Space laser radar forest structure data correction method and system

Abstract

The present application relates to a space laser radar forest structure data correction method and system, and belongs to the technical field of space laser remote sensing and forest resource monitoring. In view of the four types of errors of GEDI data, such as geometric positioning deviation, slow variation distortion, terrain widening effect and crown structure influence, the present application adopts hierarchical layout strategy to layout UAV control points, establishes the same point relationship through waveform registration and calculates the weight; based on the regular grid control points, the overall geometric correction is carried out; based on the encryption control points, the local distortion correction is carried out; the physical compensation model is established to compensate the terrain widening effect; the statistical regression model is established based on the crown structure parameters and GEDI signal quality index to correct the crown height residual error; finally, the high-precision crown height convergence result is obtained through spatial smoothing processing, and the corrected data is output. The present application realizes the gradual accuracy improvement from the whole to the local and from the geometry to the height, has low cost, high precision and wide application range.

Description

Technical Field

[0001] This invention relates to the field of space laser remote sensing and forest resource monitoring technology, and in particular to a method and system for correcting forest structure data using space lidar. Background Technology

[0002] The Global Ecosystem Dynamics Investigation (GEDI) spaceborne lidar, carried on the International Space Station, can acquire parameters such as forest canopy height, vertical structure, and ground elevation on a global scale, and is of great value in carbon storage accounting, forest resource surveys, and ecosystem research.

[0003] However, due to the limited accuracy of orbit and attitude calculations of the observation platform, the significant influence of terrain and canopy structure on laser waveforms, and differences in signal quality, GEDI data suffers from various types of errors in large-scale applications, directly affecting its reliability and widespread use. Specifically, these errors manifest in several ways: First, geometric and positioning system offsets: due to errors in orbit and attitude calculations and limitations in the accuracy of satellite-to-ground reference conversion, the GEDI footprint as a whole exhibits translational, rotational, or scale deviations, leading to inconsistencies between the observation point location and the actual ground features. Second, gradual geometric distortion: during orbital observations, small cumulative errors in attitude and geometric parameters cause gradual drift or surface distortion within the orbital segment, distorting the geometric relationships in local areas, which cannot be corrected by overall translation alone. Third, waveform broadening caused by terrain: in sloping and undulating areas, the elevation difference within the footprint coverage area causes the echo waveform to be elongated along the elevation direction, resulting in a systematic deviation of the RH height index from the true canopy top or ground height. Fourth, the influence of canopy structure: differences in forest coverage, thickness, and volume among different forest types alter laser penetration and echo energy distribution, and under weak signal conditions, may even lead to missing ground echoes, thus causing systematic biases in tree height estimation. Existing GEDI data correction methods mostly rely on digital elevation models (DEMs) or a small number of sample plots for overall deviation correction, or use affine transformation and empirical regression models. However, these methods usually cannot simultaneously solve multiple types of errors such as systematic migration, gradual distortion, terrain broadening, and canopy structure differences. They also lack reasonable unmanned aerial vehicle (UAV) control point deployment strategies, resulting in insufficient accuracy of GEDI data in large-scale applications.

[0004] Therefore, it is necessary to develop a new method that can effectively solve the above-mentioned types of errors under the condition of limited UAV cost, and realize the high-precision application of GEDI in the province and even a larger area. Summary of the Invention

[0005] This invention addresses the issue of insufficient accuracy of GEDI data in large-scale applications by proposing a spatial lidar forest structure data correction method and system. This method can resolve four types of errors in GEDI data: geometric and positioning system offset, gradually varying geometric distortion, waveform broadening caused by terrain, and canopy structure influence, thereby reducing systematic errors in forest height and terrain information.

[0006] To solve the above problems, the present invention adopts the following technical solution:

[0007] A method for correcting forest structure data using space lidar, comprising the following steps:

[0008] Step 1: According to the hierarchical deployment strategy, deploy UAV control points in a regular grid within the target area, and densify the deployment of UAV control points in complex terrain areas. After acquiring the UAV data of all the UAV control points, perform waveform registration between the UAV data and the GEDI footprint to establish the relationship between corresponding points and calculate the corresponding weights.

