A direct snow depth inversion method based on a spaceborne photon counting lidar

Abstract

The application discloses a direct snow depth inversion method based on a spaceborne photon counting laser radar, and comprises the following steps: extracting a target snow accumulation area, eliminating a thick cloud layer area and non-target ground point cloud, aligning a surface, and performing noise correction; performing deconvolution to obtain a corrected snow accumulation attenuation backscattering profile; setting an initial snow absorption coefficient, inverting a snow backscattering path length distribution without absorption, and calculating a first moment and a second moment of the distribution; based on the first moment and the second moment, simultaneously inverting snow depth, snow albedo and diffuse scattering coefficient; combining the snow albedo and the diffuse scattering coefficient to calculate an updated snow absorption coefficient; if the updated snow absorption coefficient does not satisfy a set convergence threshold, taking the updated snow absorption coefficient as a new initial snow absorption coefficient and repeating iteration; and if the convergence threshold is satisfied, outputting final snow depth. According to the application, snow depth inversion along a track can be directly realized without the aid of any other data.

Description

Technical Field

[0001] This invention belongs to the field of spaceborne lidar remote sensing technology, and in particular relates to a direct snow depth inversion method based on spaceborne photon counting lidar. Background Technology

[0002] Snow depth is a core indicator for quantitatively characterizing snow reserves and their spatiotemporal distribution, and its importance is particularly prominent in polar research. The low thermal conductivity of snow on the sea ice surface significantly alters sea ice thermodynamic processes, and snow depth is the largest source of uncertainty in sea ice inversion. Errors in snow depth measurement can lead to deviations in sea ice thickness inversion, making high-precision global snow depth detection an urgent problem to be solved in the field of cryosphere science.

[0003] While traditional in-situ snow depth measurement is highly accurate, it suffers from drawbacks such as low efficiency, sparse coverage, and poor representativeness, making it unable to meet the needs of large-scale, multi-temporal dynamic monitoring.

[0004] Satellite remote sensing provides an effective approach for global snow depth detection. For example, Chinese patent document CN102708277A discloses a snow depth inversion design method based on ant colony algorithm. First, simulated data on the relationship between radiation brightness temperature at different frequencies and snow depth are obtained through a microwave radiation transmission model of snow. Then, the simulated radiation brightness temperature data and snow depth data are discretized and converted into the path space of ants in the ant colony algorithm. Each discrete interval is taken as a possible path point, and an initial value of pheromone concentration is set at each possible path point. The path chosen by each ant represents an inversion rule, and the validity of the rule is calculated as the basis for updating the pheromone concentration. Feedback is provided to the path selection of the next ant colony. After several iterations, the final inversion rule of snow depth is obtained.

[0005] However, existing methods all have obvious limitations: microwave radiometers are only applicable to multi-year ice regions, with a resolution of only 12.5-100km; synthetic aperture radar and passive optical satellites rely on auxiliary data such as the type and reflectivity of the underlying surface; radar altimeters are limited by spatiotemporal matching errors, and the snow depth detection method of altimeters is mainly achieved by subtracting the Ku-band elevation that can penetrate the snow surface from the elevation detected by visible light laser or Ka-band microwave, which cannot penetrate the snow surface, from the elevation detected by Ku-band microwave, which can penetrate the snow surface to reach the ice surface. However, the detection accuracy is limited by spatiotemporal matching, and Ku-band microwave cannot completely penetrate the snow body to reach the underlying surface.

[0006] Therefore, global snow depth data is currently scarce. Spaceborne photon-counting lidar has direct elevation detection capabilities, enabling rapid and large-area acquisition of snow depth information. However, current snow depth retrieval methods based on spaceborne photon-counting lidar still face significant technical bottlenecks, hindering their large-scale application in global snow cover monitoring.

[0007] The current mainstream inversion method calculates snow depth by the difference between the snow surface elevation during the snow cover period and the underlying surface reference elevation during the snowless period. This method is highly dependent on historical snowless period reference data and performs poorly in seasonal snow cover areas and polar and high-altitude regions with year-round snow cover. Summary of the Invention

[0008] This invention provides a direct snow depth inversion method based on a spaceborne photon counting lidar, which can directly realize snow depth inversion along the orbit without the need for any other data.

