An along-track sea ice concentration estimation method and system based on space-borne atmospheric detection lidar ACDL

Abstract

The application discloses a method and system for estimating the along-track sea ice concentration based on an atmospheric detection laser radar ACDL on a satellite. In a two-dimensional feature space and a linear abundance solving framework, according to the fact that the laser footprint mainly shows three stable response directions in the two-dimensional feature response space, i.e. low spread high response, low spread low response and high spread response, three types of endmember response modes are constructed, the number of effective endmembers is maintained, the stability of the abundance solving is ensured, the balance between the 532 nm endmember attribute constraints is adopted, the proportion of the footprint scale continuous ice phase response is obtained, and the method can effectively represent the along-track sea ice concentration changes in the marginal ice zone, the broken ice zone and the ice-water transition zone. The complementary relationship between the 1572 nm echo form and the 532 nm optical parameters is utilized to jointly analyze the ice-water mixing state of the ocean underlying surface, and the along-track footprint scale continuous sea ice concentration estimation result is obtained.

Description

Technical Field

[0001] This invention relates to the field of lidar detection technology, specifically to a method and system for estimating sea ice concentration along orbit based on a spaceborne atmospheric sounding lidar (ACDL). This invention is applicable to scenarios where sea ice and open water are mixed, such as polar edge ice zones and ice-water transition zones. It can be used to extract the ice phase response ratio at the laser footprint scale from spaceborne lidar along-orbit observation data and generate along-orbit sea ice concentration estimation results. Background Technology

[0002] Polar sea ice extent, sea ice concentration, and the distribution of ice-water transition zones are crucial aspects of polar marine environmental monitoring. Sea ice concentration is typically used to characterize the proportion of sea ice cover to ocean area within a given region, and it plays a significant role in climate change research, air-sea interaction analysis, and polar shipping route environmental assessment. Existing sea ice concentration products are mostly obtained through passive microwave remote sensing inversion. Passive microwave remote sensing offers advantages such as large coverage area, long time series, and minimal impact from solar illumination conditions, making it suitable for large-scale sea ice change monitoring. For example, the NOAA / NSIDC passive microwave sea ice concentration product provides daily and monthly sea ice concentration data with a grid spatial resolution of 25 km. While this type of product can reflect regional-scale sea ice cover, in marginal ice zones, flake ice zones, and ice-water transition zones, a single grid may simultaneously contain multiple underlying surface states, including open water, flake ice, thin ice, and dense sea ice, making it difficult to directly characterize small-scale ice-water mixing structures.

[0003] Remote sensing techniques such as visible light, infrared, and synthetic aperture radar (SAR) can also be used to identify sea ice boundaries, sea ice types, or ice surface conditions. Visible light and infrared remote sensing provide high spatial resolution imagery, but their ability to acquire stable and clear sea ice surface information is limited under observation conditions such as cloud cover, polar night, or low solar altitude angles. SAR offers all-weather, all-day observation capabilities, but its ice-water identification results are typically affected by factors such as sea surface wind field, ice surface roughness, incident angle, and regional empirical thresholds. While these methods play a crucial role in large-scale sea ice monitoring, estimating the ice-water mixing ratio at the scale of the laser footprint along the orbit still requires processing in conjunction with the high sampling density observation characteristics of lidar itself.

[0004] Spaceborne lidar is an active remote sensing method that can acquire high spatial sampling density echo information of the underlying surface along satellite orbits. Taking ICESat-2 as an example, its ATLAS payload uses a 532 nm laser and photon counting detection method. The relevant sea ice products mainly provide information such as sea ice height, sea surface lead line, and sea ice free plate, which is suitable for sea ice altimetry and free plate estimation. This type of method can obtain high-precision elevation information along the orbit, but its sea ice products mainly focus on elevation, lead line, and free plate parameters. The expression of the ice-water mixing ratio within the laser spot usually still needs to be further inferred from the height, lead line identification, or external data. It is difficult to directly and accurately describe the ice-water mixing state of the ocean underlying surface using multi-band echo differences.

[0005] Passive microwave sea ice concentration products are suitable for large-scale, long-term sea ice monitoring, but their spatial resolution is typically on the order of tens of kilometers. For edge ice zones and ice-water transition zones, a single passive microwave grid may simultaneously contain multiple underlying surface states such as open water, broken ice, thin ice, and dense sea ice, making it difficult to directly describe the ice-water mixing state at the laser footprint scale.

[0006] On April 16, 2022, my country launched its Atmospheric Environment Monitoring Satellite (DQ-1), carrying an Aerosol and Carbon Dioxide Detection Lidar (ACDL). This became the world's first successfully operational active greenhouse gas detection payload in orbit. It possesses multi-wavelength active laser detection capabilities at 532 nm, 1064 nm, and 1572 nm. The 532 nm channel provides information on atmospheric and near-surface scattering, while the 1572 nm channel primarily serves for carbon dioxide column concentration detection and can also record surface echo response information. Existing spaceborne laser sea ice applications mostly utilize 532 nm laser altimetry data to obtain parameters such as sea ice height, sea surface lead, and free plate, mainly addressing the estimation of sea ice elevation and thickness. No effective method for estimating sea ice concentration along orbit has been found. Summary of the Invention

[0007] To address the problems existing in the prior art, this invention provides a method and system for estimating sea ice concentration along orbit based on a spaceborne atmospheric sounding lidar ACDL.

[0008] The specific technical solution of the present invention is as follows:

[0009] A method for estimating sea ice concentration along orbit based on a spaceborne atmospheric sounding lidar (ACDL) includes the following steps:

[0010] S1: Acquire co-track data of the 1572nm and 532nm channels of the onboard ACDL and preprocess the data.

[0011] S2, based on 1572nm channel data, calculate characteristic parameters to characterize the response state of the underlying surface, construct a two-dimensional characteristic response space according to the characteristic parameters, screen samples and determine representative endmembers of three types of response endmembers; the characteristic parameters include pulse broadening exponent and logarithmic pseudoreflectivity;

[0012] S3. Based on the representative endmembers, construct an endmember matrix and a three-endmember linear mixture model, and calculate the abundance of three types of response endmembers in the 1572 nm response space of the laser footprint.

[0013] S4. Based on the near-surface optical parameters reflecting the endmember ice-water properties in the 532 nm channel, the weights of the three types of response endmember ice phase properties are determined, and the ice phase response of the three types of response endmember abundances at 1572 nm is synthesized according to the weights to obtain the estimated results of sea ice concentration along the orbit.

