A method for retrieving above-ground biomass of sonneratia caseolaris by using multi-source remote sensing data

Abstract

The application discloses a kind of using multi-source remote sensing data's Sonneratia apetala aboveground biomass inversion method, is realized by multi-source remote sensing data.First, plot investigation is carried out to obtain biomass measured data, and the height data of unmanned aerial vehicle laser radar scanning are acquired.Then, the height-biomass inversion model is constructed in combination with plot investigation results, biomass measured data and height data, and the biomass distribution of Sonneratia apetala in laser radar scanning area is estimated.Then, sentinel 2 and Landsat series images are collected, spectral features and age data are extracted, regression analysis is carried out, and the Sonneratia apetala biomass inversion model based on satellite remote sensing data is constructed.Finally, the Sonneratia apetala biomass in satellite remote sensing image range is estimated using the model and satellite remote sensing data.The application realizes efficient and high-precision Sonneratia apetala aboveground biomass inversion.

Description

Technical Field

[0001] This invention relates to the field of vegetation biomass estimation, and more specifically, to a method for retrieving aboveground biomass of *Sonneratia apetala* using multi-source remote sensing data. Background Technology

[0002] Mangroves are vegetation communities growing in tropical and subtropical marine-terrestrial interface areas, including trees and shrubs. They provide important ecological services such as carbon sequestration, wind and wave protection, and siltation and beach conservation, with a significantly higher carbon sequestration capacity per unit area than terrestrial forests. *Avicennia marina* (also known as *Avicennia apetalum*) is an excellent arboreal mangrove species. Due to its tolerance to flooding, rapid growth, and strong adaptability, it has been widely introduced to southern my country over the past few decades and is an important species for mangrove restoration and afforestation of tidal flats. At the same time, large-scale afforestation of *Avicennia marina* has accumulated a large amount of biomass. Accurately estimating the aboveground biomass of *Avicennia marina* is crucial for evaluating the effectiveness of past mangrove conservation and for future carbon inventory calculations. However, the restoration areas of *Avicennia marina* are spatially fragmented, exhibiting significant spatial heterogeneity. Furthermore, the growth areas of *Avicennia marina* are mostly located in intermittently flooded silty beaches. Traditional field surveys require manual entry into the forest to measure tree structural parameters, which is difficult and inefficient, making large-scale field surveys impractical. Therefore, it is necessary to make full use of drones and satellite remote sensing technology to assist ground surveys and quickly obtain structural information and growth status of trees on the ground through LiDAR and multispectral data.

[0003] Chinese invention application No. 202311577475.2, entitled "A method for assessing carbon storage in mangrove ecosystems based on remote sensing and DeepLab V3+", classifies mangrove populations based on remote sensing images and DeepLab models, and obtains carbon density data for each population by combining field survey results. The area of ​​each population is calculated and multiplied by the corresponding carbon density.

[0004] The Chinese invention application with application number 202311156020.3, entitled "A method for estimating aboveground carbon storage of mangrove plants in the marine-terrestrial ecotone", proposes a method that first classifies coastal land cover types, uses drone photography to help select samples for the classification model, and then constructs a biomass inversion model based on field quadrat survey data.

[0005] Chinese invention application No. 202010622479.8, entitled "A method for inverting mangrove biomass using aerial imagery and laser data", uses UAV imagery and LiDAR data to calculate spectral data, texture data and structural parameter data, and constructs a biomass model by combining it with biomass data from field quadrat surveys.

[0006] The aforementioned technical solutions primarily rely on correlating spectral, textural, and point cloud information from quadrat-scale field surveys with satellite and UAV imagery to directly construct biomass inversion models. Ground-based surveys, conducted manually, are time-consuming and labor-intensive, and compared to UAV surveys, cover a smaller area with the same time investment. Furthermore, the accumulation of mangrove biomass is closely related to its age, a factor that current technologies do not yet consider. Summary of the Invention

[0007] To address the low estimation efficiency of existing mangrove biomass estimation techniques, this invention provides a method for retrieving aboveground biomass of *Sonneratia apetala* using multi-source remote sensing data. The technical solution adopted by this invention is as follows:

[0008] This invention provides a method for retrieving aboveground biomass of lobeless *Sonneratia africana* using multi-source remote sensing data, the method comprising:

[0009] A quadrat survey was conducted in the pre-defined quadrat area, and biomass was calculated based on the quadrat survey results. The calculated results were used as the actual biomass data for fitting.

[0010] The preset UAV lidar scanning area is detected, and tree height data based on 3D point cloud data inversion is obtained according to the detection results, wherein the lidar scanning area covers the sample area.

[0011] A tree height-biomass inversion model is constructed using the quadrat survey results, the measured biomass data used for fitting, and the tree height data inverted from the three-dimensional point cloud data. The tree height-biomass inversion model is then used to obtain the biomass distribution of all lobeless Sorbus aegyptiaca within the lidar scanning area.

[0012] Collect Sentinel-2 remote sensing images and Landsat series images covering the lidar scanning area, and extract features from the Sentinel-2 remote sensing images and Landsat series images respectively to obtain the spectral features and tree age data corresponding to the lidar scanning area.

