A method for positioning and repairing wind speed data in the sea-land boundary area

Abstract

The present invention provides a method for locating and repairing wind speed data in a land-sea boundary area, which relates to the technical field of data processing. The method comprises: performing terrain zoning processing on original wind speed data; extracting wind speed sequences of continuous time periods in each sub-area, constructing a wind speed change curve with spatial position as a variable, and identifying short-term drastic change sections through the wind speed change curve; combining high-frequency disturbance characteristic values ​​in the wind speed sequences of the corresponding sub-area historical time periods to determine whether the short-term drastic change sections are terrain-induced disturbances; adopting a trend continuation calculation method based on time and space weight adjustment for the original wind speed data corresponding to the suspected abnormal sections, and performing interpolation repair using confirmed valid wind speed values ​​in adjacent time periods and adjacent sub-areas; and reintegrating the repaired wind speed data into the corresponding zoned wind speed data. The present invention improves the repair accuracy of wind speed data in the land-sea boundary area.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method for positioning and repairing wind speed data in a sea-land boundary area. Background Art

[0002] In existing technologies, wind speed data acquisition in land-sea boundary areas typically relies on multiple data sources, including meteorological monitoring stations, radar wind measurement equipment, and numerical meteorological models. To ensure data consistency and integrity, interpolation algorithms are often used to spatially align wind speed data from different sources and with varying degrees of accuracy. Time series analysis or filtering algorithms are also used to repair missing or anomalous data, ensuring the continuity and availability of wind speed field data. This type of technical solution has been widely used in scenarios such as offshore wind farm site selection, weather forecasting, and coastal disaster prevention.

[0003] However, in complex terrain areas, such as those between nearshore islands and land, existing technologies may be unable to effectively distinguish between wind speed variations influenced by terrain and abnormal data interference. For example, wind speed monitoring around offshore ports is affected by the terrain channel effect, resulting in dramatic and directional wind speed fluctuations. Using a unified interpolation algorithm for data processing could misidentify actual wind speed disturbances induced by terrain as outliers and eliminate them. This could distort the data restoration results, affecting the accuracy of subsequent safety warning models based on wind speed data. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for locating and repairing wind speed data in the sea-land boundary area, aiming to solve the problems mentioned in the background technology.

[0005] In order to solve the above technical problems, the technical solutions of the present invention are as follows:

[0006] A method for locating and repairing wind speed data in a land-sea boundary area, the method comprising:

[0007] The original wind speed data is processed by terrain partitioning, and the land-sea boundary area is divided into land sub-area, nearshore sub-area and sea sub-area. Based on the terrain characteristics of different sub-areas, the corresponding terrain partition matrix is ​​constructed to obtain the partition wind speed data with partition labels.

[0008] Based on the partitioned wind speed data, the wind speed sequence of continuous time periods in each sub-region is extracted, and a wind speed change curve with spatial position as the variable is constructed. The wind speed change curve is used to identify the short-term drastic change sections and obtain the fluctuation mark data.

[0009] Based on the fluctuation mark data, combined with the high-frequency disturbance characteristic values ​​in the wind speed series of the corresponding sub-area during the same period in history, it is determined whether the short-term drastic change section is a terrain-induced disturbance. If the disturbance feature matching threshold is met, it is marked as a valid disturbance section, otherwise it is marked as a suspected abnormal section, and the section label data is obtained;

[0010] According to the segment label data, the original wind speed data corresponding to the suspected abnormal segment is interpolated and repaired using the trend continuation calculation method based on time and space weight adjustment using the confirmed valid wind speed values ​​in the adjacent time period and adjacent sub-areas to obtain the repaired wind speed data.

[0011] The repaired wind speed data are reintegrated into the corresponding partitioned wind speed data to generate the final wind speed dataset.

[0012] Preferably, a corresponding terrain partition matrix is ​​constructed based on the terrain characteristics of different sub-regions to obtain partition wind speed data with partition labels, including:

[0013] Get the spatial coordinate labels in the original wind speed data;

[0014] According to the spatial coordinate labels, the elevation values, water contact information and terrain slope parameters of the corresponding locations are extracted from the terrain data to obtain a set of terrain factors;

[0015] According to the terrain factor set, the sub-region classification rules are matched to obtain the terrain partition matrix;

[0016] According to the terrain partition matrix, partition labels are added to the original wind speed data to obtain partition wind speed data.

[0017] Preferably, based on the partitioned wind speed data, the wind speed sequence of continuous time periods in each sub-region is extracted to construct a wind speed variation curve with spatial position as a variable, including:

[0018] According to the partitioned wind speed data, extract the timestamp set of each sub-region;

[0019] According to the timestamp set, the wind speed observation values ​​of the continuous time period are extracted to obtain the sub-area wind speed sequence;

[0020] According to the sub-area wind speed sequence, the wind speed value at the spatial position is extracted at each time point, and spatial splicing processing is performed to obtain the wind speed change curve.

[0021] Preferably, identifying a short-term drastic change section through the wind speed change curve to obtain fluctuation mark data includes:

[0022] Obtain a sequence of wind speed differences between various spatial positions in a wind speed variation curve;

[0023] According to the wind speed difference sequence, the spatial segments that continuously exceed the preset wind speed difference threshold are identified to obtain the short-term drastic change segments;

[0024] Based on the short-term drastic change segments, their start and end times, spatial range and change intensity are extracted to generate fluctuation marker data.

[0025] Preferably, based on the fluctuation mark data, combined with the high-frequency disturbance characteristic value in the wind speed sequence of the corresponding sub-area during the same period in history, it is determined whether the short-term drastic change section is a terrain-induced disturbance. If the disturbance characteristic matching threshold is met, it is marked as a valid disturbance section, otherwise it is marked as a suspected abnormal section, and the section label data is obtained, including:

[0026] Extract the short-term drastic change segments based on the fluctuation mark data;

[0027] According to the short-term drastic change section, the sub-area and start and end time are determined, and the corresponding sub-area data of the same period are extracted from the historical wind speed series to obtain the historical disturbance sample data;

[0028] Based on historical disturbance sample data, the disturbance periodicity, disturbance frequency and disturbance change rate are calculated to construct high-frequency disturbance characteristic values;

[0029] According to the high-frequency disturbance characteristic value, it is determined whether the changed segment meets the disturbance matching threshold and the segment label data is obtained.

[0030] Preferably, based on the segment label data, the original wind speed data corresponding to the suspected abnormal segment is interpolated and repaired using a trend continuation calculation method based on time and space weight adjustment, using the confirmed valid wind speed values ​​in the adjacent time period and adjacent sub-areas to obtain the repaired wind speed data, including:

[0031] According to the segment label data, the time period and spatial location marked as suspected anomaly are extracted to obtain the abnormal time period and abnormal spatial location;

[0032] According to the abnormal time period, the front and back range of the point to be repaired on the time axis is determined, and the wind speed observation values ​​in the time series are extracted to obtain the wind speed data of the time neighborhood;

[0033] According to the abnormal spatial location, the wind speed observation values ​​marked as effective disturbance sections are extracted from the adjacent sub-areas to obtain the spatial neighborhood wind speed data;

[0034] Based on the temporal neighborhood wind speed data and the spatial neighborhood wind speed data, the temporal trend results and the spatial trend results are calculated respectively. Based on the weighted coefficient corresponding to the temporal fitting error and the spatial distance calculation, the two trend results are weightedly fused to obtain the interpolation repair value.

[0035] According to the interpolation repair value, the original wind speed observation value in the suspected abnormal section is replaced to obtain the repaired wind speed data.

[0036] Preferably, based on historical disturbance sample data, the disturbance periodicity, disturbance frequency and disturbance change rate are calculated to construct a high-frequency disturbance characteristic value, including:

[0037] According to the historical disturbance sample data, the wind speed change sequence of each time period is extracted to obtain the disturbance wind speed fragment data;

[0038] According to the disturbance wind speed segment data, the disturbance duration, fluctuation frequency and wind speed change amplitude of each time period are calculated, and the disturbance periodicity, disturbance frequency and disturbance change rate are obtained respectively;

[0039] The disturbance periodicity, disturbance frequency and change rate are combined to construct the high-frequency disturbance eigenvalue.

[0040] Preferably, according to the abnormal spatial position, wind speed observation values ​​marked as effective disturbance sections are extracted from adjacent sub-areas to obtain spatial neighborhood wind speed data, including:

[0041] According to the abnormal spatial location, the geographical scope of the surrounding sub-area is extracted to obtain the surrounding sub-area range data;

[0042] According to the data of the surrounding sub-areas, the wind speed observation values ​​marked as effective disturbance sections are extracted to form the initial data set;

[0043] According to the initial data set, the spatial distance between the abnormal spatial position and the wind speed observation point is calculated, and the wind speed observation values ​​within the preset spatial neighborhood radius are screened out to obtain the spatial neighborhood wind speed data.

