A method for retrieving reflectivity of coral reef substrate by multispectral satellite remote sensing

Abstract

A kind of coral reef bottom reflectivity multispectral satellite remote sensing retrieval method belongs to satellite ocean remote sensing application technical field.Based on shallow sea single scattering radiation transfer model, through the collection of adjacent pixel pairs of different depth and different bottom type and the pixel of same bottom type (such as sandy seabed) of different depth, the optimal band rotation coefficient and the ratio of blue-green band diffuse attenuation coefficient are obtained respectively, combined with the optical deep water remote sensing reflectance data of the adjacent shallow sea area, the depth-independent reflectance equation and the bottom type-independent reflectance equation are constructed respectively, and based on the two equations, the coral reef bottom reflectivity image is obtained by joint calculation.It is suitable for high-resolution multispectral satellite remote sensing image, and auxiliary data such as water depth and water property are not required, and the blue-green band reflectivity data of coral reef bottom can be directly calculated and obtained.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of satellite ocean remote sensing application, and particularly relates to a method for retrieving bottom reflectance of coral reef by multi-spectral satellite remote sensing. BACKGROUND

[0002] The bottom reflectance of shallow sea is an important parameter for the optical remote sensing research of shallow sea benthic classification, coral reef and water color. At present, the most direct method for obtaining the bottom reflectance of coral reef is based on field spectrometer measurement, but the field spectral measurement can only obtain a small amount of point data, which is obviously insufficient for large-area coral reef monitoring. In contrast, the satellite remote sensing technology can obtain large-area coral reef information, but due to the influence of water properties and water depth, the remote sensing reflectance signal obtained by the satellite sensor contains signal interference from the water body, which leads to large uncertainty in the remote sensing retrieval of the bottom reflectance of coral reef. According to the single radiation transfer model of shallow sea, to obtain the bottom reflectance of shallow sea, the diffuse attenuation coefficient of shallow sea water body and water depth data are needed, but these two parameters are often difficult to obtain by large-area measurement, so the accurate monitoring of the bottom reflectance of coral reef by satellite remote sensing also faces great difficulties.

[0003] The existing compromise method is to parameterize and assume the bottom reflectance model (such as the mixed pixel model) in the process of shallow water depth semi-analysis remote sensing retrieval, so as to obtain the bottom reflectance information while realizing the remote sensing retrieval of shallow water depth. However, this method is more suitable for high-spectral satellite data with a large number of bands, but not suitable for multi-spectral satellite data with higher spatial resolution but fewer bands, and the accuracy of the retrieved reflectance is often not high due to the influence of the input of the bottom reflectance end member data. Therefore, it is necessary to develop a bottom reflectance remote sensing retrieval method suitable for multi-spectral satellite remote sensing data. SUMMARY

[0004] The present application aims to solve the problem that the existing method for retrieving the bottom reflectance of coral reef by remote sensing is difficult to be applied to multi-spectral satellite remote sensing data, and provides a method for retrieving the bottom reflectance of coral reef by multi-spectral satellite remote sensing, which is suitable for high-resolution multi-spectral satellite remote sensing images, does not need auxiliary data such as water depth and water properties, and can directly calculate and obtain the blue-green band reflectance data of coral reef bottom.

[0005] Based on the single scattering radiation transfer model of shallow sea, the optimal band rotation coefficient and the ratio of blue-green band diffuse attenuation coefficient are respectively calculated and obtained by collecting the adjacent pixel pairs of different depths and different bottom types and the pixels of the same bottom type (such as sandy seabed) at different depths, and the depth-independent reflectance equation and the bottom type-independent reflectance equation are respectively constructed by combining the remote sensing reflectance data of the optical deep water in the adjacent shallow sea area, and the blue and green reflectance images of coral reef bottom are obtained by joint calculation based on the two equations.

