Coastal zone culture area multi-dimensional environment assessment method and cloud platform

Abstract

The invention relates to the technical field of environment assessment, in particular to a coastal zone culture area multi-dimensional environment assessment method and a cloud platform. The method comprises the following steps: continuously recording dissolved oxygen concentration, pH value, centigrade temperature, salinity unit, turbidity scattering unit, chlorophyll fluorescence intensity, ammonia nitrogen milligram per liter and nitrite milligram per liter, synchronously calling multi-source remote sensing images covering a culture area and an adjacent water area, extracting remote sensing reflectivity data through image correction and atmospheric correction, and calculating to obtain water color parameters. According to the method, the water quality sensor array is deployed at the key point of the culture area, continuous multi-parameter environmental data acquisition is carried out, multi-source remote sensing images are synchronously integrated, water color information is extracted in real time, and the timeliness and data accuracy of culture water area environmental monitoring are improved; trend decomposition and dynamic baseline construction are carried out based on the data sequence, so that the environmental fluctuation evaluation of the breeding area is more objective and accurate.

Description

Technical Field

[0001] The present invention relates to the field of environmental assessment technology, and in particular to a multi-dimensional environmental assessment method and cloud platform for coastal aquaculture areas. Background Art

[0002] The field of environmental assessment technology mainly involves the use of multidisciplinary integrated methods to systematically investigate, monitor and analyze factors such as the ecosystem, environmental quality, resource utilization and human activities in a specific area in order to assess its environmental status, changing trends and potential risks.

[0003] Existing technologies struggle to effectively identify the propagation paths of environmental anomalies, making it difficult to accurately and promptly determine the source of pollution and the scope of its impact. This leads to significant blind spots in environmental risk assessments, inefficient tracing of ecological and environmental issues, and risk management, increasing the cost of environmental remediation. For example, when water quality deteriorates in aquaculture areas, a lack of data support and propagation path analysis often makes it difficult to quickly locate the source and spread of the pollution, delaying environmental remediation efforts and leading to lower product quality and increased economic losses. Therefore, improvements are needed. Summary of the Invention

[0004] The purpose of the present invention is to solve the shortcomings of the existing technology and propose a multi-dimensional environmental assessment method and cloud platform for coastal aquaculture areas.

[0005] In order to achieve the above objectives, the present invention adopts the following technical solution, a multi-dimensional environmental assessment method for aquaculture areas in coastal zones, comprising the following steps: Water quality sensor arrays are deployed at key locations in the aquaculture area to continuously record dissolved oxygen concentration, pH value, temperature in degrees Celsius, salinity units, turbidity scattering units, chlorophyll fluorescence intensity, milligrams per liter of ammonia nitrogen, and milligrams per liter of nitrite. Multi-source remote sensing images covering the aquaculture area and adjacent waters are simultaneously retrieved. Through image correction and atmospheric correction, remote sensing reflectance data is extracted and water color parameters are calculated. The sensor array time series data and remote sensing water color parameters are integrated to generate comprehensive water environment monitoring data. Based on the sensor parameters in the comprehensive water environment monitoring data, each parameter time series is decomposed into trend terms, seasonal terms, and residual terms, and statistical control limits are set for the residual terms to construct a dynamic normal fluctuation range for each parameter, establish a dynamic baseline for water quality parameters, and combine remote sensing water color parameters to segment different water bodies by setting a water color index threshold. Then, edge points are connected and smoothed to identify the spatial distribution and boundaries of different water masses and establish a spatial distribution map of water masses; The real-time sensor parameter values ​​collected are compared with the dynamic baselines of the corresponding water quality parameters through difference calculation and threshold comparison. When the difference exceeds the preset control limit, it is marked as a potential anomaly point. The duration of the anomaly point is analyzed, and events in which single or multiple parameters deviate from the normal range are identified. The water mass boundary position and water color parameter changes in the water mass spatial distribution map are continuously monitored, and the daily pixel displacement of the boundary and the change in the water color index are recorded to obtain environmental spatiotemporal variation information.

[0006] Preferably, the method further comprises: Analyze the timing and spatial linkage of data anomalies in the environmental spatiotemporal variation information, determine the direction of anomaly propagation by calculating the time difference and spatial distance of anomalies in sensors at different locations, combine the migration direction and speed of water mass boundaries in the water mass spatial distribution map, determine the upstream impact area through hydrodynamic principles, trace the source direction of the abnormal disturbance and the potential impact sector, evaluate the degree of impact on the ecological environment of coastal waters, and generate preliminary tracing conclusions on the source of environmental anomalies.

[0007] Preferably, the method further comprises: Integrate the preliminary tracing conclusions of the environmental anomaly events and the spatial distribution map of the water masses, merge the anomaly levels of various parameters, impact ranges, and water mass stability data, compare with the survival condition thresholds of the aquaculture species and the lower limit of dissolved oxygen and the upper limit of ammonia nitrogen in the regional environmental management objectives, evaluate the current environmental status and potential risks of the aquaculture area, divide the environmental status assessment levels into different levels according to the risk matrix judgment rules, and output the aquaculture environment assessment results.

[0008] Preferably, the steps for acquiring the comprehensive water environment monitoring data are: Water quality sensor arrays were deployed at key locations in the aquaculture area, with a sampling frequency set to once every ten minutes. Dissolved oxygen concentration, pH value, Celsius temperature, salinity units, turbidity scattering units, chlorophyll fluorescence intensity, ammonia nitrogen milligrams per liter, and nitrite milligrams per liter were continuously recorded to obtain the sensor's original observation sequence. Based on the geographic coordinates of the aquaculture area, multi-source remote sensing images covering the aquaculture area and adjacent waters were retrieved. The image data were orthorectified and geo-registered, and atmospheric correction was performed using radiometric calibration and atmospheric condition parameter input. Furthermore, remote sensing reflectance data were extracted by combining target visible and near-infrared bands, and water color parameters were calculated to establish a remote sensing image water color index set. Based on the sensor original observation sequence and the remote sensing image water color index set, the geographic coordinates of each sensor in the sensor original observation sequence are spatially matched with the geographic grid of the image pixels in the remote sensing image water color index set, and the sensor data and the remote sensing image daily data are timestamp aligned using linear interpolation or time window averaging method to generate comprehensive water environment monitoring data.

[0009] Preferably, the steps for obtaining the water cluster spatial distribution map are: Based on the sensor parameters in the comprehensive water environment monitoring data, the corresponding parameter continuous time series data for the past year are extracted from the historical database, and a Gaussian smoothing filter is used to filter out random noise from each parameter time series. Then, the least squares method is used to fit the trend change straight line or curve component, and the Fourier transform is used to identify and separate the main periodic seasonal fluctuation component and the remaining random disturbance component to obtain the parameter sequence decomposition component; Based on the residual sequence of each parameter in the decomposition component of the parameter sequence, by calculating the mean and standard deviation values ​​of the residual sequence, setting the mean plus or minus three times the standard deviation as the upper and lower thresholds of statistical control, constructing the dynamic normal fluctuation reference range of each parameter, establishing a dynamic baseline of water quality parameters, and extracting corresponding remote sensing water color parameters from comprehensive water environment monitoring data. According to the numerical domain characteristics of different water color parameters, multiple threshold intervals are set, and each pixel of the remote sensing image is classified and divided to obtain a water quality baseline and a preliminary water body segmentation area; Based on the water quality baseline and the classification results of water pixels in the preliminary water body segmentation area, the Sobel or Canny edge detection operator is used to scan the image, identify the intersection pixels between water pixel areas of different categories, connect the intersection pixels with spatial adjacency to form an initial vector boundary line, and then apply the Douglas-Peucker algorithm to iteratively simplify the initial boundary line to establish a water mass spatial distribution map.

