Shallow sea water depth hyperspectral inversion method considering substrate characteristics

Abstract

The invention relates to a shallow sea water depth hyperspectral inversion method considering sediment characteristics. The method comprises the following steps: combining a hyperspectral remote sensing image with water depth data in satellite-borne laser radar data to obtain a shallow sea hyperspectral remote sensing image; carrying out spectral data sampling on the substrate at the waterline of the shallow sea hyperspectral remote sensing image, averaging the spectral data of the same kind of substrates, and then assigning the average value to the spatial distribution data of the substrate to obtain a substrate spectral image; a shallow sea hyperspectral remote sensing image and a substrate spectral image of a satellite-borne laser radar data coverage area are used as input data of a training sample, water depth data in satellite-borne laser radar data are used as output data of the training sample, and a shallow sea water depth inversion model is trained; and carrying out shallow sea water depth inversion by using the trained shallow sea water depth inversion model. The shallow sea sediment heterogeneous interference is weakened by fully excavating and utilizing the sediment characteristics, and a higher-precision method is provided for shallow sea water depth remote sensing inversion.

Description

Technical Field

[0001] The present invention relates to the field of shallow sea water depth inversion methods, and in particular to a shallow sea water depth hyperspectral inversion method taking into account bottom characteristics. Background Art

[0002] Shallow water depth data plays an important role in the study of oceans and coastal areas, and is widely used in marine ecosystem protection, coastline monitoring, channel management, and seafloor topography mapping. Traditional depth measurement methods usually rely on ship-borne echo sounders and airborne lidar technology (Pacheco, Horta et al. 2015, Kerr and Purkis 2018, Li, Tang et al. 2023). Echo sounders can provide high-precision depth data, but their data acquisition range is limited, especially in large areas of shallow waters. The efficiency is low (Chu, Cheng et al. 2023). Although LiDAR technology can obtain depth data with high spatial resolution, it is expensive (Stumpf, Holderied et al. 2003). These methods are generally not suitable for large-scale, cost-effective depth measurements, especially in remote or difficult-to-reach areas. Compared with the above methods, satellite-derived bathymetry (SDB) is not geographically restricted, has low cost and high efficiency, and is an effective method for large-scale bathymetry (Lyzenga, Malinas et al. 2006).

[0003] Water depth inversion algorithms include empirical models, semi-theoretical and semi-empirical models, and theoretical models (Chu, Cheng et al. 2023). Among them, empirical models and semi-theoretical and semi-empirical models are widely used in water depth inversion of different types of water bodies because of their relatively simple calculation process and high accuracy (Nguyen, Liquet et al. 2021). In recent years, with the improvement of remote sensing technology and computing power, especially the development of machine learning algorithms, these data-driven methods have been increasingly widely used in water depth inversion (Nagamani, Chauhan et al. 2012). Compared with traditional models, machine learning algorithms can better capture the complex nonlinear relationship between water body reflection signals and water depth, thereby improving the inversion accuracy and robustness (Wu, Mao et al. 2022, Chu, Cheng et al. 2023, Li, Tang et al. 2023).

[0004] Although SDB depth inversion technology has made significant progress, its inversion accuracy is often affected by the characteristics of the bottom sediments (Philpot 1989, Stumpf, Holderied et al. 2003, McKinna, Fearns et al. 2015). When the water depth is fixed, the bottom sediment type affects the signal strength received by the remote sensing sensor, which in turn affects the results of remote sensing depth inversion (Reichstetter, Fearns et al. 2015, Arabi, Salama et al. 2020). Studies have shown that by dividing the study area into multiple sub-regions with single and uniform bottom sediments through bottom sediment classification, and constructing depth models in each type of bottom sediment area, the influence of bottom sediment factors on the depth model can be weakened to a certain extent, thereby improving the accuracy of depth inversion (Nagamani, Chauhan et al. 2012, Ma, Tao et al. 2014, Cheng, Ma et al. 2021).

[0005] Current studies have shown that the introduction of sediment spatial characteristics can improve the accuracy of water depth inversion, but there is little research on another important feature of sediment, namely spectral characteristics. Sediment spectral information can effectively reflect the physical and chemical properties of sediment, and its performance in water bodies will undergo complex interactions with changes in water depth and water optical properties (Hochberg 2003, Jay, Guillaume et al. 2017, Liu, Deng et al. 2018, Arabi, Salama et al. 2020), affecting the water depth inversion results.