[0006] The present application comprises the following steps:

[0007] 1) According to the derivation result of the single radiation transmission process in shallow sea, a single-band logarithmic linear model is obtained through logarithmic conversion and simplification;

[0008] 2) For the blue and green bands, a depth-independent reflectance equation is constructed by eliminating the water depth using the single-band logarithmic linear model;

[0009] 3) Selecting adjacent pixel pairs of different depths and different substrates on the image, a set of optimal band rotation unit vectors is obtained by minimizing the Xi data set of the adjacent pixel pairs;

[0010] 4) Based on the optimal band rotation unit vector, a substrate type-independent reflectance equation is constructed;

[0011] 5) The blue and green band reflectances of the coral reef substrate are obtained by solving the depth-independent reflectance equation in step 2) and the substrate type-independent reflectance equation in step 4).

[0012] In step 1), the single-band logarithmic linear model is obtained by logarithmic conversion and simplification according to the derivation result of the single radiation transmission process in shallow sea as follows:

[0013]

[0014] wherein, g = k d +k, which is the sum of the diffuse attenuation coefficients of downward and upward light, r rs is the remote sensing reflectance of the shallow sea water body below the water surface, is the optical deep water value r rs immediately adjacent to the shallow sea area, b is the substrate reflectance, k d is the diffuse attenuation coefficient of the water body downward irradiance, and k is the upward radiance attenuation coefficient of the water body or seabed, and z is the water depth.

[0015] In step 2), the specific method of constructing a depth-independent reflectance equation can be:

[0016] The formula (1) single-band logarithmic linear model is calculated for the blue and green bands of the coral reef substrate, and by eliminating the water depth z, the equation can be obtained:

[0017]

[0018] In the formula, the subscripts 1 and 2 correspond to the blue and green bands used, g1 and g2 are the sum of the diffuse attenuation coefficients of downward and upward light for the blue and green bands, respectively, and g1 / g2 does not change with space assuming that the water body properties of the coral reef shallow sea area are uniform.

[0019] In step 3), the minimization of the Xi dataset for adjacent pixel pairs is solved as follows:

[0020]

[0021]

[0022] In the formula, i represents a pair of adjacent pixels, and Δsz i Let A and B represent the difference in values ​​of the i-th pixel pair after rotation, where A and B represent the different substrate types corresponding to the pixel pair, n is the number of pixel pairs, and f is the minimization function. The above formula yields a set of optimal band rotation unit vectors [α1, α2].

[0023] In step 4), the specific steps for constructing a substrate-type-independent reflectance equation can be as follows:

[0024] When the band rotation coefficient is optimal, calculations are performed for different seabed sediment types. The value should theoretically be consistent, defined as the substrate constant b, thereby obtaining the substrate-type-independent reflectance formula:

[0025]

[0026] In step 5), obtaining the blue and green band reflectance of the coral reef substrate includes:

[0027] (1) Solve for the green band substrate reflectance:

[0028] Combining formulas (2) and (4), the green band can be obtained.

[0029]

[0030] The above formula is divided by Apart from that, all other parameters can be obtained directly from the image; the green band substrate reflectance ρ can be calculated from the above formula. b2 :

[0031] (2) Solving for the blue band substrate reflectivity: The obtained green band... Substituting into formula (2), the blue band can be obtained. The blue band substrate reflectivity ρ can be further calculated. b1 :

[0032] This invention targets high-resolution multispectral satellite images and directly obtains the blue and green band reflectance data of coral reef substrate by solving equations without needing to acquire data on water depth and optical properties of the coral reef area.

[0033] Compared with the prior art, the present application has the following advantages:

[0034] 1) The present application is based on a shallow sea single scattering radiation transmission model, and the optimal wave band rotation coefficient and the blue-green wave band diffuse attenuation coefficient ratio are obtained by collecting adjacent image elements of different depths and different bottom types and the image elements of different depths of the same bottom type (such as a sandy seabed), and a depth-independent reflectivity equation and a bottom type-independent reflectivity equation are respectively constructed in combination with the optical deep water remote sensing reflectivity data of the adjacent shallow sea region, and the coral reef bottom reflectivity image is obtained based on the two sets of equations.