[0044] Preferably, the temporal trend result and the spatial trend result are calculated respectively according to the temporal neighborhood wind speed data and the spatial neighborhood wind speed data, and the two trend results are weightedly fused based on the weighted coefficient corresponding to the temporal fitting error and the spatial distance calculation to obtain the interpolation repair value, including:

[0045] According to the wind speed data of the time neighborhood, a time trend function is fitted to calculate the time extension value of the abnormal time point in the abnormal time period, and the time weight coefficient is calculated based on the trend fitting error;

[0046] According to the wind speed data of the spatial neighborhood, the spatial trend function is fitted to calculate the spatial extension value of the abnormal spatial position, and the spatial weight coefficient is calculated based on the spatial distance between the wind speed observation point and the abnormal spatial position;

[0047] According to the time weight coefficient and the space weight coefficient, the time extension value and the space extension value are weightedly fused to obtain an interpolation repair value.

[0048] Preferably, the repaired wind speed data is reintegrated into the corresponding partitioned wind speed data to generate a final wind speed data set, including:

[0049] Obtain the corresponding spatial coordinate labels and timestamp identifiers in the repaired wind speed data;

[0050] According to the spatial coordinate label and timestamp identifier, the matching wind speed observation value position is found in the corresponding partition wind speed data, and the replacement operation is performed to obtain the partition wind speed update data;

[0051] Summarize and integrate the updated wind speed data of each subarea, build a unified data structure, and generate a merged wind speed dataset;

[0052] According to the merged wind speed dataset, a data status label is added to each wind speed observation record to distinguish valid data from repaired data, and the final wind speed dataset is obtained.

[0053] The above solution of the present invention includes at least the following beneficial effects:

[0054] The present invention improves the application effect of existing technical solutions in complex terrain areas. The present invention combines terrain zoning processing with disturbance feature analysis to effectively distinguish between real disturbances induced by terrain in wind speed changes and abnormal data caused by equipment failure, data loss or other non-topographic factors, solving the limitation of existing technologies that cannot process data in complex terrain areas.

[0055] First, based on the terrain zoning of wind speed data in the land-sea boundary area, the area is divided into land sub-areas, nearshore sub-areas, and sea surface sub-areas. A terrain zoning matrix is ​​constructed according to the terrain characteristics of different sub-areas, and a partition label is assigned to each data point, thereby achieving a more refined wind speed data analysis. The wind speed variation characteristics are different under different terrain conditions. Through this zoning processing, a targeted wind speed data repair strategy can be provided for each sub-area. Unlike the traditional method of using a unified interpolation algorithm, the present invention can flexibly adjust the data repair method according to the characteristics of each sub-area, thereby improving the accuracy and adaptability of wind speed data processing.

[0056] Secondly, in the process of constructing the wind speed change curve, the present invention accurately judges the short-term drastic change in wind speed by identifying the short-term drastic change section and combining it with the high-frequency disturbance characteristic values ​​in the historical wind speed series of the same period. It can effectively distinguish between wind speed changes caused by terrain-induced disturbances and abnormal data interference, avoiding the situation in which the real terrain influence is misjudged as abnormal data and eliminated in traditional methods. Through this innovative disturbance identification and repair method, the present invention significantly improves the reliability of data repair and ensures the authenticity and continuity of the data, especially in the complex offshore island and land boundary area.

[0057] Finally, based on the trend extension calculation method with time and space weight adjustment, the present invention comprehensively considers the wind speed data of the time series and spatial neighborhood, and uses a nonlinear weighted fusion method to repair suspected abnormal data. This method can minimize the data omission or distortion problems caused by traditional repair methods while ensuring the rationality of the repair results. Through precise interpolation repair, the wind speed data set finally generated not only has a high spatial continuity, but also can provide a reliable basis for subsequent analysis based on wind speed data, especially in applications such as wind farm site selection, weather forecasting, coastal disaster prevention, etc., which greatly improves the credibility and practicality of the data.

[0058] In summary, based on the existing technology, the present invention utilizes innovative terrain zoning, disturbance identification and repair methods, as well as multiple time and space weighting strategies to solve many problems in wind speed data repair in complex terrain areas, significantly improves the accuracy and reliability of data repair, and has important practical application value and technical advantages. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a flowchart of a method for locating and repairing wind speed data in a land-sea boundary area provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0060] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0061] like Figure 1 As shown, an embodiment of the present invention provides a method for locating and repairing wind speed data in a land-sea boundary area, the method comprising:

[0062] S100, obtaining original wind speed data, wherein the original wind speed data is wind speed observation values ​​of multiple time periods collected by multiple meteorological collection devices located in the sea-land boundary area, and the original wind speed data has a spatial coordinate label and a timestamp identifier;

[0063] S200, performing terrain zoning processing on the original wind speed data, dividing the land-sea boundary area into land sub-areas, nearshore sub-areas, and sea surface sub-areas, and constructing corresponding terrain zoning matrices based on the terrain characteristics of different sub-areas to obtain zoned wind speed data with zone labels;

[0064] S300, extracting the wind speed sequence of continuous time periods in each sub-area based on the partitioned wind speed data, constructing a wind speed variation curve with spatial position as a variable, and identifying short-term drastic change sections through the wind speed variation curve to obtain fluctuation mark data;

[0065] S400: Based on the fluctuation mark data and the high-frequency disturbance characteristic values ​​in the wind speed sequence of the corresponding sub-area during the same period in history, determine whether the short-term drastic change section is a terrain-induced disturbance. If the disturbance characteristic matching threshold is met, it is marked as a valid disturbance section; otherwise, it is marked as a suspected abnormal section, and the section label data is obtained;

[0066] S500: Based on the segment label data, the original wind speed data corresponding to the suspected abnormal segment is interpolated and repaired using a trend continuation calculation method based on time and space weight adjustment, using the confirmed valid wind speed values ​​in the adjacent time period and adjacent sub-areas to obtain repaired wind speed data;

[0067] S600: Reintegrate the repaired wind speed data into the corresponding partitioned wind speed data to generate a final wind speed data set.

[0068] In an embodiment of the present invention, in the sea-land boundary area, due to the complex terrain structure and atmospheric disturbance mechanism, the wind speed data has problems such as violent fluctuations, sudden abnormalities and spatial discontinuity. The traditional meteorological data analysis model is difficult to maintain stability and effectiveness in this area. By obtaining the original wind speed data with spatial coordinate labels and timestamp identifiers, it is possible to accurately locate the data of the collection point and build a foundation for time series analysis, providing data support for subsequent processing. In the process of terrain zoning processing of the original wind speed data, the complex area is divided into land sub-areas, nearshore sub-areas and sea surface sub-areas, so that the data processing process in different areas can fully consider the differences in the impact of terrain conditions on wind speed. This zoning strategy enhances the accuracy of the interpretation of wind speed changes in subsequent analysis.

[0069] Furthermore, by extracting the wind speed sequence of continuous time periods within each sub-area and constructing a wind speed change curve with spatial position as a variable, it is possible to capture abnormal wind speed jumps in local areas. Based on the construction of the wind speed change curve, the spatial connectivity and gradient mutation identification methods are used to extract short-term drastic change sections, effectively avoiding the problem of interference in the overall trend judgment caused by local extreme values ​​or isolated jump points. By comparing and analyzing the high-frequency disturbance characteristic values ​​in the historical wind speed sequence of the same period, the difference between terrain-induced disturbances and random abnormal changes is identified, thereby completing the classification of effective disturbance sections and suspected abnormal sections, providing a basis for subsequent repair strategies.

[0070] When interpolating and repairing data using a trend continuation calculation method based on temporal and spatial weight adjustment, the interpolation and repair process balances spatial similarity and temporal consistency by introducing valid wind speed values ​​confirmed in the spatial neighborhood and the stability trend of the time series. This effectively suppresses the diffusion of local noise in the wind speed field and improves the credibility of the repair results. After the interpolation and repair is completed, the repaired wind speed data is reintegrated back into the partitioned wind speed data. The final wind speed dataset retains the spatial structural characteristics of the original data while also having high numerical integrity and anomaly repair capabilities. This provides a more stable data foundation for subsequent applications such as marine meteorological analysis, shipping route optimization, and land-sea interaction modeling.

[0071] The wind speed sequence is a set of wind speed data arranged in chronological order.

[0072] Among them, obtain the original wind speed data, specifically:

[0073] In this invention, raw wind speed data is obtained by observing the land-sea boundary over a long period of time using multiple meteorological data collection devices. Meteorological data collection devices typically include ground-based weather stations, weather buoys, weather satellites, radars, drones, and other equipment. These devices use sensors to collect wind speed information in the area in real time. Wind speed data consists of two important components:

[0074] Spatial Coordinate Labeling: Each wind speed observation point is assigned a spatial coordinate, typically a three-dimensional representation of its location on Earth using longitude, latitude, and altitude. Spatial coordinate labels ensure that data is aligned with its geographic location, facilitating spatial analysis, comparison, and further data processing.

[0075] Timestamp: Each wind speed observation is timestamped, indicating the exact time of measurement. Timestamps typically use a standard date and time format, such as "YYYY-MM-DDHH:MM:SS." This allows the data to reflect temporal variations in wind speed, providing a foundation for subsequent data recovery, trend analysis, and forecasting.

