A water body anomaly monitoring method and system based on remote sensing big data

Abstract

This invention belongs to the field of water body monitoring technology, specifically relating to a method and system for monitoring water anomalies based on remote sensing big data. The method includes acquiring updated water body data, comparing the updated data with normal water body data to identify new anomaly nodes, and updating the monitoring list based on these new anomaly nodes. This invention overcomes the limitations of using fixed thresholds by establishing a dynamic adaptive judgment benchmark based on the data's own characteristics, setting corresponding anomaly thresholds for different fluctuation nodes. This significantly reduces the false alarm rate and improves monitoring accuracy when identifying and excluding seasonal water storage regulation.

Description

Technical Field

[0001] This invention belongs to the field of water body monitoring technology, specifically relating to a method and system for monitoring water anomalies based on remote sensing big data. Background Technology

[0002] The protection and management of water resources are crucial for maintaining ecological security and sustainable development. Monitoring water anomalies helps to promptly detect changes in water quality and quantity, providing a scientific basis for risk warnings and decision-making. Through continuous monitoring of important water bodies, managers can address anomalies promptly and take targeted measures to prevent escalation, thereby safeguarding regional ecological security.

[0003] Current water anomaly monitoring technologies mainly rely on remote sensing image analysis or fixed-point sensor measurements. Without in-depth spatiotemporal correlation analysis, changes in water level, turbidity, and color caused by seasonal changes, periodic regulation of upstream reservoirs, or regular agricultural irrigation are easily misjudged by the system as abnormal events, leading to false alarms. This not only wastes a lot of verification and emergency response resources but also weakens the credibility of the monitoring system.

[0004] Furthermore, while physical measurement methods based on fixed sensors can provide high-frequency data, their monitoring range is limited to point or linear areas, making it impossible to capture the spatial formation and evolution of abnormal events such as algal blooms and pollutant diffusion. Although remote sensing images can provide a wide-area spatial view, their revisit cycle is long, which often leads to discontinuous monitoring in time, missing the critical window period for anomalies and affecting the timeliness of early warning.

[0005] To address the aforementioned problems, this invention provides a method and system for monitoring water anomalies based on remote sensing big data. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for monitoring water anomalies based on remote sensing big data, which can quickly identify water anomalies and the judgments are reliable.

[0007] The specific technical solution adopted by this invention is as follows:

[0008] A method for monitoring water anomalies based on remote sensing big data includes the following steps:

[0009] Based on the initial water area data derived from remote sensing imagery, once anomaly nodes are identified as constituting abnormal water bodies, dynamic monitoring of these abnormal water bodies is initiated. This dynamic monitoring includes:

[0010] Get updated water area data;

[0011] The updated water area data is compared with the identified normal water area data to identify new anomalous nodes;

[0012] And update the monitoring list based on the new abnormal nodes;

[0013] The determination of normal water area data includes: acquiring remote sensing images of the water area to be measured, and extracting the target water area from the remote sensing images;

[0014] The target water area is grayscaled to obtain initial water data; the initial water data is analyzed to identify fluctuation nodes, and the offset value and anomaly threshold are calculated for the fluctuation nodes based on the evaluation model;

[0015] And the data portion corresponding to the fluctuation nodes with offset values ​​less than the abnormal threshold is identified as normal water area data.

[0016] Preferably, determining abnormal nodes includes:

[0017] Data points that do not belong to the normal water area data in the initial water area data will be marked as abnormal nodes.

[0018] Preferably, after identifying anomalous nodes, the method further includes:

[0019] Identify anomalous nodes caused by seasonal water storage and exclude anomalous nodes caused by seasonal water storage from the list of anomalous nodes.

[0020] Preferably, determining that anomaly nodes constitute anomaly water bodies includes:

[0021] According to the monitoring rules, the severity of abnormal nodes is determined; abnormal nodes with the second degree of severity are marked as abnormal water bodies; and abnormal nodes with the first degree of severity are marked as mild abnormal nodes.

[0022] Preferably, judging the severity of abnormal nodes according to monitoring rules includes:

[0023] The target water area is structurally segmented to obtain multiple monitoring sub-areas, and corresponding time-series data are established for each monitoring sub-area.