[0009] A direct snow depth inversion method based on a spaceborne photon counting lidar includes the following steps: (1) The photon point cloud obtained by the spaceborne photon counting lidar is processed, the photon point cloud of the target snow area is selected, the surface is aligned and noise correction is performed to obtain the denoised snow attenuation backscattering profile. ; (2) Backscatter profile of snow accumulation attenuation Perform deconvolution to obtain the corrected snow attenuation backscattering profile. ; (3) Set the initial snow absorption coefficient and scatter the corrected snow attenuation profile. Invert the path length distribution of non-absorbing snow backscattering and calculate the first and second moments of this distribution; (4) Based on the first and second moments of the path length distribution, snow depth, snow albedo and diffuse scattering coefficient are simultaneously inverted; (5) Introduce an empirical physical model of snow optical properties, and calculate the updated snow absorption coefficient by combining the snow albedo and diffuse scattering coefficient obtained by inversion; (6) If the updated snow absorption coefficient does not meet the set convergence threshold, the updated snow absorption coefficient is used as the new initial snow absorption coefficient, and steps (3) to (5) are repeated; if it meets the threshold, the final snow depth is output.

[0010] Further, in step (1), the photon point cloud of the target snow-covered area is obtained by filtering, specifically as follows: Based on the initial screening of latitude and longitude, the photon point cloud of the target snow-covered area is obtained. The point cloud is divided into continuous independent segments at fixed intervals along the direction of the satellite orbit. The point cloud segments are statistically analyzed, and a photon count threshold is set according to the albedo of the snow. Point cloud segments with counts below the threshold are deleted, thereby removing areas obscured by clouds and fog and non-target surface areas.

[0011] Further, in step (1), the surfaces are aligned and noise correction is performed, specifically as follows: Photon point clouds of the target snow-covered area were obtained by filtering, and histograms were generated by statistically analyzing the photon height distribution at set vertical intervals. Gaussian functions were used to fit the histograms, and the average value of the Gaussian parameters for the photon distribution was calculated. and standard deviation ; average As the initial snow surface elevation for each segment, and to ensure consistent alignment of the initial snow surface elevations for all segments; (The last part, "elevation greater than...", appears to be a fragment and requires further context for accurate translation.) The photons were determined to be atmospheric scattering and solar background noise. The noise photons were averaged at the aforementioned vertical intervals to obtain the average noise. The statistical elevation was less than the average value. All effective photons are accumulated at equal vertical intervals to generate the original backscatter profile; the original backscatter profile is then subtracted from the corresponding noise baseline at each height level to obtain the denoised snow attenuation backscatter profile. .

[0012] Furthermore, in step (2), the specific process of deconvolution is as follows: Snow attenuation backscatter profile Represented as: ;in, Indicates the vertical depth relative to the snow surface; This is the actual lidar signal profile, i.e., the corrected snow attenuation backscattering profile; The system's impulse response curve; The Lucy-Richardson iterative deconvolution algorithm is used to attenuate the snow backscatter profile generated from the original photon point cloud of the target region. Correction is performed to obtain the corrected snow attenuation backscattering profile. .

[0013] Furthermore, in step (3), the backscattering profile from the corrected snow attenuation is... The path length distribution of backscattered snow without absorption was retrieved as follows: ; in, This represents the distribution of backscattering path lengths in non-absorbing snow. The initial snow absorption coefficient, , where is the backscattering path length of the photon within the snow.

[0014] Further, in step (3), the first and second moments of the distribution are calculated, specifically as follows: First moment Represented as: ; Second moment Represented as: .

[0015] Furthermore, in step (4), the specific formulas for retrieving snow depth, snow albedo, and diffuse scattering coefficient are as follows: ; ; ; in, This indicates the snow depth obtained through inversion. This represents the snow albedo obtained through inversion. This represents the diffuse scattering coefficient obtained from the inversion.

[0016] Furthermore, the specific process of step (5) is as follows: Based on the set initial snow absorption coefficient and the snow albedo obtained by inversion Calculate the equivalent radius of snow particles ; Based on the set initial snow absorption coefficient and the calculated equivalent radius of snow particles Derivation of the diffuse decay coefficient ; The updated snow absorption coefficient is derived based on the optical properties of snow. .

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention constructs a direct snow depth inversion method based on photon-counting lidar signals using multiple scattering theory. It does not use any auxiliary data and is not affected by orbital deviation or missing auxiliary data.

[0018] 2. The snow depth inverted in this invention uses the backscatter profile of a spaceborne photon counting lidar, which has extremely high along-track resolution and its accuracy is not affected by mixed pixels or ice and snow roughness. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of a direct snow depth inversion method based on a spaceborne photon counting lidar according to an embodiment of the present invention.

[0021] Figure 2 This is a comparison of snow depth in the Arctic region and IceBridge in an embodiment of the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] It should be noted that, unless otherwise specified, the features in the following embodiments and implementation methods can be combined with each other.

[0024] In this embodiment of the invention, the trajectory of the ICEBridge airborne snow radar on April 19, 2019, was selected as verification data. The inversion data used by the spaceborne photon counting lidar is the trajectory of ICESat-2 passing through the area at 13:10 on April 19, 2019.

