Rapid quality control method suitable for global real-time wind profile data

Abstract

The invention discloses a rapid quality control method suitable for global real-time wind profile data. The algorithm comprises four main parts of characteristic value inspection, threshold value inspection, internal consistency inspection and vertical consistency inspection. Wherein the characteristic value checking comprises three steps of runoff value checking, sudden change checking and wind shear checking. The quality control algorithm is not only suitable for vertical wind profile data obtained by detection or inversion of various instruments such as wind profile radars, weather radars, PILOT wind measurement, radio acoustic detection systems (RASS) and the like, but also covers various elements such as horizontal wind directions, horizontal wind speeds, vertical speeds, signal-to-noise ratios and the like. The algorithm can be deployed at a data decoding storage end, so that the data can be controlled along with quality when arriving, and the operation time efficiency of a single time can reach a millisecond level. Compared with air sounding observation, the algorithm ensures the quality control timeliness, and also maintains the strong discrimination capability of wrong data.

Description

Technical Field

[0001] The present invention relates to the technical field of meteorological data quality control algorithms, and more particularly to a rapid quality control method suitable for global real-time wind profile data. Background Art

[0002] The structural information of vertical wind profiles is of great reference significance for nowcasting and severe weather warnings. Its data can be used to determine the height of the boundary layer, estimate the turbulent energy dissipation rate, etc., and is of great value for scientific understanding and research on three-dimensional dynamic structure, physical mechanism, atmospheric boundary layer, etc. Among meteorological observation equipment, the detection instruments of vertical wind profiles mainly include wind profile radar, weather radar, pilot wind measurement, and sounding. Among them, wind profile radar inverts the scattering signal of electromagnetic waves by atmospheric turbulence into radial wind and synthesizes it into the horizontal wind field of each layer; weather radar calculates the radial velocity by observing the Doppler frequency shift of cloud and rain particles, and inverts it into a vertical wind profile; sounding balloons calculate the wind field by tracking the position of the balloon through the Beidou or L-band radar positioning system. Regardless of the data observed by the instrument, a standard unified and efficient quality control algorithm is required as the core processing link of the application front end.

[0003] In previous studies, the quality control algorithms for wind profiler radar, weather radar, pilot wind measurement and other data were implemented independently. The detection principles of different devices are different, which makes the quality control algorithms of vertical wind profile data different. This leads to the need to use different quality control methods for different data, affecting the efficiency of storage. Summary of the invention

[0004] Therefore, the technical problem to be solved by the present invention is to provide a rapid quality control method for global real-time wind profile data applicable to multiple types of instruments and multiple factors, so as to ensure the timeliness of quality control while maintaining a strong ability to identify erroneous data.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] A rapid quality control method for global real-time wind profile data includes the following steps:

[0007] Step (1), performing a characteristic value check on the elements to be checked, and then preprocessing the height elements and horizontal wind elements that pass the check;

[0008] Step (2), the data that has passed the feature value check and been preprocessed enters the limit value check;

[0009] Step (3), internal consistency check of observation data with threshold value check and detection device code 0, 3, 6;

[0010] Step (4), perform vertical consistency check on the passed data.

[0011] The above-mentioned global real-time wind profile data rapid quality control method is applied to four types of observation data uploaded by the Global Data Collection and Distribution System (CTS), namely, PILOT wind measurement, weather radar, wind profile radar, and radio acoustic detection system. The corresponding detection equipment type codes in the WMO BUFR message are 0, 3, 6, and 8 respectively.

[0012] The quality control object has 21 elements in total, including area station number, longitude, latitude, station height, year, month, day, hour, minute, second, report category, detection equipment type, sampling height, sampling altitude, geopotential height, horizontal wind direction, horizontal wind speed, u wind component, v wind component, vertical wind speed and signal-to-noise ratio;

[0013] The meaning of the quality control code is: 0 means the data is correct, 1 means the data is suspicious, 2 means the data is wrong, 8 means the data is missing or there is no observation data, and 9 means no quality control is performed.

[0014] In the above-mentioned global real-time wind profile data rapid quality control method, in step (1), the characteristic value check is to examine whether a certain observation data is missing or unobserved; if the observation value exists and is not missing, the QC code = 0, and the subsequent check is entered; if the observation value is missing or unobserved, the QC code = 8, and the subsequent check is not participated.

[0015] The above-mentioned global real-time wind profile data rapid quality control method, pre-processing of horizontal wind elements: in the global wind profile data uploaded by the global data collection and distribution system CTS, the reporting elements of the horizontal wind at each station are divided into two forms, namely, horizontal wind direction and horizontal wind speed, u wind component and v wind component. It is necessary to uniformly convert the horizontal wind direction and horizontal wind speed into u and v wind components;

[0016] If the QC codes of the horizontal wind direction Wd and horizontal wind speed Ws are both 8, but the QC codes of the u and v wind components are not 8, it means that the reporting elements are the u wind component and the v wind component. Then, it is necessary to calculate the corresponding horizontal wind direction and horizontal wind speed through the u wind component and the v wind component, and participate in the subsequent related inspections;

[0017] If the QC codes of the u and v wind components are both 8, but the QC codes of the horizontal wind direction and horizontal wind speed are not 8, it means that the reporting elements are the horizontal wind direction and horizontal wind speed. In this case, the corresponding u wind component and v wind component need to be calculated through the horizontal wind direction and horizontal wind speed for subsequent related inspections.

[0018] The specific conversion formula is as follows:

[0019] Calculate the u and v wind components based on the horizontal wind direction and horizontal wind speed:

[0020] u=-Ws×sin (Wd×π / 180.0) (1);

[0021] v=-Ws×cos (Wd×π / 180.0) (2);

[0022] Calculate the horizontal wind direction and horizontal wind speed based on the u and v wind components:

[0023]

[0024] In formula (1) to formula (5), u and v represent the u wind component and v wind component of the horizontal wind, respectively, in m / s; Wd and Ws represent the horizontal wind direction and horizontal wind speed, in degrees and m / s, respectively; α is the angle parameter, in degrees; π is the circumference of a circle, and its value is 3.1415926;

[0025] Height element preprocessing: In the global wind profile data uploaded by the global data collection and distribution system CTS, the observation height is divided into station height, sampling height, sampling altitude and geopotential height. The station height, sampling height and geopotential height need to be unified as sampling altitude elements for inspection;

[0026] If the QC code of the sampling altitude element is 8, the corresponding value is calculated by the following method and used as the sampling altitude; if any of the following conditions are not met, no processing is done on the altitude:

[0027] If the QC codes of the sampling altitude and the station altitude are both 0, the cumulative value of the station altitude and the sampling altitude will be used as the sampling altitude.

[0028] If the QC code of the geopotential height is 0, the geopotential height is converted to the sampling altitude for use;

[0029] The element value obtained according to the above preprocessing method is called the calculated quantity; in the subsequent quality control inspection, if the calculated quantity participates in the inspection and is judged to be suspicious or wrong, the calculated quantity will not be assigned a QC code, and its QC code will be assigned to all observed elements involved in the preprocessing of the calculated quantity.

[0030] In the above-mentioned global real-time wind profile data rapid quality control method, in step (2), the limit value check includes a range check and a climatological limit value check; in the following order: station height, sampling height, sampling altitude / potential height, horizontal wind direction, horizontal wind speed, vertical wind speed, signal-to-noise ratio, the observed value of each element is compared with the limit value in turn; if the observed value is not within the limit value range, the QC code = 2, and it does not participate in the subsequent check; otherwise, the QC code remains unchanged and continues to participate in the subsequent check;

[0031] If the sampling altitude of an observation record is judged to be wrong, the elements with limit values ​​given according to the altitude range will no longer be subject to limit value checks; after the check is completed, the observation data with the detection equipment type code 8 will be directly output and will no longer enter subsequent checks, while other data will continue to participate in subsequent checks.