[0024] Identify at least one adjacent sub-region that is adjacent to the monitoring sub-region containing the anomalous node;

[0025] Furthermore, by comparing the time series data corresponding to the abnormal node with the time series data corresponding to at least one adjacent sub-region, the similarity is calculated, and the severity of the abnormal node is determined based on the comparison result of the similarity with the preset similarity threshold.

[0026] Preferably, comparing the updated water area data with normal water area data to identify new anomalous nodes includes:

[0027] If a point deviates from the normal water area data in the updated water area data, that point is extracted as a suspicious node.

[0028] Obtain a time window centered on the suspicious node as the comparison interval;

[0029] And within the comparison interval, the test node with the smallest difference in data value from the suspicious node will be marked as a new abnormal node.

[0030] This invention also discloses a water anomaly monitoring system based on remote sensing big data, used to implement the above-mentioned water anomaly monitoring method based on remote sensing big data, comprising:

[0031] The water anomaly identification unit is used to identify anomaly nodes based on initial water area data derived from remote sensing images;

[0032] Anomaly severity assessment unit is used to determine the severity of anomalies and generate a monitoring activation signal when it is determined that an anomaly constitutes an abnormal water body.

[0033] And a dynamic monitoring unit, used to respond to the monitoring start signal, initiate and execute dynamic monitoring of abnormal water bodies.

[0034] Preferably, responding to the monitoring activation signal, initiating and executing dynamic monitoring of abnormal water bodies includes: acquiring updated water area data, comparing the updated water area data with normal water area data to identify new abnormal nodes, and updating the monitoring list based on the new abnormal nodes.

[0035] Preferably, the water anomaly identification unit is specifically used for:

[0036] Initial water data is analyzed to identify undulating nodes, and offset values ​​and anomaly thresholds are calculated for these nodes based on the evaluation model.

[0037] The data portion corresponding to the fluctuating nodes with offset values ​​less than the abnormal threshold is identified as normal water area data.

[0038] And data points that do not belong to the normal water area data in the initial water area data will be marked as abnormal nodes.

[0039] Preferably, the anomaly severity assessment unit is specifically used for:

[0040] The target water area is structurally segmented to obtain multiple monitoring sub-areas, and corresponding time-series data are established for each monitoring sub-area.

[0041] Furthermore, by comparing the time series data corresponding to the abnormal node with the time series data corresponding to the adjacent sub-region, the similarity is calculated, and the severity of the abnormal node is judged based on the comparison result of the similarity with the preset similarity threshold.

[0042] Beneficial effects

[0043] 1. This invention determines normal water area data by analyzing initial water area data, and calculates offset values ​​and abnormal thresholds for each fluctuation node based on the evaluation model. At the same time, it excludes abnormal nodes caused by seasonal water storage, thereby establishing a dynamic adaptive judgment benchmark based on the characteristics of the data itself, setting corresponding abnormal thresholds for different fluctuation nodes, and thus overcoming the limitations of using fixed thresholds for judgment, reducing the false alarm rate when identifying and excluding seasonal water storage.

[0044] 2. This invention divides the target water area into multiple monitoring sub-areas. When judging the severity of abnormal nodes according to monitoring rules, it calculates the similarity of time series data between the monitoring sub-area where the abnormal node is located and its adjacent sub-areas, and classifies the abnormal node into the first severity level or the second severity level. By introducing spatial correlation analysis, it achieves differentiated assessment of anomalies, effectively distinguishes between local sudden anomalies and regional slow-changing anomalies, and provides a graded basis for early warning response, making management decisions more targeted.