[0009] Step 2: Based on the UAV control points of the regular grid, perform overall geometric correction on the planar coordinates of the GEDI footprint according to the track segment or time period, including overall translation, rotation and scale adjustment, to obtain the first-level corrected coordinates;

[0010] Step 3: Based on the densely deployed UAV control points, perform local bending and distortion correction on the GEDI footprint after overall geometric correction to obtain the secondary corrected coordinates;

[0011] Step 4: Based on the terrain slope and laser beam direction, establish a physical compensation model to compensate for the terrain widening effect in the canopy height of the GEDI data, and obtain the first corrected canopy height;

[0012] Step 5: Based on the canopy structure parameters and GEDI signal quality index, establish a statistical regression model to correct the canopy height residual and obtain the secondary corrected canopy height;

[0013] Step 6: Eliminate the local scattered residuals of the secondary corrected canopy height using a spatial smoothing method to obtain the canopy height convergence result. Finally, output the corrected GEDI forest structure data, including the secondary corrected coordinates and the canopy height convergence result.

[0014] Meanwhile, the present invention also proposes a space lidar forest structure data correction system, which includes a processor and a memory. The memory stores a computer program, and the processor executes the computer program to implement the steps of the method described above.

[0015] This invention proposes a method and system for correcting forest structure data using a space lidar system. Combining a regular grid layout with densely deployed UAV control points in complex terrain areas (e.g., mountainous regions), it establishes a step-by-step elimination approach to address four main types of errors in GEDI data during large-scale applications: geometric offset, gradual distortion, terrain broadening, and canopy structure influence. Through overall geometric correction, local distortion correction, terrain broadening effect compensation, canopy and signal statistical correction, and spatial smoothing, this invention achieves a comprehensive improvement in the accuracy of GEDI data from global to local levels, and from geometric to height dimensions.

[0016] Compared with existing methods, the present invention has the following advantages:

[0017] (1) Taking into account both the whole and the local: The large-scale systematic error is eliminated by regular grid UAV control points, while the local distortion under complex terrain is corrected by densely deploying UAV control points in mountainous and other complex terrain areas, so as to achieve the dual guarantee of "consistency in the whole area + fine local conditions".

[0018] (2) Combining physics and statistics: both a physical compensation model is established using the geometric relationship between the terrain slope and the laser beam direction to compensate for the terrain widening effect, and a statistical regression model is established using canopy structure parameters and GEDI signal quality indicators to correct the residuals, ensuring that the error correction has both theoretical support and flexible adaptability.

[0019] (3) Efficient control point deployment strategy: Under the constraint of UAV cost, the hierarchical deployment strategy of "regular grid + densification in complex terrain areas" significantly improves the utilization efficiency of control points and achieves high-precision correction of large-scale GEDI data with less UAV data.

[0020] (4) Step-by-step correction process: The six-step correction approach is adopted, from overall adjustment to local distortion correction, then to terrain widening effect compensation, canopy and signal statistical correction and spatial smoothing, step by step to eliminate various errors, and finally output GEDI canopy height and planar positioning data with high accuracy and high consistency.

[0021] (5) Wide range of applications: This invention is not only applicable to GEDI data correction within a province, but can also be extended to other regions, especially large-scale surface scenes where mountains and plains coexist, as well as regions with different forest types and terrain conditions, and has good versatility and promotion value.

[0022] In summary, by introducing UAV control points and establishing a hierarchical deployment and step-by-step correction technical architecture, this invention can significantly improve the geometric positioning accuracy and elevation accuracy of GEDI data, providing reliable data support for applications such as forest resource monitoring, carbon storage assessment, and ecosystem research. Attached Figure Description

[0023] Figure 1 This is a flowchart of the space lidar forest structure data correction method described in this invention;

[0024] Figure 2 This is a schematic diagram of the layout of UAV control points in Jilin Province in an embodiment of the present invention, where blue dots represent UAV control points in a regular grid and red dots represent UAV control points with a denser layout. Detailed Implementation

[0025] The technical solution of the present invention will now be described in detail with reference to the accompanying drawings and preferred embodiments.

[0026] like Figure 1 As shown in the figure, this embodiment provides a spatial lidar forest structure data correction method based on UAV lidar data, which specifically includes the following steps one to six.

[0027] Step 1: UAV Control Point Deployment and Corresponding Point Matching. Following a hierarchical deployment strategy, UAV control points are deployed in a regular grid within the target area, with increased density in complex terrain areas. After acquiring UAV data from all control points, the UAV data is waveform-registered with the GEDI footprint to establish corresponding point relationships, and the corresponding weights are calculated. The purpose of this step is to establish comparable "corresponding" observations between GEDI and UAV, used for geometric positioning and canopy height correction in subsequent steps, providing true reference values ​​for eliminating systematic migration, local distortion, slope effects, and canopy influences.

