Marine ecological early warning online monitoring method and device, electronic equipment and storage medium

Abstract

The application provides a marine ecological early warning online monitoring method and device, electronic equipment and a storage medium, belonging to the technical field of marine ecological protection. The method comprises the following steps: acquiring satellite remote sensing data of an air-based region, performing feature enhancement on the satellite remote sensing data to obtain enhanced data, and performing spectral feature extraction on the enhanced data based on a time dimension and a space dimension to obtain a multispectral feature data set; determining a shore-based region and a sea-based region based on geographic location information in the multispectral feature data set, acquiring first environmental data of the shore-based region and second environmental data of the sea-based region; performing time-space alignment on the first environmental data and the second environmental data based on a dynamic time-space calibration algorithm to obtain target environmental data; determining a risk warning result based on the target environmental data and a risk warning model, and performing online monitoring on a target region based on the risk warning result. The application can improve the response efficiency of marine ecological early warning.

Description

Technical Field

[0001] This application belongs to the field of marine ecological protection technology, and more specifically, relates to a marine ecological early warning online monitoring method, device, electronic equipment and storage medium. Background Technology

[0002] With the increasing demand for marine ecological protection, we can currently obtain long-term, high-frequency marine ecological monitoring data by carrying out online monitoring for marine ecological early warning, patrols of marine disaster-prone areas, and patrols of beach erosion conditions. This data can provide timely and effective support for the prevention and control of ecological disasters such as red tides and green tides, and provide basic data and management basis for marine early warning management.

[0003] In response to ecological disasters such as red tides and green tides, traditional marine monitoring involves equipment with limited monitoring range, making it difficult to cover the entire area to be monitored and thus unable to achieve full-chain tracing and early warning of ecological risks. Summary of the Invention

[0004] The purpose of this application is to provide a method, device, electronic equipment and storage medium for online monitoring of marine ecological early warning, so as to expand the scope of marine monitoring, improve the response efficiency of marine ecological early warning, and realize the whole-chain tracing and early warning of ecological risks.

[0005] A first aspect of this application provides a method for online monitoring and early warning of marine ecology, comprising: Satellite remote sensing data of the air-based region is acquired, and feature enhancement is performed on the satellite remote sensing data to obtain enhanced data. Spectral features are extracted from the enhanced data based on the time and spatial dimensions to obtain a multispectral feature dataset; the air-based region is the open ocean area. Based on the geographic location information in the multispectral feature dataset, the shore-based region and the sea-based region are determined, and the first environmental data of the shore-based region and the second environmental data of the sea-based region are obtained. The shore-based region, the sea-based region and the air-based region all belong to the target region. The shore-based region, the sea-based region and the air-based region extend from the mainland to the open sea in sequence. The target environment data is obtained by spatiotemporally aligning the first and second environment data based on a dynamic spatiotemporal calibration algorithm. Risk warning results are determined based on target environmental data and risk warning models, and online monitoring of the target area is carried out based on the risk warning results.

[0006] A second aspect of this application provides a marine ecological early warning online monitoring device, comprising: The data acquisition unit is used to acquire satellite remote sensing data in the air-based region, perform feature enhancement on the satellite remote sensing data to obtain enhanced data, and extract spectral features from the enhanced data based on the time and spatial dimensions to obtain a multispectral feature dataset; the air-based region is the open ocean area; The first data processing unit is used to determine the shore-based region and the sea-based region based on the geographical location information in the multispectral feature dataset, and to acquire the first environmental data of the shore-based region and the second environmental data of the sea-based region. The shore-based region, the sea-based region and the air-based region all belong to the target region, and the shore-based region, the sea-based region and the air-based region extend from the mainland to the open sea in sequence. The second data processing unit is used to perform spatiotemporal alignment of the first environmental data and the second environmental data based on a dynamic spatiotemporal calibration algorithm to obtain the target environmental data. The risk prediction unit is used to determine the risk warning results based on the target environmental data and the risk warning model, and to conduct online monitoring of the target area based on the risk warning results.

[0007] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the above-described online monitoring method for marine ecological early warning.

[0008] In a fourth aspect of this application, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the above-described online monitoring method for marine ecological early warning.

[0009] The beneficial effects of the marine ecological early warning online monitoring method, device, electronic equipment, and storage medium provided in this application embodiment are as follows: This application's embodiments overcome the limitations of traditional monitoring in the far sea area, such as difficult coverage and low data quality, by first acquiring satellite remote sensing data of the airborne region (the open ocean area) and then performing feature enhancement and spatiotemporal spectral feature extraction. This provides accurate basic data for subsequently determining the shore-based and sea-based regions that need to be monitored. Based on the geographical location information in the multispectral feature dataset, shore-based and sea-based regions extending sequentially from the mainland to the open sea are determined, and environmental data corresponding to each region is acquired, achieving full monitoring coverage of the target area. This facilitates the full-chain tracing and early warning of ecological risks. In addition, determining the shore-based and sea-based regions to be monitored based on the multispectral feature dataset of the airborne region first can reduce the monitoring of irrelevant areas and reduce manpower and material costs. This application's embodiments also align the first and second environmental data through a dynamic spatiotemporal calibration algorithm to obtain target environmental data with a unified spatiotemporal framework, effectively eliminating spatiotemporal deviations in data from different regions and ensuring data consistency and usability. This application's embodiments determine risk warning results based on a risk warning model and conduct targeted online monitoring, and can dynamically adjust monitoring strategies based on warning results, improving the accuracy and timeliness of marine ecological risk warnings. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 A schematic flowchart of an online monitoring method for marine ecological early warning provided in an embodiment of this application; Figure 2 A distribution map of the shore-based and sea-based regions provided in an embodiment of this application; Figure 3 An architectural diagram of a preset grid cell provided in an embodiment of this application; Figure 4 This is a structural block diagram of a marine ecological early warning online monitoring device provided in one embodiment of this application; Figure 5 This is a schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0012] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0013] To facilitate understanding, the main terms used in this application will be explained first.

[0014] The shore-based region refers to the spatial area extending from the coastline towards the continental boundary and towards the upper limit of the intertidal zone. This region includes both the continental and oceanic portions. The continental portion refers to the land area where fixed shore-based monitoring stations, river estuaries, and sewage outlets are located. The oceanic portion refers to the intertidal zone and nearshore shallow waters, where the intertidal zone includes areas such as mudflats, wetlands, and tidal flats.

[0015] The sea-based area refers to the waters and seabed space extending from the lower limit of the intertidal zone to the open sea, using ocean waters, seabed surfaces, and floating data acquisition devices as carriers, within 50 nautical miles of the coastline.

