Satellite remote sensing estimation method for aboveground biomass of desert steppe shrubs

Abstract

The invention discloses a desert steppe shrub above-ground biomass satellite remote sensing estimation method, which is applied to desert steppe shrub above-ground biomass estimation and comprises the following steps: acquiring satellite remote sensing data of an unmanned aerial vehicle measurement area arranged in a desert steppe shrub area; performing gridding processing on an unmanned aerial vehicle sample plot arranged in the unmanned aerial vehicle measurement area to obtain a sample plot grid; acquiring remote sensing data corresponding to the sample plot grid according to the satellite remote sensing data of the unmanned aerial vehicle measurement area and the sample plot grid; and inputting remote sensing data corresponding to the sample plot grids into the trained relation model of the overground biomass of the shrubs in the unit area and the remote sensing data to obtain the overground biomass of the shrubs in the unit area. According to the method, through cooperative monitoring of the desert steppe shrub overground biomass from point to surface to domain, the monitoring scale limitation can be broken through, and rapid and accurate shrub overground biomass estimation can be realized in a region or a larger scale.

Description

Technical Field

[0001] The present application relates to the field of ecological remote sensing, and in particular to a satellite remote sensing estimation method for shrub biomass on the western desert grassland of Inner Mongolia Autonomous Region from point to surface and then to domain based on "air-space-ground" coordinated monitoring. Background Art

[0002] In the field of ecological and environmental monitoring, the estimation of shrub aboveground biomass is crucial for evaluating ecosystem functions, carbon cycles, and sustainable land use. In particular, in ecological research and resource management at a regional or larger scale, it is of great significance to quickly and accurately obtain shrub aboveground biomass data. However, at present, traditional ground survey methods and drone-based estimation methods are difficult to achieve rapid and accurate shrub aboveground biomass estimation at a regional or larger scale. Summary of the invention

[0003] The purpose of the embodiments of the present application is to provide a method for estimating aboveground biomass of shrubs in desert steppes by satellite remote sensing, which method can achieve rapid and accurate estimation of aboveground biomass of shrubs on a regional or larger scale.

[0004] In order to achieve the above-mentioned purpose, the embodiment of the present application provides a method for estimating aboveground biomass of shrubs in desert steppes by satellite remote sensing, which is applied to estimating aboveground biomass of shrubs in desert steppes, and comprises:

[0005] Obtain satellite remote sensing data from drone survey areas deployed in desert steppe shrub areas;

[0006] Gridding the drone sample plots deployed in the drone survey area to obtain sample plot grids;

[0007] According to the satellite remote sensing data and sample plot grid of the UAV survey area, the remote sensing data corresponding to the sample plot grid is obtained;

[0008] The remote sensing data corresponding to the sample plot grids were input into the trained relationship model between the aboveground biomass of shrubs per unit area and remote sensing data to obtain the aboveground biomass of shrubs per unit area.

[0009] Optionally, obtaining remote sensing data corresponding to the sample plot grid according to the satellite remote sensing data and the sample plot grid of the drone survey area includes:

[0010] Obtain satellite remote sensing data of the area surveyed by the drone, wherein the spatial resolution of the remote sensing data is consistent with the grid size of the sample plot;

[0011] Based on the coordinates of the center point of the sample plot grid, the remote sensing data corresponding to the sample plot grid is obtained in the acquired remote sensing data.

[0012] As an option, the process of constructing the relationship model between the aboveground biomass of shrubs per unit area and remote sensing data is as follows:

[0013] Constructing a training data set, wherein the training data set includes training samples, the training samples include input features and output labels, the input features use remote sensing data corresponding to the sample plot grid, and the output labels use aboveground biomass per unit area in the grid;

[0014] Determine a loss function and an optimizer; wherein the loss function is used to measure the difference between the model prediction result and the output label, and the optimizer is used to adjust the model parameters to minimize the loss function;

[0015] Input features into the deep learning model to get prediction results;

[0016] The loss value of the prediction result and the output label is calculated, and the gradient of the loss function to the model parameters is calculated by the back propagation algorithm. The optimizer updates the model parameters according to the gradient to gradually reduce the loss value until the loss value converges or reaches a preset number of training rounds, completing the supervised learning training of the deep learning model and obtaining the relationship model between the aboveground biomass of shrubs per unit area and remote sensing data.

[0017] As an option, the process of obtaining the aboveground biomass per unit area in the grid is as follows:

[0018] The phenotypic structural parameters of the shrubs in the drone sample plot were input into the composite model for estimating the aboveground biomass of the desert steppe shrubs to calculate the aboveground biomass of each shrub in the drone sample plot; the phenotypic structural parameters of the shrubs included the crown area and height of the shrubs;

[0019] The drone sample plots were gridded, and the total aboveground biomass of shrubs in each grid in the drone sample plots was obtained based on the aboveground biomass of each shrub in the drone sample plots;

[0020] Based on the total aboveground biomass of shrubs in each grid in the drone sample plot, the aboveground biomass per unit area in each grid was calculated.

[0021] As an option, the process of obtaining the phenotypic structural parameters of the shrubs in the drone sample plot is as follows:

[0022] Several drone sample plots were randomly set up according to the spatial distribution range of desert steppe shrubs;

[0023] Use the multispectral sensor carried by the drone to collect multispectral images of the sample site, and use the lidar sensor carried by the drone to collect lidar data of the sample site;

[0024] Perform image segmentation on the multispectral image to obtain the crown area of ​​each shrub;

[0025] The LiDAR data was inverted to obtain the height of the shrubs.