[0032] In the above-mentioned global real-time wind profile data rapid quality control method, in step (3), the internal consistency check is to examine whether the relationship between the meteorological element records observed at the same time satisfies a certain physical connection;

[0033] If Wd≠0 and Ws=0, then the QC code of Wd and Ws=2;

[0034] If the QC code of Wd = 8 and the QC code of Ws = 0, then the QC code of Ws = 2;

[0035] If the QC code of Wd = 0 and the QC code of Ws = 8, then the QC code of Ws = 2;

[0036] Wd and Ws are horizontal wind direction and horizontal wind speed, respectively;

[0037] If the above conditions are not met, it is deemed to have passed the internal consistency check, the QC code remains unchanged, and proceeds to subsequent inspections.

[0038] In the above-mentioned global real-time wind profile data rapid quality control method, in step (4), the vertical consistency check is to examine whether the changes of meteorological records within a certain vertical range have a specific regularity; the vertical consistency check is divided into three check items, namely, the rigid value check, the mutation check and the wind shear check;

[0039] The rigidity check is to check whether the average variability of horizontal wind within a certain vertical range is less than the minimum value; the vertical consistency check and wind shear check are to check whether the average variability of horizontal wind within a certain vertical range exceeds the maximum value;

[0040] The purpose of mutation check is to identify data with large vertical variability within a certain vertical range;

[0041] Wind shear checks are designed to identify data where shear values ​​between adjacent altitude levels exceed a threshold;

[0042] Before checking, first select a single vertical observation profile and sort the observation records one by one in order of sampling altitude from low to high. If an observation record meets any of the following conditions, the record will not be sorted;

[0043] Sampling altitude QC code = 2;

[0044] The sampling altitude QC code = 8 and does not meet the conditions for calculating the sampling altitude;

[0045] Either QC code of horizontal wind direction or horizontal wind speed is 2 or 1;

[0046] The QC code of any one of the u wind component and the v wind component is 2 or 1;

[0047] The QC codes for horizontal wind direction, horizontal wind speed, u wind component, and v wind component are all 8.

[0048] The above-mentioned global real-time wind profile data rapid quality control method performs rigid value checks on the sorted vertical profiles in turn. The specific algorithm of the rigid value check is:

[0049] During the inspection, start from the i-th layer with the lowest sampling altitude in the single vertical profile, and check in turn whether the observation values ​​of the i+1, i+2, i+3, ... layers above it remain the same as those of the layer to be inspected;

[0050] If the observed values ​​of the horizontal wind direction and horizontal wind speed, or the u wind component and the v wind component, of three or more consecutive altitude layers do not change, and the difference in sampling altitude between adjacent layers is within 500m, then the QC code of the horizontal wind direction and horizontal wind speed, or the u wind component and the v wind component corresponding to all layers with rigid values ​​is 2, otherwise, the QC code remains unchanged;

[0051] Then continue to check upward from the first non-rigid value layer above layer i until the top layer; observation records that fail the rigid value check are directly output and no longer participate in subsequent checks.

[0052] In the above-mentioned global real-time wind profile data rapid quality control method, mutation check is to check whether the change of meteorological records within a certain vertical range exceeds the maximum threshold based on the algorithm of increment check, vertical residual check and comprehensive judgment; mutation check is performed on the observation records of rigid value check;

[0053] (t-1) Before the inspection, first calculate the weighted sliding mean of the u and v wind components of each layer of the profile; select the sorted single vertical observation profile, take the layer i to be detected as the center, select all the height layers within 500m above and below it, and calculate the weighted sliding mean of the u and v wind components respectively according to the weighted sliding average method; the weighted sliding average method is based on the sliding average method, considering the weight value of each data, thereby eliminating the influence of accidental mutation values ​​on the average value; the greater the absolute deviation of a point from its adjacent height range, the lower its weight in the sliding average calculation process;

[0054] During calculation, if layer i is the first layer or the last layer, only the altitude layers within 500m above or below it are selected to calculate the average value; the specific calculation method is as follows:

[0055]

[0056] In the above formula (6), i is the layer to be detected, j is a layer within 500m above and below, and a i is the weighted sliding average of the observations at layer i, n u 、n d are the number of floors within 500m above and below the i-th floor, o j is the observed value of layer j, w j is the weight coefficient of the observation value in layer j;

[0057] Each layer involved in the calculation must meet Conditions to ensure system stability:

[0058]

[0059] In the above formula (7), d j is the observed value x at level j j The absolute deviation from the sliding mean of all observations in the window is calculated as shown in formula (8); if d j = 0, which makes it impossible to calculate w j , then dj=10 -9 ;

[0060]

[0061] If u +n d +1<3, the weighted sliding mean of the layer i to be detected is assigned as missing;

[0062] (t-2) Incremental check: Calculate the mean difference magnitude ZIi of the layer to be detected, which is used to calculate the average vector residual magnitude RESMN in the comprehensive judgment, so as to check whether the vertical change of the point exceeds the threshold; the mean difference magnitude calculation method is as follows:

[0063] ud i =u i -ua i (9);

[0064] vd i =v i -va i (10);

[0065]

[0066] In the above formulas (9) to (12), the subscript i represents the layer to be detected, and u i 、v i u and v wind components respectively; ua i , va i are the weighted sliding means of the u and v wind components respectively; ud i、vd i are the mean difference of u and v wind components respectively; vvdi is the mean difference of horizontal wind; ZIi is the magnitude of the mean difference; XI is the mean difference parameter corresponding to the sampling altitude of this layer. If any one of the observation value and the weighted sliding mean is missing, the corresponding vvd i and ZI i All are assigned as missing;

[0067] (t-3) Vertical residual check: If the mean difference ud of the u and v wind components of the layer i to be detected is i 、vd i If the absolute values ​​are within 10 times of the XI of the corresponding altitude layer, the vertical residual magnitude ZV of the layer to be detected is calculated. i , which is used to calculate the average vector residual magnitude RESMN in the comprehensive judgment, so as to examine whether the vertical change of the point exceeds the threshold; the vertical residual magnitude calculation method is as follows:

[0068] ① If the layer i to be detected is the bottom layer and its vertical distance to the layer i+1 is within 500m, then:

[0069] uv i =ud i -ud i+1 (13);

[0070] vv i =vd i -vd i+1 (14);

[0071] ② If the layer i to be detected is the top layer and its vertical distance from the i-1 layer is within 500m, then:

[0072] uv i =ud i -ud i-1 (15);

[0073] vv i =vd i -vd i-1 (16);

[0074] ③ If the layer i to be detected is the middle layer, and its vertical distance from the i-1 and i+1 layers is within 500m, then:

[0075] uv i =ud i -0.5(ud i-1 +ud i+1 ) (17);

[0076] vv i =vd i -0.5(vdi-1 +vd i+1 ) (18);

[0077] In the above formulas (15) to (18), ud i 、vd i are the mean difference of the u and v wind components of the i-th layer, ud i+1 、vd i+1 are the mean difference of the u and v wind components of the i+1th layer, ud i-1 、vd i-1 are the mean difference of the u and v wind components of the i-1th layer, uv i 、vv i are the vertical residuals of the u and v wind components of the i-th layer respectively, and then the vertical residual of the horizontal wind of the i-th layer to be tested is calculated, and the formula is:

[0078]

[0079] In formula (19), vvv i is the vertical residual of horizontal wind at the i-th layer; in formula (20), ZV i is the vertical residual magnitude; XV is the vertical residual threshold corresponding to the sampling altitude of this layer. If any of the physical quantities involved in the calculation is missing or does not meet any of the conditions ① to ③, the corresponding vvv i and ZV i All are assigned as missing;

[0080] (t-4) Comprehensive judgment:

[0081] ① Calculate the mean vector residual magnitude RESMN;

[0082] If ZI i With ZV i If all measurements are not missing, then RESMN=(ZI i +ZV i ) / 2;

[0083] If ZI i Missing test, but ZV i If there is no missing measurement, RESMN=ZV i ;

[0084] If ZV i Missing test, but ZI i If there is no missing measurement, RESMN=ZI i ;

[0085] ② Calculate the threshold R for judging the QC code:

[0086] If Hi≤10000m, then R=8.0;

[0087] If Hi ≥ 13000m, then R = 3.0;

[0088] If 10000m<Hi<13000m, then:

[0089]

[0090] In formula (21), H i is the sampling altitude of the layer i to be detected, in meters, m;

[0091] ③ Determine the QC code: If RESMN>R is satisfied, the QC code of the u and v wind components of layer i = 2, and the observation record is directly output and no longer involved in subsequent inspections; otherwise, the data is correct and the QC code remains unchanged.