[0014] Further, in S1, the acquisition of co-track data of the 1572nm and 532nm channels of the spaceborne ACDL, and the preprocessing of the data, includes:

[0015] The 1572 nm channel data includes the latitude and longitude of the laser footprint, the channel echo integral value of the emitted energy, the channel echo peak value, the first pulse ranging auxiliary parameters, and the channel echo waveform; the 532 nm channel data includes the near-surface depolarization ratio and the backscattering coefficient.

[0016] The preprocessing includes clear sky data screening, quality control, land-sea masking, study area screening, and orbital data registration. The study area is selected from the polar edge ice zone, the ice break zone, and the ice-water transition zone.

[0017] Furthermore, the characteristic parameters mentioned in S2 include the pulse broadening index and the logarithmic pseudo-reflectivity, wherein the pulse broadening index and the logarithmic pseudo-reflectivity are defined as follows:

[0018] ;

[0019] In the formula, The pulse broadening index, This is the integral value of the echo from the 1572 nm online channel. The peak echo value is 1572 nm in the online channel.

[0020] ;

[0021] in, To logarithmize the pseudo-reflectivity, These are the one-way distance measurement parameters from the laser to the underlying surface. This is the energy emitted by the echo pulse.

[0022] Furthermore, in S2, when constructing the two-dimensional feature response space based on the aforementioned feature parameters, each laser footprint is mapped to the original two-dimensional feature response space composed of 1572 nm channel feature parameters. The original two-dimensional feature response space is represented as follows:

[0023] ;

[0024] In the formula, Indicates the first The feature vector of each laser footprint in the original two-dimensional feature response space Indicates the first The pulse broadening index of a laser footprint, Indicates the first Logarithmic pseudoreflectivity of each laser footprint;

[0025] To reduce the impact of differences in the dimensions of different feature quantities and extreme values ​​on the feature space distribution, a standardized two-dimensional feature response space is constructed based on the median and interquartile range. The eigenvectors of a laser footprint in the standardized two-dimensional feature response space are represented as:

[0026]

[0027] in, ; ;

[0028] In the formula, For the first Standardized pulse broadening index for each sample. Indicates the first Standardized logarithmic pseudoreflectance of each sample This means taking the median over *. This indicates taking the interquartile range for *.

[0029] Furthermore, the process of screening samples and determining representative endmembers for the three types of response endmembers as described in S2 is as follows:

[0030] Construct a candidate sample filtering function for three types of response endmembers (A, B, and C) to filter samples and establish a candidate sample set.

[0031] The candidate sample selection function for the low-broadening, high-response endmember A is:

[0032] ;

[0033] The candidate sample selection function for low-broadened, low-response endmember B is:

[0034] ;

[0035] The candidate sample selection function for the high-broadness response endmember C is:

[0036] ;

[0037] For each laser footprint, three types of response endmembers are selected, and the results are obtained respectively. , , Candidate sample sets are established for the top 1% of samples with the highest values. , , Calculate within the candidate set of each type of response terminator and The medians of the two dimensions are used as representative endmembers for the three types of response endmembers in the standardized two-dimensional feature response space:

[0038]

[0039]

[0040]

[0041] In the formula, A, B, and C are the representative endmembers of the three types of response endmembers A, B, and C in the standardized two-dimensional feature response space, respectively. For the first Standardized pulse broadening index for each sample. Indicates the first Standardized logarithmic pseudoreflectance of each sample For the first Standardized pulse broadening index for each sample. Indicates the first Standardized logarithmic pseudoreflectance of each sample For the first Standardized pulse broadening index for each sample. Indicates the first Standardized logarithmic pseudoreflectance of each sample.

[0042] Furthermore, S3 describes constructing an endmember matrix and a three-endmember linear mixture model based on the representative endmembers to calculate the abundance of three types of response endmembers in the 1572 nm response space of the laser footprint, specifically as follows:

[0043] Based on the three types of response endmembers, an endmember matrix is ​​constructed, and a three-endmember linear mixture model is built:

[0044] ;

[0045] ;

[0046] The abundance of A / B / C type response endmembers for each laser footprint in the 1572 nm channel response space was calculated using the fully constrained least squares method (constraints being non-negativity and summation to 1).

[0047] The established objective function is:

[0048] ;

[0049] The full constraints are:

[0050] , , ;

[0051] ;

[0052] In the formula, For endmember matrices, For the first The feature vector of a laser footprint in a standardized two-dimensional feature response space. For the first The abundance of three types of response endmembers of a laser footprint in a normalized two-dimensional feature response space, among which , , , The abundance of response endmembers are categorized into three types: A, B, and C. This represents the model residuals.

[0053] Further, in S4, the near-surface optical parameters reflecting endmember ice phase properties in the 532 nm channel are used to determine the weights of the three types of response endmember ice phase properties. Then, based on these weights, the abundance of the three types of response endmembers at 1572 nm is synthesized to obtain the estimated sea ice concentration along the orbit. Specifically:

[0054] Based on the three types of response endmembers (A / B / C) selected from the 1572 nm channel and the representative endmembers of the three types of response endmembers, the M samples closest to the representative endmembers of each type of response endmember are selected as the representative samples of each type of response endmember.

[0055] Representative samples of various response endmembers were matched with co-track data from the 532 nm channel to extract the near-surface depolarization ratio and backscattering coefficient corresponding to the 532 nm channel. Based on statistical analysis, the distribution differences of the near-surface depolarization ratio and backscattering coefficient in representative samples of different endmembers were compared to determine the ice phase attribute weights of the three types of response endmembers.

[0056] ;

[0057] in, ;

[0058] In the formula, This is the weight vector of ice phase properties constrained by the 532 nm channel. These represent the ice phase attribute weights of the three response endmembers, A, B, and C, respectively.

[0059] The following is a model for estimating the concentration of sea ice along the track, based on the weighting constraint of the 532 nm endmember ice phase attribute and the synthesis of the abundance of the three types of 1572 nm response endmembers:

[0060] .

[0061] Furthermore, the determination of ice phase attribute weights for the three types of response endmembers is based on statistical analysis to compare the distribution differences of near-surface depolarization ratio and backscattering coefficient in different endmember representative samples, specifically as follows:

[0062] Statistical analysis shows that the backscattering responses of type B and type C response endmembers are generally stronger than those of type A response endmembers in relation to the ice phase. Depolarization ratio analysis further indicates that type A response endmembers are weaker than both type B and type C response endmembers in relation to the ice phase. Therefore, the three types of response endmembers are determined as follows:

[0063] Class A response endmembers are low-broadening, high-response endmembers, mainly corresponding to non-icy phase responses;

[0064] Type B response endmembers are low-broadening, low-response endmembers, corresponding to ice phase-related responses;

[0065] Class C response endmembers are high-broadening response endmembers, corresponding to ice phase-related responses, and are associated with wide-pulse or strong scattering responses.