[0013] Regression analysis was performed on the biomass distribution, spectral characteristics and age data of all Avicennia marinae in the lidar scanning area to construct a biomass inversion model of Avicennia marinae based on satellite remote sensing data.

[0014] The biomass of Avicennia marina based on satellite remote sensing data is used to estimate the biomass of Avicennia marina within the range of satellite remote sensing images.

[0015] As a preferred approach, a method for conducting quadrat surveys in a pre-defined quadrat area, calculating biomass based on the survey results, and using the calculated results as the actual biomass data for fitting includes:

[0016] A quadrat survey was conducted in the pre-defined quadrat area to obtain the measured tree height, diameter at breast height (DBH), and latitude and longitude coordinates of each Avicennia marina abstemii tree within the pre-defined quadrat area.

[0017] The measured tree height and diameter at breast height (DBH) of each Avicennia marina in the quadrat area were substituted into the preset Avicennia marina allometric growth equation to calculate the biomass of each Avicennia marina in the quadrat area. This calculation result was used as the measured biomass data for fitting the model. The specific calculation formula is as follows:

[0018] W = a(D) 2 H) b

[0019] Where W is biomass, D is diameter at breast height (DBH), H is tree height, and a and b are constants.

[0020] As a preferred approach, methods for detecting a pre-defined UAV lidar scanning area and obtaining tree height data based on 3D point cloud data inversion include:

[0021] By using drones equipped with LiDAR sensors to detect the preset drone lidar scanning area, three-dimensional point cloud data is obtained.

[0022] Based on the three-dimensional point cloud data, a canopy height model corresponding to the lidar scanning area is obtained;

[0023] The canopy height model is processed by segmenting individual trees using a preset watershed segmentation algorithm, and the highest point within the canopy range of each petalless Sinica tree is taken as the tree height, thus obtaining tree height data based on the inversion of three-dimensional point cloud data.

[0024] As a preferred embodiment, the method for obtaining the canopy height model corresponding to the lidar scanning area based on the three-dimensional point cloud data includes:

[0025] The 3D point cloud data is input into LiDAR360 software for point cloud denoising and ground point classification, resulting in a classified ground point set and a vegetation point cloud set.

[0026] Digital elevation models and digital surface models are generated based on the classified ground point sets and vegetation point cloud sets, respectively.

[0027] Subtracting the digital elevation model from the digital surface model yields the canopy height model corresponding to the lidar scanning area.

[0028] As a preferred approach, methods for generating digital elevation models and digital surface models based on classified ground point sets and vegetation point cloud sets, respectively, include:

[0029] Based on the classified ground point set, a digital elevation model is generated using the irregular triangular mesh interpolation method;

[0030] A digital surface model is generated based on a pre-classified set of vegetation point clouds using an inverse distance weighted interpolation method.

[0031] As a preferred embodiment, the method for constructing a tree height-biomass inversion model using the quadrat survey results, the measured biomass data used for fitting, and the tree height data inverted from the three-dimensional point cloud data, and obtaining the biomass distribution of all lobeless Sorbus aegypti within the lidar scanning area using the tree height-biomass inversion model, includes:

[0032] Based on the latitude and longitude coordinates of each Avicennia marina in the quadrat survey results, the measured biomass data used for fitting is matched one by one with the tree height data derived from the three-dimensional point cloud data.

[0033] A biomass inversion model based on lidar data for retrieving tree height was constructed using the least squares method on the matched biomass measurement data and tree height data.

[0034] The tree height-biomass inversion model based on lidar data was verified and calibrated using the measured tree height data, measured diameter at breast height data, and latitude and longitude coordinates, and finally the tree height-biomass inversion model was obtained.

[0035] The tree height data derived from the 3D point cloud data is input into the tree height-biomass inversion model to obtain the biomass distribution of all lobeless Sorbus aegyptiaca within the lidar scanning area.

[0036] As a preferred approach, methods for constructing a biomass inversion model based on lidar data to retrieve tree height using the least squares method on well-matched measured biomass data and tree height data include:

[0037] Based on the linear relationship between the diameter at breast height (DBH) and tree height of *Avicennia marina* during a certain growth period, and considering the allometric growth equation of *Avicennia marina*, a correlation formula between the biomass and tree height of *Avicennia marina* was constructed, and this nonlinear polynomial was fitted in MATLAB:

[0038] W=(aH 3 +bH 2 +cH) d

[0039] Where W is biomass, H is tree height, and a, b, c, and d are constants.

[0040] As a preferred embodiment, the method for collecting Sentinel-2 remote sensing images covering the lidar scanning area and extracting features from the Sentinel-2 remote sensing images to obtain the spectral features corresponding to the lidar scanning area includes:

[0041] Sentinel-2 remote sensing images covering the lidar scanning area were collected and preprocessed as follows:

[0042] Radiometric calibration, spatial registration, and atmospheric correction;

[0043] The preprocessed Sentinel-2 remote sensing image was resampled, and the surface reflectance data was obtained based on the resampling results.

[0044] The spectral characteristics within the range of the Sentinel-2 remote sensing image were obtained by performing remote sensing feature calculations on the surface reflectance data.

[0045] The spectral characteristics of the lidar scanning area are obtained based on the spectral characteristics within the range of the Sentinel-2 remote sensing image.

