Forest carbon reserve monitoring method based on unmanned aerial vehicle

Abstract

The invention relates to the technical field of radio wave measurement, and further relates to a forest carbon reserve monitoring method based on an unmanned aerial vehicle. The method comprises the following steps: step 1, acquiring laser pulse and spectral data of a target forest area by using a multi-band laser and a hyperspectral camera carried by an unmanned aerial vehicle; constructing a three-dimensional point cloud picture of the target forest area according to the laser pulse; carrying out atmospheric correction calculation to obtain corrected surface reflectance; 2, calculating a terrain curved surface equation of the target forest region according to the three-dimensional point cloud picture; calculating the chlorophyll content of vegetation in the target forest area by using an inversion function of the PROSAIL model; and step 3, based on a preset biomass model and a preset carbon conversion coefficient, in combination with the chlorophyll content, calculating the carbon reserve of the target forest region. According to the method, highly layered monitoring can be carried out, chlorophyll contents and leaf area densities at different height levels of the forest are captured, and the accuracy and comprehensiveness of carbon reserve estimation are ensured.

Description

Technical Field

[0001] The invention belongs to the technical field of radio measurement, and in particular relates to a forest carbon storage monitoring method based on an unmanned aerial vehicle. Background Art

[0002] As the global climate change problem becomes increasingly serious, forests, as an important part of the global carbon cycle, have received extensive attention for their carbon storage monitoring and management. Forests play a key role in the global carbon cycle, absorbing carbon dioxide (CO2) from the atmosphere through photosynthesis and fixing it in vegetation and soil, thereby mitigating the accumulation of greenhouse gases in the atmosphere. In order to better understand the role of forests in the carbon cycle, accurately measuring and monitoring forest carbon storage has become an important topic in global research and policy making.

[0003] At present, the monitoring methods of forest carbon stocks are mainly divided into two categories: ground survey-based and remote sensing-based. Ground surveys usually select sample plots in the forest, measure parameters such as the diameter, height, and growth of trees in the sample plots, and use empirical formulas to infer the biomass and carbon stocks of the sample plots based on these data. Although this method can provide more accurate carbon stock data to a certain extent, it has significant limitations. First, ground surveys are time-consuming and labor-intensive, and it is difficult to promote and apply them on a large scale. Secondly, due to the limited representativeness of the sample plots, there are often large errors in the estimation of carbon stocks in forest areas outside the sample plots. In addition, it is difficult for ground surveys to conduct detailed layered monitoring of vegetation at different height levels (such as crowns, trunks, ground cover vegetation, etc.), resulting in low accuracy of monitoring results.

[0004] In order to overcome the limitations of ground surveys, remote sensing technology is increasingly used in forest carbon storage monitoring. Remote sensing technology can quickly collect optical or laser data on large areas of forests through sensors carried by satellites, aircraft or drones, and infer the biomass and carbon storage of the forest based on the acquired data. Among them, the monitoring method based on optical remote sensing analyzes the spectral reflectance characteristics of the forest, infers the biological parameters of vegetation such as chlorophyll content and leaf area index, and then estimates the carbon storage of the forest. Common optical remote sensing technologies include multispectral and hyperspectral imaging, which mainly obtain information such as plant health status and vegetation coverage through the absorption and reflection of reflectance spectra in visible light, infrared light and other bands. In addition, the application of laser radar (LiDAR) technology in forest carbon storage monitoring has also been increasing in recent years. LiDAR technology generates three-dimensional point cloud data of the target area by emitting pulsed lasers and measuring the time difference between them reflecting back to the sensor on the vegetation surface. LiDAR technology has strong penetration and can obtain structural information below the forest canopy. Especially in high-density forests, LiDAR can effectively penetrate the tree canopy and capture the height distribution information of tree trunks and the ground. Therefore, compared with traditional optical remote sensing technology, LiDAR technology has obvious advantages in the three-dimensional monitoring of forest biomass and carbon storage. Summary of the invention

[0005] The main purpose of the present invention is to provide a forest carbon storage monitoring method based on drones, which can perform highly stratified monitoring, capture the chlorophyll content and leaf area density at different height levels of the forest, and ensure the accuracy and comprehensiveness of carbon storage estimation. Compared with the existing technology, the present invention improves the accuracy and efficiency of carbon storage monitoring, is suitable for forest carbon storage estimation in complex environments, and has strong environmental adaptability and multifunctional application potential.

[0006] In order to solve the above problems, the technical solution of the present invention is achieved as follows:

[0007] A method for monitoring forest carbon stocks based on drones, the method comprising:

[0008] Step 1: Use the multi-band laser and hyperspectral camera carried by the drone to collect laser pulses and spectral data of the target forest area; construct a three-dimensional point cloud map of the target forest area based on the laser pulses; calculate the effective reflection intensity of the laser pulse based on the received laser pulses and the parameters of the multi-band laser; combine the radiation brightness in the spectral data and the effective reflection intensity of the laser pulse to perform atmospheric correction calculations to obtain the corrected surface reflectivity;

[0009] Step 2: Calculate the terrain surface equation of the target forest area based on the 3D point cloud map; calculate the leaf area density of the vegetation in the target forest area based on the corrected surface reflectivity and the effective reflection intensity of the laser pulse; and calculate the chlorophyll content of the vegetation in the target forest area based on the leaf area density using the inversion function of the PROSAIL model;

[0010] Step 3: Based on the preset biomass model and the preset carbon conversion coefficient, combined with the chlorophyll content, calculate the carbon storage of the target forest area.

[0011] Furthermore, in step 1, the multi-band laser receives laser pulses of different wavelengths and measures the intensity of these laser pulses after being reflected by vegetation and the ground at different heights, and infers the positions at different heights by calculating the time delay to generate a three-dimensional point cloud map.

[0012] Furthermore, in step 1, the following formula is used to calculate the effective reflection intensity of the laser pulse based on the received laser pulse and the parameters of the multi-segment laser:

[0013]

[0014] Among them, I eff (λ,x,y,z) represents the effective reflection intensity of the laser pulse with wavelength λ at point (x,y,z); x is the X-axis coordinate, y is the Y-axis coordinate, and z is the Z-axis coordinate; I0(λ) represents the initial intensity of the laser pulse with wavelength λ when it is emitted; the atmospheric attenuation coefficient α(λ,s) describes the intensity loss caused by atmospheric components during the propagation of the laser pulse at wavelength λ from the starting point s=0 to s=z; σ(λ,x,y,z) is the backscattering cross section, which describes the ability of the target area to reflect the laser pulse at wavelength λ after the laser pulse reaches point (x,y,z) in the target area; R is the propagation distance, which represents the propagation distance of the laser pulse from the multi-band laser to the target area; A r is the effective receiving area of ​​the multi-band laser; s is the integral variable; η sys (λ) is the system optical efficiency, which represents the optical efficiency of the multi-band laser at wavelength λ. It reflects the performance loss of the optical components within the system and describes the efficiency of the laser pulse from emission, propagation to reception.

[0015] Furthermore, in step 1, atmospheric correction is performed using the following formula to calculate the corrected surface reflectance:

[0016]

[0017] Among them, ρ surface (λ,x,y) is the corrected surface reflectance at wavelength λ and spatial position (x,y); Lsensor (λ,x,y) is the radiation brightness received by the hyperspectral camera at wavelength λ, which includes the surface reflected light and atmospheric scattered light; T ↑ (λ) and T ↓ (λ) are the atmospheric upward and downward transmittances, respectively; T ↑ (λ) is the transmittance of the atmosphere to the light of wavelength λ from the surface to the hyperspectral camera; T ↓ (λ) is the transmittance of the atmosphere to light of wavelength λ from the sun to the earth's surface; E0(λ) represents the irradiance of the sun outside the earth's atmosphere at wavelength λ, that is, the intensity of the light that can be irradiated to the earth's surface without being affected by the atmosphere; the solar zenith angle θ refers to the angle of the sun's rays relative to the vertical surface; E down (λ) is the atmospheric downward irradiance of light with wavelength λ, which is the light intensity of light with wavelength λ in the atmosphere returning to the surface after being scattered.