[0025] like Figure 1 As shown, a direct snow depth inversion method based on a spaceborne photon counting lidar includes the following steps: S1. Extract the target snow area, remove areas with thick cloud layers and non-target surface point clouds, align the surface, and perform noise correction.

[0026] The study area was defined as having latitude and longitude of 80°N-90°N and 150°W-100°W. Based on this latitude and longitude, a preliminary selection of photon point clouds for the target area was obtained. These point clouds were then divided into continuous, independent segments at fixed intervals of 7 km along the satellite orbit. Statistical analysis was performed on these point cloud segments. Since snow cover has the highest albedo among all surface types, a photon count threshold of 30,000 per segment was set based on albedo. Segments with a count below this threshold were deleted, thus removing areas obscured by clouds and fog, as well as non-target surface areas.

[0027] For the photon point cloud of the selected target region, a histogram of photon height distribution was generated by statistically analyzing photons at vertical intervals of 0.15m. A Gaussian function was then used to fit the histogram, and the average value of the Gaussian parameters for the photon distribution was calculated. and standard deviation .Will This serves as the initial snow surface elevation for each segment, and the initial snow surfaces of all segments are aligned consistently. Elevations greater than [a certain value] are used. The photons were determined to be atmospheric scattering and solar background noise. The noise photons were averaged at the aforementioned vertical intervals to obtain the average noise. The statistical elevation was less than... All effective photons are accumulated at equal vertical intervals to generate the original backscatter profile; the original profile is then subtracted from the corresponding noise baseline at each height level to obtain the denoised snow attenuation backscatter profile. .

[0028] S2. The influence of the back pulse is eliminated by deconvolution to obtain the corrected attenuated backscatter signal profile.

[0029] In reality, due to laser pulse shape distortion, receiver optical component response delay, and interference from afterpulse effects, It is the convolution of the actual snow attenuation backscattered signal and the system impulse response, which can be expressed as: .

[0030] Where z represents the vertical depth relative to the snow surface. For the true lidar signal profile, The system impulse response curve is shown. The Lucy-Richardson iterative deconvolution algorithm is used to correct the snow attenuation backscatter profile generated from the original photon point cloud of the target region, resulting in the corrected snow attenuation backscatter profile. .

[0031] S3. Set the initial snow absorption coefficient, invert the snow backscattering path length distribution without absorption from the corrected attenuated backscattering signal profile, and calculate the first and second moments of the distribution.

[0032] Assuming initial snow absorption coefficient Then the backscattering path distribution of non-absorbing snow can be expressed as: ,in Let be the path length of the photon backscattering within the snow. Then, the first moment of the snow-scattered photon path distribution... It can be represented as: ; The second moment can be expressed as: .

[0033] S4. Based on the first and second moments of the path length distribution, the initial snow depth, snow albedo, and diffuse scattering coefficient are simultaneously inverted.

[0034] Snow depth derived from photon path distribution It can be represented as: ; The diffuse attenuation coefficient can be expressed as: ; Albedo can be expressed as: ; S5. Introduce an empirical physical model of snow optical properties, and calculate the updated snow absorption coefficient by combining the snow albedo and diffuse scattering coefficient obtained by inversion.

[0035] The equivalent radius of a snow particle, calculated from the preset absorption coefficient and albedo, can be expressed as: ; Correspondingly, the diffuse attenuation coefficient derived from the preset absorption coefficient and particle size can be expressed as: ; The snow absorption coefficient, derived from the optical properties of snow, can be expressed as: ; S6. Set the absorption coefficient convergence threshold. If it is not met, use the updated absorption coefficient as the new initial value and repeat steps S3 to S5. If it is met, output the final snow depth.

[0036] Set a relative error convergence threshold for the snow absorption coefficient. Compare the absorption coefficient calculated in step S5 with the initial value of this iteration. If the absolute value of the difference is greater than the threshold, use the value after this iteration as the new initial absorption coefficient. Repeat steps S3 to S5 for inversion of the non-absorption path length distribution, inversion of initial snow parameters, and updating of absorption coefficient. At the same time, set the maximum number of iterations and the maximum value of the albedo boundary to be less than 1. If the number of iterations reaches the upper limit or the albedo exceeds the boundary value and the convergence condition is not met, the output of this result is invalid. If the convergence condition is met, output the final snow depth and related optical characteristic parameters obtained in this iteration.

[0037] To verify the effectiveness of this invention, the snow depth retrieved by the ICESat-2 photon counting lidar based on this invention was compared with the snow depth detected by the ICEBridge (ice bridge) at the same location. The results are as follows: Figure 2 As shown, the snow depth inverted based on this invention has a high degree of consistency with the airborne snow depth.

