A machine learning-based ocean polarized lidar multiple scattering correction method

Abstract

The application discloses a kind of ocean polarization lidar multiple scattering correction methods based on machine learning, comprising: based on polarization Monte Carlo method, construct ocean polarization lidar radiation transfer numerical model, set multiple groups of lidar hardware and water body environment parameters, obtain a large number of simulation data, utilize machine learning method to construct the polarization multiple scattering correction model of different detection system, and use non-uniform water body to verify the effectiveness of model.According to the multiple scattering model of the effective attenuation coefficient of the measured signal and the lidar of different platforms, the signal depolarization ratio and water body parameter are calculated, and the polarization multiple scattering correction model is brought into the polarization multiple scattering correction model combined with the system parameters of lidar, to obtain seawater backscattering depolarization ratio.Compared with the direct seawater backscattering depolarization ratio of lidar signal depolarization ratio, the error between the seawater backscattering depolarization ratio corrected by the method and the in-situ value is significantly reduced.

Description

Technical Field

[0001] This invention belongs to the field of marine lidar signal processing technology, and in particular relates to a multiscattering correction method for marine polarization lidar based on machine learning. Background Technology

[0002] Seawater polarization characteristics are crucial information reflecting water composition and play a vital role in marine biodiversity, carbon cycle, and climate change research. Seawater polarization lidar can efficiently detect vertical profiles of upper ocean polarization characteristics, which is significant for distinguishing phytoplankton populations, accurately assessing phytoplankton biomass and upper ocean primary productivity, and thus reducing uncertainties in carbon cycle and climate change studies.

[0003] For example, Chinese patent document CN121541171A discloses a marine lidar system and a method for detecting multiple parameters of seawater; Chinese patent document CN110673157A discloses a hyperspectral resolution lidar system for detecting marine optical parameters.

[0004] However, in seawater, laser transmission involves a complex multiple scattering process. Under the influence of multiple scattering, the depolarization state of the laser changes, making it impossible for the depolarization ratio of the polarization lidar signal to accurately reflect the inherent backscattering polarization characteristics of seawater.

[0005] Existing research indicates that in strongly scattering water bodies, the signal depolarization caused by multiple scattering can be more than twice that of seawater backscattering depolarization, and this effect increases with depth. Multiple scattering depolarization severely impacts the measurement accuracy of seawater backscattering polarization characteristics by lidar. Therefore, the multiple scattering depolarization effect in polarization lidar is the main reason limiting the accuracy of seawater polarization characteristic detection. Correcting the impact of multiple scattering on the signal depolarization ratio is crucial for improving the measurement accuracy of seawater polarization characteristics by lidar and is an essential step towards promoting high-precision quantitative detection applications of polarization lidar.

[0006] Therefore, it is urgent to develop a method to correct the influence of multiple scattering on the depolarization ratio of lidar signals, so as to achieve accurate measurement of the backscattering polarization characteristics of seawater, improve the detection accuracy of lidar on the vertical profile of the polarization characteristics of the upper ocean, and enhance the understanding of the polarization characteristics of the global ocean. Summary of the Invention

[0007] To address the problem that the depolarization ratio of existing lidar signals is significantly affected by multiple scattering effects, this invention provides a machine learning-based multiple scattering correction method for marine polarization lidar, which can greatly improve the measurement accuracy of lidar for the backscattering polarization characteristics of seawater.