[0045] 3. This invention identifies anomalous nodes with the second degree of severity as anomalous water bodies and updates them to the monitoring list. Then, dynamic monitoring is initiated. By acquiring updated water body data and comparing it with the data of identified normal water bodies, new anomalous nodes are identified and the monitoring list is updated to form a dynamic monitoring system. This system tracks the development of known anomalous water bodies and discovers new anomalous nodes through efficient comparison with normal water body data. This achieves the transformation from static monitoring to dynamic tracking, improving the continuity and efficiency of monitoring. Attached Figure Description

[0046] Figure 1 This is a flowchart of the method of the present invention;

[0047] Figure 2 This is a system structure diagram of the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0049] Example 1

[0050] Please see Figure 1As shown in the figure, this embodiment provides a method for monitoring water anomalies based on remote sensing big data, which specifically includes the following steps:

[0051] S1. Acquire remote sensing images and perform structured segmentation. This involves decomposing the extracted target water area into multiple standardized and easily manageable monitoring sub-areas according to preset rules such as geographic grids. The specific steps are as follows:

[0052] Acquire remote sensing images of the water body to be measured. These images can be multispectral images containing information from multiple spectral bands, or optical images with high spatial resolution.

[0053] Extracting target water areas from remote sensing images to focus the analysis on the water itself is achieved through classification rules based on pre-labeled samples or through automatic grouping based on the spectral or texture characteristics of the image data itself, thereby separating water from land, vegetation and other land features at the pixel level.

[0054] The classification rules are based on the discrimination criteria established by the spectral and texture features of pre-labeled samples. They are used to distinguish water bodies from non-water bodies at the pixel level. The automatic grouping process based on the spectral or texture characteristics of the image data itself is specifically manifested in classifying pixels with similar spectral responses into one category.

[0055] After extracting the target water area, the target water area is structurally segmented to obtain multiple monitoring sub-regions for more refined spatial analysis and to support subsequent regional correlation judgment.

[0056] The purpose of the above-mentioned structured segmentation is to decompose the vast target water area into basic geographical units that are easy to manage and analyze, thereby enabling accurate location of anomalies.

[0057] The specific segmentation method of the structured segmentation is to overlay a preset geographic grid on the geographic coordinates of the target water area. Specifically, the segmentation method is to divide the entire water area into multiple square monitoring sub-areas with a side length of 100 meters.

[0058] Images of each monitoring sub-area acquired at different times are converted to grayscale, and the grayscale values ​​at each time point are serialized to obtain initial water area data representing the change in brightness of each sub-area over time.

[0059] S2. Establish a normal state reference and identify fluctuation nodes, as detailed below:

[0060] For each monitoring sub-area, a corresponding time-series data sequence is established based on the initial water area data formed by remote sensing images at different time points. The time-series data sequence is a data set formed by arranging the gray values ​​of a certain monitoring sub-area at multiple consecutive time points in chronological order, which constitutes the basis for analyzing the dynamic changes of water bodies.

[0061] Before analysis, noise removal is performed on the initial water data to eliminate artifacts introduced by factors such as sensor thermal noise, atmospheric scattering, or instantaneous reflections from the water surface, thereby ensuring the accuracy of subsequent analysis.

[0062] Specifically, for each monitoring sub-region, a processing method is applied to the grayscale value time series to smooth the data and filter out instantaneous high-frequency fluctuations, such as replacing the value of the center point with the median of the data points within a time window.

[0063] After processing, stable sections are identified in the initial water data after noise removal to establish a reliable reference that can represent the normal state of the sub-region.

[0064] Furthermore, in specific implementation, by calculating the variance or standard deviation of time series data within a sliding time window, when the variance or standard deviation remains below a specific stable threshold for a preset duration, the data within that time window is identified as a stationary segment.

[0065] Calculate the sample mean of all data points within the stable period, and determine the sample mean as the benchmark value representing the normal state of the monitored sub-region;

[0066] Based on this benchmark, the entire initial water area data is analyzed to identify possible abnormal fluctuation nodes. For any data point in the time series data, the difference between its data value and the benchmark is calculated. This difference is the offset value, which is used to quantify the degree to which the data point deviates from the normal state.

[0067] According to the preset evaluation rules, the corresponding abnormal threshold is calculated for each fluctuation node. The abnormal threshold is a value dynamically calculated based on the statistical characteristics of historical data of the monitored sub-area, or a fixed value set according to the water type and seasonal factors. The evaluation rules are based on the statistical distribution of historical data, such as the standard deviation multiple, or are set in combination with external factors such as water type and season.

[0068] The fluctuation nodes in the initial water area data that have an absolute value of offset less than their corresponding abnormal threshold are selected, and the data portions corresponding to these selected fluctuation nodes are collectively identified as normal water area data.