[0018] Furthermore, in step 2, the terrain surface equation is expressed by the following formula:

[0019]

[0020] Among them, S(x,y) represents the height of the ground at the position (x,y); is related to the surface gradient size The associated diffusion coefficient, anisotropic diffusion allows for stronger smoothing in flat areas, while suppressing smoothing at edges or steep areas, thus preserving terrain details, It decreases as the gradient size increases to reduce diffusion at the edges; is the surface divergence, which describes the rate and direction of change of the surface shape; is the surface gradient, which describes the inclination direction and slope of the surface; S0 is the initial terrain model generated based on the 3D point cloud map; λ is a weight parameter that controls the degree of fit between S(x, y) and S0. The larger the λ value, the closer S(x, y) is to S0; μ is the multi-band laser intensity weight; I eff (λ ground ,x,y,S) represents the effective reflection intensity of the ground laser pulse, which is the surface at wavelength λ ground The effective reflected intensity of the laser pulse under .

[0021] Furthermore, in step 2, the leaf area density is calculated by the following formula:

[0022]

[0023] Where LAD(z) represents the leaf area density at height z; λ vegis the average wavelength of light; G(θ) is the leaf projection function, and the formula is as follows:

[0024]

[0025] Furthermore, in step 2, multiple key wavelengths are selected, namely λ1, λ2, …, λ i ,…,λ n ; i is the i-th key wavelength; n is the number of key wavelengths; based on the leaf area density, the PROSAIL inversion model is used to calculate the chlorophyll content C of the vegetation in the target forest area at a height of z using the following formula: ab (z):

[0026]

[0027] in, is the inversion function of the PROSAIL model; ρ surface (λ1,λ2,…,λ i ,…,λ n ) indicates that the wavelengths are λ1,λ2,…,λ i ,…,λ n is the vector of the corrected surface reflectance corresponding to the time; LAI is the leaf area index, which is obtained by integrating LAD(z) in the z direction; θ v is the observation zenith angle; φ is the relative azimuth, which means the angle of the projection of the observation direction of the hyperspectral camera on the horizontal plane relative to the true north direction; angle β is the preset influence coefficient.

[0028] Furthermore, in step 3, the preset biomass model is expressed using the following formula:

[0029]

[0030] Where z0 is the height of the ground surface; z max is the highest height of the tree crown; a1 and a2 are the influence coefficients of the first-order derivative and second-order derivative of chlorophyll content related to plant type, respectively; b i is the preset weight coefficient of the i-th key wavelength; I eff (λ i ,z) represents the wavelength λ i is the effective reflection intensity of the laser pulse at height z; B is the biomass.

[0031] Furthermore, the preset carbon conversion coefficient is C coef , with a value of 0.45 to 0.5; the carbon storage C of the target forest area is calculated using the following formula:

[0032]

[0033] Among them, ρ soil (z) is the soil reflectivity at height z; σ(z) is the canopy backscattering cross section at height z; I eff (λ,z) ​​is the effective reflection intensity of the laser pulse at wavelength λ at height z; E abs (z) is the light absorption at height z, which represents the solar radiation energy absorbed by the canopy; E ref (z) is the reflected light energy at height z, which represents the radiation energy reflected back by the canopy; R(z) represents the propagation distance of the laser pulse from the transmitter to the canopy and then reflected back to the receiver.

[0034] The forest carbon storage monitoring method based on unmanned aerial vehicle of the present invention has the following beneficial effects: through the present invention, multi-band laser radar is combined with hyperspectral imaging technology for the first time, and the geometric structure and optical characteristics of the forest can be captured simultaneously. This data fusion method improves the system's sensitivity to different levels and different types of vegetation in the forest, and can more accurately reflect the chlorophyll content, leaf area index (LAI) and vegetation health status, thereby achieving higher precision carbon storage estimation. The present invention can accurately estimate the carbon storage of the forest at different height levels through layered monitoring in the height direction. The carbon storage distribution of forest vegetation is often highly heterogeneous, and the crown, trunk and surface vegetation contribute differently to carbon absorption and storage. Traditional monitoring methods are difficult to effectively distinguish these levels, and the present invention can capture the complex carbon distribution structure inside the forest by introducing the layered integration of leaf area density (LAD), chlorophyll content and laser reflection intensity. Specifically, the system ensures the comprehensiveness and precision of carbon storage estimation by integrating the biomass and carbon conversion coefficients at different height levels. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A schematic diagram of a method flow of a method for monitoring forest carbon reserves based on an unmanned aerial vehicle (UAV) according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] The terms "first", "second", "third", "fourth", etc. (if any) in the description and claims of the present invention and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein.

[0037] Example 1, reference Figure 1 :A method for monitoring forest carbon reserves based on drones, the method comprising:

[0038] Step 1: Use the multi-band laser and hyperspectral camera carried by the drone to collect laser pulses and spectral data of the target forest area; construct a three-dimensional point cloud map of the target forest area based on the laser pulses; calculate the effective reflection intensity of the laser pulse based on the received laser pulses and the parameters of the multi-band laser; combine the radiation brightness in the spectral data and the effective reflection intensity of the laser pulse to perform atmospheric correction calculations to obtain the corrected surface reflectivity;

[0039] Through the drone platform, a multi-band laser is used to emit laser pulses and record the returned laser signals to obtain the three-dimensional structural information of the forest area. In this way, a three-dimensional point cloud map of the target area can be constructed. This process relies on the principle of laser time-of-flight (ToF), that is, after the laser pulse is emitted from the transmitter, it encounters the target object such as the crown or the ground surface and is reflected. The sensor calculates the distance based on the return time, thereby generating accurate three-dimensional point cloud data. Compared with traditional ground measurement methods, laser radar (LiDAR) can quickly and widely obtain three-dimensional spatial data through drone platforms, greatly improving efficiency and accuracy, and can also fine-tune modeling of complex forest structures.

[0040] Combined with the effective reflection intensity of the laser pulse, the present invention further enhances the detection capability of the forest structure. The reflection intensity is closely related to factors such as the target surface material, roughness and incident angle, and it reflects the physical characteristics of the interaction between the laser pulse and the target object. By emitting pulses of different wavelengths by a multi-band laser, the present invention can better distinguish different types of vegetation and surface objects in the forest. This design breaks through the limitations of traditional single-band lidar, allowing the system to analyze the target area from multiple spectral dimensions, thereby providing a richer data basis for subsequent vegetation and carbon storage estimates. At the same time, the hyperspectral camera also plays an important role in step 1. Hyperspectral imaging technology can capture the spectral characteristics of different object surfaces by subdividing each band within the continuous spectral range. Different plant species in the forest, leaf water content, chlorophyll content and other physiological parameters are reflected through changes in spectral reflectance. Therefore, the present invention combines the spectral data obtained by the hyperspectral camera with the three-dimensional point cloud data obtained by the lidar to more comprehensively and accurately characterize the physical and biological characteristics of the target forest area. In particular, by measuring the radiance, the system can obtain the original spectral data of the target area, and the acquisition of radiance is the premise of spectral reflectance estimation. After data acquisition, the system uses the effective reflection intensity of the laser pulse and the spectral data of the hyperspectral camera for atmospheric correction, which is to eliminate the atmospheric interference of light during propagation, such as the influence of aerosols, clouds and other factors on data accuracy. The core idea of ​​atmospheric correction is to correct the atmospheric component interference in spectral data and laser reflection intensity through model calculation or reference standard substances. Under normal circumstances, the spectral reflectance will be affected by atmospheric components and be low or high. Therefore, through the atmospheric correction method, the present invention can restore the real reflection characteristics of the target object and obtain the corrected surface reflectance. Atmospheric correction is an important step in spectral remote sensing analysis. It ensures that hyperspectral data can accurately reflect the real situation of the surface, which lays a solid foundation for the calculation of biomass and carbon storage based on spectral information in subsequent steps.