[0010] Preferably, the steps for acquiring the environmental spatiotemporal variation information are: The real-time sensor parameter values ​​are compared with the upper and lower limits of the normal fluctuation range of each parameter in the dynamic baseline of the corresponding water quality parameter on a point-by-point basis, and the deviation between the current parameter measurement value and the baseline range boundary is calculated. When the deviation exceeds the pre-set allowable fluctuation threshold, the parameter data point at that moment is marked as a potential abnormal state, forming a potential water quality abnormal point sequence; Based on the potential water quality anomaly point sequence, by searching whether the data points marked as potential anomaly states in the sequence appear more than three times consecutively on the time axis, identifying abnormal events in which a single parameter or multiple combined parameters deviate from the normal range, and at the same time, by comparing with the water mass spatial distribution map, extracting the coordinate set of each water mass boundary pixel and the water color parameter of the corresponding pixel at the current moment, and obtaining the identified water quality event and boundary pixel set; Based on the identified water quality events and boundary pixel sets, by comparing the boundary pixel coordinate sets of the same identified water mass recorded in the current monitoring period with the previous monitoring period, the north-south and east-west displacement components of each corresponding boundary point in the geographic coordinate system are calculated, and then the total displacement vector is synthesized. The increase or decrease in the water color parameters of the area corresponding to the pixel is recorded to obtain environmental spatiotemporal variation information.

[0011] Preferably, the obtaining step of the environmental abnormal event preliminary tracing conclusion is: The data abnormality occurrence time sequence and spatial linkage in the environmental spatiotemporal variation information are analyzed, the identified abnormal events with consistent types and same starting time at different geographical location sensor monitoring points are screened, the time difference value of the event occurrence time is calculated, the GPS coordinate of each sensor point is measured to measure the straight-line geographical distance, the apparent direction and average rate of the abnormal signal propagation in the water body are determined through the time difference and distance ratio, and the abnormal propagation characteristic parameter set is obtained; Based on the main propagation direction of the abnormal signal indicated by the abnormal propagation characteristic parameter set, combined with the dominant direction vector and average moving speed data of the water mass boundary migration within the abnormal occurrence time extracted from the water mass spatial distribution map, the starting upstream potential influence area range of the abnormal disturbance is judged according to the regional water body natural flow characteristics, and the potential abnormal source area is divided. Based on the potential abnormal source area division, combined with the sensor points and water mass boundary variation range affected by the abnormal event recorded in the environmental spatiotemporal variation information, the negative influence degree of the abnormal event on the benthic biological community and the plankton number of the target water area in the divided area is evaluated with reference to the typical diffusion coefficient of the known pollutants under the same hydrological condition stored in the database, and the environmental abnormal event preliminary tracing conclusion is generated.

[0012] Preferably, the obtaining step of the aquaculture environment evaluation result is: The environmental abnormal event preliminary tracing conclusion is integrated, the data is subjected to dimensionless conversion processing by using maximum minimum normalization, and the standardized multidimensional environmental state data is formed. Based on the standardized multidimensional environmental state data, the standardized values of the water quality parameters associated with the breeding species are extracted, the current monitoring values are compared with the suitable survival condition threshold table of the breeding species, the compliance of each monitoring point of the breeding area is judged, the potential ecological environment risk level is compared, the environmental risk compliance degree judgment value is obtained.

[0013] Preferably, the obtaining step of the aquaculture environment evaluation result further includes: based on the environmental risk compliance degree judgment value, outputting the aquaculture environment evaluation result.

[0014] The application also provides an environmental evaluation cloud platform, which comprises a memory and a processor, and the memory stores a computer program.

[0015] Compared with the prior art, the application has the advantages and positive effects that: The present invention deploys water quality sensor arrays at key points in aquaculture areas to collect continuous multi-parameter environmental data, and synchronously integrates multi-source remote sensing images to extract water color information in real time, thereby improving the timeliness and data accuracy of aquaculture water environment monitoring. Trend decomposition and dynamic baseline construction are performed based on data sequences to make the assessment of environmental fluctuations in aquaculture areas more objective and accurate. Water body segmentation and spatial map construction are performed on remote sensing images to achieve real-time correlation analysis between the spatial distribution of water masses and environmental anomalies, which is conducive to quickly capturing environmental changes and pollution diffusion trends. By analyzing the spatiotemporal linkage characteristics of abnormal events and the direction of hydrodynamic propagation, the source of abnormal events can be quickly traced and tracked, effectively reducing the blindness and uncertainty of environmental risk assessment. In addition, risk level assessment is performed by integrating environmental status with the survival condition thresholds of aquaculture species and regional management target standards, which can achieve more accurate and detailed classification and evaluation of the environmental status of the aquaculture area, making management decisions more scientific. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 Schematic diagram of the steps of the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0018] See also Figure 1 The present invention provides a technical solution, a multi-dimensional environmental assessment method for a coastal aquaculture area, comprising the following steps: Water quality sensor arrays are deployed at key locations in the aquaculture area to continuously record dissolved oxygen concentration, pH value, temperature in degrees Celsius, salinity units, turbidity scattering units, chlorophyll fluorescence intensity, milligrams per liter of ammonia nitrogen, and milligrams per liter of nitrite. Multi-source remote sensing images covering the aquaculture area and adjacent waters are simultaneously retrieved. Through image correction and atmospheric correction, remote sensing reflectance data is extracted and water color parameters are calculated. The sensor array time series data and remote sensing water color parameters are integrated to generate comprehensive water environment monitoring data. Based on the sensor parameters in the comprehensive water environment monitoring data, the trend term, seasonal term, and residual term of each parameter time series are decomposed, and statistical control limits are set for the residual term. The dynamic normal fluctuation range of each parameter is constructed, and the dynamic baseline of water quality parameters is established. In combination with remote sensing water color parameters, different water bodies are segmented by setting the water color index threshold. Then, the edge points are connected and smoothed to identify the spatial distribution and boundaries of different water masses and establish a water mass spatial distribution map. Perform difference calculations and threshold comparisons on the real-time sensor parameter values ​​and the corresponding dynamic baselines of water quality parameters. When the difference exceeds the preset control limit, it is marked as a potential anomaly. The duration of the anomaly is analyzed to identify events where a single or multiple parameters deviate from the normal range. The water mass boundary position and water color parameter changes in the water mass spatial distribution map are continuously monitored. The daily pixel displacement of the boundary and the change in the water color index are recorded to obtain information on the spatiotemporal variation of the environment. Analyze the timing and spatial linkage of data anomalies in environmental spatiotemporal variation information. By calculating the time difference and spatial distance between anomalies at different sensor locations, determine the direction of anomaly propagation. Combined with the migration direction and speed of water mass boundaries in the water mass spatial distribution map, determine the upstream impact area based on hydrodynamic principles, trace the source direction of the anomaly disturbance and the potential impact sector, assess the degree of impact on the ecological environment of coastal waters, and generate preliminary tracing conclusions on the source of environmental anomalies. Integrate the preliminary tracing conclusions of environmental anomalies and the spatial distribution map of water masses, merge the anomaly levels of various parameters, impact range, and water mass stability data, compare them with the survival condition thresholds of aquaculture species and the lower limit of dissolved oxygen and upper limit of ammonia nitrogen in the regional environmental management objectives, evaluate the current environmental status and potential risks of the aquaculture area, divide different environmental status assessment levels according to the risk matrix judgment rules, and output the aquaculture environment assessment results.