[0028] Optionally, in the hierarchical deployment strategy, the spacing between UAV control points in the regular grid is 10 to 50 km, while the spacing between densely deployed UAV control points is 5 to 10 km. Complex terrain areas refer to mountainous areas with terrain undulations exceeding 300 meters.

[0029] The UAV data acquired in this step are reference features generated from the UAV point cloud, including:

[0030] With GEDI footprint diameter (Its value is approximately 25m) is used as a window, generated from the UAV point cloud;

[0031] Digital Elevation Model and Slope Slope aspect;

[0032] Canopy Height Model (CHM), Canopy Parameters (Coverage) ,thickness ,volume wait);

[0033] Analog waveform It is obtained by superimposing the GEDI system response and energy normalization.

[0034] Furthermore, the waveform registration process includes:

[0035] Using the GEDI footprint diameter as a window, extract the corresponding analog waveform from the UAV data;

[0036] Within the preset micro-shift search window, by maximizing the normalized cross-correlation, the shift that makes the simulated waveform most similar to the GEDI waveform is determined as the corresponding point matching result, and the maximum normalized cross-correlation number is recorded.

[0037] For the GEDI footprints ( In the micro-movement search window Internal calculation of normalized correlation coefficient :

[0038]

[0039] in, and They represent the first Both the observed waveforms of the GEDI footprint and the simulated waveforms generated from UAV data represent the vertical echo power as a function of elevation. A changing function; They represent the first Mean and standard deviation of observed waveforms for each GEDI footprint; They represent the first The mean and standard deviation of each UAV simulated waveform. The translation amount (pixels / meter) to make the two waveforms most similar. And record the maximum normalized cross-correlation number. For the first The maximum correlation coefficient of the waveform corresponding to each GEDI footprint, i.e., the maximum normalized cross-correlation coefficient. In this step, both the UAV control points of the regular grid and the densely deployed UAV control points participate in the establishment of corresponding points. Corresponding point matching provides both geometric reference ( It also provides quality metrics. This lays the foundation for subsequent weighted solutions.

[0040] After establishing the relationship between corresponding points, the weights used for subsequent weighted adjustment are calculated using the following formula:

[0041]

[0042] in, For the first The weights corresponding to each GEDI footprint; The GEDI signal quality index represents the first, second, and third signals, respectively. Sensitivity, received energy, and number of samples for each GEDI footprint; These are all empirical indices, with values ​​ranging from 0.5 to 1.0; This is a truncation function. These are the lower and upper thresholds for weight truncation, respectively.

[0043] Step one involves deploying UAV lidar measurement points in regular grids and densely populated mountainous locations to obtain high-precision terrain and canopy information. These UAV data are then aligned one by one with the GEDI footprint to establish corresponding point relationships, thus providing an accurate reference benchmark for subsequent error correction. This is the basic preparatory step for the entire method.

[0044] Step 2: Global Geometric Correction. Based on the UAV control points of the regular grid, the planar coordinates of the GEDI footprint are subjected to global geometric correction according to the track segment or time period. This includes overall translation, rotation, and scale adjustment to obtain the first-level corrected coordinates. This step can eliminate systematic translation, rotation, and scale deviations of the track segment / region and establish global geometric consistency.

[0045] This step relies on UAV control points arranged in a regular grid. Since the UAV control points are laid out according to a regular grid, covering the entire target area, they can effectively constrain large-scale translation, rotation, and scaling errors of the GEDI data.

[0046] In this step, the overall geometric correction uses an affine adjustment model, and its observation equation is:

[0047]

[0048] The solution is obtained using the weighted least squares method, and its matrix form is as follows:

[0049]

[0050] in, For the first One GEDI original footprint coordinate; Use the UAV true coordinates or the corresponding matching coordinates as the reference; ; Let be the affine transformation parameter vector to be estimated, where and They respectively represent two-dimensional affine transformations in and The correlation coefficients of direction, rotation, and scale; For the reason The design matrix is ​​composed of Design matrix transpose; It is a diagonal matrix composed of weights from step one; This is the affine transformation parameter vector estimated by the weighted least squares method.

[0051] Solving the affine transformation parameter vector using the weighted least squares method And based on the obtained affine transformation parameter vector Correct the GEDI footprint coordinates to obtain the corresponding first-level corrected coordinates. First-level corrected coordinates After overall affine correction, a large range of system offsets has been eliminated.