[0041] Step 2: Calculate the terrain surface equation of the target forest area based on the 3D point cloud map; calculate the leaf area density of the vegetation in the target forest area based on the corrected surface reflectivity and the effective reflection intensity of the laser pulse; and calculate the chlorophyll content of the vegetation in the target forest area based on the leaf area density using the inversion function of the PROSAIL model;

[0042] The three-dimensional point cloud map can accurately reflect the crown height, tree distribution and terrain changes, which makes it possible to accurately derive the terrain surface equation of the forest. This equation describes the height change of the terrain in the forest area. Especially for mountainous areas or areas with large terrain undulations, the establishment of the terrain surface equation is particularly important because it can eliminate the errors caused by terrain changes, thereby ensuring the accurate extraction of subsequent vegetation features. Next, based on the surface reflectivity and the effective reflection intensity of the laser pulse, the leaf area density of the target area can be calculated. Leaf area density is an important parameter to characterize vegetation density, which reflects the total area of ​​leaves per unit volume. Through the change of the reflection intensity of the laser pulse in different vegetation layers, combined with the surface reflectivity, the system can estimate the coverage and distribution of vegetation. Compared with traditional ground survey methods, LiDAR provides higher spatial resolution and more comprehensive coverage through UAV platforms, making the calculation of leaf area density more accurate and efficient. In addition, LiDAR can penetrate the canopy layer and provide information about different vegetation layers, which makes the estimation of leaf area density in complex forest environments more refined.

[0043] After obtaining the leaf area density, the present invention calculates the chlorophyll content of the vegetation through the inversion function of the PROSAIL model. The PROSAIL model is a physical model that links the spectral reflectance of vegetation with its physiological parameters. It combines the PROSPECT leaf model and the SAIL canopy model and can effectively simulate the spectral characteristics of vegetation. The core idea of ​​the PROSAIL model is to generate the spectral reflectance of the canopy by inputting a series of physiological parameters such as leaf structure parameters, chlorophyll content, and leaf water content. In the present invention, using the known surface reflectance and leaf area density, the system can infer the chlorophyll content in the leaf through the model inversion function. This process is actually a process of combining hyperspectral data with the physiological state of vegetation. Hyperspectral data can capture the absorption and reflection of leaves to light of different bands, and this information is the indirect expression of chlorophyll content. In the inversion process of the PROSAIL model, spectral data provides key input parameters because chlorophyll has significant differences in its absorption capacity for red light and near-infrared light. Generally, leaves with higher chlorophyll content will strongly absorb red light, while in the near-infrared light band, they will show stronger reflection. Therefore, the system analyzes these differences in the spectral data and combines it with vegetation structure information such as leaf area density to invert and calculate the chlorophyll content in the target area.

[0044] The PROSPECT model is used to describe the reflectance and transmittance of a single leaf in the visible and near-infrared bands. It simulates the interaction between light and the internal structure of the leaf, especially the effects of chlorophyll, leaf thickness, water content, etc. on the spectral reflectance characteristics. There are several key parameters in the PROSPECT model, which are used to characterize the structure and spectral absorption characteristics of the leaf, among which the chlorophyll content is one of the most important parameters. By analyzing the difference in absorption and reflection of leaves in the red light (visible light) and near-infrared light bands, the model can reflect the chlorophyll content. This is because chlorophyll has strong absorption in the red light band (about 660-680nm) and shows higher reflection in the near-infrared band (greater than 700nm). The PROSPECT model simulates the spectral reflectance and transmittance of the leaf by inputting spectral data of different bands and combining pre-set leaf physiological parameters. The SAIL model is a canopy model that is mainly used to describe the multiple scattering, absorption and reflection process of light in the entire vegetation canopy. The reflectance characteristics of the canopy are not only related to the reflectance of the leaves, but also affected by factors such as the arrangement of leaves in the canopy, vegetation density and canopy height. The input parameters of the SAIL model include leaf reflectance and transmittance (provided by the PROSPECT model), leaf area index (LAI, representing the total leaf area per unit surface area), and other canopy structure parameters. With these input parameters, the SAIL model can simulate the spectral reflectance of the entire canopy, taking into account light scattering between leaves and light transmission between canopy layers. In this way, the SAIL model provides complete spectral reflectance information for the entire vegetation system.

[0045] The inversion process is the process of inferring the chlorophyll content from known spectral reflectance data. In the present invention, the hyperspectral camera carried by the drone collects spectral reflectance data of the target forest area, which contains information of multiple bands such as red light and near-infrared light. By inputting these observation data into the PROSAIL model, the system can infer the chlorophyll content of the vegetation based on the characteristics of the spectrum. Specifically, the inversion process usually adopts an iterative optimization method, such as the least squares method or an optimization algorithm based on Bayesian inference. The system will adjust the parameters in the PROSPECT and SAIL models (such as chlorophyll content, leaf area index, leaf water content, etc.) according to the observed spectral reflectance until the error between the spectral reflectance simulated by the model and the observed data is minimized. Ultimately, these optimized parameter values ​​are the inverted vegetation physiological parameters such as chlorophyll content. The key to the inversion of chlorophyll content lies in the changes in spectral reflectance in the red and near-infrared bands. Since chlorophyll has strong absorption capacity in the red light band, the higher the chlorophyll content, the stronger the vegetation's absorption of red light and the lower the reflectivity. In the near-infrared band, due to the multiple reflections of the leaf cell structure, the reflectivity is relatively high.

[0046] Step 3: Based on the preset biomass model and the preset carbon conversion coefficient, combined with the chlorophyll content, calculate the carbon storage of the target forest area.

[0047] Biomass refers to the total amount of organic matter in forest vegetation, including the stems, leaves, roots and other parts of trees. Traditional biomass estimation methods usually rely on the collection of ground samples and laboratory analysis. Although this method is accurate, it is time-consuming and difficult to apply on a large scale. In the present invention, the estimation of biomass is completed by remote sensing data collected by drones, especially in combination with the chlorophyll content calculated in the previous step. Chlorophyll content reflects the health and growth potential of vegetation, so it is closely related to biomass. Generally, the higher the chlorophyll content of vegetation, the higher its biomass, because the photosynthesis of vegetation is more active and more organic matter is accumulated. Through this association, the present invention can infer the biomass of vegetation based on chlorophyll content, which provides an important basis for further calculation of carbon storage. In addition to chlorophyll content, leaf area index (LAI) and other vegetation parameters also play an important role in the estimation of biomass. Leaf area index refers to the total area of ​​leaves per unit surface area, which is an important parameter for measuring vegetation density and photosynthesis efficiency. With the support of the laser radar and hyperspectral sensor carried by the drone, the present invention can efficiently and accurately obtain the leaf area index of the target forest area. Based on the comprehensive analysis of the leaf area index and chlorophyll content, the system can build a biomass model. After inputting these parameters into the model, the system can estimate the biomass of vegetation in different areas. Compared with traditional ground measurement methods, this method has higher spatial resolution and coverage, and can quickly and comprehensively estimate biomass in large-scale, complex terrain forest environments.

[0048] After obtaining the estimated value of biomass, the present invention converts biomass into carbon storage through a preset carbon conversion coefficient. The carbon conversion coefficient is usually a parameter obtained based on a large number of field experiments, which represents the proportion of carbon content in a unit biomass. The carbon content of forest vegetation mainly exists in the cell walls of trees, especially in organic substances such as lignin and cellulose, so the carbon conversion coefficient may be different in different types of vegetation. In order to ensure the accuracy of carbon storage estimation, the present invention uses a variety of different types of carbon conversion coefficients, which are set based on experimental data of different types of trees and vegetation. For example, for coniferous forests and broad-leaved forests, their carbon conversion coefficients may be different. The system will identify the type of forest vegetation based on remote sensing data, and then select a suitable carbon conversion coefficient for calculation. This flexibility enables the present invention to adapt to the carbon storage estimation needs of different ecosystems. The calculation formula of carbon storage mainly involves the product relationship between biomass and carbon conversion coefficient. By multiplying the biomass in each area by the corresponding carbon conversion coefficient, the system can obtain the carbon storage in the area. Then, the system summarizes the carbon storage in each area and finally obtains the carbon storage of the entire target forest area.

[0049] Embodiment 2: In step 1, the multi-band laser receives laser pulses of different wavelengths and measures the intensity of these laser pulses after being reflected by vegetation and the ground at different heights, and infers the positions at different heights by calculating the time delay to generate a three-dimensional point cloud map.

[0050] Specifically, the reflection intensity of the laser pulse is closely related to many factors such as the material of the target object, surface roughness, tilt angle, and laser wavelength. By adopting a multi-band laser, that is, emitting and receiving laser pulses of different wavelengths, the system can obtain the reflection information of the target in different spectral ranges. This multi-band laser technology has significant advantages over single-band lasers because lasers of different wavelengths exhibit different physical phenomena when interacting with target objects. For example, lasers with shorter wavelengths may be more sensitive to reflections from leaves, while lasers with longer wavelengths may more easily penetrate the canopy and provide information about the internal structure of vegetation. Through this multi-band combination, the present invention can more comprehensively reflect the various levels of forest structure and generate a more detailed three-dimensional point cloud map.