[0026] Optionally, performing image segmentation on the multispectral image to obtain the crown area of ​​each shrub includes:

[0027] Stitching multispectral images to obtain a stitched spectral image;

[0028] The spliced ​​spectral images are input into the image segmentation model to identify the boundaries and feature information of the shrubs and background objects;

[0029] According to the boundary and feature information between the shrubs and the background objects, mark the segmented area corresponding to each shrub;

[0030] The crown area of ​​each shrub was calculated based on the segmented area corresponding to each shrub and the spatial resolution of the image.

[0031] Optionally, the laser radar data is inverted to obtain the height value of the shrub, including:

[0032] Preprocess the LiDAR data to obtain a subset of shrub point clouds, including:

[0033] The lidar data collected from different perspectives and at different times are unified into the same coordinate system, and the registration algorithm is used to stitch the lidar data together to form a complete sample point cloud data set;

[0034] De-noising the sample point cloud dataset;

[0035] The denoised sample point cloud dataset is separated into ground points and shrub points to obtain a shrub point cloud subset.

[0036] Cluster analysis is performed on the shrub point cloud subset to obtain the range of each shrub in the shrub point cloud subset, including:

[0037] According to the spatial clustering characteristics of the point cloud, the point clouds of the same shrubs are clustered into the same category, so that each clustering result corresponds to a shrub, so as to divide the range of each shrub in the shrub point cloud subset;

[0038] Perform height inversion on the point cloud data of each shrub to obtain the height of each shrub, including:

[0039] The spatial coordinate query operation is used to obtain the maximum point coordinates and the minimum point coordinates of each shrub in the vertical direction of the point cloud, where the maximum point coordinates represent the top position of the shrub, and the minimum point coordinates represent the bottom position of the shrub;

[0040] The height of each shrub in the survey plot was obtained by subtracting the coordinates of the minimum point from the coordinates of the maximum point.

[0041] As an option, the construction process of the composite model for estimating aboveground biomass of shrubs in desert steppe is as follows:

[0042] The composite model for estimating aboveground biomass of shrubs in desert steppes was obtained by fitting the phenotypic structural parameters of shrubs with the measured aboveground biomass of shrubs and the pseudo-measured aboveground biomass of shrubs.

[0043] Among them, the pseudo-measured shrub aboveground biomass was calculated by inputting the phenotypic structure parameters of the shrubs in the drone sample plot into the shrub aboveground biomass estimation model.

[0044] As an option, the construction process of the shrub aboveground biomass estimation model is as follows:

[0045] The phenotypic structural characteristics and aboveground biomass of each shrub in the survey plots set up in the desert steppe shrub area were collected;

[0046] According to the phenotypic structural characteristics and aboveground biomass of each shrub, with the phenotypic structural characteristics of the shrub as the independent variable and the aboveground biomass as the dependent variable, an aboveground biomass estimation model of shrubs was constructed by species, and the aboveground biomass estimation model of shrubs of the corresponding species was obtained.

[0047] The satellite remote sensing estimation method for the aboveground biomass of shrubs in desert steppes provided in the embodiment of the present application can break through the limitation of monitoring scale through the coordinated monitoring of the aboveground biomass of shrubs in desert steppes from point to surface and then to domain, and realize rapid and accurate aboveground biomass estimation of shrubs on a regional or larger scale, thereby obtaining more comprehensive biomass information and making up for the limitations of a single monitoring method; the accuracy and reliability of aboveground biomass estimation of shrubs can be improved, and the growth status of shrubs in desert steppes can be more accurately grasped. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A flow chart of a method for estimating aboveground biomass of shrubs in desert steppes by satellite remote sensing provided in an embodiment of the present application;

[0049] Figure 2 A flowchart of step S200 in the method for estimating aboveground biomass of shrubs in desert steppes provided in an embodiment of the present application via satellite remote sensing;

[0050] Figure 3 A flowchart of step S300 in the method for estimating aboveground biomass of shrubs in desert steppes provided in an embodiment of the present application via satellite remote sensing;

[0051] Figure 4 A flowchart of step S310 in the method for estimating aboveground biomass of shrubs in desert steppes provided in an embodiment of the present application via satellite remote sensing;

[0052] Figure 5 A flowchart of step S3110 in the method for estimating aboveground biomass of shrubs in desert steppes provided in an embodiment of the present application via satellite remote sensing;

[0053] Figure 6 A schematic diagram of the result of obtaining the aboveground biomass of shrubs per unit area in a drone sample plot by using the satellite remote sensing estimation method for aboveground biomass of shrubs in desert steppes provided in an embodiment of the present application; DETAILED DESCRIPTION

[0054] Various schemes and features of the present application are described herein with reference to the accompanying drawings. It should be understood that various modifications may be made to the embodiments of the present application. Therefore, the above description should not be considered as limiting, but only as an example of an embodiment. Other modifications within the scope and spirit of the present application will occur to those skilled in the art.

[0055] At present, the traditional ground survey method is to measure the height, crown width, ground diameter and other parameters of shrubs on the spot and estimate the biomass through empirical formulas. Although it can guarantee the accuracy of the results to a certain extent, this method requires a lot of manpower and material resources, and is time-consuming and labor-intensive. When faced with large areas, the efficiency is extremely low and cannot meet the needs of rapid estimation.

[0056] With the development of technology, drone technology has been applied to the estimation of shrub biomass. It can obtain high-resolution images and has certain advantages in estimating shrub biomass in a small area. However, due to the limitations of endurance and flight area, drones are difficult to achieve rapid and comprehensive data collection on a regional or larger scale, resulting in limitations in estimating biomass in large areas.

