote sensing images are only observational data that indirectly reflect the surface state, they cannot directly characterize vegetation structure and physiological parameters, such as leaf area index (LAI) and leaf dry matter content (LDMC). Therefore, it is necessary to establish a stable one-to-one correspondence between remotely observable spectral features and target vegetation parameters by building a lookup table based on a physical model. This will provide calculable and traceable structural support for subsequent parameter inversion and biomass estimation.
[0088] In step 301, multiple vegetation parameter samples are acquired. These vegetation parameter samples are a set of parameters generated according to preset rules, typically including but not limited to key parameters such as LAI and LDMC. Values can be selected within a preset reasonable range using a fixed step size or a random distribution, ensuring that the overall sample set covers typical vegetation states that may occur in grassland ecosystems. These parameters, as inputs, represent hypothetical combinations of different land cover structures and physiological states, forming the basic input space for subsequent model simulations.
[0089] In step 302, the aforementioned vegetation parameter samples are input into the radiative transfer model to obtain the spectral response corresponding to each parameter combination. Preferably, the PROSAIL (PROSPECT+SAIL) model is used as the radiative transfer model, where PROSPECT simulates spectral absorption and transmission behavior at the leaf scale, and SAIL simulates spectral transmission and directionality characteristics at the canopy scale, thereby achieving a full-chain physical simulation from vegetation structure parameters to surface reflectance. In practical implementation, simulated reflectance data in multiple bands can be generated for each parameter combination, or a set of representative vegetation spectral characteristic indices can be further calculated, including Normalized Difference Vegetation Index (NDVI), Modified Normalized Difference Vegetation Index (MNDVI), Ratio Vegetation Index (SR), and Vertical Vegetation Index (PVI), to enhance the feature sensitivity and discrimination accuracy in the inversion stage.
[0090] In step 303, a lookup table is constructed based on each vegetation parameter sample and its corresponding vegetation spectral feature sample. The lookup table uses spectral features as indexes and actual vegetation parameters such as LAI and LDMC as values, forming a data structure that supports reverse lookups. This lookup table can be implemented using hash mapping, KD-tree structures, or high-dimensional matrix forms to meet the performance requirements of efficient inversion and rapid localization. In the subsequent parameter inversion stage, the corresponding vegetation parameters can be obtained directly by inputting the spectral features of pixels in the remote sensing image and performing a reverse lookup in the lookup table. This replaces the traditional empirical modeling method based on regression fitting or sample-driven approaches, significantly improving the physical consistency and generalization ability of the model.
[0091] The above-mentioned lookup table construction scheme based on physical model-driven approach realizes the explicit correlation between remote sensing image observations and vegetation structure parameters. This not only ensures the interpretability and repeatability of the inversion results, but also significantly improves the adaptability of the estimation process under different remote sensing platforms and different regional types, providing a stable, transparent and efficient parameter basis for the universal estimation of grassland biomass on a large scale.
[0092] In one possible implementation, refer to Figure 4 , Figure 4 This is the fourth flowchart of the grassland biomass physical inversion method provided by the present invention. Step 101 specifically includes steps 401 to 402:
[0093] Step 401: Extract the actual vegetation spectral features corresponding to each pixel from the corrected remote sensing image, and determine the initial vegetation parameters corresponding to the actual vegetation spectral features according to the lookup table.
[0094] Step 402: Based on the latitude and longitude information of each measurement point, extract the actual vegetation parameters at the corresponding locations of each measurement point from the spatial distribution results corresponding to the initial vegetation parameters.
[0095] In this embodiment, steps 401 to 402 are used to extract the actual vegetation spectral features from the remote sensing image based on the acquired calibrated remote sensing image and lookup table, and to retrieve the actual vegetation parameters corresponding to the measurement points based on the lookup table. This step is a key step in achieving a one-to-one correspondence between the vegetation parameter retrieval and the ground measurement points. The fundamental purpose of this step is to accurately correspond the spectral feature information at the remote sensing pixel level with the spatially discrete locations of the measurement points, so as to quantify the vegetation structure state at the measurement point scale, thereby establishing a basic input for subsequent biomass estimation and parameter optimization.