[0016] The air-based area refers to the offshore area covered by observations using remote sensing satellites and drones, which is more than 50 nautical miles from the coastline.

[0017] GIS stands for Geographic Information System. The data acquisition and input module in a GIS system can bind non-spatial data to spatial locations and transform it into a GIS-recognizable format.

[0018] Spatial network refers to a visualized geographic network model that projects the location coordinates of various shore-based monitoring stations in the shore-based area and the location coordinates of various marine-based equipment in the marine-based area onto the three-dimensional space of the GIS platform, resulting in interconnected geographic spaces.

[0019] In this embodiment, the current marine ecosystem is a key component of the Earth's food chain, possessing core functions such as maintaining biodiversity, regulating climate, providing fishery resources, and supporting shipping and coastal economic development. However, it also faces multiple threats, such as red tides / green tides and water quality deterioration. These marine ecological problems are diffuse and transmissible; once damaged, recovery is lengthy and costly, necessitating a shift from passive management to proactive early warning. Therefore, this embodiment proposes an online monitoring method for marine ecological early warning, enabling timely warnings at the early stages of marine problems.

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the following description will be provided in conjunction with the accompanying drawings and specific embodiments.

[0021] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the online monitoring method for marine ecological early warning provided in this application. The method can be executed by an electronic device and may include steps S101 to S104.

[0022] S101: Acquire satellite remote sensing data of the air-based region, perform feature enhancement on the satellite remote sensing data to obtain enhanced data, and extract spectral features from the enhanced data based on the time and spatial dimensions to obtain a multispectral feature dataset.

[0023] The air-based area refers to the open sea region.

[0024] This embodiment acquires satellite remote sensing data of an airborne region periodically through sensors carried by a remote sensing satellite. This satellite remote sensing data may include raw signal data used to characterize various physical, chemical, and biological properties of the ocean surface.

[0025] Because raw satellite remote sensing data is greatly affected by external factors such as atmospheric scattering, cloud cover, and sensor noise, it can blur ecologically relevant features. Therefore, feature enhancement is needed to eliminate interference and highlight ecologically relevant features. In this embodiment, feature enhancement methods include, but are not limited to, edge and texture enhancement, spectral enhancement, or dynamic feature enhancement. Enhanced data is obtained by performing feature enhancement on the satellite remote sensing data; that is, by using feature enhancement methods such as band ratio, normalized difference index, or band difference. These feature enhancement methods are existing methods and will not be elaborated upon here.

[0026] Multispectral feature datasets are obtained by extracting spectral features from the enhanced data based on the time and spatial dimensions. This can be understood as follows: Marine ecosystems are dynamically evolving, and the marine ecosystems at different locations at the same time exhibit spatial heterogeneity. For example, red tides take several days from initial appearance to full outbreak, and red tide trends differ between nearshore and offshore areas. Therefore, it is necessary to perform spatiotemporal alignment on enhanced satellite remote sensing data based on both temporal and spatial dimensions. For instance, if the acquired satellite remote sensing data is for a specific time period, such as 1 day, 5 days, or 7 days, enhanced data acquired by the same remote sensing satellite at different time points are matched according to "spatial location" to obtain a time-series dataset for the airborne region. The time-series dataset is then processed using methods such as sliding window averaging or temporal interpolation to filter out temporal noise. Multispectral feature data is then extracted from the processed time-series data to obtain a multispectral feature dataset for the temporal dimension. Alternatively, if the acquired satellite remote sensing data spans different spatial regions, the entire airborne region can be divided into multiple "spatial units" based on ocean current regions, water depth ranges, or latitude and longitude tables. The spectral specificity index of the enhanced satellite remote sensing data from different "spatial units" at the same time, as well as the spectral similarity between adjacent "spatial units," are calculated to obtain a multispectral feature dataset for the spatial dimension.

[0027] This embodiment is based on "spatial units," matching each "spatial unit" with corresponding time-series data to form a spatiotemporal grid dataset. Each spatiotemporal grid is identified by its spatial location and time range. For example, 37°-38° North latitude, 118°-119° East longitude, month A, day B. The content of each spatiotemporal grid includes, but is not limited to, the multispectral data of that spatial unit at different time points.

[0028] S102: Determine the shore-based region and the sea-based region based on the geographical location information in the multispectral feature dataset, and obtain the first environmental data of the shore-based region and the second environmental data of the sea-based region.

[0029] Among them, the shore-based area, the sea-based area, and the air-based area are all part of the target area, which extends from the mainland to the open sea in sequence. In this embodiment, geographic location information refers to the spatial coordinate information bound to spectral features in the multispectral feature dataset, typically including latitude and longitude information. Determining the shore-based and sea-based regions based on the geographic location information in the multispectral feature dataset includes: identifying areas in the air-based region with red tide risk or other anomaly risks based on the multispectral data in the multispectral feature dataset; and determining the shore-based and sea-based regions based on the geographic location information corresponding to these areas. (Reference) Figure 2Coastal base 11 and coastal base 12 are two different coastal base regions, while sea base 21, sea base 22, and sea base 23 are different sea base regions. If areas with red tide risk in the air-based regions have been identified based on the multispectral data in the multispectral feature dataset, then the coastal base region determined based on the corresponding geographical location information of these areas can be coastal base 11, and the sea base regions can be sea base 21 and sea base 22. The coastal base regions, sea base regions, and air-based regions extend sequentially from the mainland to the open sea; that is, the coastal base regions are closest to the continental plates, and the air-based regions are farthest from the continental plates.

[0030] Because monitoring all sea-based and shore-based systems would require a significant amount of manpower and resources, this embodiment first monitors the air-based area to obtain a multispectral feature dataset. Then, based on the multispectral feature dataset, it selects the sea-based and shore-based areas that need to be monitored, making the monitoring more targeted and reducing manpower and resources.

[0031] After determining the shore-based and sea-based areas, environmental data can be collected from these areas. The primary environmental data for the shore-based areas includes, but is not limited to, water quality monitoring factors and intertidal vegetation coverage factors; the secondary environmental data for the sea-based areas includes, but is not limited to, water quality monitoring factors collected by sea-based buoys or unmanned vessels.

[0032] S103: Based on the dynamic spatiotemporal calibration algorithm, the first environmental data and the second environmental data are spatiotemporally aligned to obtain the target environmental data.

