Weather radar networked wind field self-adaptive correction method and system

Abstract

The application provides a weather radar network wind field self-adaptive correction method and system, the method comprises the following steps: according to the radar coverage condition of the multi-radar network in the target area, adopting the local least square method or the three-dimensional variation method based on multi-radar observation to carry out wind field inversion, and obtaining a first wind field; according to the second wind field obtained by the observation equipment in the target area, the first wind field is fused and corrected; based on the digital elevation model data, the terrain forced correction is carried out on the fused and corrected wind field, and the final wind field is obtained. The application can obtain more fine and more physically consistent real-time wind field under the condition of complex terrain and strong convection, and provide more reliable data support for the monitoring, prediction and mechanism research of disastrous weather.

Description

Weather radar network wind field adaptive correction method and system Technical Field

[0001] This invention relates to the field of meteorological forecasting technology, and in particular to a weather radar network wind field adaptive correction method and system. Background Technology

[0002] Wind is a key factor influencing weather evolution and forecast accuracy. Its changes directly determine the transport of water vapor and energy, and also largely control the triggering, development, and dissipation of convective systems. Low-level wind shear can regulate the organization, morphology, and intensity of convective clouds; mid-level jet streams are often closely related to the formation of heavy precipitation belts and squall lines; while upper-level divergence affects the overall atmospheric circulation pattern and the movement path of weather systems. Therefore, accurately obtaining atmospheric wind field distributions at different scales and altitudes is fundamental to revealing the dynamic mechanisms of weather systems and improving numerical weather prediction capabilities.

[0003] However, conventional ground-based observations and radiosondes have significant limitations in spatial coverage and temporal resolution, making it difficult to meet the real-time monitoring needs of small- and medium-scale weather processes, especially severe convection and typhoons. In contrast, weather radar, with its advantages of higher spatiotemporal resolution and continuous observation capabilities, has gradually become an important means of inverting atmospheric wind fields. By inverting the radial velocity of radar, the wind field structure can be reconstructed in three-dimensional space. Especially under the condition of multiple radar networks, the joint inversion of radial velocities in different azimuths can overcome the uncertainties caused by the observation geometry constraints of a single radar, thus more comprehensively characterizing the organization and evolution of atmospheric airflow. The three-dimensional wind field obtained by inversion not only helps to reveal dynamic processes such as low-level jets, shear lines, convergence zones, vortices, and updrafts, but also provides key data for numerical model verification, convective system mechanism research, and precipitation microphysical process analysis. In disaster prevention and mitigation, refined wind field information has irreplaceable value for short-term early warning of severe convection, typhoon monitoring and track forecasting, and aviation and marine meteorological services. With the widespread application of dual-polarization radar and phased array radar, as well as the integrated development of multi-source observation data such as wind profiler radar and laser wind measurement radar, weather radar wind field inversion technology has become increasingly important.

[0004] In existing research, multi-radar network wind field inversion has become the main method for obtaining the dynamic characteristics of small- and medium-scale weather systems. By jointly utilizing radial velocities in different azimuths, the atmospheric flow field can be reconstructed relatively completely in three-dimensional space, demonstrating significant advantages in monitoring typical weather processes such as convective systems, typhoons, and shear lines. To improve inversion accuracy, researchers typically employ velocity defuzzification, multi-source observation constraints, and variational assimilation methods for correction, mitigating the effects of unfavorable observation geometry, clutter interference, and missing data. These correction strategies have improved the physical consistency and spatiotemporal continuity of the inverted wind field to some extent, but significant shortcomings still exist.

[0005] On the one hand, most existing methods fail to explicitly consider the forcing effect of complex terrain on near-surface wind fields, leading to significant biases in inversion results in mountainous and coastal areas. On the other hand, commonly used parameter factors in correction methods are usually set to fixed values, lacking adaptability to differences in weather processes and local environmental changes. Furthermore, wind field corrections are mostly based on large-scale smoothing constraints, failing to adequately characterize key dynamic features such as convergence, divergence, and local vortices at the medium and small scales, limiting their application in rapidly evolving weather processes such as severe convection and typhoons. Therefore, introducing a more adaptive correction mechanism into the networked inversion framework, comprehensively considering both terrain forcing and small-scale dynamic processes, is a key direction for promoting the development of high-precision weather radar wind field inversion. Summary of the Invention

[0006] To address the problems existing in the prior art, this invention provides a weather radar network wind field adaptive correction method and system, which integrates terrain forcing and multi-source observation data constraints for adaptive correction.

[0007] This invention provides a weather radar network wind field adaptive correction method, comprising:

[0008] Based on the radar coverage conditions of the multi-radar network in the target area, the wind field is inverted using the local least squares method or the three-dimensional variational method based on multi-radar observations to obtain the first wind field.

[0009] The first wind field is fused and corrected based on the second wind field obtained from the observation equipment within the target area;

[0010] The final wind field is obtained by performing terrain-forced correction on the fused and corrected wind field based on digital elevation model data.