[0096] In step 401, the actual vegetation spectral features of each pixel must first be extracted from the calibrated remote sensing image. The calibrated remote sensing image, after radiometric calibration, atmospheric correction, and geometric correction, already possesses the true reflectance of ground features and geospatial accuracy. Depending on the selected spectral feature type, multi-band reflectance can be directly extracted from each pixel, or feature indices such as NDVI, MNDVI, SR, and PVI can be calculated based on specific combination formulas. These features are consistent with the preset spectral feature structure in the lookup table and can be used as input index items for table lookup inversion. To improve inversion accuracy, it is preferable to select an index form with high sensitivity and strong discriminative power to vegetation parameters such as LAI and LDMC as the dominant feature dimension.
[0097] Subsequently, based on the extracted spectral features, a pre-constructed lookup table is used for reverse matching to retrieve the entry that most closely matches the actual spectral features, and the corresponding initial vegetation parameters are obtained. This process essentially establishes a spatial correspondence between remote sensing observation information and radiative transfer model simulation data, thereby achieving physical inversion reasoning and replacing traditional empirical regression methods. The lookup process can employ minimum Euclidean distance, weighted minimum error, or nearest neighbor search strategies to ensure that the extracted parameter values achieve optimal similarity matching.
[0098] In step 402, to establish a direct link between the above vegetation parameter results and the ground-measured data, it is necessary to combine the latitude and longitude information of each measurement point to accurately locate its spatial corresponding pixel in the remote sensing image, and extract the actual vegetation parameters at the location of the measurement point from the already retrieved pixel-level parameter distribution results. Since the corrected remote sensing image has undergone geometric correction processing, there is a one-to-one correspondence between the latitude and longitude of the measurement point and the pixel coordinates. Therefore, accurate data assignment can be achieved through spatial lookup or interpolation matching. The extraction results should maintain parameter consistency to ensure that the actual vegetation parameters such as LAI and LDMC obtained for each measurement point can be used as input variables for the subsequent aboveground biomass estimation model.
[0099] By combining remote sensing images with lookup tables to perform spectral inversion and spatial extraction processing, the actual vegetation parameters at each measurement point can be obtained by back-calculation using a physical model without continuous ground sampling.
[0100] In one possible implementation, refer to Figure 5 , Figure 5 This is the fifth flowchart of the grassland biomass physical inversion method provided by the present invention. Step 102 specifically includes steps 501 to 504:
[0101] Step 501: Set the initial stem-to-leaf ratio corresponding to the grassland type.
[0102] In this embodiment, the purpose of step 501 is to set an initial stem-to-leaf ratio corresponding to the grassland type as the starting input for subsequent aboveground biomass estimation and parameter optimization. This step is implemented because different types of grasslands have significant differences in structural composition, especially in the mass distribution ratio of leaves to stems. Therefore, directly using a uniform, fixed stem-to-leaf ratio will be difficult to accurately reflect the actual structural characteristics of each grassland type, easily introducing systematic estimation bias. To achieve the subsequent optimization of the stem-to-leaf ratio, it is necessary to first provide a reasonable initial estimate for each grassland type, so that the optimization algorithm has a good convergence starting point and physical constraints.
[0103] Specifically, for each grassland type, an initial stem-to-leaf ratio needs to be set. This initial value can be set based on historical research data, measured values from typical sample plots, empirical values from literature, or expert knowledge and experience, preferably covering the mainstream structural range that the grassland type may exhibit. For example, for meadow-type grasslands dominated by leaves, the initial stem-to-leaf ratio can be set to a smaller value (e.g., 0.5 to 0.8), while for typical desert grasslands with robust structures and a high proportion of stems, a higher value (e.g., 1.2 to 1.5) can be set. The initial stem-to-leaf ratio can be stored as a fixed parameter vector in the model configuration file and loaded into the parameter environment of each grassland type before the cost function construction and iterative optimization begin.
[0104] When setting the initial stem-to-leaf ratio, to prevent getting stuck in local optima or numerical oscillations during the optimization process, it is preferable to use an initial value that is ecologically reasonable and mathematically stable. This is combined with grassland type coding for group management, allowing each type of grassland to be optimized independently and avoiding cross-interference. Furthermore, setting the initial value can also be used to control the search range and step size of the optimization strategy, thereby improving overall estimation efficiency.
[0105] Step 502: Calculate the estimated aboveground biomass for each measurement point based on the actual vegetation parameters and the initial stem-to-leaf ratio.
