A hierarchical detection method for ocean lidar based on differential threshold evaluation

Abstract

This invention discloses a method for detecting ocean layers using lidar (LiDAR) based on differential threshold evaluation. The method includes the following steps: After preprocessing the airborne LiDAR signal through deduplication, background signal removal, and distance correction, the valid signal range is screened. A differential function model is then established. By analyzing the amplitude of the signal slope change and using the relative intensity between layer peaks and layer bottoms as a constraint, the ocean layers are filtered from the LiDAR echo signal. This method can directly locate layer boundaries, reduce interference from the water environment and signal fluctuations, and achieve rapid and accurate detection of ocean layers.

Description

Technical Field

[0001] The present invention belongs to the technical field of ocean laser radar, and in particular relates to an ocean laser radar hierarchical detection method based on differential threshold evaluation. Background Art

[0002] Marine ecosystems are crucial for maintaining Earth's life systems and material cycles, significantly impacting human production and daily life. A thin layer, primarily composed of phytoplankton, represents the ocean's primary productivity. The layer's thickness ranges from a few centimeters to over ten centimeters, extending horizontally for thousands of meters and persisting for days or even months. The waters in which the layer resides differ significantly from the surrounding waters, containing chlorophyll levels as high as 55 times that of background water. This chlorophyll level significantly influences geochemical processes such as the carbon cycle, organic matter transport and transfer, fisheries development, and harmful algal blooms.

[0003] Layer detection technology can help better understand the underwater distribution and growth of thin layers, enabling understanding and monitoring of ocean carbon cycles and marine ecological changes. Layer detection is also the foundation for lidar inversion. Layer detection technology can accurately classify lidar signals and help determine the lidar ratios of different particles, laying the foundation for subsequent lidar inversion and obtaining bio-optical profiles.

[0004] In recent years, lidar layer detection technology has flourished, and many excellent layer detection algorithms have emerged. In atmospheric layer detection, the slope method uses a multi-point window to fit the signal slope. When a certain number of consecutive slopes are greater than zero, the layer base is determined. The wavelet covariance method (WCT) transforms the echo signal using a wavelet function and analyzes the local maximum of the wavelet transform modulus to determine the layer position. Chinese patent publication No. CN113504528A discloses an atmospheric layer detection method based on multi-scale hypothesis testing. This invention uses a fixed-size sliding window to detect the entire signal after attenuation calculation. Based on Poisson, Gaussian, or Bernoulli distributions, it determines whether the center point of the window belongs to a layer point. The threshold method is currently the most widely used layer detection method. For example, Chinese patent publication No. CN107870336A discloses an iterative detection method for the bottom of a penetrable layer for spaceborne lidar. This method uses the ratio of the attenuation-scattering ratio calculated from the layer bottom-top signal to the attenuation-scattering ratio obtained by profile scanning as a threshold. The detection of the atmospheric boundary layer is continuously optimized through an iterative algorithm. For ocean lidar layer detection, the perturbation method proposed by NOAA is currently the most widely used algorithm. Researchers at the Second Institute of Oceanography under the Ministry of Natural Resources have iteratively updated this algorithm and proposed an adaptive perturbation method for ocean layer detection. This method assumes that lidar signals decay exponentially in water, with a constant attenuation coefficient. It then uses linear fitting to identify layer bottoms and peaks.

[0005] Given the different physical properties of the ocean and atmosphere, and the distinct optical characteristics of gas and water molecules, lasers experience different attenuation and scattering characteristics when propagating through the atmosphere and ocean. The relative intensity of the ocean subsurface and water is not as strong as that of aerosols and clouds, making existing atmospheric detection algorithms unsuitable for marine environments. The perturbation method developed for ocean lidar layer detection is simple in principle, but due to prior assumptions, layer detection results are significantly affected by water characteristics and can easily result in large detection errors in the presence of large signal fluctuations. Therefore, there is an urgent need to develop an accurate and fast layer detection algorithm for the ocean. Summary of the Invention

[0006] In order to solve the above problems existing in the prior art, the present invention provides an ocean lidar layer detection method based on differential threshold evaluation, which can realize rapid and accurate detection of marine phytoplankton layers.