[0019] The steps for obtaining comprehensive water environment monitoring data are as follows: Water quality sensor arrays were deployed at key locations in the aquaculture area, with a sampling frequency set to once every ten minutes. Dissolved oxygen concentration, pH value, Celsius temperature, salinity units, turbidity scattering units, chlorophyll fluorescence intensity, ammonia nitrogen milligrams per liter, and nitrite milligrams per liter were continuously recorded to obtain the sensor's original observation sequence. Based on the geographic coordinates of the aquaculture area, multi-source remote sensing images covering the aquaculture area and adjacent waters were retrieved. The image data were orthorectified and geo-registered, and atmospheric correction was performed using radiometric calibration and atmospheric condition parameter input. Furthermore, remote sensing reflectance data were extracted by combining target visible and near-infrared bands, and water color parameters were calculated to establish a remote sensing image water color index set. Based on the original sensor observation sequence and the remote sensing image water color index set, the geographic coordinates of each sensor in the original sensor observation sequence are spatially matched with the geographic grid of the image pixels in the remote sensing image water color index set, and the sensor data and the remote sensing image daily data are timestamp aligned using linear interpolation or time window averaging methods to generate comprehensive water environment monitoring data.

[0020] Specifically, water quality sensor arrays are deployed at key points in the aquaculture area. Specifically, representative locations in the aquaculture area are selected as key points, such as the central area of ​​the aquaculture raft, the dense area of ​​the aquaculture boxes, near the water inlet and outlet where water exchange is relatively frequent, and sensitive areas where water quality anomalies have occurred in the past. At least one set of sensor combinations containing all specified parameters is deployed at each key point. The sensor probes need to be calibrated and maintained in accordance with standard operating procedures to ensure data accuracy. The sampling frequency of the data acquisition controller is then set to once every ten minutes. The basis for setting this frequency is the rate of change of key water quality parameters (such as dissolved oxygen) in the aquaculture water body under sudden circumstances, and the response time requirements of the early warning system. For example, under normal circumstances, a rapid drop in dissolved oxygen may cause stress to aquaculture organisms within tens of minutes. Therefore, a ten-minute sampling interval can provide enough data points to capture the trend of changes when the parameters change significantly. The system will automatically follow this frequency instruction and continuously measure and record the specific values ​​of dissolved oxygen concentration, pH value, Celsius temperature, salinity unit, turbidity scattering unit, chlorophyll fluorescence intensity, ammonia nitrogen milligrams per liter and nitrite milligrams per liter through each sensor. These values ​​are directly transmitted and stored in the form of digital signals to form raw data records containing timestamps, point information and measurement values ​​of each parameter, thereby obtaining the original observation sequence of the sensor.

[0021] Based on the geographical coordinate range of the aquaculture area, the geographical coordinate range is first determined, for example, 120°30'00" to 120°30'30" east longitude and 22°15'00" to 22°15'30" north latitude. Then, multi-source remote sensing images covering this aquaculture area and adjacent waters extending outward, for example, at least 5 kilometers, are retrieved from satellite data providers or public data archives. Images with higher spatial resolution (for example, 10 to 30 meters) and shorter revisit periods (for example, 5 to 16 days) are preferred to obtain sufficiently detailed and timely water meter information. Then, the obtained raw image data are preprocessed. First, orthorectification is performed, using digital elevation models (DEM) and satellite orbit parameters to eliminate geometric distortion caused by terrain undulations and sensor viewing angles. Then, precise georeferencing is performed, selecting ground control points (GCPs) or using the high-precision geolocation information of the image itself to correct the image to a unified geographic coordinate system (such as WGS84) to ensure the spatial consistency of each scene image and subsequent analysis results. Then, radiometric calibration is performed to convert the original digital quantization value (DN value) recorded by the sensor into the reflectance or radiance of the top atmosphere with actual physical meaning. This process requires the use of the calibration coefficient of the sensor itself, and then Atmospheric correction is performed to eliminate or reduce the interference of atmospheric molecules, aerosol scattering and absorption on water body information. For example, the 6S model can be used. The input atmospheric condition parameters, such as aerosol optical depth, water vapor content, ozone concentration, etc., can be obtained from synchronous ground observation stations, or estimated using the image's own band information, such as the dark pixel method to estimate aerosol optical depth. The initial reference value can be set to 0.1 and adjusted according to image quality and atmospheric conditions. Then, by selecting the target visible light band (for example, blue light about 450-515nm, green light about 525-600nm, red light about 630-690nm), the atmospheric parameters can be corrected. The remote sensing reflectance data are extracted by combined operations of the blue-green band (nm) and the near-infrared band (for example, about 750-900nm), such as the band ratio method and the normalized difference index method, and water color parameters are calculated according to specific algorithms. For example, the transparency-related index is calculated using the ratio of the blue-green band, or the suspended matter concentration-related index is calculated using the combination of red and near-infrared bands, or a specific band combination algorithm for chlorophyll concentration is adopted, such as calculating a variant of the normalized difference vegetation index (NDVI) based on the reflectance of the red and near-infrared bands to estimate the chlorophyll concentration. The water color parameters obtained by these calculations together constitute the remote sensing image water color index set.

[0022] Based on the sensor original observation sequence and remote sensing image water color index set obtained in the above steps, the two are first spatially matched. Specifically, the precise geographic coordinates (latitude and longitude) of each water quality sensor are read from the sensor original observation sequence, and the geographic grid information of the image pixels is extracted from the remote sensing image water color index set. For each sensor, the nearest neighbor method is used to find the pixel with the closest spatial position on the remote sensing image at its geographic coordinate position. Alternatively, if the sensor monitoring range is considered to be larger than a single pixel, a window of, for example, 3x3 pixels or 5x5 pixels can be defined, with the center being the pixel closest to the sensor. The average value of the water color parameters corresponding to all pixels in the window is calculated as the remote sensing water color parameter value matching the sensor. The selection of this window size is based on the spatial resolution of the remote sensing image and the actual water area represented by the sensor. For example, for a 30-meter resolution image, a 3x3 window represents an area of ​​90 meters x 90 meters. It is necessary to ensure that this selection is consistent with the representativeness of the key aquaculture points and complete the spatial matching. After time matching, the sensor data and the daily remote sensing image data are timestamp aligned. Since the sampling frequency of sensor data (every ten minutes) is much higher than the acquisition frequency of remote sensing images (usually once every few days), for the acquisition date and time of each remote sensing image, the sensor data closest to the acquisition time of the remote sensing image is screened out from the original sensor observation sequence. Alternatively, a more common method is to use the time window averaging method, that is, define a time window with the remote sensing image acquisition time as the center, for example, push 2 hours before and after to form a 4-hour time window, and then calculate the average of all sensor readings in the time window as the sensor parameter value aligned in time with the remote sensing image. In special cases, such as when the sensor data before and after the remote sensing image is acquired is missing but still needs to be used, the linear interpolation method can be used to estimate the sensor parameter value at that time point based on the most recent valid sensor reading before and after the image acquisition time point. For example, if the image acquisition time is T, the most recent valid sensor reading before and after is In time , In time , then interpolation ,Through the above-mentioned spatial proximity matching and timestamp alignment processing, the ,observation data of each sensor are integrated with the ,remotely sensed water color parameters of its corresponding time and ,spatial position, forming a record containing the timestamp, geographical location, ,parameters of each on-site sensor and the corresponding remotely ,sensed water color parameters, and thus generating comprehensive ,water environment monitoring data.