[0011] According to the present invention, a weather radar network wind field adaptive correction method is provided, wherein the radar coverage condition is that at least three or more radars form a reasonable intersection geometry within the target area. Based on the radar coverage condition of the multi-radar network within the target area, a wind field inversion is performed using a local least squares method or a multi-radar three-dimensional variational method to obtain a first wind field, including:

[0012] When the radar coverage conditions are met, the first wind field is obtained by wind field inversion using a three-dimensional variational method based on multi-radar observations.

[0013] When the radar coverage conditions are not met, the local least squares method is used to perform wind field inversion to obtain the first wind field.

[0014] According to the present invention, a weather radar network wind field adaptive correction method is provided, which uses a three-dimensional variational method based on multi-radar observations to invert the wind field to obtain the first wind field, including:

[0015] The predicted velocity of each radar station is determined based on the unit direction vector from each radar station to the observation grid point and the wind vector component of that grid point.

[0016] A cost function is constructed based on the predicted velocity and the observed radial velocity at each radar station.

[0017] Solve for the wind vector components of the grid points to minimize the cost function.

[0018] According to the adaptive wind field correction method for weather radar networks provided by the present invention, the predicted velocity of each radar station is determined by the following formula based on the unit direction vector from each radar station to the observation grid point and the wind vector component of that grid point:

[0019]

[0020] in, Let i be the predicted velocity of the i-th radar station. It is a three-dimensional wind vector. The wind vector component at this grid point. It is the unit direction vector from the i-th radar station to the observation grid point. , and yes The unit direction vectors in the x, y, and z directions;

[0021] The cost function is constructed using the following formula, based on the predicted velocity and the observed radial velocity at each radar station:

[0022]

[0023] in, It is a cost function. Here is the divergence constraint coefficient. To smooth the constraint coefficients, For the first Radial velocity observed by radar The observation error covariance matrix, It is a vector differential operator, representing the spatial differential operation on a vector field V.

[0024] According to the adaptive wind field correction method for weather radar networks provided by the present invention, in areas covered by only two radars, the first wind field is obtained by wind field inversion using the local least squares method with the following formula:

[0025] ;

[0026] in, , and It is the first The horizontal unit line-of-sight vector pointing from the radar station location to the grid point. The unit direction vectors in the x and y directions. , and These are the radial velocities observed by the two radars respectively. The horizontal wind vectors obtained from each grid point are used as the first wind field.

[0027] According to a weather radar network wind field adaptive correction method provided by the present invention, the first wind field is fused and corrected based on a second wind field obtained by observation equipment within the target area, including:

[0028] Centered on the location of the observation equipment, the surrounding first wind field, constructed based on a multi-radar network, is divided into a first-level correction zone and a second-level correction zone;

[0029] The wind field at the grid points within the first-level correction area is obtained using the observation equipment as a second wind field.

[0030] The size and correction intensity of the secondary correction zone are adjusted based on the wind field dynamics characteristics of the first wind field constructed by the multi-radar network.

[0031] Based on the adjusted size and correction intensity of the secondary correction zone, the wind field of the grid points in the secondary correction zone is the wind field resulting from the fusion of the first and second wind fields.

[0032] According to a weather radar network wind field adaptive correction method provided by the present invention, the size and correction intensity of the secondary correction area are adjusted based on the wind field dynamic characteristics of the first wind field constructed by the multi-radar network using the following formula:

[0033] ;

[0034] ;

[0035] ;

[0036] in, and The standard deviations of divergence and vorticity are given. For divergence, vorticity, For the correction strength of the secondary correction area, This is the inner boundary of the second-level correction area. This is the outer boundary of the secondary correction area. For the product grid resolution of the first wind field, To correct the maximum distance;

[0037] Based on the adjusted size and correction intensity of the secondary correction zone, the wind field at the grid points of the secondary correction zone is determined using the fused wind field of the first and second wind fields according to the following formula:

[0038] ;

[0039] ;

[0040]

[0041] in, and These are the u and v components of the wind field after correction in the second-order correction region, respectively. and Let u and v be the components of the second wind field. and Let u and v be the components of the first wind field; d is the distance weight, g is the Gaussian function, and d is the distance from the current target point to the observation device.

[0042] According to the present invention, a weather radar network wind field adaptive correction method is provided, which performs terrain-forced correction on the fused and corrected wind field based on digital elevation model data to obtain the final wind field, including:

[0043] Calculate the slope based on the terrain height in the digital elevation model data;

[0044] The slope weight is determined based on the slope, and the clearance attenuation weight is determined based on the terrain height, the height of the detection target, and the attenuation scale height.

[0045] The total weight coefficient is determined based on the slope weight and the net clearance attenuation weight.

[0046] The final wind field is obtained by performing terrain-forced correction on the fused and corrected wind field based on the total weight coefficient.