[0076] Raw wind speed data contains wind speed observations over multiple time periods. These data cover different time windows (such as hours, days, and months) and record the changes in wind speed within each time period. This raw data set provides a continuous wind speed time series, providing a complete information foundation for further analysis of wind speed changes.

[0077] In a preferred embodiment of the present invention, the above positioning can be achieved by the following steps, for example:

[0078] When the meteorological data collection device is installed, each device is assigned precise geographic coordinates (such as latitude and longitude, altitude) through satellite positioning systems such as GPS and Beidou or pre-deployment based on the geographic information system (GIS), and is bound to the collected wind speed data in real time to form raw data with spatial coordinate tags.

[0079] For example, if the coordinates of a coastal weather station are (120.5°E, 30.2°N, 10m above sea level), the wind speed data collected will be automatically associated with these coordinates.

[0080] Unique identifier of coordinates:

[0081] The spatial coordinate label serves as the "spatial address" of the data, ensuring that each wind speed observation corresponds to a unique physical location in the land-sea boundary area.

[0082] Coordinate-based spatial region division:

[0083] In the S200 terrain zoning process, the area is divided into land, nearshore, and offshore sub-areas by matching the spatial coordinates of the wind speed data with a database of terrain features (such as land boundaries, coastlines, and water depth data):

[0084] Land sub-area: The coordinates are located within a certain distance (e.g. 10km) from the coastline to the land side and the altitude is higher than sea level;

[0085] Nearshore sub-area: The coordinates are located within a certain range on the seaward side of the coastline (e.g. 0-5km sea area, water depth <10m);

[0086] Sea surface sub-area: coordinates are located in the open sea area outside the nearshore sub-area.

[0087] For example, the coordinates of a data point (120.6°E, 30.1°N) are compared with the coastline data. If it is located 3 km seaward of the coastline, it is classified as a nearshore sub-area.

[0088] Spatial mapping of the terrain partition matrix:

[0089] When constructing the partition matrix, each data point is mapped to the grid of the corresponding sub-area using the coordinates as the index, forming an association relationship of "spatial position-sub-area label" to realize the spatial positioning partitioning of the data.

[0090] Construction of wind speed curve in spatial dimension:

[0091] In step S300, a spatially continuous wind speed curve is constructed, with the spatial coordinates (e.g., longitude and latitude) of the data points as the horizontal axis and the wind speed values ​​as the vertical axis. For example, along a coastline, the coordinates (x1, y1), (x2, y2), etc. of multiple observation points correspond to wind speed values ​​v1, v2, etc., forming a curve reflecting the spatial distribution of wind speed.

[0092] Spatial location of the short-term drastic change segment:

[0093] By using conditions such as sudden changes in the curve slope and wind speed differences between adjacent points exceeding a threshold (such as 2 m / s), the system can identify sections of abnormal wind speed fluctuations and determine their specific locations within the land-sea boundary based on coordinates. For example, a section of the curve corresponding to the coordinates (120.3°E-120.4°E, 30.5°N) can be located as an abnormal wind speed area near a nearshore port.

[0094] Neighboring subregions and spatial weights:

[0095] In step S500, when repairing a suspected anomaly segment, the distance between the data point and the adjacent sub-area observation point is calculated based on the spatial coordinates of the data point. The closer the distance, the higher the spatial weight. For example, if a land data point (x0, y0) is anomaly and its adjacent land sub-area point (x1, y1) is 1 km away and the nearshore sub-area point (x2, y2) is 3 km away, the former will have a higher weight than the latter.

[0096] When calculating trend extension, the coordinates of the outlier point are taken as the center, and the coordinates and wind speed values ​​of adjacent valid points are used to generate repair values ​​based on spatial position relationships through methods such as inverse distance weighted interpolation (IDW) and Kriging interpolation to ensure that the repaired data conforms to the spatial distribution law of regional wind speed.

[0097] In a preferred embodiment of the present invention, a corresponding terrain partition matrix is ​​constructed based on the terrain characteristics of different sub-regions to obtain the partition wind speed data with partition labels, including:

[0098] Get the spatial coordinate labels in the original wind speed data;

[0099] According to the spatial coordinate labels, the elevation values, water contact information and terrain slope parameters of the corresponding locations are extracted from the terrain data to obtain a set of terrain factors;

[0100] According to the terrain factor set, the sub-region classification rules are matched to obtain the terrain partition matrix;

[0101] According to the terrain partition matrix, partition labels are added to the original wind speed data to obtain partition wind speed data.

[0102] In this embodiment of the present invention, by extracting the spatial coordinate labels of each data point in the raw wind speed data, it is possible to clearly locate the spatial attribution of the wind speed observations. Furthermore, by extracting the elevation value, water contact information, and terrain slope parameters corresponding to the coordinate location from the terrain data, a terrain factor set containing multiple spatial physical characteristics is formed. This allows data partitioning to no longer be based on simple geometric divisions, but instead reflect the actual topographic changes within the region.

[0103] By matching the sub-region classification rules with the aforementioned terrain factor set, the terrain type of the observation point can be automatically determined based on pre-set regional feature recognition rules, thereby constructing a well-structured terrain partitioning matrix. This matrix can be used as a data label mapping table to isolate the characteristics of wind speed data between different sub-regions, providing a clearer geographic basis for subsequent disturbance detection and remediation strategies.

[0104] Ultimately, the original wind speed data is assigned partition labels, forming partitioned wind speed data with geographic partitioning attributes. This data structure plays a crucial role in subsequent processing. For example, within the same region, a unified wind speed difference threshold setting and trend fitting model can be used. However, across regions, model parameters can be flexibly adjusted based on partition labels, significantly improving overall adaptability and response accuracy.

[0105] Among them, according to the spatial coordinate labels, the elevation value, water contact information and terrain slope parameters of the corresponding location are extracted from the terrain data to obtain the terrain factor set. Specifically:

[0106] When processing wind speed data, terrain information is one of the important factors affecting wind speed changes, especially in complex sea-land interface areas. To match terrain features with wind speed data, the present invention first extracts geographical feature information corresponding to wind speed observation points from the terrain data. The specific operation is as follows:

[0107] Elevation: The elevation value represents the vertical height of the observation point relative to sea level. In practical applications, wind speed variations are often affected by the undulating terrain, especially at the land-sea interface. Wind speeds can vary due to the presence of diverse terrain features such as mountains, hills, lowlands, and sea surface. By extracting the elevation information of the observation point, we can reveal the relationship between wind speed variations and topography, particularly how wind speed varies in unique terrain such as mountains, canyons, and coasts.

[0108] Water contact information: Water contact information refers to whether the observation point is in contact with or near a body of water, such as an ocean, lake, or river. Wind speeds near water bodies are often significantly affected by surface friction and climate, and may exhibit distinct wind speed patterns. Therefore, extracting water contact information can further understand the characteristics of areas affected by water in wind speed data.

[0109] Terrain Slope: Terrain slope indicates the gradient of the earth's surface, typically referring to the rate of change in surface elevation within a specific area. The greater the terrain slope, the greater the gradient of wind speed variation. This is particularly true at land-sea borders, in areas with large elevation differences, such as mountains and hills. The slope parameter helps understand how wind speed varies across different terrain conditions.

[0110] The three terrain factor sets (elevation, water contact information, and terrain slope parameters) are extracted and combined to form a terrain factor set for each wind speed observation point. These factors reflect the terrain characteristics of the wind speed observation point and provide the necessary geographic context for further analysis of wind speed data.

[0111] Among them, according to the terrain factor set, the sub-region classification rules are matched to obtain the terrain partition matrix. Specifically:

[0112] In this invention, the entire land-sea interface region is divided into different subregions, such as land subregions, nearshore subregions, and sea surface subregions. To match wind speed data to different terrain regions, a subregion classification rule matching is performed based on a set of terrain factors (including elevation, water contact information, terrain slope, etc.), thereby classifying each wind speed observation point into the appropriate terrain subregion. The specific steps of this process are as follows:

[0113] Subregion classification rules: Based on predefined rules, various parameters in a set of terrain factors (such as elevation, water contact information, and slope) are used to define the characteristics of each subregion. For example, high-elevation areas might be defined as land subregions, areas close to the coastline as nearshore subregions, and areas located in the ocean as sea surface subregions. The core of the classification rules is to match different region types based on terrain factors, ensuring that each observation point is consistent with the characteristics of the geographic region in which it is located.

[0114] Construction of a Terrain Zoning Matrix: After classification, each wind speed observation point is assigned a zoning label, identifying the sub-zonal category of each data point. This creates a terrain zoning matrix encompassing all observation points. Each row in the matrix represents a wind speed observation point, while the columns record the terrain zoning label for that point. This matrix provides a structured representation of the data in geographic space, facilitating subsequent analysis and processing.

[0115] By constructing a terrain partitioning matrix, wind speed data can be grouped and managed according to different geographical features, allowing for targeted processing of wind speed data in different regions. For example, different interpolation and restoration methods can be used for data in sea sub-regions, while more adaptable wind speed variation models can be used for data in land sub-regions or near-shore sub-regions.

