Coastal aquaculture classification method based on spectrum time sequence characteristics
By combining dual-spectral index synergistic response and temporal noise suppression processing with the XGBoost algorithm and multi-index fusion features, high-precision classification of kelp, seaweed, and cage aquaculture types was achieved. This solved the problem of difficulty in distinguishing aquaculture types with similar spectral features in existing technologies, and improved the stability and reliability of classification.
Patent Information
- Application Number
- CN202511095023.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-06
- Publication Date
- 2025-11-18
AI Technical Summary
Existing technologies struggle to effectively distinguish between aquaculture types with similar spectral characteristics, such as kelp, seaweed, and net cages, and lack a reflection of the dynamic changes in aquaculture types, making precise classification difficult.
A dual-spectral index collaborative response mechanism is adopted, combined with temporal noise suppression and model-feature adaptation optimization. The characteristic spectral responses of algae and aquaculture water are captured by NDAI and NDAWI indices. Savitzky-Golay filtering is used to suppress noise, and the XGBoost algorithm is used to classify the features fused with multiple indices.
It significantly improved the accuracy of distinguishing between kelp, asparagus, and cage aquaculture types, ensuring the stability and reliability of classification results and supporting the supervision of aquaculture resources and ecological environment assessment.
Smart Images
Figure CN120976754A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The application belongs to the technical field of remote sensing data processing and aquaculture classification, and particularly relates to a coastal aquaculture classification method based on spectral time sequence characteristics. BACKGROUND
[0002] In recent years, the aquaculture industry has developed rapidly, and the aquaculture mode has gradually transformed towards intensification and modernization, which not only guarantees the supply of aquatic products, but also effectively promotes the adjustment of fishery structure and the income increase of farmers in coastal areas. However, the rapidly expanding aquaculture industry also faces severe ecological challenges. Due to unreasonable layout and excessive development intensity, some sea areas have problems such as overcapacity, water eutrophication, and degradation of bottom sediment, which has caused great pressure on the coastal ecosystem. Under the background of marine ecological protection and coordinated development of the fishery industry, scientifically grasping the spatial distribution and structural characteristics of aquaculture, especially accurately identifying the spatial distribution of different types of aquaculture, is of great significance for aquaculture supervision, resource management, and ecological red line demarcation. Based on the need for high-quality development of local marine fishery economy, aquaculture type information is urgently needed as a scientific basis for marine planning.
[0003] Traditional aquaculture information acquisition relies on manual investigation and on-site verification, which has problems such as high cost, long cycle, limited coverage, and is difficult to meet the needs of fine management and dynamic monitoring. Current researches mostly focus on the boundary extraction and pattern recognition of aquaculture areas, and pay less attention to the fine classification of specific aquaculture categories. For example, Wu Yunren et al. proposed a method combining spatial transformation and permutation attention mechanism to realize the extraction of pond culture and net cage culture; Wang Fang et al. extracted the spatial distribution of floating raft and hanging culture in Zhelin Bay in eastern Guangdong based on Gaofen-1 satellite images using the Apriori algorithm; Yu Wenxiao et al. achieved remote sensing recognition of pond culture and net cage culture using object-oriented classification method; Liu et al. achieved global mapping of raft culture and net cage culture based on multi-source remote sensing time series. These studies have made significant progress in aquaculture mode recognition and research scope, but most of them rely on single image, which is difficult to reflect the time variation characteristics in the growth process of aquaculture, and lack of differentiation of aquaculture types.
[0004] Time series remote sensing images can represent the dynamic changes of ground objects in the growth cycle, and have been widely used in agricultural crop recognition tasks, and have shown good potential in aquaculture classification [5] . Dong Xiuchun et al. constructed a time sequence feature dataset of typical features of rice-shrimp fields using Sentinel-1 backscattering coefficients, and achieved high-precision extraction of rice-shrimp fields, conventional rice fields, and lotus fields. Li Kaiqiang The spectral response and temporal variation of different types of seawater used for aquaculture were analyzed based on multi-source remote sensing data, enabling the classification of algae, shellfish, and ordinary seawater. Chen Zhiyang et al. Based on Google Earth Engine and Sentinel-2 dense time series images, a pond aquaculture extraction and change monitoring method combining K-means clustering and hierarchical decision tree classification algorithm was proposed. Zhang et al. Based on measured data, a normalized aquaculture water index (NDAWI) was proposed, a NDAWI time series dataset was constructed, and a random forest method was used to distinguish kelp and undaria pinnatifida aquaculture water. These achievements verify the potential of time series characteristics in aquaculture classification, but the existing methods have single spectral index application, and there is still room for improvement in the mining of phenological characteristics.