[0023] The steps to obtain the water mass spatial distribution map are: Based on the sensor parameters in the comprehensive water environment monitoring data, the corresponding parameter continuous time series data of the past year were extracted from the historical database. A Gaussian smoothing filter was used to filter out random noise for each parameter time series. The least squares method was then used to fit the trend change straight line or curve components, and the Fourier transform was used to identify and separate the main periodic seasonal fluctuation components and the remaining random disturbance components to obtain the parameter series decomposition components. Based on the residual sequence of each parameter in the parameter sequence decomposition component, by calculating the mean and standard deviation of the residual sequence, setting the mean plus or minus three times the standard deviation as the upper and lower thresholds of statistical control, constructing the dynamic normal fluctuation benchmark range of each parameter, establishing a dynamic baseline of water quality parameters, and extracting the corresponding remote sensing water color parameters from the comprehensive water environment monitoring data. According to the numerical domain characteristics of different water color parameters, multiple threshold intervals are set, and each pixel of the remote sensing image is classified and divided to obtain the water quality baseline and preliminary water body segmentation area; Based on the water quality baseline and the classification results of water pixels in the preliminary water body segmentation area, the Sobel or Canny edge detection operator is used to scan the image, identify the intersection pixels between water pixel areas of different categories, connect the intersection pixels with spatial adjacency to form an initial vector boundary line, and then apply the Douglas-Peucker algorithm to iteratively simplify the initial boundary line to establish a water mass spatial distribution map.

[0024] Specifically, based on the sensor parameters in the comprehensive water environment monitoring data, such as dissolved oxygen concentration and pH value, first extract the continuous time series data of at least one year (for example, from May 1, 2023 to April 30, 2024) for each sensor parameter from the historical database. If there is a small amount of missing data, linear interpolation or neighboring mean filling methods can be used to ensure data continuity. Then, a Gaussian smoothing filter is applied to the time series data of each parameter to filter out random noise. The window size of the Gaussian smoothing filter is and standard deviation Need to be set, for example, for data with a sampling interval of 10 minutes, the window size Can be set to 5 data points (i.e. 50 minutes), standard deviation It can be set to one-sixth to one-third of the window size, e.g. The setting of these values ​​is based on the prior knowledge of the normal fluctuation characteristics and noise level of the parameters or determined through experiments, so as to retain the real short-term fluctuation information while effectively smoothing the noise. After Gaussian smoothing, the least squares method is used to fit the smoothed parameter time series to extract the trend change component. You can try to fit a linear trend. and quadratic trend , by comparing the coefficient of determination of the two ( value) to select a better fitting method, for example, if the quadratic curve The linear trend If the value is higher than 0.1, the quadratic curve is selected, otherwise the linear trend is selected. After the trend model is selected, the model coefficients (such as a, b or a, b, c) are calculated using the least squares method to obtain the trend change straight line or curve component. Then, the extracted trend component is subtracted from the smoothed sequence. The Fourier transform is applied to the obtained detrended sequence to calculate its power spectrum density. The frequency points with significant energy concentration in the power spectrum are identified. These frequency points correspond to the main periodic fluctuations in the sequence, such as the daily change period (corresponding to the frequency 1 / 24 hour) -1 ), tide-related periods (such as 1 / 12.42 hours -1 ), etc., by setting an energy threshold, for example, selecting the main frequency components whose cumulative energy accounts for more than 80% of the total energy, or selecting the frequency components whose power exceeds a specific multiple (for example, 3 times) of the average power spectral density, these main periodic seasonal fluctuation components are separated, and after subtracting the trend component and the identified main periodic seasonal fluctuation components from the original smooth sequence, the remaining part is the residual random disturbance component. Finally, for each sensor parameter, the three parameter sequence decomposition components of its trend term, seasonal term and residual term (that is, the residual random disturbance component) are obtained.

[0025] Based on the residual sequence of each parameter in the decomposition component of each parameter sequence obtained in the previous step, such as the residual sequence of dissolved oxygen concentration parameter, the residual sequence of pH value parameter, etc., for each parameter residual sequence, first calculate its mean value and standard deviation. For example, for the residual sequence of pH value , its mean , its standard deviation , and then set the upper and lower thresholds of statistical control based on these statistical values. The specific method is to use the mean plus or minus three times the standard deviation, that is, the upper control threshold and lower control threshold The choice of three times the standard deviation is based on statistical process control theory. For example, when the data is approximately normally distributed, this range can contain about 99.7% of the random fluctuation data points, which will be determined by the trend term, seasonal term and the residual term. The parameter value interval composed of the range is defined as the dynamic normal fluctuation benchmark range of the parameter. This dynamically changing benchmark range is the dynamic baseline of the water quality parameter. Next, the remote sensing water color parameters corresponding to each remote sensing image are extracted from the previously generated comprehensive water environment monitoring data, such as the chlorophyll a concentration estimate, the suspended matter concentration estimate or a specific water color index (such as the Forel-Ule index). According to the numerical domain empirical characteristics of different water color parameters or relevant water quality standards, multiple threshold intervals are set to classify each pixel in the remote sensing image. For example, for the chlorophyll a concentration parameter, according to the regional water eutrophication evaluation standard, the threshold interval can be set as: less than 5 micrograms per liter is oligotrophic; If the chlorophyll a concentration is estimated to be 12 micrograms per liter, the pixel is classified as a mesotrophic water body. Similarly, for other water color parameters such as suspended solids concentration, threshold intervals such as less than 10 milligrams per liter for low suspended solids, 10 to 50 milligrams per liter for medium suspended solids, and greater than 50 milligrams per liter for high suspended solids can also be set. The specific values ​​of these thresholds are usually based on local water quality standards, statistical analysis results of historical monitoring data, or the basis for the division of this type of water body in relevant literature. By performing this threshold judgment and classification on all pixels based on their water color parameter values, the dynamic baseline of each water quality parameter and the preliminary water body segmentation area of ​​the remote sensing image can be obtained.

[0026] Based on the water quality baseline obtained in the above steps and the classification result of the water body pixels in the preliminary water body segmentation area, the result is a raster image that divides each pixel of the remote sensing image into different categories (such as oligotrophic, mesotrophic, eutrophic water bodies, or low turbidity, medium turbidity, high turbidity water bodies) according to the water color parameter threshold. Then, in order to identify the boundaries between different categories of water body areas, the Canny edge detection operator is selected to scan and process this classified raster image. The Canny operator is selected because it has better noise suppression ability and edge positioning accuracy than operators such as Sobel. The execution of the Canny operator includes steps such as Gaussian smoothing, gradient calculation, non-maximum suppression and double threshold hysteresis connection. The kernel size of the Gaussian smoothing is set to 5x5, for example, and the standard deviation is 0. Set to 1.4, and the high and low threshold settings are adjusted according to the image contrast and the desired edge detail level, for example, the high threshold The gradient amplitude can be set to 0.2 times the maximum gradient value, low threshold It can be set to 0.4 times the high threshold, that is, The specific value is determined by testing the sample image and evaluating the edge extraction effect. For example, if the maximum gradient is 250, then is 50, is 20. After processing with the Canny operator, a binary edge image is obtained, in which the pixels marked as edges are the boundary pixels between different categories of water pixel areas. Subsequently, these identified boundary pixels are connected by tracing the pixels with the same edge mark and spatial adjacency in the eight or four neighborhoods in space, and connecting them to form continuous initial vector boundary segments. These segments represent the initial boundaries between different water masses. Since these initial boundary lines may contain a large number of redundant vertices, they are not smooth enough and the data volume is large, so they need to be simplified. The Douglas-Peucker algorithm is applied to iteratively simplify these initial vector boundary lines. The tolerance parameter (tolerance) of the Douglas-Peucker algorithm is For example, if the spatial resolution of the remote sensing image is 10 meters, you can set It is set to 1.5 times the pixel size, that is, 15 meters. This means that if the vertical distance from a point on the boundary line to the line segment connected by the two points before and after it is less than 15 meters, the point is considered a redundant point and is removed. The selection of this tolerance value requires a balance between the smoothness of the boundary and the degree of detail retention. After simplification, a smoother vector boundary line with less data volume is obtained. The collection of different water body category areas and their boundary lines obtained after edge detection, connection and smoothing simplification together constitutes the spatial distribution map of water masses.