[0092] The above-mentioned rapid quality control method for global real-time wind profile data uses observation records of mutation checks to perform wind shear checks;

[0093] First, starting from the layer i to be detected, all the altitude layers above and below the layer to be detected are numbered in sequence, and the following checks are performed:

[0094] (f-1) If the layer i to be detected is the bottom layer, and the vertical distances between the i+1 layer and the i layer, and between the i+1 layer and the i+2 layer are all within 900m, if any of the following conditions is met, the QC code of the u and v wind components = 2, otherwise, the QC code remains unchanged;

[0095] ①Mu i+1,i >0.03m -1 ·s / m and Mu i+2,i+1 ≤0.03m -1 ·s / m, and Mu i+1,i With Mu i+2,i+1 The signs are opposite;

[0096] ②Mv i+1,i >0.03m -1 ·s / m and Mv i+2,i+1 ≤0.03m -1 ·s / m, and Mv i+1,i With Mv i+2,i+1 The signs are opposite;

[0097] (f-2) If the layer i to be detected is the top layer, and the vertical distances between the i-1 layer and the i-2 layer are within 900 m, if any of the following conditions are met, the QC code of the u and v wind components = 2, otherwise, the QC code remains unchanged;

[0098] ①Mu i,i-1 >0.03m -1 ·s / m and Mu i-1,i-2 ≤0.03m-1 ·s / m, and Mu i,i-1 With Mu i-1,i-2 The signs are opposite;

[0099] ②Mv i,i-1 >0.03m -1 ·s / m and Mv i-1,i-2 ≤0.03m -1 ·s / m, and Mv i,i-1 With Mv i-1,i-2 The signs are opposite;

[0100] (f-3) If the layer i to be detected is the middle layer, and the vertical distances between the i-1 layer and the i layer, and between the i+1 layer and the i layer are all within the range of 900 m, if any of the following conditions are met, the QC code of the u and v wind components = 2, otherwise, the QC code remains unchanged;

[0101] ①Mu i+1,i >0.03m -1 ·s / m and Mu i,i-1 >0.03m -1 ·s / m, and Mu i+1,i With Mu i,i-1 The signs are opposite;

[0102] ②Mv i+1,i >0.03m -1 ·s / m and Mv i,i-1 >0.03m -1 ·s / m, and Mv i+1,i With Mv i,i-1 The signs are opposite;

[0103] In the above conditions, the calculation method of each variable is as follows:

[0104]

[0105] In the above formulas (22) to (29), u n 、v n , Z n are the u and v wind components and sampling altitude of the nth layer, respectively, n is i, i+1, i+2, i-1, i-2; Mu i+1,i 、Mv i+1,i Mu are the shear values ​​of the u and v wind components between the layers to be detected, i and i+1 respectively; i,i-1 、Mv i,i-1 Mu are the shear values ​​of the u and v wind components between the layers to be detected i and i-1 respectively; i+2,i+1 、Mv i+2,i+1 Mu are the shear values ​​of the u and v wind components between the i+1 layer and the i+2 layer respectively; i-1,i-2 、Mv i-1,i-2are the shear values ​​of the u and v wind components between the i-1 and i-2 layers, respectively;

[0106] After the above inspections are completed, all data and the final QC code are output, and quality control is completed.

[0107] The technical solution of the present invention achieves the following beneficial technical effects:

[0108] This essential control algorithm targets the vertical wind profile data uploaded by the Global Data Collection and Distribution System (CTS for short), and combines the quality control of multiple types of observation instrument data (wind profiler radar data, weather radar, PILOT and other types of wind profiler data). It also sets different quantitative parameters of different observation instruments in different algorithms, increases the quality control of observation items such as vertical velocity and signal-to-noise ratio, and can achieve intensive quality control of wind profile data of various types and factors.

[0109] The quality control algorithm of this application sets up steps such as internal consistency check, dead value check and wind shear check for typical erroneous data types of global wind profile data.

[0110] (1) After the characteristic value check, the height and horizontal wind elements are preprocessed; the u and v wind components are calculated using the horizontal wind direction and horizontal wind speed to lay the data foundation for the subsequent mutation check and wind shear check; and various height reporting elements are converted into sampling altitude elements.

[0111] (2) The quality control algorithm of this application targets typical erroneous data types of global wind profile data. It sets three steps in the vertical consistency check: rigid value check, mutation check and wind shear check. It proposes to use the residual method to check each layer of the vertical profile, and adopts the differential sliding average method to establish a reference sequence. Through the statistical analysis of historical data, the residual thresholds of the increment of single observation values ​​at different heights and the vertical direction are given, thereby improving the effect of the vertical consistency check.

[0112] In the rigid value check, the single vertical observation profile observation records need to be sorted according to the sampling altitude, and then the vertical profile layers are checked using the residual method.

[0113] In mutation inspection, the differential sliding average method is used to establish the reference threshold, and the residual threshold from the incremental and vertical directions is given through statistical analysis of historical data, which improves the effect of vertical consistency inspection.

[0114] In the wind shear check, the wind shear values ​​between adjacent altitude layers are examined to see whether they exceed the maximum threshold.

[0115] The added vertical consistency check meets the high timeliness requirements of real-time meteorological data processing services for global wind profile data quality control. This algorithm can realize rapid quality control after real-time data is decoded and before it is stored in the database. While the single-time quality control efficiency can reach milliseconds, the consistency rate of the quality-controlled data with the sounding data can reach more than 98%, maintaining a strong ability to identify erroneous data.

[0116] (3) This quality control algorithm is not only applicable to vertical wind profile data detected or inverted by multiple instruments such as wind profiler radar, weather radar, PILOT wind measurement, and radio acoustic sounding system (RASS), but also covers multiple factors such as horizontal wind direction, horizontal wind speed, vertical speed, and signal-to-noise ratio. It can also realize data quality control as soon as it arrives and be deployed at the data decoding and storage end, with a single time efficiency of up to milliseconds. BRIEF DESCRIPTION OF THE DRAWINGS

[0117] Figure 1 is a flow chart of the quality control method of the present invention; Figure 2 Schematic diagram for the rigidity examination; Figure 3 Schematic diagram of mutation inspection; Figure 4 This is a schematic diagram of wind shear inspection; Figure 5 Schematic diagram for calculating weighted sliding mean; Figure 6 The proportion of horizontal wind direction and horizontal wind speed error data that failed the rigid value check; Figure 7 This is the error data of the 03020 station at 05:00 on April 13, 2020; Figure 8 is the proportion of horizontal wind direction and horizontal wind speed error data that failed the mutation check; Fig. 9 This is the erroneous data of the sudden change in wind speed at station 71994 at 04:30 on December 23, 2019; Fig.10 The proportion of horizontal wind direction and horizontal wind speed error data that failed the wind shear check; Fig.11 This is the wind shear error data that occurred at station 08289 at 12:43 on April 20, 2020; Fig.12 This is the wind shear error data that occurred at station 06776 at 01:00 on January 22, 2020. DETAILED DESCRIPTION

[0118] like Figure 1 The quality control flow chart of this application is shown in FIG. The rapid quality control method of global real-time wind profile data of this application is specifically as follows.

[0119] 1. Quality Control Objects

[0120] The application objects are four types of observation data: PILOT wind measurement, weather radar, wind profile radar, and radio acoustic detection system. The corresponding detection equipment type codes in the WMO BUFR message are 0, 3, 6, and 8 respectively.

[0121] The quality control objects of this scheme include 21 elements, and the specific information is shown in Table 1. Among them, the detection equipment type code and its meaning are shown in Table 2.