[0005] References
[0006] [1] Wu Y, Zhang X, Liu P, et al. A method for information extraction in coastal aquaculture areas combining spatial transformation and permutation attention mechanism[J]. Journal of Dalian University of Technology, 2024, 39(02): 327-336.
[0007] [2] Wang F, Xia L, Chen Z, et al. Remote sensing recognition of coastal seawater aquaculture patterns based on association rules and object-oriented[J]. Transactions of the Chinese Society of Agricultural Engineering, 2018, 34(12): 210-217.
[0008] [3] Yu W, Wang C, Li K. Coastal land feature classification based on high-resolution images and multi-level features[J]. Spacecraft Recovery and Remote Sensing, 2023, 44(02): 140-152.
[0009] [4] Liu Y, Yang X, Wang Z, et al. Mapping the fine spatial distribution of global offshore surface seawater mariculture using remote sensing big data[J]. International Journal of Digital Earth, 2024, 17(1): 240-2418.
[0010] [5] Yue H, Cui H, Liu S. Research on the classification method of cultivated land considering spatial scale and temporal characteristics[J]. Surveying and Mapping Science, 2024, 49(09): 134-143.
[0011] [6] Dong Xiuchun, Jiang Yi, Li Zongnan, et al. Remote sensing identification of rice-shrimp fields based on Sentinel-1 time series data [J]. Remote Sensing Technology and Application, 2024, 39(02):306-314.
[0012] [7] Li Kaiqiang. Analysis and classification of spectral characteristics of seawater for coastal aquaculture based on remote sensing observation [D]. Shandong University of Science and Technology, 2020.
[0013] [8] Chen Zhiyang, Mao Dehua, Wang Zongming, et al. Extraction of aquaculture ponds in the Jianghan Plain based on time-series Sentinel-2 data [J]. Remote Sensing of Natural Resources, 2025, 37(01):169-178.
[0014] [9]Zhang C, Gao L, Lu Z, et al. Classification of Aquaculture Watersthrough Remote Sensing on the Basis of a Time-Series Water Index[J]. Journal of Coastal Research, 2022, 38(6): 1148-1162.. Summary of the Invention
[0015] To address the shortcomings of existing technologies in distinguishing between aquaculture types with similar spectral characteristics, such as kelp, asparagus, and cage culture, this invention provides a refined classification method for coastal aquaculture based on multispectral temporal features. This method achieves high-precision classification through a dual-exponential collaborative response mechanism, temporal noise suppression, and model-feature adaptation optimization, and includes the following innovative features:
[0016] 1. Dual-spectral index synergistic response mechanism
[0017] Algae-specific identification: The dynamic ratio of mean reflectance in the green and blue bands to reflectance in the near-infrared band (NDAI) is used to capture the characteristic spectral response of algae cultivation as phenological changes occur.
[0018] Enhanced identification of aquaculture water bodies: Based on the reflectance ratio of the green-blue band combination and the red-shortwave infrared band combination (NDAWI), the spectral confusion between aquaculture water bodies and open water bodies is resolved.
[0019] 2. Triple protection against timing noise suppression
[0020] Monthly resampling is used to match key phenological nodes, linear interpolation is used to fill in data missing due to clouds and fog, and Savitzky-Golay filtering (based on window size optimization settings) is used to extract anti-interference time-series trend features and suppress short-term fluctuation noise.
[0021] 3. Model-feature adaptation optimization and verification
[0022] Single-index features and multi-index fusion features are input into multiple types of machine learning models in parallel; the optimal fit of XGBoost and multi-index fusion features in distinguishing aquaculture types is verified through multi-dimensional evaluation of classification accuracy (precision / recall, etc.).
[0023] 4. Verifiable output with a closed-loop process throughout.
[0024] Based on high-resolution validation samples, training data is interpreted to generate spatial distribution maps of kelp, seaweed, and cage aquaculture, supporting aquaculture supervision and ecological assessment.
[0025] The specific technical solution adopted by this invention to solve its technical problem is as follows:
[0026] A classification method for coastal aquaculture based on spectral temporal features, comprising:
[0027] Acquire time-series remote sensing image data of the target area and calculate multiple spectral indices, including at least the Normalized Algae Index (NDAI) and the Normalized Aquaculture Water Index (NDAWI).
[0028] The spectral indices are sequentially resampled over time, filled with missing data by linear interpolation, and subjected to Savitzky-Golay filtering to construct a smooth temporal feature set.