[0007] A method for detecting ocean layers using lidar based on differential threshold evaluation is proposed. By comparing the amplitude of signal slope changes and using the relative intensity of layer peaks and bottoms as a constraint, the method can filter out ocean layers from lidar echo signals. The method includes the following steps:

[0008] (1) Deduplication of identical GPS signals for airborne lidar signals;

[0009] (2) Perform preliminary denoising on the deduplicated airborne lidar signal to remove the influence of background signals;

[0010] (3) Perform distance correction on the LiDAR signal after preliminary denoising according to the operating height;

[0011] (4) For the corrected lidar signal, smooth and denoise it along the single profile to determine the effective lidar signal range of the single profile;

[0012] (5) Establish a differential function model, set a sliding window with a fixed size of W, and calculate the slope difference of the signal in two adjacent windows;

[0013] (6) Setting the peak height threshold, determining the peak size and position of the differential function, and determining the layer bottom and layer height;

[0014] (7) Set the relative intensity screening threshold, calculate the relative intensity between the layer peak and the layer bottom, discard the unreasonable layer detection results below the threshold, and finally determine the layer position corresponding to this profile.

[0015] The present invention is based on an ocean lidar operating in the blue-green band and realizes accurate detection of ocean layers by proposing a detection technology based on differential threshold evaluation.

[0016] Taking into account the high-frequency sampling of the laser radar, in order to reduce accidental errors, in step (1), the echo signals of the same GPS positioning are averaged multiple times for the airborne laser radar signal to complete the deduplication of the same GPS signal.

[0017] In step (2), the specific process of preliminary denoising is as follows: based on the echo signal averaged multiple times, assuming that the echo signal at infinity is negligible, the mean of the 100 sampling points at the end of the signal is taken as the background signal, and the background signal is eliminated by linear operation.

[0018] In step (3), the distance correction formula is:

[0019] S M (z) = S o (z)×(nH+z) 2

[0020] Where S o and S M represent the lidar echo signal after background signal removal and the lidar echo signal after distance correction, respectively; n, H, and z represent the water refractive index, radar operating height, and water penetration depth, respectively.

[0021] In step (4), the process of determining the effective lidar signal range of a single profile is as follows:

[0022] (4-1) Calculate the change of adjacent sampling points for a single profile

[0023] (4-2) Find the depth z where the signal decays to one thousandth of the maximum intensity r , as the maximum depth of effective signal;

[0024] (4-3) Set the window width d, at the selected depth z r The following slides the window in turn and calculates the mean φ(z) of the change in the window;

[0025] (4-4) Using φ(z) as the threshold to judge the degree of signal fluctuation, if at a certain depth Z noise The change of multiple consecutive sampling points Exceeding the threshold φ(z), determine the depth Z noise The signal after is invalid; take the depth 2m-Z noise The echo signal within the range is a valid signal S E .

[0026] The specific process of step (5) is:

[0027] (5-1) At depth 2m-Z noise Perform logarithmic processing on single profile within the range

[0028] LogS(z)=ln(S E (z))

[0029] (5-2) Set a fixed-size sliding window W and calculate the x value for a single profile. z The slope value of the signal in the two windows before and after the center is F(x z ) and G(x z )(x≤z r ) represent the calculated slope values;

[0030] (5-3) Define the difference function model:

[0031] Fun = G(x) - F(x).

[0032] In step (6), the quantile of the differential function Fun is used as the peak height threshold, and the peak of the differential function Fun is found. The values ​​below the peak height threshold are discarded, and the first peak position is the layer height, and the last peak position is the layer bottom.

[0033] In step (7), the effective signal S E After compensation, logarithmic processing is performed to obtain the logarithmic signal standard deviation sigma. Sigma is used as the relative intensity screening threshold. The layer with the relative intensity between the layer peak and the layer bottom higher than the screening threshold is the effective layer. The specific formula is:

[0034]

[0035]

[0036] Where z r is the maximum depth of the effective signal; α0 is the attenuation coefficient of the lidar echo signal obtained by the perturbation method; D is the relative intensity of the layer peak; z down 、z up are floor bottom and floor height respectively; S res It is the equation of the straight line consisting of the sampling points corresponding to the layer bottom and layer height.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] This invention provides a new method for ocean lidar layer detection, using signal variation trends as a starting point to accurately locate the left and right boundaries of layers in the echo signal. Traditional methods are particularly susceptible to signal fluctuations and can produce significant detection errors, especially in conditions with poor signal-to-noise ratios. The differential threshold assessment method described in this invention directly locates layer boundaries, minimizing interference from signal fluctuations and enabling rapid and accurate detection of ocean layers. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 This is a flow chart of a method for ocean lidar layer detection based on differential threshold evaluation according to the present invention;

[0040] Figure 2 Result diagram of the differential model in an embodiment of the present invention;

[0041] Figure 3 This is a result diagram of single-profile hierarchical detection in an embodiment of the present invention;

[0042] Figure 4 This is a relative signal strength diagram of the levels obtained by detecting 126 profiles in an embodiment of the present invention;

[0043] Figure 5 is an intensity diagram of the original signal in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] 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.