[0006] There is currently a lack of effective solutions on how to mine and integrate the spatial and spectral characteristics of the seabed to further reduce the interference of seabed heterogeneity on water depth remote sensing inversion and improve inversion accuracy. Summary of the invention

[0007] In view of the problems in the related art, the present invention proposes a shallow sea water depth hyperspectral inversion method taking into account the bottom characteristics to overcome the above technical problems existing in the existing related art.

[0008] To this end, the specific technical solution adopted by the present invention is as follows: a shallow sea water depth hyperspectral inversion method taking into account the bottom characteristics comprises the following steps:

[0009] S1. Obtain hyperspectral remote sensing images and satellite-borne laser radar data and perform preprocessing, extract seawater pixels in the hyperspectral remote sensing images, and combine with the water depth data in the satellite-borne laser radar data to remove the pixels with water depth greater than the water depth threshold H in the seawater pixels, and obtain shallow sea hyperspectral remote sensing images;

[0010] S2. Obtain spatial distribution data of the bottom sediment based on supervised classification of shallow sea hyperspectral remote sensing images, sample spectral data of the bottom sediment at the water edge of the shallow sea hyperspectral remote sensing images and average the spectral data of the same type of bottom sediments to obtain the spectrum of each type of bottom sediment, and assign the spectrum of the bottom sediment to the spatial distribution data to obtain a bottom sediment spectral image, thereby realizing bottom sediment spectral compounding, wherein the bottom sediment includes at least three types: coral, coral sand, and debris;

[0011] S3, establishing a shallow sea water depth inversion model, wherein the input data of the training samples include shallow sea hyperspectral remote sensing images and bottom spectral images of the area covered by the satellite-borne laser radar data, and the output data of the model training samples are the water depth data in the satellite-borne laser radar data; the model is trained using the training samples; the shallow sea hyperspectral remote sensing images obtained in step S1 and the bottom spectral images obtained in step S2 are input into the trained shallow sea water depth inversion model, and the shallow sea water depth inversion results are output.

[0012] The present invention connects the spectral data of the bottom type with the shallow sea hyperspectral remote sensing image as the input of the shallow sea water depth inversion model. The present invention fully explores and utilizes the bottom characteristics to weaken the heterogeneous interference of the seabed bottom, and provides a higher precision method for the remote sensing inversion of shallow sea water depth.

[0013] Furthermore, step S1 specifically includes the following steps:

[0014] S11. Perform geometric correction, atmospheric correction and flare correction on the ZY-01 hyperspectral remote sensing image; perform mask processing on the cloud and land parts of the remote sensing image in the area, and retain the sea water pixels;

[0015] S12, extract the sea surface and seabed photon point cloud data from the ICESat-2 ATL03 lidar data respectively, and perform refraction correction on the seabed photon point cloud data. Obtain the tidal height when the ZY-01 hyperspectral remote sensing image and ICESat-2 lidar data were collected from the tidal values ​​provided by the tide gauge station near the study area, and carry out tidal correction;

[0016] S13. Geo-register the point cloud water depth data with the hyperspectral image data of the study area, and extract the range where the water depth is greater than the water depth threshold H from the point cloud data, and define the image pixels corresponding to the range as deep-sea pixels. Extract the spectral characteristics of the deep-sea pixels, and segment the hyperspectral image of the study area based on the spectral characteristics of the deep-sea pixels to obtain deep-sea hyperspectral remote sensing images and shallow-sea hyperspectral remote sensing images.

[0017] Furthermore, step S2 specifically includes the following steps:

[0018] S21, taking the preprocessed shallow sea hyperspectral remote sensing image as input;

[0019] S22. Construct sample criteria for bottom sediment classification, collect samples of different bottom sediments in the study area, and the bottom sediments include at least three categories: coral, coral sand, and debris. Use the maximum likelihood method to conduct supervised classification in the study area to obtain the spatial distribution data of the bottom sediments.