[0038] The embodiments described above provide a detailed explanation of the technical solutions and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, additions, and equivalent substitutions made within the scope of the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A direct snow depth inversion method based on a spaceborne photon counting lidar, characterized in that, Includes the following steps: (1) The photon point cloud obtained by the spaceborne photon counting lidar is processed, the photon point cloud of the target snow area is selected, the surface is aligned and noise correction is performed to obtain the denoised snow attenuation backscattering profile. ; (2) Backscatter profile of snow accumulation attenuation Perform deconvolution to obtain the corrected snow attenuation backscattering profile. ; (3) Set the initial snow absorption coefficient and scatter the corrected snow attenuation profile. Invert the path length distribution of non-absorbing snow backscattering and calculate the first and second moments of this distribution; (4) Based on the first and second moments of the path length distribution, snow depth, snow albedo and diffuse scattering coefficient are simultaneously inverted; (5) Introduce an empirical physical model of snow optical properties, and calculate the updated snow absorption coefficient by combining the snow albedo and diffuse scattering coefficient obtained by inversion; (6) If the updated snow absorption coefficient does not meet the set convergence threshold, the updated snow absorption coefficient is used as the new initial snow absorption coefficient, and steps (3) to (5) are repeated; if it meets the threshold, the final snow depth is output.

2. The direct snow depth inversion method based on spaceborne photon counting lidar according to claim 1, characterized in that, In step (1), the photon point cloud of the target snow-covered area is obtained by filtering, specifically as follows: Based on the initial screening of latitude and longitude, the photon point cloud of the target snow-covered area is obtained. The point cloud is divided into continuous independent segments at fixed intervals along the direction of the satellite orbit. The point cloud segments are statistically analyzed, and a photon count threshold is set according to the albedo of the snow. Point cloud segments with counts below the threshold are deleted, thereby removing areas obscured by clouds and fog and non-target surface areas.

3. The direct snow depth inversion method based on spaceborne photon counting lidar according to claim 2, characterized in that, In step (1), the surfaces are aligned and noise correction is performed, specifically as follows: Photon point clouds of the target snow-covered area were obtained by filtering, and histograms were generated by statistically analyzing the photon height distribution at set vertical intervals. Gaussian functions were then used to fit the histograms, and the average value of the Gaussian parameters for the photon distribution was calculated. and standard deviation ; average As the initial snow surface elevation for each segment, and to ensure consistent alignment of the initial snow surface elevations for all segments; (The last part, "elevation greater than...", appears to be a fragment and requires further context for accurate translation.) The photons were determined to be atmospheric scattering and solar background noise. The noise photons were averaged at the aforementioned vertical intervals to obtain the average noise. The statistical elevation was less than the average value. All effective photons are accumulated at equal vertical intervals to generate the original backscatter profile; the original backscatter profile is then subtracted from the corresponding noise baseline at each height level to obtain the denoised snow attenuation backscatter profile. .

4. The direct snow depth inversion method based on spaceborne photon counting lidar according to claim 1, characterized in that, In step (2), the specific process of deconvolution is as follows: Snow attenuation backscatter profile Represented as: ;in, Indicates the vertical depth relative to the snow surface; This is the actual lidar signal profile, i.e., the corrected snow attenuation backscattering profile; The system's impulse response curve; The Lucy-Richardson iterative deconvolution algorithm is used to attenuate the snow backscatter profile generated from the original photon point cloud of the target region. Correction is performed to obtain the corrected snow attenuation backscattering profile. .

5. The direct snow depth inversion method based on spaceborne photon counting lidar according to claim 4, characterized in that, In step (3), the backscattering profile from the corrected snow attenuation is... The path length distribution of backscattered snow without absorption was retrieved as follows: ； in, This represents the distribution of backscattering path lengths in non-absorbing snow. The initial snow absorption coefficient, , where is the backscattering path length of the photon within the snow.

6. The direct snow depth inversion method based on spaceborne photon counting lidar according to claim 5, characterized in that, In step (3), the first and second moments of the distribution are calculated, specifically as follows: First moment Represented as: ； Second moment Represented as: 。 7. The direct snow depth inversion method based on spaceborne photon counting lidar according to claim 6, characterized in that, In step (4), the specific formulas for retrieving snow depth, snow albedo, and diffuse scattering coefficient are as follows: ； ； ； in, This indicates the snow depth obtained through inversion. This represents the snow albedo obtained through inversion. This represents the diffuse scattering coefficient obtained from the inversion.

8. The direct snow depth inversion method based on spaceborne photon counting lidar according to claim 1, characterized in that, The specific process of step (5) is as follows: Based on the set initial snow absorption coefficient and the snow albedo obtained by inversion Calculate the equivalent radius of snow particles ; Based on the set initial snow absorption coefficient and the calculated equivalent radius of snow particles Derivation of the diffuse decay coefficient ; The updated snow absorption coefficient is derived based on the optical properties of snow. .

Images