[0033] In this embodiment, the dynamic spatiotemporal calibration algorithm is based on the asynchronous temporal sampling and spatial location deviation of data from different sources. Through interpolation and data completion operations, it achieves temporal synchronization and spatial matching of data from different sources. The dynamic spatiotemporal calibration algorithm can be Dynamic Time Warping (DTW) or Spatio-Temporal Kriging (STK), etc. Spatiotemporal alignment means that the dynamic spatiotemporal calibration algorithm ensures that the first and second environmental data satisfy both temporal and spatial consistency.

[0034] In this embodiment, the data acquisition frequencies of the first environmental data and the second environmental data may differ. For example, the first environmental data may be obtained from monitoring stations in a shore-based area, with a data acquisition frequency of 2F. The second environmental data may be obtained from sea-based buoys or unmanned vessels in a sea-based area. Because the sea-based area covers a large area, its data acquisition frequency is lower than that of the shore-based area, and can be F. As a result, the amount of first and second environmental data collected within the same time period will be different. To process these two types of data simultaneously, the second environmental data, which has a smaller amount of data, can be supplemented with data from the first environmental data in the time dimension to maintain temporal consistency.

[0035] In this embodiment, the first environmental data and the second environmental data are collected from different locations. Therefore, it is necessary to spatially align the first environmental data and the second environmental data based on the spatial dimension to maintain spatial consistency for subsequent processing.

[0036] The target environmental data obtained by spatiotemporally aligning the first and second environmental data eliminates the spatiotemporal bias between the two types of data and can be directly used for subsequent marine ecological early warning analysis.

[0037] S104: Determine the risk warning results based on the target environmental data and risk warning model, and conduct online monitoring of the target area based on the risk warning results.

[0038] In this embodiment, the risk warning model is an algorithmic model constructed based on historical ecological risk data and historical environmental parameters of the target area. It can predict potential marine ecological risk types, risk levels, and risk areas by analyzing abnormal features in the current target environmental data. In this embodiment, the risk warning model can be a Long Short-Term Memory (LSTM) network model. The input data of this model are historical ecological risk data and historical environmental parameters of the target area, and the output data are the marine ecological risk types, risk levels, and risk areas of the target area.

[0039] Risk warning results are determined based on target environmental data and a risk warning model. This includes inputting target environmental data into the risk warning model to obtain the results, or inputting target environmental data along with other influencing factors into the model. Other influencing factors refer to non-environmental parameters that significantly impact marine ecological risks, including meteorological factors such as wind speed, wind direction, and precipitation; human activity factors such as aquaculture wastewater discharge and vessel traffic density in shore-based areas; and hydrodynamic factors such as tidal cycles and seabed topography. The risk warning results include information such as the risk area, risk level, probability of occurrence, and expected duration. This information serves as the basis for adjusting online monitoring strategies.

[0040] Online monitoring of target areas based on risk warning results can be understood as follows: if the current online monitoring strategy is implemented, and the risk warning results obtained based on the risk warning model indicate that certain areas are prone to red tides or other marine ecological disasters, then the current online monitoring strategy can be adjusted to other monitoring strategies. For example, the data collection frequency of shore-based, sea-based, and air-based areas can be increased, the monitoring equipment in shore-based areas can be increased, the accuracy of data collection in sea-based areas can be improved, and the accuracy of marine ecological warnings can be improved.

[0041] As can be seen from the above, this application embodiment overcomes the limitations of traditional monitoring in the far sea area, such as difficult coverage and low data quality, by first acquiring satellite remote sensing data of the air-based region (far sea area) and then performing feature enhancement and spatiotemporal spectral feature extraction. This provides accurate basic data for subsequently determining the shore-based and sea-based regions that need to be monitored. Based on the geographical location information in the multispectral feature dataset, the shore-based and sea-based regions extending sequentially from the mainland to the far sea are determined, and the corresponding environmental data for each region are acquired, achieving full monitoring coverage of the target area. This is beneficial for realizing full-chain tracing and early warning of ecological risks. In addition, determining the shore-based and sea-based regions that need to be monitored based on the multispectral feature dataset of the air-based region can reduce the monitoring of irrelevant areas and reduce manpower and material costs. This application embodiment also aligns the first and second environmental data through a dynamic spatiotemporal calibration algorithm to obtain target environmental data with a unified spatiotemporal framework, effectively eliminating spatiotemporal deviations in data from different regions and ensuring data consistency and usability. This application embodiment determines the risk warning results based on the risk warning model and conducts targeted online monitoring. It can also dynamically adjust the monitoring strategy according to the warning results, improving the accuracy and timeliness of marine ecological risk warning.

[0042] In one embodiment of this application, the first environmental data includes the first location coordinates corresponding to each shore-based monitoring station, and the second environmental data includes the second location coordinates corresponding to each sea-based device. First environmental data from the shore-based region and second environmental data from the sea-based region are acquired. Based on a dynamic spatiotemporal calibration algorithm, the first and second environmental data are spatiotemporally aligned to obtain target environmental data, including: First environmental data of the shore-based area is acquired based on the first data acquisition frequency; Second environmental data for the sea-based area are acquired based on a second data acquisition frequency, where the first data acquisition frequency is greater than the second data acquisition frequency. Using the first environmental data as a benchmark, time series interpolation is performed on the second environmental data to obtain the second target environmental data; Import the first and second location coordinates into the GIS system to construct a spatial network; The target environment data is obtained based on the first environmental data, the second target environment data, and the spatial network.

[0043] In this embodiment, the first environmental data includes the first location coordinates corresponding to each shore-based monitoring station, that is, the geographic coordinates corresponding to each shore-based monitoring station that collects the data. The second environmental data includes the second location coordinates corresponding to each shore-based monitoring station, that is, the geographic coordinates corresponding to each sea-based device that collects the data, such as a buoy monitoring device, an underwater sensor, or a mobile sampling device on a research vessel.

[0044] In this embodiment, the first data acquisition frequency of the shore-based region is greater than the second data acquisition frequency of the sea-based region, for at least one of the following reasons: Equipment operating environment, equipment maintenance environment, and data transmission characteristics.

[0045] In this embodiment, the shore-based area is located in a fixed area on land or near the shore, free from direct interference from extreme marine environments such as wind, waves, ocean currents, and high salt spray. The equipment operates in a stable environment and does not need to deal with issues such as seawater corrosion, equipment drift, and underwater pressure, resulting in a low hardware wear rate. The sea-based area is significantly affected by the marine dynamic environment, and the equipment must withstand high salt spray corrosion, wind and wave impacts, and underwater high pressure. Increasing the acquisition frequency will exacerbate equipment energy consumption and hardware load, shortening equipment lifespan. Therefore, the first data acquisition frequency of the shore-based monitoring station is greater than the second data acquisition frequency of the sea-based equipment.