[0027] The steps for obtaining environmental spatiotemporal variation information are as follows: The real-time sensor parameter values ​​are compared with the upper and lower limits of the normal fluctuation range of each parameter in the dynamic baseline of the corresponding water quality parameter on a point-by-point basis, and the deviation between the current parameter measurement value and the baseline range boundary is calculated. When the deviation exceeds the pre-set allowable fluctuation threshold, the parameter data point at that moment is marked as a potential abnormal state, forming a potential water quality abnormal point sequence; Based on the sequence of potential water quality anomaly points, the system searches for data points marked as potential anomaly states in the sequence and finds whether they appear more than three times consecutively on the time axis. This system identifies abnormal events in which a single parameter or a combination of multiple parameters deviates from the normal range. At the same time, the system compares the spatial distribution map of water masses and extracts the coordinate set of the boundary pixels of each water mass at the current moment and the water color parameters of the corresponding pixels to obtain the identified water quality events and boundary pixel sets. Based on the identified water quality events and boundary pixel sets, by comparing the boundary pixel coordinate sets of the same identified water mass recorded in the current monitoring period with the previous monitoring period, the north-south and east-west displacement components of each corresponding boundary point in the geographic coordinate system are calculated, and then the total displacement vector is synthesized. The increase or decrease in the water color parameters of the area corresponding to the pixel is recorded to obtain environmental spatiotemporal variation information.

[0028] Specifically, the real-time collected sensor parameter values, such as dissolved oxygen concentration of 5.8 mg / L and pH value of 7.9, are compared with the upper and lower limits of the normal fluctuation range of the parameter at the current moment in the previously constructed dynamic baseline of the corresponding water quality parameter. For example, the dynamic baseline predicts that the normal fluctuation range of dissolved oxygen concentration at the current moment is 6.5 mg / L to 7.5 mg / L. The values ​​are compared hourly (for example, sensor readings every ten minutes) and point-by-point (for each sensor). Then, the deviation of the current parameter measurement value from the baseline range boundary is calculated. If the current measurement value is within the normal fluctuation range, the deviation is zero or no special processing is performed. If the current measurement value exceeds the range, for example, the dissolved oxygen concentration of 5.8 mg / L is lower than the lower limit of 6.5 mg / L, the deviation is 0.7 mg / L (calculated as 6.5 minus 5.8). When the calculated deviation, such as 0.7 mg / L, exceeds one The setting of this allowable fluctuation threshold is based on the parameter characteristics and management requirements. For example, for dissolved oxygen, the threshold can be set to 0.2 mg per liter, and for pH value, it can be set to 0.1 unit. These values ​​are based on experience, taking into account the measurement error of the sensor itself and the small normal disturbance of the water body to avoid triggering potential anomalies too frequently. For example, the dynamic baseline range of a parameter is [10, 20], and its allowable fluctuation threshold is set to 1. If the real-time value is 8.5, then its deviation from the lower limit of 10 is 1.5. Because 1.5 is greater than 1, it is marked as a potential abnormal state. If the real-time value is 9.5, the deviation is 0.5, which is less than 1, it is not marked. All the marked data points of potential abnormal state, together with their timestamps, sensor locations and parameter types, form a potential water quality abnormal point sequence.

[0029] Based on the potential water quality anomaly point sequence formed in the previous link, the sequence is searched to check whether the data points marked as potential anomaly states appear continuously on the time axis for the same parameter of the same sensor for a preset number of times or more. The preset number is usually set to three times. For example, if the dissolved oxygen parameter is marked as a potential anomaly state at three consecutive sampling times of 10:00, 10:10, and 10:20 at point A sensor, the system determines that an abnormal event has occurred in which the dissolved oxygen parameter deviates from the normal range. If it only appears at 10:00 and 10:20, and the middle 10:10 is normal, the condition of three consecutive times is not met. The setting of "three times" is to filter out isolated pseudo-anomalies that may be caused by instantaneous interference or data transmission errors, and to improve the reliability of abnormal event identification. In addition to single parameters, the system will also check whether there are multiple combined parameters that meet their respective continuous anomalies at the same time. Conditions, for example, if dissolved oxygen is lower than its dynamic baseline lower limit for three consecutive times and exceeds the allowable fluctuation threshold, and ammonia nitrogen is higher than its dynamic baseline upper limit for three consecutive times and exceeds the allowable fluctuation threshold, it will be identified as a more serious complex abnormal event. Once such an abnormal event in which a single parameter or multiple combined parameters deviate from the normal range is identified, the system will record the start time, duration, parameters involved and sensor location of the event, and synchronously retrieve the water mass spatial distribution map corresponding to the time when the abnormal event occurred (or the closest remote sensing image acquisition time), and extract the boundary pixel coordinate set of each water mass divided on the map from the map, as well as the water color parameter values ​​(such as chlorophyll concentration, turbidity, etc.) of these boundary pixels themselves and the pixels inside their adjacent water masses. By integrating this information, an identified water quality event and a boundary pixel set containing the details of the identified water quality event and the boundary conditions of the water mass when the event occurred are obtained.

[0030] Based on the identified water quality events and boundary pixel sets obtained in the above steps, the system further analyzes the spatial dynamic changes of water masses by comparing the boundary pixel coordinate sets of the same spatially continuous water mass recorded in the current monitoring cycle (for example, the spatial distribution map of water masses corresponding to the remote sensing image acquired and processed today) with the boundary pixel coordinate sets of the same spatially continuous water mass recorded in the previous monitoring cycle (for example, the spatial distribution map of water masses corresponding to the remote sensing image acquired and processed today). The same water mass is usually identified based on the proximity of its centroid position in the images of the two cycles (for example, the difference does not exceed 30% of the average water mass diameter) and the area change rate (for example, not more than 20%). If it is determined to be the same water mass, its boundaries are matched. For example, the boundary lines of the water masses in the two cycles are sampled to obtain a certain number of boundary points. Then, based on the spatial proximity of the points and the consistency of the boundary line direction, the corresponding boundary points are found. For each pair of successfully matched corresponding boundary points, their geographic coordinates in the two monitoring cycles (for example, and ), calculate its north-south displacement component in the geographic coordinate system ( ) and the east-west displacement component ( ), and then the total displacement vector is obtained by vector synthesis calculation, and its magnitude is And the direction is Determine, at the same time, for pixels near the boundary of the water mass or pixels in a specific area inside the water mass, record the specific numerical changes of the corresponding water color parameters (such as chlorophyll fluorescence intensity or turbidity scattering unit) in the current monitoring period relative to the previous monitoring period, whether it increases or decreases, as well as the absolute amount of change and the relative rate of change. For example, if the chlorophyll fluorescence intensity in a certain area changes from 10 units in the previous monitoring period to 15 units in the current period, it is recorded as an increase of 5 units and the rate of change is 50%. These boundary displacement information and water color parameter change information are summarized to obtain environmental spatiotemporal variation information.

[0031] The steps for obtaining preliminary tracing conclusions of environmental abnormal events are as follows: Analyze the temporal sequence and spatial linkage of data anomalies in environmental spatiotemporal variation information, screen identified anomaly events with the same type and start time at sensor monitoring points in different geographical locations, calculate the time difference between the event occurrence times, and use the GPS coordinates of each sensor point to measure the straight-line geographical distance. The apparent direction and average rate of propagation of the anomaly signal in the water body are determined by the ratio of time difference to distance, and a set of anomaly propagation characteristic parameters is obtained. Based on the main propagation direction of the abnormal signal indicated by the abnormal propagation characteristic parameter set, combined with the dominant direction vector and average movement speed data of the inner boundary migration of the water mass when the abnormality occurs extracted from the water mass spatial distribution map, according to the natural flow characteristics of the regional water body, the upstream potential impact area of ​​the origin of the abnormal disturbance is determined, and the potential abnormal source area is delineated; Based on the delineation of potential abnormal source areas, and combined with the sensor points and water mass boundary change ranges affected by the abnormal events recorded in the environmental spatiotemporal variation information, and referring to the typical diffusion coefficients of known pollutants under similar hydrological conditions stored in the database, the negative impact of this abnormal event on the benthic community and plankton population in the target waters within the delineated area is evaluated, and a preliminary tracing conclusion of the environmental abnormal event is generated.