[0047] According to the adaptive wind field correction method for weather radar networks provided by the present invention, the final wind field is obtained by performing terrain-forced correction on the fused and corrected wind field based on the total weight coefficient using the following formula:

[0048]

[0049] in, For the final wind field, To integrate the corrected wind field, The slope.

[0050] This invention also provides a weather radar network wind field adaptive correction system, characterized in that it includes:

[0051] The inversion module is used to perform wind field inversion based on the radar coverage conditions of the multi-radar network in the target area, using the local least squares method or the three-dimensional variational method based on multi-radar observations, to obtain the first wind field.

[0052] The first correction module is used to perform fusion correction on the first wind field based on the second wind field obtained by the observation equipment in the target area;

[0053] The second correction module is used to perform terrain-forced correction on the fused and corrected wind field based on digital elevation model data to obtain the final wind field.

[0054] This invention provides an adaptive correction method and system for wind fields in a weather radar network. Addressing the shortcomings of existing multi-radar network wind field inversion correction methods, it proposes an adaptive correction scheme that integrates topographic forcing and constraints from multi-source observation data. The core idea is to introduce high-resolution digital elevation model information into the inversion framework to perform topographically sensitive corrections on the near-surface wind field, thereby more realistically reflecting the dynamic effects of complex underlying surfaces in mountains, valleys, and coastal areas. Simultaneously, various weights and parameter factors are no longer set with fixed values ​​during the correction process, but are automatically adjusted based on weather processes and the spatiotemporal consistency of multi-source data (such as wind profiler radar and laser wind radar), achieving dynamic adaptability to different weather processes. Most importantly, this invention incorporates a mechanism for identifying and preserving key dynamic features such as small-scale convergence, divergence, and local vortices into the correction algorithm, avoiding the detail weakening problem caused by traditional smoothing constraints. Through this method, more refined and physically consistent real-time wind fields can be obtained under complex terrain and strong convection conditions, providing more reliable data support for the monitoring, forecasting, and mechanistic research of severe weather. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0056] Figure 1 is a flowchart illustrating the adaptive correction method for wind field in a weather radar network provided by the present invention.

[0057] Figure 2 is a schematic diagram of target wind field inversion using the multi-radar three-dimensional variational method in the adaptive correction method for wind field in a weather radar network provided by the present invention.

[0058] Figure 3 is a schematic diagram of the multi-radar network wind field (CAPPI) data area (a) and the intermediate layer interpolation (b) of the redundancy height H in the weather radar network wind field adaptive correction method provided by the present invention.

[0059] Figure 4 is a schematic diagram of adaptive weighted fusion of the first and second level correction areas in the adaptive correction method for wind field in weather radar networking provided by the present invention.

[0060] Figure 5 is a schematic diagram of the influence of terrain on wind field in the adaptive correction method for wind field in weather radar networking provided by the present invention.

[0061] Figure 6 is a schematic diagram of the terrain-forced wind field correction method in the weather radar networking wind field adaptive correction method provided by the present invention.

[0062] Figure 7 shows the independent observation verification results of the adaptive fusion corrected wind field and 9 wind profiler stations at 850 hPa in the weather radar networking wind field adaptive correction method provided by the present invention.

[0063] Figure 8 is a schematic diagram of the weather radar network wind field adaptive correction system provided by the present invention. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0065] The following describes a weather radar network wind field adaptive correction method of the present invention with reference to Figure 1, including:

[0066] Step 101: Based on the radar coverage conditions of the multi-radar network in the target area, wind field inversion is performed using the local least squares method or the three-dimensional variational method based on multi-radar observations to obtain the first wind field.

[0067] Step 102: Perform fusion correction based on the second wind field and the first wind field obtained by the observation equipment in the target area;

[0068] Step 103: Perform terrain-forced correction on the fused and corrected wind field based on the digital elevation model data to obtain the final wind field.

[0069] After performing necessary quality control procedures such as velocity deblurring on the weather radar base data, the data is first uniformly transformed from polar coordinates to Cartesian coordinates, and then subjected to standard refraction. First, calculate the beam center altitude and horizontal distance using the Earth approximation:

[0070]

[0071] in, The altitude of the radar antenna. For the Earth's radius, Then, adjust the polar coordinates according to the orientation. Mapping to Cartesian coordinates :

[0072]

[0073]

[0074]

[0075] in, To detect distance, Radar elevation angle, This is the radar azimuth angle.

[0076] To account for the differences in radar coverage conditions in different regions, wind field inversion was performed using either the local least squares method or the multi-radar three-dimensional variational method.

[0077] The wind field constructed by the multi-radar network is called the NDOP wind field (first wind field), and its data are fused and corrected using data from vertical observation equipment such as wind profiler radar and laser wind radar (hereinafter collectively referred to as WPR, second wind field).