[0046] Shore-based monitoring stations can be fixedly installed on the coast, offering flexible power supply options and providing continuous and stable power support. Equipment maintenance can be quickly accessed via land transportation, resulting in low operation and maintenance costs and high efficiency. In contrast, sea-based equipment relies on batteries or offshore solar power for power. Increased data collection frequency leads to rapid energy consumption and shortens the equipment's operating cycle. Furthermore, equipment maintenance requires the use of research vessels, drones, etc., making operation and maintenance more difficult and costly. Therefore, the primary data collection frequency of shore-based monitoring stations is higher than that of sea-based equipment.

[0047] Onshore areas can rely on land-based communication networks, which have sufficient transmission bandwidth and stable signals. However, sea-based areas mainly rely on satellite transmission, which has problems such as limited bandwidth, high transmission costs, and signals that are easily affected by weather. If the acquisition frequency is increased, the amount of data will exceed the satellite's transmission capacity, leading to data transmission delays or loss, or significantly increasing communication costs. Therefore, it is necessary to control the amount of data by reducing the acquisition frequency to ensure that critical data can be transmitted effectively.

[0048] In this embodiment, because the data acquisition frequency of the second environmental data is less than that of the first environmental data, the amount of data in the second environmental data is also less than that in the first environmental data. Therefore, the first environmental data can be used as a benchmark to perform time series interpolation on the second environmental data to obtain the interpolated second environmental data, i.e., the second target environmental data. Here, time series interpolation means using the time nodes of the first environmental data as a reference, and estimating the missing time node data based on existing, continuous time point data in the second environmental data using mathematical algorithms such as linear interpolation or cubic spline interpolation.

[0049] In this embodiment, the data supplementation of the second environmental data is not limited to time series interpolation alone. Alternatively, the second environmental data and the first environmental data can be input into a trained regression model to obtain the second target environmental data. This regression model can learn the correlation between environmental data in the shore-based region and the sea-based region.

[0050] Import the first and second location coordinates into the GIS system to construct a spatial network; based on the first environmental data, the second target environmental data, and the spatial network, obtain the target environmental data, which can be understood as: By importing the primary location coordinates of the shore-based monitoring stations and the secondary location coordinates of the marine-based equipment into a GIS system, the GIS system can accurately locate all monitoring points to their corresponding geographic spaces, forming a discrete spatial monitoring point layer. The primary and secondary location coordinates can be latitude and longitude, or distances from fixed locations, etc. After establishing the spatial network, the primary and secondary environmental data can be filled into the corresponding positions in the spatial network, resulting in a spatial network with bidirectional location-data correlation. This facilitates the analysis of the diffusion relationships between various parameters and enables timely early warning of potential phenomena such as red tides.

[0051] As can be seen from the above, this embodiment acquires environmental data from shore-based and sea-based areas at different acquisition frequencies, interpolating the low-frequency sea-based data using the high-frequency shore-based data as a benchmark, ensuring the consistency and integrity of the data in the temporal dimension. The location coordinates of the two areas are imported into a GIS system to construct a spatial network, achieving precise matching in the spatial dimension. This embodiment processes the first and second environmental data through both spatiotemporal dimensions to obtain the target environmental data, improving the quality and efficiency of multi-source heterogeneous data fusion and providing reliable support for marine risk early warning and online monitoring of marine ecosystems.

[0052] In one embodiment of this application, both the first environmental data and the second target environmental data include multiple identical water quality monitoring factors; the target environmental data is obtained based on the first environmental data, the second target environmental data, and a spatial network, including: For the same water quality monitoring factor, determine the overlapping monitoring periods for shore-based and sea-based areas; Based on the environmental data of the shore-based area corresponding to the overlapping monitoring period, the environmental data of the sea-based area corresponding to the overlapping monitoring period is calibrated using a calibration model constructed based on the random forest algorithm. The target environment data is obtained based on the first environmental data, the calibrated second target environment data, and the space network.

[0053] In this embodiment, the first environmental data and the second target environmental data may include multiple identical water quality monitoring factors. These water quality monitoring factors are core parameters used to characterize water quality, such as dissolved oxygen, chemical oxygen demand, ammonia nitrogen concentration, chlorophyll a concentration, and pH value. Overlapping monitoring periods refer to the time intervals in which the first environmental data and the second target environmental data coincide in time. Because the first data collection frequency is higher than the second data collection frequency, they will form a large number of overlapping periods with completely identical timestamps. During these overlapping periods, because the marine-based equipment is located at sea level or seabed for extended periods, it is subject to severe wind, waves, and seawater erosion, which may cause data anomalies. Therefore, the values ​​in the second target environmental data can be calibrated based on the values ​​corresponding to the same water quality monitoring factors in the first environmental data.

[0054] The calibration model can be a machine learning model built based on the random forest algorithm to correct biases in marine environmental data. The input data for the calibration model consists of first environmental data and second target environmental data to be calibrated, and the output is the calibrated second target environmental data. In this embodiment, the parameters that need to be set for the calibration model include the number of decision trees, the maximum depth, the minimum number of leaf node samples, and the number of feature selections. The number of decision trees is the total number of basic decision trees in the model. Setting a reasonable number ensures a balance between the model's calibration accuracy and computational efficiency, for example, 50. Setting the maximum depth prevents overfitting; the maximum depth can be 10. Setting the minimum number of leaf node samples avoids overfitting to individual abnormal monitoring data; for example, it can be 2. Randomly selecting the number of features for each tree enhances the model's diversity and improves calibration robustness; for example, it can be 70% of the total number of features.

[0055] In this embodiment, by calibrating multiple water quality monitoring factors, the problem of data anomalies caused by equipment malfunctions and environmental interference in marine-based data is solved. After data calibration, the final target environmental data can be generated by combining the first environmental data, the calibrated second target environmental data, and the spatial network, which can significantly improve the accuracy of marine ecological risk early warning.

[0056] In one embodiment of this application, target environment data is obtained based on first environmental data, second target environment data, and a spatial network, including: The first environmental data and the second target environmental data are projected onto the preset grid cells of the spatial network, and the preset grid cells without data are selected as the target grid cells. The data in the target grid cell is completed based on the first environmental data and the second target environmental data; All data in the completed spatial network will be used as target environment data.

[0057] In this embodiment, the preset grid unit of the spatial network is a pre-defined standardized spatial unit used to divide the monitoring area. The division of the preset grid unit can be based on the distance between each shore-based monitoring station. For example, if the distance between each shore-based monitoring station is 5km, then the preset grid unit can be a 5km×5km grid.

