A method for optimizing spaceborne ocean lidar data based on signal-to-noise ratio constraints

Abstract

The present invention discloses a method for optimizing satellite-borne ocean lidar data based on signal-to-noise ratio constraints, comprising the following steps: finding the sea surface position based on satellite-borne lidar data; calculating the signal-to-noise ratio (SNRss) of the vertical channel attenuated backscatter profile, performing a signal-to-noise ratio constrained calculation averaging step, and calculating the number of profiles to be averaged using a theoretical formula; finding the sea surface position of each profile within 100 km of the current profile and aligning them; finding profiles whose total attenuated backscatter coefficients differ from the current profile by less than 50% and averaging them; and finally performing ocean optical parameter inversion calculations. The present invention can improve the signal-to-noise ratio while maintaining the homogeneity of the average profile, thereby reducing the error in the inverted data. This method has reference significance for the development and data processing of future satellite-borne ocean polarization lidars and has broad applicability.

Description

Technical Field

[0001] The present invention belongs to the technical field of laser radar, and in particular relates to a method for optimizing spaceborne ocean laser radar data based on signal-to-noise ratio constraints. Background Art

[0002] Marine ecosystems are complex entities encompassing a vast array of species operating across a wide range of time and space. Phytoplankton is the foundation of most marine ecosystems, with annual net photosynthetic carbon sequestration roughly equivalent to that of all terrestrial plants. Research on the optical properties of the ocean is crucial for understanding phytoplankton distribution, the ocean carbon cycle, and global climate change. As the only spaceborne lidar in orbit for over a decade, CALIOP's active detection technology complements the inability of traditional passive ocean color satellites to detect water at night, providing valuable information for studying global marine phytoplankton distribution, climate, and environmental change.

[0003] The research and application of using CALIOP data to invert the optical properties of seawater began in 2013. The Oregon State University research team first proposed the inversion of the backscattering coefficient b of marine particles based on the signal of the orthogonal polarization channel of CALIOP 532nm. bp feasible method, and through b bp Quantitative calculation of particulate organic carbon and phytoplankton biomass in the global ocean proves that spaceborne lidar can promote the understanding of the global plankton system. In 2014, NASA improved the above algorithm and removed the transient response of the lidar. In 2021, it proposed that there was polarization crosstalk between the dual polarization channels of the 532nm satellite-borne lidar, which greatly affected the accuracy of the data, and gave a correction scheme. The Chinese patent document with publication number CN115792946A discloses an inversion method for detecting subsurface water elements by satellite-borne lidar, which uses a neural network method to independently invert the optical properties of seawater. It matches the data corresponding to the satellite-borne lidar and the existing water body parameters according to time and longitude and latitude, and then uses the total depolarization ratio, water body backscatter coefficient and latitude as variables, and the water body parameters as true values ​​to input into the deep neural network for learning, to obtain a trained deep learning network to predict water body parameters.

[0004] As a spaceborne lidar, CALIOP's signal-to-noise ratio is necessarily much lower than that of typical ground-based or airborne lidars, resulting in low accuracy in the inversion results. However, existing research using CALIOP to invert the optical properties of seawater is based on the original algorithm proposed in 2013, focusing on the overall picture while ignoring the inversion of details. At the same time, there has been no detailed description or comparative analysis of the low signal-to-noise ratio problem in ocean data applications, nor has a solution been proposed. As a result, current ocean optical results cannot balance spatial resolution and accuracy. In downstream data processing, the simplest and most efficient solution to improve the signal-to-noise ratio is signal averaging. However, a high number of averaging times will lead to an averaged profile in areas such as uneven coasts or where phytoplankton changes dramatically. The final result cannot represent the optical characteristics of the ocean scene, and too low an averaging number cannot effectively eliminate errors.

[0005] Therefore, there is an urgent need for a scale-varying average spaceborne ocean lidar optimization algorithm that can adapt to the differences in signal performance in different regions. Summary of the Invention

[0006] The present invention provides a method for optimizing spaceborne ocean lidar data based on signal-to-noise ratio constraints. By using the SNR (or signal strength) to constrain the number of signal averaging times, the signal-to-noise ratio can be improved while maintaining the homogeneity of the average profile, thereby reducing the inversion data error.