[0051] A three-dimensional point cloud is a spatial data expression method, which consists of a large number of three-dimensional coordinate points, each of which represents a specific spatial position. In the present invention, the laser radar constructs a three-dimensional point cloud map of forest vegetation and the ground surface by collecting laser reflection echoes at different heights. This point cloud map can not only describe the topographic undulations of the forest, but also clearly reflect the distribution of vegetation at different height levels, including the crown, trunk, vegetation under the tree and other levels. Compared with traditional two-dimensional remote sensing data, the three-dimensional point cloud map can more accurately depict the spatial structure of the forest, so that the system can take into account the biomass and carbon content of different vegetation layers when analyzing forest carbon reserves. In addition, the laser radar infers the vegetation and surface positions at different heights by calculating the time delay, which provides a solid data foundation for subsequent forest carbon reserve estimation. In a forest environment, the laser pulse will not only be reflected by the crown, but also penetrate some leaves to reach the vegetation layer below and even the ground surface. This multi-level data acquisition method enables the system to build a complete three-dimensional structural model from the crown to the ground surface, which is crucial for the accurate estimation of forest biomass and carbon reserves. Compared with traditional ground survey methods, LiDAR can cover large areas through UAV platforms and can quickly and accurately obtain three-dimensional spatial data, greatly improving work efficiency. By combining the reflection intensity of laser pulses with three-dimensional point clouds, the system can not only provide spatial information of the target area, but also infer the density and composition of different vegetation layers through reflection intensity. The reflection intensity of laser pulses reflects the characteristics of the interaction between light and matter. Therefore, by analyzing the reflection intensity of lasers of different wavelengths at different heights, the distribution characteristics of different vegetation layers such as crowns, leaves, and trunks can be identified.

[0052] Embodiment 3: In step 1, the following formula is used to calculate the effective reflection intensity of the laser pulse according to the received laser pulse and the parameters of the multi-segment laser:

[0053]

[0054] Among them, I eff (λ,x,y,z) represents the effective reflection intensity of the laser pulse with wavelength λ at point (x,y,z); x is the X-axis coordinate, y is the Y-axis coordinate, and z is the Z-axis coordinate; I0(λ) represents the initial intensity of the laser pulse with wavelength λ when it is emitted; the atmospheric attenuation coefficient α(λ,s) describes the intensity loss caused by atmospheric components during the propagation of the laser pulse at wavelength λ from the starting point s=0 to s=z; σ(λ,x,y,z) is the backscattering cross section, which describes the ability of the target area to reflect the laser pulse at wavelength λ after the laser pulse reaches point (x,y,z) in the target area; R is the propagation distance, which represents the propagation distance of the laser pulse from the multi-band laser to the target area; A r is the effective receiving area of ​​the multi-band laser; s is the integral variable; ηsys (λ) is the system optical efficiency, which represents the optical efficiency of the multi-band laser at wavelength λ. It reflects the performance loss of the optical components within the system and describes the efficiency of the laser pulse from emission, propagation to reception.

[0055] Specifically, I0(λ) in the formula represents the initial intensity of the laser pulse when it is emitted, which is the energy emitted by the laser at a specific wavelength λ. This part describes the basic power output of the laser. Lasers of different wavelengths have different initial energy distributions. In a forest environment, using a multi-band laser to emit pulses of different wavelengths can obtain reflection information at different wavelengths, thereby distinguishing different vegetation types in the forest. Compared with a single wavelength laser, this multi-band technology can provide richer spectral information and provide more dimensional data support for subsequent analysis. Next, the laser pulse will be affected by the attenuation effect when propagating in the atmosphere. This phenomenon is explained by the formula Describe it. The atmospheric attenuation coefficient α(λ,s) reflects the energy loss of the laser pulse during propagation due to atmospheric components (such as aerosols, clouds and water vapor). When the laser pulse propagates in the air, the energy gradually weakens as the propagation distance increases. This part can be understood as the laser pulse interacting with particles and molecules in the atmosphere when passing through the atmosphere, and part of the energy is absorbed or scattered. This attenuation effect varies with the wavelength and altitude. Lasers of different wavelengths have different attenuation rates in the atmosphere, and short-wavelength lasers usually attenuate faster than long-wavelength lasers. Therefore, in the present invention, by modeling the attenuation effect of lasers of different wavelengths, the system can more accurately estimate the laser reflection intensity at different altitudes in the forest, thereby improving the accuracy of carbon stock estimation.

[0056] When a laser pulse reaches a target area (such as a tree canopy or the ground in a forest), some of the energy is reflected by the target back to the sensor. to describe. Here, σ(λ, x, y, z) represents the backscattering cross section, which reflects the reflection ability of the target object at a specific wavelength. The size of the backscattering cross section depends on the material, surface structure and wavelength of the target object. For example, the reflection abilities of leaves, tree trunks and the ground are different. When the laser pulse reaches these different surfaces, the energy reflected back will also be different. By introducing the backscattering cross section, the present invention can more carefully analyze the reflection characteristics of vegetation at different levels in the forest, and then infer the overall structure and vegetation type of the forest. The propagation distance R of the laser pulse from the emission point to the target area also has an important influence on the reflection intensity. According to the principle of spherical diffusion, the energy of the laser pulse will spread as the propagation distance increases, and the intensity of the reflected signal decreases inversely proportional to the square of the distance. This phenomenon is explained by 4πR 2It is described by the diffusion of laser energy in space. The farther the distance, the weaker the reflected signal. Therefore, when calculating the effective reflection intensity, the distance of the laser pulse propagation must be taken into account, which helps to correct the intensity difference caused by the distance. The receiving area A of the multi-band laser r It is an important parameter for the system to receive reflected signals. The larger the receiving area, the more reflected photons the system can capture, and the higher the signal strength. This parameter is usually determined by the design of the laser. By optimizing the receiving area, the efficiency and quality of signal reception can be improved. In a forest environment, the efficiency of signal reception is crucial because the canopy and vegetation in the forest will cause multiple reflections and scattering of the signal, increasing the complexity of signal reception.

[0057] Finally, the system optical efficiency η sys (λ) describes the performance loss of the optical components in the system during the process from the emission to the reception of the laser pulse. This part includes the influence of the lens, filter and other optical devices that the laser pulse passes through during the transmission process. The reduction of optical efficiency may be caused by the absorption, reflection and scattering effects of the device. Therefore, in the system design, optimizing the optical efficiency is an important means to improve the accuracy of the calculation of the laser reflection intensity. In the present invention, considering the optical efficiency of the system, the measurement data can be further corrected to ensure that the calculation result of the laser reflection intensity is more accurate. Unlike the traditional single-band laser radar, the present invention can obtain the reflection information at different wavelengths in the forest by using a multi-band laser, and then conduct a detailed analysis of the vegetation at different heights. The effective reflection intensity of the laser pulse provides key input data for the subsequent biomass and carbon storage estimation. The system can more accurately monitor the forest carbon storage by modeling the intensity change of the laser pulse. Compared with the traditional ground survey method, the present invention uses the long-range detection capability of the laser radar to quickly and widely obtain high-precision forest structure data without direct contact with the vegetation, which greatly improves the work efficiency and also improves the spatial resolution and accuracy of the data. This innovative remote sensing monitoring method makes the estimation of forest carbon stocks more scientific and accurate, and has important practical application value.

[0058] Embodiment 4: In step 1, atmospheric correction is performed by the following formula to calculate the corrected surface reflectivity:

[0059]

[0060] Among them, ρ surface (λ,x,y) is the corrected surface reflectance at wavelength λ and spatial position (x,y); L sensor (λ,x,y) is the radiation brightness received by the hyperspectral camera at wavelength λ, which includes the surface reflected light and atmospheric scattered light; T ↑(λ) and T ↓ (λ) are the atmospheric upward and downward transmittances, respectively; T ↑ (λ) is the transmittance of the atmosphere to the light of wavelength λ from the surface to the hyperspectral camera; T ↓ (λ) is the transmittance of the atmosphere to light of wavelength λ from the sun to the earth's surface; E0(λ) represents the irradiance of the sun outside the earth's atmosphere at wavelength λ, that is, the intensity of the light that can be irradiated to the earth's surface without being affected by the atmosphere; the solar zenith angle θ refers to the angle of the sun's rays relative to the vertical surface; E down (λ) is the atmospheric downward irradiance of light with wavelength λ, which is the light intensity of light with wavelength λ in the atmosphere returning to the surface after being scattered.