[0058] refer to Figure 3 The first location coordinate corresponds to the first environmental data, and the second location coordinate corresponds to the second target environmental data. After projecting the first environmental data and the second target environmental data onto the preset grid cells of the spatial network, it can be seen that there is no corresponding data in grid cell 300. Grid cell 300 is the preset grid cell with unfilled data that has been filtered out, which is the target grid cell. When the distribution of shore-based monitoring stations or sea-based equipment is unreasonable, or when shore-based monitoring station equipment is damaged or sea-based equipment drifts with the waves, the following will occur: Figure 3 The spatial network shown is a distribution pattern. If the target grid cell has no data, it will affect the assessment of subsequent risk warning results. Therefore, the data in the target grid cell can be supplemented based on the first environmental data and the second target environmental data, and all the data in the supplemented spatial network can be used as the target environmental data.

[0059] In one embodiment, data in the target grid cell is completed based on first environmental data and second target environmental data, including: Acquire first environmental data and second target environmental data within a preset range around the target grid cell, and use the first environmental data and second target environmental data within the preset range as reference data; determine the influence weight of each reference data on the target grid cell based on the spatial distance between each reference data and the target grid cell and the dynamic influence factor of the marine environment; perform weighted calculation based on each reference data and the influence weight corresponding to each reference data to obtain the initial complete data of the target grid cell; The initial completion data is corrected based on the multispectral feature dataset of the empty base region to obtain the final completion data of the target grid cell; A consistency check is performed on the data of all completed target grid cells. If any target grid cell has abnormal data, the data completion step is re-executed. The consistency check on the data of all completed target grid cells includes: calculating the difference between the completed data of adjacent target grid cells; if the difference exceeds a preset threshold, a target grid cell is determined to have abnormal data.

[0060] Both the preset range and the preset threshold can be set based on experience. For example, the preset range is divided based on distance, typically selecting adjacent grids of 3×3, 5×5, or 9×9 centered on the target grid cell. Adjacent grids of the target grid cell belong to the preset range, while non-adjacent grids do not. This division method is a conventional approach in grid division. The preset threshold is a critical value used to determine whether the difference in the completed data of adjacent target grid cells is abnormal. Setting this preset threshold based on experience means that, for existing historical measured data or historical valid grid data in both shore-based and sea-based areas, the data difference between all adjacent grids can be statistically analyzed, based on statistical... The criterion uses the statistical maximum value as the preset threshold.

[0061] The dynamic influencing factors of the marine environment include ocean current velocity, tidal height, and seawater temperature gradient. Based on the spatial distance between each reference data point and the target grid cell, and the dynamic influencing factors of the marine environment, the influence weight of each reference data point on the target grid cell is determined. This includes: calculating the basic weight corresponding to each reference data point based on the spatial distance between each reference data point and the target grid cell; spatial distance and basic weight are negatively correlated; calculating dynamic correction coefficients based on ocean current velocity, tidal height, and seawater temperature gradient; and multiplying each basic weight by the dynamic correction coefficient to obtain the influence weight of each reference data point on the target grid cell.

[0062] The initial completed data is corrected based on the multispectral feature dataset of the empty base region to obtain the final completed data of the target grid cell. This includes: extracting the spectral feature parameters corresponding to the target grid cell from the multispectral feature dataset; determining the environmental correction value corresponding to the spectral feature parameters based on the preset mapping relationship between spectral features and environmental data; and fusing the initial completed data and the environmental correction value to obtain the final completed data of the target grid cell.

[0063] In this embodiment, the influence weight of each reference data is calculated by combining the spatial distance between the reference data and the target grid cell, as well as dynamic marine environmental factors such as ocean current velocity and tidal height. The initial completed data is obtained by weighting, with higher basic weights for closer spatial distances. Then, based on the multispectral feature data of the spatial base area, environmental correction values ​​are determined according to the mapping relationship between spectral features and environmental data to optimize the initial data and obtain the final completed data. Finally, by verifying whether the difference between adjacent grid data exceeds a threshold, the consistency and accuracy of the completed data are ensured, achieving accurate completion of the target grid cell data.

[0064] As can be seen from the above, projecting the first environmental data and the second target environmental data onto the preset grid cells in the spatial network can accurately locate the data distribution. By filtering the target grid cells without data filling, the missing data areas can be quickly identified, and the missing data in the target grid cells can be filled in using the existing data in a targeted manner. This effectively solves the problem of data gaps, ensures the integrity and comprehensiveness of the data, and makes the final target environmental data more accurately reflect the overall environmental situation, thereby improving the reliability of the risk warning analysis results.

[0065] In one embodiment of this application, the sea-based region includes multiple sea-based sub-regions, and the shore-based region includes multiple shore-based sub-regions; the marine ecological early warning online monitoring method further includes: Based on real-time hydrodynamic data and historical ecological risk database within the target area, the marine-based and shore-based sub-regions are dynamically classified to obtain multiple sub-regions corresponding to different environmental risk types. For each type of environmental risk, upstream and downstream labels are made for multiple sub-regions corresponding to different environmental risk types, and regional risk transmission weights are assigned to the multiple sub-regions after upstream and downstream labeling. Among them, determining the risk warning results based on target environmental data and risk warning models includes: The target environmental data is divided into target environmental data corresponding to each sub-region based on multiple sub-regions. The environmental feature set of the target environmental data corresponding to each sub-region is extracted. The environmental feature set and regional risk transmission weight are input into the risk warning model to obtain the risk warning result.

[0066] In this embodiment, the sea-based and shore-based regions are dynamically divided, breaking the limitations of fixed regional classification. This allows for the creation of multiple sub-regions based on different environmental risk types, facilitating early warning of areas most prone to red tides, hypoxia, and land-based pollution. The dynamic classification in this embodiment can be based on machine learning or rule-matching algorithms to establish a correlation model between hydrodynamic data and historical risks. Based on this model, multiple sub-regions corresponding to different environmental risk types are obtained. For example, if the real-time sewage discharge flow rate of a shore-based sub-region is >5 m³ / s (far exceeding the historical average), and the historical data shows that the "land-based pollution risk" occurrence rate under this discharge flow rate reaches 80%, then this shore-based sub-region is classified as a "highly sensitive land-based pollution sub-region." If the real-time tidal current velocity of a sea-based sub-region is <0.2 m / s (favorable for algal aggregation), and the historical data shows that "red tide risk" frequently occurs under this velocity, then it is classified as a "red tide prone sub-region."