[0029] Based on the smoothed time series feature set, single-exponential time series features and multi-exponential fused time series features are constructed respectively;
[0030] The single-exponential time-series features and the multi-exponential fused time-series features are input into multiple machine learning models for training, and the optimal model-feature combination is selected by comparing accuracy.
[0031] The target region is classified using the optimal model-feature combination, and the aquaculture classification results are output.
[0032] Furthermore, the normalized aquaculture water index NDAWI is calculated as the normalized ratio of (green band reflectance + blue band reflectance) to (red band reflectance + shortwave infrared band reflectance); the normalized algae index NDAI is calculated as the normalized ratio of (green band reflectance + blue band reflectance) / 2 to near-infrared band reflectance.
[0033] Furthermore, the classification results distinguish between kelp, asparagus, and cage aquaculture.
[0034] Furthermore, the optimal model and feature combination is a combination of the XGBoost algorithm and multi-exponential fusion features.
[0035] Furthermore, the spectral indices also include the Normalized Difference Vegetation Index (NDVI) and the Normalized Difference Chlorophyll Index (NDCI).
[0036] Furthermore, the time resampling uses the first day of each month as the time node.
[0037] Furthermore, the Savitzky-Golay filter employs a filter with a maximum window size of 5 and uses a quadratic polynomial fitting.
[0038] Furthermore, the multi-index fusion feature is a combination of NDAI, NDAWI, NDVI, and NDCI.
[0039] Furthermore, the evaluation metrics used in the accuracy comparison include accuracy, precision, recall, and F1 score.
[0040] Furthermore, the classification results were verified using high-resolution remote sensing imagery.
[0041] And a computer device including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the method described above.
[0042] A non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described above.
[0043] Compared with the prior art, the present invention and its preferred embodiments have at least the following beneficial effects:
[0044] 1. Dual-index synergy improves the accuracy of distinguishing aquaculture types.
[0045] By employing a synergistic response mechanism between the Algae Specificity Index (NDAI) and the Aquaculture Water Index (NDAWI), the problem of spectral confusion between algae cultivation and open water bodies, such as kelp and seaweed, is effectively solved, and the ability to distinguish between aquaculture types with similar spectral characteristics is significantly enhanced.
[0046] 2. Temporal noise suppression ensures feature stability
[0047] Monthly resampling identifies key phenological periods, and triple time-series optimization combining linear interpolation and Savitzky-Golay filtering effectively suppresses cloud and fog interference and short-term fluctuation noise, extracts stable spectral time-series features, and improves classification robustness.
[0048] 3. Model-feature adaptation optimization decision reliability
[0049] By using a parallel training and accuracy comparison mechanism of single-index and multi-index fusion features, the optimal fit between XGBoost and multi-index fusion features is verified, providing a highly reliable decision-making basis for aquaculture classification.
[0050] 4. Enhanced reliability of results through a closed-loop process.
[0051] The validation loop based on high-resolution sample interpretation ensures the accuracy of the spatial distribution results of kelp, seaweed, and cage aquaculture, directly supporting the needs of aquaculture resource supervision and ecological environment assessment. Attached Figure Description
[0052] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:
[0053] Figure 1 This is a general flowchart of an embodiment of the present invention;
[0054] Figure 2 This is a model classification accuracy diagram according to an embodiment of the present invention;
[0055] Figure 3 This is a diagram showing the optimal combination classification results of an embodiment of the present invention. Detailed Implementation
[0056] To make the features and advantages of the present invention more apparent and understandable, specific embodiments are described below in detail:
[0057] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used in this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.
[0058] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.
[0059] Currently, many studies focus on aquaculture area boundary extraction and pattern recognition, with less emphasis on fine-grained classification of aquaculture species. However, the distribution characteristics of different aquaculture types are crucial for developing scientific aquaculture plans and environmental monitoring. Therefore, the ability to efficiently and cost-effectively acquire multi-category spatial distribution information of aquaculture areas is closely related to remote sensing image classification based on aquaculture temporal characteristics, and has significant strategic importance for regional fisheries management and marine ecological protection. This invention proposes an aquaculture classification method based on spectral temporal characteristics, including: high-resolution image sample interpretation, Sentinel-2 image spectral index calculation, temporal data processing, training with multiple machine learning algorithms and multiple index feature combinations, classification accuracy analysis, mapping of optimal combination classification results, and spatial pattern analysis. This method achieves automatic identification and fine-grained classification of multiple aquaculture targets, providing scientific basis and data support for optimizing regional aquaculture spatial layout, ecological environment monitoring, and refined management of aquatic resources.