[0007] A method for optimizing spaceborne ocean lidar data based on signal-to-noise ratio constraints, comprising:

[0008] (1) Import the attenuated backscatter profile data of the spaceborne lidar, filter the profile data, and find the sea surface position;

[0009] (2) Integrate the signal strength of the vertical polarization channel sea surface position of the current profile, read the noise signal value at the same time, and then calculate the signal-to-noise ratio (SNR) of the vertical channel ss ;

[0010] (3) Using signal-to-noise ratio (SNR) ss The theoretical formula of the relationship with the average number of times is used to calculate the theoretical average profile number Avenum. If it is less than 15, it is set to 15 times;

[0011] (4) Find the peak sea surface position of each profile in the 100 km profile near the current profile and align them; calculate the total attenuation backscatter coefficient of the aligned sea surface and the three height bins below the sea surface;

[0012] (5) Calculate the profiles within 100 km of the current profile whose total attenuation backscatter coefficient is within 50% of the current profile, and count the number of adjacent averaged profiles count_near_total;

[0013] (6) Compare the theoretical average number of profiles and the number of adjacent averageable profiles, and select the average number of profiles to average the parallel and vertical channels;

[0014] (7) Use the averaged dual-channel signal to calculate the depolarization ratio and particle backscattering coefficient b bp Inversion calculation.

[0015] In step (1), when filtering the profile data, the signal peak position is used as the sea surface position, and the extension lines with incorrect peak positions, that is, the profiles outside the range of ±120 m of the ground elevation value, are filtered out.

[0016] The specific process of step (2) is:

[0017] Integrate the signal intensity of the vertical polarization channel at the sea surface position of the current profile and calculate the number of ocean subsurface signal particles s m , the formula is as follows:

[0018]

[0019] Where h is Planck's constant, ν is the photon frequency, r is the distance from the satellite to the sampling point, and surface is the sea surface bin position. Read the noise signal value and calculate the single profile signal-to-noise ratio (SNR). ss :

[0020]

[0021] Among them, N f is the detector noise factor, σ b 2 and σ n 2 are the standard deviations of the statistical fluctuations of the solar background illumination and dark current, respectively, expressed in terms of the number of detected photoelectrons.

[0022] In step (3), if SNR ss If the value is >200, no profile averaging is performed. Otherwise, the theoretical average profile number Avenum is calculated using the theoretical formula of the relationship between the signal-to-noise ratio and the number of averages. The formula is as follows:

[0023]

[0024] Where, SNR ideal is the ideal expansion line number, which is set to 300; if the theoretical average profile number Avenum is lower than 15, it is set to 15 times.

[0025] In step (4), the integrated total attenuated backscatter coefficient of the aligned sea surface and the three height bins below the sea surface is calculated using the following formula:

[0026]

[0027] Where r is the distance from the satellite to the sampling point, surface is the sea surface height bin position, and β' is the attenuated backscatter coefficient.

[0028] In step (5), before counting the number of adjacent averaged profiles, data with erroneous peak positions and data with an integral value of the total attenuated backscatter coefficient less than or equal to 0 are excluded.

[0029] In step (6), if the theoretical average profile number is greater than or equal to the number of adjacent averageable profiles, the adjacent averageable profile number is selected as the actual average profile;

[0030] If the theoretical average number of profiles is less than the number of adjacent averageable profiles, the Avenum profile with the theoretical average number of profiles closest to the central position profile is selected for averaging.

[0031] The specific process of step (7) is:

[0032] First, read the sea surface wind speed data v and calculate the average square wave slope of the ocean surface 2 >, the formula is:

[0033]

[0034] Then, the model is used to invert the sea surface backscatter value:

[0035]

[0036] Where θ is the incident angle of the lidar system, and ρ represents the Fresnel reflectivity;

[0037] Using the ratio δ of the perpendicular and parallel channel signals T Calculate the vertical polarization component β′ of the water column integrated attenuation backscatter w+ :

[0038]

[0039] Final b bp The value is determined by β′ w+ Multiplying by a constant yields:

[0040] b bp =β′ w+ / 0.32 / 0.98 2 .

