A sea-land wind model-based low and medium altitude wind field prediction system and method

Abstract

The application relates to the field of wind field prediction, and discloses a low-altitude wind field prediction system and method based on a sea-land wind model, which comprises the following steps: obtaining wind field records along a coastal observation belt and taking the fine time dislocation formed between different devices as a synchronous reference; layer by layer, energy distribution in the wind field records which changes with height, temperature difference and humidity is split; the change clues of wind field element groups reflecting fluctuation, deflection and weak rotation trend are limitedly coded, the mutual pulling degree of disturbance in each period, the energy transfer direction and the mutation sensitivity are remolded into structured data segments; unstable interaction in local areas in the wind speed, wind direction and vertical temperature difference relationship is found by using a wind field expression body, and the disturbance transmission relationship between nodes in these areas is extracted; and micro-disturbance events are compared with the long-term deviation distribution of the overall observation records. The application has the advantages of improving the capturing and expression ability of micro-scale wind field disturbance.

Description

Technical Field

[0001] This invention relates to the field of wind field forecasting, specifically to a mid-to-low-level wind field prediction system and method based on a land-sea breeze model. Background Technology

[0002] In approach segment management at coastal airports, dispatch systems typically need to make real-time corrections to aircraft glide trajectories, speed control, and automated guidance thresholds based on mid- and low-altitude wind field prediction data. However, during the initial intrusion phase of sea-land wind circulation, extremely narrow-scale micro-pulsations in wind speed occur at altitudes below 200 meters near the coast. These micro-pulsations often last less than 30 seconds in time and have a spatial scale of less than 800 meters. The data is scattered across lidar echo sequences, second-by-second wind speeds from shore-based meteorological towers, sea surface temperature difference buoy records, and historical samples of similar wind fields. Existing data processing workflows typically perform independent analysis on various data types, lacking the ability to align time-series fragments and reconstruct fine-grained features across devices and modes. This results in these micro-pulsations being treated as noise and filtered out before fusion, causing wind field prediction models to suffer from persistent low-frequency biases during the rapid intrusion phase of sea-land winds. Against this backdrop, even if numerical models can provide trend judgments on the overall structure of sea-land winds, the lack of a complete reconstruction of these micro-scale disturbances in the input data makes it difficult for the models to identify their triggering effect on mid- and low-altitude wind shear. For scenarios relying on fine-grained wind field forecasts, such as short-term wind shear warnings for coastal airports and the establishment of safety buffer zones for near-shore low-altitude logistics routes, this seemingly minor data fragmentation problem has become a key factor limiting prediction accuracy. Therefore, it is essential to design a mid-to-low-altitude wind field prediction system and method based on a sea-land wind model that improves the ability to capture and represent microscale wind field disturbances. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a mid-to-low-altitude wind field prediction system and method based on a land-sea breeze model, which has the advantage of improving the ability to capture and express microscale wind field disturbances, and solves the problems mentioned in the background technology.

[0004] To achieve the aforementioned goal of improving the ability to capture and represent microscale wind field disturbances, this invention provides the following technical solution: a method for predicting mid-to-low-level wind fields based on a land-sea breeze model, comprising the following steps: Wind field records were acquired in the coastal observation zone, and a comparable time frame was established for various records by using the slight time discrepancies between different devices as a synchronization reference. Based on a comparable time frame, the energy distribution in the wind field records that varies with altitude, temperature difference and humidity is broken down layer by layer to identify short-term wind speed jumps and local ruptures of the inversion layer when the sea wind begins to advance toward the land, forming a group of wind field elements that reflect different disturbance scales. The changes in wind field elements reflecting pulsation, deflection, and weak rotation trends are restricted and encoded. The mutual influence of disturbances, energy transfer direction, and abrupt change sensitivity at each time period are reshaped into structured data fragments. A wind field expression body representing small disturbances is constructed through a stepwise dilution and compensation method. By using the wind field representation, we can find local areas where unstable interactions occur in the relationship between wind speed, wind direction, and vertical temperature difference. We can extract the perturbation transmission relationship between nodes in these areas and screen out the micro-perturbation events that cause low-altitude anomalies by detecting the resonance behavior of node thresholds in short intervals. By comparing the micro-perturbation events with the long-term deviation distribution of the overall observation records, the data retrieval order is rearranged according to the deviation trend, and the reference weights of different segments are adjusted.

[0005] The preferred process for establishing comparable timeframes for various types of records is as follows: Wind speed, wind direction, and temperature and humidity parameters are obtained through nearshore wind towers and sea surface temperature difference buoys. The subtle offsets caused by differences in the internal time base of the equipment are detected, and the offsets are used as a synchronization reference to align the observed data along the time axis. Based on the time alignment results, and combined with the sampling performance, response delay, and spatial distribution of the observation equipment, a comparable time frame with a unified reference point is constructed.