[0069] S3. Exclude non-abnormal factors such as seasonal storage, as follows:

[0070] In the initial water data, data points whose absolute offset value is greater than or equal to their anomaly threshold will be initially identified as anomaly nodes, indicating that the point has a significant deviation from the normal state.

[0071] Since not all anomalous nodes represent unexpected events such as water quality deterioration or pollution, it is necessary to identify and exclude nodes caused by normal human activities such as seasonal water storage to improve the accuracy of monitoring.

[0072] Seasonal water level regulation refers to predictable, periodic artificial adjustments to water levels for purposes such as irrigation, power generation, and flood control. It also manifests as significant changes in grayscale values ​​in remote sensing images. Misjudging such events as anomalies can trigger unnecessary alarms.

[0073] For the reasons mentioned above, by introducing historical water storage records or referring to established hydrological patterns, the occurrence time, magnitude of change, and rate of change of the current abnormal nodes are compared. Specifically:

[0074] Historical water storage records include the opening and closing times of reservoir gates, planned water release volumes, and historical water level change curves; hydrological patterns include periodic water level decline patterns caused by agricultural irrigation needs in specific seasons.

[0075] If the combined similarity between the characteristics of an abnormal node, such as the rate of change and duration, and multiple key indicators of historical water storage event patterns is greater than or equal to a preset matching threshold, then the node is identified as a node caused by seasonal water storage and excluded from the set of abnormal nodes, and does not proceed to the subsequent severity judgment process.

[0076] The comprehensive similarity is a calculation model used to quantify the degree of matching between the features of the current abnormal node and the historical flood control event patterns. Its specific definition is as follows:

[0077] ;

[0078] In the formula, The overall similarity score represents the index that quantifies the degree of similarity between the current abnormal event and the historical flood control pattern; the higher the value, the greater the similarity.

[0079] The weighting coefficients represent the relative importance of the rate of change and duration in the similarity calculation, respectively, satisfying the following conditions: ;

[0080] This indicates the current rate of change, which means the speed at which the grayscale value corresponding to the current abnormal node changes.

[0081] This indicates the current duration, which means the length of time the current abnormal state has lasted;

[0082] This represents the historical average rate of change, which means the rate of change of the average gray value obtained from historical water storage records.

[0083] This indicates the historical average duration, which is the average duration obtained from historical water storage records.

[0084] S4. Determine the severity and classify it as follows:

[0085] After arranging the remaining abnormal nodes in chronological order, the severity of each abnormal node is determined according to preset monitoring rules, which involve comparing the characteristics of the abnormal node with its spatial neighborhood. This allows for differentiated management of different levels of abnormalities, thereby rationally allocating monitoring and response resources. Specific judgment methods include:

[0086] Identify at least one adjacent sub-region that is spatially adjacent or close to the monitoring sub-region containing the anomalous node. Compare the sampling information corresponding to the anomalous node with the sampling information of the at least one adjacent sub-region at the same time point to calculate the similarity between the two, thereby determining whether the anomaly is an isolated point event or a local manifestation of a large-scale change.

[0087] The sampling information includes a dataset of multiple dimensions of features that can reflect the state of water bodies extracted from remote sensing images. In addition to grayscale values, it also includes water color index, which characterizes the degree of eutrophication of water bodies, and transparency inversion value, which reflects the content of suspended solids in water bodies.

[0088] Similarity is calculated using methods that measure the degree of similarity between two datasets, preferably cosine similarity or Euclidean distance.

[0089] If the calculated similarity is lower than the preset similarity threshold, it means that the characteristics of the abnormal node are significantly different from those of its surrounding waters, showing strong local mutation. This characteristic is usually related to abnormal events such as point source pollution. In this case, the abnormal node is judged to have the second degree of severity.

[0090] If the calculated similarity is greater than or equal to the preset similarity threshold, it means that the anomaly may be part of a large-scale gradual process and its urgency is relatively low. Therefore, the anomaly node is judged to have the first severity.