[0078] To address the impact of terrain on the wind field, this embodiment introduces DEM (Digital Elevation Model) data to perform terrain-forced correction on the fused wind field, adjusting the near-surface portion of the wind field.

[0079] This embodiment addresses the shortcomings of existing multi-radar network wind field inversion and correction methods by proposing an adaptive correction scheme that integrates topographic forcing and multi-source observation data constraints. The core idea is to introduce high-resolution digital elevation model information into the inversion framework to perform topographically sensitive corrections on the near-surface wind field, thereby more realistically reflecting the dynamic effects of complex underlying surfaces such as mountains, valleys, and coastal areas. Simultaneously, various weights and parameter factors are no longer set with fixed values ​​during the correction process, but are automatically adjusted based on weather processes and the spatiotemporal consistency of multi-source data (such as wind profiler radar and laser wind radar), achieving dynamic adaptability to different weather processes. Most importantly, this invention incorporates a mechanism for identifying and preserving key dynamic features such as small-scale convergence, divergence, and local vortices into the correction algorithm, avoiding the detail weakening problem caused by traditional smoothing constraints. Through this method, more refined and physically consistent real-time wind fields can be obtained under complex terrain and strong convection conditions, providing more reliable data support for the monitoring, forecasting, and mechanistic research of severe weather.

[0080] Based on the above embodiments, the radar coverage condition in this embodiment is that at least three or more radars form a reasonable intersection geometry within the target area. According to the radar coverage condition of the multi-radar network within the target area, wind field inversion is performed using a local least squares method or a multi-radar three-dimensional variational method to obtain the first wind field, including:

[0081] When the radar coverage conditions are met, the first wind field is obtained by wind field inversion using a three-dimensional variational method based on multi-radar observations.

[0082] When the radar coverage conditions are not met, the local least squares method is used to perform wind field inversion to obtain the first wind field.

[0083] When at least three or more radars form a reasonable intersection geometry within the target area (intersection angle between 30 and 150°), and previous research has shown that the optimal spacing between two radars should be 0.55 times the maximum detection range of the radars, and the detection range of S-band weather radar is 230 km, the detection area of ​​each radar is selected within 120 km. If the above conditions are met, the three-dimensional variational (3DVAR) method based on multi-radar observation is adopted. By introducing observation terms and mass conservation constraints into the cost function, a three-dimensional wind field with good spatial continuity is obtained.

[0084] When the above conditions cannot be met, i.e., the number of radars is less than three or the intersection angle is insufficient (intersection angle is less than 30° or greater than 150°), the local least squares method is used for inversion.

[0085] Based on the above embodiments, this embodiment uses a three-dimensional variational method based on multi-radar observations to obtain the first wind field through wind field inversion, including:

[0086] The predicted velocity of each radar station is determined based on the unit direction vector from each radar station to the observation grid point and the wind vector component of that grid point.

[0087] A cost function is constructed based on the predicted velocity and the observed radial velocity of each radar station.

[0088] Solve for the wind vector components of the grid points to minimize the cost function.

[0089] Using radar radial velocity as an observation constraint, and assuming a three-dimensional wind vector... For each radar within the detection area, its predicted velocity is:

[0090]

[0091] in, Let i be the predicted velocity of the i-th radar station. It is a three-dimensional wind vector. The wind vector component at this grid point. It is the unit direction vector from the i-th radar station to the observation grid point. , and yes The unit direction vectors in the x, y, and z directions;

[0092] By incorporating the radial velocities of all radars, a cost function is constructed based on the predicted velocities and observed radial velocities of each radar station using the following formula:

[0093]

[0094] in, It is a cost function. Here is the divergence constraint coefficient. To smooth the constraint coefficients, For the first Radial velocity observed by radar The observation error covariance matrix, It is a vector differential operator, representing the spatial differential operation on a vector field V.

[0095] Solve using the above formula. The wind field of the region was obtained, and the schematic diagram of the target wind field inversion using the multi-radar three-dimensional variational method is shown in Figure 2.

[0096] Based on the above embodiments, this embodiment uses the local least squares method to invert the wind field in an area covered by only two radars to obtain the first wind field, including:

[0097] Let the horizontal wind vector at each grid point within the detection area be... The radial velocities observed by the two radars are respectively: and ;

[0098] For the The radar station is located at... The grid point locations are Then by the first The horizontal unit line-of-sight vector pointing from the radar station location to the grid point is:

[0099]

[0100] in, and Representing grid points and the first Given the horizontal coordinates of the radar station location, the radial velocity is:

[0101]

[0102] in, and yes The unit direction vectors in the x and y directions of the vector. To account for observation errors, the observation equations of the two radars are written in matrix form:

[0103]

[0104] Due to unknown quantities ( This provides two independent observation equations for the two radars, which can be solved using least squares:

[0105]

[0106] in, , , To obtain the horizontal wind vector at each grid point .