[0122] Table 1 Elements involved in the quality control process

[0123]

[0124] Table 2 Detection equipment types and their meanings

[0125] Detection device type code meaning 0 Barometer used with wind measuring equipment (PILOT) 3 Weather Radar 6 Wind profiler radar 8 Radio Acoustic Observation System (RASS)

[0126] 2. Quality Control Code

[0127] The quality control code (QC code for short) is a number that identifies the quality of the data. In the process of quality control of the data, the quality control code is continuously adjusted according to the quality control results of each inspection step. After all steps have been inspected, the quality control code obtained is the final quality identifier of the data. The definition of the quality control code is shown in Table 3.

[0128] Table 3 Quality control codes and their meanings

[0129] Quality Control Code meaning 0 Data is correct 1 Suspicious data 2 Data Error 8 Data missing or no observation data 9 No quality control

[0130] 3. Quality Control Process

[0131] The quality control flow chart of this program is as follows Figure 1 As shown in the figure, the scheme includes four steps, namely, characteristic value check, limit value check, internal consistency check and vertical consistency check. Among them, the vertical consistency check includes three sub-steps, namely, rigid value check, mutation check and wind shear check. Among them, the first three steps are carried out in sequence according to each observation record, and the vertical consistency check is carried out in sequence according to each vertical observation profile. The corresponding relationship between the quality control steps and each element is shown in Table 4. The specific algorithms of each step are introduced below.

[0132] Table 4 Correspondence between quality control steps and various elements

[0133]

[0134] Quality control algorithm at each step:

[0135] 3.1. Eigenvalue Check

[0136] Definition: Eigenvalue check is a check to see whether a certain observation is missing or unobserved.

[0137] Specific algorithm: If the observation value exists and is not missing (see Table 5 for specific parameters), QC code = 0, and proceed to the subsequent inspection; if the observation value is missing or there is no observation value, QC code = 8, and do not participate in the subsequent inspection. The elements involved in the inspection are shown in Appendix 4.

[0138] Table 5 Common characteristic values ​​of meteorological elements

[0139]

[0140] 3.2 Height and horizontal wind preprocessing

[0141] The GTS global real-time wind profile data received by the National Meteorological Information Center is in BUFR format. According to statistics, there are nearly ten types of BUFR templates in the data currently received. For the same observed physical quantity, different reporting centers may upload it with different reporting elements. This situation mainly occurs in the two physical quantities of horizontal wind and height.

[0142] For example, the reporting elements of horizontal wind are divided into two forms, namely horizontal wind direction and horizontal wind speed, u wind component and v wind component. The reporting elements of height are divided into three forms, namely sampling height (height above the measuring station), sampling altitude and geopotential height. The corresponding coding fields are shown in Appendix 6.

[0143] Table 6 Types of elements for reporting height in global real-time wind profile data in BUFR format

[0144] Serial number Feature name unit Reporting Fields 1 Sampling height (height above the measuring station) m 0 07 006 2 Sampling altitude m 0 07 007、0 10 007 3 Geopotential Height gpm 0 07 009

[0145] In the data, multiple types of reporting elements are set with corresponding element codes. In the quality control process, the same physical quantity represents the same physical meaning. Therefore, after the characteristic value check, different reporting elements of the same physical quantity can be pre-processed for subsequent use. The processing principles are as follows:

[0146] (1) Horizontal wind

[0147] In the subsequent quality control inspections, the steps of checking according to horizontal wind direction and horizontal wind speed are limit value inspection and internal consistency inspection; the steps of checking according to u and v wind components are mutation inspection and wind shear inspection; and the rigid value inspection can be carried out according to horizontal wind direction and horizontal wind speed, as well as u and v wind components. In order to meet the requirements of the elements required for different inspection steps, the corresponding element values ​​need to be calculated in advance. The specific conversion formula is as follows:

[0148] Calculate the u and v wind components based on the horizontal wind direction and horizontal wind speed:

[0149] u=-Ws×sin (Wd×π / 180.0) (1);

[0150] v=-Ws×cos (Wd×π / 180.0) (2);

[0151] Calculate the horizontal wind direction and horizontal wind speed based on the u and v wind components:

[0152]

[0153] In formula (1) to formula (5), u and v represent the u wind component and v wind component of the horizontal wind, respectively, in m / s; Wd and Ws represent the horizontal wind direction and horizontal wind speed, in degrees and m / s, respectively; α is an angle parameter, in degrees; π is pi, and its value is 3.1415926.

[0154] The conditions for conversion between different elements are as follows:

[0155] ① If the QC codes of horizontal wind direction and horizontal wind speed are both 8, but the QC codes of u and v wind components are not 8, it means that the reporting elements are u and v wind components. Then calculate the corresponding horizontal wind direction and horizontal wind speed and participate in subsequent related inspections.

[0156] ② If the QC codes of the u and v wind components are both 8, but the QC codes of the horizontal wind direction and horizontal wind speed are not 8, it means that the reporting elements are horizontal wind direction and horizontal wind speed. Then calculate the corresponding u and v wind components and participate in subsequent related inspections.

[0157] (2) Height

[0158] Except for the separate inspection of various altitude reporting elements when checking the station altitude, sampling altitude (altitude above the station) and sampling altitude limit value, when checking other subsequent steps, all altitude reporting elements are unified as sampling altitude elements for inspection. If the QC code of the sampling altitude element is 8, the corresponding value can be calculated by the following method and used as the sampling altitude. If any of the following conditions are not met, no processing will be done on the altitude.

[0159] ① If the QC code of the sampling height (height above the measuring station) and the measuring station height are both 0, the cumulative value of the measuring station height and the sampling height (height above the measuring station) can be used as the sampling altitude. That is:

[0160] Sampling altitude = station altitude + sampling altitude (height above station altitude)

[0161] ② According to the relationship between geopotential height and geometric altitude in the "Conventional High-altitude Meteorological Observation Business Specifications (2010 Edition)", geopotential height can be used approximately as geometric altitude. Therefore, if the QC code of geopotential height is 0, the geopotential height is regarded as the sampling altitude. That is:

[0162] Sampling altitude = geopotential height

[0163] The element value obtained by the above preprocessing method is called the calculated quantity. In the subsequent quality control inspection, if the calculated quantity is involved in the inspection and is judged to be suspicious or wrong, the calculated quantity will not be assigned a QC code, and its QC code will be assigned to all observed elements involved in the preprocessing of the calculated quantity.

[0164] For example, in a certain inspection step, the calculated u and v wind components should be judged as wrong, then the QC code of the horizontal wind direction and horizontal wind speed involved in the calculation of the u and v wind components is assigned as 2.

[0165] 3.3. Limit value check

[0166] Definition: Boundary value check includes range check and climatological limit value check. Range check refers to whether the meteorological record is within its range, and climatological limit value check refers to whether the meteorological record exceeds its climatological limit value.

[0167] Specific algorithm:

[0168] The data that has passed the characteristic value check and preprocessing enters the limit value check. The limit value range of each element is shown in Table 7. During the inspection, the observed value of each element is compared with the limit value in turn according to the element number sequence in Table 7. If the observed value is not within the limit value range, the QC code = 2 and does not participate in the subsequent inspection; otherwise, the QC code remains unchanged and continues to participate in the subsequent inspection.

[0169] If the sampling altitude of an observation record is judged to be wrong, the elements with limit values ​​given according to the altitude range will no longer be subject to limit value check. After the inspection is completed, the observation data with the detection equipment type code of 8 will be directly output and will no longer enter the subsequent inspection, while other data will continue to participate in the subsequent inspection.

[0170] Table 7 Limit value parameter range

[0171]

[0172] 3.4 Internal consistency check

[0173] Definition: Internal consistency check is to examine whether the relationship between the records of meteorological elements observed at the same time satisfies certain physical connections.