[0091] Accordingly, abnormal nodes judged to have the first severity level are classified as mild abnormal nodes and recorded, but high-level alarms are not immediately triggered; abnormal nodes judged to have the second severity level are marked as abnormal water bodies, and a real-time water body monitoring mechanism is triggered to include the monitoring sub-area where the abnormal water body is located in the priority monitoring list; the monitoring list includes all monitoring sub-areas that are identified as abnormal water bodies and need to be monitored in a priority and continuous manner; at the same time, the earliest abnormal node corresponding to the abnormal water body is recorded as the first abnormal node.

[0092] The adjacent sub-regions are other monitoring sub-regions that are geographically adjacent to the currently analyzed monitoring sub-region or located within a predetermined distance.

[0093] S5. Identify and uniformly process consecutive anomalies, as follows:

[0094] After determining the severity level of a single anomalous node, a spatiotemporal correlation analysis is performed on the anomalous node to gain a more macroscopic understanding of the overall situation of the anomalous event. The specific steps for performing the spatiotemporal correlation analysis on the anomalous node are as follows:

[0095] All identified anomalous nodes are sorted and categorized, including slightly anomalous nodes and nodes identified as anomalous water bodies.

[0096] The distribution density of abnormal nodes in the spatiotemporal dimension is statistically analyzed, that is, the number or frequency of abnormal nodes in a unit space and a unit time, so as to discover clusters of abnormal nodes that are spatially or temporally concentrated. Based on whether the distribution density meets the preset continuity condition, it is determined whether multiple abnormal nodes constitute a continuous abnormal event.

[0097] The pre-defined continuity condition is specifically defined as follows:

[0098] If abnormal nodes appear in two or more consecutive imaging cycles within three or more spatially adjacent or neighboring monitoring sub-regions, and this is determined to constitute a continuous abnormal event, it means that a large-scale or long-lasting abnormal event has occurred.

[0099] At this point, among the abnormal nodes that constitute a series of abnormal events, one or more baseline nodes are set, that is, the abnormal node with the second most severe severity or the earliest abnormal node in time is selected as the baseline node.

[0100] Within a series of anomalous events, other anomalous nodes that are not the baseline node are processed in a unified manner with the baseline node. This unified processing specifically refers to:

[0101] All the abnormal nodes that constitute this continuous abnormal event are managed as a whole event, and preferably, they are assigned the same event identifier;

[0102] The location, time, and severity of the baseline node are used as representative information for the continuous abnormal events, thereby simplifying the subsequent tracking, analysis, and alarm processes.

[0103] S6. Dynamic monitoring and identification of new abnormal nodes;

[0104] After the initial calibration is completed, the updated water area data for the water area to be tested is continuously acquired and processed. The updated water area data is compared with the previously determined benchmark value and anomaly threshold representing the normal state. When the deviation between the data value and the benchmark value of a data point in the updated water area data exceeds the anomaly threshold, the data point is initially extracted as a suspicious node.

[0105] To further confirm whether the suspicious node is a genuine anomaly rather than transient noise, its spatiotemporal context is verified. The specific verification method is as follows:

[0106] Check whether there are other data points that also exceed the abnormal threshold within a preset time window or spatial neighborhood centered on the suspicious node;

[0107] If it exists, the suspicious node is confirmed as a new anomalous node, because this indicates that the deviation has a certain degree of persistence or spatial clustering; after confirmation, the new anomalous node will also undergo the severity judgment and continuous anomaly analysis process.

[0108] The preset time window is preferably one imaging cycle before and after, and the spatial neighborhood is preferably an adjacent monitoring sub-region.

[0109] Based on the newly identified anomaly nodes and their analysis results, the monitoring list is updated to form a dynamic monitoring system that can continuously respond to changes in water areas.

[0110] Example 2

[0111] Please see Figure 2 As shown, this embodiment provides a water anomaly monitoring system based on remote sensing big data, used to implement the above-mentioned water anomaly monitoring method based on remote sensing big data, including:

[0112] The water anomaly identification unit is used to process remote sensing images to identify potential anomaly nodes.

[0113] Specifically, remote sensing images of the water body to be measured are acquired, and the target water area is extracted from the remote sensing images using techniques such as image segmentation or water index; the target water area is then processed into grayscale to obtain initial water data representing the water body state.