[0020] S23, sampling spectral data of the bottom material near the water edge of the shallow sea hyperspectral remote sensing image and averaging the spectral data of the same type of bottom material to obtain the spectrum of each type of bottom material;

[0021] S24. Based on the sediment spatial distribution data and the sediment water spectral data, the sediment spectrum is added to the corresponding sediment spatial distribution data to obtain a spectral image of the sediment, thereby realizing the composite of the partitioned sediment spectrum.

[0022] The shallow sea water depth inversion model of the present invention is any machine learning model.

[0023] Preferably, the shallow sea water depth inversion model uses a LGBM (Light Gradient Boosting Machine) water depth inversion model or a RF (Random Forest) water depth inversion model.

[0024] Among them, LGBM is an efficient machine learning framework based on the gradient boosted tree (GBT) algorithm. Developed by Microsoft Research Asia, it is particularly suitable for processing large data sets and high-dimensional features. LGBM uses a histogram-based decision tree algorithm and a depth-constrained leaf growth strategy to significantly speed up training and reduce memory consumption. By integrating multiple decision trees, LGBM is able to capture the complex nonlinear relationship between water depth and hyperspectral features, thereby providing more accurate depth estimates.

[0025] Among them, RF is an ensemble learning algorithm based on the decision tree method. It builds many decision trees during the training process and outputs the mode of their predictions for classification tasks and the average value for regression tasks. By using bootstrap sampling and random feature selection, RF reduces the risk of overfitting and enhances the generalization ability of the model.

[0026] According to another aspect of the present invention, a system for implementing the above-mentioned shallow sea water depth hyperspectral inversion method taking into account the bottom characteristics is provided, the system comprising: a data preprocessing module, a bottom space and spectral feature extraction module and an algorithm establishment module;

[0027] The data preprocessing module is used to obtain and preprocess hyperspectral remote sensing images and satellite-borne laser radar data, extract seawater pixels in the hyperspectral remote sensing images, and combine the water depth data in the satellite-borne laser radar data to remove the pixels with water depth greater than the water depth threshold H in the seawater pixels, so as to obtain shallow sea hyperspectral remote sensing images;

[0028] The bottom sediment spatial and spectral feature extraction module obtains the spatial distribution data of the bottom sediment based on the supervised classification of shallow sea hyperspectral remote sensing images, samples the spectral data of the bottom sediment at the water edge of the shallow sea hyperspectral remote sensing images and averages the spectral data of the same type of bottom sediment to obtain the spectrum of each type of bottom sediment, and assigns the spectrum of the bottom sediment to the spatial distribution data to obtain the bottom sediment spectral image, realizing the bottom sediment spectral compound, where the bottom sediment includes at least three types: coral, coral sand, and debris;

[0029] The algorithm establishment module is used to establish a shallow sea water depth inversion model. The input data of the training samples include shallow sea hyperspectral remote sensing images and bottom spectral images of the area covered by the satellite-borne laser radar data. The output data of the model training samples are the water depth data in the satellite-borne laser radar data. The model is trained using the training samples. The shallow sea hyperspectral remote sensing image obtained in step S1 and the bottom spectral image obtained in step S2 are input into the trained shallow sea water depth inversion model, and the shallow sea water depth inversion result is output.

[0030] The beneficial effects of the present invention are:

[0031] 1. The present invention proposes a shallow sea water depth hyperspectral inversion that takes into account the bottom characteristics, fully explores and utilizes the bottom characteristics to weaken the heterogeneous interference of the seabed, and provides a higher precision method for shallow sea water depth remote sensing inversion.

[0032] 2. The present invention proposes a strategy for extracting the spatial characteristics of the bottom sediment based on supervised classification of remote sensing images and extracting the spectral characteristics of the bottom sediment based on the water edge of the image, thereby improving the accuracy of water depth inversion.

[0033] 3. The present invention is based on the commonly used LGBM machine learning algorithm and RF machine learning algorithm. The algorithm usage scenarios are consistent with the classic algorithms, which is conducive to the promotion and application of the method of the present invention and provides technical support for applications such as island and reef construction, navigation safety, and ecological protection. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1 is a flow chart of shallow sea water depth hyperspectral inversion taking into account bottom characteristics according to an embodiment of the present invention;

[0036] Figure 2 It is an overview map of the study area in the example;

[0037] Figure 3It is the ICESat-2 point cloud trajectory map of the study area in the example;

[0038] Figure 4 This is the bottom sediment classification result map of the study area in the example;

[0039] Figure 5 This is the inversion result map of shallow sea water depth in the study area in the example;

[0041] Figure 6 4 is a system block diagram of a shallow sea water depth hyperspectral inversion system taking into account bottom characteristics according to an embodiment of the present invention.