[0066] The weights for the ice phase attributes using binary endmembers are as follows:

[0067]

[0068] in, This indicates that the Class A response endmember is not included in the ice phase response. , q indicates that type B response terminators and type C response terminators are included in the ice phase response;

[0069] The following is a model for estimating the concentration of sea ice along the orbit based on the synthesis of the abundance of three types of endmember responses at 1572 nm under a 532 nm channel constraint:

[0070] .

[0071] This invention provides a system for estimating sea ice concentration along orbit based on a spaceborne atmospheric sounding lidar (ACDL), comprising the following modules:

[0072] The data acquisition module acquires co-track data from the 1572nm and 532nm channels of the onboard ACDL and preprocesses the data.

[0073] The representative endmember determination module calculates characteristic parameters for characterizing the response state of the underlying surface based on 1572nm channel data, constructs a two-dimensional characteristic response space based on the characteristic parameters, filters samples, and determines representative endmembers for three types of response endmembers; the characteristic parameters include pulse broadening exponent and logarithmic pseudoreflectivity;

[0074] The endmember abundance calculation module constructs an endmember matrix and a three-endmember linear mixture model based on the representative endmembers, and calculates the abundance of three types of response endmembers in the 1572 nm response space of the laser footprint.

[0075] The ice phase attribute constraint and sea ice concentration estimation module determines the weights of three types of response endmember ice phase attributes based on the near-surface optical parameters reflecting the endmember ice phase attributes in the 532 nm channel, and synthesizes the ice phase response of the three types of response endmember abundances at 1572 nm according to the weights to obtain the sea ice concentration estimation results along the track.

[0076] An electronic device includes a memory and a processor, the memory storing a computer program, and the processor being configured to invoke and run the computer program stored in the memory to perform the method described above.

[0077] Compared with the prior art, the present invention has the following advantages:

[0078] This invention presents a method for estimating along-orbit sea ice concentration based on a spaceborne atmospheric sounding lidar (ACDL). Employing a two-dimensional feature space and linear abundance calculation framework, and based on the fact that the laser footprint in the two-dimensional feature response space mainly exhibits three stable response directions—low broadening high response, low broadening low response, and high broadening response—three types of endmember response modes are constructed, maintaining an effective number of endmembers and stability of abundance calculation. Simultaneously, by balancing the constraints of 532 nm endmember attributes, the proportion of continuous ice phase response at the footprint scale is obtained. This method can effectively represent along-orbit sea ice concentration variations in edge ice zones, broken ice zones, and ice-water transition zones.

[0079] This invention utilizes the DQ-1 ACDL satellite, which simultaneously possesses 1572 nm echo information and 532 nm optical parameters. The 1572 nm echo provides pulse broadening and echo intensity response information, while the 532 nm data provides optical parameters such as depolarization ratio and backscattering coefficient. By using the satellite-borne lidar's along-track footprint observation information, the continuous ice phase response ratio at the along-track footprint scale can be obtained, preserving the continuous variation information of the mixing state of sea ice and open water, thereby supplementing the shortcomings of coarse-resolution sea ice concentration products in representing small-scale mixing zones.

[0080] This invention leverages the multi-wavelength active laser detection capability of the DQ-1 ACDL payload, primarily utilizing the 1572nm channel echo information and the 532nm channel near-surface optical parameters for joint processing. Compared to some spaceborne laser-based sea ice studies that estimate sea ice height, sea surface lead, and free plate based on 532nm altimetry information, this invention further utilizes the complementary relationship between the 1572nm echo morphology and the 532nm optical parameters to jointly analyze the ice-water mixing state of the ocean surface, obtaining continuous sea ice concentration estimation results at the orbital footprint scale. Attached Figure Description

[0081] Figure 1 This is a schematic diagram of the overall process of the method of the present invention;

[0082] Figure 2 This invention presents the two-dimensional characteristic response space of the 1572 nm pulse broadening exponential-logarithmic pseudoreflectivity and the envelope diagram of the A / B / C response endmembers.

[0083] Figure 3 This is a normalized echo waveform diagram of the representative regions of the three types of 1572 nm response endmembers A, B, and C in this invention;

[0084] Figure 4 Box plots showing the peak amplitude distribution of the representative regions of the three types of 1572 nm response endmembers (A / B / C) in this invention;

[0085] Figure 5 This is a typical orbital research range and a background distribution map of NSIDC sea ice concentration.

[0086] Figure 6 The figure shows the 1572 nm response abundance, 532 nm near-surface optical parameters, and DQ-1 sea ice concentration along the typical orbit of this invention.

[0087] Figure 7 This is a scatter plot showing the DQ-1 sea ice concentration estimation results of this invention and the verification of NSIDC SIC products. Detailed Implementation

[0088] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. After reading this invention, any modifications of the invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0089] Example 1:

[0090] This example provides a method for estimating sea ice concentration along orbit based on a spaceborne (DQ-1) atmospheric sounding lidar (ACDL). Figure 1 As shown, it includes the following steps:

[0091] S1: Acquire co-track data of the 1572nm and 532nm channels of the onboard ACDL and preprocess the data.

[0092] S2, based on 1572nm channel data, calculate characteristic parameters to characterize the response state of the underlying surface, construct a 1572nm two-dimensional characteristic response space according to the characteristic parameters, screen samples and determine representative endmembers of three types of response endmembers; the characteristic parameters include pulse broadening exponent and logarithmic pseudoreflectivity;

[0093] S3, based on representative endmembers, constructs an endmember matrix and a three-endmember linear mixture model to calculate the abundance of three types of response endmembers in the 1572 nm response space of the laser footprint;

[0094] S4. Based on the near-surface optical parameters reflecting the endmember ice-water properties in the 532 nm channel, the weights of the three types of response endmember ice phase properties are determined, and the ice phase response of the three types of response endmember abundances at 1572 nm is synthesized according to the weights to obtain the estimated results of sea ice concentration along the track.

[0095] This invention utilizes the 1572 nm echo characteristics to construct response endmembers and calculate their abundance. Simultaneously, it constrains the ice-water properties of the response endmembers using 532 nm optical parameters along the same track, thereby obtaining sea ice concentration estimation results at the track footprint scale. Existing passive microwave sea ice concentration products have coarse resolution and are difficult to directly describe the ice-water mixing state at the laser footprint scale. This invention can obtain the continuous ice phase response ratio at the track footprint scale and further form sea ice concentration estimation results, improving the expressive power of sea ice concentration products in small-scale ice-water mixing regions. Passive microwave SIC products typically reflect the average sea ice coverage ratio within a large grid (25 km × 25 km), while this invention calculates SIC point-by-point for the DQ-1 laser footprint, generating a continuous variation curve along the track. In the ice-water transition region, this curve can show a continuous change (or from high to low) from low sea ice concentration areas to high sea ice concentration areas, thus supplementing the small-scale ice-water mixing structure that coarse-resolution SIC products struggle to express.