[0114] The initial water area data is analyzed to identify fluctuation nodes. Based on the preset evaluation rules, the offset value and the corresponding abnormal threshold are calculated for each fluctuation node. The offset value represents the degree to which the node deviates from the normal state, while the abnormal threshold defines the acceptable range of normal fluctuations.

[0115] The fluctuation node represents the region where the state of the water body changes in time or space.

[0116] After the calculation is completed, the data portion associated with the fluctuation node whose offset value is less than its corresponding anomaly threshold is identified as normal water area data. This portion of normal water area data constitutes the benchmark for subsequent anomaly comparison.

[0117] Accordingly, data points that do not belong to the normal water area data in the initial water area data will be marked as abnormal nodes.

[0118] Furthermore, the water anomaly identification unit is also configured to identify and exclude anomaly nodes caused by seasonal water storage regulation, specifically:

[0119] For reservoirs or rivers affected by seasonal hydrological changes, the water anomaly identification unit combines historical data from the same period or known regulation and storage plans to determine whether an anomaly node is caused by normal water level rises and falls. If it is caused by normal water level rises and falls, it is excluded from the set of anomaly nodes to avoid false alarms.

[0120] The anomaly severity assessment unit is activated after the water anomaly identification unit identifies the anomaly node. It is used to determine the severity of the anomaly node and decide whether to initiate higher-level dynamic monitoring.

[0121] Specifically, the following steps are taken to conduct a detailed assessment:

[0122] The target water area is structurally segmented, for example, into multiple grid-like monitoring sub-areas, and a corresponding time-series data sequence is established for each monitoring sub-area. This time-series data sequence records the history of changes in the water state of the sub-area over time.

[0123] When an anomalous node appears, the anomaly severity assessment unit determines the monitoring sub-region containing the anomalous node and identifies at least one adjacent sub-region. By comparing the time series data sequence corresponding to the anomalous node with the time series data sequence corresponding to at least one adjacent sub-region, the similarity between them is calculated. The similarity here reflects whether the anomaly occurs in a local isolation or has regional diffusion.

[0124] Based on preset monitoring rules, the calculated similarity is compared with a preset similarity threshold, and the severity of the abnormal node is determined based on the comparison result. Specifically:

[0125] If the similarity is very low, it indicates that the anomaly is significantly different from the state of the surrounding area, which may be a serious anomaly, and it is classified as the second most serious.

[0126] If the similarity is high, it may be a slight or transitional change, which is judged as the first severity and marked as a mild anomalous node;

[0127] When one or more anomalous nodes are identified as having the second degree of severity, they are determined to constitute an anomalous water body that requires close monitoring, and a monitoring activation signal is generated accordingly to trigger subsequent dynamic monitoring procedures.

[0128] The dynamic monitoring unit, in response to the monitoring activation signal generated by the anomaly severity assessment unit, initiates and executes dynamic monitoring of the abnormal water body. Upon receiving the monitoring activation signal, the unit immediately starts working, and its dynamic monitoring process includes the following steps:

[0129] Actively acquiring updated water data about the anomalous water area means increasing the frequency of remote sensing image acquisition or calling up higher resolution data sources;

[0130] The updated water area data is compared with the normal water area data identified by the water anomaly identification unit to identify new anomaly nodes.

[0131] When a data point in the updated water data deviates from the baseline value by more than the anomaly threshold, the point is extracted as a suspicious node. A time window centered on the suspicious node is obtained as a comparison interval. Within this comparison interval, the test node with the smallest difference in data value from the suspicious node is found and finally marked as a new anomaly node. This helps to more accurately lock the evolution trajectory of anomalies in a continuously changing data stream.

[0132] Based on the newly identified anomaly nodes, the global monitoring list is updated. This list records all active and historical anomalies in water bodies, their severity, spatiotemporal location, and evolution status. By continuously updating this list, the entire lifecycle of anomaly events can be tracked and managed.