[0042] In the figure: 1. Data preprocessing module; 2. Sediment spatial and spectral feature extraction module; 3. Algorithm establishment module. DETAILED DESCRIPTION

[0043] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in the field should be able to understand other possible implementations and advantages of the present invention. The components in the figures are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0044] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1 The flowchart of the shallow sea water depth hyperspectral inversion method taking into account the bottom characteristics of the present invention is shown, and the method comprises the following steps:

[0045] S1. Obtain hyperspectral remote sensing images and satellite-borne lidar data and perform preprocessing to extract seawater pixels in the hyperspectral remote sensing images. Combined with the water depth data in the satellite-borne lidar data, remove the pixels with water depth greater than the water depth threshold H in the seawater pixels to obtain shallow sea hyperspectral remote sensing images.

[0046] This example uses a certain sea area and reef as the experimental area (such as Figure 2 The remote sensing image data is the ZY-01 hyperspectral satellite image, which was acquired at 17:04 (UTC) on January 13, 2023. The wavelength of ZY-01 is 450nm-700nm, with a total of 36 bands and a spatial resolution of 30m. The water depth data is ICESat-2 ATL03. Refraction correction is performed on the extracted effective seabed photon points, and the reference surface of the point cloud water depth data is converted to the local mean sea level (such as Figure 3 shown). Figure 3 In the figure, the black straight line segment is the coverage area of ​​the satellite-borne lidar data. It can be seen that the water depth data in the point cloud data is local data and cannot cover the entire test area.

[0047] This step S1 specifically includes the following steps:

[0048] S11. Use professional remote sensing processing software ENVI to perform geometric correction, atmospheric correction and flare correction on ZY-01 hyperspectral remote sensing images; and perform mask processing on the cloud and land parts of the remote sensing images in the area to retain seawater pixels;

[0049] S12. Use PhotonLabeler software to extract sea surface and seabed photon point cloud data from ICESat-2 ATL03 lidar data, perform refraction correction on the seabed photon point cloud data, obtain the tidal height when ZY-01 hyperspectral remote sensing images and ICESat-2 lidar data were collected from the tidal values ​​provided by the tide gauge station near the study area, and carry out tidal correction.

[0050] S13, geo-register the point cloud water depth data with the hyperspectral image data of the study area, extract the range with a water depth greater than 20m from the point cloud data, and define the image pixels corresponding to the range as deep-sea pixels, extract the spectral features of the deep-sea pixels, and segment the hyperspectral image of the study area based on the spectral features of the deep-sea pixels to obtain deep-sea hyperspectral remote sensing images and shallow-sea hyperspectral remote sensing images. In addition, the deep-sea range can be obtained through nautical chart data, and the shallow-sea hyperspectral remote sensing images can be obtained after masking the hyperspectral image data.

[0051] S2, based on the supervised classification of shallow sea hyperspectral remote sensing images, obtain the spatial distribution data of the bottom, sample the spectral data of the bottom at the water edge of the shallow sea hyperspectral remote sensing image and average the spectral data of the same bottom, obtain the spectrum of each type of bottom, and assign the spectrum of the bottom to the spatial distribution data to obtain the bottom spectrum image, realize the bottom spectrum composite, wherein the bottom includes at least three categories: coral, coral sand, and debris. Step S2 specifically includes the following steps:

[0052] S21, taking the preprocessed hyperspectral remote sensing image as input;

[0053] S22, constructing a sample criterion for bottom sediment classification, collecting samples of different bottom sediments in the study area. In this embodiment, the bottom sediments include three categories: coral, coral sand, and debris. The maximum likelihood method is used to conduct supervised classification on the study area to obtain bottom sediment spatial distribution data;

[0054] S23, sampling spectral data of the bottom material near the water edge of the shallow sea hyperspectral remote sensing image and averaging the spectral data of the same type of bottom material to obtain the spectrum of each type of bottom material;

[0055] S24. Based on the sediment spatial distribution data and the sediment water spectral data, the sediment spectrum is added to the corresponding sediment spatial distribution data to obtain a spectral image of the sediment, thereby realizing the composite of the partitioned sediment spectrum.