[0046] As a preferred embodiment, the method for collecting Landsat series images covering the lidar scanning area and extracting features from the Landsat series images to obtain tree age data corresponding to the lidar scanning area includes:

[0047] Collect a long-term Landsat image set covering the LiDAR scanning area and perform the following preprocessing:

[0048] Radiation correction and cloud removal;

[0049] NDVI time series data is constructed by calculating the NDVI value of each pixel in the preprocessed Landsat long-term image set;

[0050] Based on the NDVI time series data, the long-term dynamic changes of pixels are obtained, and the tree age of each pixel is calculated according to the changes, to obtain the tree age data corresponding to the Landsat series image range, specifically:

[0051] The NDVI time series data is processed by moving average. The year when the NDVI value after moving average processing is higher than a preset threshold is taken as the starting year for calculating tree age. The distance between the current time and the tree age is taken as the tree age. When there is a continuous sharp drop in NDVI, the starting time point of tree age is recalculated.

[0052] The tree age data corresponding to the LiDAR scanning area is obtained based on the tree age data corresponding to the range of the Landsat series images.

[0053] As a preferred approach, a method for constructing a biomass inversion model of Avicennia marina based on satellite remote sensing data involves performing regression analysis on the biomass distribution, spectral characteristics, and tree age data of all Avicennia marina species corresponding to the lidar scanning area.

[0054] First, the biomass distribution of all lobeless Sorrelia biomass corresponding to the lidar scanning area is statistically analyzed using a grid according to the spatial resolution of the Sentinel-2 remote sensing image.

[0055] Then, the biomass distribution, spectral characteristics, and tree age data of all Avicennia marina species corresponding to the lidar scanning area were input into a preset random forest model for regression analysis to construct an Avicennia marina biomass inversion model based on satellite remote sensing data.

[0056] Compared with the prior art, the beneficial effects of this invention are:

[0057] This invention utilizes a lidar mounted on a drone to rapidly survey the biomass of lobeless mangroves in localized areas, ignoring terrain obstacles. Compared to traditional survey methods, this significantly improves efficiency and yields more detailed survey information.

[0058] This invention utilizes multi-source remote sensing data (UAV-mounted lidar and multi-source satellite remote sensing data) to fuse multi-dimensional spatiotemporal remote sensing information for biomass inversion, thereby improving inversion accuracy. Each data source has its own advantages: lidar data can acquire high-precision three-dimensional structural information of ground features; Sentinel-2 satellite imagery has high spatial resolution and rich spectral information; and Landsat series satellite imagery provides the longest continuous temporal data, recording the long-term dynamic changes of ground features. Compared to constructing a biomass inversion model using only spectral information, incorporating tree age information can characterize the cumulative effect of tree height and biomass, further enhancing the effectiveness of regional-scale biomass inversion. Attached Figure Description

[0059] Figure 1 This embodiment provides a flowchart of a method for retrieving aboveground biomass of *Sonneratia apetala* using multi-source remote sensing data.

[0060] Figure 2 This is a diagram showing the canopy height model results provided in this embodiment;

[0061] Figure 3 This is a partial single-tree segmentation result diagram provided in this embodiment;

[0062] Figure 4 The aboveground biomass distribution of lobeless mangrove based on UAV lidar data provided in this embodiment;

[0063] Figure 5 This embodiment provides some spectral index features;

[0064] Figure 6 This is a diagram showing the tree age inversion results provided in this embodiment;

[0065] Figure 7 The image shows the results of the biomass inversion of the lobeless Avicennia marina region provided in this embodiment. Detailed Implementation

[0066] The accompanying drawings are for illustrative purposes only and should not be construed as limiting the invention.

[0067] It should be understood that the described embodiments are merely some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of the embodiments of this application.

[0068] The terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the embodiments of this application. The singular forms “a,” “the,” and “the” used in the embodiments of this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0069] In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims. In the description of this application, it should be understood that the terms "first," "second," "third," etc., are used only to distinguish similar objects and are not necessarily used to describe a specific order or sequence, nor should they be construed as indicating or implying relative importance. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0070] Furthermore, in the description of this application, unless otherwise stated, "multiple" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. The invention will be further described below with reference to the accompanying drawings and embodiments.

[0071] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0072] Example 1

[0073] Please refer to Figure 1 This embodiment provides a method for retrieving aboveground biomass of lobeless *Sonneratia alopecuroides* using multi-source remote sensing data. The method includes:

[0074] S1: Conduct quadrat surveys in the pre-defined quadrat areas, calculate biomass based on the quadrat survey results, and use the calculated results as the actual biomass data for fitting.

[0075] In one specific embodiment, a method for conducting quadrat surveys in a pre-defined quadrat area, calculating biomass based on the quadrat survey results, and using the calculated results as the measured biomass data for fitting includes:

[0076] A quadrat survey was conducted in the pre-defined quadrat area to obtain the measured tree height, diameter at breast height (DBH), and latitude and longitude coordinates of each Avicennia marina abstemii tree in the pre-defined quadrat area.