[0174] The specific algorithm is:

[0175] (1) If Wd≠0 and Ws=0, then the QC code of Wd and Ws=2;

[0176] (2) If the QC code of Wd = 8 and the QC code of Ws = 0, then the QC code of Ws = 2;

[0177] (3) If the QC code of Wd = 0 and the QC code of Ws = 8, then the QC code of Ws = 2;

[0178] Wd and Ws are horizontal wind direction and horizontal wind speed, respectively;

[0179] If the above conditions are not met, it is deemed to have passed the internal consistency check, the QC code remains unchanged, and proceeds to subsequent inspections.

[0180] 3.5. Vertical consistency check

[0181] Vertical consistency check is to examine whether the changes in meteorological records within a certain vertical range have specific regularities.

[0182] The vertical consistency check is divided into three inspection items, namely, the rigid value check, the mutation check and the wind shear check. Among them, the rigid value check is to check whether the average variability of the horizontal wind within a certain vertical range is less than the minimum value, and the vertical consistency check and wind shear check are to check whether the average variability of the horizontal wind within a certain vertical range exceeds the maximum value. The mutation check is to identify data with large vertical variability within a vertical range of 1000m; the wind shear check is to identify data with shear values ​​exceeding the threshold between adjacent altitude layers. It can be considered that the wind shear check is an enhanced check of the mutation check.

[0183] 3.5.1. Take a single vertical observation profile

[0184] Before checking, first select a single vertical observation profile and sort the observation records one by one in order of sampling altitude from low to high. If an observation record meets any of the following conditions, the record will not be sorted;

[0185] Sampling altitude QC code = 2;

[0186] The sampling altitude QC code = 8 and does not meet the conditions for calculating the sampling altitude;

[0187] Either QC code of horizontal wind direction or horizontal wind speed is 2 or 1;

[0188] The QC code of any one of the u wind component and the v wind component is 2 or 1;

[0189] The QC codes for horizontal wind direction, horizontal wind speed, u wind component, and v wind component are all 8.

[0190] Method for selecting a single vertical observation profile: If the following element values ​​in multiple observation records are the same, they are considered to be the same observation: (1) Observation time, i.e. year, month, day, hour, minute and second; (2) Area station number; (3) Detection equipment type code; (4) Report category.

[0191] 3.5.2. Dead Value Check

[0192] Definition: The rigid value check is also called the minimum variation rate check. It examines the phenomenon that the observed data remains unchanged due to the failure of the detection instrument and other reasons in the transmission and recording process, resulting in untrue data.

[0193] Specific algorithm: The sorted vertical profiles are checked for rigidity in turn. The specific algorithm for rigidity check is as follows: Figure 2 As shown:

[0194] During the inspection, start from the i-th layer with the lowest sampling altitude in the single vertical profile, and check in turn whether the observation values ​​of the i+1, i+2, i+3, ... layers above it remain the same as those of the layer to be inspected;

[0195] If the observed values ​​of the horizontal wind direction and horizontal wind speed, or the u wind component and the v wind component, of three or more consecutive altitude layers do not change, and the difference in sampling altitude between adjacent layers is within 500m, then the QC code of the horizontal wind direction and horizontal wind speed, or the u wind component and the v wind component corresponding to all layers with rigid values ​​is 2, otherwise, the QC code remains unchanged;

[0196] Then continue to check upward from the first non-rigid value layer above layer i until the top layer; observation records that fail the rigid value check are directly output and no longer participate in subsequent checks.

[0197] 3.5.3 Mutation Check

[0198] Definition: Mutation check is a check that examines whether the changes in meteorological records within a certain vertical range exceed the maximum threshold based on algorithms such as incremental check, vertical residual check and comprehensive judgment.

[0199] Specific algorithms, such as Figure 3 As shown:

[0200] Sudden changes are checked through observation records of rigid value checks. When checking, start from the lowest level of the single vertical profile sampling altitude and check upward layer by layer. If the layer to be checked is recorded as i, the altitude layers above the layer to be checked are numbered i+1, i+2, ..., and the altitude layers below the layer to be checked are numbered i-1, i-2, ....

[0201] (t-1) Before checking, calculate the weighted sliding mean of the u and v wind components of each layer of the profile, such as Figure 5As shown in the figure, a single vertical observation profile after sorting is selected, and all height layers within 500m above and below the layer to be detected are selected as the center. The weighted sliding mean of the u and v wind components are calculated respectively according to the weighted sliding average method. The weighted sliding average method is based on the sliding average method, and the weight value of each data is considered to eliminate the influence of accidental mutation values ​​on the average value. The greater the absolute deviation of a point from its adjacent height range, the lower its weight in the sliding average calculation process.

[0202] During calculation, if layer i is the first layer or the last layer, only the altitude layers within 500m above or below it are selected to calculate the average value; the specific calculation method is as follows:

[0203]

[0204] In the above formula (6), i is the layer to be detected, j is a layer within 500m above and below, and a i is the weighted sliding average of the observations at layer i, n u 、n d are the number of floors within 500m above and below the i-th floor, o j is the observed value of layer j, w j is the weight coefficient of the observation value in layer j;

[0205] Each layer involved in the calculation must meet Conditions to ensure system stability:

[0206]

[0207] In the above formula (7), d j is the absolute deviation between the j-th layer observation and the sliding mean of all observations in the window. The calculation method is shown in formula (8). j = 0, which makes it impossible to calculate w j , then dj=10 -9 ;

[0208]

[0209] If u +n d +1<3, the weighted sliding mean of the layer i to be detected is assigned as missing;

[0210] (t-2) Incremental check: Calculate the mean difference magnitude ZIi of the layer to be detected, which is used to calculate the average vector residual magnitude RESMN in the comprehensive judgment, so as to check whether the vertical change of the point exceeds the threshold; the mean difference magnitude calculation method is as follows:

[0211] ud i =u i -uai (9);

[0212] vd i =v i -va i (10);

[0213]

[0214] In the above formulas (9) to (12), the subscript i represents the layer to be detected, and u i 、v i u and v wind components respectively; ua i , va i are the weighted sliding means of the u and v wind components respectively; ud i 、vd i are the mean difference of u and v wind components respectively; vvdi is the mean difference of horizontal wind; ZIi is the magnitude of the mean difference; XI is the mean difference parameter corresponding to the sampling altitude of this layer, as shown in Table 8. If any one of the observation value and the weighted sliding mean is missing, the corresponding vvd i and ZI i All are assigned as missing;

[0215] Table 8 Mean difference threshold

[0216]

[0217]

[0218] (t-3) Vertical residual check: If the absolute values ​​of the mean difference udi and vdi of the u and v wind components of the layer to be detected i are within 10 times of the corresponding altitude layer XI, the vertical residual magnitude ZVi of the layer to be detected is calculated, which is used to calculate the average vector residual magnitude RESMN in the comprehensive judgment, so as to examine whether the vertical change of the point exceeds the threshold; the vertical residual magnitude calculation method is as follows:

[0219] ① If the layer i to be detected is the bottom layer and its vertical distance to the layer i+1 is within 500m, then:

[0220] uv i =ud i -ud i+1 (13);

[0221] vv i =vd i -vd i+1 (14);

[0222] ② If the layer i to be detected is the top layer and its vertical distance from the i-1 layer is within 500m, then:

[0223] uv i =ud i -ud i-1 (15);

[0224] vv i =vd i -vd i-1 (16);

[0225] ③ If the layer i to be detected is the middle layer, and its vertical distance from the i-1 and i+1 layers is within 500m, then:

[0226] uv i =ud i -0.5(ud i-1 +ud i+1 ) (17);

[0227] vv i =vd i -0.5(vd i-1 +vd i+1 ) (18);

[0228] In the above formulas (15) to (18), ud i 、vd i are the mean difference of the u and v wind components of the i-th layer, ud i+1 、vd i+1 are the mean difference of the u and v wind components of the i+1th layer, ud i-1 、vd i-1 are the mean difference of the u and v wind components of the i-1th layer, uv i 、vv i are the vertical residuals of the u and v wind components of the i-th layer respectively, and then the vertical residual of the horizontal wind of the i-th layer to be tested is calculated, and the formula is:

[0229]

[0230] In formula (19), vvv i is the vertical residual of horizontal wind at the i-th layer; in formula (20), ZV i is the vertical residual magnitude; XV is the vertical residual threshold corresponding to the sampling altitude of this layer, and the specific values ​​are shown in Table 9; if any of the physical quantities involved in the calculation is missing or does not meet any of the conditions ① to ③, the corresponding vvv i and ZV i All are assigned as missing.