[0057] Satellite remote sensing technology has the advantages of wide coverage and short observation period, and can quickly obtain surface information over a large area. Based on this, combined with the rich spectral, texture and other information of satellite remote sensing images, a shrub aboveground biomass estimation model adapted to it is established, which can achieve rapid and accurate estimation of shrub aboveground biomass on a regional or larger scale, which is of great significance to ecological environment monitoring and management.

[0058] The satellite remote sensing estimation method for the aboveground biomass of desert steppe shrubs provided in this application is used for satellite remote sensing estimation of the aboveground biomass of desert steppe shrubs. Among them, desert steppe shrubs refer to shrub plants growing in desert steppe areas. Desert steppe areas are a specific ecological region with dual characteristics of desert and steppe. The vegetation is mainly shrubs, and also contains a very small amount of herbaceous plants, etc. The ecological environment is relatively fragile and unique. The aboveground biomass of shrubs refers to the total amount of organic matter in the aboveground parts of all shrubs in a certain area. It is an important indicator for measuring the growth status of shrubs, ecosystem productivity, and carbon storage.

[0059] The satellite remote sensing estimation method for desert steppe shrub biomass provided in the present application realizes the coordinated monitoring of desert steppe shrub biomass from point to surface and then to domain through the combination of satellite remote sensing technology with unmanned aerial vehicle remote sensing technology and ground measured data (for example, shrub biomass and shrub phenotypic structure parameters, etc.), and can realize rapid and accurate shrub biomass estimation on a regional or larger scale, thereby providing a technical reference for desert steppe resource surveys, ecological environment estimation and grassland carbon estimation.

[0060] The following is a detailed description of the satellite remote sensing estimation method for desert steppe shrub biomass provided in the embodiment of the present application in conjunction with the accompanying drawings. Figure 1 The flowchart of the method for estimating aboveground biomass of shrubs in desert steppes by satellite remote sensing provided in the embodiment of the present application is as follows: Figure 1 As shown, the method comprises the following steps:

[0061] S100: Obtain satellite remote sensing data of the drone survey area deployed in the first area of ​​the desert steppe.

[0062] The first desert steppe area can be a specific area divided according to specific research purposes, geographical features or management needs. In the first desert steppe area, drones are used to plan and deploy survey areas according to specific research tasks and data requirements.

[0063] Remote sensing data obtained from satellites about drone surveys of desert steppe shrub areas can be used to identify different types of land features (e.g., vegetation, soil, water bodies, etc.) and monitor the growth status of vegetation.

[0064] Satellite remote sensing data includes remote sensing reflectivity data and vegetation index. Remote sensing reflectivity data refers to the ratio of radiation reflected from surface objects to incident radiation received by satellite sensors, which reflects the reflection characteristics of surface objects to electromagnetic waves of different wavelengths. Different ground objects, such as vegetation, water bodies, soil, etc., have obvious differences in remote sensing reflectivity in different bands. Taking vegetation as an example, green plants have a higher absorption rate and lower reflectivity in the visible red light band; while in the near-infrared band, due to the characteristics of plant cell structure, the reflectivity is higher. By analyzing remote sensing reflectivity data, information such as the type, status and distribution of ground objects can be obtained.

[0065] Vegetation index is obtained by mathematical calculation of reflectivity of different bands in satellite remote sensing data. It can reflect vegetation growth, coverage, biomass and other information. Common vegetation indices include Normalized Difference Vegetation Index (NDVI), Difference Vegetation Index (DVI), Enhanced Vegetation Index (EVI) and Soil Adjusted Vegetation Index (SAVI). The more luxuriant the vegetation growth, the higher the NDVI value; conversely, the lower the NDVI value.

[0066] S200: Gridding the drone sample plots deployed in the drone survey area to obtain sample plot grids.

[0067] Specifically, multiple drone plots can be deployed in the drone survey area as needed. In order to better manage and analyze the data of the drone plots, the drone plots are gridded. That is, each drone plot is divided into several regular grid units, which have a certain size and shape (such as square or rectangle). Through gridding, it is more convenient to count and analyze the data in the plot. For example, in a desert steppe plot, it is divided into 10m×10m grids, and each grid can be used as an independent analysis unit to calculate indicators such as vegetation index and remote sensing reflectivity in the area.

[0068] S300, acquiring remote sensing data corresponding to the sample plot grid according to the satellite remote sensing data of the drone survey area and the sample plot grid;

[0069] In order to obtain the remote sensing data corresponding to the sample plot grid, it is first necessary to perform preprocessing such as radiation calibration, atmospheric correction, image rectification, image enhancement, and image fusion on the satellite remote sensing data.

[0070] After the preprocessing is completed, the geographic coordinate information of the sample plot grid is spatially matched with the preprocessed satellite remote sensing data. Through the geographic information system (GIS) technology, the location range of each sample plot grid on the satellite remote sensing image can be accurately identified. Then, the spectral data and other information of all pixel points in the corresponding sample plot grid are extracted from the satellite remote sensing data. These extracted data are the remote sensing data corresponding to the sample plot grid.

[0071] S400, inputting the satellite remote sensing data corresponding to the sample plot grid into the trained relationship model between the aboveground biomass of shrubs per unit area and remote sensing data to obtain the aboveground biomass of shrubs per unit area.

[0072] Specifically, the remote sensing reflectance data or vegetation index can be input into the relationship model between the aboveground biomass of shrubs per unit area and the remote sensing data, and the relationship model between the aboveground biomass of shrubs per unit area and the remote sensing data outputs the aboveground biomass of shrubs within the unit area.