[0077] The measured tree height and diameter at breast height (DBH) of each Avicennia marina in the quadrat area were substituted into the preset Avicennia marina allometric growth equation to calculate the biomass of each Avicennia marina in the quadrat area. This calculation result was used as the measured biomass data for fitting the model. The specific calculation formula is as follows:

[0078] W = a(D) 2 H) b

[0079] Where W is biomass, D is diameter at breast height (DBH), H is tree height, and a and b are constants.

[0080] S2: Detect the preset UAV lidar scanning area and obtain tree height data based on the detection results inverted from 3D point cloud data, wherein the lidar scanning area covers the sample area;

[0081] In one specific embodiment, the method for detecting a preset UAV lidar scanning area and obtaining tree height data based on 3D point cloud data inversion according to the detection results includes:

[0082] By using drones equipped with LiDAR sensors to detect the preset drone lidar scanning area, three-dimensional point cloud data is obtained.

[0083] Based on the three-dimensional point cloud data, a canopy height model corresponding to the lidar scanning area is obtained;

[0084] The canopy height model is processed by segmenting individual trees using a preset watershed segmentation algorithm, and the highest point within the canopy range of each petalless Sinica tree is taken as the tree height, thus obtaining tree height data based on the inversion of three-dimensional point cloud data.

[0085] In one specific embodiment, the method for obtaining the canopy height model corresponding to the lidar scanning area based on the three-dimensional point cloud data includes:

[0086] The 3D point cloud data is input into LiDAR360 software for point cloud denoising and ground point classification, resulting in a classified ground point set and a vegetation point cloud set.

[0087] Digital elevation models and digital surface models are generated based on the classified ground point sets and vegetation point cloud sets, respectively.

[0088] Subtracting the digital elevation model from the digital surface model yields the canopy height model corresponding to the lidar scanning area.

[0089] In one specific embodiment, the method for generating a digital elevation model and a digital surface model based on a classified set of ground points and a set of vegetation point clouds, respectively, includes:

[0090] Based on the classified ground point set, a digital elevation model is generated using the irregular triangular mesh interpolation method;

[0091] A digital surface model is generated based on a pre-classified set of vegetation point clouds using an inverse distance weighted interpolation method.

[0092] S3: Construct a tree height-biomass inversion model using the quadrat survey results, the measured biomass data used for fitting, and the tree height data inverted based on three-dimensional point cloud data, and obtain the biomass distribution of all lobeless Sorbus aegyptiaca within the lidar scanning area using the tree height-biomass inversion model;

[0093] In a specific embodiment, the method for constructing a tree height-biomass inversion model using the quadrat survey results, the measured biomass data used for fitting, and the tree height data inverted based on 3D point cloud data, and obtaining the biomass distribution of all lobeless Avicennia marina within the lidar scanning area using the tree height-biomass inversion model, includes:

[0094] Based on the latitude and longitude coordinates of each Avicennia marina tree in the quadrat survey results, the measured biomass data used for fitting is matched one by one with the tree height data derived from the three-dimensional point cloud data.

[0095] A biomass inversion model based on lidar data for retrieving tree height was constructed using the least squares method on the matched biomass measurement data and tree height data.

[0096] The tree height-biomass inversion model based on lidar data was verified and calibrated using the measured tree height data, measured diameter at breast height data, and latitude and longitude coordinates, and finally the tree height-biomass inversion model was obtained.

[0097] The tree height data derived from the 3D point cloud data is input into the tree height-biomass inversion model to obtain the biomass distribution of all lobeless Sorbus aegyptiaca within the lidar scanning area.

[0098] In a specific embodiment, the method for constructing a biomass inversion model based on lidar data to invert tree height using the least squares method on matched biomass measurement data and tree height data includes:

[0099] Based on the linear relationship between the diameter at breast height (DBH) and tree height of *Avicennia marina* during a certain growth period, and considering the allometric growth equation of *Avicennia marina*, a correlation formula between the biomass and tree height of *Avicennia marina* was constructed, and this nonlinear polynomial was fitted in MATLAB:

[0100] W=(aH 3 +bH 2 +cH) d

[0101] Where W is biomass, H is tree height, and a, b, c, and d are constants.

[0102] S4: Collect Sentinel-2 remote sensing images and Landsat series images covering the lidar scanning area, extract features from Sentinel-2 remote sensing images and Landsat series images respectively, and obtain spectral features and tree age data corresponding to the lidar scanning area.

[0103] In one specific embodiment, the method for collecting Sentinel-2 remote sensing images covering the lidar scanning area and extracting features from the Sentinel-2 remote sensing images to obtain the spectral features corresponding to the lidar scanning area includes:

[0104] Sentinel-2 remote sensing images covering the lidar scanning area were collected and preprocessed as follows:

[0105] Radiometric calibration, spatial registration, and atmospheric correction;

[0106] The preprocessed Sentinel-2 remote sensing image was resampled, and the surface reflectance data was obtained based on the resampling results.

[0107] The spectral characteristics within the range of the Sentinel-2 remote sensing image were obtained by performing remote sensing feature calculations on the surface reflectance data.

[0108] The spectral characteristics of the lidar scanning area are obtained based on the spectral characteristics within the range of the Sentinel-2 remote sensing image.