[0067] In this embodiment, for sub-regions with different environmental risk types, their risk transmission paths are clearly defined and assigned weights. The aim is to reflect the degree of impact of upstream sub-region risks on downstream sub-regions, thus better reflecting the actual environment. Based on the "direction of material migration" in real-time hydrodynamic data, such as runoff flowing towards the sea in shore-based sub-regions and tidal currents flowing from sub-region A to sub-region B in sea-based sub-regions, upstream and downstream sub-regions of the same risk type are labeled. For example, in the "land-based pollution risk category," the shore-based sub-region of the estuary is the source of pollution and can be marked as the "upstream sub-region," the sea-based sub-region within 5 km outside the estuary is marked as the "first-level downstream sub-region," and the sea-based sub-region 10 km further outside is marked as the "second-level downstream sub-region," and so on.

[0068] Based on the flow velocity, sub-region spacing, and historical transmission efficiency in the hydrodynamic data, a corresponding regional risk transmission weight is assigned to each sub-region using the analytic hierarchy process (AHP) or entropy weight method. The closer the sub-regions are, the more direct the risk transmission, and the higher the regional risk transmission weight. Historical transmission efficiency refers to the probability that pollution will occur in the downstream sub-region within 12 hours after pollution in the upstream sub-region in the past; the higher the probability, the higher the regional risk transmission weight. For example, the regional risk transmission weight of an upstream shoreline sub-region to the first-level downstream marine sub-region is 0.8 (direct impact), to the second-level downstream marine sub-region is 0.3 (reduced impact), and the regional risk transmission weight between sub-regions that are not upstream or downstream is 0 (no direct impact).

[0069] In this embodiment, after determining the regional risk transmission weights for multiple sub-regions, the target environmental data can be divided into target environmental data corresponding to each sub-region based on the multiple sub-regions. Environmental feature sets of the target environmental data corresponding to each sub-region are extracted, and the environmental feature sets and regional risk transmission weights are input into the risk warning model to obtain the risk warning result. The risk warning model can be a random forest model, a graph neural network model, or a long short-term memory network model, etc.

[0070] In this embodiment, if the risk warning model is a Long Short-Term Memory (LSTM) network model, the working principle of inputting the environmental feature set and regional risk transmission weights into the risk warning model to obtain the risk warning result is as follows: the environmental feature set and regional risk transmission weights are weighted to obtain a comprehensive feature vector, and then this comprehensive feature vector is input into the risk warning model to obtain the risk warning result. The processing methods for other models can refer to their respective data processing specifications, and will not be elaborated here.

[0071] As can be seen from the above, this embodiment dynamically classifies sea-based and shore-based sub-regions to obtain sub-regions corresponding to different environmental risk types. It also labels each sub-region with upstream and downstream information and assigns regional risk transmission weights, clearly presenting the risk transmission path and impact level, thus helping to quickly identify risk areas. Furthermore, this embodiment divides the target environmental data into sub-regions and extracts the corresponding environmental feature sets for each sub-region. Inputting multiple environmental feature sets and the regional risk transmission weights for each sub-region into the risk warning model enables precise risk warnings based on the actual risk transmission situation, improving the timeliness and accuracy of marine ecological early warnings.

[0072] In one embodiment of this application, the risk warning result includes the risk warning level corresponding to each sub-region; After inputting the environmental feature set and regional risk transmission weights into the risk warning model to obtain the risk warning results, the following steps are also included: Based on the correlation between risk warning level and the spatial location of sub-regions, a dynamic risk diffusion path prediction map is generated. Cross-regional risk prediction results were obtained based on dynamic risk diffusion path prediction maps and satellite remote sensing data.

[0073] In this embodiment, after obtaining the risk warning results including the risk warning levels of each sub-region, the correlation between the risk warning level and the spatial location of the sub-region is utilized, that is, the spatial distribution and mutual influence of different risk levels. A dynamic risk diffusion path prediction map is generated by a specific algorithm to intuitively present the direction and range of possible risk diffusion. This prediction map is then combined with satellite remote sensing data. Satellite remote sensing data can provide information such as the geographical environment on a large scale and in real time, which helps to judge the influence of geographical, environmental and other factors that may affect the risk diffusion process. Finally, a comprehensive analysis is conducted to obtain the cross-regional risk prediction results.

[0074] In one embodiment, a dynamic risk diffusion path prediction map is generated based on the correlation between risk warning level and the spatial location of the sub-region, including: Sub-regions with risk warning levels greater than preset warning levels are identified as risk sources. The initial risk diffusion boundary is determined based on the upstream and downstream labels and adjacency relationships of the sub-regions. The initial risk diffusion boundary is the sub-regions with risk warning levels less than or equal to preset warning levels that are spatially associated with the risk source. Based on the preset time step and the real-time risk warning level of each sub-region, the risk diffusion range and the risk warning level of each sub-region are iteratively calculated using the risk diffusion rule model after each preset time step until the preset time is reached, generating a dynamic risk diffusion path prediction map.

[0075] In this embodiment, the risk warning level may include, but is not limited to, level 1, level 2, level 3, and level 4, with higher levels indicating higher risks. The preset warning level can be level 2. The preset time step and preset duration can be set according to the monitoring cycle or based on experience, and the preset duration is composed of multiple preset time steps added together sequentially.

[0076] The risk diffusion rule model takes the difference in risk warning levels at different times, the regional risk transmission weight, the spatial distance between sub-regions, and real-time hydrodynamic data within the target area as input parameters, and the diffusion rate of risk between different sub-regions as output parameters. The diffusion rate of risk between different sub-regions is positively correlated with the difference in risk warning levels and the regional risk transmission weight, and negatively correlated with the spatial distance between sub-regions.

[0077] The risk diffusion rule model is used to iteratively calculate the risk diffusion range and risk warning level of each sub-region after each preset time step until the preset time is reached, generating a dynamic risk diffusion path prediction map. This can be understood as follows: after each calculation of the risk diffusion range and risk warning level of each sub-region, in the GIS system, the real-time risk warning level of each sub-region is marked with differentiated colors (such as red → yellow → blue) using the spatial distribution of the sub-regions as the base map, and the risk diffusion path is drawn with arrowed lines (arrow thickness represents diffusion intensity, and line color represents time stage). After multiple preset time steps, a dynamic risk diffusion path prediction map is obtained.

[0078] In this embodiment, the online monitoring method for marine ecological early warning also includes: If, within a preset time period, the risk warning level of a sub-region is detected to be higher than the preset warning level, the sub-region will be identified as a new risk source and the risk diffusion path will be adjusted to generate a new dynamic risk diffusion path prediction map.

[0079] In one embodiment of this application, the risk warning result includes the risk warning level corresponding to each sub-region; Online monitoring of the target area based on risk warning results includes: Based on the risk warning level corresponding to each sub-region, a target monitoring strategy is selected from the adaptive monitoring strategy matching library, and differentiated online monitoring is carried out for different sub-regions based on the target monitoring strategy.