[0106] To optimize the parameters in the aboveground biomass retrieval process, a computational path must be constructed from the input vegetation parameters to the output estimated values. Based on this, a cost function is defined by quantifying the error between the estimated and measured data. Therefore, constructing a preliminary aboveground biomass estimate using the initial stem-to-leaf ratio and known actual vegetation parameters is not only a prerequisite for constructing the cost function but also the logical starting point for subsequent parameter optimization.
[0107] In a specific implementation, the actual vegetation parameters determined in steps 401 to 402 for each measurement point are first used as input, preferably including leaf area index (LAI) and leaf dry matter content (LDMC). These parameters characterize the structural density and biological mass of the surface vegetation and are core variables affecting grassland yield estimation. Then, based on the structural physics model and combined with the initial stem-to-leaf ratio corresponding to the current grassland type, the estimated aboveground biomass value for each measurement point is calculated.
[0108] The estimated aboveground biomass can be calculated using the following formula:
[0109]
[0110] in, This represents the estimated aboveground biomass at the i-th measurement point for the j-th grassland type. Let the initial stem-to-leaf ratio be that of the j-th grassland type. Let be the leaf area index of the i-th measurement point for the j-th grassland type. Let represent the leaf dry matter content at the i-th measurement point for the j-th grassland type.
[0111] Step 503: Compare the estimated values of each aboveground biomass with the corresponding measured values of aboveground biomass to construct the cost function corresponding to the grassland type.
[0112] To optimally determine the stem-to-leaf ratio corresponding to each grassland type, an error evaluation function must be constructed using actual observation data as a reference, thereby providing a clear numerical feedback mechanism for the parameter optimization process. The construction of the cost function is not only the core driver of the optimization iteration but also the foundation for ensuring the ecological interpretability and data consistency of the optimization results. Therefore, in this embodiment, the purpose of step 503 is to construct a cost function specific to each grassland type based on the deviation between the estimated aboveground biomass and the corresponding measured aboveground biomass.
[0113] In a specific implementation, for each grassland type j, all measurement points i=1,2,...,n belonging to that type are summarized, and the estimated aboveground biomass value of each measurement point obtained in step 502 is used. Corresponding measured aboveground biomass values By comparing the results, calculating the squared relative error, and then averaging and taking the square root over all measurement points, the following cost function is constructed:
[0114]
[0115] In the formula, n is the number of measurement points in the measurement area, i is a positive integer between 1 and n, and j is the grassland type. Let j be the cost function corresponding to the j-th grassland type. This represents the estimated aboveground biomass at the i-th measurement point for the j-th grassland type. This represents the measured aboveground biomass at the i-th measurement point for the j-th grassland type. The smaller the value, the closer the estimated value is to the measured value.
[0116] Step 504: Iteratively optimize the initial stem-to-leaf ratio based on the cost function. Terminate the iteration when the number of iterations is greater than the first threshold or the error of the cost function is less than the second threshold, and output the optimized stem-to-leaf ratio.
[0117] In this embodiment, the purpose of step 504 is to use the cost function corresponding to the grassland type constructed in the previous step to perform adaptive iterative optimization of the initial stem-to-leaf ratio for that grassland type, and output the optimized stem-to-leaf ratio after satisfying the preset convergence condition. This iterative process is necessary because different grassland types have natural differences in their growth structure. Relying solely on one-time empirical settings often fails to obtain the optimal structural parameters that match the measured aboveground biomass. By progressively updating the cost function as the evaluation index, estimation errors can be significantly reduced while ensuring physical interpretability, and the model can adaptively adjust to regional ecological differences.
[0118] In this step, the cost function is the scalar function defined above in the form of normalized root mean square relative error. Its numerical value directly characterizes the degree to which the estimated aboveground biomass corresponding to the current stem-to-leaf ratio deviates from the measured value; the initial stem-to-leaf ratio is a preset starting parameter for each grassland type before the iteration begins; the first threshold is used to limit the maximum number of iterations to prevent long-term computation from leading to excessively low convergence efficiency; the second threshold specifies the minimum allowable error limit of the cost function, when... A value below this threshold indicates that the model has reached acceptable accuracy. The final output optimized stem-to-leaf ratio refers to the structural parameter that significantly reduces estimation error after satisfying any termination condition.