[0008] A machine learning-based method for multiple scattering correction of marine polarization lidar includes: (1) Construct a numerical model of radiative transfer for marine polarization lidar; (2) Set multiple sets of different lidar system parameters, water body inherent optical property parameters, and seawater Mueller matrix parameters; use the radiative transfer numerical model to perform simulation and obtain lidar echo signals under different parameters; (3) After preprocessing the obtained echo signal data, calculate the signal depolarization ratio; take the laser radar system parameters, the inherent optical properties of the water body and the signal depolarization ratio of the corresponding water depth as input, and take the seawater backscattering depolarization ratio as output to construct a neural network with a single hidden layer; and train the corresponding polarization multiple scattering correction model for three different marine laser radar detection systems: spaceborne, airborne and shipborne. (4) A non-uniform water body model was set up to verify the polarization multiple scattering correction model; (5) Calculate the signal depolarization ratio and effective attenuation coefficient based on the measured echo signals of lidar from different detection platforms; combine the multiple scattering correction model of the effective attenuation coefficient to obtain the water absorption coefficient and backscattering coefficient from the measured echo signals of lidar. (6) Based on the absorption coefficient and backscattering coefficient obtained by inversion, combined with the lidar system parameters, and substituted into the trained polarization multiple scattering correction model, the seawater backscattering depolarization ratio of the lidar measured echo signal is calculated.

[0009] In step (1), a numerical model of radiation transfer for marine polarization lidar is constructed based on the polarization Monte Carlo simulation method. The numerical model of radiation transfer is used to simulate and obtain the echo signals of the parallel polarization channel and the vertical polarization channel.

[0010] In step (2), the lidar system parameters include the lidar's height H above the sea surface and the telescope's field of view (FOV); the inherent optical properties of the water body include the water absorption coefficient. Water backscattering coefficient .

[0011] In the simulation, the lidar altitude and field of view parameters for three different marine lidar detection systems—spaceborne, airborne, and shipborne—are set as shown in Table 1.

[0012] Table 1

[0013] The method for obtaining the inherent optical properties parameters of water bodies is as follows: The absorption coefficient was calculated using a type of aquatic bio-optical model based on the chlorophyll concentration in seawater. scattering coefficient Backscattering coefficient Represented as: ; in, This refers to the chlorophyll concentration in seawater. The backscattering ratio, The scattering coefficient of water molecules is denoted as . The scattering coefficient of phytoplankton in Class I water bodies. The wavelength of the lidar.

[0014] In step (2), the expression for the parameters of the seawater Mueller matrix is ​​as follows: ; in, For the Muller matrix of seawater, The scattering angle is... Given the scattering phase function, the depolarization ratio of seawater backscattering is obtained based on the parameters of the seawater Mueller matrix. : ; in, , , This represents the non-zero elements in the normalized parameterized seawater scattering phase matrix. express exist Values The value corresponding to the time.

[0015] element , , The expression is as follows: ; ; ; in, p (90°) represents the degree of polarization when the scattering angle is 90°, and its value is between 0 and 1. It represents The scattering angle corresponding to the minimum value and =90° deviation; exponential term Used to describe and Due to the influence of multiple scattering effects of large particles in the ocean at small angles, the parameters... and They are reciprocals of each other.

[0016] In step (3), the obtained echo signal data is preprocessed and the signal de-bias ratio is calculated, specifically as follows: Calculate the effective detection depth of lidar based on the optical properties of water. We selected 0.3m as the depth interval for the simulated echo signal, and obtained the parallel polarization channel echo signal and vertical polarization channel echo signal from the lidar simulation. We first removed obvious abnormal signal points caused by random simulation errors, and then smoothed the signal to remove abnormal points that deviated from the smoothed signal. Solve for the signal debias ratio using the processed signal. As shown in the following formula: ; in, Represents depth The echo signal from the parallel polarization channel, Represents depth z The echo signal in the vertical polarization channel, For depth The 180° backscattering coefficient of the parallel polarization channel. For depth z The 180° backscattering coefficient of the vertical polarization channel.

[0017] In step (4), the non-uniform water body model adopts a Gaussian distribution of chlorophyll vertical concentration.

[0018] In step (5), the multiple scattering correction model based on the effective attenuation coefficient is expressed as follows: ; in, For effective attenuation coefficient, The depth of the water body , , For parameter items, The water absorption coefficient, Water backscattering coefficient.