[0116] In a preferred embodiment of the present invention, based on the partitioned wind speed data, a wind speed sequence of continuous time periods in each sub-region is extracted to construct a wind speed variation curve with spatial position as a variable, including:

[0117] According to the partitioned wind speed data, extract the timestamp set of each sub-region;

[0118] According to the timestamp set, the wind speed observation values ​​of the continuous time period are extracted to obtain the sub-area wind speed sequence;

[0119] According to the sub-area wind speed sequence, the wind speed value at the spatial position is extracted at each time point, and spatial splicing processing is performed to obtain the wind speed change curve.

[0120] In this embodiment of the present invention, after acquiring the zoned wind speed data, extracting wind speed observations for consecutive time periods within each sub-zone and constructing a wind speed variation curve are key steps in identifying spatial differences. First, by extracting a timestamp set for each sub-zone, the integrity and continuity of the data in the temporal dimension are ensured, avoiding trend analysis offsets caused by missing time points. Furthermore, wind speed observations within consecutive time periods are extracted in chronological order to form a wind speed sequence unique to each sub-zone, laying the foundation for subsequent spatiotemporal fusion modeling in terms of data structure.

[0121] In constructing the wind speed curve, we extract the wind speed values ​​at all current locations at each time point and perform spatial splicing to make the wind speed distribution characteristics in the spatial dimension explicit. This curve not only preserves the temporal trend of wind speed values ​​but also introduces spatial variation factors, forming a composite wind speed characteristic expression that combines spatial resolution and temporal dynamics.

[0122] This curve allows for more accurate identification of sudden wind speed changes within a spatial context, such as extreme value concentration and localized disturbance amplification, providing a stronger physical basis for subsequent disturbance detection and interpolation repair. In particular, constructing a wind speed curve dominated by spatial variables can significantly improve the ability to adapt to and identify complex disturbance patterns in areas with strong spatial gradients, such as the land-sea boundary.

[0123] Among them, according to the partitioned wind speed data, the timestamp set of each sub-region is extracted, specifically:

[0124] After completing the terrain partitioning of the raw wind speed data, we obtain wind speed observation data with partition labels. Each wind speed record contains three key pieces of information: spatial coordinates, timestamp, and wind speed value. To facilitate subsequent temporal continuity analysis, the data within each sub-region must first be organized by time.

[0125] During this process, the wind speed data for each subregion is traversed and the timestamps of all observations are extracted. This collection of timestamps represents the temporal distribution of all wind speed data for that subregion. The extracted timestamps can be standardized to a uniform format (e.g., UTC) to ensure that subsequent analysis avoids data mismatches or misordering due to inconsistent time formats.

[0126] The purpose of this step is to build a timeline basis for each sub-area to ensure that data sorting, splicing, and trend analysis can be performed based on a unified time dimension in subsequent operations.

[0127] Among them, according to the timestamp set, the wind speed observation values ​​of the continuous time period are extracted to obtain the sub-area wind speed sequence. Specifically:

[0128] After the timestamps are sorted, the wind speed data for each sub-area will be analyzed in time series, based on the order of the time axis. This step focuses on selecting temporally continuous data segments from each sub-area's wind speed data, meaning data sequences without significant time gaps or discontinuities.

[0129] This extraction of continuous time segments can be accomplished by setting a maximum permissible time interval threshold, for example, retaining only data segments where the interval between adjacent timestamps does not exceed a certain standard length. This process aims to eliminate data segments with missing, discontinuous, or observational anomalies, ensuring the temporal consistency and integrity of the underlying data used to construct the spatial wind speed curve.

[0130] Finally, the wind speed observations extracted from each sub-area over a continuous period of time are organized into a time-ordered wind speed sequence. These wind speed sequences serve as an important data foundation for subsequent spatial dimension analysis.

[0131] Among them, according to the sub-area wind speed sequence, the wind speed value at the spatial position is extracted at each time point, and spatial splicing processing is performed to obtain the wind speed change curve. Specifically:

[0132] After obtaining the wind speed time series in each sub-area, each time point will be used as a slice to extract and process the data of multiple spatial observation points at the same time to form a spatial wind speed distribution map.

[0133] Specifically, at each time point, the wind speed values ​​and corresponding spatial coordinates of all observation points in the subarea at that time are extracted. These spatial points are then connected or concatenated in order of their coordinates (for example, based on spatial distance, location index, etc.) to form a wind speed curve in the spatial dimension. This method can demonstrate the spatial distribution and changes of wind speed at a specific time.

[0134] In practice, spatial splicing can be performed using simple linear sorting or by optimizing the spatial sequence based on factors such as terrain characteristics and wind direction. The resulting wind speed curve not only reflects the spatial distribution of wind speed at a given moment, but also provides a graphical and structured analysis basis for identifying abnormal wind speed fluctuations and areas of short-term, drastic changes.

[0135] This spatial wind speed variation curve will serve as the core data structure for identifying fluctuation segments in subsequent steps and has important application value.

[0136] In a preferred embodiment of the present invention, the short-term drastic change section is identified through the wind speed change curve to obtain the fluctuation mark data, including:

[0137] Obtain a sequence of wind speed differences between various spatial positions in a wind speed variation curve;

[0138] According to the wind speed difference sequence, the spatial segments that continuously exceed the preset wind speed difference threshold are identified to obtain the short-term drastic change segments;

[0139] Based on the short-term drastic change segments, their start and end times, spatial range and change intensity are extracted to generate fluctuation marker data.

[0140] In an embodiment of the present invention, by constructing a wind speed variation curve and identifying short-term, drastic change segments through this curve, it is possible to accurately locate disturbances in wind speed data. By extracting a sequence of wind speed differences between spatial locations in the wind speed variation curve, spatial gradient information can be digitized, thereby identifying local mutation points or segments. Compared to traditional methods of determining disturbances based on the overall mean or variance, the use of difference sequences has higher sensitivity and resolution, making it particularly suitable for detecting small-scale but significant disturbances.

[0141] During the recognition process, by setting a wind speed difference threshold and determining whether consecutive spatial points meet this threshold, isolated disturbance noise can be effectively eliminated, retaining only disturbance segments with true evolutionary structure. This continuous disturbance recognition mechanism not only enhances the stability of anomaly detection but also improves the recognition rate of terrain-induced disturbances, as these disturbances often exhibit spatial clustering characteristics.

[0142] The start and end times, spatial extent, and intensity of each short-term, drastic change segment are extracted and encapsulated as fluctuation marker data, giving the entire disturbance identification process clear spatial-temporal boundaries and intensity information. This type of structured data representation not only facilitates subsequent comparison and matching with historical disturbance samples, but can also be used for tasks such as visualization, disturbance feature modeling, and uncertainty assessment. Overall, this processing method significantly improves the ability to respond to sudden, non-stationary disturbances, providing data support for further determining their physical causes and whether repair is necessary.

[0143] Among them, the wind speed difference sequence between each spatial position in the wind speed change curve is obtained, specifically:

[0144] In the previous steps, we generated a time- and space-based wind speed curve, describing the wind speed distribution at each spatial location at a given moment. To identify areas experiencing drastic changes, we first need to calculate the wind speed differences between these locations.

[0145] The process involves iterating through each time point and calculating the wind speed differences between adjacent spatial locations at that time. These differences reflect the local fluctuations in wind speed. Areas with large differences typically indicate significant changes in wind speed, possibly due to terrain features or other disturbances.

[0146] These difference series provide the foundational data for subsequent identification of areas of dramatic change, allowing us to focus on areas of large fluctuations while ignoring stable areas or noise points. In this way, we can filter out potential disturbance areas from a large amount of wind speed data, preparing for subsequent data repair.

[0147] Among them, according to the wind speed difference sequence, the spatial segments that continuously exceed the preset disturbance threshold are identified, and the short-term drastic change segments are obtained. Specifically:

[0148] After obtaining a series of wind speed differences between spatial locations, the next step is to identify areas experiencing significant wind speed fluctuations. This process begins by setting a disturbance threshold to determine which areas experience wind speed fluctuations outside the normal range. This threshold can be determined through historical data analysis or empirical rules.

[0149] When the wind speed difference between adjacent spatial locations continuously exceeds the preset disturbance threshold, it is considered that a short-term drastic change has occurred in the area. This change may be caused by sudden fluctuations in wind speed, weather changes, or terrain factors.

[0150] Once a section of significant change is identified, its start and end time and spatial extent are marked for each section, ensuring the clear location of the disturbance in time and space. Furthermore, the intensity of the change, defined as the absolute change in wind speed difference, is recorded. This marked data will serve as a crucial basis for subsequent repairs and analysis.

[0151] Among them, according to the short-term drastic change segment, its start and end time, spatial range and change intensity are extracted to generate fluctuation mark data. Specifically:

[0152] During this process, identified short-term, dramatic fluctuations are converted into specific fluctuation marker data. This data contains detailed information about each dramatic fluctuation segment: start time, end time, spatial extent, and intensity. This data structure not only clearly describes the spatiotemporal distribution of wind speed variations but also provides essential information for subsequent disturbance analysis and anomaly remediation.

[0153] Start and end time: record the specific time range of each drastic change segment, providing a basis for time series analysis;

[0154] Spatial range: Marks the spatial range covered by the drastic change segment to help locate the change area in geographic space;

[0155] Intensity of change: By calculating the magnitude of the wind speed difference within the section, the intensity of the change is determined, helping to judge the nature and impact of the disturbance.