[0109] In one specific embodiment, the method for collecting Landsat series images covering the lidar scanning area and extracting features from the Landsat series images to obtain tree age data corresponding to the lidar scanning area includes:

[0110] Collect a long-term Landsat image set covering the LiDAR scanning area and perform the following preprocessing:

[0111] Radiation correction and cloud removal;

[0112] NDVI time series data is constructed by calculating the NDVI value of each pixel in the preprocessed Landsat long-term image set;

[0113] Based on the NDVI time series data, the long-term dynamic changes of pixels are obtained, and the tree age of each pixel is calculated according to the changes, to obtain the tree age data corresponding to the Landsat series image range, specifically:

[0114] The NDVI time series data is processed by moving average. The year when the NDVI value after moving average processing is higher than a preset threshold is taken as the starting year for calculating tree age. The distance between the current time and the tree age is taken as the tree age. When there is a continuous sharp drop in NDVI, the starting time point of tree age is recalculated.

[0115] The tree age data corresponding to the LiDAR scanning area is obtained based on the tree age data corresponding to the range of the Landsat series images.

[0116] S5: Perform regression analysis on the biomass distribution, spectral characteristics and age data of all Avicennia marinae corresponding to the lidar scanning area, and construct an Avicennia marinae biomass inversion model based on satellite remote sensing data;

[0117] In a specific embodiment, the method for constructing a biomass inversion model of Avicennia marina based on satellite remote sensing data by performing regression analysis on the biomass distribution, spectral characteristics, and tree age data of all Avicennia marina aphyllum trees corresponding to the lidar scanning area includes:

[0118] First, the biomass distribution of all lobeless Sorrelia biomass corresponding to the lidar scanning area is statistically analyzed using a grid according to the spatial resolution of the Sentinel-2 remote sensing image.

[0119] Then, the biomass distribution, spectral characteristics, and tree age data of all Avicennia marina species corresponding to the lidar scanning area were input into a preset random forest model for regression analysis to construct an Avicennia marina biomass inversion model based on satellite remote sensing data.

[0120] S6: Estimate the biomass of Avicennia marina within the satellite remote sensing image range using the aforementioned biomass inversion model based on satellite remote sensing data.

[0121] Example 2

[0122] Please refer to Figure 1 This embodiment provides a method for retrieving aboveground biomass of lobeless *Sonneratia alopecuroides* using multi-source remote sensing data. The method includes:

[0123] S1: Conduct quadrat surveys in the pre-defined quadrat areas, calculate biomass based on the quadrat survey results, and use the calculated results as the actual biomass data for fitting.

[0124] In one specific embodiment, a method for conducting quadrat surveys in a pre-defined quadrat area, calculating biomass based on the quadrat survey results, and using the calculated results as the measured biomass data for fitting includes:

[0125] A quadrat survey was conducted in the pre-defined quadrat area to obtain the measured tree height, diameter at breast height (DBH), and latitude and longitude coordinates of each Avicennia marina abstemii tree in the pre-defined quadrat area.

[0126] The measured tree height and diameter at breast height (DBH) of each Avicennia marina in the quadrat area were substituted into the preset Avicennia marina allometric growth equation to calculate the biomass of each Avicennia marina in the quadrat area. This calculation result was used as the measured biomass data for fitting the model. The specific calculation formula is as follows:

[0127] W = a(D) 2 H) b

[0128] Where W is biomass, D is diameter at breast height (DBH), H is tree height, and a and b are constants.

[0129] Specifically, based on the acquired dataset of the growth range of *Avicennia marina*, the survey area and location were determined, and further, aerial photography areas were selected using Google Earth and remote sensing imagery. Considering the growth density of *Avicennia marina* forests and the convenience of ground surveys, areas with different tree densities were selected, and several field measurement points were set up. The tree height and diameter at breast height of *Avicennia marina* were investigated using a 10*10 meter quadrat scale.

[0130] S2: Detect the preset UAV lidar scanning area and obtain tree height data based on the detection results inverted from 3D point cloud data, wherein the lidar scanning area covers the sample area;

[0131] In one specific embodiment, the method for detecting a preset UAV lidar scanning area and obtaining tree height data based on 3D point cloud data inversion according to the detection results includes:

[0132] By using drones equipped with LiDAR sensors to detect the preset drone lidar scanning area, three-dimensional point cloud data is obtained.

[0133] Specifically, under favorable weather and wind conditions, and at low tide (when the tidal flats are mostly exposed), a drone equipped with a LiDAR sensor was used to collect lobe-less lidar data of the Kalanchoe blakeana near the sample plot area. The drone flew at an altitude of approximately 100m, shooting vertically downwards, with a lateral and forward overlap of more than 75%.

[0134] Based on the three-dimensional point cloud data, a canopy height model corresponding to the LiDAR scanning area is obtained. The canopy height model is as follows: Figure 2 As shown;

[0135] Please refer to Figure 3 The canopy height model is processed by single-tree segmentation using a preset watershed segmentation algorithm, and the highest point within the canopy range of each petalless sago palm is taken as the tree height, thus obtaining tree height data based on the inversion of three-dimensional point cloud data.

