A multi-spectral remote sensing water depth inversion method and computer readable storage medium

Abstract

The present invention proposes a multispectral remote sensing water depth inversion method and a computer-readable storage medium. The method includes preprocessing the multispectral remote sensing image to obtain the water surface reflectivity data of the target area; performing water depth inversion based on the water surface reflectivity data and the UMOPE model to preliminarily obtain the water depth and the optical characteristic related parameters of the water body; establishing a relationship function between the optical characteristic related parameters of the water body and the water depth, and replacing the optical characteristic related parameters of the water body in the UMOPE model according to the relationship function to improve the UMOPE model; re-performing water depth inversion based on the improved UMOPE model and the water surface reflectivity data to obtain the final water depth. The present invention improves the depth measurement accuracy and has important application value.

Description

Technical Field

[0001] The present invention relates to the technical field of water depth inversion methods, and in particular to a multi-spectral remote sensing water depth inversion method and a computer-readable storage medium. Background Art

[0002] Water depth information is essential for assessing the health of water bodies, monitoring water quality changes and ecosystem services, and is also an important reference for port construction, waterway maintenance and marine engineering planning. Traditional water depth measurements require the use of ships and acoustic sounding equipment, which usually provide point or line data. They are costly and time-sensitive, and are difficult to meet the depth measurement needs of large areas or remote waters. Compared with traditional methods, remote sensing depth measurement can collect water depth data over a large area more quickly and cost-effectively.

[0003] With the advancement of remote sensing technology, data processing capabilities and computing methods, remote sensing water depth inversion methods are constantly being optimized. Among them, physical models based on optical properties and radiation transfer theory, such as the UMOPE algorithm (unmixing-based multispectral optimization process exemplar method), can more accurately simulate the propagation process of light in water bodies, while inverting water depth, inherent optical properties of water bodies and benthic reflectivity, and can usually obtain good bathymetric results. Therefore, it is currently an ideal remote sensing water depth inversion method. The UMOPE algorithm contains 7 unknown parameters. In the inversion process, the existence of multiple parameters will inevitably produce additional degrees of freedom, thereby introducing more local minimum values ​​in the solution process and reducing the reliability of the model output. Therefore, it is necessary to optimize the model parameters to improve the accuracy of water depth inversion. Summary of the invention

[0004] In view of the above problems, the present invention provides a multi-spectral remote sensing water depth inversion method and a computer-readable storage medium.

[0005] In order to solve the above technical problems, in a first aspect, the present invention provides a multi-spectral remote sensing water depth inversion method, comprising:

[0006] Preprocess the multispectral remote sensing images to obtain the water surface reflectance data of the target area;

[0007] Performing water depth inversion based on the water surface reflectivity data and the UMOPE model to preliminarily obtain parameters related to water depth and optical characteristics of water bodies;

[0008] Establishing a relationship function between the optical characteristic related parameters of the water body and the water depth, and replacing the optical characteristic related parameters of the water body in the UMOPE model according to the relationship function to improve the UMOPE model;

[0009] The water depth inversion is re-performed based on the improved UMOPE model and the water surface reflectivity data to obtain the final water depth.

[0010] Furthermore, the preprocessing includes image cropping and atmospheric correction.

[0011] Furthermore, the optical property related parameters of the water body include the absorption coefficient of phytoplankton at 440nm , light absorption coefficient of colored dissolved organic matter and debris at 440nm and the backscattering coefficient of particles at 400nm .

[0012] Furthermore, the improved UMOPE model is:

[0013]

[0014]

[0015]

[0016]

[0017]

[0018]

[0019]

[0020]

[0021]

[0022]

[0023]

[0024]

[0025]