[0006] Preferably, the process of decomposing the energy distribution in the wind field record as it varies with altitude, temperature difference, and humidity layer by layer is as follows: Within a comparable timeframe, wind field records are segmented according to different observation heights, and the vertical transfer relationship of wind energy is analyzed in conjunction with the evolution trend of land-sea temperature difference over time. By identifying energy attenuation patterns from high to low altitudes, key energy segments reflecting inter-layer ventilation, humidity traction effects, and changes in land-sea temperature gradients are extracted, and these extracted energy segments are used to construct a layer-by-layer perturbation archive.

[0007] Preferably, the process of forming a group of wind field elements that reflects different disturbance scales is as follows: Retrieve key segments from the layer-by-layer disturbance archives that reflect phenomena such as airflow acceleration, airflow compression, or temperature reflection. Use short time windows to capture signs of sudden increases in wind speed or rapid changes in direction, and determine the correspondence between the signs and the stages of sea breeze propulsion; The analysis of local inversions, sudden drops, or rapid rises in the temperature profile serves as a basis for determining the rupture of the inversion layer. The abnormal signs are integrated into wind field element groups according to the scope of the disturbance, the response status of the upper and lower layers, and the intensity of energy fluctuations.

[0008] Preferably, the process of reshaping the mutual influence of disturbances, the direction of energy transfer, and the sensitivity to mutations at different time periods into structured data fragments is as follows: The wind field elements are divided into three types of feature segments according to the airflow motion mode: pulsating, deflecting, and weakly swirling. During the encoding process, a constraint factor is set for the constraint conditions of each type of feature segment. The encoded results are compared and verified to eliminate artifacts caused by equipment noise and short-term environmental oscillations, forming structured data fragments for perturbation correlation calculations.

[0009] Preferably, the process of constructing a wind field representation of subtle perturbations through stepwise dilution and compensation is as follows: The structured data segments are grouped according to the strength of the disturbance from high to low, and information dilution operations are performed on each group. The key response to weak perturbations is recovered from the results after dilution using a local compensation rule. The diluted and compensated fragments were combined according to time reference and observation height to construct a wind field representation that reflects subtle disturbances.

[0010] Preferably, the process of extracting the disturbance propagation relationships between nodes in these regions is as follows: Search for regions in the wind field representation where wind speed changes and wind direction shifts are synchronously enhanced; Points where the vertical temperature difference shrinks or reverses in a short period of time are marked as low-altitude instability factors. By performing correlation analysis on the nodes involved in these areas, the disturbance traction mode, response delay duration and inter-layer influence range between nodes are extracted to form the disturbance transmission relationship.

[0011] Preferably, the process for screening out micro-disturbance events that trigger low-altitude anomalies is as follows: Based on the disturbance propagation relationship, multiple thresholds are set, including the magnitude of sudden wind speed increase, the angle of sudden wind direction deflection, the amount of sudden shrinkage of humidity gradient, and the intensity of temperature difference reversal, to characterize the sensitive response limit of nodes under instantaneous disturbances. In the wind field representation, find the set of nodes that continuously touch multiple thresholds or cross different threshold intervals in a very short time along the disturbance propagation path, and determine whether the set of nodes shows resonance signs of phase convergence and response amplification by combining the associated path. Path consistency verification is performed on the node set showing signs of resonance to confirm the traction direction, interlayer amplification and extension range of the node set in the disturbance transmission relationship. The node combination that continuously amplifies energy along the main transmission path is identified as a micro-disturbance event.

[0012] Preferably, the process of rearranging the data retrieval order and adjusting the reference weights of different segments based on the deviation trend is as follows: By comparing the screened micro-perturbation events with the deviation trajectories formed in long-term observations, the rarity of the deviation trajectories in the large-scale climate background and their impact on stability can be determined. Based on the comparison results, the retrieval order of wind field records is adjusted, the reference weight of high-impact segments is increased, and the weight of segments with strong stability but weak impact is decreased, and the weighted wind field data is output.

[0013] A mid-to-low-level wind field prediction system based on a land-sea breeze model includes: Time calibration module: Constructs a unified reference time frame based on slight time misalignments between multiple devices; Layer-by-layer decomposition module: The generated energy distribution is decomposed layer by layer to form a group of wind field elements that reflect different disturbance scales; Perturbation coding module: It converts pulsation, deflection and micro-spin features into coding fragments, and constructs a wind field representation that reflects small perturbations through dilution and compensation. Anomaly screening module: Identifies local areas in the wind field representation where the relationship between wind speed, wind direction and vertical temperature difference is unstable, and screens out micro-disturbance events based on the threshold resonance performance of short intervals; Weighting adjustment module: Compares micro-perturbation events with long-term deviation trends, rearranges the data retrieval order, and adjusts the reference weights of different segments.