[0045] The present invention is based on an ocean lidar operating in the blue-green band and proposes an ocean lidar layer detection method based on differential threshold evaluation. A large number of echo signals are averaged to reduce accidental errors; the effective signal range is determined after denoising, smoothing and distance correction; and a differential function model is established based on the effective signal to ultimately determine the effective layer.

[0046] The embodiment of the present invention takes a set of data from March 2018 as an example. Figure 1 As shown, a marine lidar layer detection method based on differential threshold evaluation includes:

[0047] In step (1), 161,598 echo signals were obtained. After 100 averaging and GPS deduplication, 126 valid profiles were finally obtained, each with 800 sampling points.

[0048] Step (2) removes background noise signals from the 126 profiles. Taking a single profile as an example, after removing the abnormal signal at the end, the signal at a depth of 70 to 88 meters is taken as the background signal value, and its mean is calculated and then linearly calculated to remove the background noise.

[0049] Step (3): remove the background noise of the echo signal S o (z) Perform distance correction.

[0050] S M (z) = S o (z)×(nH+z) 2

[0051] Among them, n is 1.33 and H is 613.8874m.

[0052] Step (4): Determine the distance-corrected echo signal S M (z) The effective range. Due to the influence of sea waves, the depth below 2m is taken as the effective signal. The removal of tail noise signal specifically includes the following steps:

[0053] (4-1) Calculate the change of adjacent sampling points for this single profile

[0054] (4-2) Find the depth z where the signal decays to one thousandth of the maximum intensity r .

[0055] (4-3) Set the window width d = 10, at the selected depth z r =31.3534m and below, the windows are sequentially slid and the mean φ(z) of the variation within the window is calculated.

[0056] (4-4) Using φ(z) as the threshold to judge the degree of signal fluctuation, at depth Z noise= Changes in multiple consecutive sampling points after 32.3684m Exceeding the threshold φ(z), determine the depth Z noise The signal after is invalid. Take the depth 2m-Z noise The echo signal within the range is a valid signal S E .

[0057] Step (5): using the effective echo signal S E Establishing a differential function model includes the following points:

[0058] (5-1) Perform logarithmic processing on this profile within the valid range.

[0059] LogS(z)=ln(S E (z))

[0060] (5-2) Set the sliding window W = 20, and calculate the x z The slope value of the signal in the two windows before and after the center is F(x z ) and G(x z )(x≤z r ) represent the calculated slope values, such as Figure 2 (a) shows F(x z ) and G(x z )(x≤z r ) curve that changes with depth.

[0061] (5-3) Define the difference function model, such as Figure 2 (b) shows the differential function model curve.

[0062] Fun=G(x)-F(x)

[0063] Step (6) uses the quantile of the differential function Fun as the threshold, finds the peak value of the differential function Fun, and discards the values ​​below the threshold, such as Figure 2 As shown in (b), the circled position is the peak of the function after filtering, and the selected threshold is 0.0247. The first peak position is the layer height, and the last peak position is the layer bottom.

[0064] Step (7): Compensate the effective signal S E To compensate, the logarithmic signal within the effective range is fitted, the fitting slope α0 is 0.0829, and the compensation signal is:

[0065] S r (z) = S E (z)×exp(2α0z)

[0066] Perform logarithmic processing on the compensation signal to obtain the standard deviation sigma of the logarithmic signal. Use sigma as the threshold to determine whether the primary level is the real level.

[0067]

[0068]

[0069] Where z r is the maximum depth of the effective signal; α0 is the attenuation coefficient of the lidar echo signal obtained by the perturbation method; D is the relative intensity of the layer peak; z down 、z up are floor bottom and floor height respectively; S res It is the equation of the straight line consisting of the sampling points corresponding to the layer bottom and layer height.

