A Forward Iterative Inversion Method for Marine Lidar Based on Instability Suppression

Abstract

The present invention discloses a forward iterative inversion method for marine lidar based on instability suppression, including: S1, preprocessing the collected original lidar data to obtain a discrete signal D with a length of N, and obtaining a boundary condition #imgabs0# corresponding to the depth position of D(0); S2, narrowing the solution range of the lidar ratio S p by detecting physical meaning errors; S3, realizing the accurate solution of S p by optimizing the objective function F(S p ); S4, according to the #imgabs1# value obtained in S3, calculating the K d (z) and β π (z) profiles through the forward Fernald inversion iteration formula and outputting the results. The present invention realizes the accurate solution of the lidar ratio S p , and only needs the boundary condition #imgabs2# of the water surface layer to achieve a robust forward inversion. Compared with the traditional backward inversion algorithm, the boundary conditions required by the present invention are easier to obtain and are also easier to combine with technologies such as water color remote sensing.

Description

Technical Field

[0001] The present invention belongs to the technical field of marine lidar, and in particular relates to a forward iterative inversion method for marine lidar based on instability suppression. Background Art

[0002] Current marine optical detection means include in-situ instruments, passive remote sensing, and active remote sensing. Among them, lidar in the field of active remote sensing has received extensive attention due to its advantages such as being able to achieve day and night detection, being not restricted by latitude, and being able to obtain marine vertical profile information, and has become an important marine detection means. Elastic scattering lidar is relatively simple and has a low cost. It is the most widely used marine lidar at present and will still occupy an important position in marine mapping for a long time to come. However, for elastic scattering lidar, it is difficult to solve the attenuation coefficient k lidar (z) and the 180-degree volume scattering coefficient β π (z) from a single echo signal equation. Therefore, it is necessary to develop a suitable inversion algorithm.

[0003] Existing classical inversion algorithms can be roughly divided into two categories. The first category of algorithms is based on the assumption that k lidar (z) appears in the integral term and its small-scale variation is weakened. By estimating the average value of k lidar (z) within the full depth range, the β π (z) profile is inverted. Such algorithms cannot reflect the profile of k lidar (z) and are usually applicable to Class I waters with relatively weak attenuation coefficient changes. The errors generated by them largely stem from the assumption that the varying part of the attenuation coefficient can be ignored and the signal-to-noise ratio of the lidar. The second category of algorithms assumes that there is a certain known relationship between k lidar (z) and β π (z), and uses the method of numerical integration to solve the lidar equation, such as the Fernald method and the Klett method. Compared with the first category of algorithms, the second category of algorithms can invert k lidar (z) and β π (z) simultaneously and have a wider application range. However, due to the instability of forward iteration, almost all marine lidar inversions using the second category of algorithms currently adopt backward iteration, which means that the algorithm requires a boundary parameter at a deep water location. The problem is that the boundary parameter at a deep water location is often difficult to obtain. We cannot, like ground-based atmospheric lidar, use a computable pure molecular scattering approximation as the boundary parameter for backward inversion.

[0004] Therefore, it is urgent to propose a forward inversion method that can overcome instability to provide more comprehensive theoretical support for lidar to detect the marine vertical profile. Summary of the Invention

[0005] Aiming at the defects of unstable forward inversion and easy generation of outliers in existing marine lidars, the present invention provides a forward iterative inversion method for marine lidars based on instability suppression, realizing the precise solution of lidar ratio S p and only requiring the boundary conditions of the water surface layer to achieve robust forward inversion.

[0006] Without considering the influence of multiple scattering, the lidar signal generated by the water body at a depth of z is corrected for environmental parameters, system parameters, and distance square, expressed as:

[0007]

[0008] where D represents the lidar signal after correction for system environmental parameters and distance square, and β π represents the 180° volume scattering function, and k lidar represents the lidar attenuation coefficient and represents the field of view angle.

[0009] The following discussion will be based on a commonly used 532 nm wavelength, large field of view shipborne lidar. At this time, k lidar can be approximated as K d . Based on such an approximation and considering the optical properties of water molecules and the remaining seawater components respectively, D is re-expressed as:

[0010]

[0011] where is the 180-degree volume scattering coefficient of seawater molecules (1.94×10 -4 m -1 sr -1 ); K dw is the diffuse attenuation coefficient of seawater molecules (0.0452 m -1 ); and K dp are the 180-degree volume scattering coefficient and diffuse attenuation coefficient of the remaining seawater components respectively.

[0012] Introduce the classical assumptions of the Fernald algorithm:

[0013]

[0014] According to this assumption, the forward Fernald inversion iteration formula can be obtained by numerical integration as:

[0015]

[0016] where Δz is the spatial resolution of the lidar, and S w is the extinction-scattering ratio of pure seawater molecules, Sp is the extinction-scattering ratio of other seawater components,

[0017] A forward iterative inversion method for marine lidar based on instability suppression, comprising the following steps:

[0018] S1. Preprocess the collected original lidar data to obtain a discrete signal D of length N, and obtain the boundary conditions corresponding to the depth position of D(0).

