A forest vertical structure information inversion method and system based on radar interferometry

Abstract

The application provides a forest vertical structure information inversion method and system based on radar interferometry, and the method comprises the following steps: performing small-look angle processing on high-resolution SAR interferometry data with a spatial baseline, calculating small-look angle InSAR phase height, extracting the statistical distribution of the small-look angle interferometric phase height in a set spatial window, and generating a small-look angle interferometric phase height histogram; based on the statistical characteristics of the small-look angle interferometric phase height histogram, establishing a regression estimation model of the ground elevation under the forest, determining the model parameters, and inverting the terrain; extracting two crown top indicators, fusing the two indicator results by setting a threshold, and generating the final forest vertical structure information. The application has the advantages that: only single-baseline and single-polarization radar interferometric data are used, more information inside the crown and under the forest can be captured, and the sensitivity to the forest vertical structure is improved; the forest vertical profile, tree height and terrain can be simultaneously inverted.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of forest remote sensing mapping, and particularly relates to a forest vertical structure information inversion method and system based on radar interferometry, which is a technical method for inverting forest vertical structure information (including vertical structure profile, forest height and underlying surface terrain elevation) by using high-resolution (such as spaceborne L-band bistatic monopolar single baseline) interferometric synthetic aperture radar (InSAR) data with spatial baseline, combined with spaceborne lidar data calibration. BACKGROUND

[0002] Forest, as an important part of the terrestrial ecosystem, the medium-high resolution (<100m) measurement of its biophysical parameters (such as tree height, biomass) is of great significance for monitoring the heterogeneity of tropical forests, estimating carbon storage and evaluating ecosystem functions. Spaceborne remote sensing technology is the main means to achieve the above goals. The commonly used observation technologies include optical sensors, lidar and synthetic aperture radar (SAR) systems.

[0003] Spaceborne lidar (such as NASA's GEDI, ICESat-2 / ATLAS) can provide high-precision discrete point tree height observation, but its sampling mode is discrete point, which is difficult to realize large-scale continuous mapping; optical remote sensing sensors (such as Landsat series) rely on vegetation reflectance, and have insufficient penetration ability for dense forests, and cannot obtain the structure information below the canopy; the bistatic interferometric measurement technology in synthetic aperture radar system has unique advantages in forest three-dimensional structure inversion due to its microwave penetration ability, accurate height measurement ability and sensitivity to vertical structure. For example, the German TanDEM-X satellite adopts a bistatic interferometric mode, but the system uses X-band (~3.1cm), which has limited penetration ability for forests, and most of the data uses single baseline, single polarization observation mode, which makes the polarization interference or tomography method, which is the mainstream of measuring forest tree height, understory terrain and vertical structure profile, not applicable.

[0004] The Lutan-1 satellite mission is the world's first spaceborne L-band bistatic InSAR (Interferometric Synthetic Aperture Radar) mission. Its primary data acquisition mode is the same as the TanDEM-X mission, using a single-polarization, single-baseline mode, providing a new data source for forest vertical structure detection. The L-band (wavelength 15-30 cm) has stronger forest penetration capabilities, capturing more information about the canopy interior and understory. However, there is currently a lack of forest vertical structure inversion methods based on this type of single-polarization, single-baseline bistatic interferometric radar data. Existing methods mostly rely on physical scattering model assumptions and have poor adaptability to single-polarization, single-baseline data. Therefore, there is an urgent need to develop suitable technical solutions to fully leverage the advantages of data acquired by Lutan-1 and satellites with similar observation systems. Summary of the Invention

[0005] The purpose of this application is to overcome the shortcomings of existing technologies in terms of continuous mapping, single-baseline single-polarization data adaptability, and inversion accuracy.

[0006] To achieve the above objectives, this application proposes a method for inverting forest vertical structure information based on radar interferometry, including:

[0007] Step S1: Perform small-look-number processing on high-resolution SAR interferometric data with spatial baselines (such as L-band bistation single-polarization single-baseline), calculate the small-look-number InSAR phase height using the phase-height conversion formula, extract the statistical distribution of the small-look-number interferometric phase height within a set spatial window, and generate a small-look-number interferometric phase height histogram.