[0156] Fluctuation marker data effectively distinguishes natural disturbances induced by terrain from wind speed anomalies caused by equipment failure, measurement errors, or other abnormal factors. This marker data is passed to the subsequent data repair step to ensure that only those disturbance segments that meet the criteria are repaired, thus avoiding the mistaken repair of true terrain disturbances.

[0157] In a preferred embodiment of the present invention, based on the fluctuation mark data and the high-frequency disturbance characteristic value in the wind speed sequence of the corresponding sub-area during the same period of history, it is determined whether the short-term drastic change section is a terrain-induced disturbance. If the disturbance characteristic matching threshold is met, it is marked as a valid disturbance section; otherwise, it is marked as a suspected abnormal section. The section label data is obtained, including:

[0158] Extract the short-term drastic change segments based on the fluctuation mark data;

[0159] According to the short-term drastic change section, the sub-area and start and end time are determined, and the corresponding sub-area data of the same period are extracted from the historical wind speed series to obtain the historical disturbance sample data;

[0160] Based on historical disturbance sample data, the disturbance periodicity, disturbance frequency and disturbance change rate are calculated to construct high-frequency disturbance characteristic values;

[0161] According to the high-frequency disturbance characteristic value, it is determined whether the changed segment meets the disturbance matching threshold and the segment label data is obtained.

[0162] In this embodiment of the present invention, for identified short-term, drastic fluctuations, a similarity match is performed using the high-frequency disturbance characteristic values ​​in historical wind speed samples to determine whether the disturbance is terrain-induced. This judgment process incorporates multiple disturbance characteristic parameters, including disturbance periodicity, disturbance frequency, and disturbance change rate, forming a high-frequency disturbance characteristic value indicator system. Compared with traditional methods that rely on a single change threshold for discrimination, this multi-parameter combined analysis approach significantly enhances the ability to identify patterns in complex disturbance contexts.

[0163] In practice, the relevant attributes of each drastically changing segment in the current fluctuation marker data, including its subregion affiliation and time range, are first extracted. Wind speed observations for the same subregion and time period are then extracted from the historical wind speed series. This spatial and temporal consistency comparison ensures the physical homology of the compared data, avoiding misjudgments due to regional differences or seasonal shifts.

[0164] Subsequently, various disturbance feature parameters are calculated from historical disturbance samples, and a multi-dimensional disturbance feature value is constructed. During the matching process, not only are individual feature items compared to see if they meet the standards, but also based on the set similarity evaluation mechanism, such as weighted comprehensive scoring or threshold interval comparison, it is determined whether the current disturbance meets the typical characteristics of high-frequency disturbances. This matching mechanism can accurately distinguish between conventional disturbances and potential anomalies, improve the accuracy of anomaly identification, and provide a judgment basis for subsequent repair steps. Ultimately, the drastic change segments that meet the matching conditions are marked as valid disturbance segments, otherwise they are marked as suspected anomaly segments, generating clearly structured segment label data and improving transparency and controllability.

[0165] Among them, according to the short-term drastic change section, the sub-area and start and end time are determined, and the corresponding sub-area time period data are extracted from the historical wind speed series to obtain the historical disturbance sample data. Specifically:

[0166] In this step, based on the short-term, drastic change segments identified in the previous step, the geographic subregion to which these segments belong is first determined. This process allows the change segments to be mapped to specific geographic locations within the land-sea interface (e.g., land subregions, nearshore subregions, or offshore subregions), and the specific spatiotemporal locations where the changes occurred are identified. This approach provides accurate geographic location for subsequent disturbance analysis.

[0167] Once the sub-region is determined, the next step is to extract the wind speed data from the corresponding sub-region's historical wind speed series, which coincides with the period of drastic changes. The historical wind speed series refers to wind speed observations over a period of time within the region, which contains the patterns of wind speed changes over different time periods.

[0168] The extracted historical wind speed data will be used to construct historical disturbance sample data. This sample data will help analyze the wind speed variation characteristics of the sub-area over similar time periods in the past. By comparing the similarity between the current period of rapid change and the historical data, it is possible to identify whether the change is caused by terrain-induced disturbances or abnormal fluctuations caused by other non-natural factors (such as equipment failure and measurement errors).

[0169] This process provides data support for subsequent disturbance matching, enabling matching and judgment of current disturbances based on historical data.

[0170] Among them, according to the high-frequency disturbance characteristic value, it is determined whether the change segment meets the disturbance matching threshold and the segment label data is obtained. Specifically:

[0171] After extracting historical disturbance sample data, the system then determines whether the current short-term, drastic change segment has similar characteristics to historical disturbances based on high-frequency disturbance characteristics (such as disturbance duration and frequency). If the similarity between the current segment's disturbance characteristics and historical disturbances reaches a preset disturbance matching threshold, the change segment is marked as a valid disturbance segment; otherwise, it is marked as a suspected abnormal segment.

[0172] This matching process is achieved through a threshold matching mechanism, which conducts a comprehensive analysis based on multiple characteristics such as the duration, frequency, and amplitude of the disturbance. Specifically, when parameters such as the duration or frequency of the disturbance are detected to be outside the set range, it indicates that the disturbance is a significant abnormal event, which may be caused by equipment problems or data collection errors.

[0173] This approach allows for precise distinction between true disturbances induced by terrain and erroneous data caused by anomalies, effectively ensuring the accuracy of subsequent data restoration. Ultimately, segment labels are generated for each segment of short-term, drastic change. These data record the segment type (valid disturbance or suspected anomaly) and its specific location in time and space.

[0174] In a preferred embodiment of the present invention, based on the segment label data, the original wind speed data corresponding to the suspected abnormal segment is interpolated and repaired using the confirmed valid wind speed values ​​in the adjacent time period and adjacent sub-areas using a trend continuation calculation method based on time and space weight adjustment to obtain the repaired wind speed data, including:

[0175] According to the segment label data, the time period and spatial location marked as suspected anomaly are extracted to obtain the abnormal time period and abnormal spatial location;

[0176] According to the abnormal time period, the front and back range of the point to be repaired on the time axis is determined, and the wind speed observation values ​​in the time series are extracted to obtain the wind speed data of the time neighborhood;

[0177] According to the abnormal spatial location, the wind speed observation values ​​marked as effective disturbance sections are extracted from the adjacent sub-areas to obtain the spatial neighborhood wind speed data;

[0178] Based on the temporal neighborhood wind speed data and the spatial neighborhood wind speed data, the temporal trend results and the spatial trend results are calculated respectively. Based on the weighted coefficient corresponding to the temporal fitting error and the spatial distance calculation, the two trend results are weightedly fused to obtain the interpolation repair value.

[0179] According to the interpolation repair value, the original wind speed observation value in the suspected abnormal section is replaced to obtain the repaired wind speed data.

[0180] In this embodiment of the present invention, for wind speed segments identified as suspected anomalies, interpolation repair is performed by constructing temporal and spatial neighborhood wind speed data, performing trend continuation calculations on each, and fusing the predicted results. Compared to a single trend repair method, this method incorporates data support from both temporal and spatial dimensions, significantly improving the accuracy and stability of the repaired values.

[0181] First, all time periods and spatial locations to be repaired are extracted using segment label data. Temporal neighborhood wind speed data is then constructed by combining the preceding and following observation points on the time axis. This temporal neighborhood reflects the changing trends of the same measurement point over time and is suitable for fitting stable time series models and obtaining temporal trend forecasts. Concurrently, wind speed observations marked as valid disturbances are extracted from the subregions surrounding the anomalous spatial location to form spatial neighborhood wind speed data. This neighborhood construction method fully considers spatial similarity and avoids contamination of the repair calculations by anomalous data.

[0182] In the trend extension stage, trend functions are fitted based on the temporal neighborhood and spatial neighborhood respectively to obtain two sets of prediction values. To ensure that the weighted fusion results have physical credibility, the fitting error and spatial distance of each trend prediction value are calculated, and the time weight coefficient and spatial weight coefficient are constructed accordingly. These weights reflect the credibility and representativeness of each prediction result in the overall trend fitting. Finally, the two sets of prediction values ​​are weighted and fused according to the weight coefficient to obtain the interpolated repair value, which is used to replace the corresponding abnormal observation value in the original wind speed data. The wind speed data repaired in this way maintains both local consistency and overall continuity, effectively improving the integrity and quality of the data set, and can better support subsequent spatial wind field modeling and analysis tasks.

[0183] Among them, according to the segment label data, the time period and spatial location marked as suspected anomaly are extracted to obtain the abnormal time period and abnormal spatial location. Specifically:

[0184] The previous steps have generated segment labels for each variation segment. In this step, we use this label data to extract the time periods and spatial locations marked as suspected anomalies from all wind speed observations. This process allows us to filter out those wind speed data segments marked as suspected anomalies for further analysis and processing.

[0185] Suspected anomalies are data that, after analysis, may not conform to conventional wind speed patterns. These data often exhibit unusual fluctuations in time and space, potentially due to factors such as equipment failure and observation errors. By filtering segment-labeled data, the temporal and spatial locations of suspected anomalies can be precisely extracted, providing target data for subsequent repair work.