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] ​1. The present invention optimizes the inversion method of spaceborne ocean lidar data, uses a signal-to-noise ratio constraint model, and performs averaging processing on varying scales to maintain the accuracy of weak signal profiles and the specificity of strong signal profiles, taking into account both the spatial resolution and accuracy of satellite observation data.

[0043] 2. Based on the basic principles of satellite-borne lidar, the present invention can improve the overall signal-to-noise ratio of the satellite's vertical polarization channel profile, better eliminate noise such as solar background noise and detector dark current, thereby reducing inversion data errors and improving the accuracy of data products.

[0044] 3. The present invention has reference significance for the development and data processing of future spaceborne ocean polarization lidar and has wide applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a flow chart of a method for optimizing spaceborne ocean lidar data based on signal-to-noise ratio constraints according to the present invention;

[0046] Figure 2 The selection of the average number of times and the trajectory position map in the embodiment of the present invention;

[0047] Figure 3 The trajectory comparison using the signal-to-noise ratio constraint method and different averaging times method in the present invention is performed, and the correlation and root mean square error are calculated by matching with MODIS data;

[0048] Figure 4 It is the probability density diagram of the signal-to-noise ratio before and after averaging and the distribution diagram of the signal-to-noise ratio before and after averaging. DETAILED DESCRIPTION

[0049] The present invention will be described in further detail below with reference to the accompanying drawings and examples. It should be noted that the following examples are intended to facilitate understanding of the present invention and do not have any limiting effect on the present invention.

[0050] like Figure 1 As shown in the figure, a method for optimizing spaceborne ocean lidar data based on signal-to-noise ratio constraint is proposed. Based on the spaceborne atmospheric lidar CALIOP, the vertical channel single profile SNR is calculated. ss , a method used to constrain the number of signal averaging times can improve the signal-to-noise ratio while maintaining the homogeneity of the average profile, thereby reducing the inversion data error.

[0051] The present invention uses a track of CALIOP near the China Sea on February 24, 2010 as an example. Figure 2 The right figure includes the following steps:

[0052] Step (1) import and quality screening of CALIOP L1B satellite-borne ocean polarization lidar data, and process them profile by profile. For the current profile, basic quality screening is first performed, using the signal peak position as the sea surface position, and filtering out profiles with peak position errors, that is, profiles outside the range of ±120m of the ground elevation value;

[0053] Step (2) integrates the signal strength of the vertical polarization channel sea surface position of the current profile, and reads the noise signal value at the same time, and then calculates the signal-to-noise ratio (SNR) of the vertical channel. ss :

[0054] (2-1) The CALIOP signal-to-noise ratio calculation formula is as follows:

[0055]

[0056] (2-2) Where σ m By the root mean square error σ sm , solar background noise σ b , dark current noise σ n constitute:

[0057]

[0058] Among them, N f is the detector noise factor. The molecular signal s m It can be written as the number of photoelectrons detected in a height bin range n m The magnitude of the RMS noise σ sm is n m The square root of . Similarly, the other noise terms, σ b 2 and σ n 2 (the standard deviation of the statistical fluctuations of the solar background illumination and dark current), which can be expressed in terms of the number of detected photoelectrons.

[0059] Therefore, the signal-to-noise ratio calculation formula for a single profile is:

[0060]

[0061] (2-3) In the above formula, the noise signal can be directly read from the CALIOP L1B data file, so only s m Calculation: Number of photoelectrons detected within the height bin range n m =bin energy value / photon energy, then the ocean subsurface s m It can be considered that the ratio of the integral value of the signal power at the sea surface to the photon energy is:

[0062]

[0063] Where h is Planck's constant, ν is the photon frequency, r is the distance from the satellite to the sampling point, and surface is the ocean surface bin position. Because the CALIOP calibration coefficient directly provides the number of photoelectrons, division by photon energy and distance is not necessary in this example. P(r) is the signal power at the ocean surface, which can be calculated from the satellite raw data file.

[0064] Step (3), if SNR ss If the value is >200, no profile averaging is performed. Otherwise, the average number of times is calculated based on the signal-to-noise ratio constraint. The theoretical average profile number Avenum is calculated using the theoretical formula of the relationship between the signal-to-noise ratio and the average number of times. The ideal SNR is taken as ideal For 300:

[0065]

[0066] If Avenum is less than 15, it is forced to 15 times.