[0032] Specifically, based on the data anomaly occurrence time series (from identified water quality events) and spatial linkage (from sensor network and water mass boundary changes) contained in the environmental spatiotemporal variation information obtained in the previous link, an in-depth analysis is conducted. First, the identified abnormal events with the same type (for example, the dissolved oxygen concentration is lower than the lower limit of the dynamic baseline and exceeds the allowable fluctuation threshold) and similar starting time (considering that it takes time for the anomaly to propagate in the water body, the "same" here should be understood as having a time series correlation. For example, the anomaly of sensor B starts within a reasonable time window after sensor A. The window is estimated based on the distance between the two points and the maximum possible flow rate. For example, if the distance is 100 meters and the maximum flow rate is 0.1 meters per second, the reasonable time window may be 1000 seconds, or about 17 minutes later) are selected. For this group of abnormal events with time series correlation selected, the time difference between each sensor recording the start time of the event is accurately calculated (for example, sensor A records the time , sensor B records the time , then the time difference ), and use the pre-stored GPS coordinates of each sensor point (for example, sensor A is (longitude_A, latitude_A), sensor B is (longitude_B, latitude_B)) to calculate the straight-line geographic distance between the two points by using the spherical distance formula or approximating it to plane coordinates within a small range ( ), by calculating the ratio of distance to time difference ( ), determine the apparent direction (from sensor A to sensor B) and average rate of propagation of the abnormal signal in the water body. If at least three sensors detect the same type of anomaly in a certain sequence, and the calculated apparent propagation direction and rate are consistent between each sensor pair (for example, the direction angle difference is less than 30 degrees, and the rate difference is less than 50%), then this is considered to be a reliable anomaly propagation path. The calculated propagation direction, average rate, and the sensor sequence involved are integrated to obtain the anomaly propagation characteristic parameter set.

[0033] Based on the main propagation direction of the abnormal signal indicated by the abnormal propagation characteristic parameter set obtained in the previous step, for example, it is determined to be from northwest to southeast. At the same time, the dominant direction vector and average movement speed data of the water mass boundary migration extracted from the water mass spatial distribution map during the time period of the abnormality are retrieved and combined. For example, the water mass spatial distribution map analysis shows that the boundary of a water mass overlapping with the abnormal area also shows an average migration of 50 meters per hour to the southeast. If the migration direction of this water mass is basically consistent with the abnormal propagation direction detected by the sensor (for example, the angle is less than 45 degrees), the two confirm each other. Further combined with the known natural flow characteristics of the regional water body, these characteristics may come from historical hydrological survey data, tide table information or simplified hydrodynamic model output, indicating the typical flow field pattern of the aquaculture area under specific seasons or tidal conditions. For example, the database shows that during the low tide of the current season, the dominant flow direction in the area is from northwest to southeast. Combining these three aspects of information -- The direction of anomaly propagation determined by the sensor, the direction of water mass migration inverted from remote sensing images, and the inherent water flow pattern of the region -- are used together to determine the scope of the upstream potential impact area of ​​the origin of the abnormal disturbance. The specific method is to start from the location of the sensor that first detected the anomaly or the boundary area of ​​the water mass where a significant change occurred, and trace back in the direction opposite to the direction of anomaly propagation (i.e., the source direction, such as the northwest direction). The tracing distance can be estimated based on the duration of the abnormal event, the propagation rate, and the water flow diffusion characteristics. For example, if the anomaly propagation rate is 100 meters per hour and the event has lasted for 2 hours, then trace back at least 200 meters upstream, and based on the possibility of lateral diffusion of the water flow, a fan-shaped or rectangular area is delineated as the upstream potential impact area. For example, with the sensor where the anomaly first appeared as the vertex, it extends 200 meters along the source direction (such as the northwest direction), and expands 30 degrees on both sides to form a fan-shaped area, thereby establishing the potential anomaly source area delineation.

[0034] Based on the delineation of potential abnormal source areas completed in the above steps, and in combination with the previously acquired environmental spatiotemporal variation information, a list of sensor points specifically affected in this abnormal event (for example, sensors S1, S3, and S5 all showed dissolved oxygen below the threshold) and the specific range of water mass boundaries where significant changes occurred (for example, the southern boundary of water mass M shrank inward by about 30 meters, and the corresponding water color parameter chlorophyll fluorescence intensity decreased by 40%), next, refer to a pre-built database that stores typical diffusion coefficients of various known pollutants (such as nitrogen and phosphorus in different forms, specific organic matter, heavy metals, etc.) under hydrological conditions similar to the current aquaculture area (for example, flow rate range 0.05-0.1 meters per second, water depth 5-10 meters, specific salinity range). For example, for ammonia nitrogen, the longitudinal diffusion coefficient may be 0.5 under low flow conditions in summer. , horizontally 0.05 , using these diffusion coefficients, we can preliminarily estimate the range of possible diffusion impact of pollutants released from the potential abnormal source area (for example, the type is related to the monitored abnormal parameters, such as high ammonia nitrogen abnormalities, ammonia nitrogen pollutants are considered) within a certain period of time, and compare it with the actual affected sensor points and water mass change range for verification. Then, based on the nature of the abnormal event (such as a sharp drop in dissolved oxygen to 2 mg / L, or ammonia nitrogen soaring to 1.0 mg / L), duration (such as 6 hours), impact range (covering about 20% of the aquaculture area), and compared with the tolerance threshold of the aquaculture species to these parameters (for example, the suffocation point of a certain fish species for dissolved oxygen is 3 mg / L), evaluate the possible impact of this abnormal event on the designated area and downstream. The assessment can be a graded result, such as slight impact (such as short-term stress but recoverable), moderate impact (such as partial mortality or growth inhibition), and severe impact (such as large-scale mortality or ecological structure change). The specific grading standards need to be set in advance. For example, if the dissolved oxygen level is continuously lower than the lower survival limit of the target aquaculture species for more than 2 hours and the coverage area exceeds 10% of the aquaculture area, it is judged as a moderate impact. Combining these analyses, a preliminary tracing conclusion of the environmental abnormality event is generated, which includes potential source pointing, impact range assessment and ecological impact degree judgment.

[0035] The steps for obtaining the results of the aquaculture environment assessment are as follows: Integrate the preliminary tracing conclusions of environmental anomalies, perform dimensionless conversion on the data using maximum and minimum normalization, and form standardized multi-dimensional environmental status data; Based on standardized multi-dimensional environmental status data, standardized values ​​of water quality parameters associated with aquaculture species are extracted. Current monitoring values ​​are compared with the threshold table of suitable survival conditions for the aquaculture species, and compliance is quantitatively compared item by item. The compliance of current environmental factors at each monitoring point in the aquaculture area with the standards and the potential ecological and environmental risk level are evaluated to obtain the environmental risk compliance judgment value. Based on the environmental risk compliance judgment value, the aquaculture environment assessment results are output.