[0008] Step S2: Based on the statistical characteristics of the small-view-number interferometric phase height histogram, establish a forest understory ground elevation regression estimation model, and use differentiated strategies to determine model parameters for different laser sample scenarios to invert the terrain;

[0009] Step S3: Combining the small-view-number interferometric phase height histogram with lidar waveform analysis, extract two canopy top indicators from the histogram, fuse the results of the two indicators by setting a threshold, and generate the final forest vertical structure information.

[0010] As an improvement to the above method, the small number of views is 1-4 views.

[0011] As an improvement to the above method, the phase-height conversion formula is as follows:

[0012] h φ =Δφ / κ z

[0013] Among them, h φκ represents the small-view InSAR phase height; Δφ represents the interferometric phase difference. z The vertical wavenumber of the interference.

[0014] As an improvement to the above method, the set space window is 20-50m.

[0015] As an improvement to the above method, the forest understory ground elevation regression estimation model is as follows:

[0016] h g =α·μ-β·σ

[0017] Among them, h g α represents the ground elevation under the forest canopy; α and β are regression coefficients, i.e., model parameters; μ and σ are the mean and standard deviation of the small-look-number interferometric phase height histogram, respectively.

[0018] As an improvement to the above method, the step of determining model parameters using a differentiated strategy for different laser sample scenarios includes:

[0019] Set α = 1, allowing β to vary with space;

[0020] When the number of laser sampling points corresponding to a single radar image is greater than the set threshold for the number of sampling points, the statistical relationship between the coefficient β and the tree height is established based on the tree height measured by the laser radar, and the coefficient β is determined by a linear regression model.

[0021] When the number of laser sampling points corresponding to a single-view radar image is less than or equal to the set sampling point number threshold, a statistical relationship between coefficient β and σ is established, using the small-view interferometric phase height standard deviation σ as a characteristic.

[0022] As an improvement to the above method, step S2 further includes: optimizing the inverted terrain by combining the lowest peak value of the forest understory terrain in the histogram, including:

[0023] For bare land and medium-to-high vegetation, if there is a minimum peak in the small-view histogram within the range of the topographic inversion value plus or minus a set range, then the minimum peak value is considered to correspond to the optimized topographic inversion value.

[0024] For low vegetation, if the histogram still has values ​​but no lowest peak in the range of the topographic inversion values ​​plus or minus a set range, then the position where the second derivative of the histogram in this range is 0 is taken as the optimized topographic inversion value.

[0025] For ultra-high vegetation, if the histogram has no value within the range of the topographic inversion value plus or minus a set range, the topographic inversion value is directly used as the optimized topographic inversion value.

[0026] As an improvement to the above method, the two canopy top indices include: canopy top indices at RH98 and 3dB attenuation point.

[0027] The application also provides a forest vertical structure information inversion system based on radar interferometry, which is realized based on the above method, and the system comprises:

[0028] A small-look-angle interferometric phase height histogram generation module is configured to perform small-look-angle processing on high-resolution (e.g., L-band bistatic single-polarization single-baseline) SAR interferometric measurement data with a spatial baseline, calculate small-look-angle InSAR phase height through a phase-height conversion formula, extract the statistical distribution of small-look-angle interferometric phase height within a set spatial window, and generate a small-look-angle interferometric phase height histogram.

[0029] A terrain inversion module is configured to establish an under-forest ground elevation regression estimation model based on the statistical characteristics of the small-look-angle interferometric phase height histogram, determine model parameters by using a differentiated strategy for different laser sample scenes, and invert terrain.

[0030] A forest vertical structure information generation module is configured to combine the small-look-angle interferometric phase height histogram and a laser radar waveform analysis method, extract two crown top indicators in the histogram, fuse the results of the two indicators by setting a threshold, and generate final forest vertical structure information.

[0031] Compared with the prior art, the application has the following advantages:

[0032] 1. The application is based on high-resolution (particularly star-borne L-band bistatic single-polarization single-baseline) SAR interferometric measurement data with a spatial baseline. L-band has strong penetration ability for forests and can capture more internal crown information, thereby improving the sensitivity to forest vertical structure.

[0033] 2. The proposed “small-look-angle interferometric phase height histogram” method does not rely on physical scattering model assumptions, has strong adaptability to single-polarization and single-baseline data, and reduces the demand for complex parameters.