[0026] in, is the subsurface remote sensing reflectivity, is the optical deep water subsurface remote sensing reflectivity, is an exponential function with the natural constant e as base, is the zenith angle of incident sunlight, is the zenith angle of satellite observation, is the optical path scattering factor of water, is the optical path scattering factor at the bottom of water, is the diffuse radiation attenuation coefficient of water body, For water depth, is the bottom albedo, is the circumference of a circle, is the water surface reflectance data obtained from multispectral remote sensing images. is the ratio of the backscattering coefficient to the diffuse attenuation coefficient, is the water absorption coefficient, is the backscattering coefficient, is the central wavelength of the satellite sensor, is the albedo of sand, is the albedo of the coral, is the albedo of seagrass, , , are proportional coefficients, is the pure water absorption coefficient, is the phytoplankton absorption coefficient, is the absorption coefficient of colored dissolved organic matter and debris, is the logarithm with constant e as base, , These are all empirical parameters. is the backscattering coefficient of pure water, is the backscattering coefficient of suspended particles, is the absorption coefficient of phytoplankton at 440 nm The relationship function with water depth is is the light absorption coefficient of colored dissolved organic matter and debris at 440nm The relationship function with water depth is is the backscattering coefficient of particles at 400nm Function of the water depth.

[0027] Furthermore, the relationship function is an exponential function.

[0028] In a second aspect, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program implements the above method when executed by a processor.

[0029] Compared with the prior art, the beneficial effects of the present invention include: the present invention first obtains the water depth and the optical characteristic related parameters of the water body through the UMOPE model inversion, then improves the UMOPE model by establishing the relationship function between the optical characteristic related parameters of the water body and the water depth, and then re-inverts based on the improved UMOPE model to obtain the water depth spatial distribution result. Compared with the direct inversion using the UMOPE algorithm, the depth measurement accuracy is improved, and it has important application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 A flow chart of a multi-spectral remote sensing water depth inversion method according to an embodiment of the present invention;

[0031] Figure 2 is a graph of the absorption coefficient of phytoplankton at 440nm versus water depth;

[0032] Figure 3 is the relationship between the light absorption coefficient of colored dissolved organic matter and debris at 440nm and water depth;

[0033] Figure 4 is the relationship between the backscattering coefficient of particles at 400nm and water depth;

[0034] Figure 5 The water depth distribution map of Ganquan Island area was obtained by inverting the improved UMOPE model;

[0035] Figure 6 This is a schematic diagram of the matching between the water depth in the Ganquan Island area obtained by inversion based on the UMOPE model and the measured water depth;

[0036] Figure 7 This is a schematic diagram of the matching between the water depth in the Ganquan Island area obtained by inversion based on the improved UMOPE model and the measured water depth. DETAILED DESCRIPTION

[0037] The present invention is further explained below in conjunction with the accompanying drawings and specific embodiments. The embodiments are implemented based on the technical solution of the present invention. It should be understood that these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention.

[0038] like Figure 1 As shown, an embodiment of the present invention provides a multi-spectral remote sensing water depth inversion method, comprising:

[0039] The multispectral remote sensing image is preprocessed to obtain the water surface reflectance data of the target area. Specifically, the preprocessing includes image cropping and atmospheric correction.

[0040] The water depth inversion is performed based on the water surface reflectance data and the UMOPE model to obtain the water depth and the optical characteristics of the water body. Specifically, the optical characteristics of the water body include the absorption coefficient of phytoplankton at 440nm. , light absorption coefficient of colored dissolved organic matter and debris at 440nm and the backscattering coefficient of particles at 400nm .

[0041] The above UMOPE model is:

[0042]

[0043]

[0044]

[0045]

[0046]

[0047]

[0048]

[0049]

[0050]

[0051]

[0052]

[0053]

[0054]

[0055] in, is the subsurface remote sensing reflectivity, is the optical deep water subsurface remote sensing reflectivity, is an exponential function with the natural constant e as base, is the zenith angle of incident sunlight, is the zenith angle of satellite observation, is the optical path scattering factor of water, is the optical path scattering factor at the bottom of water, is the diffuse radiation attenuation coefficient of water body, For water depth, is the bottom albedo, is the circumference of a circle, is the water surface reflectance data obtained from multispectral remote sensing images. is the ratio of the backscattering coefficient to the diffuse attenuation coefficient, is the water absorption coefficient, is the backscattering coefficient, is the central wavelength of the satellite sensor, is the albedo of sand, is the albedo of the coral, is the albedo of seagrass, , , are proportional coefficients, is the pure water absorption coefficient, is the phytoplankton absorption coefficient, is the absorption coefficient of colored dissolved organic matter and debris, is the logarithm with constant e as base, , These are all empirical parameters. is the backscattering coefficient of pure water, is the backscattering coefficient of suspended particles.