[0036] Specifically, the preliminary environmental anomaly event source tracing conclusion obtained in the foregoing steps is integrated, and the conclusion includes information such as an abnormal parameter type (for example, dissolved oxygen, ammonia nitrogen), an abnormal occurrence degree (for example, a minimum value of dissolved oxygen is 2.0 mg / L, a maximum value of ammonia nitrogen is 0.8 mg / L), an affected spatial range (for example, an affected area is 500 m2), an event duration (for example, 4 hours), and a preliminary impact assessment level on aquatic organisms (for example, moderate impact). For key quantitative indicators in these data, such as dissolved oxygen concentration, ammonia nitrogen concentration, pH value, affected range area, duration, and the like, a maximum minimum normalization method is used for dimensionless conversion processing to eliminate the influence of different parameter dimensions and numerical range differences. Specifically, for each parameter to be normalized , the normalized value of the parameter is calculated as , where and are the preset minimum value and maximum value of the parameter, respectively. These preset and are determined according to the physical meaning, ecological meaning, or historical data range of the parameter. For example, for the dissolved oxygen concentration, can be set to 0 mg / L (absolute minimum value), can be set to 15 mg / L (common supersaturation value or upper limit of instrument range). If the current dissolved oxygen concentration is 2.0 mg / L, the normalized value is . For the ammonia nitrogen concentration, can be set to 0 mg / L, can be set according to the acute toxicity reference value of the cultured species, for example, set to 2.0 mg / L. If the current ammonia nitrogen is 0.8 mg / L, the normalized value is . Similarly, the range (for example, minimum 0 m2, maximum entire culture area such as 10,000 m2) and duration (for example, minimum 0 hours, maximum reference management period or longest historical event such as 24 hours) are also normalized. All these normalized parameter values (ranging from 0 to 1) jointly constitute the standardized multi-dimensional environmental state data.

[0037] Based on the standardized multi-dimensional environmental state data formed in the previous step, first, according to the type of the main breeding species in the current breeding area, for example, if the breeding species is grouper, the current monitoring values of the water quality parameters closely related to the growth and survival of the grouper are extracted from the standardized data (these values are the actual measured values before normalization and known before the traceability conclusion and the formation of standardized data), for example, the current dissolved oxygen concentration is 2.0 mg / L, the ammonia nitrogen concentration is 0.8 mg / L, the pH value is 7.2, the water temperature is 28°C, and the salinity is 25 ppt. At the same time, the system loads the preset suitable survival condition threshold table for this grouper species, which details the suitable range, warning range and dangerous range of each key water quality parameter, for example, the suitable survival condition threshold table for grouper specifies that the suitable range of dissolved oxygen is greater than 5 mg / L, the warning range is 3-5 mg / L, and the dangerous range is less than 3 mg / L; the suitable range of ammonia nitrogen (total ammonia nitrogen) is less than 0.5 mg / L, the warning range is 0.5-1.0 mg / L, and the dangerous range is greater than 1.0 mg / L; the suitable range of pH is 7.5-8.5, the warning range is 7.0-7.5 or 8.5-9.0, and the dangerous range is less than 7.0 or greater than 9.0. Comparing the currently monitored dissolved oxygen 2.0 mg / L with the threshold table, it is determined that it is in the dangerous range, the current ammonia nitrogen 0.8 mg / L is in the warning range, and the current pH value 7.2 is in the warning range. For each related parameter, such item-by-item compliance quantitative comparison is carried out, and according to the comparison result, the compliance of each environmental element of each monitoring point or comprehensive area in the current breeding area is judged, and the potential ecological environmental risk level is determined accordingly, for example, the risk level rules can be set as follows: if any key parameter enters the dangerous range, the overall risk level is "high risk"; if no parameter enters the dangerous range, but at least one parameter enters the warning range, it is "medium risk"; if all parameters are within the suitable range, it is "low risk". According to this rule, since the dissolved oxygen is in the dangerous range, the potential ecological environmental risk level of the current breeding area is rated as "high risk", and this level or the comprehensive score obtained by calculating the risk (for example, high risk counts 3 points, medium risk counts 2 points, and low risk counts 1 point, and then weighted average) is the environmental risk compliance judgment value obtained.

[0038] Based on the environmental risk compliance judgment value obtained in the previous step, for example, the judgment value is a "high risk" level, or a numerical risk score such as 2.8 (for example, the score range is 0-3, the higher the risk, the greater the risk), the system uses this judgment value as the core content, organizes and outputs the final aquaculture environment assessment results, and the assessment results are intended to clearly and comprehensively reflect the current environmental conditions of the aquaculture area and the risks it poses to aquaculture activities. The specific output content includes at least: a clear overall environmental risk level, such as "high risk", and (if a scoring system is adopted) the corresponding risk index, such as "risk index: 2.8 / 3.0". At the same time, the main driving parameters that lead to this risk level judgment and their current status are listed, for example, "Main risk factor: dissolved oxygen concentration is 2.0 mg per liter, which is at a dangerous level (grouper danger threshold: less than 3 mg per liter) per liter); the total ammonia nitrogen concentration is 0.8 mg per liter, which is at the warning level (grouper warning threshold: 0.5 to 1.0 mg per liter)". It should also include the timestamp or time period of the monitoring data on which the assessment is based, as well as the aquaculture area or specific monitoring point information involved. If the system has included a preliminary judgment on the cause of the anomaly or an estimate of the scope of impact in the previous steps (such as the preliminary tracing conclusion of the environmental anomaly event), this information can also be selectively summarized and included in the assessment results. For example, "Preliminary tracing suggests that there may be abnormal disturbances in the upstream area A, affecting water mass B, covering an area of ​​approximately 500 square meters." The final aquaculture environment assessment results are presented in the form of a structured report, or displayed to the user through a visual interface (such as the indicator light color and alarm information on the dashboard), and can also be used as a data interface for other management systems to call.

[0039] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A multi-dimensional environmental assessment method for coastal aquaculture areas, characterized by: The following steps are involved: Water quality sensor arrays are deployed at key locations in the aquaculture area to continuously record dissolved oxygen concentration, pH value, temperature in degrees Celsius, salinity units, turbidity scattering units, chlorophyll fluorescence intensity, milligrams per liter of ammonia nitrogen, and milligrams per liter of nitrite. Multi-source remote sensing images covering the aquaculture area and adjacent waters are simultaneously retrieved. Through image correction and atmospheric correction, remote sensing reflectance data is extracted and water color parameters are calculated. The sensor array time series data and remote sensing water color parameters are integrated to generate comprehensive water environment monitoring data. Based on the sensor parameters in the comprehensive water environment monitoring data, each parameter time series is decomposed into trend terms, seasonal terms, and residual terms, and statistical control limits are set for the residual terms to construct a dynamic normal fluctuation range for each parameter, establish a dynamic baseline for water quality parameters, and combine remote sensing water color parameters to segment different water bodies by setting a water color index threshold. Then, edge points are connected and smoothed to identify the spatial distribution and boundaries of different water masses and establish a spatial distribution map of water masses; The real-time sensor parameter values ​​collected are compared with the dynamic baselines of the corresponding water quality parameters through difference calculation and threshold comparison. When the difference exceeds the preset control limit, it is marked as a potential anomaly point. The duration of the anomaly point is analyzed, and events in which single or multiple parameters deviate from the normal range are identified. The water mass boundary position and water color parameter changes in the water mass spatial distribution map are continuously monitored, and the daily pixel displacement of the boundary and the change in the water color index are recorded to obtain environmental spatiotemporal variation information.

2. The multi-dimensional environmental assessment method for coastal aquaculture areas according to claim 1 is characterized in that: The method further comprises: Analyze the timing and spatial linkage of data anomalies in the environmental spatiotemporal variation information, determine the direction of anomaly propagation by calculating the time difference and spatial distance of anomalies in sensors at different locations, combine the migration direction and speed of water mass boundaries in the water mass spatial distribution map, determine the upstream impact area through hydrodynamic principles, trace the source direction of the abnormal disturbance and the potential impact sector, evaluate the degree of impact on the ecological environment of coastal waters, and generate preliminary tracing conclusions on the source of environmental anomalies.