[0034] 3. The method can realize large-scale continuous mapping. Forest vertical structure information products can be generated based on data obtained by the domestic Land Explorer 1 satellite (the first star-borne L-band bistatic interferometric radar satellite in the world) or other satellites with similar observation systems, thereby breaking through the limitations of discrete sampling of star-borne laser radars.

[0035] 4. The inversion precision is high. For tropical forests with a height of up to 45 m, the tree height inversion precision is about 5 m (relative error 10-15%), and the terrain inversion error varies with slope as 3 m (<2°), 4 m (2°-6°), 7 m (6°-25°), and 9 m (>25°), thereby meeting the Chinese DEM precision standard.

[0036] 5. The method can simultaneously invert forest vertical profile, tree height, and terrain, thereby providing multi-dimensional data support for forest biomass estimation, ecological change monitoring, and the like. Attached Figure Description

[0037] Figure 1 The flowchart shown is a method for inverting forest vertical structure information based on radar interferometry.

[0038] Figure 2 The diagram shows a schematic of the underlying surface topography (DTM) inversion strategy.

[0039] Figure 3 The diagram shown is a schematic diagram of the fusion of forest height inversion indicators.

[0040] Figure 4 The image shows histograms of bare land and forest generated using real data from the LuTan-1 satellite.

[0041] Figure 5 The diagram shown is a schematic diagram of the optimized terrain inversion generated using real data from the LuTan-1 satellite, which distinguishes four types of histograms: bare land, low vegetation, medium-high vegetation, and very high vegetation. Detailed Implementation

[0042] The technical solution of this application will be described in detail below with reference to the accompanying drawings.

[0043] This application aims to utilize high-resolution SAR interferometric data with existing spatial baselines (such as spaceborne L-band bistation single-polarization, single-baseline) and combine it with spaceborne lidar data for calibration, to achieve high-precision, large-scale inversion of forest vertical structure information (including vertical structure profiles, forest height, and underlying surface topographic elevation) through the "small-viewpoint interferometric phase-height histogram" method.

[0044] The method described in this application is not only applicable to spaceborne L-band bistatic radar interferometric data, such as China's LuTan-1. It can be applied to any advanced radar interferometric satellite, whether bistatic (one transmitter, two receivers) or heavy orbit (single transmitter, single receiver) interferometric radar satellite, as long as there is a certain spatial baseline between the two observations performing the interferometry, thus being sensitive to the vertical structure information of the forest, and possessing high resolution (in this application, high resolution refers to a resolution less than 6m) to support radar waves penetrating forest gaps (forest windows).

[0045] Example 1

[0046] like Figure 1 As shown, the forest vertical structure information inversion method based on radar interferometry proposed in this application includes:

[0047] Step 1: Generate a small-look-number interferometric phase height histogram: Perform small-look-number (1-4 looks) processing on high-resolution (less than 6m) bistatic L-band InSAR single-look complex data (SLC) to balance noise suppression and spatial heterogeneity preservation; calculate the small-look-number InSAR phase height using the phase-height conversion formula, the formula is as follows:

[0048] h φ =Δφ / κ z (1)

[0049] Where Δφ is the interference phase difference, κ z The vertical wavenumber of the interferometer is used. By combining fuzzy height (HoA) correction for periodic jumps and extracting the statistical distribution of the small-aperture interferometric phase height within a relatively large spatial window (20-50m), a histogram is generated as a representation of the forest's vertical structure profile. This histogram can be compared with the waveform of a full-waveform lidar (such as GEDI) to verify its ability to capture vertical stratification features.

[0050] Step 2: Invert the underlying surface topography (DTM): such as Figure 2 As shown, based on the statistical characteristics (mean μ, standard deviation σ) of the small-look-number interferometric phase height histogram, a regression estimation model for forest understory ground elevation is established:

[0051] h g =α·μ-β·σ (2)