[0073] The satellite remote sensing estimation method for the aboveground biomass of shrubs in desert steppes in this application breaks through the limitation of a single scale and can realize the scale-up of aboveground biomass of shrubs from single shrubs to drone scale and then to remote sensing scale. This multi-scale estimation capability can flexibly select the appropriate scale for estimating aboveground biomass of shrubs according to different research needs and application scenarios. For detailed research in local areas, we can focus on single shrubs and drone scales to obtain high-precision data; and for large-scale desert steppe monitoring, remote sensing scales can provide macroscopic biomass distribution information, thereby improving the accuracy and comprehensiveness of aboveground biomass estimates.

[0074] Through the satellite remote sensing estimation method of desert steppe shrub aboveground biomass in this application, it is possible to regularly obtain aboveground biomass data at different scales and timely grasp the changes in aboveground biomass of shrubs over time. This is helpful for studying the dynamic evolution of desert steppe ecosystems, such as the impact of climate change, human activities and other factors on shrub growth and biomass, and providing a scientific basis for ecological protection and management.

[0075] In terms of desert grassland resource survey, the satellite remote sensing estimation method for shrub biomass in desert grassland applied in this application can accurately estimate the shrub biomass, providing key data for understanding the vegetation resource status of desert grassland. These data can be used to evaluate indicators such as the ecological carrying capacity and vegetation coverage of desert grassland, which will help to rationally plan and utilize desert grassland resources and avoid over-exploitation and destruction.

[0076] In ecological and environmental assessment, shrub aboveground biomass is an important indicator. The aboveground biomass of shrubs per unit area obtained by the satellite remote sensing estimation method for desert steppe shrub aboveground biomass in this application can more comprehensively reflect the health status and ecological function of the desert steppe ecosystem. Through the analysis of biomass, changes in the stability and biodiversity of the ecosystem can be evaluated, providing scientific guidance for the protection and restoration of the ecological environment.

[0077] In terms of carbon content estimation, accurate shrub aboveground biomass data is the basis for carbon storage calculation. The satellite remote sensing estimation method for desert steppe shrub aboveground biomass in this application can provide reliable biomass estimation results, thereby improving the accuracy of desert steppe carbon content estimation. This is of great significance for studying the role of desert steppe in the global carbon cycle, evaluating its carbon sequestration capacity, and formulating strategies to address climate change.

[0078] In the above step S200 of the embodiment of the present application, Figure 2 As shown, the remote sensing data corresponding to the sample plot grid is obtained based on the satellite remote sensing data and the sample plot grid of the drone survey area, including:

[0079] S210, obtaining satellite remote sensing data of the area surveyed by the drone, wherein the spatial resolution of the remote sensing data is consistent with the grid size of the sample plot.

[0080] The spatial resolution of remote sensing data refers to the minimum ground distance or target size that satellite remote sensing images can distinguish. The higher the spatial resolution, the richer the details that can be displayed on the image, and the stronger the ability to identify ground objects.

[0081] In order to better match and combine satellite remote sensing data with observation data in the sample plot grid, the spatial resolution of the acquired satellite remote sensing data is required to be the same as the size of the sample plot grid. In this way, when analyzing satellite remote sensing data, each sample plot grid can be directly matched with the corresponding satellite image pixel, which is convenient for data comparison, verification and comprehensive analysis, thereby improving the accuracy and reliability of the research results. For example, if the size of the sample plot grid is 5m×5m, then it is necessary to obtain satellite remote sensing data with a spatial resolution of 5m so that the situation in each sample plot grid can be analyzed more accurately.

[0082] S220. Based on the coordinates of the center point of the sample plot grid, remote sensing data corresponding to the sample plot grid is obtained from the acquired remote sensing data.

[0083] When the coordinates of the center points of the sample plot grid are known, geographic information system (GIS) technology can be used to spatially match these coordinates with remote sensing data. The remote sensing data is presented in the form of images, and each pixel in the image corresponds to specific geographic coordinates and spectral information.

[0084] Specifically, it is to find the location range of the sample plot grid in the remote sensing image according to the coordinates of the center point, and then extract the remote sensing data of all pixel points within the sample plot grid. These data can include the spectral values ​​of different bands in the area. By analyzing and processing these spectral values, the feature information of the objects in the sample plot grid can be further obtained. For example, the health status of vegetation in the sample plot grid and the moisture of the soil can be judged by analyzing the reflectivity of different bands.

[0085] In the above step S300 of the embodiment of the present application, if Figure 3 As shown in FIG. 1 , the construction process of the relationship model between the aboveground biomass of shrubs per unit area and remote sensing data is as follows:

[0086] S310, constructing a training data set, wherein the training data set includes training samples, and the training samples include input features and output labels. The input features use remote sensing data corresponding to the sample plot grid, and the output labels use aboveground biomass per unit area in the grid.

[0087] You can also construct validation and test datasets as needed. The validation dataset is used to adjust model hyperparameters during training to avoid overfitting, while the test samples are used to evaluate the performance of the model after training.

[0088] S320, determining a loss function and an optimizer; wherein the loss function is used to measure the difference between the model prediction result and the output label, and the optimizer is used to adjust the model parameters to minimize the loss function;

[0089] S330, inputting the input features into the deep learning model to obtain a prediction result;

[0090] S340, calculate the loss value of the prediction result and the output label, calculate the gradient of the loss function to the model parameters through the back propagation algorithm, and the optimizer updates the model parameters according to the gradient to gradually reduce the loss value until the loss value converges or reaches the preset number of training rounds, thereby completing the supervised learning training of the deep learning model and obtaining the relationship model between the aboveground biomass of shrubs per unit area and the remote sensing data.

[0091] It should be noted that the constructed training data set can also be used to construct a relationship model between shrub biomass per unit area and remote sensing data based on a machine learning model. Specifically, the machine learning model can be a random forest model.