[0096] While the 1572 nm band primarily serves for atmospheric carbon dioxide column concentration detection, its surface echo signals also contain surface reflection information. For oceanic ice-water transition zones, the peak value, integral value, and waveform broadening of the 1572 nm echo can reflect the different responses of the underlying surface to short-wave infrared lasers. Simultaneously, the near-surface depolarization ratio and backscattering coefficient in the 532 nm band can provide optical parameters related to the near-surface ice-water surface scattering characteristics. Therefore, the DQ-1 ACDL can not only provide along-track laser footprint observations but also utilize the complementary relationship between the 1572 nm echo morphology and the 532 nm optical parameters to jointly analyze the ice-water mixing state.

[0097] This invention fully utilizes the characteristics of the DQ-1 ACDL dual-band payload to achieve joint processing of the 1572 nm echo response and 532 nm optical parameters. The 1572 nm response endmember primarily reflects the echo morphology and cannot be directly equated to a specific ice-water type based solely on two-dimensional characteristics. This invention utilizes the near-surface depolarization ratio and backscattering coefficient of the 532 nm endmember on the same platform and in the same orbit to provide auxiliary constraints for incorporating the ice phase response into the A / B / C types of 1572 nm response endmembers, thereby further converting the 1572 nm endmember abundance into sea ice concentration estimation results with ice phase implications.

[0098] Example 2:

[0099] A further design feature of this embodiment is that, in step S1, the co-track data of the 1572nm and 532nm channels of the onboard ACDL are acquired, and the data is preprocessed. The specific design is as follows:

[0100] The 1572 nm channel data includes the latitude and longitude of the laser footprint, the channel echo integral value of the emitted energy, the channel echo peak value, the first pulse ranging auxiliary parameters, and the channel echo waveform; the 532 nm channel data includes the near-surface depolarization ratio and the backscattering coefficient.

[0101] This embodiment uses DQ-1 ACDL payload data as an example, utilizing the 1572 nm online channel and the 532 nm profile data from the same platform to identify the ice-water mixing state of the marine underlying surface and estimate sea ice concentration in the ice-water transition zone of eastern Greenland. The DQ-1 ACDL 1572 nm band includes both online and offline channels. The online channel is located at a carbon dioxide absorption-sensitive position, and its echo intensity may be affected by atmospheric CO2 absorption and optical path changes; the offline channel is relatively less affected by CO2 absorption. This embodiment previously compared and analyzed the online and offline data, showing that both exhibit echo response trends related to the ice-water transition within the study area. This application selects the online channel as an example to demonstrate that even with some atmospheric absorption in the online channel, a stable 1572 nm response endmember can still be extracted, and along-track estimation results consistent with sea ice concentration changes can be obtained, thus demonstrating the robustness of this method. The offline channel can be used as an alternative input or auxiliary verification channel for this method.

[0102] The following variables were extracted from the data obtained from the DQ-1 ACDL L2B satellite probe:

[0103] (1) Footprint latitude (Lidar_Foot_Latitude);

[0104] (2) Footprint longitude (Lidar_Foot_Longitude);

[0105] (3) 1572 nm PulseA emission energy Pulse_Energy_1572_PulseA;

[0106] (4) 1572 nm online channel echo integral value Pon_Integral_Value;

[0107] (5) 1572 nm online channel echo peak value (Pon_Peak_Value);

[0108] (6) Distance_WaveStartToLidar_1572_FirstPulse, a one-way distance measurement parameter from the 1572 nm laser to the underlying surface;

[0109] (7) 1572 nm online channel echo waveform Pon_Pulse_Wave;

[0110] (8) Depolarization ratio VDR and backscattering coefficient β in 532 nm profile data.

[0111] Data preprocessing includes clear-sky data screening, quality control, land-sea masking, study area screening, and orbital data registration.

[0112] Basic data validity screening is performed, with screening criteria including: valid latitude and longitude, transmitted energy greater than 0, ranging parameters greater than 0, and valid echo integral and peak values ​​greater than 0. To suppress the influence of weak signals and noise, a peak echo threshold is preferably set for the online channel. mV.

[0113] The study area was selected, and this invention chose the ice-water transition zone in the eastern waters of Greenland as the research scope. An example area within the ice-water transition zone of eastern Greenland used for endmember extraction is as follows: , .

[0114] Simultaneously, a land-sea masking technique was employed using land vector boundaries to eliminate land points, retaining only the oceanic surface footprint. The study area was selected in the ice-water transition zone between the NSIDC sea ice concentration area and open water, ensuring that along-orbit observations covered the changing scenarios from low to high sea ice concentration.

[0115] Example 3:

[0116] In a further design of this embodiment, in S2, characteristic parameters for characterizing the response state of the underlying surface are calculated based on 1572nm channel data. A two-dimensional characteristic response space of 1572nm is constructed according to the characteristic parameters. Samples are screened and representative endmembers of three types of response endmembers are determined. The characteristic parameters include the pulse broadening exponent and the logarithmic pseudoreflectivity, specifically as follows:

[0117] Based on 1572 nm online channel data, two features describing the underlying surface response state are extracted: the pulse broadening exponent and the logarithmic pseudoreflectivity are defined as follows:

[0118] ;

[0119] In the formula, The pulse broadening index, This is the integral value of the echo from the 1572 nm online channel. The peak value of the echo in the online channel is 1572 nm.

[0120] The pulse broadening index describes the extent to which the echo energy expands over time. When the echo energy is concentrated within a narrow pulse, the integral value is smaller relative to the peak value, and the broadening index is lower. When there is volume scattering, multiple scattering, or complex rough surface response, the echo waveform widens, the integral value increases relative to the peak value, and the broadening index rises.

[0121] ;

[0122] in, To logarithmize the pseudo-reflectivity, These are the one-way distance measurement parameters from the laser to the underlying surface. This is the energy emitted by the echo pulse.

[0123] This feature, by introducing a distance-squared term and an emission energy term, relatively normalizes the echo intensity for different orbits or observation distances. Since this parameter is not the true surface reflectivity after complete radiometric calibration, it is called pseudo-reflectivity.