[0070] The hierarchical recognition results of the selected contours in this example are as follows Figure 3 As shown in Figure 1, the dotted lines in (a) are the selected levels, and the dotted lines in (b) are the sigma reference lines. By comparing with sigma, it can be seen that the detected level is the real level. The detection results of 126 profiles are shown in Figure 1. Figure 4 As shown, Figure 5 The original signal comparison is basically consistent, verifying the effectiveness of the present invention.

[0071] 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 marine lidar layer detection method based on differential threshold evaluation, characterized in that: The following steps are involved: (1) Deduplication of identical GPS signals for airborne lidar signals; (2) Perform preliminary denoising on the deduplicated airborne lidar signal to remove the influence of background signals; (3) Perform distance correction on the LiDAR signal after preliminary denoising according to the operating height; (4) For the corrected lidar signal, smooth and denoise it along the single profile to determine the effective lidar signal range of the single profile; (5) Establish a differential function model, set a sliding window with a fixed size of W, and calculate the slope difference of the signal in two adjacent windows; (6) Setting the peak height threshold, determining the peak size and position of the differential function, and determining the layer bottom and layer height; (7) Set the relative intensity screening threshold, calculate the relative intensity between the layer peak and the layer bottom, discard the unreasonable layer detection results below the threshold, and finally determine the layer position corresponding to this profile.

2. The ocean lidar layer detection method based on differential threshold evaluation according to claim 1 is characterized in that: In step (1), for the airborne laser radar signal, the echo signal of the same GPS positioning is averaged multiple times to complete the deduplication of the same GPS signal.

3. The ocean lidar layer detection method based on differential threshold evaluation according to claim 1 is characterized in that: In step (2), the specific process of performing preliminary denoising is as follows: taking the mean of the last 100 sampling points of the signal as the background signal, and performing linear operation to eliminate the background signal.

4. The ocean lidar layer detection method based on differential threshold evaluation according to claim 1 is characterized in that: In step (3), the distance correction formula is: S M (z)=S o (z)×(nH+z) 2 Where S o and S M represent the lidar echo signal after background signal removal and the lidar echo signal after distance correction, respectively; n, H, and z represent the water refractive index, radar operating height, and water penetration depth, respectively.

5. The ocean lidar layer detection method based on differential threshold evaluation according to claim 1 is characterized in that: In step (4), the process of determining the effective lidar signal range of a single profile is as follows: (4-1) Calculate the change of adjacent sampling points for a single profile (4-2) Find the depth z where the signal decays to one thousandth of the maximum intensity r , as the maximum depth of effective signal; (4-3) Set the window width d, at the selected depth z r The following slides the window in turn and calculates the mean φ(z) of the change in the window; (4-4) Using φ(z) as the threshold to judge the degree of signal fluctuation, if at a certain depth Z noise The change of multiple consecutive sampling points Exceeding the threshold φ(z), determine the depth Z noise The signal after is invalid; considering the influence of sea waves, take the depth 2m-Z noise The echo signal within the range is a valid signal S E .

6. The ocean lidar layer detection method based on differential threshold evaluation according to claim 5 is characterized in that: The specific process of step (5) is: (5-1) At depth 2m-Z noise Perform logarithmic processing on single profile within the range LogS(z)=ln(S E (z)) (5-2) Set a fixed-size sliding window W and calculate the x value for a single profile. z The slope value of the signal in the two windows before and after the center is F(x z ) and G(x z ) represent the calculated slope values, x≤z r ; (5-3) Define the difference function model: Fun = G(x) - F(x).

7. The ocean lidar layer detection method based on differential threshold evaluation according to claim 6 is characterized in that: In step (6), the quantile of the differential function Fun is used as the peak height threshold, and the peak of the differential function Fun is found. The values ​​below the peak height threshold are discarded, and the first peak position is the layer height, and the last peak position is the layer bottom.

8. The ocean lidar layer detection method based on differential threshold evaluation according to claim 7 is characterized in that: In step (7), the effective signal S E After compensation, logarithmic processing is performed to obtain the logarithmic signal standard deviation sigma. Sigma is used as the relative intensity screening threshold. The layer with the relative intensity between the layer peak and the layer bottom higher than the screening threshold is the effective layer. The specific formula is: Where z r is the maximum depth of the effective signal; α0 is the attenuation coefficient of the lidar echo signal obtained by the perturbation method; D is the relative intensity of the layer peak; z down 、z up are floor bottom and floor height respectively; S res It is the equation of the straight line consisting of the sampling points corresponding to the layer bottom and layer height.

Images