[0231] Table 9 Vertical residual threshold

[0232]

[0233]

[0234] (t-4) Comprehensive judgment:

[0235] ① Calculate the mean vector residual magnitude RESMN;

[0236] If ZI i With ZV i If all measurements are not missing, then RESMN=(ZI i +ZV i ) / 2; if ZI i Missing test, but ZV i If there is no missing measurement, RESMN=ZV i ;if ZV i Missing test, but ZI i If there is no missing measurement, RESMN=ZI i ;

[0237] ② Calculate the threshold R for judging the QC code:

[0238] If Hi≤10000m, then R=8.0;

[0239] If Hi ≥ 13000m, then R = 3.0;

[0240] If 10000m<Hi<13000m, then:

[0241]

[0242] In formula (21), H i is the sampling altitude of the layer i to be detected, in meters, m;

[0243] ③ Determine the QC code: If RESMN>R is satisfied, the QC code of the u and v wind components of layer i = 2, and the observation record is directly output and no longer involved in subsequent inspections; otherwise, the data is correct and the QC code remains unchanged.

[0244] 3.5.4 Wind shear check

[0245] Definition: A wind shear check is a check to see if the wind shear value between adjacent altitude levels exceeds the maximum threshold.

[0246] Specific algorithm: Wind shear check is performed through observation records of mutation check; e.g. Figure 4 As shown:

[0247] First, starting from the layer i to be detected, all the altitude layers above and below the layer to be detected are numbered in sequence, and the following checks are performed:

[0248] (f-1) If the layer i to be detected is the bottom layer, and the vertical distances between the i+1 layer and the i layer, and between the i+1 layer and the i+2 layer are all within 900m, if any of the following conditions is met, the QC code of the u and v wind components = 2, otherwise, the QC code remains unchanged;

[0249] ①Mu i+1,i >0.03m -1 ·s / m and Mu i+2,i+1 ≤0.03m -1 ·s / m, and Mu i+1,i With Mu i+2,i+1 The signs are opposite;

[0250] ②Mv i+1,i >0.03m -1 ·s / m and Mv i+2,i+1 ≤0.03m -1 ·s / m, and Mv i+1,i With Mv i+2,i+1 The signs are opposite;

[0251] (f-2) If the layer i to be detected is the top layer, and the vertical distances between the i-1 layer and the i-2 layer are within 900 m, if any of the following conditions are met, the QC code of the u and v wind components = 2, otherwise, the QC code remains unchanged;

[0252] ①Mu i,i-1 >0.03m -1 ·s / m and Mu i-1,i-2 ≤0.03m -1 ·s / m, and Mu i,i-1 With Mu i-1,i-2 The signs are opposite;

[0253] ②Mv i,i-1 >0.03m -1 ·s / m and Mv i-1,i-2 ≤0.03m -1 ·s / m, and Mv i,i-1 With Mv i-1,i-2 The signs are opposite;

[0254] (f-3) If the layer i to be detected is the middle layer, and the vertical distances between the i-1 layer and the i layer, and between the i+1 layer and the i layer are all within the range of 900 m, if any of the following conditions are met, the QC code of the u and v wind components = 2, otherwise, the QC code remains unchanged;

[0255] ①Mu i+1,i >0.03m -1 ·s / m and Mu i,i-1 >0.03m -1 ·s / m, and Mu i+1,i With Mui,i-1 The signs are opposite;

[0256] ②Mv i+1,i >0.03m -1 ·s / m and Mv i,i-1 >0.03m -1 ·s / m, and Mv i+1,i With Mv i,i-1 The signs are opposite;

[0257] In the above conditions, the calculation method of each variable is as follows:

[0258]

[0259]

[0260] In the above formulas (22) to (29), u n 、v n , Z n are the u and v wind components and sampling altitude of the nth layer, respectively, n is i, i+1, i+2, i-1, i-2; Mu i+1,i 、Mv i+1,i Mu are the shear values ​​of the u and v wind components between the layers to be detected, i and i+1 respectively; i,i-1 、Mv i,i-1 Mu are the shear values ​​of the u and v wind components between the layers to be detected i and i-1 respectively; i+2,i+1 、Mv i+2,i+1 Mu are the shear values ​​of the u and v wind components between the i+1 layer and the i+2 layer respectively; i-1,i-2 、Mv i-1,i-2 are the shear values ​​of the u and v wind components between the i-1 and i-2 layers, respectively;

[0261] After the above inspections are completed, all data and the final QC code are output, and quality control is completed.

[0262] For vertical consistency check, add the check instance of station 03020.

[0263] (1) Dead value check

[0264] Figure 6 This is the proportion of horizontal wind direction and horizontal wind speed error data that failed the rigid value check from December 2019 to June 2020. Overall, the daily error rate is less than 0.06%. Figure 7 These are the u and v wind component profiles observed at the 03020 station at 05:00 on April 13, 2020. It can be seen from the figure that at the 1000m, 1500m, 2000m, and 2500m altitude layers, the observation records of the u and v wind components have rigid values ​​(green diamond points), which should be erroneous data caused by instrument failure or transmission errors.

[0265] (2) Mutation inspection

[0266] Figure 8 This is the proportion of horizontal wind direction and horizontal wind speed error data that failed the mutation check from December 2019 to June 2020. Overall, the daily error rate was lower than 0.6%, especially after April 2020, the error rate was below 0.2%. Fig. 9 This is an example of erroneous data of a sudden change in wind speed at station 71994 at 04:30 on December 23, 2019. As can be seen from the figure, the vertical profile of the horizontal wind speed increases as the altitude increases, but a smaller value appears at the 2000m altitude layer, causing the discontinuity of the wind speed curve. In the mutation check, this point was judged as an error.

[0267] (3) Wind shear check

[0268] Fig.10 This is the proportion of horizontal wind direction and horizontal wind speed error data that failed the wind shear check from December 2019 to June 2020. Overall, the daily error rate can be as high as 1.0%, which is slightly higher than the mutation check. This also shows that the wind shear check can detect more error data based on the mutation check. Fig.11 This is an example of wind shear error data that occurred at 08289 station at 12:43 on April 20, 2020. As can be seen from the figure, at the 900m altitude layer, a shear appeared in both the u and v wind component profiles. This error was not identified in the mutation check, but was detected in the wind shear check.

[0269] In addition, the wind shear check also has a strong ability to identify the situation of single shear in horizontal wind. Fig.12 This is the u and v wind component profile of the 06776 station at 01:00 on January 22, 2020. As can be seen from the figure, at the bottom layer of the profile, the horizontal wind speed has a sudden change value, but the wind shear check can accurately identify it.

[0270] When making quality control algorithms, two types of checks are generally set up. One is rapid quality control, which generally only implements threshold-type checks. Its characteristics are fast speed but no quality guarantee. The other is comprehensive quality control. The typical difference is that it requires consistency checks with data from adjacent time periods (previous and next time periods), which is usually called time consistency checks. The final quality control code is comprehensively judged by combining the results of various types of checks. Therefore, comprehensive quality control will affect efficiency. Its characteristics are slow speed but reliable quality.

[0271] The fast quality algorithm of this application does not perform time consistency check, but strengthens the vertical consistency check, which improves the timeliness and ensures the correctness. That is, the quality algorithm of this application is based on the high reliability of meeting the consistency rate of more than 98% between the quality control data and the sounding data, and the quality control speed is not affected. The single-time quality control timeliness can still reach the conventional timeliness level of fast quality control.

[0272] The data were selected for comparison with the data from the same-site observation stations of global soundings, and the rapid quality control method was used to control the quality of wind profiler radar data and weather radar data. The data accuracy is shown in Table 10. The judgment accuracy of the u wind component and the v wind component reached more than 98%.