[0019] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention utilizes a numerical model of radiation transmission of marine polarization lidar and sets hardware and environmental parameters to obtain a large number of lidar simulation signals. Based on machine learning methods, it constructs polarization multiple scattering correction models for different detection systems to eliminate the influence of multiple scattering on the signal depolarization ratio, thereby improving the measurement accuracy of marine lidar on the polarization characteristics of seawater.

[0020] 2. The multiple scattering polarization correction method proposed in this invention can be applied to marine lidar systems with various detection mechanisms. It can select a suitable model based on the actual measurement situation to obtain a high-precision seawater polarization characteristic profile. Attached Figure Description

[0021] 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.

[0022] Figure 1 This is a schematic diagram of the construction process of the polarization multiple scattering correction model in this invention.

[0023] Figure 2 This is a flowchart illustrating a machine learning-based multiple scattering correction method for marine polarization lidar proposed in this invention.

[0024] Figure 3 This is a comparison chart of the actual seawater depolarization ratio, the predicted seawater depolarization ratio, and the simulated signal depolarization ratio in a uniform water body according to an embodiment of the present invention.

[0025] Figure 4 This is a comparison chart of the actual seawater deflection ratio, the predicted seawater deflection ratio, and the simulated signal deflection ratio in a non-uniform water body according to an embodiment of the present invention. Detailed Implementation

[0026] 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.

[0027] It should be noted that, unless otherwise specified, the features in the following embodiments and implementation methods can be combined with each other.

[0028] The specific construction process of the polarization multiple scattering correction model proposed in this invention is as follows: Figure 1 As shown, multiple sets of radar parameters, inherent optical property parameters of water bodies, and Mueller matrix parameters of seawater are set to simulate different water environments and lidar hardware conditions. These are then substituted into a numerical model of the radiative transmission of a marine polarization lidar to obtain simulated echo signals. A polarization multiple scattering correction model for different detection systems is constructed using machine learning methods, and the effectiveness of the model is verified using a non-uniform water body. This improves the measurement accuracy of the polarization characteristics of seawater by the marine lidar.

[0029] like Figure 2As shown, a machine learning-based multiple scattering correction method for marine polarization lidar first establishes a polarization multiple scattering correction model based on the marine polarization lidar radiative transfer numerical model. Then, based on the measured lidar signals from different platforms and multiple scattering models with effective attenuation coefficients, the signal depolarization ratio and water parameters are calculated. These are then combined with lidar system parameters and input into the polarization multiple scattering correction model to obtain a more accurate seawater backscattering depolarization ratio. This removes the influence of multiple scattering on the signal depolarization ratio, significantly improving the measurement accuracy of the lidar for the polarization characteristics of seawater backscattering.

[0030] The specific steps of this embodiment of the invention are as follows: The first step is to construct a numerical model of the radiation transmission of a marine polarized lidar based on the polarization Monte Carlo simulation method. The echo signals from the vertical and parallel receiving channels of the lidar are shown in the following equation: ; in, Represents depth The echo signal from the parallel polarization channel of the lidar at that location. Represents depth The vertical polarization channel signal at that location, To emit laser pulse energy, For the lidar system constant, The speed of light in a vacuum. The laser pulse width, Atmospheric transmittance, For sea surface transmittance, For the telescope's receiving area, The refractive index of seawater, The distance at which the laser radar reaches the sea surface. This represents the 180° backscattering coefficient of the parallel channel. This is the 180° backscattering coefficient of the vertical channel. This represents the effective attenuation coefficient of the lidar.

[0031] The second step involves setting multiple sets of different lidar system parameters, using a shipborne lidar as an example. The lidar system parameters that have the greatest impact on signal depolarization due to multiple scattering include: the lidar's height above the sea surface. The telescope receives the field of view (FOV). The lidar used in the simulation has a height range of 3:1:30m and a field of view of 50:20:300mrad.