[0124] When constructing the two-dimensional feature response space based on the feature parameters, each laser footprint is mapped to the original two-dimensional feature response space composed of 1572 nm channel feature parameters. The 1572 nm two-dimensional feature response space is represented as:

[0125] ;

[0126] In the formula, Indicates the first The feature vector of each laser footprint in the original two-dimensional feature response space Indicates the first The pulse broadening index of a laser footprint, Indicates the first Logarithmic pseudoreflectivity of each laser footprint;

[0127] Pulse broadening index and logarithmic pseudoreflectivity Truncating the values ​​and removing extreme outliers below 0.5% and above 99.5%, a standard two-dimensional feature response space is constructed using the median and interquartile range. The eigenvectors of a laser footprint in the standardized two-dimensional feature response space are represented as:

[0128]

[0129] in, ; ;

[0130] In the formula, For the first Standardized pulse broadening index for each sample. Indicates the first Standardized logarithmic pseudoreflectance of each sample This means taking the median over *. This indicates taking the interquartile range for *.

[0131] In the ice-water transition zone sample of this embodiment, the effective footprint mainly exhibits three stable response directions in the two-dimensional feature response space: low broadening high response, low broadening low response, and high broadening response. Therefore, this invention constructs type A response endmembers, type B response endmembers, and type C response endmembers to characterize the above three main response modes.

[0132] In this embodiment, within the two-dimensional feature space and linear abundance calculation framework, if only two endmembers are used, one of the three main response directions needs to be merged into the other endmembers. This can easily lead to some footprints being approximated by interpolation using only two endmembers, weakening the expressive power of endmember abundance for different echo response modes. Adding more endmembers increases the number of abundance variables and endmember attribute assignments. With limited two-dimensional observation features, additional endmembers may introduce situations where multiple abundance combinations can reconstruct the same observation point, reducing the stability of abundance calculation. Therefore, a three-endmember structure is preferred. This structure can cover the main variation directions of the 1572 nm echo in both response strength and waveform broadening dimensions, while maintaining a balance between the number of endmembers, the stability of abundance calculation, and the subsequent 532 nm endmember attribute constraints.

[0133] This invention uses FCLS as the constraint solution method for the endmember abundance calculation stage of this application, ensuring that the A / B / C abundance of each footprint satisfies the non-negativity and sum-to-one constraints, so that the contribution of the A / B / C response endmembers can be represented in a continuous proportional form, avoiding the output of only discrete ice-water category results.

[0134] The samples were screened and representative endmembers for the three types of response endmembers were determined, specifically as follows:

[0135] Construct a candidate sample filtering function for three types of response endmembers (A, B, and C) to filter samples and establish a candidate sample set.

[0136] The candidate sample selection function for the low-broadening, high-response endmember A is:

[0137] ;

[0138] The candidate sample selection function for low-broadened, low-response endmember B is:

[0139] ;

[0140] The candidate sample selection function for the high-broadness response endmember C is:

[0141] ;

[0142] For each laser footprint, three types of response endmembers are selected, and the results are obtained respectively. , , Candidate sample sets are established for the top 1% of samples with the highest values. , , Calculate within the candidate set of each type of response terminator and The medians of the two dimensions are used as representative endmembers for the three types of response endmembers in the standardized two-dimensional feature response space:

[0143]

[0144]

[0145]

[0146] In the formula, A, B, and C are the representative endmembers of the three types of response endmembers A, B, and C in the standardized two-dimensional feature response space, respectively. For the first Standardized pulse broadening index for each sample. Indicates the first Standardized logarithmic pseudoreflectance of each sample For the first Standardized pulse broadening index for each sample. Indicates the first Standardized logarithmic pseudoreflectance of each sample For the first Standardized pulse broadening index for each sample. Indicates the first The standardized logarithmic pseudo-reflectance of each sample. Using the median of the candidate set to determine the endmember coordinates can reduce the influence of individual extreme samples on the endmember positions, making the A / B / C endmembers more representative of the stable center positions in their respective response directions.

[0147] In this embodiment, the candidate sample set mentioned above is... and The combination of the medians of the two standardized feature dimensions serves as the representative endmembers of the A / B / C class response endmembers in the standardized two-dimensional feature response space. A specific example is shown below:

[0148]

[0149]

[0150] .

[0151] Example 4:

[0152] In this embodiment, the design further involves constructing an endmember matrix and a three-endmember linear mixture model based on representative endmembers in S3, and calculating the abundance of three types of response endmembers in the 1572 nm response space of the laser footprint.

[0153] An endmember matrix is ​​established based on the representative coordinates of the three types of response endmembers, and a three-endmember linear mixture model is constructed:

[0154] ;

[0155] ;

[0156] The abundance of A / B / C type response endmembers for each laser footprint in the 1572 nm channel response space was calculated using the fully constrained least squares method (constraints being non-negativity and summation to 1).

[0157] The established objective function is:

[0158] ;

[0159] The full constraints are:

[0160] , , ;

[0161] ;

[0162] In the formula, For endmember matrices, For the first The feature vector of a laser footprint in a standardized two-dimensional feature response space. For the first The abundance of three types of response endmembers of a laser footprint in the standardized two-dimensional feature response space, among which , , , The abundance of response endmembers are categorized into three types: A, B, and C. This represents the model residuals.

[0163] This invention calculates the 1572 nm three-endmember response abundance using fully constrained least squares and computes the SiC under 532 nm endmember attribute constraints, outputting continuous proportional values ​​rather than simple ice / water binary labels. This method can be used to provide information on along-rail sea ice concentration variations in marginal ice zones, fractal ice zones, and ice-water transition zones.

[0164] This invention utilizes the complementary relationship between 1572 nm channel data and 532 nm optical parameters. The 1572 nm channel is primarily used to extract echo morphology features and calculate the abundance of the A / B / C type response endmembers. However, this abundance primarily represents the composition of the 1572 nm response mode and should not be directly equated to ice-water properties. The 532 nm channel constrains the ice phase properties of the A / B / C type response endmembers through near-surface depolarization ratio and backscattering coefficient. The 1572 nm channel addresses the question of "how much of each type of response accounts for," while the 532 nm channel addresses the question of "which responses are included in the ice phase." Together, they form the DQ-1 sea ice concentration estimation result.

[0165] Example 5:

[0166] In a further design of this embodiment, in S4, based on the near-surface optical parameters reflecting the endmember ice-water properties in the 532 nm channel, the weights of the three types of response endmember ice phase properties are determined, and the ice phase response is synthesized based on the abundance of the three types of response endmembers at 1572 nm according to the weights to obtain the estimated results of the sea ice concentration along the track.

[0167] Based on the three types of response endmembers (A / B / C) selected from the 1572 nm channel and the representative endmembers of the three types of response endmembers, the M samples closest to the representative endmembers of each type of response endmember are selected. In this example, 150 samples are selected as the representative samples of each type of response endmember.