[0052] Among them, h gLet α be the ground elevation under the forest canopy, and (α, β) be the regression coefficients. The following assumption is that α = 1, while allowing β to vary spatially. A differentiated strategy is adopted for different laser sample scenarios. In scenarios with abundant laser samples (tens of thousands of laser sampling points corresponding to a single radar image), a statistical relationship (positive correlation) between coefficient β and tree height is established using the tree height of the lidar as a feature. The coefficient β is determined through a linear regression model (β = 0 for bare ground, β ≈ 2 for tall trees). In scenarios with limited laser samples (only a few hundred to a few thousand laser sampling points corresponding to a single radar image), a statistical relationship (positive correlation) between coefficient β and σ is established using the standard deviation of the small-look-number interferometric phase height as a feature, obtaining the initial values ​​for terrain inversion. Combined with the lowest peak value of the forest canopy terrain in the histogram, four histogram types—bare ground, low vegetation, medium-high vegetation, and very high vegetation—are distinguished to optimize terrain inversion. Specifically: For bare land and medium-to-high vegetation, in the interval near the initial value of topographic inversion (with a small amount added or subtracted), if the small-view histogram has a minimum peak, then the minimum peak value (with a first derivative of 0) is considered to correspond to the optimized topographic inversion value; for low vegetation, in the interval near the initial value of topographic inversion, if the histogram still has values ​​but does not have a minimum peak, then the position where the second derivative of the histogram in this interval is 0 is taken as the optimized topographic inversion value; for very tall vegetation, in the interval near the initial value of topographic inversion, if the histogram has no values, then the initial value is directly used to replace the optimized topographic inversion value.

[0053] Step 3: Invert forest height: such as Figure 3 As shown, based on the small-view-number interferometric phase height histogram and drawing on the waveform analysis method of lidar, two canopy top indices are extracted: RH98 (98% relative height, referring to the lidar RH98 definition) and the 3dB attenuation point (referring to the power attenuation point of the first peak of the canopy top in the histogram). The results of the two indices are fused by thresholding, for example, using 7.5m as the threshold. Below this value, the 3dB attenuation point result is used, and above this value, the RH98 result is used to improve the inversion accuracy.

[0054] The forest vertical structure information inversion method based on radar interferometry proposed in this application is compatible with data from the domestically produced LuTan-1 satellite and can be extended to other similar high-resolution InSAR data with spatial baselines. The method is verified below using LuTan-1 satellite data as an example:

[0055] Step 1: Generate a small-aperture interferometric phase height histogram

[0056] Interferograms are generated by performing small-view (e.g., 2x2, i.e., 4 views) processing on the single-view complex data (SLC) of LuTan-1. The phase is converted to height and a phase height histogram is generated by extracting the data within the spatial window (e.g., 50m). The size of the histogram segment interval (which can be set to 0.7-1m) needs to balance the vertical profile details and noise. The generated histogram can be compared with the waveform of the spaceborne lidar (e.g., GEDI), showing that it can effectively capture the vertical stratification characteristics of the forest.

[0057] Histograms of bare land and forest generated using real data from LuTan-1 are shown below. Figure 4 As shown, the histogram of bare land exhibits a more concentrated narrow Gaussian distribution, while the histogram of forest vegetation shows a more pronounced asymmetric and stratified structure.

[0058] Step 2: Invert the underlying surface topographic elevation (DTM)

[0059] Based on the statistical characteristics (mean μ, standard deviation σ) of the small-view-number interferometric phase height histogram, a regression estimation model for forest ground elevation is established, as shown in formula (2).

[0060] Based on the sample size of the spaceborne lidar in the study area, differentiated processing is performed. For areas with sufficient lidar samples, a statistical regression model of coefficient β and tree height is established with lidar tree height as a feature (β = 0 for bare land, β ≈ 2 for tall trees). The spatial distribution of coefficient β is determined by the interpolation result of tree height of spaceborne lidar, and then substituted into the model formula (2) to calculate DTM. For areas with sparse lidar, a statistical regression model of coefficient β and σ is established with small-look-number interferometric phase height standard deviation σ as a feature for initial estimation. The terrain inversion is optimized by combining the lowest peak value of the understory terrain in the histogram, and four histogram types are distinguished: bare land, low vegetation, medium-high vegetation, and ultra-high vegetation.