[0032] Set the seawater chlorophyll concentration [Chl] between 0.01 and 3.00 mg / m³. 3Within the range, intervals are 0.01-0.10 (0.01 intervals), 0.10-1.00 (0.10 intervals), and 0.50-3.00 (0.50 intervals). The absorption coefficient was calculated using a type of aquatic bio-optical model. scattering coefficient The formula is as follows: ; ; The scattering characteristics of seawater particles are represented using the Fournier-Forand phase function: ; in, , Backscattering ratio Based on a type of aquatic bio-optical model It can be represented as: ; in, ; ; ; .

[0033] wavelength Set to 532nm, backscattering coefficient It can be represented as: ; The polarization scattering characteristics of seawater can be represented using the normalized Mueller matrix: ; in, Equal to the scattering phase function The expressions for the other elements are as follows: ; ; ; in, p (90°) represents the degree of polarization when the scattering angle is 90°, and its value is between 0 and 1. It represents The scattering angle corresponding to the minimum value and =90° deviation. (Exponential term) Used to describe and Due to the influence of multiple scattering effects of large particles in the ocean at small angles, the parameters... and They are reciprocals of each other.

[0034] The depolarization ratio of seawater backscattering can be obtained from the Mueller matrix expression of seawater. As shown in the following formula: ; Analysis of the effect of seawater Mueller matrix parameters on seawater backscattering depolarization ratio The effect of determining the depolarization ratio of seawater backscattering. The range of input parameters that have a significant impact. The depolarization ratio of seawater backscattering is mainly affected by... p The (90°) angle has a smaller impact than the other parameters, so the remaining parameters are fixed to the fitted result, and only the (90°) angle is set. p The (90°) variation is used to simulate the change in the depolarization ratio of seawater backscattering. p The (90°) parameter is selected from any value in the range of 0.60-0.90 with an interval of 0.01. Simulations were performed using a numerical model of marine polarized lidar radiation transmission to obtain lidar echo signals under different parameters.

[0035] The third step, based on the obtained echo signal data, is to first perform data preprocessing to remove outliers with large errors and then calculate the signal depolarization ratio. To avoid inversion errors caused by the signal-to-noise ratio of lidar at greater water depths, four optical thicknesses were selected as the optimal fitting depths for multiple scattering polarization correction. Based on the optical properties of the water, the effective detection depth of the lidar was calculated. As shown in the following formula: ; ; We selected a depth of 0.3m as the interval for the simulated echo signals to obtain the simulated LiDAR echo signals. We first removed obvious abnormal signal points caused by random simulation errors, and then used the smoothdata function to smooth the signals, removing abnormal points that deviated far from the smoothed signals, thus ensuring the accuracy of the model input data.

[0036] The signal depolarization ratio is calculated using the echo signals from both vertical and parallel LiDAR channels. As shown in the following formula: ; Based on water absorption coefficient Water backscattering coefficient LiDAR altitude H LiDAR field of view (FOV), water depth and signal debias ratio As input, the depolarization ratio of seawater backscattering As output, a neural network with a single hidden layer (containing 12 hidden nodes) is constructed. Polarization multiple scattering correction models are trained for three different marine lidar detection systems: spaceborne, airborne, and shipborne. The constructed seawater backscattering depolarization ratio model is as follows: ; in, This represents the depolarization ratio of seawater backscattering. Represents the water absorption coefficient. Represents the backscattering coefficient of water. The FOV represents the field of view of the lidar, representing the lidar's altitude. Represents water depth. This represents the signal debiasing ratio.

[0037] The accuracy of the polarization multiple scattering correction model was calculated to evaluate its effectiveness. The average absolute percentage error between the predicted seawater depolarization ratio and the actual seawater depolarization ratio was 7.65%, while the average absolute percentage error between the simulated signal depolarization ratio and the actual seawater depolarization ratio was 138.87%. It can be seen that the polarization multiple scattering correction model significantly improves the accuracy of the seawater depolarization ratio, which proves that the polarization multiple scattering correction model has good reliability.