[0056] After checking the tide table, the tidal height at the time of acquiring the image of the islands and reefs in the test area was 0.9m (refer to the nearby tide gauge station). The water depth of the point cloud of the test area has been corrected to the instantaneous sea surface water depth at the time of image acquisition, and the 0-20m water depth is extracted as the experimental water depth data of the present invention. The water depth uses the mean water depth of the ZY-01 image pixel, that is, the average of the measured water depth data within a single pixel is used as the water depth value of the pixel.

[0057] In this example, three typical types of bottom samples were collected in the study area, including coral, coral sand, and debris. The maximum likelihood method was used to classify the bottom of the islands and reefs. The classification results are as follows: Figure 4 As shown in Figure 1, the land area has been masked.

[0058] S3, establishing a shallow sea water depth inversion model, wherein the input data of the training samples include shallow sea hyperspectral remote sensing images and bottom spectral images of the area covered by the satellite-borne laser radar data, and the output data of the model training samples are the water depth data in the satellite-borne laser radar data; the model is trained using the training samples; the shallow sea hyperspectral remote sensing images obtained in step S1 and the bottom spectral images obtained in step S2 are input into the trained shallow sea water depth inversion model, and the shallow sea water depth inversion results are output.

[0059] Wherein, the shallow sea water depth inversion model can be any machine learning model. The samples can be divided into training samples and test samples according to the proportion, the training samples are used to train the model, and the test samples are used to evaluate the trained model. In this embodiment, the LGBM water depth inversion model taking into account the bottom characteristics and the RF water depth inversion model taking into account the bottom characteristics are respectively selected for learning and classification.

[0060] In specific applications, the present invention uses the root mean square error (RMSE), the mean absolute error (MAE) and the coefficient of determination (R 2 ) quantitatively evaluates the accuracy of the water depth inversion results output by the water depth inversion model. The smaller the RMSE and MAE, the higher the accuracy of the water depth inversion. The calculation formula of the quantitative index is as follows:

[0061]

[0062] n represents the number of water depth points; H estimation is the inverted water depth; H lidar is the water depth extracted from ICESat-2; and They are the average water depth extracted by ICESat-2 and the average water depth inverted.

[0063] Comparing the water depth inversion results of the classic method and the method proposed by the invention that takes into account the bottom characteristics, each method includes two sets of results of the LGBM algorithm and the RF algorithm. The LGBM algorithm that takes into account the bottom characteristics is as follows: Figure 5 As shown in (a), the RF algorithm taking into account the bottom characteristics is as follows Figure 5 As shown in (b), the reef outline is clear, the underwater terrain features are smooth, and there is less noise. The water depth inversion accuracy is shown in Table 1:

[0064] Table 1 Shallow sea depth inversion accuracy evaluation table

[0065]

[0066] From Table 1, compared with the classic LGBM and RF algorithms, the LGBM and RF algorithms that take into account the bottom characteristics have the lowest RMSE and MAE errors, and the correlation coefficient R 2 At the very least, the hyperspectral water depth inversion method proposed in the present invention which takes into account the bottom characteristics has excellent water depth inversion performance.

[0067] As can be seen from the table, the water depth inversion results of the classic LGBM algorithm are RMSE, MAE, R 2 They are 1.30m, 0.67m and 0.83 respectively.

[0068] The RMSE, MAE, and R of the water depth inversion results of the LGBM algorithm considering the bottom characteristics 2 They are 1.05m, 0.55m and 0.89 respectively.

[0069] The water depth inversion results of the classic RF algorithm are RMSE, MAE, and R 2 They are 1.29m, 0.64m and 0.83 respectively.

[0070] The RMSE, MAE, and R of the water depth inversion results of the RF algorithm considering the bottom characteristics 2 They are 1.05m, 0.52m and 0.89 respectively.