[0061] like Figure 5 As shown, the specific distinctions between the four histogram types are as follows: For bare land (Type I: σ is relatively small, generally around 2-3 meters) and medium-high vegetation (Type III: σ is relatively large), in the interval near the initial value of terrain inversion (adding or subtracting a small amount, see the red dashed line segment in the figure), if the small-view histogram has a minimum peak (corresponding to the red dot in the figure) above the noise threshold (corresponding to the green dashed line in the figure), then the minimum peak value (first derivative is 0) is considered to correspond to the optimized terrain inversion value; For low vegetation (Type II: σ is slightly smaller, slightly larger than Type I), in the interval near the initial value of terrain inversion, the histogram still has values ​​above the noise threshold but does not have a minimum peak, then the position where the second derivative of the histogram in this interval is 0 is taken as the optimized terrain inversion value; For very tall vegetation (Type IV: σ is relatively large, similar to Type III), in the interval near the initial value of terrain inversion, the histogram has no values ​​above the noise threshold, then the initial value is directly used to replace the optimized terrain inversion value.

[0062] Step 3: Invert forest height

[0063] Based on the small-view-number interferometric phase height histogram, and drawing on lidar waveform analysis methods, two canopy top indices are extracted from the histogram: RH98 (98% relative height, referring to the lidar RH98 definition) and the 3dB attenuation point (referring to the power attenuation point of the first peak at the canopy top of the histogram). The results of the two indices are fused by thresholding, for example, using 7.5m as the threshold. Below this value, the 3dB attenuation point result is used, and above this value, the RH98 result is used, in order to improve the inversion accuracy.

[0064] Example 2

[0065] This application also provides a forest vertical structure information inversion system based on radar interferometry, implemented using the above method, the system comprising:

[0066] The module for generating small-look-number interferometric phase height histograms is used to perform small-look-number processing on high-resolution SAR interferometric measurement data with spatial baselines (such as spaceborne L-band bistationary single-polarization single-baseline data). It calculates the small-look-number InSAR phase height using the phase-height conversion formula, extracts the statistical distribution of the small-look-number interferometric phase height within a set spatial window, and generates a small-look-number interferometric phase height histogram.

[0067] The terrain inversion module is used to establish a forest understory ground elevation regression estimation model based on the statistical characteristics of the small-look-number interferometric phase height histogram. Differentiated strategies are used to determine model parameters for different laser sample scenarios to invert the terrain.

[0068] A module for generating forest vertical structure information is used to combine small-viewpoint interferometric phase height histograms with lidar waveform analysis methods to extract two canopy top indicators from the histograms. By setting thresholds, the results of the two indicators are fused to generate the final forest vertical structure information.

[0069] This application may also provide a computer device, including: at least one processor, memory, at least one network interface, and a user interface. The various components in this device are coupled together via a bus system. It is understood that the bus system is used to implement communication between these components. In addition to a data bus, the bus system also includes a power bus, a control bus, and a status signal bus.

[0070] The user interface can include a display, keyboard, or clicking device. Examples include a mouse, trackball, touchpad, or touchscreen.

[0071] It is understood that the memory in the embodiments disclosed in this application may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory may be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate Synchronous DRAM (DDRSDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM). The memories described herein are intended to include, but are not limited to, these and any other suitable types of memory.

[0072] In some implementations, the memory stores elements such as executable modules or data structures, or subsets thereof, or extended sets thereof: operating systems and applications.

[0073] The operating system includes various system programs, such as the framework layer, core library layer, and driver layer, used to implement various basic business functions and handle hardware-based tasks. The application programs include various applications, such as media players and browsers, used to implement various application functions. Programs implementing the methods of the embodiments of this disclosure can be included in the application programs.

[0074] In the above embodiments, the processor can also invoke programs or instructions stored in memory, specifically programs or instructions stored in an application program, for the following purposes:

[0075] Follow the steps described above.

[0076] The above methods can be applied to or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above methods can be completed by integrated logic circuits in the processor's hardware or by software instructions. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the disclosed methods, steps, and logic block diagrams. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the disclosed methods can be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above methods.

[0077] It is understood that the embodiments described in this application can be implemented using hardware, software, firmware, middleware, microcode, or a combination thereof. For hardware implementation, the processing unit can be implemented in one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers, microprocessors, other electronic units for performing the functions described in this application, or combinations thereof.

[0078] For software implementation, the technology of this application can be implemented by executing the functional modules (e.g., procedures, functions, etc.) of this application. The software code can be stored in memory and executed by a processor. The memory can be implemented in the processor or externally.

[0079] This application may also provide a non-volatile storage medium for storing a computer program. When the computer program is executed by a processor, it can implement the steps in the above method embodiments.