[0136] Specifically, a watershed segmentation algorithm is used to further segment individual trees based on the canopy height model. The resolution, minimum and maximum values, and Gaussian smoothing parameters should be adjusted appropriately according to the different survey subjects to achieve better segmentation results. Here, the minimum tree height is set to 3 meters, and the Gaussian smoothing factor is 1.5. The highest point within the canopy is used to represent the tree height, and the tree height information for each tree is statistically analyzed.

[0137] In one specific embodiment, the method for obtaining the canopy height model corresponding to the lidar scanning area based on the three-dimensional point cloud data includes:

[0138] The 3D point cloud data is input into LiDAR360 software for point cloud denoising and ground point classification, resulting in a classified ground point set and a vegetation point cloud set.

[0139] Digital elevation models and digital surface models are generated based on the classified ground point sets and vegetation point cloud sets, respectively.

[0140] Subtracting the digital elevation model from the digital surface model yields the canopy height model corresponding to the lidar scanning area.

[0141] In one specific embodiment, the method for generating a digital elevation model and a digital surface model based on a classified set of ground points and a set of vegetation point clouds, respectively, includes:

[0142] Based on the classified ground point set, a digital elevation model is generated using the irregular triangular mesh interpolation method;

[0143] A digital surface model with a spatial resolution of 0.5m is generated using an inverse distance weighted interpolation method based on a classified set of vegetation point clouds.

[0144] S3: Construct a tree height-biomass inversion model using the quadrat survey results, the measured biomass data used for fitting, and the tree height data inverted based on three-dimensional point cloud data, and obtain the biomass distribution of all lobeless Sorbus aegyptiaca within the lidar scanning area using the tree height-biomass inversion model;

[0145] In a specific embodiment, the method for constructing a tree height-biomass inversion model using the quadrat survey results, the measured biomass data used for fitting, and the tree height data inverted based on 3D point cloud data, and obtaining the biomass distribution of all lobeless Avicennia marina within the lidar scanning area using the tree height-biomass inversion model, includes:

[0146] Based on the latitude and longitude coordinates of each Avicennia marina tree in the quadrat survey results, the measured biomass data used for fitting is matched one by one with the tree height data derived from the three-dimensional point cloud data.

[0147] A biomass inversion model based on lidar data for retrieving tree height was constructed using the least squares method on the matched biomass measurement data and tree height data.

[0148] The tree height-biomass inversion model based on lidar data was verified and calibrated using the measured tree height data, measured diameter at breast height data, and latitude and longitude coordinates, and finally the tree height-biomass inversion model was obtained.

[0149] The tree height data derived from the 3D point cloud data is input into the tree height-biomass inversion model to obtain the biomass distribution of all lobeless Sorbus aegyptiaca within the lidar scanning area.

[0150] In a specific embodiment, the method for constructing a biomass inversion model based on lidar data to invert tree height using the least squares method on matched biomass measurement data and tree height data includes:

[0151] It should be noted that, according to research, the diameter at breast height (DBH) of the apetalous saffron tree has a good linear relationship with its height over a certain growth period.

[0152] Based on the linear relationship between the diameter at breast height (DBH) and tree height of *Avicennia marina* during a certain growth period, and considering the allometric growth equation of *Avicennia marina*, a correlation formula between the biomass and tree height of *Avicennia marina* was constructed, and this nonlinear polynomial was fitted in MATLAB:

[0153] W=(aH 3 +bH 2 +cH) d

[0154] Where W is biomass, H is tree height, and a, b, c, and d are constants.

[0155] S4: Collect Sentinel-2 remote sensing images and Landsat series images covering the lidar scanning area, extract features from Sentinel-2 remote sensing images and Landsat series images respectively, and obtain spectral features and tree age data corresponding to the lidar scanning area.

[0156] In one specific embodiment, please refer to Figure 5 The method for collecting Sentinel-2 remote sensing images covering the lidar scanning area and extracting features from the Sentinel-2 remote sensing images to obtain the spectral features corresponding to the lidar scanning area includes:

[0157] Sentinel-2 remote sensing images covering the lidar scanning area were collected and preprocessed as follows:

[0158] Radiometric calibration, spatial registration, and atmospheric correction;

[0159] The preprocessed Sentinel-2 remote sensing image was resampled, and the surface reflectance data was obtained based on the resampling results.

[0160] The spectral characteristics within the range of the Sentinel-2 remote sensing image were obtained by performing remote sensing feature calculations on the surface reflectance data.

[0161] The spectral characteristics of the lidar scanning area are obtained based on the spectral characteristics within the range of the Sentinel-2 remote sensing image.

[0162] In one specific embodiment, please refer to Figure 6 The method for collecting Landsat series images covering the lidar scanning area, extracting features from the Landsat series images, and obtaining tree age data corresponding to the lidar scanning area includes:

[0163] Collect a long-term Landsat image set covering the LiDAR scanning area and perform the following preprocessing:

[0164] Radiation correction and cloud removal;

[0165] NDVI time series data is constructed by calculating the NDVI value of each pixel in the preprocessed Landsat long-term image set;

[0166] Based on the NDVI time series data, the long-term dynamic changes of pixels are obtained, and the tree age of each pixel is calculated according to the changes, to obtain the tree age data corresponding to the Landsat series image range, specifically:

[0167] The NDVI time series data is processed by moving average. The year when the NDVI value after moving average processing is higher than a preset threshold is taken as the starting year for calculating tree age. The distance between the current time and the tree age is taken as the tree age. When there is a continuous sharp drop in NDVI, the starting time point of tree age is recalculated.