[0052] This step solves for the parameters of the affine adjustment model separately for each track segment or time slice. That is, the parameters of the affine adjustment model are established and solved independently for each track segment or time slice to avoid error propagation caused by cross-segment mixing. After solving, the adjustment residuals and coordinate covariance can be output as quality assurance (QA).

[0053] In step two, an affine adjustment model is established by using UAV control points laid out in a regular grid. The GEDI footprint is then translated, rotated, and scaled as a whole. This process effectively eliminates the systematic geometric and positioning offset problem of GEDI data over a large area, laying the framework for subsequent more refined local corrections.

[0054] Step 3: Local Distortion Correction. Based on the densely deployed UAV control points, local curvature and distortion corrections are performed on the GEDI footprint after overall geometric correction to obtain secondary corrected coordinates. This step can remove slowly varying nonlinear distortions and local curvatures within the track segment, ensuring local geometric accuracy.

[0055] This step relies on a densely deployed UAV control point network. Because densely deployed UAV control points in complex terrain areas (such as mountainous regions) can capture local nonlinear distortions and curvature characteristics, fine compensation can be performed on complex terrain areas after the overall framework is adjusted.

[0056] Furthermore, in step three, the local bending and distortion correction uses a quadratic surface residual model or a thin plate spline model to fit the residuals of the first-level correction coordinates in order to eliminate the gradual geometric distortion within the track segment.

[0057] The quadratic surface residual model is as follows:

[0058] Residuals after global geometric correction , The fit is as follows:

[0059]

[0060] The thin plate spline (TPS) model is as follows:

[0061]

[0062] in, ; This represents the planar residual field between the global geometrically corrected UAV reference and the UAV reference. Geographic coordinates representing any GEDI original footprint (with systematic offset and distortion); Indicates the coordinates of the UAV control points; for TPS coefficients in the direction (with smoothing regularization for robust solution). subscript " " indicates the first TPS coefficients for each UAV control point express The transpose of. Among them, and The coefficients to be estimated for the quadratic surface residual model are used to describe the GEDI footprint in... direction and The variation of planar residuals in the direction; This represents the Euclidean distance norm.

[0063] Optionally, the TPS model should be preferred in mountainous or complex areas; a quadratic surface residual model can be used in plains areas. The first-order corrected coordinates... Further correction to second-level corrected coordinates The second-level corrected coordinates further eliminate local gradual distortions, resulting in the final high-precision planar position result.

[0064] After completing the overall framework adjustment in step two, step three requires eliminating the residual nonlinear geometric distortions in complex terrain areas. Using a densely deployed network of UAV control points, the bending and stretching characteristics of GEDI data in localized areas can be fitted. This is equivalent to performing detailed correction on the locally "rubber-stretched" portions after overall geometric correction. This process can resolve the problem of gradually varying geometric distortions, making the geometric relationships in local areas more realistic and reliable.

[0065] Step 4: Physical Correction for Terrain Broadening Height. Based on the terrain slope and laser beam direction, a physical compensation model is established to compensate for the terrain broadening effect in the canopy height of the GEDI data, resulting in a corrected canopy height. The purpose of this step is to explicitly compensate for the system height deviation caused by waveform broadening due to terrain (or slope).

[0066] The physical compensation model established in this step is as follows:

[0067]

[0068] in, The corrected canopy height is the height after slope effect compensation, which eliminates the systematic error caused by terrain widening. The slope angle is obtained from the DEM. The angle between the laser beam direction and the maximum slope direction can be obtained by finding the angle between the track direction and the slope direction. If the laser beam direction is unknown, the worst-case scenario or the neighborhood average can be used to approximate it. The diameter of the GEDI footprint; This refers to the surface roughness index. , These are the model coefficients obtained through joint calibration using UAV ground truth data.

[0069] It can be determined by, for example, least squares or robust regression methods.

[0070] This step involves a physical correction for the slope effect. In mountainous areas and regions with steep slopes, the laser waveform of GEDI is often elongated due to the slope, causing systematic errors in tree height estimation. Therefore, by introducing the geometric relationships of slope angle, aspect, and footprint size, a physical compensation model can be established to model and compensate for this broadening effect, thereby effectively eliminating the height deviation caused by terrain. This step utilizes both regular grid UAV control points and densely distributed UAV control points. The regular grid UAV control points provide overall constraints for calculating the global trend of the slope-waveform broadening relationship; the densely distributed UAV control points provide higher resolution data locally in complex terrain areas, helping to accurately fit error compensation under steep slope terrain.