[0019] S2. Narrow the solution range of the lidar ratio S by detecting physical meaning errors, specifically: p as follows:

[0020] (2.a) Define the initial range of the lidar ratio S p as

[0021] (2.b) Let

[0022] (2.c) Using as the initial value, substitute the current value of S p and the signal D into the forward Fernald inversion iterative formula to obtain an inversion profile β π (n);

[0023] (2.d) Determine whether there is π in β If the judgment result is true, execute (2.f); if the judgment result is false, execute (2.e);

[0024] (2.e) Determine whether there is π in β If the judgment result is true, execute (2.g); if the judgment result is false, execute (2.h);

[0025] (2.f) Let Return to (2.b) and run in a loop;

[0026] (2.g) Let Return to (2.b) and run in a loop;

[0027] (2.h) End S2 and execute S3;

[0028] S3. Achieve the exact solution of S by optimizing the objective function F(S p ), specifically: p as follows:

[0029] (3.a) Extract what is obtained in S2 Let

[0030] (3.b) Let Calculate

[0031] (3.c) Judge whether i <= n. If the judgment is true, execute (3.d); if the judgment is false, execute (3.g);

[0032] (3.d) Calculate

[0033] (3.e) Judge V tmp < V 0 ; if the judgment is true, let

[0034] (3.f) Let i = i + 1, and jump to execute (3.c);

[0035] (3.g) Store the current value, end S3, and execute S4;

[0036] S4. According to the value obtained from S3, through the forward Fernald inversion iterative formula, calculate K d (z) and β π (z) profiles and output the results.

[0037] Furthermore, in step S1, the acquisition of boundary conditions is specifically: measure the boundary conditions through in-situ measurement or water color remote sensing

[0038] In step S3, the calculation method of the objective function F(S p ) is: taking as the initial value, substituting the function parameters S p and the signal D into the forward Fernald inversion iterative formula, obtaining the inversion profile β π (n) with a length of N, and returning the value where is the median value of the β π profile.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] The present invention does not require a pre-assumed lidar ratio S p , realizes the automated solution of S p , and overcomes the defect that the traditional algorithm requires a preset S p ; the present invention realizes a robust forward inversion without generating outliers, overcoming the instability of the traditional forward inversion; the present invention is more convenient in obtaining boundary conditions, without measuring or estimating the boundary conditions at deep water, only requiring a boundary condition at the water surface layer It can be inverted; the water surface boundary conditions required by the present invention can be obtained by ocean color remote sensing, which is conducive to the combination of active and passive remote sensing. Description of the Drawings

[0041] Figure 1 It is a flow chart of a forward iteration inversion method for an ocean lidar based on instability suppression according to the present invention;

[0042] Figure 2 It is a partial flow chart of step S2 in the present invention;

[0043] Figure 3 It is a partial flow chart of step S3 in the present invention;

[0044] Figure 4 It is a comparison chart of the optical inversion result and the in-situ detection result in the embodiment of the present invention. Detailed Embodiment

[0045] The present invention will be further described in detail below with reference to the drawings and embodiments. It should be noted that the following embodiments are intended to facilitate the understanding of the present invention and do not limit it in any way.

[0046] As Figure 1 shown, a forward iteration inversion method for an ocean lidar based on instability suppression mainly includes 4 steps: S1 preprocesses the lidar signal data to complete denoising and calibration; S2 and S3 complete the solution of the lidar ratio; step S4 completes the calculation and output of the final inversion result.

[0047] As Figure 2 shown, the algorithm of step S2 is a process of continuous binary iteration. S2 will narrow the solution range of S p by detecting physical meaning errors:

[0048] (2.a) Initialize the iteration number loop to 1 and initialize the solution range p of S to [0, 1000];

[0049] (2.b) Judge whether loop < 60. If the judgment result is true, execute (2.c) sequentially; if the judgment result is false, jump to execute (2.i);

[0050] (2.c) Let

[0051] (2.d) Use as the initial value, substitute the current value of S p and the signal D into the forward Fernald inversion iteration formula to obtain the inversion profile β π (n) with a length of N;

[0052] (2.e) Determine β π (n) to check if it exists If the judgment result is true, jump to and execute (2.g); if the judgment result is false, execute (2.f) sequentially;

[0053] (2.f) Determine β π (n) to check if it exists If the judgment result is true, jump to and execute (2.h); if the judgment result is false, jump to and execute (2.j);

[0054] (2.g) loop++, return to (2.b) and run in a loop;

[0055] (2.h) loop++, return to (2.b) and run in a loop;

[0056] (2.i) Stop executing the program and report an error;

[0057] (2.j) Execute S3.