[0035] 2) Compared with the existing method for obtaining the coral reef bottom reflectivity based on the bottom reflectivity model parameterization and the shallow sea water depth semi-analysis inversion method, the present application is suitable for high-resolution multispectral satellite remote sensing images, and the auxiliary data such as water depth and water property are not required, and the blue-green wave band reflectivity data of the coral reef bottom can be directly calculated and obtained. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 is a blue-green wave band sandy seabed different depth position sampling point X1-X2 scatter plot and its linear fitting in the method embodiment of the present application;

[0037] Figure 2 is a coral reef bottom reflectivity satellite remote sensing inversion result graph in the method embodiment of the present application. Wherein, a is the blue wave band, and b is the green wave band. DETAILED DESCRIPTION

[0038] The present application is suitable for high-resolution multispectral satellite remote sensing images, and the related parameters required for the coral reef bottom reflectivity inversion are obtained through the image itself, without the auxiliary data such as water depth and water property, so that the coral reef bottom reflectivity remote sensing inversion is realized.

[0039] The technical solution adopted in the embodiment of the present application is:

[0040] 1) According to the derivation result of the shallow sea single radiation transmission process, the single wave band logarithmic linear model can be obtained through logarithmic conversion and simplification:

[0041]

[0042] Wherein, g=k d +k, that is, the sum of the diffuse attenuation coefficients of the downward and upward light, r rs is the remote sensing reflectivity below the water surface of the shallow sea water body, is the optical deep water r rs value adjacent to the shallow sea region, b is the bottom reflectivity, k dk is the water or seafloor upward radiance attenuation coefficient, and z is the water depth.

[0043] 2) Respectively, the formula (1) is calculated by using the single-band logarithmic linear model of the blue and green wave bands of the coral reef bottom. Through the elimination of water depth z, the equation can be obtained:

[0044]

[0045] In the formula, the subscripts 1 and 2 correspond to the blue and green wave bands used respectively, g1 and g2 are the sum of the downward and upward light diffusion attenuation coefficients of the blue and green wave bands respectively, and g1 / g2 does not change with space assuming that the water properties of the coral reef shallow sea area are uniform.

[0046] The above formula (2) constructs a depth-independent reflectance equation.

[0047] 3) Select adjacent pixel pairs of different depths and different substrates on the image, and obtain a set of optimal band rotation unit vectors [α1, α2] by minimizing the Xi data set of adjacent pixel pairs:

[0048]

[0049]

[0050] In the formula, i represents a certain adjacent pixel pair, Δsz i is the difference between the values after rotation of the i-th pixel pair, A and B represent different substrate types corresponding to the pixel pair, n is the number of pixel pairs, and f is the minimization function.

[0051] 4) When the band rotation coefficient is optimal, the values of calculated for different seafloor substrate types should theoretically be consistent, defined as the substrate constant b, thereby obtaining the substrate type-independent reflectance formula:

[0052]

[0053] 5) Joint formula (2) and (4), the green wave band

[0054]

[0055] The above formula, except , other parameters can be directly obtained from the image. Therefore, the green wave band substrate reflectivity ρ b2 can be obtained from the above formula:

[0056] 6) The green wave band obtained is brought into formula (2), and the blue wave band can be obtained Further, the reflectivity of the blue band substrate p can be obtained b1 :

[0057] The specific implementation process of the technical solutions of the present application will be described in detail below in combination with the drawings and a specific embodiment:

[0058] Step one: Obtain high-resolution multispectral satellite images of the study area, carry out atmospheric correction to obtain reflectivity images, and convert the reflectivity data into underwater remote sensing reflectivity data r w When the satellite image self-positioning error is low, geometric correction processing needs to be carried out first; when the image has obvious flare interference, flare correction processing also needs to be carried out;

[0059] Step two: Through visual interpretation, select deep water area pixels adjacent to the visible shallow water area of the seabed substrate, and calculate the average values of the blue and green bands According to the formula , respectively, to obtain the blue and green band X image data;

[0060] Step three: Select adjacent pixel point sets of Xi data of different substrate types at different distances along the coastline (representing different depths), and calculate the optimal band rotation coefficient [α1, α2] using the optimal algorithm (formula 2);

[0061] Step four: Select X1~X2 data of the same seabed substrate (such as sandy) at multiple different depth positions on the image, and use the least squares method to establish a linear regression formula (such as Figure 1 ) between the two, to obtain the ratio g1 / g2 of the sum of the upward and downward diffuse attenuation coefficients of the blue and green bands;

[0062] Step five: Combine the near-infrared band image, visually interpret the water edge line position, and select a certain sample amount of blue and green band pixel values at the typical water edge line position, calculate the value, and obtain the substrate constant b value.