[0071] Step 5: Canopy and Signal Statistical Correction. Based on canopy structure parameters and the GEDI signal quality index, a statistical regression model is established to correct the canopy height residual, resulting in a secondary corrected canopy height. This step uses canopy structure and signal quality to explain the remaining bias.

[0072] After performing step four, the residual is obtained:

[0073]

[0074] in, To correct the canopy height, This is the true value for UAV height.

[0075] The expression for a statistical regression model (such as weighted regression or ridge regression) is as follows:

[0076]

[0077] The solution yields:

[0078]

[0079] The secondary corrected canopy height was obtained:

[0080]

[0081] in, These are canopy coverage, thickness, and volume, which are obtained from CHM or point cloud statistics.

[0082] GEDI signal quality indicators;

[0083] These are terrain slope and surface roughness, respectively.

[0084] The design moment is composed of the aforementioned canopy structure parameters, GEDI signal quality index, and terrain parameters; express transpose;

[0085] These are the estimated regression coefficients obtained through the regression model;

[0086] It is a diagonal matrix composed of weights from step one;

[0087] This is the random error term;

[0088] For constant terms;

[0089] These are the regression coefficients corresponding to the canopy structure parameters;

[0090] These are the regression coefficients corresponding to the GEDI signal quality index;

[0091] These are the regression coefficients corresponding to the terrain parameters.

[0092] For random noise;

[0093] This is the height residual between the canopy height and the true UAV value for a single correction.

[0094] Optionally, Huber / Tukey loss is used to iteratively update the weights to suppress outliers. To reduce the impact of outliers (outliers) on the regression model, a robust regression method is employed, which reduces the influence of outliers by iteratively adjusting the weights.

[0095] Even after the corrections in steps two through four, some small-scale, scattered residual errors will still exist. Step five involves establishing a statistical regression model by combining structural characteristics such as canopy thickness, coverage, and volume, as well as GEDI's own signal quality indicators, to correct the residuals and eliminate the final residuals.

[0096] Step Six: Spatial Residual Smoothing and Output. Spatial smoothing methods are used to eliminate localized, scattered residuals in the secondary correction of canopy height, resulting in a converged canopy height result. The final output is the corrected GEDI forest structure data, including the secondary correction coordinates and the converged canopy height result. This step smooths the small-scale spatial residual field and outputs the final product, the corrected GEDI forest structure data, addressing the influence of canopy structure and eliminating the final residuals.

[0097] Alternatively, spatial smoothing methods can employ thin plate spline (TPS) or Kriging interpolation to achieve spatial residual smoothing and eliminate fragmented errors.

[0098] right The final canopy height is obtained by performing TPS or Kriging:

[0099]

[0100] in, For position The height residual at the location; In order to be in Based on this, the secondary corrected canopy height was obtained by statistical regression correction using canopy characteristic parameters and GEDI signal quality index; The reference canopy height is the true value obtained from UAV lidar data; For the final corrected canopy height result, in Based on this, a spatial smoothing term is added to eliminate local scattered residuals, representing the high-precision GEDI canopy height output by this invention.

[0101] This step can also output uncertainty and quality control results, including:

[0102] Output adjustment covariance Regression RMSE, pixel-level 68% / 95% confidence interval;

[0103] in accordance with A QA mask is generated using the residual threshold to label low-confidence regions.

[0104] In step six, local anomalies are eliminated using spatial smoothing methods to ensure that the final output is both accurate and continuous. This step primarily addresses the influence of canopy structure and ultimately converges all remaining errors.

[0105] This step primarily relies on a densely deployed UAV control point network, as areas with complex canopy structures often occur in mountainous regions, and a densely deployed network of UAV control points provides reliable samples for statistical modeling. Simultaneously, regular grid-based UAV control points also participate to ensure statistical robustness in plains areas and large-scale regions.

[0106] After the above six steps, the final output of this invention is fully corrected GEDI data, with its planar coordinates and height information systematically improved. The entire evolution process of the result of this method is as follows:

[0107] In terms of planar coordinates, the footprint coordinates of the original GEDI data are as follows: After the overall geometric correction in step two, the first-level corrected coordinates are obtained. This coordinate system has eliminated large-scale systematic translations, rotations, and scale shifts. Further, in step three, through local distortion correction, the first-level corrected coordinates are corrected for nonlinear bending and gradual distortion to obtain the final second-level corrected coordinates. .therefore, The transformation process completely corresponds to the gradual elimination of planar positioning errors, in which... This indicates a high-precision planar position that has been corrected using a regular grid and a densely deployed UAV control point network.