[0107] The wind field components are obtained by inversion on a regular Cartesian grid using variational or least squares methods. Following this, as shown in part (a) of Figure 3, horizontal wind field data at multiple altitudes are constructed according to observation needs. A fixed altitude is selected. Due to the target height Wind field components may not necessarily fall on the regular grid layer, requiring further interpolation. The interpolation method involves selecting a data redundancy thickness T and choosing... and Interpolate the two layers of data. This method allows you to obtain data of any specified height. The horizontal wind field distribution is constructed, and then the equal height wind field (CAPPI) product is obtained, as shown in part (b) of Figure 3.

[0108] Based on the above embodiments, this embodiment performs fusion correction on the first wind field according to the second wind field obtained by the observation equipment in the target area, including:

[0109] Centered on the location of the observation equipment, the surrounding first wind field, constructed based on a multi-radar network, is divided into a first-level correction zone and a second-level correction zone;

[0110] The wind field at the grid points within the first-level correction area is obtained using the observation equipment as a second wind field.

[0111] The size and correction intensity of the secondary correction zone are adjusted based on the wind field dynamics characteristics of the first wind field constructed by the multi-radar network.

[0112] Based on the adjusted size and correction intensity of the secondary correction zone, the wind field of the grid points in the secondary correction zone is the wind field resulting from the fusion of the first and second wind fields.

[0113] Since the WPR equipment sites are generally spaced more than 50 km apart, directly using distance-weighted or fixed-range-weighted fusion will result in gaps or excessive overlap in the correction range between adjacent sites. At the same time, it will spread the uncertainty of point observations to more distant areas, weaken small- and medium-scale structures such as frontal zones and wind shear, and even lead to unreasonable convergence or divergence of local wind fields.

[0114] Therefore, this invention adopts an adaptive weighted fusion of WPR wind field data and NDOP wind field data, giving higher weights to WPR data in areas where WPR data is more reliable, and reducing the weights of WPR data in areas where NDOP data is more reliable.

[0115] The adaptive weighted fusion of WPR data employs Gaussian weighting and adaptive weighting methods, using the location of the WPR data as the center to partition and weight the NDOP wind field of the surrounding network radar.

[0116] First, the correction area is divided into a primary correction area and a secondary correction area. The primary correction area is relatively small, and the wind field at the grid points within this area is entirely based on the observation results of WPR data. The secondary correction area is not a fixed area, but is adaptively adjusted based on the wind field dynamic characteristics such as divergence and vorticity calculated by NDOP. By calculating the divergence and vorticity distribution around the target point, areas with significant dynamic structures or strong wind shear are identified.

[0117] When the divergence and vorticity in the observation area are small, the wind field changes gently, and the atmosphere is in a near-homogeneous state, the radius of the secondary correction zone can be appropriately enlarged to improve the correction effect as much as possible based on WPR data. When the divergence or vorticity is significantly enhanced in certain directions, it indicates that there are obvious convergence, divergence, or rotation characteristics in the local wind field. Excessive smoothing correction may weaken these dynamic structures. Therefore, the correction radius should be appropriately reduced in these directions to reduce the influence of WPR data on the networked NDOP wind field and preserve the original small- and medium-scale characteristics.

[0118] The weights of the second-level correction region are expressed as a Gaussian function, with parameters for its inner and outer ranges. and The correction is adaptively adjusted based on the distribution of divergence and vorticity. In this way, the size and intensity of the secondary correction zone can vary with the dynamic characteristics of the wind field, making full use of real-time WPR data without disrupting the balance of the original dynamic structure of the network wind field.

[0119] Based on the above embodiments, this embodiment adjusts the size and correction intensity of the secondary correction zone according to the wind field dynamics characteristics of the first wind field constructed by the multi-radar network, including:

[0120] The vorticity and divergence of the first wind field constructed based on a multi-radar network are calculated. The correction intensity is determined based on the vorticity and divergence of the first wind field. The correction weight formula is as follows:

[0121]

[0122] in, and The standard deviations of divergence and vorticity are used to control the weighted attenuation range. For divergence, vorticity, The correction strength for the secondary correction zone;

[0123] vorticity and divergence are calculated based on the NDOP wind field. Divergence vorticity .in, The wind field is respectively The directional component. Based on the calculation results of vorticity and divergence, the rotation and expansion characteristics of the wind field in the local region are obtained, which serve as the input to the aforementioned weighting function.

[0124] Calculate the first-level correction area. ,in:

[0125]

[0126] Calculate the range of the second-level correction area. ,in:

[0127]

[0128] in, For the product grid resolution of the first wind field, To correct the maximum distance, It forms the outer boundary of the first-level correction area and the inner boundary of the second-level correction area. This is the outer boundary of the secondary correction area;

[0129] Based on the calculated primary and secondary correction zones, as shown in Figure 4, the wind field in the primary correction zone is directly corrected using the second wind field, while the wind field in the secondary correction zone is corrected using distance weighting and correction intensity. The distance weighting... Using the decay function:

[0130]

[0131] Where g is a Gaussian function and d is the distance from the current target point to the observation device;

[0132] The formula for correcting the wind field in the secondary correction zone is:

[0133] ;

[0134] ;

[0135] in, and For the u and v components of the wind field after correction in the second-level correction area, , For the u and v components of the second wind field, , These are the u and v components of the first wind field.