[0168] Representative samples of various response endmembers were matched with co-track data from the 532 nm channel. Nearest neighbor matching was performed on the dual-band data. For the matched 532 nm profile, optical parameters were extracted within the near-surface altitude range. The altitude range selected in this example is: The near-surface depolarization ratio (VDR) and backscattering coefficient β corresponding to the 532 nm channel were extracted. Based on statistical analysis, the distribution differences of VDR and backscattering coefficient in different endmember representative samples were compared to determine the ice phase attribute weights of the three types of response endmembers.

[0169] ;

[0170] in, ;

[0171] In the formula, This is the weight vector of ice phase properties constrained by the 532 nm channel. These represent the ice phase attribute weights of the three response endmembers, A, B, and C, respectively.

[0172] The following is a model for estimating the concentration of sea ice along the orbit based on the synthesis of the abundance of three types of endmember responses at 1572 nm under a 532 nm channel constraint:

[0173] .

[0174] Example 6:

[0175] This embodiment further involves determining the ice facies attribute weights of the three types of response endmembers based on statistical analysis and comparison of the distribution differences of the near-surface deflection ratio (VDR) and backscattering coefficient β in different endmember representative samples. Specifically:

[0176] Statistical analysis shows that the backscattering responses of type B and type C response endmembers are generally stronger than those of type A response endmembers in relation to the ice phase. Depolarization ratio analysis further indicates that type A response endmembers are weaker than both type B and type C response endmembers in relation to the ice phase. Therefore, the three types of response endmembers are determined as follows:

[0177] Class A response endmembers are low-broadening, high-response endmembers, mainly corresponding to non-icy phase responses;

[0178] Type B response endmembers are low-broadening, low-response endmembers, corresponding to ice phase-related responses;

[0179] Class C response endmembers are high-broadening response endmembers, corresponding to ice phase-related responses, and are associated with wide-pulse or strong scattering responses.

[0180] The weights for the ice phase attributes using binary endmembers are as follows:

[0181]

[0182] in, This indicates that the Class A response endmember is not included in the ice phase response. , q indicates that type B response terminators and type C response terminators are included in the ice phase response;

[0183] The following is a model for estimating the concentration of sea ice along the orbit based on the synthesis of the abundance of three types of endmember responses at 1572 nm under a 532 nm channel constraint:

[0184] .

[0185] In this embodiment, statistical analysis was performed on the near-surface depolarization ratio and backscattering coefficient of the extracted 532 nm channel. Representative samples from the three response endmembers (A, B, and C) showed differences in the statistical distribution of 532 nm optical parameters. The backscattering coefficient primarily reflects the differences in near-surface scattering intensity corresponding to different endmembers, while the depolarization ratio helps reflect the differences in polarization state changes corresponding to different endmembers. Statistical comparison results show that the backscattering responses corresponding to endmembers B and C are generally stronger than those corresponding to endmember A. The depolarization ratio, as an auxiliary diagnostic measure, also supports the existence of attribute differences between endmember A and endmembers B and C. Therefore, endmember A is set to not include the ice phase response, while endmembers B and C are set to include the ice phase response, specifically:

[0186] A-terminal: Low-broadening, high-response terminal, mainly corresponding to non-icy phase responses;

[0187] B-terminal: Low-broadening, low-response terminal, corresponding to ice-phase related response;

[0188] C-terminal: High-broadness response end-terminal, corresponding to ice phase-related response, which is related to wide pulse or strong scattering response;

[0189] This example uses binary endmember ice phase attribute weights:

[0190]

[0191] in, This indicates that the A-endmember is not included in the ice phase response. , q indicates that the B-endmember and C-endmember are included in the ice phase response.

[0192] The 532 nm near-surface depolarization ratio and backscattering coefficient are used in this method to determine the endmember glacial facies weights and are not involved in the FCLS calculation of the 1572 nm three-endmember abundance. (1572 nm three-endmember abundance) Used to represent the relative contributions of the three types of response endmembers (A, B, and C) in each footprint. This is used to determine whether each endmember includes the ice phase response. The combination of these two factors yields the DQ-1 sea ice concentration estimation result:

[0193]

[0194] In this embodiment, due to ,therefore:

[0195] .

[0196] This invention provides a system for estimating sea ice concentration along orbit based on a spaceborne atmospheric sounding lidar (ACDL), comprising the following modules:

[0197] The data acquisition module acquires co-track data from the 1572nm and 532nm channels of the onboard ACDL and preprocesses the data.

[0198] The representative endmember determination module calculates characteristic parameters for characterizing the response state of the underlying surface based on 1572nm channel data, constructs a two-dimensional characteristic response space based on the characteristic parameters, filters samples, and determines representative endmembers for three types of response endmembers; the characteristic parameters include pulse broadening exponent and logarithmic pseudoreflectivity;

[0199] The endmember abundance calculation module constructs an endmember matrix and a three-endmember linear mixture model based on the representative endmembers, and calculates the abundance of three types of response endmembers in the 1572 nm response space of the laser footprint.

[0200] The ice phase attribute constraint and sea ice concentration estimation module determines the weights of three types of response endmember ice phase attributes based on the near-surface optical parameters reflecting the endmember ice phase attributes in the 532 nm channel, and synthesizes the ice phase response of the three types of response endmember abundances at 1572 nm according to the weights to obtain the sea ice concentration estimation results along the track.

[0201] Example 7:

[0202] This example provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor is used to call and run the computer program stored in the memory to perform the method described above.

[0203] Verification example:

[0204] This embodiment is used to verify the differences in the 1572 nm echo response of the three types of response endmembers (A / B / C) and to verify the effectiveness of the DQ-1 ACDL sea ice concentration estimation results of this invention.

[0205] This example first extracts the waveform characteristics and peak amplitude distribution of the representative regions of the three types of response endmembers (A, B, and C) to verify the differences in the 1572 nm echo response of the three types of response endmembers.

[0206] Figure 2 The figure illustrates the envelopes of the A / B / C response endmembers in the original two-dimensional feature response space. As can be seen from the figure, the laser footprint samples are mainly distributed in three response directions in the two-dimensional feature space. The A-type response endmembers are located in the low-broadening, high-echo-response region; the B-type response endmembers are located in the low-broadening, low-echo-response region; and the C-type response endmembers are located in the high-broadening, high-echo-response region. The determination of the A / B / C-type response endmembers is based on the standardized two-dimensional feature response space. Figure 2 The original two-dimensional feature response space is used for display to preserve the physical meaning of the pulse broadening exponent and the logarithmic pseudoreflectivity.

[0207] To further verify the differences in the actual echo waveforms corresponding to the three types of response endmembers (A, B, and C), this embodiment, based on the three types of response endmembers (A, B, and C) determined in Embodiment 3, calculates the distance between each sample and its corresponding endmember in the standardized two-dimensional characteristic response space, and selects several samples closest to the endmembers as representative region samples to analyze their original 1572 nm echo waveforms.