3. The multi-dimensional environmental assessment method for coastal aquaculture areas according to claim 2, characterized in that: The method further comprises: Integrate the preliminary tracing conclusions of the environmental anomaly events and the spatial distribution map of the water masses, merge the anomaly levels of various parameters, impact ranges, and water mass stability data, compare with the survival condition thresholds of the aquaculture species and the lower limit of dissolved oxygen and the upper limit of ammonia nitrogen in the regional environmental management objectives, evaluate the current environmental status and potential risks of the aquaculture area, divide the environmental status assessment levels into different levels according to the risk matrix judgment rules, and output the aquaculture environment assessment results.

4. The multi-dimensional environmental assessment method for coastal aquaculture areas according to claim 1, characterized in that: The steps for obtaining the comprehensive water environment monitoring data are as follows: Water quality sensor arrays were deployed at key locations in the aquaculture area, with a sampling frequency set to once every ten minutes. Dissolved oxygen concentration, pH value, Celsius temperature, salinity units, turbidity scattering units, chlorophyll fluorescence intensity, ammonia nitrogen milligrams per liter, and nitrite milligrams per liter were continuously recorded to obtain the sensor's original observation sequence. Based on the geographic coordinates of the aquaculture area, multi-source remote sensing images covering the aquaculture area and adjacent waters were retrieved. The image data were orthorectified and geo-registered, and atmospheric correction was performed using radiometric calibration and atmospheric condition parameter input. Furthermore, remote sensing reflectance data were extracted by combining target visible and near-infrared bands, and water color parameters were calculated to establish a remote sensing image water color index set. Based on the sensor original observation sequence and the remote sensing image water color index set, the geographic coordinates of each sensor in the sensor original observation sequence are spatially matched with the geographic grid of the image pixels in the remote sensing image water color index set, and the sensor data and the remote sensing image daily data are timestamp aligned using linear interpolation or time window averaging method to generate comprehensive water environment monitoring data.

5. The multi-dimensional environmental assessment method for coastal aquaculture areas according to claim 1, characterized in that: The steps for obtaining the water cluster spatial distribution map are: Based on the sensor parameters in the comprehensive water environment monitoring data, the corresponding parameter continuous time series data for the past year are extracted from the historical database, and a Gaussian smoothing filter is used to filter out random noise from each parameter time series. Then, the least squares method is used to fit the trend change straight line or curve component, and the Fourier transform is used to identify and separate the main periodic seasonal fluctuation component and the remaining random disturbance component to obtain the parameter sequence decomposition component; Based on the residual sequence of each parameter in the decomposition component of the parameter sequence, by calculating the mean and standard deviation values ​​of the residual sequence, setting the mean plus or minus three times the standard deviation as the upper and lower thresholds of statistical control, constructing the dynamic normal fluctuation reference range of each parameter, establishing a dynamic baseline of water quality parameters, and extracting corresponding remote sensing water color parameters from comprehensive water environment monitoring data. According to the numerical domain characteristics of different water color parameters, multiple threshold intervals are set, and each pixel of the remote sensing image is classified and divided to obtain a water quality baseline and a preliminary water body segmentation area; Based on the water quality baseline and the classification results of water pixels in the preliminary water body segmentation area, the Sobel or Canny edge detection operator is used to scan the image, identify the intersection pixels between water pixel areas of different categories, connect the intersection pixels with spatial adjacency to form an initial vector boundary line, and then apply the Douglas-Peucker algorithm to iteratively simplify the initial boundary line to establish a water mass spatial distribution map.

6. The multi-dimensional environmental assessment method for aquaculture areas in coastal zones according to claim 1, characterized in that: The steps for obtaining the environmental spatiotemporal variation information are as follows: The real-time sensor parameter values ​​are compared with the upper and lower limits of the normal fluctuation range of each parameter in the dynamic baseline of the corresponding water quality parameter on a point-by-point basis, and the deviation between the current parameter measurement value and the baseline range boundary is calculated. When the deviation exceeds the pre-set allowable fluctuation threshold, the parameter data point at that moment is marked as a potential abnormal state, forming a potential water quality abnormal point sequence; Based on the potential water quality anomaly point sequence, by searching whether the data points marked as potential anomaly states in the sequence appear more than three times consecutively on the time axis, identifying abnormal events in which a single parameter or multiple combined parameters deviate from the normal range, and at the same time, by comparing with the water mass spatial distribution map, extracting the coordinate set of each water mass boundary pixel and the water color parameter of the corresponding pixel at the current moment, and obtaining the identified water quality event and boundary pixel set; Based on the identified water quality events and boundary pixel sets, by comparing the boundary pixel coordinate sets of the same identified water mass recorded in the current monitoring period with the previous monitoring period, the north-south and east-west displacement components of each corresponding boundary point in the geographic coordinate system are calculated, and then the total displacement vector is synthesized. The increase or decrease in the water color parameters of the area corresponding to the pixel is recorded to obtain environmental spatiotemporal variation information.

7. The multi-dimensional environmental assessment method for coastal aquaculture areas according to claim 2, characterized in that: The steps for obtaining the preliminary tracing conclusion of the environmental abnormality event are as follows: Analyze the temporal sequence and spatial linkage of data anomalies in the environmental spatiotemporal variation information, screen identified anomaly events of the same type and with the same start time at sensor monitoring points in different geographical locations, calculate the time difference between the event occurrence times, and use the GPS coordinates of each sensor point to measure the straight-line geographical distance. Determine the apparent direction and average rate of propagation of the anomaly signal in the water body through the ratio of time difference to distance, and obtain a set of anomaly propagation characteristic parameters; Based on the main propagation direction of the abnormal signal indicated by the abnormal propagation characteristic parameter set, combined with the dominant direction vector and average movement speed data of the inner boundary migration of the water mass when the abnormality occurs extracted from the water mass spatial distribution map, according to the natural flow characteristics of the regional water body, the upstream potential impact area of ​​the starting point of the abnormal disturbance is determined, and the potential abnormal source area is delineated; Based on the delineation of the potential abnormal source area, and combined with the sensor points and water mass boundary change ranges affected by the abnormal events recorded in the environmental spatiotemporal variation information, and referring to the typical diffusion coefficients of known pollutants under similar hydrological conditions stored in the database, the negative impact of this abnormal event on the benthic community and plankton population in the target waters within the delineated area is evaluated, and a preliminary tracing conclusion of the environmental abnormal event is generated.

8. The multi-dimensional environmental assessment method for coastal aquaculture areas according to claim 3 is characterized in that: The steps for obtaining the aquaculture environment assessment results are as follows: Integrate the preliminary tracing conclusions of the environmental anomaly events, perform dimensionless conversion on the data using maximum and minimum normalization, and form standardized multi-dimensional environmental status data; Based on the standardized multi-dimensional environmental status data, the standardized values ​​of water quality parameters associated with the aquaculture species are extracted, and the current monitoring values ​​are compared with the threshold table of suitable living conditions for the aquaculture species for item-by-item compliance. The compliance of the current environmental factors at each monitoring point in the aquaculture area with the standards and the potential ecological environmental risk level are evaluated to obtain the environmental risk compliance judgment value.

9. The multi-dimensional environmental assessment method for coastal aquaculture areas according to claim 8, characterized in that: The step of obtaining the aquaculture environment assessment result further includes: outputting the aquaculture environment assessment result based on the environmental risk compliance determination value.

10. The environmental assessment cloud platform of the multi-dimensional environmental assessment method for coastal aquaculture areas according to any one of claims 1 to 9, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the processor runs the computer program, the method according to any one of claims 1 to 9 is implemented.

Images