[0136] Based on the above embodiments, this embodiment performs terrain forcing correction on the fused and corrected wind field based on digital elevation model data to obtain the final wind field, including:

[0137] Calculate the slope based on the terrain height in the digital elevation model data;

[0138] The slope weight is determined based on the slope, and the clearance attenuation weight is determined based on the terrain height, the height of the detection target, and the attenuation scale height.

[0139] The total weight coefficient is determined based on the slope weight and the net clearance attenuation weight.

[0140] The final wind field is obtained by performing terrain-forced correction on the fused and corrected wind field based on the total weight coefficient.

[0141] As shown in Figure 5, under complex terrain conditions, the influence of topography on the wind field is significant. Mountains and hills exert strong lifting, bypassing, and blocking effects on airflow. When the incoming wind speed is high and the mountain is relatively steep, the airflow is forced to rise, often leading to convergence and upward motion on the windward side, while sinking, vortices, or wind shadow areas easily appear on the leeward side, resulting in a distinctly asymmetrical structure in the local wind field. When the wind speed is low or the stability is high, the topography may produce a blocking effect, causing the airflow to bypass the mountain, forming a channel effect or acceleration zone, resulting in a strong gradient in the local wind speed distribution. Narrow terrains such as canyons and river valleys constrain airflow like pipes, significantly increasing wind speed and tending to align with the topographic axis, while plateaus or basins often form relatively closed local circulations.

[0142] To address the impact of topography on the wind field, this embodiment introduces DEM (Digital Elevation Model) data to perform topographic-forced correction on the fused wind field, adjusting the near-surface portion of the wind field. As shown in Figure 6, the specific method for topographic-forced wind field correction is as follows:

[0143] According to terrain height Calculate slope and slope Establish slope based on actual terrain and net clearance attenuation The weight functions are as follows:

[0144]

[0145]

[0146] in, To control the normalization constant of the slope factor, To detect the target's altitude, For terrain height, The height represents the attenuation scale.

[0147] It reflects the steepness of the terrain undulations, when The larger the slope, the steeper the mountain slope, and the easier it is for airflow to be blocked and lifted; the corresponding slope weight... An increase indicates a stronger influence of topography on the wind field. On the other hand, this is determined by the clearance height. To calculate the height difference relative to the top of the mountain, when the clearance height The smaller the value, the closer the airflow is to the mountainside, and the more easily it is controlled by the topographic forcing effect, hence the lower the clearance weight. The weighting of slope and clearance increases accordingly; conversely, when the clearance height is large, it indicates that the airflow is much higher than the mountain and is less affected by the terrain, so the weighting tends to decrease. Through the combined effect of slope weighting and clearance weighting, the correction magnitude of the wind field under different terrain conditions can be dynamically adjusted, resulting in enhanced ascent and deflection on the windward slope and a weakening or backflow trend on the leeward slope, making the corrected wind field more consistent with the actual atmospheric flow patterns.

[0148] The total weighting coefficient for terrain correction is obtained by combining the two weighting factors. ;

[0149] The final wind field is obtained by applying terrain-forced correction to the fused and corrected wind field using the following formula based on the total weighting coefficient:

[0150]

[0151] in, For the final wind field, To integrate the corrected wind field, The slope.

[0152] The wind field retrieved by weather radar network often cannot completely cover all detection areas. This is mainly because the radar has a limited detection range, and the signal attenuates severely at long distances, making it difficult to obtain effective observations. At the same time, under clear sky conditions with no precipitation or few targets, the radial velocity signal is significantly weakened or even missing, resulting in incomplete wind field information in some areas. This leads to gaps or discontinuities in the final retrieval results in space.

[0153] The NDOP wind field after WPR data fusion and terrain-forced correction may still have missing wind field data in some areas. In order to ensure the integrity of the wind field, Gaussian weighted interpolation is needed to fill the gaps in the corrected NDOP wind field.

[0154] At each missing measurement point Within the neighborhood, only existing effective wind field grid points are selected. Participate in interpolation, with Gaussian weights related to distance. Perform normalized weighted average (where, (The number of adjacent grids around the missing measurement point) yields: and This method does not change the local characteristics of the corrected area, but only smoothly extends the surrounding information in the missing area in a distance-decreasing manner, thereby obtaining a continuous, complete and consistent NDOP final wind field with the neighborhood without introducing complex constraints.