[0119] Specific implementation methods can employ a single-parameter golden section search or an iterative one-dimensional adaptive step-size method based on error gradients. The algorithm first uses an initial stem-to-leaf ratio... Starting from this point, calculate the corresponding cost function. Then, the stem-to-leaf ratio is adjusted according to the current error gradient direction. Recalculate .like (Where ϵ is the set error convergence threshold, i.e., the second threshold, and the preferred value of the second threshold is 5%) or the iteration count reaches the first threshold. (If the preferred value for the first threshold is 5000), then convergence is determined and output. The optimal stem-to-leaf ratio is used for this grassland type; otherwise, iteration continues until... The process continues until any of the above termination conditions are met. To improve numerical stability, the step size can be adaptively scaled according to the magnitude of the error decrease, avoiding oscillations or premature convergence.
[0120] The above-mentioned cost function-driven iterative optimization scheme enables the stem-to-leaf ratio parameter to automatically approximate the optimal solution from the initial empirical value. While maintaining the physical meaning, it significantly reduces the deviation between the model output and the actual measurement, and improves the accuracy and robustness of aboveground biomass estimation under different grassland types.
[0121] The grassland aboveground biomass physical inversion system provided by this invention is described below. The grassland aboveground biomass physical inversion system described below can be referred to in conjunction with the grassland aboveground biomass physical inversion method described above. (Refer to...) Figure 6 , Figure 6 This is a schematic diagram of the physical inversion system for grassland biomass provided by the present invention. The system includes:
[0122] The first processing module is used to determine the actual vegetation parameters of each measurement point in the measurement area based on the lookup table and the corrected remote sensing image of the measurement area. The lookup table includes multiple preset vegetation parameters and preset vegetation spectral features corresponding to each preset vegetation parameter. All measurement points in the measurement area correspond to one type of grassland.
[0123] The second processing module is used to construct a cost function based on the actual vegetation parameters and the measured values of aboveground biomass at the corresponding measurement points, and to iteratively solve the optimal stem-to-leaf ratio corresponding to the grassland type under the preset iteration termination condition.
[0124] The third processing module is used to obtain the optimized aboveground biomass value for each measurement point based on the optimized stem-to-leaf ratio and various actual vegetation parameters.
[0125] The fourth processing module is used to generate a ground biomass distribution map of the entire region based on the optimized ground biomass values of all measurement points and the grassland type distribution map of all measurement areas.
[0126] In one possible implementation, the first processing module is further configured to:
[0127] Extract the actual vegetation spectral features corresponding to each pixel from the corrected remote sensing image, and determine the initial vegetation parameters corresponding to the actual vegetation spectral features according to the lookup table;
[0128] Based on the latitude and longitude information of each measurement point, the actual vegetation parameters at the corresponding locations of each measurement point are extracted from the spatial distribution results corresponding to the initial vegetation parameters.
[0129] In one possible implementation, the second processing module is further configured to:
[0130] Set the initial stem-to-leaf ratio corresponding to the grassland type;
[0131] Based on the actual vegetation parameters and the initial stem-to-leaf ratio, the estimated aboveground biomass at each measurement point was calculated.
[0132] By comparing the estimated values of each aboveground biomass with the corresponding measured values of aboveground biomass, a cost function corresponding to each grassland type is constructed.
[0133] The initial stem-to-leaf ratio is iteratively optimized based on the cost function. The iteration is terminated when the number of iterations exceeds the first threshold or the error of the cost function is less than the second threshold, and the optimized stem-to-leaf ratio is output.
[0134] In one possible implementation, the first processing module is further configured to:
[0135] Obtain multiple vegetation parameter samples;
[0136] The vegetation parameter samples are input into the radiative transfer model to obtain the vegetation spectral feature samples corresponding to each vegetation parameter sample output by the radiative transfer model.
[0137] A lookup table is constructed based on each vegetation parameter sample and its corresponding vegetation spectral feature sample.
[0138] In one possible implementation, the first processing module is further configured to:
[0139] Acquire the first initial remote sensing image of the measurement area;
[0140] The grayscale values in the first initial remote sensing image are converted into radiance values to obtain the second initial remote sensing image;
[0141] Atmospheric correction processing is performed on the second initial remote sensing image using a preset atmospheric correction model, and geometric correction processing is performed on the second initial remote sensing image using image conversion to obtain a corrected remote sensing image.