[0058] As Figure 3 shown, S3 will optimize the objective function F(S p ) to achieve the exact solution of S p :

[0059] (3.a) Extract what is obtained from S2 Let

[0060] (3.b) Let Calculate

[0061] (3.c) Judge if i <= n, if the judgment is true, execute (3.d); if the judgment is false, execute (3.g);

[0062] (3.d) Calculate F(S p ) is calculated as follows: with as the initial value, substitute the function parameter S p and the signal D into the forward Fernald inversion iterative formula to obtain the inversion profile β π (n), and return the value where is the median value of the β π profile;

[0063] (3.e) Judge V tmp < V 0 ; if the judgment is true, let V 0 = V tmp ,

[0064] (3.f) Let i = i + 1, and then jump to execute (3.c);

[0065] (3.g) Store the current value, end S3, and execute S4.

[0066] S4 will, according to the value obtained from S3, calculate K d (z) and β π (z) profiles through Fernald forward iteration and output the results.

[0067] The optical parameter results obtained by inversion according to the embodiment are as Figure 4 shown. Among them, the 'x'-marked line represents the optical parameters detected by the in-situ instrument at the whole hour, the solid line represents the average inversion result of 51 lidar signal profiles within ±5 minutes of the whole hour, and the error bars represent the standard deviation of ±1. It can be seen that the result of the forward inversion is stable and does not produce outliers, which proves that the inversion algorithm has successfully overcome the instability. It can also be seen that the coincidence degree between the inversion result and the in-situ measurement result is good, indicating the effectiveness of the inversion method.

[0068] In this embodiment, by using the forward Fernald inversion algorithm based on the solution of lidar ratio of the present invention, the detection and inversion of the water body optical parameter K d , β π profiles are realized, and the comparison with the detection results of the in-situ instrument is good, verifying the effectiveness of the present invention.

[0069] The above-described embodiments have detailed 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 used to limit the present invention. Any modifications, supplements, and equivalent replacements 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 forward iterative inversion method for marine lidar based on instability suppression, characterized in that, it includes the following steps: S1. Preprocess the collected original lidar data to obtain a discrete signal D of length N, and obtain the boundary conditions corresponding to the depth position of D(0). S2, by detecting physical meaning errors, narrow the solution range of lidar ratio S p is The specific process is as follows: (2.a) Define the lidar ratio S p The initial range is (2.b) Let (2.c) Using as the initial value, substitute the current S p value and signal D into the forward Fernald inversion iteration formula to obtain the inversion profile β π (n) with a length of N; (2.d) Determine β π in (n) whether there is If the judgment result is true, then execute (2.f); if the judgment result is false, then execute (2.e); (2.e) Determine β π if it exists in (n) If the judgment result is true, execute (2.g); if the judgment result is false, execute (2.h); is the 180-degree volume scattering coefficient of seawater molecules; (2.f) Let Return to the loop operation in (2.b); (2.g) Let Return (2.b) loop runs; (2.h) End S2 and execute S3; S3. By optimizing the objective function F(S p ), the exact solution of S p is achieved, specifically as follows: (3.a) Obtained by extracting S2 Let (3.b) Let Calculate (3.c) Judge if i < 10 3 , if the judgment is true, execute (3.d); if the judgment is false, execute (3.g) (3.d) Calculate (3.e) Judge V tmp <V 0 ; If the judgment is true, let V 0 =V tmp , (3.f) Let i = i + 1, and jump to execute (3.c); (3.g) Store the current value, end S3, and execute S4; S4, starting with as the initial value, substitute the function parameter S p and the signal D into the forward Fernald inversion iteration formula to obtain the inversion profile β π (n) of length N, and return the value where is the median value of the β π profile.

2. The forward iterative inversion method for marine lidar based on instability suppression according to claim 1, characterized in that, in step S1, the preprocessing of the collected original lidar data is specifically as follows: Considering the optical properties of water molecules and other seawater components respectively, perform range correction and noise removal on the original lidar data to obtain a signal D that obeys the following equation: where z represents the depth; K dw is the diffuse attenuation coefficient of seawater molecules; and K dp are respectively the 180-degree volume scattering coefficient and the diffuse attenuation coefficient of the remaining seawater components.

3. The forward iterative inversion method for marine lidar based on instability suppression according to claim 2, characterized in that, In step S1, the boundary condition acquisition specifically includes: measuring the boundary condition through in-situ measurement or water color remote sensing 4. The forward iterative inversion method for marine lidar based on instability suppression according to claim 3, characterized in that, the forward Fernald inversion iteration formula is: where Δz is the spatial resolution of the lidar, and S w is the extinction-scattering ratio of pure seawater molecules.

Images