[0273] Table 10

[0274]

Claims

1. A rapid quality control method for global real-time wind profile data, characterized in that: The steps include: Step (1), performing a characteristic value check on the elements to be checked, and then preprocessing the height elements and horizontal wind elements that pass the check; Step (2), the data that has passed the feature value check and been preprocessed enters the limit value check; Step (3), internal consistency check of observation data with threshold value check and detection device code 0, 3, 6; Step (4) performs a vertical consistency check on the passed data.

2. A rapid quality control method for global real-time wind profile data according to claim 1, characterized in that: The objects of application are four types of observation data: PILOT wind measurement, weather radar, wind profile radar, and radio acoustic detection system. The corresponding detection equipment type codes in the WMO BUFR message are 0, 3, 6, and 8 respectively; The quality control object has 21 elements in total, including area station number, longitude, latitude, station height, year, month, day, hour, minute, second, report category, detection equipment type, sampling height, sampling altitude, geopotential height, horizontal wind direction, horizontal wind speed, u wind component, v wind component, vertical wind speed and signal-to-noise ratio; The meaning of the quality control code is: 0 means the data is correct, 1 means the data is suspicious, 2 means the data is wrong, 8 means the data is missing or there is no observation data, and 9 means no quality control is performed.

3. A rapid quality control method for global real-time wind profile data according to claim 2, characterized in that: In step (1), the characteristic value check is to examine whether a certain observation data is missing or unobserved; if the observation value exists and is not missing, the QC code = 0, and the subsequent check is carried out; if the observation value is missing or unobserved, the QC code = 8, and the subsequent check is not carried out.

4. A rapid quality control method for global real-time wind profile data according to claim 2, characterized in that: Preprocessing of horizontal wind elements: In the global wind profile data uploaded by the Global Data Collection and Distribution System (CTS), the reporting elements of horizontal wind at each station are divided into two forms, namely horizontal wind direction and horizontal wind speed, u wind component and v wind component. It is necessary to uniformly convert horizontal wind direction and horizontal wind speed into u and v wind components. If the QC codes of the horizontal wind direction Wd and horizontal wind speed Ws are both 8, but the QC codes of the u and v wind components are not 8, it means that the reporting elements are the u wind component and the v wind component. Then, it is necessary to calculate the corresponding horizontal wind direction and horizontal wind speed through the u wind component and the v wind component, and participate in the subsequent related inspections; If the QC codes of the u and v wind components are both 8, but the QC codes of the horizontal wind direction and horizontal wind speed are not 8, it means that the reporting elements are the horizontal wind direction and horizontal wind speed. In this case, the corresponding u wind component and v wind component need to be calculated through the horizontal wind direction and horizontal wind speed for subsequent related inspections. The specific conversion formula is as follows: Calculate the u and v wind components based on the horizontal wind direction and horizontal wind speed: u=-Ws×sin (Wd×π / 180.0) (1); v=-Ws×cos (Wd×π / 180.0) (2); Calculate the horizontal wind direction and horizontal wind speed based on the u and v wind components: In formula (1) to formula (5), u and v represent the u wind component and v wind component of the horizontal wind, respectively, in m / s; Wd and Ws represent the horizontal wind direction and horizontal wind speed, in degrees and m / s, respectively; α is the angle parameter, in degrees; π is the circumference of a circle, and its value is 3.1415926; Height element preprocessing: In the global wind profile data uploaded by the Global Data Collection and Distribution System, referred to as CTS, there are four types of observation height upload forms, namely, station height, sampling height, sampling altitude and geopotential height. The station height, sampling height and geopotential height need to be unified as sampling altitude elements for inspection; If the QC code of the sampling altitude element is 8, the corresponding value is calculated by the following method and used as the sampling altitude; if any of the following conditions are not met, no processing is done on the altitude: If the QC codes of the sampling altitude and the station altitude are both 0, the cumulative value of the station altitude and the sampling altitude will be used as the sampling altitude. If the QC code of the geopotential height is 0, the geopotential height is converted to the sampling altitude for use; The element value obtained according to the above preprocessing method is called the calculated quantity; in the subsequent quality control inspection, if the calculated quantity participates in the inspection and is judged to be suspicious or wrong, the calculated quantity will not be assigned a QC code, and its QC code will be assigned to all observed elements involved in the preprocessing of the calculated quantity.

5. A rapid quality control method for global real-time wind profile data according to claim 1, characterized in that: In step (2), the limit value check includes a range check and a climatological limit value check; the observed value of each element is compared with the limit value in the following order: station height, sampling height, sampling altitude / geopotential height, horizontal wind direction, horizontal wind speed, vertical wind speed, signal-to-noise ratio. If the observed value is not within the limit value range, the QC code = 2 and does not participate in the subsequent check; otherwise, the QC code remains unchanged and continues to participate in the subsequent check; If the sampling altitude of an observation record is judged to be wrong, the elements with limit values ​​given according to the altitude range will no longer be subject to limit value checks; after the check is completed, the observation data with the detection equipment type code 8 will be directly output and will no longer enter subsequent checks, while other data will continue to participate in subsequent checks.

6. A rapid quality control method for global real-time wind profile data according to claim 1, characterized in that: In step (3), the internal consistency check is to examine whether the relationship between the records of meteorological elements observed at the same time satisfies a certain physical connection; If Wd≠0 and Ws=0, then the QC code of Wd and Ws=2; If the QC code of Wd = 8 and the QC code of Ws = 0, then the QC code of Ws = 2; If the QC code of Wd = 0 and the QC code of Ws = 8, then the QC code of Ws = 2; Wd and Ws are horizontal wind direction and horizontal wind speed, respectively; If the above conditions are not met, it is deemed to have passed the internal consistency check, the QC code remains unchanged, and proceeds to subsequent inspections.

7. A rapid quality control method for global real-time wind profile data according to claim 1, characterized in that: In step (4), the vertical consistency check is to examine whether the changes in meteorological records within a certain vertical range have a specific regularity; the vertical consistency check is divided into three check items, namely, the rigid value check, the mutation check and the wind shear check; The rigidity check is to check whether the average variability of horizontal wind within a certain vertical range is less than the minimum value; the vertical consistency check and wind shear check are to check whether the average variability of horizontal wind within a certain vertical range exceeds the maximum value; The purpose of mutation check is to identify data with large vertical variability within a certain vertical range; Wind shear checks are designed to identify data where shear values ​​between adjacent altitude levels exceed a threshold; Before checking, first select a single vertical observation profile and sort the observation records one by one in order of sampling altitude from low to high. If an observation record meets any of the following conditions, the record will not be sorted; Sampling altitude QC code = 2; The sampling altitude QC code = 8 and does not meet the conditions for calculating the sampling altitude; Either QC code of horizontal wind direction or horizontal wind speed is 2 or 1; The QC code of any one of the u wind component and the v wind component is 2 or 1; The QC codes for horizontal wind direction, horizontal wind speed, u wind component, and v wind component are all 8.

8. A rapid quality control method for global real-time wind profile data according to claim 7, characterized in that: The sorted vertical profiles are checked for rigidity in turn. The specific algorithm for rigidity check is: During the inspection, start from the i-th layer with the lowest sampling altitude in the single vertical profile, and check in turn whether the observation values ​​of the i+1, i+2, i+3, ... layers above it remain the same as those of the layer to be inspected; If the observed values ​​of the horizontal wind direction and horizontal wind speed, or the u wind component and the v wind component, of three or more consecutive altitude layers do not change, and the difference in sampling altitude between adjacent layers is within 500m, then the QC code of the horizontal wind direction and horizontal wind speed, or the u wind component and the v wind component corresponding to all layers with rigid values ​​is 2, otherwise, the QC code remains unchanged; Then continue to check upward from the first non-rigid value layer above layer i until the top layer; observation records that fail the rigid value check are directly output and no longer participate in subsequent checks.