[0061] Specifically, ρ in the formula surface (λ,x,y) represents the surface reflectance after atmospheric correction, which is the value we need to get in the end. The surface reflectance describes the ability of the surface (such as leaves and tree trunks in the forest) to reflect sunlight at a specific wavelength λ. Since light is affected by atmospheric components when passing through the atmosphere, resulting in scattering, absorption and other phenomena, the uncorrected reflectance cannot accurately reflect the actual situation of the surface, so atmospheric correction is required. In the right part of the formula, first π·[L sensor (λ,x,y)-L path (λ)], where L sensor (λ,x,y) represents the radiance received by the hyperspectral camera at wavelength λ. This includes the light reflected from the ground and the scattered light from the atmosphere. Since the light received by the sensor contains not only the light reflected from the ground but also the light scattered from the atmosphere, L must be subtracted from it. path (λ), which is the atmospheric path radiance. This radiance is caused by light scattering caused by atmospheric components (such as aerosols, water vapor, etc.). After subtracting this part, the influence of atmospheric scattering on the radiance received by the sensor can be effectively eliminated, thus being closer to the actual reflection of the surface.

[0062] Next It is the inverse of the atmospheric upward transmittance. ↑ (λ) represents the transmittance of the atmosphere to light of wavelength λ when the light from the surface to the hyperspectral camera passes through the atmosphere. Since there are various components in the atmosphere, these components have different degrees of absorption and scattering of light of different wavelengths, so this transmittance is needed to correct the energy loss of light when passing through the atmosphere. By introducing the uplink transmittance, the system can ensure that the calculated surface reflectivity will not be deviated due to the absorption and scattering effects of the atmosphere. The next part of the formula [E0(λ)·cosθ·T ↓ (λ)+E down(λ)] describes the incident light reaching the surface. This part can be divided into two parts: the first part is E0(λ)·cosθ·T ↓ (λ), where E0(λ) represents the solar irradiance outside the Earth's atmosphere, which is the intensity of sunlight without atmospheric absorption and scattering. ↓ (λ), the downward atmospheric transmittance, is used to correct for the energy loss of sunlight due to atmospheric absorption and scattering in the process of reaching the earth's surface from outside the atmosphere. Multiplying by cosθ is to take into account the solar zenith angle θ, which is the angle of incidence of sunlight relative to the vertical surface. When the sun is lower in the sky, the angle of incidence is larger and the light needs to pass through more atmosphere, so the energy of the light is further attenuated. By introducing the zenith angle, the formula can more accurately describe the actual light intensity reaching the earth's surface. Part II E down (λ) is the scattered light in the atmosphere. This part is due to the irradiance of sunlight returning to the surface after being scattered in the atmosphere. The scattering of light by molecules, aerosols and other particles in the atmosphere will cause part of the light to deviate from the original propagation path and reach the surface again. Although this part of light does not come directly from the sun, it will also affect the reflection of the surface. Therefore, when calculating the surface reflectivity, this part of the atmospheric downward irradiance E must be taken into account. down (λ) is taken into account to ensure the accuracy of the calculation. Finally, the formula It is the normalization of the effective reflection intensity of the laser pulse. eff (λ,x,y,0) reflects the light reflection characteristics of the target area. By combining the surface reflectivity with the effective reflection intensity of the laser pulse, the system can eliminate the light reflection differences caused by different materials in the target area, ensuring that the calculation results of the reflectivity are more objective.

[0063] Embodiment 5: In step 2, the terrain surface equation is expressed by the following formula:

[0064]

[0065] Among them, S(x,y) represents the height of the ground at the position (x,y); is related to the surface gradient size The associated diffusion coefficient, anisotropic diffusion allows for stronger smoothing in flat areas, while suppressing smoothing at edges or steep areas, thus preserving terrain details, It decreases as the gradient size increases to reduce diffusion at the edges; is the surface divergence, which describes the rate and direction of change of the surface shape; is the surface gradient, which describes the inclination direction and slope of the surface; S0 is the initial terrain model generated based on the 3D point cloud map; λ is a weight parameter that controls the degree of fit between S(x, y) and S0. The larger the λ value, the closer S(x, y) is to S0; μ is the multi-band laser intensity weight; I eff (λ ground ,x,y,S) represents the effective reflection intensity of the ground laser pulse, which is the surface at wavelength λ ground The effective reflected intensity of the laser pulse under .

[0066] Specifically, the left side of the equation Describes the rate of change of the surface height S(x,y) over time t. This means that the surface height S(x,y) evolves over time, gradually adjusting from the initial terrain model S0 to the terrain distribution that conforms to the actual situation. This evolution process is similar to the level set method in multi-scale geometry processing, which gradually optimizes the shape of the terrain surface during the iteration process to make it more consistent with the actual geographical characteristics. Next, Represents the gradient of a surface, describing the slope or inclination direction of the terrain at position (x, y). The magnitude of the gradient Indicates the steepness of terrain changes. In relatively flat areas, the gradient is small; on steep slopes or where the terrain changes suddenly, the gradient is large. This gradient information provides an important basis for the accuracy of the terrain model.

[0067] In the formula is a diffusion coefficient that adjusts the smoothness of the terrain according to the gradient. The concept of anisotropic diffusion is applied here to control the strength of diffusion under different terrain characteristics. Specifically, when When the terrain is relatively flat, the diffusion coefficient is larger, which means that the model performs stronger smoothing in these areas, thereby reducing noise or irregular changes; Larger, that is, at steep edges or areas with dramatic changes, the diffusion coefficient will be reduced accordingly, thereby suppressing smoothing and protecting the details of the terrain. This method ensures that when the model handles complex terrain, it will not lose important terrain features such as ridges, valleys or other parts with significant terrain changes due to excessive smoothing. In addition, represents the surface divergence, which describes the rate and direction of change of the surface shape. The divergence term plays a key role here because it can reflect the local geometric characteristics of the surface, that is, whether the surface at a certain location is convex or concave. This calculation is crucial to accurately capture the overall morphology of the terrain, especially when dealing with complex natural terrain, such as undulating terrain and diverse terrain features in forests. The second term λ(S0-S) in the formula is a fitting term used to fit the current surface S(x, y) to the initial terrain model S0. Here S0 is the initial terrain model generated by the three-dimensional point cloud map, which is directly extracted from the lidar data. The role of this part of the term is to adjust the fit between the surface and the initial model through the weight parameter λ. When the λ value is large, the current surface S(x, y) will be closer to the initial model S0; when the λ value is small, the surface will have more degrees of freedom to adjust itself, so as to better adapt to the actual terrain characteristics. Through this fitting, the terrain model can strike a balance between the actual data and the initial estimate. Finally, μ·I eff (λ ground ,x,y,S) is a term related to the effective reflection intensity of the laser pulse. The effective reflection intensity of the laser pulse I eff (λ ground ,x,y,S) is the surface at wavelength λ ground The intensity of reflected light under the ground directly reflects the characteristics of the surface material. This item is used to directly introduce the data measured by the lidar into the terrain model, ensuring that the model is not only adjusted according to the geometric shape, but also accurately fits the physical characteristics of the surface. The stronger the laser reflection intensity, the greater the ability of the surface to reflect light, which is usually related to the smoothness or undulation of the terrain structure. By introducing this parameter, the terrain model can better reflect the actual physical characteristics, thereby improving the reliability and accuracy of the model.

[0068] LiDAR uses the time-of-flight (ToF) method to measure the time from the laser emission to the target surface reflecting back to the sensor by emitting a pulsed laser. Based on the propagation time of the laser pulse and the speed of light, the system can calculate the distance between the target object (such as the ground, trees, etc.) and the laser. The multi-band laser on the drone will simultaneously emit laser pulses of different wavelengths, which will be reflected back to the drone's receiver after contacting the ground or vegetation. By analyzing these reflected signals, the system can obtain countless three-dimensional coordinate points in the target area. The coordinates of each point reflect the spatial position of the point, that is, the specific height and position of the surface of the target object in three-dimensional space. Through the flight of the drone, the LiDAR system scans and emits laser pulses in the target area. The system converts the collected reflection data into spatial coordinate points to form a dense three-dimensional point cloud. These point cloud data are usually composed of thousands or even millions of three-dimensional points, each of which is represented in the form of (x, y, z). Here, x and y are the coordinates of the point on the horizontal plane, and z represents the height of the point, reflecting the height of the ground or vegetation at that location. Through these point cloud data, the system can roughly outline the spatial form of the entire area.