[0168] The tree age data corresponding to the LiDAR scanning area is obtained based on the tree age data corresponding to the range of the Landsat series images.

[0169] S5: Perform regression analysis on the biomass distribution, spectral characteristics and age data of all Avicennia marinae corresponding to the lidar scanning area, and construct an Avicennia marinae biomass inversion model based on satellite remote sensing data;

[0170] In a specific embodiment, the method for constructing a biomass inversion model of Avicennia marina based on satellite remote sensing data by performing regression analysis on the biomass distribution, spectral characteristics, and tree age data of all Avicennia marina aphyllum trees corresponding to the lidar scanning area includes:

[0171] Please refer to Figure 4 First, the biomass distribution of all lobeless Sorrelia biomass corresponding to the LiDAR scanning area is statistically analyzed using a grid according to the spatial resolution of the Sentinel-2 remote sensing image.

[0172] Specifically, the biomass distribution of all lobeless mangroves corresponding to the lidar scanning area is statistically analyzed using a grid with a spatial resolution of 10*10 meters.

[0173] Then, the biomass distribution, spectral characteristics, and tree age data of all Avicennia marina species corresponding to the lidar scanning area were input into a preset random forest model for regression analysis to construct an Avicennia marina biomass inversion model based on satellite remote sensing data.

[0174] S6: Please refer to Figure 7 The biomass of Avicennia marina, based on satellite remote sensing data, is estimated within the satellite remote sensing image range using the aforementioned biomass inversion model for Avicennia marina.

[0175] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A method for retrieving aboveground biomass of *Avicennia marina* without petals using multi-source remote sensing data, characterized in that, The method includes: A quadrat survey was conducted in the pre-defined quadrat area, and biomass was calculated based on the quadrat survey results. The calculated results were used as the actual biomass data for fitting. The preset UAV lidar scanning area is detected, and tree height data based on 3D point cloud data inversion is obtained according to the detection results, wherein the lidar scanning area covers the sample area. A tree height-biomass inversion model is constructed using the quadrat survey results, the measured biomass data used for fitting, and the tree height data inverted from the three-dimensional point cloud data. The tree height-biomass inversion model is then used to obtain the biomass distribution of all lobeless Sorbus aegyptiaca within the lidar scanning area. Collect Sentinel-2 remote sensing images and Landsat series images covering the lidar scanning area, and extract features from the Sentinel-2 remote sensing images and Landsat series images respectively to obtain the spectral features and tree age data corresponding to the lidar scanning area. Regression analysis was performed on the biomass distribution, spectral characteristics and age data of all Avicennia marinae in the lidar scanning area to construct a biomass inversion model of Avicennia marinae based on satellite remote sensing data. The biomass of Avicennia marina based on satellite remote sensing data is used to estimate the biomass of Avicennia marina within the range of satellite remote sensing images.

2. The method for retrieving aboveground biomass of *Sonneratia alopecuroides* using multi-source remote sensing data according to claim 1, characterized in that, Methods for conducting quadrat surveys in pre-defined quadrat areas, calculating biomass based on the survey results, and using the calculated results as the actual biomass data for fitting include: A quadrat survey was conducted in the pre-defined quadrat area to obtain the measured tree height, diameter at breast height (DBH), and latitude and longitude coordinates of each Avicennia marina abstemii tree in the pre-defined quadrat area. The measured tree height and diameter at breast height (DBH) of each Avicennia marina in the quadrat area were substituted into the preset Avicennia marina allometric growth equation to calculate the biomass of each Avicennia marina in the quadrat area. This calculation result was used as the measured biomass data for fitting the model. The specific calculation formula is as follows: W=a(D 2 H) b Where W is biomass, D is diameter at breast height (DBH), H is tree height, and a and b are constants.

3. The method for retrieving aboveground biomass of *Sonneratia apetala* using multi-source remote sensing data according to claim 1, characterized in that... Methods for detecting a pre-defined UAV lidar scanning area and obtaining tree height data based on 3D point cloud data inversion include: By using drones equipped with LiDAR sensors to detect the preset drone lidar scanning area, three-dimensional point cloud data is obtained. Based on the three-dimensional point cloud data, a canopy height model corresponding to the lidar scanning area is obtained; The canopy height model is processed by segmenting individual trees using a preset watershed segmentation algorithm, and the highest point within the canopy range of each petalless Sinica tree is taken as the tree height, thus obtaining tree height data based on the inversion of three-dimensional point cloud data.

4. The method for retrieving aboveground biomass of *Sonneratia apetala* using multi-source remote sensing data according to claim 3, characterized in that, The method for obtaining the canopy height model corresponding to the lidar scanning area based on the three-dimensional point cloud data includes: The three-dimensional point cloud data is input into LiDAR360 software to perform point cloud denoising and ground point classification in sequence, resulting in a classified ground point set and a vegetation point cloud set. Digital elevation models and digital surface models are generated based on the classified ground point sets and vegetation point cloud sets, respectively. Subtracting the digital elevation model from the digital surface model yields the canopy height model corresponding to the lidar scanning area.