[0208] The specific process involves calculating the distance from each laser footprint sample to the three response endmembers A, B, and C in a standardized two-dimensional feature space. Samples are then assigned based on the nearest endmember, and several samples closest to the representative endmember are selected as the waveform analysis region for each endmember class. For example, 150 representative samples are selected for each endmember class.

[0209] Since the endmember representative region samples are determined through a standardized two-dimensional feature response space, their corresponding footprint indexes can be used to look up the Pon_Pulse_Wave echo waveform data in the original HDF5 file. Background estimation and subtraction are performed on the original waveform. Using the echo peak position as the alignment reference, waveform windows of the 60 bins before and 120 bins after the peak are extracted and normalized according to the peak value of each waveform. Figure 3 The normalized median waveform and interquartile ranges for regions A, B, and C are shown, with the median waveform as the main curve and the 25%–75% quantile range as the shaded area. The figure shows that the waveforms of the type A and type B response endmembers are generally narrower, with energy concentrated near the peak; the waveform of the type C response endmember is relatively wider, with a longer post-peak decay process, exhibiting a more pronounced broadening characteristic.

[0210] Figure 4 The figure illustrates the differences in the distribution of peak amplitudes across regions A, B, and C. As shown, the peak amplitude of type A response endmembers is generally higher than that of types B and C response endmembers, consistent with the characteristics of low broadening and high response; the peak amplitude of type B response endmembers is generally lower, consistent with the characteristics of low broadening and low response; while the peak amplitude of type C response endmembers is not as high as that of type A response endmembers, but considering... Figure 3 The wide pulse characteristic in the waveform exhibits a significant waveform broadening response.

[0211] Test example:

[0212] This example uses data from the 1572 nm and 532 nm channels on the same observation orbit of the DQ-1 ACDL on March 21, 2023, as a typical case. This orbit passes through the ice-water transition zone in eastern Greenland, covering the change from areas of lower sea ice concentration to areas of higher sea ice concentration along the orbital direction.

[0213] Figure 5This figure illustrates the spatial location and study area of ​​this typical orbit within the NSIDC sea ice concentration background field. The figure shows NSIDC sea ice concentration data products; the red line represents the DQ-1 ACDL orbit selected in this application, and the blue box represents the target region studied in this application. The NSIDC (National Snow and Ice Data Center) sea ice concentration data product is also shown.

[0214] Figure 6 The results show the 1572 nm logarithmic pseudoreflectivity, pulse broadening index, A / B / C type endmember abundance, 532 nm near-surface debiasing ratio, and DQ-1 sea ice concentration estimation for this orbit.

[0215] In the results along the track, , , These represent the relative contributions of the three types of 1572 nm response endmembers (A, B, and C) in different footprints. After constraining the 532 nm endmember attributes, the along-track sea ice concentration estimation model is as follows:

[0216]

[0217] As the orbit moves from low to high latitudes through the ice-water transition zone, this estimate can reflect the spatial variation trend of sea ice concentration from low to high.

[0218] NSIDC SIC verification example:

[0219] To evaluate the estimated DQ-1 sea ice concentration, this embodiment uses NSIDC passive microwave sea ice concentration products for verification.

[0220] Figure 7 A scatter plot comparing the DQ-1 sea ice concentration estimation results with the NSIDC SIC product is presented. The validation results show that the sea ice concentration estimation results along the track obtained by this invention are consistent with the spatial variation trend of the NSIDC passive microwave product; at the same time, compared with the NSIDC SIC product, this invention can provide continuous variation information at the track footprint scale, which is more suitable for characterizing the small-scale ice-water mixing structure inside the edge ice zone and the ice-water transition zone.

Claims

1. A method for estimating sea ice concentration along orbit based on a spaceborne atmospheric sounding lidar (ACDL), characterized in that, Includes the following steps: S1: Acquire co-track data of the 1572nm and 532nm channels of the onboard ACDL and preprocess the data. S2, calculate the characteristic parameters used to characterize the response state of the underlying surface based on the 1572nm channel data, construct a two-dimensional characteristic response space according to the characteristic parameters, screen samples and determine the representative endmembers of the three types of response endmembers; The characteristic parameters include the pulse broadening exponent and the logarithmic pseudoreflectivity; S3. Based on the representative endmembers, construct an endmember matrix and a three-endmember linear mixture model, and calculate the abundance of three types of response endmembers in the 1572 nm response space of the laser footprint. S4. Based on the near-surface optical parameters reflecting the endmember ice-water properties in the 532 nm channel, the weights of the three types of response endmember ice phase properties are determined, and the ice phase response of the three types of response endmember abundances at 1572 nm is synthesized according to the weights to obtain the estimated results of sea ice concentration along the orbit.

2. The method for estimating along-orbit sea ice concentration based on spaceborne atmospheric sounding lidar (ACDL) according to claim 1, characterized in that, In S1, the acquisition of co-track data for the 1572nm and 532nm channels of the spaceborne ACDL, and the preprocessing of the data, include: The 1572 nm channel data includes the latitude and longitude of the laser footprint, the channel echo integral value of the emitted energy, the channel echo peak value, the first pulse ranging auxiliary parameters, and the channel echo waveform; the 532 nm channel data includes the near-surface depolarization ratio and the backscattering coefficient. The preprocessing includes clear sky data screening, quality control, land-sea masking, study area screening, and orbital data registration. The study area is selected from the polar edge ice zone, the ice break zone, and the ice-water transition zone.

3. The method for estimating along-orbit sea ice concentration based on spaceborne atmospheric sounding lidar ACDL according to claim 1, characterized in that, The characteristic parameters mentioned in S2 include the pulse broadening index and the logarithmic pseudo-reflectivity, wherein the pulse broadening index and the logarithmic pseudo-reflectivity are defined as follows: ； In the formula, The pulse broadening index, This is the integral value of the echo from the 1572 nm online channel. The peak echo value is 1572 nm in the online channel. ； in, To logarithmize the pseudo-reflectivity, These are the one-way distance measurement parameters from the laser to the underlying surface. This is the energy emitted by the echo pulse.

4. The method for estimating along-orbit sea ice concentration based on spaceborne atmospheric sounding lidar ACDL according to claim 3, characterized in that, In S2, when constructing the two-dimensional feature response space based on the aforementioned feature parameters, each laser footprint is mapped to the original two-dimensional feature response space composed of 1572 nm channel feature parameters. The original two-dimensional feature response space is represented as follows: ； In the formula, Indicates the first The feature vector of each laser footprint in the original two-dimensional feature response space Indicates the first The pulse broadening index of a laser footprint, Indicates the first Logarithmic pseudoreflectivity of each laser footprint; A standardized two-dimensional feature response space is constructed based on the median and interquartile range. The eigenvectors of a laser footprint in the standardized two-dimensional feature response space are represented as: ； in, ; ; In the formula, For the first Standardized pulse broadening index for each sample. Indicates the first Standardized logarithmic pseudoreflectance of each sample This means taking the median over *. This indicates taking the interquartile range for *.