[0069] Although the 3D point cloud contains rich spatial information, many of the points actually represent the height of vegetation such as tree crowns and trunks, while what we need when generating the initial terrain model is the surface height. Therefore, the point cloud must be filtered and processed to remove the points representing vegetation and extract the surface points. This is usually achieved through an algorithm called classification or separation. The algorithm automatically distinguishes between surface points and vegetation points based on the height distribution of the points, the intensity of the laser echo, and the spatial distribution characteristics. For example, the reflection points of the tree crown are usually high and have a certain continuity, while the surface points are usually located at a lower position and are distributed more gently. Once the points representing the surface are extracted, it is necessary to generate a continuous terrain model through interpolation technology. Since the points in the 3D point cloud are discrete and cannot be directly used to represent a continuous terrain surface, these discrete points must be interpolated. Common interpolation methods include nearest neighbor interpolation, bilinear interpolation, and Kriging interpolation. These methods estimate the height value at an unknown position based on the spatial position of the existing points, thereby generating a smooth and continuous surface model. After interpolation, the system usually represents the terrain as a digital elevation model or digital terrain model. These models divide the terrain into fine grids, and the height value in each grid represents the average surface height of the area. Through this gridding method, the system can generate a high-resolution initial terrain model, which provides a basis for subsequent terrain correction. The generated initial terrain model S0 (x, y) is the product of interpolation and gridding of the three-dimensional point cloud, which reflects the surface height distribution of the target area. This model not only represents the overall characteristics of the terrain, such as hills, plains, valleys, etc., but also retains local surface details. In the context of the present invention, this initial model is used as the reference model S0 in the terrain surface equation, that is, during the terrain optimization process, the current terrain surface will be gradually adjusted to better match the actual terrain data.

[0070] Example 6: In step 2, the leaf area density is calculated by the following formula:

[0071]

[0072] Where LAD(z) represents the leaf area density at height z; λ veg is the average wavelength of light; G(θ) is the leaf projection function, and the formula is as follows:

[0073]

[0074] Specifically, the core part of the formula is calculated by calculating the effective reflection intensity ∑ of the laser pulse at height z λ I eff (λ,x,y,z), and compare it with the reflection intensity ∑ λ I eff(λ,x,y,0) is compared to infer the attenuation of the laser pulse when passing through vegetation. This attenuation directly reflects the leaf area density of the vegetation. Since the lidar has the ability to penetrate the tree canopy, it can not only measure the reflected signal on the surface of the tree canopy, but also capture the signal after the laser penetrates the leaves. Therefore, by analyzing the laser reflection intensity layer by layer, the system can accurately calculate the leaf area density at each height level. In this calculation process, the formula is transformed by logarithm. To process the ratio of laser reflection intensity. The physical significance of this operation is to linearize the nonlinear change of reflection intensity, so that the system can more accurately estimate the energy loss of laser pulses in the vegetation layer. When the laser penetrates the vegetation, it is affected by the leaf density and the degree of attenuation changes with height. Therefore, by processing the logarithm of the reflection intensity, the trend of vegetation density changing with height can be effectively captured. This logarithmic transformation can amplify subtle signal changes, especially in high-density leaf areas, and can improve the accuracy of leaf area density calculation.

[0075] In addition, the formula Represents the partial derivative with respect to height z, which reflects the rate at which leaf area density changes with height. By deriving the gradient change in the height direction, the system can accurately capture the distribution of leaf area density at different heights. This is particularly important for the accurate estimation of forest carbon storage, because different height levels of forests (such as the canopy layer, understory vegetation layer, etc.) have significant differences in carbon absorption and biomass accumulation. The calculation of leaf area density also introduces the leaf projection function G(θ), which describes the effect of the incident angle of sunlight relative to the leaf surface. The projection area of ​​the leaf depends on the incident angle of the sunlight. When the light is perpendicular to the leaf surface, the leaf can receive more light energy and the projection area is larger; when the light is parallel to the leaf surface, the projection area is smaller. This function allows the system to correct the actual projection area of ​​the leaves, ensuring that the calculation of leaf area density can reflect the actual photosynthetically active area at different incident angles. The leaf projection function is an important geometric correction factor in photosynthesis simulation, because the interaction of light with leaves at different incident angles will affect the photosynthetic efficiency of plants, thereby affecting the estimation of carbon storage. In order to further eliminate the interference of surface reflectance on the calculation of leaf area density, the formula also introduces the surface reflectance correction term ρ surface (λ veg ,x,y). Since LiDAR measures not only the reflection of vegetation but also the reflection of the ground surface at different wavelengths, the ground reflectance must be included in the calculation to ensure that the reflected signal only reflects the characteristics of vegetation. Surface reflectance correction is a very critical step in remote sensing analysis, especially in forest monitoring, where the reflective properties of the surface (such as exposed soil or water bodies) may affect the correct estimation of leaf area density.

[0076] Embodiment 7: In step 2, multiple key wavelengths are selected, namely λ1, λ2, ..., λ i ,…,λ n ; i is the i-th key wavelength; n is the number of key wavelengths; based on the leaf area density, the PROSAIL inversion model is used to calculate the chlorophyll content C of the vegetation in the target forest area at a height of z using the following formula: ab (z):

[0077]

[0078] in, is the inversion function of the PROSAIL model; ρ surface (λ1,λ2,…,λ i ,…,λ n ) indicates that the wavelengths are λ1,λ2,…,λ i ,…,λ n is the vector of the corrected surface reflectance corresponding to the time; LAI is the leaf area index, which is obtained by integrating LAD(z) in the z direction; θ v is the observation zenith angle; φ is the relative azimuth, which means the angle of the projection of the observation direction of the hyperspectral camera on the horizontal plane relative to the true north direction; angle β is the preset influence coefficient.

[0079] Specifically, in this calculation formula, multiple key wavelengths are selected, which are λ1, λ2, …, λ n , corresponding to the reflection and absorption characteristics of plants in different spectral ranges, especially in the red and near-infrared bands. This is because chlorophyll has a strong absorption of light in the red band (600-700nm), and exhibits a strong reflection characteristic in the near-infrared band. By acquiring the reflectance information at these wavelengths, the system can more comprehensively characterize the spectral characteristics of vegetation, and then use the PROSAIL inversion model to deduce the chlorophyll content. The core of the inversion process lies in the PROSAIL model. The PROSAIL model is a complex model that combines the leaf optical model (PROSPECT) and the canopy radiation transfer model (SAIL). The PROSPECT model mainly describes the spectral characteristics of a single leaf, taking into account the influence of leaf structure, chlorophyll content, water content and other parameters on light absorption and reflection; while the SAIL model integrates multiple leaves into a whole, simulates the optical characteristics of the entire vegetation canopy, and takes into account factors such as multiple scattering and reflection of light in the canopy and the incident angle of light and leaves. In the present invention, the inversion function of the PROSAIL model is used to estimate chlorophyll content from surface reflectance. Specifically, ρ in the formula surface (λ1,λ2,…,λn ) is a reflectance vector that represents the corrected surface reflectance at multiple key wavelengths. The core purpose of selecting these wavelengths is to capture the absorption and reflection characteristics of chlorophyll in the red and near-infrared light bands, so as to deduce the chlorophyll content through the inversion algorithm.