[0060] like Figure 1 As shown, the specific implementation process of the coastal aquaculture classification scheme based on Sentinel-2 multispectral data and multiple index time-series features provided in this embodiment of the invention includes:
[0061] Step 1: Field investigation of kelp, asparagus and cage aquaculture, and acquisition of high-resolution multispectral remote sensing images and Sentinel-2 multispectral remote sensing images of the study area based on aquaculture phenology; visual interpretation of sample points based on high-resolution images to obtain sample data of kelp, asparagus and cage aquaculture.
[0062] Step 2: Calculate four spectral indices based on the processed image data; specifically: calculate four spectral indices based on Sentinel-2 time series images: Normalized Algae Index (NDAI), Normalized Aquaculture Water Index (NDAWI), Normalized Chlorophyll Index (NDCI), and Normalized Vegetation Index (NDVI), and construct a multi-temporal index feature dataset.
[0063] Step 3: Preprocess the four types of spectral index image data; specifically, perform temporal resampling, linear interpolation, and Savitzky-Golay filtering on the multi-temporal index data to construct smooth spectral temporal features;
[0064] Step 4: Classify aquaculture types using four machine learning algorithms and different combinations of exponential features; specifically: use single-exponential features and multi-exponential fusion features as model inputs respectively, and conduct classification experiments using various machine learning algorithms (random forest (RF), support vector machine (SVM), gradient boosting decision tree (GBDT), extreme gradient boosting algorithm (XGBoost)) to compare the classification performance under different model and feature combinations;
[0065] Step 5: Utilize the accuracy index to evaluate the model, determine the optimal classification model and feature combination, and generate an aquaculture classification result map. Specifically, the classification accuracy is evaluated using accuracy, precision, recall, and F1 score to determine the optimal classification model and feature combination.
[0066] In this embodiment, data acquisition includes the following:
[0067] Acquire Jilin-1 multispectral imagery from February 2024 and Sentinel-2 multitemporal data from 2023-2024, visually interpret sample points, and divide the dataset into training and validation sets.
[0068] In this embodiment, the exponent calculation includes the following:
[0069] Four spectral indices were calculated: Normalized Algae Index (NDAI), Normalized Difference Vegetation Index (NDVI), Normalized Aquaculture Water Index (NDAWI), and Normalized Chlorophyll Index (NDCI). The formulas for their calculation are as follows:
[0070] NDAI: ;
[0071] NDAWI: ;
[0072] NDCI: ;
[0073] NDVI: ;
[0074] Where ρ GREEN ρ BLUE ρ NIR ρ RED ρ SWIR These represent the surface reflectance values for the corresponding green, blue, near-infrared, red, and shortwave infrared bands in the Sentinel-2 data, respectively.
[0075] In this embodiment, time-series data processing specifically includes the following:
[0076] (1) Time series data are resampled, with the first day of each month as the target time step, and missing data are filled by linear interpolation;
[0077] (2) Apply Savitzky-Golay filter to smooth the data, set the window size to 5, and use quadratic polynomial fitting to extract the long-term trend;
[0078] (3) The four spectral indices are fused to obtain multi-band synthesized multi-index fused data.
[0079] In this embodiment, the steps for building and training the model include the following:
[0080] (1) Classification was performed using RF, SVM, GBDT and XGBoost algorithms respectively, based on single-index and multi-index fusion features;
[0081] (2) RF sets up 500 decision trees, with a maximum depth of 20 for each tree; GBDT sets up 500 decision trees, with a maximum depth of 10 for each tree; SVM uses radial basis function kernels and enables probability estimation; XGBoost sets up 500 trees, with a maximum depth of 10 for each tree and uses log loss as the evaluation metric.
[0082] In this embodiment, the model accuracy evaluation includes the following:
[0083] Classification accuracy is evaluated using accuracy, precision, recall, and F1 score. The formulas for calculating these metrics are as follows:
[0084] Accuracy: ;
[0085] Precision: ;
[0086] Recall: ;
[0087] F1 score: ;
[0088] Wherein, TP (True Positive) represents the number of samples correctly classified as the target class; TN (True Negative) represents the number of samples correctly classified as non-target classes; FP (False Positive) represents the number of samples misclassified as the target class; and FN (False Negative) represents the number of samples misclassified as non-target classes. Based on the evaluation results of different model and feature combinations, XGBoost and multi-index fusion features are the optimal combination for aquaculture classification.
[0089] This embodiment of the scheme comprehensively utilizes high-resolution imagery and time-series remote sensing data, combined with various spectral indices and machine learning algorithms, to achieve fine classification of three aquaculture types: kelp, seaweed, and cage culture.