[0038] like Figure 3 As shown, this example demonstrates a comparison between the actual seawater de-biasing ratio, the predicted seawater de-biasing ratio, and the simulated signal de-biasing ratio. In this example, p(90°) is 0.65. It is 6m. FOV The concentration is 0.2 rad, and [Chl] is 0.01 mg / m³ according to (a) to (i). 3 0.05 mg / m 3 0.1 mg / m 3 0.3 mg / m 3 0.6 mg / m 3 0.8 mg / m 3 1.0 mg / m 3 2.0 mg / m 3 3.0 mg / m 3 The straight line in the figure represents the simulated seawater deflection ratio. The dashed line represents the debiasing ratio of the simulated signal. The dotted line represents the predicted sea level deviation. It can be clearly seen that the predicted seawater depolarization ratio is closer to the simulated seawater depolarization ratio than the simulated signal depolarization ratio. Considering factors such as signal-to-noise ratio, water surface reflection, and signal jitter, there are certain differences between the predicted and simulated seawater depolarization ratios. The reason for classifying the seawater depolarization ratio by different seawater chlorophyll concentrations [Chl] in the figure is that seawater chlorophyll concentration [Chl] is a key factor affecting the signal depolarization ratio error. If the seawater chlorophyll concentration [Chl] is higher, the multiple scattering effect will be relatively stronger, leading to a larger signal depolarization ratio error. Therefore, this classification can cover most water body scenarios detected by lidar as much as possible.

[0039] The fourth step is to set up a non-uniform water body to verify the polarization multiple scattering correction model, compare the error between the simulated seawater backscattering depolarization ratio and the model's predicted depolarization ratio, calculate the correlation between the two, and ensure the effectiveness of the seawater backscattering depolarization ratio prediction model.

[0040] The non-uniform water body model adopts a Gaussian distribution of chlorophyll vertical concentration: ; in, Background chlorophyll concentration, from surface concentration [Chl] surf Start with slope It decays linearly with depth. At the highest concentration, This is the location with the highest concentration. The width of the Gaussian peak. It is a dimensionless depth parameter, defined as geometric depth. With the depth of the true light layer The ratio, that is: ; The dimensionless chlorophyll concentration is defined as chlorophyll concentration. Average concentration of the true light layer The ratio, that is: ; Among them, the average chlorophyll concentration in the euphotic layer Defined as: ; Different surface chlorophyll concentrations [Chl] surf It is divided into 9 categories, from S1 to S9. S1 represents [Chl]. surf Less than 0.04 mg / m 3 The water body, S2 represents [Chl]. surf Between 0.04 and 0.08 mg / m³ 3 The water body within the specified area, S3 represents [Chl].surf Between 0.08 and 0.12 mg / m 3 The water body within the specified area, S4 represents [Chl]. surf In the range of 0.12-0.2 mg / m 3 The water bodies within the specified area, S5 stands for [Chl]. surf At 0.2-0.3 mg / m 3 The water body within the specified area, S6 represents [Chl]. surf At 0.3-0.4 mg / m 3 The water bodies within the specified area, S7 represents [Chl]. surf In 0.4-0.8 mg / m 3 The water body within the specified area, S8 represents [Chl]. surf Between 0.8 and 2.2 mg / m 3 The water bodies within the specified area, S9 represents [Chl]. surf Between 2.2 and 4.0 mg / m 3 The water bodies within the range correspond to different trophic levels, from low to high. Based on the chlorophyll concentration distribution in non-uniform water bodies, lidar echo signal simulations were performed to verify and optimize the constructed polarization multiple scattering correction model, ensuring the reliability of the seawater backscattering depolarization ratio prediction model. For cases where the polarization multiple scattering correction model exhibits low accuracy in non-uniform water bodies, it is necessary to analyze the sensitivity of the model's accuracy to changes in various parameters. Parameters that significantly impact model accuracy and lead to lower accuracy should be assigned higher weights, or the model should be segmented, with adapted models constructed for different parameter ranges under the same trophic level. This approach aims to improve model accuracy until the seawater backscattering depolarization ratio prediction model achieves high reliability.