[0092] For example, based on the coordinates of the center point of the sample plot grid, the remote sensing data corresponding to 4500 sample plot grids are extracted from the acquired remote sensing data and used as sample data. The 4500 sample data are randomly divided into 8:2, that is, 80% of the sample data are used for model training and 20% of the sample data are used for model testing.

[0093] The relationship model between shrub aboveground biomass per unit area and remote sensing data is constructed based on machine learning models or deep learning models:

[0094] AGB e =G(b1,b2,…b n ) (1)

[0095] In formula (1), AGB e represents the aboveground biomass of shrubs per unit area; G represents a machine learning model or a deep learning model; b i Represents remote sensing data, which can be remote sensing reflectance value or vegetation index.

[0096] In the above step S310 of the embodiment of the present application, Figure 4 As shown, the process of obtaining the aboveground biomass per unit area in the grid includes:

[0097] S3110. Input the phenotypic structural parameters of the shrubs in the UAV sample plot into the composite model for estimating the aboveground biomass of the desert steppe shrubs, and calculate the aboveground biomass of each shrub in the UAV sample plot; wherein the phenotypic structural parameters of the shrubs include the crown area and height value of the shrubs.

[0098] S3120. According to the aboveground biomass of each shrub in the drone sample plot, the total aboveground biomass of the shrubs in each grid in the drone sample plot is obtained.

[0099] S3130. Calculate the aboveground biomass per unit area in each grid based on the total aboveground biomass of shrubs in each grid in the drone sample plot.

[0100] Specifically, in the above step S3110, if Figure 5 As shown, the process of obtaining the phenotypic structural parameters of the shrubs in the drone sample plot includes:

[0101] S31110. According to the spatial distribution range of desert steppe shrubs, several drone sampling plots were randomly set up.

[0102] S31120. Use the multispectral sensor carried by the drone to collect spectral images of the sample site, and use the lidar sensor carried by the drone to collect lidar data of the sample site.

[0103] Among them, laser radar (LiDAR) is an active optical remote sensing technology that obtains information by emitting laser pulses to the target area and receiving laser signals reflected from the target surface. In the survey plots of desert steppe shrubs, the laser radar equipment can be installed on drones, enabling it to cover the entire selected plot range for data collection. The laser pulse will be reflected when it encounters different parts of the shrubs, such as branches and trunks. The reflected light carries relevant information such as the spatial position and distance of the shrubs. The laser radar receiving device captures the reflected light and then records a large amount of point cloud data. These point cloud data contain relevant parameters of the laser reflected by different objects at various locations in the plot.

[0104] S31130. Perform image segmentation on the multispectral image of the desert steppe shrubs in the drone sample plot to obtain the crown area of ​​each shrub in the drone sample plot, including:

[0105] The multispectral images are stitched to obtain a stitched spectral image.

[0106] The stitched spectral images are input into the image segmentation model to identify the boundaries and feature information between the shrubs and the background objects.

[0107] The image segmentation model can adopt the SAM model (Segment Anything Model). According to the resolution, size and segmentation accuracy of the multispectral image of the UAV, the parameters such as the segmentation threshold and number of iterations of the SAM model can be appropriately adjusted to optimize the segmentation effect of the shrubs.

[0108] The spectral images used for segmentation and recognition are input into the SAM model one by one. Based on its powerful visual feature learning ability, the SAM model automatically identifies the boundaries and features of different objects in the image (including bushes and background objects, etc.) and generates corresponding segmentation results. The segmentation results can be presented in the form of annotated images or binary images.

[0109] In order to improve the accuracy of segmentation of each shrub, the segmentation results can be further optimized by using optimization methods based on morphological processing or optimization methods based on filtering technology.

[0110] According to the boundary and feature information of the shrubs and the background, the segmented areas corresponding to each shrub are marked. The segmented areas of each shrub can be marked and distinguished by different numbers, colors, etc.

[0111] The crown area of ​​each shrub was calculated based on the segmented area corresponding to each shrub and the spatial resolution of the image.

[0112] For each shrub segmentation area that has been marked, the number of pixels in the area is calculated using the pixel statistics method according to the spatial resolution of the multispectral image, and then multiplied by the actual area corresponding to each pixel to obtain the crown area of ​​each shrub. For example, if the spatial resolution of the image is 0.1m / pixel, and the number of pixels in the segmentation area of ​​a shrub is 1000, then the crown area of ​​the shrub is 0.1×0.1×1000=10m 2 .

[0113] It should be noted that in order to evaluate the accuracy of the segmentation results, a certain number of representative shrubs can be selected in the sample plot, and their true crown area can be obtained through field measurement, and the crown area can be used as a reference standard. The crown area calculated based on the SAM model segmentation is compared and analyzed with the field measurement value, and error indicators such as the root mean square error and mean absolute error are calculated, and the accuracy of the segmentation results is evaluated using the error indicators.

[0114] S31140. Invert the laser radar data of the desert steppe shrubs in the drone sample plot to obtain the height value of each shrub in the drone sample plot, including:

[0115] Preprocess the LiDAR data to obtain a subset of shrub point clouds, including:

[0116] The lidar data collected from different perspectives and at different times are unified into the same coordinate system, and the registration algorithm is used to stitch the lidar data together to form a complete sample point cloud data set;

[0117] The sample point cloud data set is denoised to remove isolated abnormal points (i.e., noise points) in the point cloud caused by factors such as instrument errors. Specifically, a statistical filtering method can be used to set reasonable distance thresholds and other parameters according to the spatial distribution characteristics of the point cloud, and points whose distance to the surrounding point clouds is greater than the preset distance threshold and points that do not conform to the overall distribution law are judged as noise points and removed.