[0014] Compared with existing technologies, this invention provides a mid-to-low-altitude wind field prediction system and method based on a land-sea breeze model, which has the following beneficial effects: This invention establishes a synchronous alignment and comparable time frame for multi-source heterogeneous wind field data from coastal observation zones, enabling direct comparison of measurement data from different devices and providing a foundation for accurate analysis of sea-land wind dynamics. Layer-by-layer decomposition of wind field energy distribution and identification of short-term jumps and inversion rupture characteristics fully reflect minute fluctuations in wind field disturbances across dimensions of height, temperature difference, and humidity, forming quantifiable wind field element sets and enhancing the analytical capabilities for sea breeze intrusion and local wind field evolution. Through restrictive coding and progressive dilution and compensation processing, weak disturbances such as pulsations, deflections, and micro-rotations are transformed into structured data fragments, generating a wind field representation that describes subtle energy transfer and disturbance coupling relationships, providing a high-resolution foundation for micro-disturbance event identification. Utilizing the wind field representation for disturbance propagation and resonance detection at local unstable nodes enables high-precision screening of micro-disturbance events, making the prediction of low-altitude anomalies both spatiotemporally sensitive and quantifiable. By comparing micro-disturbance events with long-term deviation distributions and dynamically adjusting the data retrieval order and segment reference weights based on deviation trends, the influence of key disturbance events in wind field simulation and prediction is enhanced, and the model's responsiveness to extreme events or anomalous wind field changes is strengthened. It can accurately capture wind field disturbance details in the land-sea interaction zone, optimize data utilization efficiency, and improve the accuracy, stability, and real-time performance of mid- and low-altitude wind field prediction. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the method of the present invention; Figure 2 This is a schematic diagram of the structure of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example 1: Please refer to Figure 1 As shown in the figure, a method for predicting mid-to-low-level wind fields based on a land-sea breeze model in an embodiment of the present invention includes the following steps: S1: Obtain wind field records in the coastal observation zone and establish comparable time frames for various records by using the slight time misalignment between different devices as a synchronization reference.

[0018] The process of establishing comparable time frames for various types of records in S1 is as follows: Wind speed, wind direction, and temperature and humidity parameters are obtained through nearshore wind towers and sea surface temperature difference buoys. Basic meteorological elements such as wind speed, wind direction, temperature, humidity, and air pressure are continuously collected by nearshore wind towers, sea surface temperature difference buoys, and auxiliary laser echo arrays deployed in the coastal observation zone. The wind towers use layered ultrasonic anemometers and temperature and humidity sensors to capture the wind direction deflection characteristics at different altitudes, while the temperature difference buoys record sea surface temperature difference changes and water vapor flux using high-frequency sampling. During the collection process, the various devices locally mark the raw measurement values ​​with millisecond-level timestamps and transmit them back to a unified data access terminal through wired or low-latency wireless links, forming a multi-source wind field raw record that includes the temperature and humidity exchange characteristics of the land-sea interface. The system detects subtle offsets caused by differences in the internal time base of the equipment, uses these offsets as a synchronization reference, and aligns the observed data along the time axis. It reads the operating parameters of the internal clock source, including crystal oscillator frequency stability, sampling cycle error, and communication superposition delay, and calculates the logical time difference between each device within the same reference time period. It uses characteristic wind speed transition points, temperature difference inflection points, or air pressure transient points captured simultaneously by the wind tower and temperature difference buoy as natural synchronization markers. It performs cross-correlation matching on the time series of each record, extracts the tiny time offsets between devices within the range of seconds or milliseconds, and uses the obtained offsets as a global synchronization reference value. It then performs linear or piecewise time resampling on the original data to achieve precise alignment of all records in the time dimension. Based on the time alignment results, and considering the sampling performance, response delay, and spatial distribution of the observation equipment, a comparable time frame with a unified reference point is constructed. After time alignment, the effective time resolution of each observation equipment in capturing micro-disturbances in the wind field is calculated according to its sampling frequency, instantaneous response capability, deployment height, and land-sea distribution location. Based on this, a quality evaluation scale describing the comparability of different data sources is established. Under this scale, high-frequency equipment (such as buoys) records are used as a fine-grained master time axis, and low-frequency equipment (such as stratified wind towers) sampling points are mapped to the master time axis through interpolation and delay compensation. Furthermore, a unified reference point (such as the height of a virtual reference layer on the land-sea boundary) is determined by combining the spatial deployment location of the equipment, thus constructing a comparable time frame with a unified benchmark, synchronous correction capability, and spatial normalization characteristics.

[0019] S2: Based on the comparable time frame, the energy distribution in the wind field record is broken down layer by layer as the altitude, temperature difference and humidity change, and the signs of short-term wind speed jumps and local ruptures of the inversion layer when the sea wind begins to advance towards the land are identified, forming a group of wind field elements that reflect different disturbance scales.