[0063] Step six: Apply the calculated α1, α2, g1 / g2 and substrate constant b of the coral reef area to the entire image, and apply them to formula 4 and formula 5, respectively, to obtain the remote sensing inversion results of the blue and green band coral reef substrate rate of the entire image (such as Figure 2 ).

[0064] The above only describes the embodiments of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent process transformation using the content of the specification and drawings, or direct or indirect application in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for retrieving the reflectivity of coral reef bottom by multispectral satellite remote sensing, characterized in that The method comprises the following steps: 1) According to the derivation result of the single radiation transmission process in shallow sea, a single-band logarithmic linear model is obtained through logarithmic conversion and simplification as follows: wherein, g = k d +k, the sum of the diffuse attenuation coefficients of downwelling and upwelling light, r rs is the remote sensing reflectance of the shallow water body below the water surface, is the optical depth of the water r rs value immediately adjacent to the shallow water region, b is the bottom reflectance, k d is the diffuse attenuation coefficient of the water body for downwelling irradiance, k is the attenuation coefficient of the water body or seafloor for upwelling radiance, and z is the water depth. 2) For the blue and green bands, a depth-independent reflectance equation is constructed by eliminating water depth by using the single-band logarithmic linear model; The specific method of constructing a depth-independent reflectance equation is: The single-band logarithmic linear model of formula (1) is calculated for the blue and green bands of the coral reef substrate respectively, and through eliminating water depth z, the equation is obtained: In the formula, the subscripts 1 and 2 correspond to the blue and green bands used respectively, g1 and g2 are the total sum of the downward and upward light diffuse attenuation coefficients of the blue and green bands respectively, and it is assumed that the water properties of the coral reef shallow sea area are uniform, so g1 / g2 does not change with space; 3) Selecting adjacent pixel pairs of different depths and different substrates on the image, a set of optimal band rotation unit vectors is obtained by minimizing the Xi data set of the adjacent pixel pairs; 4) Based on the optimal band rotation unit vector, a substrate type-independent reflectance equation is constructed; 5) Jointing the depth-independent reflectance equation in step 2) and the substrate type-independent reflectance equation in step 4), the blue and green band reflectances of the coral reef substrate are solved; The solving of the blue and green band reflectances of the coral reef substrate comprises: (1) Solving the green band substrate reflectance: Combining equation (2) and equation (4), the green wave band is obtained The other parameters can be obtained directly from the image except for the green band bottom-of-atmosphere reflectance p which is obtained from the equation b2 : (2) Obtain the reflectivity of the bottom material in the blue band: put the obtained reflectivity of the bottom material in the green band into formula (2), that is, obtain the reflectivity of the bottom material in the blue band Further obtain the reflectivity of the bottom material in the blue band b1 :

2. The method according to claim 1, wherein the method is characterized by In step 3), the Xi data set of the adjacent pixel pairs is minimized as follows: where i represents a certain adjacent pixel pair, Δsz i is the difference of the value of the i-th pixel pair after rotation, A and B represent different underlying types corresponding to the pixel pair, n is the number of pixel pairs, and f is a minimization function. A set of optimal band rotation unit vectors [α1, α2] is obtained through the above formula.

3. The method according to claim 1, wherein the method is characterized by In step 4), the specific steps of constructing a substrate type-independent reflectance equation are: When the band rotation coefficient is optimal, calculations are performed for different seabed sediment types. The value should theoretically be consistent, defined as the substrate constant b, thereby obtaining the substrate-type-independent reflectance formula:

Images