[0186] Once the abnormal time period and spatial location are determined, these data will serve as key areas for further repair and verification. Wind speed observation data in these suspected abnormal areas will be further analyzed to try to identify the underlying causes and implement repairs.

[0187] Among them, according to the abnormal time period, the front and back range of the point to be repaired on the time axis is determined, and the wind speed observation values ​​in the time series are extracted to obtain the time neighborhood wind speed data. Specifically:

[0188] After identifying the suspected anomaly time period, the next step is to conduct an in-depth analysis of the wind speed data within that time period. This process begins by determining the time range before and after the point to be repaired. This time range is determined based on the temporal characteristics of wind speed observations, and typically several time points before and after the anomaly period are selected as the repair range.

[0189] Once the time range is determined, wind speed observations are extracted from that time period and the temporal neighborhood wind speed data for that period is generated. The temporal neighborhood dataset contains the time series data before and after the point to be repaired, providing the necessary historical data foundation for interpolation repair.

[0190] This temporal neighborhood data typically includes wind speed observations adjacent to the repair point, reflecting the changing trend of wind speed over time. By analyzing this neighborhood data, the wind speed value at the repair point can be predicted and repaired, ensuring data continuity and consistency over time.

[0191] In this way, it is possible to effectively recover missing or erroneous wind speed data due to equipment problems or other factors, and ensure the naturalness and rationality of data repair.

[0192] In a preferred embodiment of the present invention, based on historical disturbance sample data, the disturbance periodicity, disturbance frequency and disturbance change rate are calculated to construct a high-frequency disturbance characteristic value, including:

[0193] According to the historical disturbance sample data, the wind speed change sequence of each time period is extracted to obtain the disturbance wind speed fragment data;

[0194] According to the disturbance wind speed segment data, the disturbance duration, fluctuation frequency and wind speed change amplitude of each time period are calculated, and the disturbance periodicity, disturbance frequency and disturbance change rate are obtained respectively;

[0195] The disturbance periodicity, disturbance frequency and change rate are combined to construct the high-frequency disturbance characteristic value; among them,

[0196] ,

[0197] is the high-frequency disturbance characteristic value, which is used to measure the intensity of high-frequency disturbance in wind speed data and reflect the severity of wind speed changes;

[0198] is the perturbation frequency, which indicates the number of perturbation segments that occur per unit time; it is calculated by the ratio of the number of perturbation segments to the total duration;

[0199] is the disturbance change rate; it represents the amplitude of wind speed change in each disturbance segment; it is obtained by dividing the wind speed difference within the disturbance segment by the duration;

[0200] is the standard deviation of wind speed variation; it represents the fluctuation range of wind speed observations in the disturbance segment, reflecting the degree of wind speed variation;

[0201] is the adjustment coefficient, which is used to control the influence of disturbance duration on the high-frequency disturbance characteristic value;

[0202] is the disturbance duration, which indicates the duration of each disturbance segment; it represents the time span when the wind speed changes drastically;

[0203] is the mean wind speed, which represents the average wind speed in the disturbance segment;

[0204] is the number of perturbation segments, which indicates the total number of perturbation segments identified in a given time window;

[0205] is the total length of the analysis period, indicating the total length of time used to calculate the disturbance frequency;

[0206] 、 For the The maximum wind speed and minimum wind speed of each disturbance segment are used to calculate the wind speed variation amplitude of the disturbance segment;

[0207] For the The duration of the disturbance segment; used to calculate the disturbance duration;

[0208] For the wind speed observations; used to calculate the standard deviation of wind speed variation;

[0209] is the total number of wind speed observation points in the disturbance segment; it is used to calculate the standard deviation of wind speed variation.

[0210] In this embodiment of the present invention, through in-depth analysis of historical disturbance sample data, the periodicity, frequency, and rate of change of the disturbance can be calculated, thereby constructing high-frequency disturbance feature values, further improving the ability to identify and classify disturbance features in wind speed data. Using the disturbed wind speed segment data, it is possible to extract wind speed variation sequences within each time period, analyze the fluctuation characteristics, and calculate the duration of the disturbance, the frequency of fluctuations, and the magnitude of wind speed changes, thus providing accurate data support for the quantitative assessment of disturbance characteristics.

[0211] Specifically, when calculating the duration of a disturbance, the start and end times of the fluctuation segments are detected and time-scaled to ensure stability and consistency in the duration calculation. For the frequency of fluctuations, the frequency data is obtained by counting the number of disturbance events within a period and then comprehensively evaluated in combination with the magnitude of wind speed changes. In the rate of change calculation, the magnitude of wind speed changes within each disturbance segment is quantified to further determine the intensity and impact range of the disturbance.

[0212] Among them, according to the disturbance wind speed segment data, the disturbance duration, fluctuation frequency and wind speed change amplitude of each time period are calculated, and the disturbance periodicity, disturbance frequency and disturbance change rate are obtained respectively. Specifically:

[0213] In this process, the first step is to extract the disturbance wind speed segments from the wind speed data. These segments are obtained by identifying the sections with sharp changes in the wind speed curve, usually those with sharp changes and short duration. The following three calculations are performed on these disturbance segments:

[0214] Disturbance Duration: Disturbance duration refers to the time span over which each wind speed disturbance occurs, typically defined by the start and end of the dramatic change in wind speed. By identifying how wind speed changes over a specific time period, the duration of each disturbance can be precisely determined.

[0215] Fluctuation frequency: Fluctuation frequency represents the number of wind speed disturbances that occur per unit time. The frequency of disturbances is determined by counting how often these disturbances occur in the time series. A high fluctuation frequency may indicate frequent wind speed fluctuations in the area, potentially due to external climate conditions or topographical factors.

[0216] Wind Speed ​​Variation: The wind speed variation refers to the maximum change in wind speed within each disturbance segment. The variation is calculated by calculating the difference between the maximum and minimum wind speeds within a disturbance segment. Disturbances with larger variation amplitudes often indicate more severe wind speed fluctuations, potentially impacting wind farms or weather forecasts.

[0217] By calculating these characteristics, we can assign specific values ​​to each disturbance segment, reflecting the disturbance's periodicity (disturbance duration), frequency (frequency of fluctuations), and intensity (magnitude of wind speed changes). These calculations provide the foundational data for the subsequent construction of disturbance characteristic values.

[0218] Among them, the disturbance periodicity, disturbance frequency and change rate are combined to construct the high-frequency disturbance characteristic value. Specifically:

[0219] Building on the previous process, we combine the three characteristics of disturbance periodicity, disturbance frequency, and wind speed variation to generate a comprehensive indicator called the high-frequency disturbance eigenvalue. This eigenvalue comprehensively describes the characteristics of wind speed disturbances, taking into account factors such as the duration, frequency, and intensity of the disturbance, and more accurately reflects the actual impact of wind speed disturbances.

[0220] Specifically, these three characteristics are combined according to a preset weighting rule to produce a comprehensive index. The contribution of different disturbance characteristics (such as duration, frequency, and amplitude) to this index can be adjusted based on actual application requirements. As a quantitative indicator of wind speed disturbance, the high-frequency disturbance characteristic value provides a reliable basis for subsequent disturbance analysis and data repair, and is particularly valuable in wind speed anomaly detection and forecasting.

[0221] In this way, the multi-dimensional information of wind speed disturbance can be integrated into a single, easy-to-understand eigenvalue, making subsequent analysis and repair more concise and efficient.

[0222] In a preferred embodiment of the present invention, based on the abnormal spatial position, wind speed observation values ​​marked as effective disturbance sections are extracted from adjacent sub-areas to obtain spatial neighborhood wind speed data, including:

[0223] According to the abnormal spatial location, the geographical scope of the surrounding sub-area is extracted to obtain the surrounding sub-area range data;

[0224] According to the data of the surrounding sub-areas, the wind speed observation values ​​marked as effective disturbance sections are extracted to form the initial data set;

[0225] According to the initial data set, the spatial distance between the abnormal spatial position and the wind speed observation point is calculated, and the wind speed observation values ​​within the preset spatial neighborhood radius are screened out to obtain the spatial neighborhood wind speed data.

[0226] In this embodiment of the present invention, the geographic scope of the surrounding sub-areas is extracted by analyzing the spatial location of the anomaly, and wind speed observations marked as valid disturbance sections are then screened to form a spatial neighborhood wind speed dataset. First, based on the spatial location of the anomaly, the geographic scope data of the surrounding sub-areas is extracted. This process ensures that the wind speed characteristics and spatial continuity of the neighboring area are fully considered during the repair process. In this way, wind speed observation points within the affected area can be accurately identified, thus preventing interference with the overall repair effect caused by local anomaly data.

[0227] After extracting wind speed observations marked as valid disturbances, the spatial distance between the anomaly's spatial location and neighboring wind speed observation points is calculated to filter out observations within a preset spatial neighborhood radius. This screening process not only ensures the spatial consistency of the data but also effectively avoids repair errors caused by sparse or discontinuous data. In this way, each anomaly segment can be effectively spatially repaired, ensuring that the repaired wind speed data remains spatially continuous and consistent, while also providing a reliable data foundation for subsequent trend continuation calculations.