[0020] The process of breaking down the energy distribution in the wind field record as it changes with altitude, temperature difference, and humidity in S2 is as follows: Within a comparable timeframe, wind field records are segmented according to different observation heights, and the vertical transfer relationship of wind energy is analyzed in conjunction with the evolution trend of land-sea temperature difference over time. Based on the height-stratified observation records provided by laser echo, atmospheric profile radar, or nearshore wind towers, wind speed, wind direction, and turbulent energy indicators within the same time period are segmented by height. The observation height is divided into stepped intervals such as 10m, 30m, 50m, 100m, and higher. For each interval, wind speed modulus, wind direction angle sequence, turbulent kinetic energy, and temperature and humidity linkage indicators are extracted. The evolution sequence of land-sea temperature difference ΔT (sea surface-land surface) over time obtained from sea surface temperature difference buoys is synchronously added to the analysis. By calculating the growth or convergence rate of ΔT, the vertical transfer intensity of thermally driven updrafts or downdrafts is determined. The energy intensity of different height layers is correlated with the time derivative d(ΔT) / dt of ΔT to identify whether there is a lag, advance, or synchronous relationship in the energy coupling degree between layers, thereby establishing a dynamic mapping model of wind energy transfer in the vertical direction. By identifying energy attenuation patterns from high to low altitudes, key energy segments reflecting inter-layer ventilation, humidity traction effects, and changes in land-sea temperature gradients are extracted. These segments are then used to construct a layer-by-layer perturbation profile. The energy attenuation rate dE / dh with altitude, the coherence coefficient between different layers, the traction strength of humidity gradients (such as specific humidity q and relative humidity RH with altitude) on energy distribution, and the thermal gradient effect of land-sea temperature differences between vertical layers are calculated. Based on these indicators, several key energy segments are automatically extracted, such as energy surge segments that occur when inter-layer ventilation is enhanced, energy traction segments caused by sudden increases in humidity or enhanced sea surface evaporation, and energy attenuation segments caused by the decline of land-sea temperature gradients. Each energy segment is accompanied by its triggering conditions, duration, intensity value, and corresponding altitude range. These segments are arranged from high to low altitude to construct a layer-by-layer perturbation profile, which is used to identify inter-layer change characteristics during sea breeze intrusion, land breeze retreat, or low-level shear formation, enabling precise tracking of the sources and evolution paths of perturbations in the vertical structure of the wind field.

[0021] The process of forming wind field element groups reflecting different disturbance scales in S2 is as follows: The system retrieves key segments reflecting airflow acceleration, airflow compression, or temperature reflection phenomena from the layer-by-layer perturbation archives. It scans the wind speed, wind direction, and temperature profile sequences at each altitude level and identifies key segments of airflow acceleration (e.g., local wind speed increase exceeding 15% of the average), airflow compression (e.g., areas of abrupt changes in horizontal wind speed gradient), and temperature reflection phenomena (e.g., short-term reverse temperature changes with altitude) by setting thresholds. For each segment, it records its start and end times, affected altitude range, energy intensity, and corresponding temperature and humidity conditions to form a preliminary anomaly candidate set. The system uses automatic threshold detection and signal peak analysis algorithms, combined with historical observation data, for correction to ensure the accuracy of key segment identification. By utilizing short time windows to capture signs of sudden increases in wind speed or rapid shifts in direction, and determining the correspondence between these signs and the stage of sea breeze propagation, the layered disturbance archive is divided into several short time windows according to the time series. Within each window, the rate of change of wind speed and the rate of shift of wind direction are monitored. When a sudden increase in local wind speed exceeding 2-3 m / s in a short period of time is detected, or the wind direction shift angle exceeds 20°, this period is marked as a wind field disturbance sign. By comparing it with the sea-land temperature difference and the sea breeze propagation time series of the coastal observation zone, it is determined whether the wind field disturbance sign is consistent with the sea breeze intrusion stage, thereby eliminating transient fluctuations not driven by sea breeze and improving the pertinence and reliability of disturbance identification. The analysis of local temperature inversions, sudden drops, or rapid rises in the temperature profile serves as the basis for determining the rupture of the inversion layer. Temperature gradient information of each layer is extracted from the temperature profile, and differential or derivative calculation methods are used to identify the intervals of local temperature inversions, sudden drops, or rapid rises. These intervals are coupled with changes in wind speed and humidity for analysis. If an abnormal temperature segment is accompanied by an increase in wind speed or a change in direction, it is judged that the inversion layer may be ruptured. The trigger height, duration, amplitude, and corresponding energy change of each abnormal segment are recorded to quantify the intensity and range of local inversion layer rupture, providing quantitative indicators for generating wind field element groups. Anomalies are integrated into wind field element groups based on their disturbance impact range, upper and lower layer response patterns, and energy fluctuation intensity. Events such as sudden increases in wind speed, directional shifts, and temperature reversals are comprehensively coded and correlated based on their spatial impact range (horizontal distance and vertical height), inter-layer response patterns (upper layers influencing middle and lower layers or middle and lower layers reacting to upper layers), and energy fluctuation amplitude. These anomalies are packaged using a multi-dimensional data structure to form wind field element groups. Each element group includes disturbance type, timestamp, impact level, energy intensity, and triggering conditions.