[0080] In this embodiment, the risk warning results include the risk warning level corresponding to each sub-region. Based on these different sub-region risk warning levels, a search and matching process is performed in a pre-built adaptive monitoring strategy matching library. This library stores various monitoring strategies corresponding to different risk levels, allowing each sub-region to accurately find a matching target monitoring strategy based on its own risk warning level. According to the selected target monitoring strategy, differentiated online monitoring is conducted for each sub-region to ensure the targeting and effectiveness of the monitoring.

[0081] This application's embodiment describes a method for selecting target monitoring strategies and implementing differentiated online monitoring based on risk warning levels. On one hand, it can rationally allocate monitoring resources according to the actual risk status of each sub-region. For sub-regions with risk warning levels higher than the preset warning level, a monitoring strategy that increases data collection frequency is adopted to ensure timely detection of potential problems. For sub-regions with risk warning levels lower than or equal to the preset warning level, a relatively simplified monitoring strategy is adopted to avoid resource waste and improve the efficiency of monitoring resource utilization. On the other hand, differentiated monitoring can more accurately capture the ecological changes in each sub-region, providing more accurate and detailed data support for subsequent marine ecological protection and risk response, and helping to improve the performance and reliability of the entire marine ecological early warning online monitoring system.

[0082] Corresponding to the marine ecological early warning online monitoring method in the above embodiments, Figure 4 This is a structural block diagram of a marine ecological early warning online monitoring device provided in one embodiment of this application. For ease of explanation, only the parts relevant to the embodiment of this application are shown. References Figure 4 The marine ecological early warning online monitoring device 40 includes: a data acquisition unit 41, a first data processing unit 42, a second data processing unit 43, and a risk prediction unit 44.

[0083] Among them, the data acquisition unit 41 is used to acquire satellite remote sensing data of the air-based region, perform feature enhancement on the satellite remote sensing data to obtain enhanced data, and extract spectral features from the enhanced data based on the time and spatial dimensions to obtain a multispectral feature dataset; the air-based region is the offshore area; The first data processing unit 42 is used to determine the shore-based region and the sea-based region based on the geographical location information in the multispectral feature dataset, and to obtain the first environmental data of the shore-based region and the second environmental data of the sea-based region. The shore-based region, the sea-based region and the air-based region all belong to the target region. The shore-based region, the sea-based region and the air-based region extend from the mainland to the open sea in sequence. The second data processing unit 43 is used to perform spatiotemporal alignment of the first environmental data and the second environmental data based on a dynamic spatiotemporal calibration algorithm to obtain the target environmental data. The risk prediction unit 44 is used to determine the risk warning result based on the target environmental data and the risk warning model, and to conduct online monitoring of the target area based on the risk warning result.

[0084] In one embodiment of this application, the first environmental data includes the first location coordinates corresponding to each shore-based monitoring station, and the second environmental data includes the second location coordinates corresponding to each sea-based device. The first data processing unit 42 is specifically used for: First environmental data of the shore-based area is acquired based on the first data acquisition frequency; Second environmental data for the sea-based area are acquired based on a second data acquisition frequency, where the first data acquisition frequency is greater than the second data acquisition frequency. Using the first environmental data as a benchmark, time series interpolation is performed on the second environmental data to obtain the second target environmental data; Import the first and second location coordinates into the GIS system to construct a spatial network; The target environment data is obtained based on the first environmental data, the second target environment data, and the spatial network.

[0085] In one embodiment of this application, both the first environmental data and the second target environmental data include multiple identical water quality monitoring factors. The first data processing unit 42 is specifically used for: For the same water quality monitoring factor, determine the overlapping monitoring periods for shore-based and sea-based areas; Based on the environmental data of the shore-based area corresponding to the overlapping monitoring period, the environmental data of the sea-based area corresponding to the overlapping monitoring period is calibrated using a calibration model constructed based on the random forest algorithm. The target environment data is obtained based on the first environmental data, the calibrated second target environment data, and the space network.

[0086] In one embodiment of this application, the first data processing unit 42 is specifically used for: The first environmental data and the second target environmental data are projected onto the preset grid cells of the spatial network, and the preset grid cells without data are selected as the target grid cells. The data in the target grid cell is completed based on the first environmental data and the second target environmental data; All data in the completed spatial network will be used as target environment data.

[0087] In one embodiment of this application, the sea-based region includes multiple sea-based sub-regions, and the shore-based region includes multiple shore-based sub-regions. The marine ecological early warning online monitoring device 40 also includes a regional risk transmission weight calculation unit; The regional risk transmission weight calculation unit is used for: Based on real-time hydrodynamic data and historical ecological risk database within the target area, the marine-based and shore-based sub-regions are dynamically classified to obtain multiple sub-regions corresponding to different environmental risk types. For each type of environmental risk, upstream and downstream labels are applied to multiple sub-regions, and regional risk transmission weights are assigned to these sub-regions after upstream and downstream labeling.

[0088] Risk prediction unit 44 is specifically used for: The target environmental data is divided into target environmental data corresponding to each sub-region based on multiple sub-regions. The environmental feature set of the target environmental data corresponding to each sub-region is extracted. The environmental feature set and regional risk transmission weight are input into the risk warning model to obtain the risk warning result.

[0089] In one embodiment of this application, the risk warning result includes the risk warning level corresponding to each sub-region. The risk prediction unit 44 is further specifically used for: Based on the correlation between risk warning level and the spatial location of sub-regions, a dynamic risk diffusion path prediction map is generated. Cross-regional risk prediction results were obtained based on dynamic risk diffusion path prediction maps and satellite remote sensing data.

[0090] In one embodiment of this application, the risk warning result includes the risk warning level corresponding to each sub-region; the risk prediction unit 44 is specifically used for: Based on the risk warning level corresponding to each sub-region, a target monitoring strategy is selected from the adaptive monitoring strategy matching library, and differentiated online monitoring is carried out for different sub-regions based on the target monitoring strategy.

[0091] See Figure 5 , Figure 5 This is a schematic block diagram of an electronic device provided according to an embodiment of this application. Figure 5 The electronic device 500 in this embodiment may include one or more processors 501, one or more input devices 502, one or more output devices 503, and one or more memories 504. The processors 501, input devices 502, output devices 503, and memories 504 communicate with each other via a communication bus 505. The memories 504 store computer programs, including program instructions. The processors 501 execute the program instructions stored in the memories 504. Specifically, the processors 501 are configured to invoke the program instructions to perform the functions of each module / unit in the above-described device embodiments, for example... Figure 4 The functions of the data acquisition unit 41, the first data processing unit 42, the second data processing unit 43, and the risk prediction unit 44 are shown.