[0155] In evaluating the effectiveness of wind field fusion correction, to avoid self-verification using observations involved in the assimilation process, this paper adopts an independent observation verification method. The specific method is as follows: all 52 wind profiler radar data are divided into two parts, with 43 used for wind field fusion correction and the remaining 9 used as independent observation datasets, not participating in the fusion process, but only used to verify the accuracy and reliability of the correction results. By comparing the corrected wind field with the independent observations in terms of wind speed, wind direction deviation, and root mean square error, the performance and applicability of the correction method can be objectively evaluated. The verification and evaluation of the wind field data from 9 wind profiler datasets and the adaptive fusion correction wind field data retrieved from the weather radar network at 850 hPa was selected from 20:00 on April 20th to 02:00 on April 21st, 2024. The results are shown in Figure 7. From the wind speed comparison (a), the correlation coefficient reaches 0.94, indicating that the adaptive fusion correction wind field has a high degree of consistency with the actual wind speed. The deviation is -0.24 ms. -1 This indicates that the wind speed corrected by adaptive fusion is slightly lower than the profile observation overall, but the deviation is small; the root mean square error is 1.81 ms. -1 The absolute deviation is 1.50 ms. -1 This reflects that the adaptively fused corrected wind field can reproduce the actual wind speed characteristics well in a large-scale context. The wind direction comparison (b) further shows that the adaptively fused corrected wind field has a high consistency rate with the actual situation, with nearly 99% of samples having wind direction differences controlled within ±20°, and 100% within ±30°. This indicates that the retrieved wind field is basically consistent with the actual observation in direction, especially in capturing the dominant circulation direction well.

[0156] Comprehensive analysis results show that the adaptive fusion correction method based on terrain forcing and multi-source data constraints can significantly improve the quality of wind fields retrieved from radar networks. By introducing constraints from vertical observation data such as wind profiler radar and terrain forcing weights, this method effectively improves the problem of missing data in complex terrain areas and the low-altitude near-surface layer while ensuring the overall continuity of the wind field, and enhances the consistency between the wind field and actual observations. Comparative evaluation shows that the wind field after adaptive fusion correction has a higher correlation and smaller deviation with wind profiler observations in wind speed and wind direction, can reasonably characterize key dynamic structures such as low-level convergence and shear, and is highly consistent with the spatial distribution of heavy precipitation echo areas. This indicates that this method not only improves the spatial integrity and dynamic rationality of the radar network wind field, but also better reveals the true atmospheric flow field characteristics during severe weather processes, providing reliable support for refined monitoring and forecasting.

[0157] The adaptive correction system for wind field in a weather radar network provided by this invention is described below. The adaptive correction system for wind field in a weather radar network described below can be referred to in correspondence with the adaptive correction method for wind field in a weather radar network described above.

[0158] As shown in Figure 8, the system includes an inversion module 1401, a first correction module 1402, and a second correction module 1403; wherein:

[0159] The inversion module 1401 is used to perform wind field inversion based on the radar coverage conditions of the multi-radar network in the target area, using the local least squares method or the three-dimensional variational method based on multi-radar observations, to obtain the first wind field.

[0160] The first correction module 1402 is used to perform fusion correction on the first wind field based on the second wind field obtained by the observation equipment in the target area;

[0161] The second correction module 1403 is used to perform terrain-forced correction on the fused and corrected wind field based on digital elevation model data to obtain the final wind field.

[0162] This embodiment incorporates high-resolution digital elevation model information into the inversion framework to perform topographically sensitive corrections to the near-surface wind field, thereby more realistically reflecting the dynamic effects of complex underlying surfaces in mountains, valleys, and coastal areas. Simultaneously, during the correction process, various weights and parameter factors are no longer set with fixed values ​​but are automatically adjusted based on weather processes and the spatiotemporal consistency of multi-source data (such as wind profiler radar and laser wind radar), achieving dynamic adaptability to different weather processes. Most importantly, this invention incorporates a mechanism for identifying and preserving key dynamic features such as small-scale convergence, divergence, and local vortices into the correction algorithm, avoiding the detail reduction problem caused by traditional smoothing constraints. Through this method, more refined and physically consistent real-time wind fields can be obtained under complex terrain and strong convection conditions, providing more reliable data support for the monitoring, forecasting, and mechanistic research of severe weather.

[0163] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A weather radar network wind field adaptive correction method, characterized in that, include: Based on the radar coverage conditions of the multi-radar network within the target area, wind field inversion is performed using the local least squares method or a three-dimensional variational method based on multi-radar observations to obtain the first wind field; based on the second wind field obtained by the observation equipment within the target area, the first wind field is fused and corrected; based on digital elevation model data, the fused and corrected wind field is subjected to terrain-forced correction to obtain the final wind field; based on the second wind field obtained by the observation equipment within the target area, the first wind field is fused and corrected, including: taking the location of the observation equipment as the center, dividing the surrounding first wind field constructed based on the multi-radar network into a first-level correction area and a second-level correction area; The wind field of the grid points in the first-level correction area is obtained by the second wind field obtained by the observation equipment; the size and correction intensity of the second-level correction area are adjusted according to the wind field dynamic characteristics of the first wind field constructed by the multi-radar network; and the wind field of the grid points in the second-level correction area is obtained by fusing the first wind field and the second wind field according to the adjusted size and correction intensity of the second-level correction area.