[0022] S3: Restrictive coding is performed on the changes in wind field elements reflecting pulsation, deflection and weak rotation trends. The degree of mutual influence of disturbances, energy transfer direction and abrupt change sensitivity at each time period are reshaped into structured data fragments. A wind field expression body representing small disturbances is constructed through progressive dilution and compensation.

[0023] The process in S3 of reshaping the mutual influence of disturbances, the direction of energy transfer, and the sensitivity to mutations at different time periods into structured data fragments is as follows: The wind field element group is divided into three types of feature segments according to the airflow motion mode: pulsating, deflecting, and weak swirling. During the encoding process, constraint factors are set for the constraints of each type of feature segment. The motion characteristics of the generated wind field element group are analyzed, and the wind speed, wind direction, and temperature and humidity change patterns are mapped to three types of disturbance features: pulsating segment represents a segment in which the wind speed rises and falls rapidly over time and has periodic changes; deflecting segment represents a segment in which the wind direction changes significantly in a short period of time and is accompanied by local energy transfer; weak swirling segment represents a segment in which a weak rotating flow field or local vortex is formed. For each type of feature segment, constraints and constraint factors are set. For example, a threshold is set for pulsating segment to exclude low-amplitude noise fluctuations; the angle change rate is constrained to not exceed a certain extreme value for deflecting segment; and the rotation intensity and duration range are limited for weak swirling segment. The encoded results are compared and verified to eliminate artifacts caused by equipment noise and short-term environmental oscillations, forming structured data fragments for disturbance correlation calculations. The encoded results are compared and verified with multiple reference benchmarks. Using equipment noise models and sensor error characteristics, abnormal signals that may be caused by measurement errors or short-term interferences are identified. Combined with short-term environmental disturbance data (such as radar echo jitter and instantaneous temperature and humidity fluctuations), pseudo-features caused by non-real disturbances are screened out. Through multi-source comparison and statistical verification, the remaining fragments are organized into standardized structured data. Each fragment contains disturbance type, time interval, impact level, and intensity information, thus forming structured data fragments that can be directly used for wind field disturbance correlation calculations.

[0024] The process of constructing a wind field representation of subtle perturbations in S3 through stepwise dilution and compensation is as follows: The structured data segments are grouped according to the strength of the disturbance from high to low, and information dilution is performed on each group. The coded and verified structured data segments are sorted according to the disturbance intensity index (such as wind speed change amplitude, deflection angle or rotation intensity) and divided into three groups: high, medium and low. For each group, information dilution is performed, that is, while retaining key feature information, the influence of secondary fluctuations is reduced by smoothing, downsampling or signal hierarchical filtering to reduce the interference of high frequency noise on micro-disturbance identification. The key responses of weak disturbances are recovered from the results after dilution using local compensation rules. Local compensation is performed on low-intensity or weak disturbance segments to recover their potential key responses. The compensation rules include interpolation estimation using disturbance information from adjacent time steps or upper and lower layer observation heights, amplification correction of short-term wind speed abrupt changes, direction deflections, or temperature reversals weakened by dilution, and reference to historical sea-land wind propagation patterns and short-term resonance characteristics to re-embed key peaks and energy transfer paths of the micro-disturbance signal into the segments. Time series repair algorithms and inter-layer energy correction methods are used to achieve the identifiability and continuity of micro-disturbance features. The diluted and compensated fragments were combined according to time reference and observation altitude to construct a wind field representation reflecting subtle disturbances. The combination process included time axis alignment, multi-level altitude data synchronization, and unified labeling of disturbance type labels to ensure the consistency of the wind field representation in time and space. The representation can simultaneously reflect the distribution and interaction of pulsations, deflections, and weak vortex-like micro-disturbances, providing basic data for low-altitude anomaly detection and disturbance propagation analysis. This was achieved through multi-dimensional array mapping, three-dimensional tensor construction, and joint indexing of time and altitude, ensuring that the representation retains the characteristics of subtle disturbances while facilitating automated analysis by computer algorithms.

[0025] S4: Use the wind field representation to find local areas where unstable interactions occur in the relationship between wind speed, wind direction and vertical temperature difference. Extract the perturbation transmission relationship between nodes in these areas, and screen out the micro-perturbation events that cause low-altitude anomalies by detecting the resonance behavior of node thresholds in short intervals.