[0056] The cost function of the UMOPE model is:

[0057]

[0058] in, is the calculation result of the cost function, is the number of satellite sensor bands used, that is, The number of The water surface reflectivity data obtained by inverting the UMOPE model. Within the constraint range of unknown parameters, the least squares method is used to solve the minimum value of the above cost function to obtain the water depth and parameters. , , In addition, the proportionality coefficient can also be obtained , , The constraint intervals of the unknown parameters are shown in Table 1:

[0059] Table 1: Constraint interval table of unknown parameters

[0060]

[0061] A relationship function between the optical property related parameters of the water body and the water depth is established, and the optical property related parameters of the water body in the UMOPE model are replaced according to the relationship function to improve the UMOPE model. Specifically, the established relationship function is as follows: P = f1(H), G = f2(H), X = f3(H). The above f1(H), f2(H), and f3(H) can all be exponential functions, but are not limited to this. It depends on the specific data. If the area or satellite is changed, other functions may also be used.

[0062] Substituting the above relationship function into the UMOPE model, the improved UMOPE model can be obtained as follows:

[0063]

[0064]

[0065]

[0066]

[0067]

[0068]

[0069]

[0070]

[0071]

[0072]

[0073]

[0074]

[0075]

[0076] The water depth inversion is performed again based on the improved UMOPE model and the water surface reflectivity data to obtain the final water depth. Specifically, the water depth inversion is performed by using the least squares method to solve the minimum value of the cost function to obtain the final water depth. .

[0077] The number of satellite sensor bands used for Landsat-8 images is 4, the central wavelength They are 443nm, 483nm, 561nm and 655nm respectively, and the central wavelengths are The corresponding empirical parameters , They are shown in Table 2:

[0078] Table 2: Central wavelength and empirical parameters , Comparison Table

[0079]

[0080] An example of using airborne radar bathymetry data as the measured reference water depth for remote sensing water depth inversion and accuracy verification is given below:

[0081] (1) Taking Ganquan Island in the Xisha Islands in the South China Sea as an example study area, Landsat-8 remote sensing images of the area were collected, atmospheric correction was performed on them using ACOLITE software, and the images were cropped with reference to the boundaries of islands and reefs to obtain the water surface reflectance data of the Ganquan Island area.

[0082] (2) Using the UMOPE model to invert the water depth and parameters , , , and establish parameters , , The relationship functions with water depth are as follows: Figures 2 to 4 As shown, the expressions of f1(H), f2(H), and f3(H) are:

[0083]

[0084]

[0085]

[0086] Substitute the above relationship function into the UMOPE model to obtain the improved UMOPE model.

[0087] (3) The improved UMOPE model was used to invert the water depth in the Ganquan Island area. The spatial distribution of the water depth is shown in the following figure: Figure 5 shown.

[0088] (4) The two water depth results were matched with the measured water depth to verify the accuracy of the water depth inversion of the UMOPE model and the improved UMOPE model, such as Figure 6 and Figure 7 As shown, by calculating the determination coefficient of the UMOPE model and the improved UMOPE model The root mean square error RMSE of the water depth inversion results is used to compare and analyze the model accuracy. The calculation method of the root mean square error RMSE of the water depth inversion result is as follows:

[0089]

[0090]

[0091] in, For the The actual water depth of the measured water depth point, is the first one obtained by inversion of the improved UMOPE model The water depth at the measured water depth point, is the mean measured water depth, is the total number of measured water depth points, The determination coefficient of the UMOPE model is is 0.71, while the determination coefficient of the improved UMOPE model is The inverted water depth based on the improved UMOPE model has a higher linear correlation with the measured water depth and better consistency. The root mean square error (RMSE) of the water depth inversion result of the UMOPE model is 2.4 m, and the root mean square error (RMSE) of the water depth inversion result of the improved UMOPE model is 1.25 m. The error between the water depth inverted based on the improved UMOPE model and the measured water depth is smaller, and the water depth inverted based on the improved UMOPE model is more accurate.