[0041] The fifth step involves calculating the signal depolarization ratio and effective attenuation coefficient based on the measured signals from different lidar platforms. This is combined with a multiple scattering correction model using the effective attenuation coefficient, and further considered based on the inherent optical properties of water and the diffuse attenuation coefficient. Based on the empirical relationship between them, the water absorption coefficient was calculated. and backscattering coefficient .

[0042] Effective attenuation coefficient of lidar The multiple scattering correction model is shown in the following equation: ; in, z The depth of the water body is m1, m2, and m3, which are parameters related to the backscattering coefficient. b b The relationship between them is shown in the following formula: ; ; ; Step 6: Based on the absorption coefficient obtained from the inversion and backscattering coefficient By combining the parameters of the lidar system and inputting the trained polarization multiple scattering correction model, the backscattering depolarization ratio of the seawater in the lidar measured signal is calculated.

[0043] like Figure 4 As shown, this example illustrates the relationship between the depolarization ratio of real seawater, the predicted depolarization ratio of seawater, and the depolarization ratio of the simulated signal in a non-uniform water body. p (90°) is 0.65. The water depth is 15m, the field of view (FOV) is 0.1 rad, and the water types from (a) to (i) are S1 to S9, respectively. The straight lines in the figure represent the actual seawater deflection ratio. The dashed line represents the debiasing ratio of the simulated signal. The dotted line represents the predicted sea level deviation. It can be clearly seen that the predicted seawater depolarization ratio is closer to the actual seawater depolarization ratio than the simulated signal depolarization ratio. However, considering factors such as signal-to-noise ratio, water surface reflection, and signal jitter, there are still some differences between the predicted and simulated seawater depolarization ratios. In the validation of non-uniform water bodies, the detection depth was uniformly set to 50m.

[0044] In non-uniform water bodies, the average absolute percentage error between the predicted seawater depolarization ratio and the actual seawater depolarization ratio is 9.5%, while the average absolute percentage error between the simulated signal depolarization ratio and the actual seawater depolarization ratio is 123.1%. Compared with the average absolute percentage error accuracy in uniform water bodies, the two are very close. Furthermore, non-uniform water bodies are not used for model training but only for verifying the model's accuracy. This demonstrates the effectiveness of the seawater backscattering depolarization ratio prediction model and the polarization multiple scattering correction algorithm.

[0045] This invention utilizes machine learning methods to construct corresponding polarization multiple scattering correction models for different marine lidar detection systems. Using the input parameters required by these models, the backscattering depolarization ratio of seawater can be obtained. The predicted results show good consistency with the actual values, demonstrating the reliability of this invention.

[0046] 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 machine learning-based method for multiple scattering correction of marine polarization lidar, characterized in that, include: (1) Construct a numerical model of radiative transfer for marine polarization lidar; (2) Set multiple sets of different lidar system parameters, water body inherent optical property parameters, and seawater Mueller matrix parameters; use the radiative transfer numerical model to perform simulation and obtain lidar echo signals under different parameters; (3) After preprocessing the obtained echo signal data, calculate the signal depolarization ratio; take the laser radar system parameters, the inherent optical properties of the water body and the signal depolarization ratio of the corresponding water depth as input, and take the seawater backscattering depolarization ratio as output to construct a neural network with a single hidden layer; and train the corresponding polarization multiple scattering correction model for three different marine laser radar detection systems: spaceborne, airborne and shipborne. (4) A non-uniform water body model was set up to verify the polarization multiple scattering correction model; (5) Calculate the signal depolarization ratio and effective attenuation coefficient based on the measured echo signals of lidar from different detection platforms; combine the multiple scattering correction model of the effective attenuation coefficient to obtain the water absorption coefficient and backscattering coefficient from the measured echo signals of lidar. (6) Based on the absorption coefficient and backscattering coefficient obtained by inversion, combined with the lidar system parameters, and substituted into the trained polarization multiple scattering correction model, the seawater backscattering depolarization ratio of the lidar measured echo signal is calculated.