[0026] The process of extracting the disturbance propagation relationships between nodes in these regions in S4 is as follows: The system searches for regions in the wind field representation where wind speed changes and wind direction shifts exhibit a synchronously increasing relationship. It scans the wind speed and wind direction change sequences of each node in the wind field representation, calculates the correlation coefficient between wind speed and wind direction within a local time window, and identifies regions exhibiting a synchronously increasing trend. Synchronous increasing relationship refers to the phenomenon where wind speed increases or decreases sharply while wind direction shifts within a short time scale. The system uses sliding correlation analysis, multi-scale time series alignment, and dynamic threshold detection methods to mark highly correlated regions. Points where the vertical temperature difference shrinks or reverses in a short period of time are marked as low-altitude instability factors; short-term variation analysis of temperature difference data at each observation height in the wind field expression body is performed, and nodes that shrink or reverse in a specific time window are identified by calculating the gradient change rate and the local reversal amplitude. These points are marked as low-altitude instability factors, which means that these locations may be the initial source of inversion layer rupture or local thermal disturbance. Correlation analysis is performed on the nodes involved in these areas to extract the disturbance traction mode, response lag duration, and inter-layer influence range between nodes, forming a disturbance transmission relationship. The correlation analysis includes calculating the time lag, response amplitude, and cross-layer influence range of disturbances between nodes to determine the traction mode (such as downwind propagation, countercurrent conduction, or inter-layer coupling) and response lag characteristics. Through comprehensive evaluation of different disturbance paths, a data structure describing the disturbance transmission relationship between nodes is generated for micro-disturbance event identification and anomalous wind field prediction.

[0027] The process by which the micro-disturbance events that caused the low-altitude anomalies were screened out in S4 is as follows: Based on the disturbance propagation relationship, multiple thresholds are set, including the magnitude of sudden wind speed increase, the angle of sudden wind direction velocities, the amount of sudden shrinkage of humidity gradient, and the intensity of temperature reversal, to characterize the sensitivity response limits of nodes under instantaneous disturbances. Multiple thresholds are set for each observation node in the low-altitude wind field to quantify the sensitivity response limits of nodes under instantaneous disturbances. Specific thresholds include the magnitude of sudden wind speed increase, the angle of sudden wind direction velocities, the amount of sudden shrinkage of humidity gradient, and the intensity of temperature reversal. Each threshold can be obtained statistically based on historical observation data and high-resolution simulation results. For example, the 99th percentile or extreme value fluctuation range can be used as a reference value. Continuous time-series information of nodes is collected through radar wind measurement, high-frequency wind speed and direction towers, and micro-meteorological stations. The wind speed change, wind direction change rate, humidity gradient change rate, and temperature difference change rate of each node within a short time window are calculated and compared with the preset thresholds. To adapt to seasonal changes or sudden weather conditions, an adaptive threshold adjustment mechanism is introduced so that the thresholds are dynamically updated with the long-term statistical characteristics of the node status, thereby accurately characterizing the sensitivity response limits of nodes under micro-disturbance conditions. In the wind field representation, we search for node sets that continuously touch multiple thresholds or cross different threshold intervals in a very short time along the disturbance propagation path. We then combine the associated paths to determine whether the node sets show signs of resonance with phase convergence and response amplification. We compare the changes in wind speed, wind direction, humidity, and temperature difference of each node at each sampling time with the thresholds to generate a binary trigger matrix. We search for sequences of continuous triggering or rapid threshold crossing in the time dimension and cluster them by combining the spatial proximity relationship or path topology between nodes to form a candidate node set. We use a graph structure to represent the disturbance propagation path and use the node threshold triggering state as a node attribute. We use path search and subgraph matching algorithms to identify high trigger density node combinations along the main propagation direction. For example, if three consecutive layers of nodes all show a sudden increase in wind speed exceeding the threshold within 1 minute, they are identified as a candidate micro-disturbance node set. For node sets exhibiting resonance characteristics, path consistency verification is performed to confirm the traction direction, inter-layer amplification, and extension range of the node set in the disturbance transmission relationship. Node combinations that continuously amplify energy along the main transmission path are identified as micro-disturbance events. Time delay correction and phase alignment are performed on the wind speed, wind direction, and temperature and humidity change sequences of the node set. Correlation coefficients or synchronization indices between nodes are calculated. Simultaneously, frequency domain analysis or wavelet transform is used to detect the amplification trend of response amplitude. If nodes show synchronous acceleration changes accompanied by energy accumulation within the same time window, resonance characteristics are identified. The energy amplification of nodes along the main disturbance transmission path is checked to see if it accumulates in a gradient manner, and whether the traction direction is consistent with the disturbance path is confirmed. The superposition of node responses at vertical levels is analyzed to assess the cross-layer extension capability of the disturbance. Using a three-dimensional wind field reconstruction model, the energy changes of the node set are mapped to a spatial grid. By setting energy growth thresholds and continuity screening rules, node combinations that continuously amplify energy along the main transmission path are identified as micro-disturbance events. The trigger time, path nodes, and amplification amplitude are recorded to provide quantitative basis for low-altitude anomaly prediction or risk warning.

[0028] S5: Compare the micro-perturbation events with the long-term deviation distribution of the overall observation records, rearrange the data retrieval order according to the deviation trend, and adjust the reference weights of different segments.