[0092] Based on the above embodiments, those skilled in the art can easily understand that the present invention also provides a computer-readable storage medium storing a computer program, which implements the above method when executed by a processor.

[0093] The above is only a preferred embodiment of the present invention. It should be noted that for ordinary technicians in this technical field, other parts not specifically described belong to the prior art or common knowledge. Several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered as the protection scope of the present invention.

Claims

1. A multispectral remote sensing water depth inversion method, characterized in that: include: Preprocessing the multispectral remote sensing image to obtain water surface reflectance data of the target area, wherein the preprocessing includes image clipping and atmospheric correction; Based on the water surface reflectance data and the UMOPE model, water depth inversion is performed to preliminarily obtain water depth and optical property related parameters of the water body, wherein the optical property related parameters of the water body include the absorption coefficient P of phytoplankton at 440nm, the light absorption coefficient G of colored dissolved organic matter and debris at 440nm, and the backscattering coefficient X of particulate matter at 400nm; Establishing a relationship function between the optical characteristic related parameters of the water body and the water depth, and replacing the optical characteristic related parameters of the water body in the UMOPE model according to the relationship function to improve the UMOPE model; Re-performing water depth inversion based on the improved UMOPE model and the water surface reflectivity data to obtain the final water depth; The improved UMOPE model is: u(λ)=b b (λ) / (a(λ)+b b (l)) k(λ)=a(λ)+b b (l) ρ=B1ρ1+B2ρ2+B3ρ3 a(λ)=a w (λ)+a φ (λ)+a g (l) a φ (λ)=[a0(λ)+a1(λ)ln(f1(H))]f1(H) a g (λ)=f2(H)exp[-0.0166(λ-440)] b b (λ)=b bw (λ+b bp (l) Among them, r rs is the subsurface remote sensing reflectivity, is the remote sensing reflectance of the subsurface layer in the optical deep water area, exp is an exponential function with the natural constant e as the base, θ w is the zenith angle of sunlight incidence, θ is the zenith angle of satellite observation, is the optical path scattering factor of water, is the optical path scattering factor of the bottom of the water, κ is the radiation diffuse attenuation coefficient of the water body, H is the water depth, ρ is the bottom albedo, π is the circumference, R rs is the water surface reflectance data obtained from multispectral remote sensing images, u is the ratio of backscattering coefficient to diffuse attenuation coefficient, a is the water body absorption coefficient, b b is the backscattering coefficient, λ is the central wavelength of the satellite sensor, ρ1 is the albedo of sand, ρ2 is the albedo of coral, ρ3 is the albedo of seagrass, B1, B2, B3 are all proportional coefficients, a w is the pure water absorption coefficient, a φ is the phytoplankton absorption coefficient, a μ is the absorption coefficient of colored dissolved organic matter and debris, ln is the logarithm with constant e as base, a o , a1 are empirical parameters, b bw is the backscattering coefficient of pure water, b bp is the backscattering coefficient of suspended particles, f1(H) is the function of the absorption coefficient P of phytoplankton at 440nm and water depth, f2(H) is the function of the light absorption coefficient G of colored dissolved organic matter and debris at 440nm and water depth, and f3(H) is the function of the backscattering coefficient X of particles at 400nm and water depth; The two water depth results were matched with the measured water depth to verify the accuracy of the water depth inversion of the UMOPE model and the improved UMOPE model. The determination coefficient R of the UMOPE model and the improved UMOPE model was calculated. 2 The root mean square error RMSE of the water depth inversion results is used to compare the model accuracy. The determination coefficient R 2 The calculation method of the root mean square error RMSE of the water depth inversion result is as follows: Among them, H true_i is the true water depth of the i-th measured water depth point, H model_i is the water depth of the ith measured water depth point obtained by inversion of the improved UMOPE model, is the mean of the measured water depth, m is the total number of measured water depth points, i = 1, 2, ... m.

2. A multi-spectral remote sensing water depth inversion method according to claim 1, characterized in that: The relationship function is an exponential function.

3. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 2 is implemented.

Images