[0118] The denoised sample point cloud dataset is separated from ground points and shrub points to obtain a shrub point cloud subset. Specifically, a morphological filtering algorithm can be used to distinguish ground points from shrub points by analyzing the morphological features of the terrain. The filtering operation obtains a point cloud subset containing only shrubs, so as to facilitate the subsequent inversion of the height of the shrubs.

[0119] Cluster analysis is performed on the shrub point cloud subset to obtain the range of each shrub in the shrub point cloud subset, including:

[0120] According to the spatial clustering characteristics of the point cloud, the point clouds of the same shrubs are clustered into the same category, so that each clustering result corresponds to a shrub, in order to divide the range of each shrub in the shrub point cloud subset. Specifically, the K-Means clustering algorithm based on spatial distance or the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering algorithm based on density can be used to perform cluster analysis on the separated shrub point cloud subset. It should be noted that if there are shrubs that are too densely grown and have adhesion or there are some misclassified point clouds, the clustered point cloud can be further optimized using morphological opening and closing operations and other processing methods to remove some small, unreasonable connected parts or fill holes.

[0121] Perform height inversion on the point cloud data of each shrub to obtain the height of each shrub, including:

[0122] The spatial coordinate query operation is used to obtain the maximum point coordinates and the minimum point coordinates of each shrub in the vertical direction of the point cloud, where the maximum point coordinates represent the top position of the shrub, and the minimum point coordinates represent the bottom position of the shrub;

[0123] By subtracting the coordinates of the minimum point from the coordinates of the maximum point, the height of each shrub in the survey plot is obtained, thus completing the inversion of the height of each shrub in the entire plot.

[0124] Specifically, in the above step S3110, the construction process of the composite model for estimating aboveground biomass of shrubs in desert steppe is as follows:

[0125] The phenotypic structural parameters of shrubs were fitted with the measured aboveground biomass of shrubs and the pseudo-measured aboveground biomass of shrubs to obtain a composite model for estimating aboveground biomass of shrubs in desert steppes.

[0126] Among them, the pseudo-measured shrub aboveground biomass was calculated by inputting the phenotypic structure parameters of the shrubs in the drone sample plot into the shrub aboveground biomass estimation model.

[0127] Furthermore, the construction process of the shrub aboveground biomass estimation model is as follows:

[0128] The phenotypic structural characteristics and aboveground biomass of each shrub in the survey plots set up in the desert steppe shrub area were collected;

[0129] According to the phenotypic structural characteristics and aboveground biomass of each shrub, with the phenotypic structural characteristics of the shrub as the independent variable and the aboveground biomass as the dependent variable, an aboveground biomass estimation model of shrubs was constructed by species, and the aboveground biomass estimation model of shrubs of the corresponding species was obtained.

[0130] The following is a specific example to illustrate in detail the method for estimating the aboveground biomass of desert steppe shrubs using satellite remote sensing provided in the embodiment of the present application.

[0131] Three 10m×10m survey plots were randomly set up in a desert steppe shrub area.

[0132] In three 10m×10m survey plots, the shrub information was investigated plant by plant, and the name, geographical location (latitude and longitude) and phenotypic structural parameters of the shrub were recorded. Among them, the phenotypic structural parameters can include phenotypic structural characteristics, coverage, dominant shrubs and plant numbers, and shrub ground diameter. Finally, the aboveground biomass of the shrubs was collected plant by plant within the sample plot.

[0133] When conducting a survey of shrubs one by one, you can refer to the "Technical Regulations for the Third National Land Survey" (TD / T 1055-2019). (1) Collect vegetation coverage in desert steppe shrub areas; (2) Record the dominant species and number of shrubs in the sample plot; (3) Measure the height of each shrub in centimeters, accurate to 0.1 cm; (4) Measure the north-south width and east-west width of the shrub crown in meters 2 , accurate to 0.01m 2 (5) Ground diameter: The diameter of the shrub close to the ground surface, in millimeters, accurate to 0.1 mm.

[0134] Specifically, in three 10m×10m survey plots, a total of 119 samples of four dominant shrub species (Tetraena mongolica, Rhizoma Cyperi, Rhizoma Cyperi and Rhizoma Cyperi) were collected, including the name (type), geographical location (latitude and longitude), phenotypic structural parameters and aboveground biomass of each shrub.

[0135] Seventeen 2km×2km drone sampling plots were randomly set up in the Sihemu Nature Reserve.

[0136] Use drones to collect shrub spectral images and lidar data. Specifically, the DJI M3 multispectral drone was used to collect shrub spectral data, and a total of red, green, and blue band data were obtained. The L2 sensor carried by the DJI M350RTK was used to obtain lidar data. The drone flew at an altitude of 80m, with projection parameters of the 2000 geodetic coordinate system and a spatial resolution of 5cm.

[0137] It should be noted that the spatial resolution determines the actual ground area represented by each pixel in the drone image. High resolution means that each pixel represents a smaller ground area, and low resolution means the opposite. For a 5cm spatial resolution, each pixel represents a 5cm×5cm ground area. In other words, for an image with a 5cm spatial resolution, each pixel in the image actually represents a 5cm×5cm square area.

[0138] Specifically, the images of each drone flight were stitched together in the DJI Zhitu software, and a total of 17 scenes of drone flight data were obtained.

[0139] According to the "Technical Regulations for the Third National Land Survey" (TD / T 1055-2019), the phenotypic structural parameters of shrubs were investigated, that is, random sampling survey experiments were conducted on shrubs in the drone sample plots, and the name, geographical location (latitude and longitude), shrub height, crown north-south width and east-west width, shrub ground diameter, etc. of each shrub were recorded.