[0142] Figure 7 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 7As shown, the electronic device may include: a processor 710, a communication interface 720, a memory 730, and a communication bus 740, wherein the processor 710, the communication interface 720, and the memory 730 communicate with each other through the communication bus 740. The processor 710 can call logical instructions in the memory 730 to execute a grassland aboveground biomass physical inversion method. This method includes: determining the actual vegetation parameters of each measurement point in the measurement area based on a lookup table and calibrated remote sensing images of the measurement area; wherein the lookup table includes multiple preset vegetation parameters and preset vegetation spectral characteristics corresponding to each preset vegetation parameter, and all measurement points in the measurement area correspond to one grassland type; constructing a cost function based on each actual vegetation parameter and the measured aboveground biomass value of the corresponding measurement point, and iteratively solving the optimized stem-to-leaf ratio corresponding to the grassland type under a preset iteration termination condition; obtaining the optimized aboveground biomass value of each measurement point based on the optimized stem-to-leaf ratio and each actual vegetation parameter; and generating an aboveground biomass distribution map of the entire area based on the optimized aboveground biomass values of all measurement points and the grassland type distribution map of all measurement areas.
[0143] Furthermore, the logical instructions in the aforementioned memory 730 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0144] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the grassland aboveground biomass physical inversion method provided by the above methods. The method includes: determining the actual vegetation parameters of each measurement point in the measurement area based on a lookup table and calibrated remote sensing images of the measurement area; wherein the lookup table includes multiple preset vegetation parameters and preset vegetation spectral features corresponding to each preset vegetation parameter, and all measurement points in the measurement area correspond to one grassland type; constructing a cost function based on each actual vegetation parameter and the measured aboveground biomass value of the corresponding measurement point, and iteratively solving the optimized stem-to-leaf ratio corresponding to the grassland type under a preset iteration termination condition; obtaining the optimized aboveground biomass value of each measurement point based on the optimized stem-to-leaf ratio and each actual vegetation parameter; and generating an aboveground biomass distribution map of the entire area based on the optimized aboveground biomass value of all measurement points and the grassland type distribution map of all measurement areas.
[0145] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the grassland aboveground biomass physical inversion method provided by the above methods. The method includes: determining the actual vegetation parameters of each measurement point in the measurement area based on a lookup table and calibrated remote sensing images of the measurement area; wherein the lookup table includes multiple preset vegetation parameters and preset vegetation spectral features corresponding to each preset vegetation parameter, and all measurement points in the measurement area correspond to a grassland type; constructing a cost function based on each actual vegetation parameter and the measured aboveground biomass value of the corresponding measurement point, and iteratively solving the optimized stem-to-leaf ratio corresponding to the grassland type under a preset iteration termination condition; obtaining the optimized aboveground biomass value of each measurement point based on the optimized stem-to-leaf ratio and each actual vegetation parameter; and generating an aboveground biomass distribution map of the entire area based on the optimized aboveground biomass value of all measurement points and the grassland type distribution map of all measurement areas.
[0146] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.
[0147] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.
[0148] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do 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 method for physical inversion of grassland aboveground biomass, characterized in that, include: Based on a lookup table and calibrated remote sensing images of the measurement area, the actual vegetation parameters of each measurement point in the measurement area are determined. The lookup table includes multiple preset vegetation parameters and preset vegetation spectral features corresponding to each preset vegetation parameter. All measurement points in the measurement area correspond to one type of grassland. Determining the actual vegetation parameters of each measurement point in the measurement area includes: extracting the actual vegetation spectral features corresponding to each pixel from the calibrated remote sensing image; performing a reverse lookup on the spectral features of each pixel in the calibrated remote sensing image according to the lookup table to determine the initial vegetation parameters corresponding to the actual vegetation spectral features; and extracting the actual vegetation parameters at the corresponding locations of each measurement point from the spatial distribution results corresponding to the initial vegetation parameters based on the latitude and longitude information of each measurement point. Based on the actual vegetation parameters and the measured aboveground biomass values at the corresponding measurement points, a cost function is constructed, and the optimized stem-to-leaf ratio corresponding to the grassland type is iteratively solved under a preset iteration termination condition. The construction of the cost function includes: setting an initial stem-to-leaf ratio corresponding to the grassland type; calculating the estimated aboveground biomass value at each measurement point based on the actual vegetation parameters and the initial stem-to-leaf ratio; and comparing each estimated aboveground biomass value with its corresponding measured aboveground biomass value to construct the cost function corresponding to the grassland type. Based on the optimized stem-to-leaf ratio and each of the actual vegetation parameters, the optimized aboveground biomass value for each of the measurement points is obtained; Based on the optimized aboveground biomass values of all the measurement points and the grassland type distribution maps of all the measurement areas, a aboveground biomass distribution map of the overall region is generated.