2. The machine learning-based multiple scattering correction method for marine polarization lidar according to claim 1, characterized in that, In step (1), a numerical model of radiation transfer for marine polarization lidar is constructed based on the polarization Monte Carlo simulation method. The numerical model of radiation transfer is used to simulate and obtain the echo signals of the parallel polarization channel and the vertical polarization channel.

3. The machine learning-based multiple scattering correction method for marine polarization lidar according to claim 1, characterized in that, In step (2), the lidar system parameters include the lidar's height H above the sea surface and the telescope's field of view (FOV); the inherent optical properties of the water body include the water absorption coefficient. Water backscattering coefficient .

4. The machine learning-based multiple scattering correction method for marine polarization lidar according to claim 3, characterized in that, The method for obtaining the inherent optical properties parameters of water bodies is as follows: The absorption coefficient was calculated using a type of aquatic bio-optical model based on the chlorophyll concentration in seawater. scattering coefficient Backscattering coefficient Represented as: ； in, This refers to the chlorophyll concentration in seawater. The backscattering ratio, The scattering coefficient of water molecules is denoted as . The scattering coefficient of phytoplankton in Class I water bodies. The wavelength of the lidar.

5. The machine learning-based multiple scattering correction method for marine polarization lidar according to claim 1, characterized in that, In step (2), the expression for the parameters of the seawater Mueller matrix is ​​as follows: ； in, For the Muller matrix of seawater, The scattering angle is... Given the scattering phase function, the depolarization ratio of seawater backscattering is obtained based on the parameters of the seawater Mueller matrix. : ； in, , , This represents the non-zero elements in the normalized parameterized seawater scattering phase matrix. express exist Values The value corresponding to the time.

6. The machine learning-based multiple scattering correction method for marine polarization lidar according to claim 5, characterized in that, element , , The expression is as follows: ； ； ； in, p (90°) represents the degree of polarization when the scattering angle is 90°, and its value is between 0 and 1. It represents The scattering angle corresponding to the minimum value and =90° deviation; exponential term Used to describe and Due to the influence of multiple scattering effects of large particles in the ocean at small angles, the parameters... and They are reciprocals of each other.

7. The machine learning-based multiple scattering correction method for marine polarization lidar according to claim 1, characterized in that, In step (3), the obtained echo signal data is preprocessed and the signal de-bias ratio is calculated, specifically as follows: Calculate the effective detection depth of lidar based on the optical properties of water. We selected 0.3m as the depth interval for the simulated echo signal, and obtained the parallel polarization channel echo signal and vertical polarization channel echo signal from the lidar simulation. We first removed obvious abnormal signal points caused by random simulation errors, and then smoothed the signal to remove abnormal points that deviated from the smoothed signal. Solve for the signal debias ratio using the processed signal. As shown in the following formula: ； in, Represents depth The echo signal from the parallel polarization channel, Represents depth z The echo signal in the vertical polarization channel, For depth The 180° backscattering coefficient of the parallel polarization channel. For depth z The 180° backscattering coefficient of the vertical polarization channel.

8. The machine learning-based multiple scattering correction method for marine polarization lidar according to claim 1, characterized in that, In step (4), the non-uniform water body model adopts a Gaussian distribution of chlorophyll vertical concentration.

9. The machine learning-based multiple scattering correction method for marine polarization lidar according to claim 1, characterized in that, In step (5), the multiple scattering correction model based on the effective attenuation coefficient is expressed as follows: ； in, For effective attenuation coefficient, The depth of the water body , , For parameter items, The water absorption coefficient, Water backscattering coefficient.

Images