[0029] The process in S5 of rearranging the data retrieval order and adjusting the reference weights of different segments based on the deviation trend is as follows: By comparing the screened micro-disturbance events with the deviation trajectories formed in long-term observations, the rarity of the deviation trajectories in the large-scale climate background and their impact on stability are determined. The key parameter sequences such as wind speed, wind direction, temperature and humidity at the time of the micro-disturbance event are matched with data in the historical deviation trajectory database in the spatial and temporal dimensions to calculate indicators such as deviation amplitude, rate of change and duration of anomalies. Through statistical analysis, the rarity of the deviation trajectory in the large-scale climate background is assessed, that is, whether it is a rare or extreme event in history. Combining the spatial consistency and temporal continuity of disturbance propagation, the degree of impact of the deviation on the overall stability of the low-level wind field is assessed. Based on the comparison results, the retrieval order of wind field records is adjusted. High-impact segments are given increased reference weights, while segments with strong stability but weak impact are given decreased weights, resulting in weighted wind field data. Highly rare or high-impact segments are prioritized during data retrieval and analysis, with increased reference weights to ensure their dominant role in wind field reconstruction or micro-disturbance simulation. Segments with strong stability but low impact are given decreased reference weights and their retrieval order is delayed to minimize their impact on the overall analysis results. Wind field records are divided into continuous time segments, and an impact score is calculated for each segment. The index order is adjusted according to the score. Weighting coefficients are applied in data fusion or interpolation calculations. For example, segments with deviations exceeding twice the historical standard deviation and lasting longer than 5 minutes have their weight increased to 1.5 times the default value, while segments with fluctuations less than the standard deviation and short durations have their weight reduced to 0.5 times.

[0030] Example 2: As Figure 2 As shown, a mid-to-low-level wind field prediction system based on a land-sea breeze model includes: Time calibration module: Constructs a unified reference time frame based on slight time misalignments between multiple devices; Layer-by-layer decomposition module: The generated energy distribution is decomposed layer by layer to form a group of wind field elements that reflect different disturbance scales; Perturbation coding module: It converts pulsation, deflection and micro-spin features into coding fragments, and constructs a wind field representation that reflects small perturbations through dilution and compensation. Anomaly screening module: Identifies local areas in the wind field representation where the relationship between wind speed, wind direction and vertical temperature difference is unstable, and screens out micro-disturbance events based on the threshold resonance performance of short intervals; Weighting adjustment module: Compares micro-perturbation events with long-term deviation trends, rearranges the data retrieval order, and adjusts the reference weights of different segments.

[0031] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0032] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for predicting mid-to-low-level wind fields based on a land-sea breeze model, characterized in that, Includes the following steps: Wind field records were acquired in the coastal observation zone, and a comparable time frame was established for various records by using the slight time discrepancies between different devices as a synchronization reference. Based on a comparable time frame, the energy distribution in the wind field records that varies with altitude, temperature difference and humidity is broken down layer by layer to identify short-term wind speed jumps and local ruptures of the inversion layer when the sea wind begins to advance toward the land, forming a group of wind field elements that reflect different disturbance scales. The changes in wind field elements reflecting pulsation, deflection, and weak rotation trends are restricted and encoded. The mutual influence of disturbances, energy transfer direction, and abrupt change sensitivity at each time period are reshaped into structured data fragments. A wind field expression body representing small disturbances is constructed through a stepwise dilution and compensation method. By using the wind field representation, we can find local areas where unstable interactions occur in the relationship between wind speed, wind direction, and vertical temperature difference. We can extract the perturbation transmission relationship between nodes in these areas and screen out the micro-perturbation events that cause low-altitude anomalies by detecting the resonance behavior of node thresholds in short intervals. By comparing the micro-perturbation events with the long-term deviation distribution of the overall observation records, the data retrieval order is rearranged according to the deviation trend, and the reference weights of different segments are adjusted.

2. The method for predicting mid-to-low-level wind fields based on a land-sea breeze model according to claim 1, characterized in that, The process of establishing comparable timeframes for various types of records is as follows: Wind speed, wind direction, and temperature and humidity parameters are obtained through nearshore wind towers and sea surface temperature difference buoys. The subtle offsets caused by differences in the internal time base of the equipment are detected, and the offsets are used as a synchronization reference to align the observed data along the time axis. Based on the time alignment results, and combined with the sampling performance, response delay, and spatial distribution of the observation equipment, a comparable time frame with a unified reference point is constructed.

3. The method for predicting mid-to-low-level wind fields based on a land-sea breeze model according to claim 2, characterized in that, The process of breaking down the energy distribution in wind field records as it changes with altitude, temperature difference, and humidity layer by layer is as follows: Within a comparable timeframe, wind field records are segmented according to different observation heights, and the vertical transfer relationship of wind energy is analyzed in conjunction with the evolution trend of land-sea temperature difference over time. By identifying energy attenuation patterns from high to low altitudes, key energy segments reflecting inter-layer ventilation, humidity traction effects, and changes in land-sea temperature gradients are extracted, and these extracted energy segments are used to construct a layer-by-layer perturbation archive.