[0140] Specifically, the basic information of 810 shrubs was randomly collected, and the 810 shrubs were divided into 17 types according to the shrub categories. The 17 shrubs were divided into 4 groups, belonging to Tetraselaki, Red Sand, Long-leaf Red Sand and Pearlwood.

[0141] Based on the aboveground biomass and phenotypic structural characteristics of the four dominant shrub species collected, estimation models of shrub surface structure parameters and aboveground biomass were established for the four dominant shrub types. The grouped shrub phenotypic structural characteristics were substituted into the aboveground biomass estimation models corresponding to Tetraena mume, Rhizophora chinensis, Rhizophora longifolia and Rhizophora chinensis, respectively, to obtain the pseudo-measured aboveground biomass of each group of shrubs.

[0142] Based on the measured and pseudo-measured aboveground biomass of shrubs and the phenotypic structural characteristics of shrubs, a composite model for estimating aboveground biomass was fitted for the Tetraena mugwort conservation area:

[0143] AGB=F(X) (2)

[0144] In formula (2), AGB represents the aboveground biomass of shrubs, F(…) represents the fitting function, and X represents the phenotypic structural characteristics of shrubs.

[0145] The drone images of the 17 sample plots were spliced ​​in DJI Maps, and the SAM image segmentation technology was used to segment the drone images to obtain the phenotypic structural parameters of all shrubs in the drone sample plots. The phenotypic structural parameters were input into the composite model for estimating aboveground biomass for the Tetrana mugwort conservation area (i.e., formula (2)) to calculate the aboveground biomass of each shrub.

[0146] Within the scope of the 17 drone sample plots, 30m×30m grids were established respectively. According to the boundaries of the drone sample plots, the grids outside the sample plot boundary and those intersecting with the sample plot boundary were eliminated, and only the grids completely within the sample plot boundary were retained, and a total of 4500 grids were obtained. The total aboveground biomass of shrubs in each grid was counted, and the aboveground biomass of shrubs per unit area in each grid was calculated.

[0147] AGB per =AGB sum / k 2 (3)

[0148] In formula (3), AGB per AGB represents the aboveground biomass of shrubs per unit area in each grid; sum represents the total aboveground biomass of shrubs in each grid; k represents the width or height of the grid, and in the embodiment of the present application, k is 30 m.

[0149] According to the grid size set in the UAV sample site, select remote sensing data that is consistent with the grid size. Specifically, Landsat 8 satellite remote sensing data with a spatial resolution of 30m can be selected. First, perform image preprocessing operations such as radiation calibration, atmospheric correction, geometric correction, and image enhancement on the satellite remote sensing data. Then, the normalized difference vegetation index (NDVI), difference vegetation index (DVI), enhanced vegetation index (EVI), soil adjusted vegetation index (SAVI), etc. are calculated based on the reflectivity of each band.

[0150] The corresponding remote sensing data is extracted according to the coordinates of the center point of each grid, and the remote sensing data is input into the trained relationship model between the aboveground biomass of shrubs per unit area and remote sensing data to calculate the following: Figure 6 The aboveground biomass of shrubs per unit area in the Tetraphyllum truncatum conservation area is shown. Figure 6Different colors are used to represent the aboveground biomass of shrubs per unit area, in grams (g / m 2 ). The legend shows how the colors correspond to the range of shrub aboveground biomass per unit area:

[0151] The red area represents the highest aboveground biomass per unit area, with a value of up to 305.621 g / m 2 The blue area represents the lowest aboveground biomass, with a value of 11.1932 g / m 2 Green and yellow areas represent intermediate levels of biomass. This color coding allows us to intuitively see the differences in biomass between different areas within the protected area.

[0152] from Figure 6 As can be seen in the figure, the distribution of shrub aboveground biomass is not uniform. The yellow and orange areas are relatively concentrated, indicating that the shrub aboveground biomass in these areas is high; while the blue and light blue areas are scattered everywhere, indicating that the shrub aboveground biomass in these places is low. Figure 6 The intuitive display of shrub aboveground biomass distribution can provide data support for related research and conservation decisions.

[0153] It should also be understood that the first, second, third, fourth and various numerical numbers involved in this document are only distinctions made for the convenience of description and are not intended to limit the scope of the present application.

[0154] It should be understood that the term "and / or" in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0155] In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in a processor or an instruction in the form of software. The steps of the method disclosed in conjunction with the embodiment of the present application can be directly embodied as a hardware processor for execution, or a combination of hardware and software modules in a processor for execution. The software module can be located in a storage medium mature in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in a memory, and the processor reads the information in the memory and completes the steps of the above method in conjunction with its hardware. To avoid repetition, it is not described in detail here.

[0156] In various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0157] Those skilled in the art will appreciate that the various illustrative logical blocks (ILBs) and steps described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or in a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0158] In the several embodiments provided in the present application, it should be understood that the disclosed method for estimating biomass on desert steppe shrub land by satellite remote sensing can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units (or modules) is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0159] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0160] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0161] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integration. The available medium can be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk), etc.

[0162] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for estimating aboveground biomass of shrubs in desert steppes by satellite remote sensing, characterized in that: Applications in the estimation of aboveground biomass of shrubs in desert steppes include: Obtain satellite remote sensing data from drone survey areas deployed in desert steppe shrub areas; Gridding the drone sample plots deployed in the drone survey area to obtain sample plot grids; According to the satellite remote sensing data and sample plot grid of the UAV survey area, the remote sensing data corresponding to the sample plot grid is obtained; The remote sensing data corresponding to the sample plot grids were input into the trained relationship model between the aboveground biomass of shrubs per unit area and remote sensing data to obtain the aboveground biomass of shrubs per unit area.