[0071] According to another embodiment of the present invention, a shallow sea water depth hyperspectral inversion system taking into account bottom characteristics is provided, the system comprising: a data preprocessing module 1, a bottom space and spectral feature extraction module 2 and an algorithm establishment module 3;

[0072] The data preprocessing module 1 is used to obtain and preprocess the hyperspectral remote sensing image and satellite-borne laser radar data, extract the seawater pixels in the hyperspectral remote sensing image, and combine the water depth data in the satellite-borne laser radar data to remove the pixels whose water depth is greater than the water depth threshold H in the seawater pixels, so as to obtain the shallow sea hyperspectral remote sensing image;

[0073] The bottom sediment spatial and spectral feature extraction module 2 obtains the spatial distribution data of the bottom sediment based on the supervised classification of the shallow sea hyperspectral remote sensing image, samples the spectral data of the bottom sediment at the water edge of the shallow sea hyperspectral remote sensing image and averages the spectral data of the same type of bottom sediment to obtain the spectrum of each type of bottom sediment, and assigns the spectrum of the bottom sediment to the spatial distribution data to obtain the bottom sediment spectral image, and realizes the bottom sediment spectral compound, wherein the bottom sediment includes at least three types: coral, coral sand, and debris;

[0074] The algorithm establishes module 3, which is used to establish a shallow sea water depth inversion model. The input data of the training samples include shallow sea hyperspectral remote sensing images and bottom spectral images of the area covered by the satellite-borne laser radar data. The output data of the model training samples are the water depth data in the satellite-borne laser radar data. The model is trained using the training samples. The shallow sea hyperspectral remote sensing image obtained in step S1 and the bottom spectral image obtained in step S2 are input into the trained shallow sea water depth inversion model, and the shallow sea water depth inversion result is output.

[0075] In summary, with the help of the above technical scheme of the present invention, the present invention proposes a shallow sea water depth hyperspectral inversion method taking into account the bottom characteristics, fully explores and utilizes the bottom characteristics to weaken the heterogeneous interference of the seabed bottom, and provides a higher precision method for shallow sea water depth remote sensing inversion; the present invention proposes a strategy for extracting bottom spatial characteristics based on supervised classification of remote sensing images and extracting bottom spectral characteristics based on image water edges, which explores and utilizes the bottom spatial characteristics and bottom spectral characteristics, and improves the depth inversion accuracy; the present invention is based on the commonly used LGBM machine learning algorithm and the RF machine learning algorithm, and the algorithm usage scenario is consistent with the classical algorithm, which is conducive to the promotion and application of the method of the present invention, and provides technical support for applications such as island and reef construction, navigation safety, and ecological protection.

[0076] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A shallow sea water depth hyperspectral inversion method taking into account bottom characteristics, comprising the following steps: S1. Obtain hyperspectral remote sensing images and satellite-borne laser radar data and perform preprocessing, extract seawater pixels in the hyperspectral remote sensing images, and combine with the water depth data in the satellite-borne laser radar data to remove the pixels with water depth greater than the water depth threshold H in the seawater pixels, and obtain shallow sea hyperspectral remote sensing images; S2. Obtain spatial distribution data of the bottom sediment based on supervised classification of shallow sea hyperspectral remote sensing images, sample spectral data of the bottom sediment at the water edge of the shallow sea hyperspectral remote sensing images and average the spectral data of the same type of bottom sediments to obtain the spectrum of each type of bottom sediment, and assign the spectrum of the bottom sediment to the spatial distribution data to obtain a bottom sediment spectral image, thereby realizing bottom sediment spectral compounding, wherein the bottom sediment includes at least three types: coral, coral sand, and debris; S3, establishing a shallow sea water depth inversion model, wherein the input data of the training samples include shallow sea hyperspectral remote sensing images and bottom spectral images of the area covered by the satellite-borne laser radar data, and the output data of the model training samples are the water depth data in the satellite-borne laser radar data; the model is trained using the training samples; the shallow sea hyperspectral remote sensing images obtained in step S1 and the bottom spectral images obtained in step S2 are input into the trained shallow sea water depth inversion model, and the shallow sea water depth inversion results are output.