[0067] Step (4): 100 km near the current profile, since the horizontal resolution of CALIOP is 333 m, it is equivalent to taking 300 adjacent profiles, finding the peak sea surface position of each average profile, and aligning them, and calculating the total attenuation backscatter coefficient of the aligned sea surface and the three bins below the sea surface.

[0068]

[0069] where β' is the attenuated backscatter coefficient.

[0070] Step (5) searches for profiles within 100 km of the current profile whose total attenuation backscatter coefficient integral value is within 50% of the current profile, and excludes data whose peak position is outside the range of 120 m above and below the ground elevation data, and data whose total attenuation backscatter coefficient integral value is less than or equal to 0. Finally, count the number of adjacent averageable profiles (count_near_total).

[0071] Step (6), compare the theoretical average profile number Avenum and the adjacent average profile number count_near_total,

[0072] A. If the theoretical average profile number is greater than or equal to the number of adjacent averageable profiles, the averageable profile number is selected as the actual average profile, because if the distance is too large, spatial differences may be introduced;

[0073] B. If the theoretical average number of profiles is less than the number of adjacent averageable profiles, the theoretical average number Avenum of profiles closest to the center position profile will be selected for averaging.

[0074] The average number of final selections is Figure 2 As shown in the left figure, the signal-to-noise ratio is greatly improved before and after averaging. Figure 4 . Figure 4 In the middle, the left figure is the probability density diagram of the signal-to-noise ratio before and after averaging, with the horizontal axis being the SNR before averaging and the vertical axis being the SNR after averaging; the right figure is the distribution diagram of the signal-to-noise ratio before and after averaging, proving that averaging can bring most signal-to-noise ratios within the ideal range.

[0075] Step (7) calculates the depolarization ratio using the ratio of the averaged vertical and parallel channel sea area integrals, Figure 3 The comparison of CALIOP trajectory inversion using the signal-to-noise ratio constraint method and different averaging times methods is shown, and the correlation and root mean square error are calculated by matching with MODIS data. The correlation and root mean square error of the corresponding profile of the proposed method and MODIS are both better. Finally, the sea surface wind speed calibration method is used to perform b bp Inversion calculation. First read the sea surface wind speed data v and calculate the average square wave slope of the ocean surface 2 >,

[0076]

[0077] The model is then used to invert the sea surface backscatter value:

[0078]

[0079] Where θ is the incident angle of the CALIOP lidar system, which is 0.3° and was adjusted to 3° after November 28, 2007. ρ is the Fresnel reflectivity, which is approximately 0.0209 in the 532 nm channel.

[0080] to b bp The estimation is mainly based on the 532nm vertical polarization channel signal in the CALIOP data, because the signal of this channel is almost entirely from the backscattering of water particles, while the influence of sea surface Fresnel reflection and water molecule scattering can be ignored. In order to avoid the error caused by calibration of atmospheric transmittance, the ratio of the vertical and parallel channel signals δ T Since the contribution of sea surface reflection in the vertical channel can be ignored, the vertical polarization component of the water column integrated attenuated backscatter can be written as:

[0081]

[0082] δ w The parameters include all scatterers in the water, which are difficult to invert directly from CALIOP data due to the current coarse vertical resolution. Behrenfeld, based on the experimental data of Voss, Fry and Kokhanovsky, assumes that δ w ​=0.1, due to δ T The value is relatively small (~0.01), so δ w The uncertainty of the derived b bp Therefore, the above formula can be rewritten as

[0083]

[0084] The final b bp The value can be determined by β′ w+ Multiplying by a constant yields:

[0085] b bp =β′ w+ / 0.32 / 0.98 2

[0086] The embodiments described above provide a detailed description of the technical solutions and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, supplements and equivalent substitutions made within the scope of the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for optimizing spaceborne ocean lidar data based on signal-to-noise ratio constraints, characterized in that: include: (1) Import the attenuated backscatter profile data of the spaceborne lidar, filter the profile data, and find the sea surface position; (2) Integrate the signal strength of the vertical polarization channel sea surface position of the current profile, read the noise signal value at the same time, and then calculate the signal-to-noise ratio (SNR) of the vertical channel ss ; (3) Using signal-to-noise ratio (SNR) ss The theoretical formula of the relationship with the average number of times is used to calculate the theoretical average profile number Avenum. If it is less than 15, it is set to 15 times; (4) Find the peak sea surface position of each profile in the 100 km profile near the current profile and align them; Calculate the total attenuated backscatter coefficient of the aligned sea surface and the three height bins below the sea surface; (5) Calculate the profiles within 100 km of the current profile, find the profiles whose total attenuation backscatter coefficient difference with the current profile is within 50%, and count the number of adjacent averaged profiles count_near_total; (6) Compare the theoretical average number of profiles and the number of adjacent averageable profiles, and select the average number of profiles to average the parallel and vertical channels; (7) Use the averaged dual-channel signal to calculate the depolarization ratio and particle backscattering coefficient b bp Inversion calculation.

2. The method for optimizing spaceborne ocean lidar data based on signal-to-noise ratio constraints according to claim 1, characterized in that: In step (1), when filtering the profile data, the signal peak position is used as the sea surface position, and the extension lines with incorrect peak positions, that is, the profiles outside the range of ±120 m of the ground elevation value, are filtered out.

3. The method for optimizing spaceborne ocean lidar data based on signal-to-noise ratio constraints according to claim 1, characterized in that: The specific process of step (2) is: Integrate the signal intensity of the vertical polarization channel at the sea surface position of the current profile and calculate the number of ocean subsurface signal particles s m , the formula is as follows: Where h is Planck's constant, ν is the photon frequency, r is the distance from the satellite to the sampling point, and surface is the sea surface height bin position; read the noise signal value and calculate the single profile signal-to-noise ratio SNR ss : Among them, N f is the detector noise factor, σ b 2 and σ n 2 are the standard deviations of the statistical fluctuations of the solar background illumination and dark current, respectively, expressed in terms of the number of detected photoelectrons.

4. The method for optimizing spaceborne ocean lidar data based on signal-to-noise ratio constraints according to claim 1, wherein: In step (3), if SNR ss If the value is >200, no profile averaging is performed. Otherwise, the theoretical average profile number Avenum is calculated using the theoretical formula of the relationship between the signal-to-noise ratio and the number of averages. The formula is as follows: Where, SNR ideal is the ideal expansion line number, which is set to 300; if the theoretical average profile number Avenum is lower than 15, it is set to 15 times.

5. The method for optimizing spaceborne ocean lidar data based on signal-to-noise ratio constraints according to claim 1, wherein: In step (4), the integrated total attenuated backscatter coefficient of the aligned sea surface and the three height bins below the sea surface is calculated using the following formula: Where r is the distance from the satellite to the sampling point, surface is the sea surface height bin position, and β' is the attenuated backscatter coefficient.

6. The method for optimizing spaceborne ocean lidar data based on signal-to-noise ratio constraints according to claim 1, characterized in that: In step (5), before counting the number of adjacent averaged profiles, data with erroneous peak positions and data with an integral value of the total attenuated backscatter coefficient less than or equal to 0 are excluded.

7. The method for optimizing spaceborne ocean lidar data based on signal-to-noise ratio constraints according to claim 1, wherein: In step (6), if the theoretical average profile number is greater than or equal to the number of adjacent averageable profiles, the adjacent averageable profile number is selected as the actual average profile; If the theoretical average number of profiles is less than the number of adjacent averageable profiles, the Avenum profile with the theoretical average number of profiles closest to the central position profile is selected for averaging.

8. The method for optimizing spaceborne ocean lidar data based on signal-to-noise ratio constraints according to claim 1, wherein: The specific process of step (7) is: First, read the sea surface wind speed data v and calculate the average square wave slope of the ocean surface 2 >, the formula is:​ Then, the model is used to invert the sea surface backscatter value: Where θ is the incident angle of the lidar system, and ρ represents the Fresnel reflectivity; Using the ratio δ of the perpendicular and parallel channel signals T Calculate the vertical polarization component β′ of the water column integrated attenuation backscatter w+ : Final b bp The value is determined by β′ w+ Multiplying by a constant yields: b bp ＝β′ w+ / 0.32 / 0.98 2 。

Images