[0080] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application and are not intended to limit it. Although this application has been described in detail with reference to the embodiments, those skilled in the art should understand that modifications or equivalent substitutions to the technical solutions of this application do not depart from the spirit and scope of the technical solutions of this application, and should all be covered within the scope of the claims of this application.

Claims

1. A method for inverting forest vertical structure information based on radar interferometry, comprising: Step S1: Perform small-look-number processing on high-resolution SAR interferometric data with spatial baselines, calculate the small-look-number InSAR phase height using the phase-height conversion formula, extract the statistical distribution of the small-look-number interferometric phase height within a set spatial window, and generate a small-look-number interferometric phase height histogram. Step S2: Based on the statistical characteristics of the small-view-number interferometric phase height histogram, establish a forest understory ground elevation regression estimation model, and use differentiated strategies to determine model parameters for different laser sample scenarios to invert the terrain; Step S3: Combining the small-view-number interferometric phase height histogram with lidar waveform analysis, extract two canopy top indices from the histogram, fuse the results of the two indices by setting a threshold, and generate the final forest vertical structure information; The regression estimation model for forest understory elevation is as follows: ； in, This refers to the elevation of the forest floor. These are the regression coefficients, i.e., the model parameters; and These are the mean and standard deviation of the small-aperture interferometric phase height histogram, respectively. The method for determining model parameters using a differentiated strategy for different laser sample scenarios includes: set up ,allow Changes with space; When the number of laser sampling points corresponding to a single-scene radar image exceeds a set threshold for the number of sampling points, a coefficient is established based on the tree height measured by the laser radar. The statistical relationship between tree height and tree height was determined using a linear regression model. ; When the number of laser sampling points corresponding to a single radar image is less than or equal to the set threshold for the number of sampling points, the standard deviation of the small-view interferometric phase height is used. Establish coefficients based on characteristics. and Statistical relationships.

2. The forest vertical structure information inversion method based on radar interferometry according to claim 1, characterized in that, The small number of views is 1-4.

3. The forest vertical structure information inversion method based on radar interferometry according to claim 1, characterized in that, The phase-height conversion formula is as follows: ； in, The small-view InSAR phase height; For the interference phase difference, The vertical wavenumber of the interference.

4. The forest vertical structure information inversion method based on radar interferometry according to claim 1, characterized in that, The set space window is 20-50m.

5. The forest vertical structure information inversion method based on radar interferometry according to claim 1, characterized in that, Step S2 further includes: optimizing the inverted terrain by combining the lowest peak value of the understory terrain in the histogram, including: For bare land and medium-to-high vegetation, if there is a minimum peak in the small-view histogram within the range of the topographic inversion value plus or minus a set range, then the minimum peak value is considered to correspond to the optimized topographic inversion value. For low vegetation, if the histogram still has values ​​but no lowest peak in the range of the topographic inversion values ​​plus or minus a set range, then the position where the second derivative of the histogram in this range is 0 is taken as the optimized topographic inversion value. For ultra-high vegetation, if the histogram has no value within the range of the topographic inversion value plus or minus a set range, the topographic inversion value is directly used as the optimized topographic inversion value.

6. The forest vertical structure information inversion method based on radar interferometry according to claim 1, characterized in that, The two canopy top indices include: the canopy top indices at RH98 and the 3dB decay point.

7. A forest vertical structure information inversion system based on radar interferometry, implemented according to the method described in any one of claims 1-6, characterized in that, The system includes: The module for generating small-look-number interferometric phase height histograms is used to perform small-look-number processing on high-resolution SAR interferometric data with spatial baselines. It calculates the small-look-number InSAR phase height using the phase-height conversion formula, extracts the statistical distribution of the small-look-number interferometric phase height within a set spatial window, and generates a small-look-number interferometric phase height histogram. The terrain inversion module is used to establish a forest understory ground elevation regression estimation model based on the statistical characteristics of the small-look-number interferometric phase height histogram. It employs differentiated strategies to determine model parameters for different laser sample scenarios and inverts the terrain. A module for generating forest vertical structure information is used to combine small-viewpoint interferometric phase height histograms with lidar waveform analysis methods to extract two canopy top indicators from the histograms. By setting thresholds, the results of the two indicators are fused to generate the final forest vertical structure information.

Images