5. The method for estimating along-orbit sea ice concentration based on spaceborne atmospheric sounding lidar ACDL according to claim 4, characterized in that, The process of screening samples and determining representative endmembers for the three types of response endmembers, as described in S2, is as follows: Construct a candidate sample filtering function for three types of response endmembers (A, B, and C) to filter samples and establish a candidate sample set. The candidate sample selection function for the low-broadening, high-response endmember A is: ； The candidate sample selection function for low-broadened, low-response endmember B is: ； The candidate sample selection function for the high-broadness response endmember C is: ； For each laser footprint, three types of response endmembers are selected, and the results are obtained respectively. , , Candidate sample sets are established for the top 1% of samples with the highest values. , , Calculate within the candidate set of each type of response terminator and The medians of the two dimensions are used as representative endmembers for the three types of response endmembers in the standardized two-dimensional feature response space: ； ； ； In the formula, A, B, and C are the representative endmembers of the three types of response endmembers A, B, and C in the standardized two-dimensional feature response space, respectively. For the first Standardized pulse broadening index for each sample. Indicates the first Standardized logarithmic pseudoreflectance of each sample For the first Standardized pulse broadening index for each sample. Indicates the first Standardized logarithmic pseudoreflectance of each sample For the first Standardized pulse broadening index for each sample. Indicates the first Standardized logarithmic pseudoreflectance of each sample.

6. The method for estimating along-orbit sea ice concentration based on spaceborne atmospheric sounding lidar ACDL according to claim 5, characterized in that, The method described in S3, which constructs an endmember matrix and a three-endmember linear mixture model based on the representative endmembers, and calculates the abundance of three types of response endmembers in the 1572 nm response space of the laser footprint, is as follows: Based on the three types of response endmembers, an endmember matrix is ​​constructed, and a three-endmember linear mixture model is built: ； ； The abundance of the three types of response endmembers (A, B, and C) in the 1572 nm channel response space for each laser footprint was calculated using the fully constrained least squares method. The established objective function is: ； The full constraints are: ， ， ； ； In the formula, For endmember matrices, For the first The feature vector of a laser footprint in a standardized two-dimensional feature response space. For the first The abundance of three types of response endmembers of a laser footprint in a normalized two-dimensional feature response space, among which , , , The abundance of response endmembers are categorized into three types: A, B, and C. This represents the model residuals.

7. The method for estimating along-orbit sea ice concentration based on spaceborne atmospheric sounding lidar ACDL according to claim 1, characterized in that, As described in S4, based on the near-surface optical parameters reflecting endmember ice phase properties in the 532 nm channel, the weights of three types of response endmember ice phase properties are determined. Then, based on these weights, the abundance of the three types of response endmembers at 1572 nm is synthesized to obtain the estimated sea ice concentration along the orbit. Specifically: Based on the three types of response endmembers (A / B / C) selected from the 1572 nm channel and the representative endmembers of the three types of response endmembers, the M samples closest to the representative endmembers of each type of response endmember are selected as the representative samples of each type of response endmember. Representative samples of various response endmembers were matched with co-track data from the 532 nm channel to extract the near-surface depolarization ratio and backscattering coefficient corresponding to the 532 nm channel. Based on statistical analysis, the distribution differences of the near-surface depolarization ratio and backscattering coefficient in representative samples of different endmembers were compared to determine the ice phase attribute weights of the three types of response endmembers. ； in, ; ; In the formula, This is the weight vector of ice phase properties constrained by the 532 nm channel. These represent the ice phase attribute weights of the three response endmembers, A, B, and C, respectively. The following is a model for estimating the concentration of sea ice along the track, based on the weighting constraint of the 532 nm endmember ice phase attribute and the synthesis of the abundance of the three types of 1572 nm response endmembers: 。 8. The method for estimating along-orbit sea ice concentration based on spaceborne atmospheric sounding lidar ACDL according to claim 7, characterized in that, The method involves statistically analyzing and comparing the distribution differences of near-surface deflection ratio and backscattering coefficient in different endmember representative samples to determine the ice facies attribute weights of the three types of response endmembers, specifically as follows: Statistical analysis shows that the backscattering responses of type B and type C response endmembers are generally stronger than those of type A response endmembers in relation to the ice phase. Depolarization ratio analysis further indicates that type A response endmembers are weaker than both type B and type C response endmembers in relation to the ice phase. Therefore, the three types of response endmembers are determined as follows: Class A response endmembers are low-broadening, high-response endmembers, mainly corresponding to non-icy phase responses; Type B response endmembers are low-broadening, low-response endmembers, corresponding to ice phase-related responses; Class C response endmembers are high-broadening response endmembers, corresponding to ice phase-related responses, and are associated with wide-pulse or strong scattering responses. The weights for the ice phase attributes using binary endmembers are as follows: ； in, This indicates that the Class A response endmember is not included in the ice phase response. , q indicates that type B response terminators and type C response terminators are included in the ice phase response; The following is a model for estimating the concentration of sea ice along the orbit based on the synthesis of the abundance of three types of endmember responses at 1572 nm under a 532 nm channel constraint: 。 9. A system for estimating sea ice concentration along orbit based on a spaceborne atmospheric sounding lidar (ACDL), characterized in that, Includes the following modules: The data acquisition module acquires co-track data from the 1572nm and 532nm channels of the onboard ACDL and preprocesses the data. The representative end-member determination module calculates characteristic parameters for characterizing the response state of the underlying surface based on 1572nm channel data, constructs a two-dimensional characteristic response space based on the characteristic parameters, filters samples, and determines representative end-members for three types of response end-members. The characteristic parameters include the pulse broadening exponent and the logarithmic pseudoreflectivity; The endmember abundance calculation module constructs an endmember matrix and a three-endmember linear mixture model based on the representative endmembers, and calculates the abundance of three types of response endmembers in the 1572 nm response space of the laser footprint. The ice phase attribute constraint and sea ice concentration estimation module determines the weights of three types of response endmember ice phase attributes based on the near-surface optical parameters reflecting the endmember ice phase attributes in the 532 nm channel, and synthesizes the ice phase response of the three types of response endmember abundances at 1572 nm according to the weights to obtain the sea ice concentration estimation results along the track.

10. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor being configured to invoke and run the computer program stored in the memory to perform the method as described in any one of claims 1 to 7.

Images