[0108] In terms of height, the original canopy height or related indicator of GEDI is as follows: After physical correction for the slope effect in step four, the waveform broadening error caused by the terrain was eliminated, resulting in a corrected canopy height. In step five, a statistical regression model is constructed using canopy features and signal quality parameters to compensate for the remaining system bias, resulting in a secondary corrected canopy height. Subsequently, in step six, local stray residuals are eliminated using a spatial smoothing method to obtain the final convergence result. .therefore, The processes correspond to terrain broadening error elimination, canopy structure and signal influence correction, and final spatial consistency optimization, respectively. This is the final altitude result output by the present invention, which includes both large-scale consistency and high precision in local mountainous areas.

[0109] Through steps one through six above, the present invention can significantly improve the planar positioning accuracy and canopy height accuracy of GEDI data in a province or even a larger area under the condition of limited UAV cost, and achieve the unity of large-area coverage and local high accuracy.

[0110] The method of this invention utilizes a hierarchical control point deployment strategy of "regular grid + mountainous densification," combined with affine adjustment, surface or thin-plate spline fitting, slope geometric compensation, and canopy and signal statistical correction, to progressively eliminate four main types of errors in GEDI data (geometric and positioning system offset, gradually varying geometric distortion, waveform broadening caused by terrain, and the influence of canopy structure). This method, while ensuring limited UAV cost investment, can achieve overall accuracy improvement of GEDI data over a large area and guarantee high accuracy in locally complex regions, thereby enabling high-quality application of canopy height and topographic elements across provinces and even larger regions.

[0111] (1) To address the offset of the geometry and positioning system, UAV grid control points are set up over a large area, and affine or quadratic polynomial adjustment methods are used to perform overall geometric correction on the GEDI footprint, thereby eliminating uniform translation, rotation and scale deviations.

[0112] (2) To address the gradually changing geometric distortion, denser UAV control points are set up in mountainous and complex areas, and surface fitting or thin plate spline methods are used to correct the local residual field in the track section to eliminate the gradually accumulating drift and bending.

[0113] (3) In response to the waveform broadening caused by the terrain, a physical compensation model is established by combining the geometric relationship between slope, aspect and footprint diameter to correct the systematic height deviation caused by the slope effect;

[0114] (4) To address the impact of canopy structure, structural parameters such as canopy thickness, coverage, and volume, as well as signal quality indicators, are introduced. Statistical regression methods are used to model the residuals to compensate for height estimation errors caused by differences in forest type and signal conditions.

[0115] The following example, using Jilin Province as the target region, illustrates the correction method of the present invention.

[0116] Jilin Province, located in Northeast China, has a complex topography, including the Changbai Mountains in the east and the Songliao Plain in the west. When GEDI data is applied in this region, it is often constrained by problems such as large-scale geometric offsets, local distortions in mountainous areas, slope widening, and the influence of canopy structure, leading to systematic errors in forest height and topographic information. To solve this problem, this invention specifically adopts the following solution:

[0117] Step 1: Match the UAV control points with the corresponding GEDI-UAV control points.

[0118] Within Jilin Province, approximately 900 UAV control points were initially established using a regular grid with 10km intervals. These control points covered the entire study area, including the Songliao Plain and some gentle slope areas, and were primarily used for geometric correction of the overall framework.

[0119] Subsequently, an area with terrain undulations exceeding 300 meters was delineated in the Changbai Mountains. Within this area, approximately 400 UAV control points were densely deployed at 5km intervals. These points were primarily used for mountain distortion correction, slope widening compensation, and canopy interference elimination. High-resolution point cloud data was collected from all UAV control points, and after processing, digital elevation models, canopy height models, and corresponding terrain and vegetation parameters were obtained.

[0120] Figure 2 The image shows a schematic diagram of the UAV control point layout in Jilin Province. Blue dots represent regular grid UAV control points spaced 10km apart, primarily used to eliminate systematic geometric and positioning offsets in GEDI data over large areas, ensuring the uniformity of the overall geometric framework. Red dots represent denser UAV control points spaced 5km apart in mountainous areas, used for fine-tuning against gradual distortions and slope effects in complex terrain. The brown contour lines in the background reflect the terrain undulations, highlighting the necessity of denser control point deployment in mountainous areas.

[0121] Using the GEDI footprint diameter (approximately 25m) as a window, corresponding analog waveforms are extracted from UAV data and correlated with GEDI waveforms to obtain the corresponding point relationships. Weights are calculated based on the correlation coefficient and GEDI signal parameters (sensitivity, received energy, echo sampling number) to provide a basis for subsequent weighted adjustment.