2. The method according to claim 1, characterized in that The remote sensing data corresponding to the sample plot grid is obtained based on the satellite remote sensing data of the drone survey area and the sample plot grid, including: Obtain satellite remote sensing data of the area surveyed by the drone, wherein the spatial resolution of the remote sensing data is consistent with the grid size of the sample plot; Based on the coordinates of the center point of the sample plot grid, the remote sensing data corresponding to the sample plot grid is obtained in the acquired remote sensing data.

3. The method according to claim 1, characterized in that The construction process of the relationship model between the aboveground biomass of shrubs per unit area and remote sensing data is as follows: Constructing a training data set, wherein the training data set includes training samples, the training samples include input features and output labels, the input features use remote sensing data corresponding to the sample plot grid, and the output labels use aboveground biomass per unit area in the grid; Determine a loss function and an optimizer; wherein the loss function is used to measure the difference between the model prediction result and the output label, and the optimizer is used to adjust the model parameters to minimize the loss function; Input features into the deep learning model to get prediction results; The loss value of the prediction result and the output label is calculated, and the gradient of the loss function to the model parameters is calculated by the back propagation algorithm. The optimizer updates the model parameters according to the gradient to gradually reduce the loss value until the loss value converges or reaches a preset number of training rounds, completing the supervised learning training of the deep learning model and obtaining the relationship model between the aboveground biomass of shrubs per unit area and remote sensing data.

4. The method according to claim 3, characterized in that The process of obtaining the aboveground biomass per unit area in the grid is as follows: The phenotypic structural parameters of the shrubs in the drone sample plot were input into the composite model for estimating the aboveground biomass of the desert steppe shrubs to calculate the aboveground biomass of each shrub in the drone sample plot; the phenotypic structural parameters of the shrubs included the crown area and height of the shrubs; The drone sample plots were gridded, and the total aboveground biomass of shrubs in each grid in the drone sample plots was obtained based on the aboveground biomass of each shrub in the drone sample plots; Based on the total aboveground biomass of shrubs in each grid in the drone sample plot, the aboveground biomass per unit area in each grid was calculated.

5. The method according to claim 4, characterized in that The process of obtaining the phenotypic structural parameters of the shrubs in the drone sample plot is as follows: Several drone sample plots were randomly set up according to the spatial distribution range of desert steppe shrubs; Use the multispectral sensor carried by the drone to collect multispectral images of the sample site, and use the lidar sensor carried by the drone to collect lidar data of the sample site; Perform image segmentation on the multispectral image to obtain the crown area of ​​each shrub; The LiDAR data was inverted to obtain the height of the shrubs.

6. The method according to claim 5, characterized in that The multispectral image is segmented to obtain the crown area of ​​each shrub, including: Stitching multispectral images to obtain a stitched spectral image; The spliced ​​spectral images are input into the image segmentation model to identify the boundaries and feature information of the shrubs and background objects; According to the boundary and feature information between the shrubs and the background objects, mark the segmented area corresponding to each shrub; The crown area of ​​each shrub was calculated based on the segmented area corresponding to each shrub and the spatial resolution of the image.

7. The method according to claim 5, characterized in that The inversion of the laser radar data to obtain the height value of the shrub includes: Preprocess the LiDAR data to obtain a subset of shrub point clouds, including: The lidar data collected from different perspectives and at different times are unified into the same coordinate system, and the registration algorithm is used to stitch the lidar data together to form a complete sample point cloud data set; De-noising the sample point cloud dataset; The denoised sample point cloud dataset is separated into ground points and shrub points to obtain a shrub point cloud subset. Cluster analysis is performed on the shrub point cloud subset to obtain the range of each shrub in the shrub point cloud subset, including: According to the spatial clustering characteristics of the point cloud, the point clouds of the same shrubs are clustered into the same category, so that each clustering result corresponds to a shrub, so as to divide the range of each shrub in the shrub point cloud subset; Perform height inversion on the point cloud data of each shrub to obtain the height of each shrub, including: The spatial coordinate query operation is used to obtain the maximum point coordinates and the minimum point coordinates of each shrub in the vertical direction of the point cloud, where the maximum point coordinates represent the top position of the shrub, and the minimum point coordinates represent the bottom position of the shrub; The height of each shrub in the survey plot was obtained by subtracting the coordinates of the minimum point from the coordinates of the maximum point.

8. The method according to claim 4, characterized in that The construction process of the composite model for estimating aboveground biomass of desert steppe shrubs is as follows: The composite model for estimating aboveground biomass of shrubs in desert steppes was obtained by fitting the phenotypic structural parameters of shrubs with the measured aboveground biomass of shrubs and the pseudo-measured aboveground biomass of shrubs. Among them, the pseudo-measured shrub aboveground biomass was calculated by inputting the phenotypic structure parameters of the shrubs in the drone sample plot into the shrub aboveground biomass estimation model.

9. The method according to claim 8, characterized in that The construction process of the shrub aboveground biomass estimation model is as follows: The phenotypic structural characteristics and aboveground biomass of each shrub in the survey plots set up in the desert steppe shrub area were collected; According to the phenotypic structural characteristics and aboveground biomass of each shrub, with the phenotypic structural characteristics of the shrub as the independent variable and the aboveground biomass as the dependent variable, an aboveground biomass estimation model of shrubs was constructed by species, and the aboveground biomass estimation model of shrubs of the corresponding species was obtained.

Images