[0080] The formula also takes into account the leaf area index (LAI), which is obtained by vertical integration of the leaf area density (LAD). The leaf area index reflects the total area of ​​leaves per unit surface area and is an important indicator for describing the degree of vegetation coverage and photosynthetic capacity. The leaf area density (LAD) describes the total area of ​​leaves per unit volume. Its change with height z directly reflects the distribution of leaves at different height levels of the forest. By combining LAD and LAI, the PROSAIL model can more accurately describe the three-dimensional structural characteristics of the canopy, and on this basis, invert the optical properties to obtain the chlorophyll content. Observation zenith angle θ v and the relative azimuth angle φ are also important factors in this formula. The zenith angle describes the angle of incidence of the sun's rays perpendicular to the ground, while the relative azimuth describes the projection angle of the observation direction on the horizontal plane. These two geometric factors affect the transmission path of light in the vegetation canopy and its reflection and absorption characteristics. When the incident angle of light with the leaves is small, the light will experience more multiple scattering and reflection in the canopy, which in turn affects the observed intensity of reflected light. Therefore, the introduction of these geometric angles in the PROSAIL model helps to improve the accuracy of the inversion results and ensure that the calculation of chlorophyll content can reflect the actual lighting conditions and the spatial structure of the vegetation. In addition, the last term in the formula Describes the effect of leaf area density on chlorophyll content as it changes with altitude. The gradient of leaf area density in the altitude direction represents the change in leaf distribution at different height levels, and β is an influence coefficient used to adjust the contribution of leaf area density to chlorophyll content. The physical meaning of this item is that the leaf area density at different heights will have different effects on the absorption and reflection characteristics of light. Generally, leaves at higher altitudes will receive more direct sunlight, while leaves at lower altitudes will be more affected by scattered light. Therefore, with the change in altitude, the chlorophyll content will also show a certain trend of change. By adjusting the gradient of leaf area density, the formula can better capture the chlorophyll distribution characteristics at different height levels.

[0081] Example 8: In step 3, the preset biomass model is expressed using the following formula:

[0082]

[0083] Where z0 is the height of the ground surface; z maxis the highest height of the tree crown; a1 and a2 are the influence coefficients of the first-order derivative and second-order derivative of chlorophyll content related to plant type, respectively; b i is the preset weight coefficient of the i-th key wavelength; I eff (λ i ,z) represents the wavelength λ i is the effective reflection intensity of the laser pulse at height z; B is the biomass.

[0084] Specifically, the formula integrates the chlorophyll content and laser reflection intensity at different height levels of the forest to obtain the biomass distribution of the entire vegetation vertical profile. The integration in the formula starts from the ground surface height z0 to the highest point of the canopy z max This means that the system has captured all biomass changes from trunks, leaves to the top of the crown through integration in the height direction. Since the structures and characteristics of vegetation in the forest are different at different height levels, the biomass distribution characteristics of the forest can be fully reflected through height integration. First, the formula takes into account the effect of chlorophyll content changes with height on biomass. Chlorophyll is a key substance for photosynthesis in plants. Its content reflects the photosynthetic capacity of vegetation, which in turn affects the rate of biomass accumulation. In the present invention, the chlorophyll content C ab The first and second derivatives of (z) are introduced into the formula to reflect the rate and acceleration of change of chlorophyll content with height, respectively. By considering the rate of change of chlorophyll content, the system is able to capture the differences in photosynthetic efficiency of vegetation at different height levels. Specifically, the first derivative It indicates the rate at which chlorophyll content changes with altitude, reflecting the difference in photosynthesis capacity of vegetation at different altitudes; while the second-order derivative It describes the acceleration of this change rate, that is, the impact of altitude change on the rate of change of chlorophyll content. By introducing these derivative terms, the formula can more accurately describe the spatial variation of chlorophyll distribution in the forest, especially at the altitude level where the chlorophyll content changes sharply.

[0085] Changes in chlorophyll content are combined with leaf area density (LAD) to describe the distribution of leaves at different height levels. Leaf area density LAD (z) refers to the total area of ​​leaves per unit volume, which reflects the leaf density of plants at different heights. Since leaves are the main site of photosynthesis, the combination of LAD and chlorophyll content can provide a more accurate biomass estimation model. The denser the distribution of leaves, the more biomass accumulates, and vice versa. Therefore, the introduction of leaf area density ensures that the biomass calculation not only takes into account the changes in chlorophyll content, but also includes the spatial distribution characteristics of leaves, greatly improving the accuracy of biomass estimation. In addition, the formula also combines the effective reflection intensity I of the laser pulse eff (λi ,z), which reflects the reflective properties of vegetation at different wavelengths. Since the reflection intensity of the laser pulse is related to the structure and composition of the vegetation, the system can further optimize the biomass estimation by measuring the laser reflection intensity at multiple key wavelengths. Vegetation has different reflective properties at different wavelengths, especially in the near-infrared and visible light bands, where the reflective capabilities of leaves and trunks vary significantly. By weighting the laser reflection intensities at multiple wavelengths, the system can extract more spectral information related to biomass. i The weight coefficients for each wavelength are different. The reflectance characteristics at different wavelengths contribute differently to the biomass. Therefore, these weight coefficients can be adjusted according to different vegetation types and measurement conditions to improve the adaptability and accuracy of the model. This combination of multi-wavelength laser reflectance with chlorophyll content and leaf area density makes the calculation of biomass not only rely on the geometric information of the lidar, but also makes full use of the spectral reflectance characteristics of the vegetation. This innovative combination makes biomass estimation more accurate, especially in forest ecosystems with complex structures. This method can better capture the heterogeneity of vegetation, thereby improving the accuracy of carbon stock estimation.

[0086] Example 9: Assuming the preset carbon conversion coefficient is C coef , with a value of 0.45 to 0.5; the carbon storage C of the target forest area is calculated using the following formula:

[0087]

[0088] Among them, ρ soil (z) is the soil reflectivity at height z; σ(z) is the canopy backscattering cross section at height z; I eff (λ,z) ​​is the effective reflection intensity of the laser pulse at wavelength λ at height z; E abs (z) is the light absorption at height z, which represents the solar radiation energy absorbed by the canopy; E ref (z) is the reflected light energy at height z, which represents the radiation energy reflected back by the canopy; R(z) represents the propagation distance of the laser pulse from the transmitter to the canopy and then reflected back to the receiver.

[0089] Specifically, biomass B is the core of the entire calculation. It is derived from the previous model and reflects the total mass of forest vegetation at different height levels. However, biomass is not directly equivalent to carbon storage and needs to be calculated through the carbon conversion coefficient C. coefConvert it into carbon storage. The carbon conversion coefficient is a preset value, usually between 0.45 and 0.5, which represents the proportion of carbon in biomass, that is, about 45% to 50% of biomass is carbon. This coefficient is based on a large number of field experiments and studies, so it is introduced as a constant in the formula to ensure a reasonable conversion between biomass and carbon storage. The calculation of carbon storage is based on a comprehensive consideration of vegetation layers in the height z direction. At different height levels, the characteristics of vegetation vary significantly, especially between the crown and the trunk. In order to capture these changes more accurately, the chlorophyll content C is introduced into the formula. ab The chlorophyll content directly reflects the photosynthetic capacity of vegetation, while the leaf area density represents the total area of ​​leaves per unit volume. The combination of these two factors reflects the photosynthetic efficiency of vegetation and its leaf distribution at different heights. By combining these two parameters, the formula can more accurately describe the carbon absorption capacity of vegetation at different height layers.

[0090] Soil reflectivity ρ soil (z) and canopy backscattering cross section σ(z) are two other important parameters. Soil reflectivity is used to describe the light reflection characteristics between the ground surface and vegetation, especially at lower altitudes, the reflectivity of the soil may significantly affect the propagation and reflection of light. The canopy backscattering cross section σ(z) describes the scattering characteristics of light in the canopy, especially at the top of the canopy, where the multiple scattering effect of light will have an important impact on the measurement results of the lidar. The introduction of these two parameters ensures that the reflection and scattering characteristics of the laser pulse can be accurately captured and corrected when it passes through vegetation at different altitudes. The effective reflection intensity I of the laser pulse eff (λ,z) ​​is a key variable that reflects the reflection intensity of the laser pulse at a height of z at a wavelength of λ. By using multi-band lidar technology, the system can collect reflection data of forest vegetation at different wavelengths. The reflection characteristics of different wavelengths can help distinguish different types of vegetation and their health status. The advantage of lidar is that it can penetrate the tree canopy and provide structural information inside the vegetation, which makes the estimation of carbon storage more accurate. The reflection intensity of the laser pulse is closely related to the density, structure and reflection characteristics of the leaves. Therefore, by analyzing the laser reflection intensity at different height layers, the system can effectively infer the carbon storage of each layer of vegetation. In addition, the amount of light absorbed E abs (z) and the reflection amount E ref (z) is also an important component of the formula. abs (z) represents the solar radiation energy absorbed by vegetation at height z. The amount of light absorbed directly affects the photosynthesis capacity of vegetation, thereby affecting carbon fixation. ref(z) represents the reflected light energy at the same height, which represents the radiation that is not absorbed by vegetation. Therefore, the ratio between absorption and reflection is It is an important indicator to measure the efficiency of vegetation in absorbing light energy at different altitudes. Through this ratio, the system can evaluate the carbon absorption capacity at different altitudes and then deduce the carbon storage. The propagation distance R(z) of the laser pulse is also a variable that cannot be ignored. Since the laser pulse will experience energy attenuation in the process of being emitted from the transmitter to the canopy and then reflected back to the receiver, the inverse square term of the propagation distance R(z) This ensures that this energy attenuation effect can be reasonably taken into account. The farther the distance, the greater the energy loss of the laser signal. Therefore, when calculating carbon reserves, the distance must be corrected to ensure the accuracy of the reflection intensity.