[0122] Step 2: Overall geometric correction.

[0123] Across the province, an affine adjustment model was established using UAV control points with a regular grid to comprehensively adjust the planar coordinates of the GEDI footprint, eliminating translation, rotation, and scale errors. After correction, the positioning offset of the GEDI footprint in the plain area decreased from the original 30–50 meters to 8–12 meters.

[0124] Step 3: Local distortion correction.

[0125] In the Changbai Mountains, a thin-plate spline model was established based on the coordinate residuals of densely deployed UAV control points to correct the local curvature and distortion of the GEDI footprint, resulting in a second-order corrected coordinate system. This step further reduced the planar error of the GEDI footprint in the mountainous area from 15–20 meters to 5–8 meters.

[0126] Step 4: Physical correction of terrain widening height (slope effect compensation).

[0127] Slope and aspect parameters were extracted using a digital elevation model (DEM), and combined with the GEDI footprint size, a slope-widening relationship model, i.e., a physical compensation model, was established to compensate for the GEDI height based on terrain. After compensation, the estimation error of the canopy height in mountainous areas was reduced by approximately 20%.

[0128] Step 5: Canopy and signal statistical correction.

[0129] Based on the canopy coverage, thickness, volume, and other characteristics provided by the UAV control points, as well as the GEDI signal sensitivity, received energy, and echo sampling number, a statistical regression model is constructed to eliminate the systematic bias of canopy interference on GEDI height and obtain a secondary corrected canopy height.

[0130] Step 6: Spatial residual smoothing and output.

[0131] A spatial smoothing method is used to eliminate local scattered residuals in the secondary correction of canopy height, resulting in GEDI height data with spatial continuity and high consistency (i.e., canopy height convergence results). Finally, GEDI height data and secondary correction coordinates are output.

[0132] After processing using the method of this invention, the positioning and elevation accuracy of GEDI data in Jilin Province were significantly improved. The experimental results are shown in Table 1.

[0133] Table 1 Comparison of GEDI data correction effects

[0134]

[0135] Another embodiment of the present invention provides a spatial lidar forest structure data correction system based on UAV lidar data. This system is used to implement the correction method described in the above embodiments. The system mainly includes a processor and a memory, wherein the memory stores a computer program. When the processor executes the computer program, the system is configured to perform the various steps of the aforementioned correction method.

[0136] Specifically, the processor includes a data acquisition and registration module, a first-level geometric correction module, a second-level local distortion correction module, a terrain widening compensation module, a statistical regression correction module, and a spatial smoothing output module.

[0137] The data acquisition and registration module is used to implement the function of step one, that is, according to the hierarchical deployment strategy, UAV control points are deployed in a regular grid within the target area, and UAV control points are deployed more densely in complex terrain areas. After acquiring the UAV data of all UAV control points, the UAV data is waveform registered with the GEDI footprint to establish the relationship between corresponding points and calculate the corresponding weights.

[0138] The first-level geometric correction module is used to implement the function of step two, that is, to perform overall geometric correction of the plane coordinates of the GEDI footprint based on the regular grid UAV control points according to the track segment or time period, including overall translation, rotation and scale adjustment, to obtain the first-level corrected coordinates.

[0139] The secondary local distortion correction module is used to implement the function of step three, that is, based on the densely deployed UAV control points, to perform local bending and distortion correction on the GEDI footprint after overall geometric correction, and obtain the secondary corrected coordinates.

[0140] The terrain widening compensation module is used to achieve the function of step four, that is, based on the terrain slope and the direction of the laser beam, a physical compensation model is established to compensate for the terrain widening effect of the canopy height in the GEDI data, so as to obtain a corrected canopy height.

[0141] The statistical regression correction module is used to implement the function of step five, which is to establish a statistical regression model based on the canopy structure parameters and GEDI signal quality index, correct the canopy height residual, and obtain the secondary corrected canopy height.

[0142] The spatial smoothing output module is used to implement the function of step six, that is, to eliminate the local scattered residuals of the secondary corrected canopy height through the spatial smoothing method, obtain the canopy height convergence result, and finally output the corrected GEDI forest structure data, including the secondary corrected coordinates and the canopy height convergence result.

[0143] The memory also stores preset parameters such as empirical indices, weight cutoff thresholds, calibration coefficients, and regression model coefficients for use by various modules. This system can run efficiently on computers or embedded devices, achieving automated and high-precision correction of GEDI data from geometric to elevation perspectives. Using this system to process GEDI data achieves the same technical effects as the method embodiments, namely, significantly improving the planar positioning accuracy and canopy height accuracy of GEDI data, providing reliable data support for applications such as forest resource monitoring and carbon storage assessment.