[0092] It should be understood that, in the embodiments of this application, the processor 501 may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors, off-the-shelf programmable gate arrays or other programmable logic devices, etc.

[0093] Input device 502 may include a touchpad, etc., and output device 503 may include a display, speaker, etc.

[0094] The memory 504 may include read-only memory and random access memory, and provides instructions and data to the processor 501. A portion of the memory 504 may also include non-volatile random access memory.

[0095] In another embodiment of this application, a computer-readable storage medium is provided. This computer-readable storage medium stores a computer program, which includes program instructions. When executed by a processor, the program instructions implement all or part of the processes in the methods described above. Alternatively, the computer program can instruct related hardware to complete the process. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0096] The computer-readable storage medium can be an internal storage unit of the electronic device in any of the foregoing embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device. Furthermore, the computer-readable storage medium can include both internal and external storage units of the electronic device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0097] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for online monitoring and early warning of marine ecology, characterized in that, include: Satellite remote sensing data of an air-based region is acquired, and feature enhancement is performed on the satellite remote sensing data to obtain enhanced data. Spectral features are extracted from the enhanced data based on the time and spatial dimensions to obtain a multispectral feature dataset; the air-based region is the open ocean area. Based on the geographic location information in the multispectral feature dataset, the shore-based region and the sea-based region are determined, and the first environmental data of the shore-based region and the second environmental data of the sea-based region are obtained. The shore-based region, the sea-based region and the air-based region all belong to the target region. The shore-based region, the sea-based region and the air-based region extend from the mainland to the open sea in sequence. The target environment data is obtained by spatiotemporally aligning the first environment data and the second environment data based on a dynamic spatiotemporal calibration algorithm. Based on the target environmental data and the risk warning model, the risk warning result is determined, and the target area is monitored online based on the risk warning result.

2. The method as described in claim 1, characterized in that, The first environmental data includes the first location coordinates corresponding to each shore-based monitoring station, and the second environmental data includes the second location coordinates corresponding to each sea-based device. The process of acquiring first environmental data of the shore-based region and second environmental data of the sea-based region, and then aligning the first environmental data and the second environmental data in time and space based on a dynamic spatiotemporal calibration algorithm to obtain target environmental data includes: First environmental data of the shore-based area is acquired based on a first data acquisition frequency. Second environmental data of the sea-based area is acquired based on a second data acquisition frequency, wherein the first data acquisition frequency is greater than the second data acquisition frequency; Using the first environmental data as a reference, time series interpolation is performed on the second environmental data to obtain the second target environmental data; Import the first and second location coordinates into the GIS system to construct a spatial network; The target environment data is obtained based on the first environmental data, the second target environment data, and the spatial network.

3. The method as described in claim 2, characterized in that, Both the first environmental data and the second target environmental data include multiple identical water quality monitoring factors; obtaining the target environmental data based on the first environmental data, the second target environmental data, and the spatial network includes: For the same water quality monitoring factor, determine the overlapping monitoring periods for the shore-based area and the sea-based area; Based on the environmental data of the shore-based area corresponding to the overlapping monitoring period, the environmental data of the sea-based area corresponding to the overlapping monitoring period is calibrated using a calibration model, which is constructed based on the random forest algorithm. The target environment data is obtained based on the first environmental data, the calibrated second target environment data, and the space network.

4. The method as described in claim 2, characterized in that, The process of obtaining the target environment data based on the first environmental data, the second target environment data, and the spatial network includes: The first environmental data and the second target environmental data are projected onto the preset grid cells of the spatial network, and the preset grid cells without data are filtered out as target grid cells; The data in the target grid cell is completed based on the first environmental data and the second target environmental data; All data in the completed spatial network will be used as target environment data.

5. The method according to any one of claims 1 to 4, characterized in that, The sea-based region includes multiple sea-based sub-regions, and the shore-based region includes multiple shore-based sub-regions; the method further includes: Based on real-time hydrodynamic data and historical ecological risk database within the target area, the marine-based sub-region and the shore-based sub-region are dynamically classified to obtain multiple sub-regions corresponding to different environmental risk types. The upstream and downstream labels are performed on the multiple sub-regions corresponding to the different environmental risk types, and regional risk transmission weights are assigned to the multiple sub-regions after upstream and downstream labeling. The determination of risk warning results based on the target environment data and risk warning model includes: Based on the multiple sub-regions, the target environmental data is divided into target environmental data corresponding to each sub-region. The environmental feature set of the target environmental data corresponding to each sub-region is extracted. The environmental feature set and the regional risk transmission weight are input into the risk warning model to obtain the risk warning result.

6. The method as described in claim 5, characterized in that, The risk warning results include the risk warning level corresponding to each sub-region; After inputting the environmental feature set and the regional risk transmission weights into the risk warning model to obtain the risk warning result, the method further includes: Based on the correlation between the risk warning level and the spatial location of the sub-region, a dynamic risk diffusion path prediction map is generated. Cross-regional risk prediction results are obtained based on the dynamic risk diffusion path prediction map and the satellite remote sensing data.

7. The method as described in claim 5, characterized in that, The risk warning results include the risk warning level corresponding to each sub-region; The online monitoring of the target area based on the risk warning result includes: Based on the risk warning level corresponding to each sub-region, a target monitoring strategy is selected from the adaptive monitoring strategy matching library, and differentiated online monitoring is performed on different sub-regions based on the target monitoring strategy.

8. A marine ecological early warning online monitoring device, characterized in that, include: The data acquisition unit is used to acquire satellite remote sensing data of an air-based region, perform feature enhancement on the satellite remote sensing data to obtain enhanced data, and extract spectral features from the enhanced data based on the time and spatial dimensions to obtain a multispectral feature dataset; the air-based region is the open ocean area; The first data processing unit is used to determine the shore-based region and the sea-based region based on the geographical location information in the multispectral feature dataset, and to acquire the first environmental data of the shore-based region and the second environmental data of the sea-based region. The shore-based region, the sea-based region and the air-based region all belong to the target region, and the shore-based region, the sea-based region and the air-based region extend from the mainland to the open sea in sequence. The second data processing unit is used to perform spatiotemporal alignment of the first environmental data and the second environmental data based on a dynamic spatiotemporal calibration algorithm to obtain target environmental data. The risk prediction unit is used to determine the risk warning result based on the target environmental data and the risk warning model, and to perform online monitoring of the target area based on the risk warning result.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images