[0133] The above description is merely a preferred embodiment of this application and is not intended to limit this application. For those skilled in the art, this application can have various modifications and variations. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for monitoring water anomalies based on remote sensing big data, characterized in that, Includes the following steps: Based on the initial water area data derived from remote sensing images, when anomaly nodes are identified as constituting abnormal water bodies, dynamic monitoring of the abnormal water bodies is initiated. The dynamic monitoring of the abnormal water bodies includes: acquiring updated water area data; comparing the updated water area data with the identified normal water area data to identify new anomaly nodes; and updating the monitoring list based on the new anomaly nodes. The determination of normal water area data includes: acquiring remote sensing images of the water area to be measured and extracting the target water area from the remote sensing images; performing grayscale processing on the target water area to obtain initial water area data; analyzing the initial water area data to identify fluctuation nodes, calculating offset values ​​and anomaly thresholds for fluctuation nodes based on the evaluation model, and determining the data portion corresponding to fluctuation nodes with offset values ​​less than the anomaly threshold as normal water area data. Identifying anomalous nodes includes: marking data points in the initial water area data that do not belong to the normal water area data as anomalous nodes; after marking anomalous nodes, the method further includes: identifying anomalous nodes caused by seasonal water storage and excluding anomalous nodes caused by seasonal water storage from the anomalous nodes. Determining whether an anomalous node constitutes an anomalous water body includes: judging the severity of the anomalous node according to monitoring rules; marking anomalous nodes with the second degree of severity as anomalous water bodies; and marking anomalous nodes with the first degree of severity as mild anomalous nodes. According to the monitoring rules, determining the severity of anomaly nodes includes: structurally segmenting the target water area to obtain multiple monitoring sub-regions and establishing corresponding time-series data for each monitoring sub-region; identifying at least one adjacent sub-region that is adjacent to the monitoring sub-region containing the anomaly node; and calculating the similarity by comparing the time-series data corresponding to the anomaly node with the time-series data corresponding to at least one adjacent sub-region, and determining the severity of the anomaly node based on the comparison result of the similarity with a preset similarity threshold.

2. The method for monitoring water anomalies based on remote sensing big data according to claim 1, characterized in that, The updated water area data is compared with normal water area data to identify new anomalous nodes, including: If a point deviates from the normal water area data in the updated water area data, that point is extracted as a suspicious node. Obtain a time window centered on the suspicious node as the comparison interval; And within the comparison interval, the test node with the smallest difference in data value from the suspicious node will be marked as a new abnormal node.

3. A water anomaly monitoring system based on remote sensing big data, used to implement the water anomaly monitoring method based on remote sensing big data as described in claim 2, characterized in that, include: The water anomaly identification unit is used to identify anomaly nodes based on initial water area data derived from remote sensing images; Anomaly severity assessment unit is used to determine the severity of anomalies and generate a monitoring activation signal when it is determined that an anomaly constitutes an abnormal water body. And a dynamic monitoring unit, used to respond to the monitoring start signal, initiate and execute dynamic monitoring of abnormal water bodies.

4. A water anomaly monitoring system based on remote sensing big data according to claim 3, characterized in that, In response to the monitoring activation signal, dynamic monitoring of abnormal water bodies is initiated and executed, including: acquiring updated water area data, comparing the updated water area data with normal water area data to identify new abnormal nodes, and updating the monitoring list based on the new abnormal nodes.

5. A water anomaly monitoring system based on remote sensing big data according to claim 4, characterized in that, The water anomaly identification unit is specifically used for: Initial water data is analyzed to identify undulating nodes, and offset values ​​and anomaly thresholds are calculated for these nodes based on the evaluation model. The data portion corresponding to the fluctuating nodes with offset values ​​less than the abnormal threshold is identified as normal water area data. And data points that do not belong to the normal water area data in the initial water area data will be marked as abnormal nodes.

6. A water anomaly monitoring system based on remote sensing big data according to claim 5, characterized in that, The anomaly severity assessment unit is specifically used for: The target water area is structurally segmented to obtain multiple monitoring sub-areas, and corresponding time-series data are established for each monitoring sub-area. Furthermore, by comparing the time series data corresponding to the abnormal node with the time series data corresponding to the adjacent sub-region, the similarity is calculated, and the severity of the abnormal node is judged based on the comparison result of the similarity with the preset similarity threshold.

Images