5. The method for retrieving aboveground biomass of *Sonneratia apetala* using multi-source remote sensing data according to claim 4, characterized in that... Methods for generating digital elevation models and digital surface models based on classified ground point sets and vegetation point cloud sets, respectively, include: Based on the classified ground point set, a digital elevation model is generated using the irregular triangular mesh interpolation method; A digital surface model is generated based on a pre-classified set of vegetation point clouds using an inverse distance weighted interpolation method.

6. The method for retrieving aboveground biomass of lobeless *Avicennia marina* using multi-source remote sensing data according to claim 2, characterized in that, The method for constructing a tree height-biomass inversion model using the quadrat survey results, the measured biomass data used for fitting, and the tree height data inverted from 3D point cloud data, and obtaining the biomass distribution of all lobeless Sorbus aegypti within the lidar scanning area using the tree height-biomass inversion model, includes: Based on the latitude and longitude coordinates of each Avicennia marina tree in the quadrat survey results, the measured biomass data used for fitting is matched one by one with the tree height data derived from the three-dimensional point cloud data. A biomass inversion model based on lidar data for retrieving tree height was constructed using the least squares method on the matched biomass measurement data and tree height data. The tree height-biomass inversion model based on lidar data was verified and calibrated using the measured tree height data, measured diameter at breast height data, and latitude and longitude coordinates, and finally the tree height-biomass inversion model was obtained. The tree height data derived from the 3D point cloud data is input into the tree height-biomass inversion model to obtain the biomass distribution of all lobeless Sorbus aegyptiaca within the lidar scanning area.

7. The method for retrieving aboveground biomass of *Sonneratia alopecuroides* using multi-source remote sensing data according to claim 6, characterized in that, Methods for constructing a biomass inversion model based on lidar data to retrieve tree height using the least squares method for matched biomass measurement data and tree height data include: Based on the linear relationship between diameter at breast height (DBH) and tree height of *Avicennia marina* during a certain growth period, and considering the allometric growth equation of *Avicennia marina*, a correlation formula between biomass and tree height of *Avicennia marina* was constructed, and a nonlinear polynomial fitting was performed in MATLAB: W=(aH 3 +bH 2 +cH) d Where W is biomass, H is tree height, and a, b, c, and d are constants.

8. The method for retrieving aboveground biomass of lobeless *Sonneratia africana* using multi-source remote sensing data according to claim 1, characterized in that, The method for collecting Sentinel-2 remote sensing images covering the lidar scanning area and extracting features from the Sentinel-2 remote sensing images to obtain the spectral features corresponding to the lidar scanning area includes: Sentinel-2 remote sensing images covering the lidar scanning area were collected and preprocessed as follows: Radiometric calibration, spatial registration, and atmospheric correction; The preprocessed Sentinel-2 remote sensing image was resampled, and the surface reflectance data was obtained based on the resampling results. The spectral characteristics within the range of the Sentinel-2 remote sensing image were obtained by performing remote sensing feature calculations on the surface reflectance data. The spectral characteristics of the lidar scanning area are obtained based on the spectral characteristics within the range of the Sentinel-2 remote sensing image.

9. The method for retrieving aboveground biomass of lobeless *Avicennia marina* using multi-source remote sensing data according to claim 1, characterized in that, The method for collecting Landsat series images covering the lidar scanning area, extracting features from the Landsat series images, and obtaining tree age data corresponding to the lidar scanning area includes: Collect a long-term Landsat image set covering the LiDAR scanning area and perform the following preprocessing: Radiation correction and cloud removal; NDVI time series data is constructed by calculating the NDVI value of each pixel in the preprocessed Landsat long-term image set; Based on the NDVI time series data, the long-term dynamic changes of pixels are obtained, and the tree age of each pixel is calculated according to the changes, to obtain the tree age data corresponding to the Landsat series image range, specifically: The NDVI time series data is processed by moving average. The year when the NDVI value after moving average processing is higher than a preset threshold is taken as the starting year for calculating tree age. The distance between the current time and the tree age is taken as the tree age. When there is a continuous sharp drop in NDVI, the starting time point of tree age is recalculated. The tree age data corresponding to the LiDAR scanning area is obtained based on the tree age data corresponding to the range of the Landsat series images.

10. The method for retrieving aboveground biomass of lobeless Avicennia marina using multi-source remote sensing data according to claim 1, characterized in that, The method for constructing a biomass inversion model of Avicennia marina based on satellite remote sensing data by performing regression analysis on the biomass distribution, spectral characteristics, and tree age data of all Avicennia marina species corresponding to the lidar scanning area includes: First, the biomass distribution of all lobeless Sorrelia biomass corresponding to the lidar scanning area is statistically analyzed using a grid according to the spatial resolution of the Sentinel-2 remote sensing image. Then, the biomass distribution, spectral characteristics, and tree age data of all Avicennia marina species corresponding to the lidar scanning area were input into a preset random forest model for regression analysis to construct an Avicennia marina biomass inversion model based on satellite remote sensing data.

Images