[0091] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A forest carbon storage monitoring method based on drones, characterized in that: The method comprises: Step 1: Use the multi-band laser and hyperspectral camera carried by the drone to collect laser pulses and spectral data of the target forest area; construct a three-dimensional point cloud map of the target forest area based on the laser pulses; calculate the effective reflection intensity of the laser pulse based on the received laser pulses and the parameters of the multi-band laser; combine the radiation brightness in the spectral data and the effective reflection intensity of the laser pulse to perform atmospheric correction calculations to obtain the corrected surface reflectivity; Step 2: Calculate the terrain surface equation of the target forest area based on the 3D point cloud map; calculate the leaf area density of the vegetation in the target forest area based on the corrected surface reflectivity and the effective reflection intensity of the laser pulse; and calculate the chlorophyll content of the vegetation in the target forest area based on the leaf area density using the inversion function of the PROSAIL model; Step 3: Based on the preset biomass model and the preset carbon conversion coefficient, combined with the chlorophyll content, calculate the carbon storage of the target forest area.

2. The method for monitoring forest carbon reserves based on an unmanned aerial vehicle according to claim 1, characterized in that: In step 1, the multi-band laser receives laser pulses of different wavelengths and measures the intensity of these laser pulses after being reflected by vegetation and the ground at different heights, and infers the positions at different heights by calculating the time delay to generate a three-dimensional point cloud map.

3. The method for monitoring forest carbon reserves based on an unmanned aerial vehicle according to claim 2, characterized in that: In step 1, the effective reflection intensity of the laser pulse is calculated based on the received laser pulse and the parameters of the multi-segment laser using the following formula: Among them, I eff (λ,x,y,z) represents the effective reflection intensity of the laser pulse with wavelength λ at point (x,y,z); x is the X-axis coordinate, y is the Y-axis coordinate, and z is the Z-axis coordinate; I0(λ) represents the initial intensity of the laser pulse with wavelength λ when it is emitted; the atmospheric attenuation coefficient α(λ,s) describes the intensity loss caused by atmospheric components during the propagation of the laser pulse at wavelength λ from the starting point s=0 to s=z; σ(λ,x,y,z) is the backscattering cross section, which describes the ability of the target area to reflect the laser pulse at wavelength λ after the laser pulse reaches point (x,y,z) in the target area; R is the propagation distance, which represents the propagation distance of the laser pulse from the multi-band laser to the target area; A r is the effective receiving area of ​​the multi-band laser; s is the integral variable; η sys (λ) is the system optical efficiency, which represents the optical efficiency of the multi-band laser at wavelength λ. It reflects the performance loss of the optical components within the system and describes the efficiency of the laser pulse from emission, propagation to reception.

4. The method for monitoring forest carbon reserves based on an unmanned aerial vehicle according to claim 3, characterized in that: In step 1, atmospheric correction is performed using the following formula to calculate the corrected surface reflectance: Among them, ρ surface (λ,x,y) is the corrected surface reflectance at wavelength λ and spatial position (x,y); L sensor (λ,x,y) is the radiation brightness received by the hyperspectral camera at wavelength λ, which includes the surface reflected light and atmospheric scattered light; T ↑ (λ) and T ↓ (λ) are the atmospheric upward and downward transmittances, respectively; T ↑ (λ) is the transmittance of the atmosphere to the light of wavelength λ from the surface to the hyperspectral camera; T ↓ (λ) is the transmittance of the atmosphere to light of wavelength λ from the sun to the earth's surface; E0(λ) represents the irradiance of the sun outside the earth's atmosphere at wavelength λ, that is, the intensity of the light that can be irradiated to the earth's surface without being affected by the atmosphere; the solar zenith angle θ refers to the angle of the sun's rays relative to the vertical surface; E down (λ) is the atmospheric downward irradiance of light with wavelength λ, which is the light intensity of light with wavelength λ in the atmosphere returning to the surface after being scattered.

5. The method for monitoring forest carbon reserves based on an unmanned aerial vehicle according to claim 4, characterized in that: In step 2, the terrain surface equation is expressed by the following formula: Among them, S(x,y) represents the height of the ground at the position (x,y); is related to the surface gradient size The associated diffusion coefficient, anisotropic diffusion allows for stronger smoothing in flat areas, while suppressing smoothing at edges or steep areas, thus preserving terrain details, It decreases as the gradient size increases to reduce diffusion at the edges; is the surface divergence, which describes the rate and direction of change of the surface shape; is the surface gradient, which describes the inclination direction and slope of the surface; S0 is the initial terrain model generated based on the 3D point cloud map; λ is a weight parameter that controls the degree of fit between S(x, y) and S0. The larger the λ value, the closer S(x, y) is to S0; μ is the multi-band laser intensity weight; I eff (λ ground ,x,y,S) represents the effective reflection intensity of the ground laser pulse, which is the surface at wavelength λ ground The effective reflected intensity of the laser pulse under .

6. The method for monitoring forest carbon reserves based on an unmanned aerial vehicle according to claim 5, characterized in that: In step 2, leaf area density is calculated by the following formula: Where LAD(z) represents the leaf area density at height z; λ veg is the average wavelength of light; G(θ) is the leaf projection function, and the formula is as follows:

7. The method for monitoring forest carbon reserves based on an unmanned aerial vehicle according to claim 6, characterized in that: In step 2, multiple key wavelengths are selected, namely λ1, λ2, …, λ i ,…,λ n ; i is the i-th key wavelength; n is the number of key wavelengths; based on the leaf area density, the PROSAIL inversion model is used to calculate the chlorophyll content C of the vegetation in the target forest area at a height of z using the following formula: ab (z): in, is the inversion function of the PROSAIL model; ρ surface (λ1,λ2,…,λ i ,…,λ n ) indicates that the wavelengths are λ1,λ2,…,λ i ,…,λ n is the vector of the corrected surface reflectance corresponding to the time; LAI is the leaf area index, which is obtained by integrating LAD(z) in the z direction; θ v is the observation zenith angle; φ is the relative azimuth, which means the angle of the projection of the observation direction of the hyperspectral camera on the horizontal plane relative to the true north direction; angle β is the preset influence coefficient.

8. The method for monitoring forest carbon reserves based on an unmanned aerial vehicle according to claim 7, characterized in that: In step 3, the preset biomass model is expressed using the following formula: Where z0 is the height of the ground surface; z max is the highest height of the tree crown; a1 and a2 are the influence coefficients of the first-order derivative and second-order derivative of chlorophyll content related to plant type, respectively; b i is the preset weight coefficient of the i-th key wavelength; I eff (λ i ,z) represents the wavelength λ i is the effective reflection intensity of the laser pulse at height z; B is the biomass.

9. The method for monitoring forest carbon reserves based on an unmanned aerial vehicle according to claim 8, characterized in that: Assume the preset carbon conversion factor is C coef , with a value of 0.45 to 0.5; the carbon storage C of the target forest area is calculated using the following formula: Among them, ρ soil (z) is the soil reflectivity at height z; σ(z) is the canopy backscattering cross section at height z; I eff (λ,z) ​​is the effective reflection intensity of the laser pulse at wavelength λ at height z; E abs (z) is the light absorption at height z, which represents the solar radiation energy absorbed by the canopy; E ref (z) is the reflected light energy at height z, which represents the radiation energy reflected back by the canopy; R(z) represents the propagation distance of the laser pulse from the transmitter to the canopy and then reflected back to the receiver.

Images