[0090] This embodiment takes a bay in a certain province as the research object. It investigates the phenology of three aquaculture species—kelp, Gracilaria, and cage culture—in the study area, determining the algae sowing and harvesting periods. Based on this, it acquires Jilin-1 multispectral imagery (0.5-meter resolution) from February 2024 and Sentinel-2 multi-temporal data (10-meter resolution) from 2023-2024. Sample points are visually interpreted, and training and validation sets are defined. Using Jilin-1 imagery, 1259 sample data points are obtained, and Sentinel-2 time-series data from 2023-2024 are acquired. A time-series dataset is constructed, including an aquaculture sample set, four spectral index features (NDAI, NDVI, NDAWI, and NDCI), and index fusion features. Four machine learning algorithms (RF, SVM, GBDT, and XGBoost) and five features are used to classify the three aquaculture species—kelp, Gracilaria, and cage culture—as shown in the results. Figure 2 , Figure 3 As shown.
[0091] Based on the same inventive concept, this invention also provides a computer device, comprising: one or more processors, and a memory for storing one or more computer programs; the programs include program instructions, and the processor executes the program instructions stored in the memory. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, used to implement one or more instructions, specifically for loading and executing one or more instructions stored in a computer storage medium to implement the above-described method.
[0092] It should be further explained that, based on the same inventive concept, the present invention also provides a computer storage medium storing a computer program, which, when executed by a processor, performs the above-described method. This storage medium can be any combination of one or more computer-readable media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.
[0093] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.
[0094] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.
[0095] This invention is not limited to the preferred embodiment described above. Anyone inspired by this invention can derive other various forms of classification methods for coastal aquaculture based on spectral time-series characteristics. All equivalent variations and modifications made within the scope of the patent application of this invention shall fall within the scope of this invention.
Claims
1. A classification method for coastal aquaculture based on spectral temporal characteristics, characterized in that: Acquire time-series remote sensing image data of the target area and calculate multiple spectral indices, including at least the Normalized Algae Index (NDAI) and the Normalized Aquaculture Water Index (NDAWI). The spectral indices are sequentially resampled over time, filled with missing data by linear interpolation, and subjected to Savitzky-Golay filtering to construct a smooth temporal feature set. Based on the smoothed time series feature set, single-exponential time series features and multi-exponential fused time series features are constructed respectively; The single-exponential time-series features and the multi-exponential fused time-series features are input into multiple machine learning models for training, and the optimal model-feature combination is selected by comparing accuracy. The target region is classified using the optimal model-feature combination, and the aquaculture classification results are output.
2. The coastal aquaculture classification method based on spectral time-series characteristics according to claim 1, characterized in that: The normalized aquaculture water index NDAWI is calculated as the normalized ratio of (green band reflectance + blue band reflectance) to (red band reflectance + shortwave infrared band reflectance); the normalized algae index NDAI is calculated as the normalized ratio of (green band reflectance + blue band reflectance) / 2 to near-infrared band reflectance.
3. The coastal aquaculture classification method based on spectral temporal characteristics according to claim 1, characterized in that: The classification results distinguish the categories of kelp, asparagus, and cage culture.
4. The coastal aquaculture classification method based on spectral time-series characteristics according to claim 1, characterized in that: The optimal model and feature combination is a combination of the XGBoost algorithm and multi-index fusion features.
5. The coastal aquaculture classification method based on spectral time-series characteristics according to claim 1, characterized in that: The spectral indices also include the Normalized Difference Vegetation Index (NDVI) and the Normalized Chlorophyll Index (NDCI).
6. The coastal aquaculture classification method based on spectral time-series characteristics according to claim 1, characterized in that: The time resampling is based on the first day of each month.
7. The coastal aquaculture classification method based on spectral time-series characteristics according to claim 1, characterized in that: The Savitzky-Golay filter uses a filter with a maximum window size of 5 and is fitted with a quadratic polynomial.
8. The coastal aquaculture classification method based on spectral time-series characteristics according to claim 1, characterized in that: The evaluation metrics used in the accuracy comparison include accuracy, precision, recall, and F1 score.
9. The coastal aquaculture classification method based on spectral time-series characteristics according to claim 1, characterized in that: The multi-index fusion feature is a combination of NDAI, NDAWI, NDVI, and NDCI.
10. The coastal aquaculture classification method based on spectral time-series characteristics according to claim 1, characterized in that: The classification results were verified using high-resolution remote sensing imagery.