[0228] Ultimately, by effectively processing these spatial neighborhood data, high-quality wind speed repair data is generated, providing strong support for the integrity and accuracy of the wind speed dataset. This technical solution demonstrates excellent adaptability and refined processing capabilities in complex terrain areas, particularly at the interface between land and sea.

[0229] These characteristic calculation results can fully reflect the essential laws of wind speed changes and provide the necessary data support for the subsequent classification and repair of disturbances. By combining factors such as disturbance periodicity, frequency, and rate of change, a more comprehensive and efficient high-frequency disturbance characteristic value can be constructed, providing a stronger theoretical basis for the repair and refinement of wind speed data. In this way, not only the accuracy of wind speed data repair is improved, but also the sensitivity to wind speed changes under complex terrain conditions is enhanced, effectively solving the limitation of conventional models that fail to fully consider disturbance details.

[0230] In a preferred embodiment of the present invention, the temporal trend result and the spatial trend result are calculated respectively based on the temporal neighborhood wind speed data and the spatial neighborhood wind speed data, and the two trend results are weightedly fused based on the weighting coefficient corresponding to the temporal fitting error and the spatial distance calculation to obtain the interpolation repair value, including:

[0231] According to the wind speed data of the time neighborhood, a time trend function is fitted to calculate the time extension value of the abnormal time point in the abnormal time period, and the time weight coefficient is calculated based on the trend fitting error;

[0232] According to the wind speed data of the spatial neighborhood, the spatial trend function is fitted to calculate the spatial extension value of the abnormal spatial position, and the spatial weight coefficient is calculated based on the spatial distance between the wind speed observation point and the abnormal spatial position;

[0233] According to the time weight coefficient and the space weight coefficient, the time extension value and the space extension value are weightedly fused to obtain an interpolation repair value; wherein,

[0234] ,

[0235] is the interpolation repair value, which represents the wind speed observation value after interpolation repair and is used to replace suspected abnormal data;

[0236] The time trend forecast value is the wind speed forecast value fitted based on the wind speed data of the time neighborhood, which reflects the trend of wind speed changes in the time series;

[0237] The spatial trend prediction value is the wind speed prediction value fitted based on the wind speed data of the spatial neighborhood, which reflects the trend of wind speed change at the spatial location;

[0238] is the adjustment coefficient, a constant used to adjust the influence of the temporal trend and spatial trend prediction values ​​on the final interpolation repair value. The larger the value, the greater the impact of the predicted trend value on the repair result;

[0239] is the time weight coefficient, which is calculated based on the time trend fitting error and is used to indicate the reliability of the time trend forecast value. The smaller the error, the larger the time weight coefficient.

[0240] is the spatial weight coefficient, which is calculated based on the spatial distance between the wind speed observation point and the abnormal point in the spatial neighborhood. It is used to indicate the reliability of the spatial trend prediction value. The closer the distance, the larger the spatial weight coefficient.

[0241] is the time trend fitting error, which represents the difference between the time trend prediction value and the actual observation value, usually expressed as the residual sum of squares or mean square error;

[0242] is the spatial distance, which represents the spatial distance between the wind speed observation point and the abnormal spatial location, and is used to calculate the spatial weight coefficient;

[0243] For the Actual wind speed values, wind speed observations in the time neighborhood;

[0244] For the Wind speed forecast value, forecast value after time trend fitting;

[0245] is the number of wind speed observations in the temporal neighborhood, indicating the number of neighborhood wind speed observation points used to calculate the temporal trend;

[0246] For the The Euclidean distance between the spatial neighborhood wind speed observation point and the anomaly point represents the distance from the spatial neighborhood wind speed observation point to the location to be repaired;

[0247] is the number of spatial neighborhood wind speed observation points, indicating the number of neighborhood wind speed observation points used to calculate the spatial trend.

[0248] In an embodiment of the present invention, during the interpolation repair process, the trend results of the temporal neighborhood wind speed data and the spatial neighborhood wind speed data are calculated separately, and the corresponding weight coefficients are calculated based on the fitting error and the spatial distance. The two types of trend results are weighted and fused to obtain the interpolation repair value. Specifically, during the temporal trend calculation process, based on the wind speed observations within the temporal neighborhood, the temporal extension value is calculated using the trend fitting model, and the temporal weight coefficient is calculated based on the fitting error. At the same time, during the spatial neighborhood calculation process, the spatial trend function is fitted using the wind speed data within the spatial neighborhood, and the corresponding spatial weight coefficient is calculated based on the spatial distance.

[0249] By introducing temporal and spatial weight coefficients, this method allows the interpolation repair process to comprehensively consider the stability of the time series and the similarity of the spatial distribution, ensuring that the repair results are more consistent with the actual situation. Specifically, the temporal weight coefficient depends on the error of the temporal trend fitting; smaller errors will make the temporal trend results occupy a greater weight in the weighted fusion; while the spatial weight coefficient depends on the relative distance of the observation points within the spatial neighborhood; smaller spatial distances will make the spatial trend results occupy a greater weight in the weighted fusion.

[0250] The interpolated restoration values ​​obtained through weighted fusion not only improve the accuracy of data restoration but also fully reflect the temporal and spatial characteristics of wind speed data. This technical solution is particularly suitable for wind speed data restoration in complex areas, effectively resolving data errors caused by temporal fluctuations or spatial differences, and improving the credibility and stability of the dataset.

[0251] Among them, according to the wind speed data of the time neighborhood, the time trend function is fitted, the time extension value of the abnormal time point in the abnormal time period is calculated, and the time weight coefficient is calculated based on the trend fitting error. Specifically:

[0252] In this step, trend fitting is first performed based on the temporal neighborhood wind speed data, i.e., the wind speed observations that are temporally adjacent to the point to be repaired. Trend fitting involves analyzing the temporal neighborhood data to extract the trend pattern of wind speed changes. This can be achieved using various regression analysis methods (such as linear regression and polynomial regression), with the specific fitting method depending on the characteristics of the data.

[0253] After trend fitting, a trend function is obtained, which represents the temporal trend of wind speed changes during the abnormal time period. Based on this trend function, the time extension value of the abnormal time point is calculated, that is, the reasonable wind speed value at that time point is estimated based on historical data.

[0254] During the fitting process, the trend fitting error—the difference between the fitted function and the actual observed data—is calculated. Areas with smaller fitting errors indicate more accurate predictions from the trend function, while areas with smaller fitting errors indicate significant deviations from the fitting results. Based on this error, a time weighting coefficient is calculated, reflecting the credibility of the time trend function. The smaller the error, the larger the time weighting coefficient, indicating that the time trend contributes more to the restoration results.

[0255] Among them, according to the spatial neighborhood wind speed data, the spatial trend function is fitted, the spatial extension value of the abnormal spatial position is calculated, and the spatial weight coefficient is calculated based on the spatial distance between the wind speed observation point and the abnormal spatial position. Specifically:

[0256] After calculating the temporal trend, we next analyze the spatially adjacent wind speed data—that is, the wind speed observations spatially adjacent to the point to be repaired—to perform spatial trend fitting. The goal of spatial trend fitting is to infer the wind speed trend at the anomaly location based on the wind speed data of neighboring points. Spatial trend fitting methods are similar to temporal trend fitting and can include methods such as regression analysis.

[0257] By fitting the spatial neighborhood data, a spatial trend function is derived, which represents how wind speed varies with spatial location within that spatial range. Based on this spatial trend function, the spatial extension value of the abnormal spatial location can be calculated, that is, the wind speed value at that location can be estimated.

[0258] At the same time, a spatial weight coefficient is calculated based on the spatial distance between the wind speed observation point and the anomaly's spatial location. Observation points that are closer in spatial distance have a greater impact on the repair results, while observation points that are farther away have a smaller impact. This method prioritizes wind speed data from nearby observation points, allowing for more accurate inference of the repair value for the anomaly.

[0259] In a preferred embodiment of the present invention, the repaired wind speed data is reintegrated into the corresponding partitioned wind speed data to generate a final wind speed data set, including:

[0260] Obtain the corresponding spatial coordinate labels and timestamp identifiers in the repaired wind speed data;

[0261] According to the spatial coordinate label and timestamp identifier, the matching wind speed observation value position is found in the corresponding partition wind speed data, and the replacement operation is performed to obtain the partition wind speed update data;

[0262] Summarize and integrate the updated wind speed data of each subarea, build a unified data structure, and generate a merged wind speed dataset;

[0263] According to the merged wind speed dataset, a data status label is added to each wind speed observation record to distinguish valid data from repaired data, and the final wind speed dataset is obtained.

[0264] In this embodiment of the present invention, after completing interpolation and repair of wind speed data in suspected anomaly sections, to ensure data structure integrity and consistent data usage, this method reintegrates the repaired wind speed data into the corresponding subarea wind speed data, constructing a final wind speed dataset with spatial continuity and credibility. This step further improves the organization and usability of the wind speed dataset, providing complete and reliable data support for subsequent meteorological analysis, model building, and early warning.