[0144] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0145] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A method for correcting forest structure data using space lidar, characterized in that, Includes the following steps: Step 1: According to the hierarchical deployment strategy, deploy UAV control points in a regular grid within the target area, and densify the deployment of UAV control points in complex terrain areas. After acquiring the UAV data of all the UAV control points, perform waveform registration between the UAV data and the GEDI footprint to establish the relationship between corresponding points and calculate the corresponding weights. Step 2: Based on the UAV control points of the regular grid, perform overall geometric correction on the planar coordinates of the GEDI footprint according to the track segment or time period, including overall translation, rotation and scale adjustment, to obtain the first-level corrected coordinates; Step 3: Based on the densely deployed UAV control points, perform local bending and distortion correction on the GEDI footprint after overall geometric correction to obtain the secondary corrected coordinates; Step 4: Based on the terrain slope and laser beam direction, establish a physical compensation model to compensate for the terrain widening effect in the canopy height of the GEDI data, and obtain the first corrected canopy height; Step 5: Based on the canopy structure parameters and GEDI signal quality index, establish a statistical regression model to correct the canopy height residual and obtain the secondary corrected canopy height; Step 6: Eliminate the local scattered residuals of the secondary corrected canopy height using a spatial smoothing method to obtain the canopy height convergence result. Finally, output the corrected GEDI forest structure data, including the secondary corrected coordinates and the canopy height convergence result.

2. The method for correcting forest structure data using space lidar according to claim 1, characterized in that, The waveform registration process described in step one includes: Using the GEDI footprint diameter as a window, extract the corresponding analog waveform from the UAV data; Within the preset micro-shift search window, by maximizing the normalized cross-correlation, the translation amount that makes the simulated waveform most similar to the GEDI waveform is determined as the corresponding point matching result, and the maximum normalized cross-correlation number is recorded.

3. The spatial lidar forest structure data correction method according to claim 2, characterized in that, The formula for calculating the weights mentioned in step one is as follows: in, For the first The weights corresponding to each GEDI footprint; They represent the first Sensitivity, received energy, and number of samples for each GEDI footprint; All are experience indices; For the first The maximum normalized cross-correlation number corresponding to each GEDI footprint; This is a truncation function. These are the lower and upper thresholds for weight truncation, respectively.

4. The method for correcting forest structure data using space lidar according to claim 1, characterized in that, The overall geometric correction described in step two uses an affine adjustment model, and its observation equation is: in, The original GEDI footprint coordinates, Use the UAV true coordinates or the corresponding matching coordinates as the reference; Solving the affine transformation parameter vector using the weighted least squares method And based on the obtained affine transformation parameter vector Correct the GEDI footprint coordinates to obtain the corresponding first-level corrected coordinates. .

5. The method for correcting forest structure data using space lidar according to claim 1, characterized in that, In step three, the local bending and distortion correction is performed by fitting the residuals of the first-level correction coordinates using a quadratic surface residual model or a thin plate spline model.

6. The method for correcting forest structure data using space lidar according to claim 1, characterized in that, The physical compensation model is as follows: in, This is a correction of the canopy height. The diameter of the GEDI footprint; This refers to the surface roughness index. The angle between the laser beam direction and the maximum slope direction; The slope angle; These are the coefficients jointly calibrated with the UAV true value.

7. The method for correcting forest structure data using space lidar according to claim 1, characterized in that, The expression for the statistical regression model is: in, This is a correction of the height residual between the canopy height and the true UAV value; These are canopy coverage, thickness, and volume, respectively. For constant terms; These are the coefficients of the canopy structure parameters; These are the signal quality parameter coefficients; These are terrain parameter coefficients; This is a random noise term.

8. The method for correcting forest structure data using space lidar according to claim 1, characterized in that, The spatial smoothing method employs either the thin-plate spline method or the Kriging interpolation method.

9. The method for correcting forest structure data using space lidar according to claim 1, characterized in that, In the hierarchical deployment strategy, the spacing between UAV control points in the regular grid is 10 to 50 km, and the spacing between UAV control points in the densely deployed grid is 5 to 10 km. The complex terrain area is a mountainous area with terrain undulations exceeding 300 meters.

10. A space lidar forest structure data correction system, comprising a processor and a memory, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 9.

Images