2. The method for shallow sea water depth hyperspectral inversion taking into account bottom characteristics according to claim 1 is characterized in that: Step S1 specifically includes the following steps: S11. Perform geometric correction, atmospheric correction and flare correction on the ZY-01 hyperspectral remote sensing image; perform mask processing on the cloud and land parts of the remote sensing image in the area, and retain the sea water pixels; S12, extract the sea surface and seabed photon point cloud data from the ICESat-2 ATL03 lidar data respectively, and perform refraction correction on the seabed photon point cloud data. Obtain the tidal height when the ZY-01 hyperspectral remote sensing image and ICESat-2 lidar data were collected from the tidal values ​​provided by the tide gauge station near the study area, and carry out tidal correction; S13. Geo-register the point cloud water depth data with the hyperspectral image data of the study area, and extract the range where the water depth is greater than the water depth threshold H from the point cloud data, and define the image pixels corresponding to the range as deep-sea pixels. Extract the spectral characteristics of the deep-sea pixels, and segment the hyperspectral image of the study area based on the spectral characteristics of the deep-sea pixels to obtain deep-sea hyperspectral remote sensing images and shallow-sea hyperspectral remote sensing images.

3. The method for shallow sea water depth hyperspectral inversion taking into account bottom characteristics according to claim 1 is characterized in that: Step S2 specifically includes the following steps: S21, taking the preprocessed shallow sea hyperspectral remote sensing image as input; S22. Construct sample criteria for bottom sediment classification, collect samples of different bottom sediments in the study area, and the bottom sediments include at least three categories: coral, coral sand, and debris. Use the maximum likelihood method to conduct supervised classification in the study area to obtain the spatial distribution data of the bottom sediments. S23, sampling spectral data of the bottom material near the water edge of the shallow sea hyperspectral remote sensing image and averaging the spectral data of the same type of bottom material to obtain the spectrum of each type of bottom material; S24. Based on the sediment spatial distribution data and the sediment water spectral data, the sediment spectrum is added to the corresponding sediment spatial distribution data to obtain a spectral image of the sediment, thereby realizing the composite of the partitioned sediment spectrum.

4. The method for shallow sea water depth hyperspectral inversion taking into account bottom characteristics according to claim 1 is characterized in that: The shallow water depth inversion model is any machine learning model.

5. The method for shallow sea water depth hyperspectral inversion taking into account bottom characteristics according to claim 1 is characterized in that: The shallow sea water depth inversion model is a LGBM water depth inversion model or a RF water depth inversion model.

6. The method for shallow sea water depth hyperspectral inversion taking into account bottom characteristics according to claim 1 is characterized in that: The water depth threshold H = 20 meters.

7. A shallow sea water depth hyperspectral inversion system taking into account bottom characteristics, used to implement the shallow sea water depth hyperspectral inversion method taking into account bottom characteristics as described in any one of claims 1 to 6, characterized in that: The system comprises: a data preprocessing module (1), a bottom sediment space and spectral feature extraction module (2) and an algorithm establishment module (3); The data preprocessing module (1) is used to obtain and preprocess the hyperspectral remote sensing image and the satellite-borne laser radar data, extract the seawater pixels in the hyperspectral remote sensing image, and combine the water depth data in the satellite-borne laser radar data to remove the pixels whose water depth is greater than the water depth threshold H in the seawater pixels, so as to obtain the shallow sea hyperspectral remote sensing image; A bottom sediment spatial and spectral feature extraction module (2) is used to obtain spatial distribution data of the bottom sediment based on supervised classification of shallow sea hyperspectral remote sensing images, sample spectral data of the bottom sediment at the water edge of the shallow sea hyperspectral remote sensing images and average the spectral data of the same type of bottom sediments to obtain the spectrum of each type of bottom sediment, and assign the spectrum of the bottom sediment to the spatial distribution data to obtain a bottom sediment spectral image, thereby realizing bottom sediment spectral compounding, wherein the bottom sediment includes at least three types: coral, coral sand, and debris; The algorithm establishment module (3) is used to establish a shallow sea water depth inversion model, wherein the input data of the training samples include shallow sea hyperspectral remote sensing images and bottom spectral images of the area covered by the satellite-borne laser radar data, and the output data of the model training samples are the water depth data in the satellite-borne laser radar data; the model is trained using the training samples; the shallow sea hyperspectral remote sensing images obtained in step S1 and the bottom spectral images obtained in step S2 are input into the trained shallow sea water depth inversion model, and the shallow sea water depth inversion results are output.

Images