[0265] Specifically, before performing the integration operation, the spatial coordinate labels and timestamps corresponding to each piece of repaired wind speed data are first extracted to ensure that each piece of data can be accurately located in the spatial and temporal dimensions, avoiding data misalignment or overlap during the integration process. This location information serves as a unique index for data insertion, allowing the repaired data to accurately replace the original observations at the corresponding locations in the partitioned wind speed data, effectively retaining the verified reliable data while eliminating any anomalies or missing data.

[0266] After the replacement operation, the updated data from each subregion is aggregated and integrated into a consistent, merged wind speed dataset. This process eliminates the need to distinguish between data source regions. Instead, the data table is reconstructed in chronological order and spatial layout based on a unified data structure. This step ensures spatial continuity of wind speed data across the entire land-sea boundary, facilitating subsequent multidimensional wind speed field modeling, spatial interpolation visualization, and historical data comparison.

[0267] Furthermore, to enhance data traceability and transparency, a status tag was added to each wind speed data record during the integration process, explicitly identifying whether the data has been repaired and whether it is an original, valid observation. The introduction of status tags allows subsequent users to perform conditional filtering, weighted modeling, or anomaly prediction based on data type when processing the dataset, thereby improving the accuracy and flexibility of data usage. This feature not only improves data management efficiency but also provides a convenient interface for subsequent automated meteorological processing workflows.

[0268] Overall, this step not only achieves seamless integration between the repaired wind speed data and the partitioned wind speed data, solving the problems of data fragmentation and untraceable sources in the traditional repair process, but also enhances the credibility management capabilities of the dataset through the status labeling mechanism, and constructs a final wind speed dataset with standardized structure, complete content, and controllable quality.

[0269] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for locating and repairing wind speed data in the sea-land boundary area, characterized in that: The method comprises: The original wind speed data is processed by terrain partitioning, and the land-sea boundary area is divided into land sub-area, nearshore sub-area and sea sub-area. Based on the terrain characteristics of different sub-areas, the corresponding terrain partition matrix is ​​constructed to obtain the partition wind speed data with partition labels. Based on the partitioned wind speed data, the wind speed sequence of continuous time periods in each sub-region is extracted, and a wind speed change curve with spatial position as the variable is constructed. The wind speed change curve is used to identify the short-term drastic change sections and obtain the fluctuation mark data. Based on the fluctuation mark data, combined with the high-frequency disturbance characteristic values ​​in the wind speed series of the corresponding sub-area during the same period in history, it is determined whether the short-term drastic change section is a terrain-induced disturbance. If the disturbance feature matching threshold is met, it is marked as a valid disturbance section, otherwise it is marked as a suspected abnormal section, and the section label data is obtained; According to the segment label data, the original wind speed data corresponding to the suspected abnormal segment is calculated by the trend extension calculation method based on time and space weight adjustment. The valid wind speed values ​​in the previous and next range time periods and adjacent sub-areas are used for interpolation and repair to obtain the repaired wind speed data. Specifically, the following steps are used: According to the segment label data, the time period and spatial location marked as suspected anomaly are extracted to obtain the abnormal time period and abnormal spatial location; According to the abnormal time period, the front and back range of the point to be repaired on the time axis is determined, and the wind speed observation values ​​in the time series are extracted to obtain the wind speed data of the time neighborhood; According to the abnormal spatial location, the wind speed observation values ​​marked as effective disturbance sections are extracted from the adjacent sub-areas to obtain the spatial neighborhood wind speed data; According to the wind speed data of the time neighborhood, a time trend function is fitted to calculate the time extension value of the abnormal time point in the abnormal time period, and the time weight coefficient is calculated based on the trend fitting error; According to the wind speed data of the spatial neighborhood, the spatial trend function is fitted to calculate the spatial extension value of the abnormal spatial position, and the spatial weight coefficient is calculated based on the spatial distance between the wind speed observation point and the abnormal spatial position; Performing weighted fusion on the time extension value and the space extension value according to the time weight coefficient and the space weight coefficient to obtain an interpolation repair value; According to the interpolation repair value, the original wind speed observation value in the suspected abnormal section is replaced to obtain the repaired wind speed data; The repaired wind speed data are reintegrated into the corresponding partitioned wind speed data to generate the final wind speed dataset.

2. The method for locating and repairing wind speed data in the sea-land boundary area according to claim 1 is characterized in that: Based on the terrain characteristics of different sub-areas, the corresponding terrain partition matrix is ​​constructed to obtain the partition wind speed data with partition labels, including: Get the spatial coordinate labels in the original wind speed data; According to the spatial coordinate labels, the elevation values, water contact information and terrain slope parameters of the corresponding locations are extracted from the terrain data to obtain a set of terrain factors; According to the terrain factor set, the sub-region classification rules are matched to obtain the terrain partition matrix; According to the terrain partition matrix, partition labels are added to the original wind speed data to obtain partition wind speed data.

3. The method for locating and repairing wind speed data in the sea-land boundary area according to claim 2 is characterized in that: Based on the partitioned wind speed data, the wind speed sequence of each sub-region in a continuous time period is extracted, and a wind speed variation curve with spatial position as the variable is constructed, including: According to the partitioned wind speed data, extract the timestamp set of each sub-region; According to the timestamp set, the wind speed observation values ​​of the continuous time period are extracted to obtain the sub-area wind speed sequence; According to the sub-area wind speed sequence, the wind speed value at the spatial position is extracted at each time point, and spatial splicing processing is performed to obtain the wind speed change curve.

4. The method for locating and repairing wind speed data in the land-sea boundary area according to claim 3 is characterized in that: The wind speed variation curve is used to identify the short-term drastic change section and obtain the fluctuation mark data, including: Obtain a sequence of wind speed differences between various spatial positions in a wind speed variation curve; According to the wind speed difference sequence, the spatial segments that continuously exceed the preset wind speed difference threshold are identified to obtain the short-term drastic change segments; Based on the short-term drastic change segments, their start and end times, spatial range and change intensity are extracted to generate fluctuation marker data.

5. The method for locating and repairing wind speed data in the land-sea boundary area according to claim 4 is characterized in that: Based on the fluctuation mark data, combined with the high-frequency disturbance characteristic values ​​in the wind speed series of the corresponding sub-area during the same period in history, it is determined whether the short-term drastic change section is a terrain-induced disturbance. If the disturbance feature matching threshold is met, it is marked as a valid disturbance section, otherwise it is marked as a suspected abnormal section. The section label data is obtained, including: Extract the short-term drastic change segments based on the fluctuation mark data; According to the short-term drastic change section, the sub-area and start and end time are determined, and the corresponding sub-area data of the same period are extracted from the historical wind speed series to obtain the historical disturbance sample data; Based on historical disturbance sample data, the disturbance periodicity, disturbance frequency and disturbance change rate are calculated to construct high-frequency disturbance characteristic values; According to the high-frequency disturbance characteristic value, it is determined whether the changed segment meets the disturbance matching threshold and the segment label data is obtained.

6. The method for locating and repairing wind speed data in the land-sea boundary area according to claim 5 is characterized in that: Based on historical disturbance sample data, the disturbance periodicity, disturbance frequency and disturbance change rate are calculated to construct high-frequency disturbance characteristic values, including: According to the historical disturbance sample data, the wind speed change sequence of each time period is extracted to obtain the disturbance wind speed fragment data; According to the disturbance wind speed segment data, the disturbance duration, fluctuation frequency and wind speed change amplitude of each time period are calculated, and the disturbance periodicity, disturbance frequency and disturbance change rate are obtained respectively; The disturbance periodicity, disturbance frequency and change rate are combined to construct the high-frequency disturbance eigenvalue.

7. The method for locating and repairing wind speed data in the sea-land boundary area according to claim 1 is characterized in that: According to the abnormal spatial location, the wind speed observation values ​​marked as effective disturbance sections are extracted from the adjacent sub-areas to obtain the spatial neighborhood wind speed data, including: According to the abnormal spatial location, the geographical scope of the surrounding sub-area is extracted to obtain the surrounding sub-area range data; According to the data of the surrounding sub-areas, the wind speed observation values ​​marked as effective disturbance sections are extracted to form the initial data set; According to the initial data set, the spatial distance between the abnormal spatial position and the wind speed observation point is calculated, and the wind speed observation values ​​within the preset spatial neighborhood radius are screened out to obtain the spatial neighborhood wind speed data.

8. The method for locating and repairing wind speed data in the land-sea boundary area according to claim 1 is characterized in that: The repaired wind speed data is reintegrated into the corresponding partitioned wind speed data to generate the final wind speed dataset, including: Obtain the corresponding spatial coordinate labels and timestamp identifiers in the repaired wind speed data; According to the spatial coordinate label and timestamp identifier, the matching wind speed observation value position is found in the corresponding partition wind speed data, and the replacement operation is performed to obtain the partition wind speed update data; Summarize and integrate the updated wind speed data of each subarea, build a unified data structure, and generate a merged wind speed dataset; According to the merged wind speed dataset, a data status label is added to each wind speed observation record to distinguish valid data from repaired data, and the final wind speed dataset is obtained.

Images