2. The method for physical inversion of grassland aboveground biomass according to claim 1, characterized in that, The iterative solution of the optimized stem-to-leaf ratio corresponding to the grassland type under the preset iteration termination condition includes: The initial stem-to-leaf ratio is iteratively optimized according to the cost function. The iteration is terminated when the number of iterations is greater than the first threshold or the error of the cost function is less than the second threshold, and the optimized stem-to-leaf ratio is output.
3. The method for physical inversion of grassland aboveground biomass according to claim 1, characterized in that, The step of comparing each of the estimated aboveground biomass values with their corresponding measured aboveground biomass values to construct a cost function corresponding to the grassland type includes: The cost function is expressed by the following formula: ； In the formula, n is the number of measurement points in the measurement area, i is a positive integer between 1 and n, and j is the grassland type. Let j be the cost function corresponding to the j-th grassland type. This represents the estimated aboveground biomass at the i-th measurement point for the j-th grassland type. This represents the measured aboveground biomass at the i-th measurement point for the j-th grassland type.
4. The method for physical inversion of grassland aboveground biomass according to claim 1, characterized in that, Also includes: Obtain multiple vegetation parameter samples; The vegetation parameter samples are input into the radiative transfer model to obtain the vegetation spectral feature samples output by the radiative transfer model corresponding to each of the vegetation parameter samples. The lookup table is constructed based on each of the vegetation parameter samples and their corresponding vegetation spectral feature samples.
5. The method for physical inversion of grassland aboveground biomass according to claim 1, characterized in that, Also includes: Acquire the first initial remote sensing image of the measurement area; The grayscale values in the first initial remote sensing image are converted into radiance values to obtain the second initial remote sensing image; The second initial remote sensing image is subjected to atmospheric correction processing using a preset atmospheric correction model, and simultaneously subjected to geometric correction processing using an image conversion method to obtain the corrected remote sensing image.
6. A physical inversion system for grassland aboveground biomass, characterized in that, include: The first processing module is used to determine the actual vegetation parameters of each measurement point in the measurement area based on a lookup table and a corrected remote sensing image of the measurement area. The lookup table includes multiple preset vegetation parameters and preset vegetation spectral features corresponding to each preset vegetation parameter. All measurement points in the measurement area correspond to a single grassland type. Determining the actual vegetation parameters of each measurement point in the measurement area includes: extracting the actual vegetation spectral features corresponding to each pixel from the corrected remote sensing image; performing a reverse lookup on the spectral features of each pixel in the corrected remote sensing image based on the lookup table to determine the initial vegetation parameters corresponding to the actual vegetation spectral features; and extracting the actual vegetation parameters at the corresponding locations of each measurement point from the spatial distribution results corresponding to the initial vegetation parameters based on the latitude and longitude information of each measurement point. The second processing module is used to construct a cost function based on the actual vegetation parameters and the measured values of aboveground biomass at the corresponding measurement points, and iteratively solve the optimized stem-to-leaf ratio corresponding to the grassland type under a preset iteration termination condition; wherein, constructing the cost function includes: setting an initial stem-to-leaf ratio corresponding to the grassland type; calculating the estimated aboveground biomass value at each measurement point based on the actual vegetation parameters and the initial stem-to-leaf ratio; comparing each estimated aboveground biomass value with the corresponding measured aboveground biomass value to construct the cost function corresponding to the grassland type; The third processing module is used to obtain the optimized aboveground biomass value for each measurement point based on the optimized stem-leaf ratio and each of the actual vegetation parameters. The fourth processing module is used to generate a ground biomass distribution map of the entire region based on the optimized ground biomass values of all the measurement points and the grassland type distribution maps of all the measurement areas.
7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the grassland aboveground biomass physical inversion method as described in any one of claims 1 to 5.
8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the grassland biomass physical inversion method as described in any one of claims 1 to 5.
9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the grassland biomass physical inversion method as described in any one of claims 1 to 5.