2. The weather radar network wind field adaptive correction method according to claim 1, characterized in that, The radar coverage condition is that at least three or more radars form a reasonable intersection geometry within the target area. Based on the radar coverage condition of the multi-radar network within the target area, the wind field is inverted using the local least squares method or the multi-radar three-dimensional variational method to obtain the first wind field. This includes: when the radar coverage condition is met, the wind field is inverted using the three-dimensional variational method based on multi-radar observations to obtain the first wind field; when the radar coverage condition is not met, the wind field is inverted using the local least squares method to obtain the first wind field.

3. The weather radar network wind field adaptive correction method according to claim 2, characterized in that, The first wind field is obtained by using a three-dimensional variational method based on multi-radar observations. This includes: determining the predicted velocity of each radar station based on the unit direction vector from each radar station to the observation grid point and the wind vector component of that grid point; constructing a cost function based on the predicted velocity of each radar station and the radial velocity observed by each radar station; and solving for the wind vector component of the grid point to minimize the cost function.

4. The weather radar network wind field adaptive correction method according to claim 3, characterized in that, The predicted velocity for each radar station is determined using the following formula, based on the unit direction vector from each radar station to the observation grid point and the wind vector component at that grid point: ;in, Let i be the predicted velocity of the i-th radar station. It is a three-dimensional wind vector. The wind vector component at this grid point. It is the unit direction vector from the i-th radar station to the observation grid point. 、 and yes The unit direction vectors in the x, y, and z directions are given; the cost function is constructed using the following formula based on the predicted velocity and the observed radial velocity at each radar station: ;in, It is a cost function. Here is the divergence constraint coefficient. To smooth the constraint coefficients, For the first Radial velocity observed by radar The observation error covariance matrix, It is a vector differential operator, representing the spatial differential operation on a vector field V.

5. The weather radar network wind field adaptive correction method according to claim 2, characterized in that, In an area covered by only two radars, the first wind field is obtained by inverting the wind field using the local least squares method with the following formula: ;in, ， and It is the first The horizontal unit line-of-sight vector pointing from the radar station location to the grid point. The unit direction vectors in the x and y directions. ， and These are the radial velocities observed by the two radars respectively. The horizontal wind vectors obtained from each grid point are used as the first wind field.

6. The weather radar network wind field adaptive correction method according to claim 1, characterized in that, The size and correction intensity of the secondary correction zone are adjusted according to the wind field dynamics characteristics of the first wind field constructed by the multi-radar network using the following formula: ； ； ;in, and The standard deviations of divergence and vorticity are given. For divergence, vorticity, For the correction strength of the secondary correction area, This is the inner boundary of the second-level correction area. This is the outer boundary of the secondary correction area. For the product grid resolution of the first wind field, To correct the maximum distance, the wind field at the grid points of the secondary correction zone is adjusted using the following formula, based on the adjusted size and correction intensity of the secondary correction zone: the wind field is the result of fusing the first and second wind fields. ； ； ;in, and These are the u and v components of the wind field after correction in the second-order correction region, respectively. and Let u and v be the components of the second wind field. and Let u and v be the components of the first wind field; d is the distance weight, g is the Gaussian function, and d is the distance from the current target point to the observation device.

7. The weather radar network wind field adaptive correction method according to claim 1, characterized in that, The final wind field is obtained by performing terrain-forced correction on the fused and corrected wind field based on digital elevation model data, including: calculating the slope based on the terrain height in the digital elevation model data; determining the slope weight based on the slope; determining the airspace attenuation weight based on the terrain height, the height of the detection target, and the attenuation scale height; determining the total weight coefficient based on the slope weight and the airspace attenuation weight; and performing terrain-forced correction on the fused and corrected wind field based on the total weight coefficient to obtain the final wind field.

8. The weather radar network wind field adaptive correction method according to claim 7, characterized in that, The final wind field is obtained by applying terrain-forced correction to the fused and corrected wind field using the following formula based on the total weighting coefficient: ;in, For the final wind field, To integrate the corrected wind field, The slope.

9. A weather radar network wind field adaptive correction system, characterized in that, The adaptive wind field correction method for weather radar network as described in any one of claims 1-8 includes: an inversion module, used to perform wind field inversion using a local least squares method or a three-dimensional variational method based on multi-radar observations according to the radar coverage conditions of the multi-radar network in the target area, to obtain a first wind field; a first correction module, used to perform fusion correction on the first wind field based on a second wind field obtained by observation equipment in the target area; and a second correction module, used to perform terrain-forced correction on the fused and corrected wind field based on digital elevation model data, to obtain a final wind field.

Images