9. A rapid quality control method for global real-time wind profile data according to claim 7, characterized in that: Mutation check is to check whether the change of meteorological records within a certain vertical range exceeds the maximum threshold based on the algorithm of increment check, vertical residual check and comprehensive judgment; mutation check is performed through observation records of rigid value check; (t-1) Before the inspection, first calculate the weighted sliding mean of the u and v wind components of each layer of the profile; select the sorted single vertical observation profile, take the layer to be detected i as the center, select all the height layers within 500m above and below it, and calculate the weighted sliding mean of the u and v wind components respectively according to the weighted sliding average method; The weighted moving average method is based on the moving average method, taking into account the weight value of each data, thereby eliminating the impact of accidental mutation values ​​on the average value; The greater the absolute deviation between a point and the observed values ​​in its adjacent height range, the lower its weight in the sliding average calculation process; During calculation, if layer i is the first layer or the last layer, only the altitude layers within 500m above or below it are selected to calculate the average value; the specific calculation method is as follows: In the above formula (6), i is the layer to be detected, j is a layer within 500m above and below, and a i is the weighted sliding average of the observations at layer i, n u 、n d are the number of floors within 500m above and below the i-th floor, o j is the observed value of layer j, w j is the weight coefficient of the observation value in layer j; Each layer involved in the calculation must meet Conditions to ensure system stability: In the above formula (7), d j is the observed value x at level j j The absolute deviation from the sliding mean of all observations in the window is calculated as shown in formula (8); if d j = 0, which makes it impossible to calculate w j , then dj=10 -9 ; If u +n d +1<3, the weighted sliding mean of the layer i to be detected is assigned as missing; (t-2) Incremental check: Calculate the mean difference magnitude ZIi of the layer to be detected, which is used to calculate the average vector residual magnitude RESMN in the comprehensive judgment, so as to check whether the vertical change of the point exceeds the threshold; the mean difference magnitude calculation method is as follows: ud i =u i -do i (9); vice president i =v i -va i (10); In the above formulas (9) to (12), the subscript i represents the layer to be detected, and u i 、v i u and v wind components respectively; ua i , va i are the weighted sliding means of the u and v wind components respectively; ud i 、vd i are the mean difference of u and v wind components respectively; vvdi is the mean difference of horizontal wind; ZIi is the magnitude of the mean difference; XI is the mean difference parameter corresponding to the sampling altitude of this layer. If any one of the observation value and the weighted sliding mean is missing, the corresponding vvd i and ZI i All are assigned as missing; (t-3) Vertical residual check: If the mean difference ud of the u and v wind components of the layer i to be detected is i 、vd i If the absolute values ​​are within 10 times of the XI of the corresponding altitude layer, the vertical residual magnitude ZV of the layer to be detected is calculated. i , used to calculate the average vector residual magnitude RESMN in the comprehensive judgment, so as to examine whether the vertical change of the point exceeds the threshold; The vertical residual magnitude is calculated as follows: ① If the layer i to be detected is the bottom layer and its vertical distance to the layer i+1 is within 500m, then: UV i =out i -out i+1 (13); vv i =vd i -vd i+1 (14); ② If the layer i to be detected is the top layer and its vertical distance from the i-1 layer is within 500m, then: UV i =out i -out i-1 (15); vv i =vd i -vd i-1 (16); ③ If the layer i to be detected is the middle layer, and its vertical distance from the i-1 and i+1 layers is within 500m, then: UV i =out i -0.5(out i-1 +out i+1 ) (17); vv i =vd i -0.5(vd i-1 +vd i+1 ) (18); In the above formulas (15) to (18), ud i 、vd i are the mean difference of the u and v wind components of the i-th layer, ud i+1 、vd i+1 are the mean difference of the u and v wind components of the i+1th layer, ud i-1 、vd i-1 are the mean difference of the u and v wind components of the i-1th layer, uv i 、vv i are the vertical residuals of the u and v wind components of the i-th layer respectively, and then the vertical residual of the horizontal wind of the i-th layer to be tested is calculated, and the formula is: In formula (19), vvv i is the vertical residual of horizontal wind at the i-th layer; in formula (20), ZV i is the vertical residual magnitude; XV is the vertical residual threshold corresponding to the sampling altitude of this layer. If any of the physical quantities involved in the calculation is missing or does not meet any of the conditions ① to ③, the corresponding vvv i and ZV i All are assigned as missing; (t-4) Comprehensive judgment: ① Calculate the mean vector residual magnitude RESMN; If ZI i With ZV i If all measurements are not missing, then RESMN=(ZI i +ZV i ) / 2; If ZI i Missing test, but ZV i If there is no missing measurement, RESMN=ZV i ; If ZV i Missing test, but ZI i If there is no missing measurement, RESMN=ZI i ; ② Calculate the threshold R for judging the QC code: If Hi≤10000m, then R=8.0; If Hi ≥ 13000m, then R = 3.0; If 10000m<Hi<13000m, then: In formula (21), H i is the sampling altitude of the layer i to be detected, in meters, m; ③ Determine the QC code: If RESMN>R is satisfied, the QC code of the u and v wind components of layer i = 2, and the observation record is directly output and no longer involved in subsequent inspections; otherwise, the data is correct and the QC code remains unchanged.

10. A rapid quality control method for global real-time wind profile data according to claim 7, characterized in that: Wind shear check through observation records of sudden change check; First, starting from the layer i to be detected, all the altitude layers above and below the layer to be detected are numbered in sequence, and the following checks are performed: (f-1) If the layer i to be detected is the bottom layer, and the vertical distances between the i+1 layer and the i layer, and between the i+1 layer and the i+2 layer are all within 900m, if any of the following conditions is met, the QC code of the u and v wind components = 2, otherwise, the QC code remains unchanged; ①Mu i+1,i >0.03m -1 ·s / m and Mu i+2,i+1 ≤0.03m -1 ·s / m, and Mu i+1,i With Mu i+2,i+1 The signs are opposite; ②Mv i+1,i >0.03m -1 ·s / m and Mv i+2,i+1 ≤0.03m -1 ·s / m, and Mv i+1,i With Mv i+2,i+1 The signs are opposite; (f-2) If the layer i to be detected is the top layer, and the vertical distances between the i-1 layer and the i-2 layer are within 900 m, if any of the following conditions are met, the QC code of the u and v wind components = 2, otherwise, the QC code remains unchanged; ①Mu i,i-1 >0.03m -1 ·s / m and Mu i-1,i-2 ≤0.03m -1 ·s / m, and Mu i,i-1 With Mu i-1,i-2 The signs are opposite; ②Mv i,i-1 >0.03m -1 ·s / m and Mv i-1,i-2 ≤0.03m -1 ·s / m, and Mv i,i-1 With Mv i-1,i-2 The signs are opposite; (f-3) If the layer i to be detected is the middle layer, and the vertical distances between the i-1 layer and the i layer, and between the i+1 layer and the i layer are all within the range of 900 m, if any of the following conditions are met, the QC code of the u and v wind components = 2, otherwise, the QC code remains unchanged; ①Mu i+1,i >0.03m -1 ·s / m and Mu i,i-1 >0.03m -1 ·s / m, and Mu i+1,i With Mu i,i-1 The signs are opposite; ②Mv i+1,i >0.03m -1 ·s / m and Mv i,i-1 >0.03m -1 ·s / m, and Mv i+1,i With Mv i,i-1 The signs are opposite; In the above conditions, the calculation method of each variable is as follows: In the above formulas (22) to (29), u n 、v n , Z n are the u and v wind components and sampling altitude of the nth layer, respectively, n is i, i+1, i+2, i-1, i-2; Mu i+1,i 、Mv i+1,i Mu are the shear values ​​of the u and v wind components between the layers to be detected, i and i+1 respectively; i,i-1 、Mv i,i-1 Mu are the shear values ​​of the u and v wind components between the layers to be detected i and i-1 respectively; i+2,i+1 、Mv i+2,i+1 Mu are the shear values ​​of the u and v wind components between the i+1 layer and the i+2 layer respectively; i-1,i-2 、Mv i-1,i-2 are the shear values ​​of the u and v wind components between the i-1 and i-2 layers, respectively; After the above inspections are completed, all data and the final QC code are output, and quality control is completed.

Images