4. The method for predicting mid-to-low-altitude wind fields based on a land-sea breeze model according to claim 3, characterized in that, The process of forming a group of wind field elements that reflect different disturbance scales is as follows: Retrieve key segments from the layer-by-layer disturbance archives that reflect phenomena such as airflow acceleration, airflow compression, or temperature reflection. Use short time windows to capture signs of sudden increases in wind speed or rapid changes in direction, and determine the correspondence between the signs and the stages of sea breeze propulsion; The analysis of local inversions, sudden drops, or rapid rises in the temperature profile serves as a basis for determining the rupture of the inversion layer. The abnormal signs are integrated into wind field element groups according to the scope of the disturbance, the response status of the upper and lower layers, and the intensity of energy fluctuations.

5. The method for predicting mid-to-low-level wind fields based on a land-sea breeze model according to claim 4, characterized in that, The process of reshaping the mutual influence of disturbances, the direction of energy transfer, and the sensitivity to mutations at different time periods into structured data fragments is as follows: The wind field elements are divided into three types of feature segments according to the airflow motion mode: pulsating, deflecting, and weakly swirling. During the encoding process, a constraint factor is set for the constraint conditions of each type of feature segment. The encoded results are compared and verified to eliminate artifacts caused by equipment noise and short-term environmental oscillations, forming structured data fragments for perturbation correlation calculations.

6. The method for predicting mid-to-low-level wind fields based on a land-sea breeze model according to claim 5, characterized in that, The process of constructing a wind field representation of subtle perturbations through stepwise dilution and compensation is as follows: The structured data segments are grouped according to the strength of the disturbance from high to low, and information dilution operations are performed on each group. The key response to weak perturbations is recovered from the results after dilution using a local compensation rule. The diluted and compensated fragments were combined according to time reference and observation height to construct a wind field representation that reflects subtle disturbances.

7. The method for predicting mid-to-low-level wind fields based on a land-sea breeze model according to claim 6, characterized in that, The process of extracting the disturbance propagation relationships between nodes in these regions is as follows: Search for regions in the wind field representation where wind speed changes and wind direction shifts are synchronously enhanced; Points where the vertical temperature difference shrinks or reverses in a short period of time are marked as low-altitude instability factors. By performing correlation analysis on the nodes involved in these areas, the disturbance traction mode, response delay duration and inter-layer influence range between nodes are extracted to form the disturbance transmission relationship.

8. The method for predicting mid-to-low-level wind fields based on a land-sea breeze model according to claim 7, characterized in that, The process of identifying micro-disturbance events that trigger low-altitude anomalies is as follows: Based on the disturbance propagation relationship, multiple thresholds are set, including the magnitude of sudden wind speed increase, the angle of sudden wind direction deflection, the amount of sudden shrinkage of humidity gradient, and the intensity of temperature difference reversal, to characterize the sensitive response limit of nodes under instantaneous disturbances. In the wind field representation, find the set of nodes that continuously touch multiple thresholds or cross different threshold intervals in a very short time along the disturbance propagation path, and determine whether the set of nodes shows resonance signs of phase convergence and response amplification by combining the associated path. Path consistency verification is performed on the node set showing signs of resonance to confirm the traction direction, interlayer amplification and extension range of the node set in the disturbance transmission relationship. The node combination that continuously amplifies energy along the main transmission path is identified as a micro-disturbance event.

9. A method for predicting mid-to-low-level wind fields based on a land-sea breeze model according to claim 8, characterized in that, The process of rearranging the data retrieval order and adjusting the reference weights of different segments based on the deviation trend is as follows: By comparing the screened micro-perturbation events with the deviation trajectories formed in long-term observations, the rarity of the deviation trajectories in the large-scale climate background and their impact on stability can be determined. Based on the comparison results, the retrieval order of wind field records is adjusted, the reference weight of high-impact segments is increased, and the weight of segments with strong stability but weak impact is decreased, and the weighted wind field data is output.

10. A mid-to-low-altitude wind field prediction system based on a land-sea breeze model, applied to the method described in any one of claims 1-9, characterized in that, include: Time calibration module: Constructs a unified reference time frame based on slight time misalignments between multiple devices; Layer-by-layer decomposition module: The generated energy distribution is decomposed layer by layer to form a group of wind field elements that reflect different disturbance scales; Perturbation coding module: It converts pulsation, deflection and micro-spin features into coding fragments, and constructs a wind field representation that reflects small perturbations through dilution and compensation. Anomaly screening module: Identifies local areas in the wind field representation where the relationship between wind speed, wind direction and vertical temperature difference is unstable, and screens out micro-disturbance events based on the threshold resonance performance of short intervals; Weighting adjustment module: Compares micro-perturbation events with long-term deviation trends, rearranges the data retrieval order, and adjusts the reference weights of different segments.

Images