Foundation microwave radiometer inversion secondary product quality control optimization method

Abstract

The invention discloses a ground-based microwave radiometer inversion secondary product quality control optimization method, which comprises the following steps of: checking an allowable value: checking whether detected data is in a physical range allowed by an observation value or a statistical value of a checking element; aiming at the temperature profile, a temperature threshold value layering setting method is adopted; aiming at the temperature profile, carrying out collaborative quality control by utilizing sounding data; checking a main change range, namely checking whether the detected data is in a climatic change range allowed by the element and the level of the observation station to which the detected data belongs; checking the vertical consistency; and carrying out data evaluation by adopting average deviation, root-mean-square error, standard deviation of deviation distribution and Euclidean distance. On the basis of an existing rapid quality control scheme, aiming at the defects in the aspect of quality control threshold value parameter setting, the effect of the original quality control scheme is improved in three aspects of temperature threshold value layered setting, collaborative quality control carried out with sounding data of the same time, and standard deviation threshold value parameter dynamic adjustment of the temperature vertical variation rate.

Description

Technical Field

[0001] The present invention relates to the technical field of quality product control methods, and more specifically to a quality control optimization method for secondary products inverted by a ground-based microwave radiometer. Background Art

[0002] Obtaining atmospheric temperature and humidity profile information through various detection methods (MWR, radiosonde, satellite remote sensing observation, etc.) is of great significance to the research in the fields of atmospheric vertical structure development, weather forecast, severe convective weather monitoring, artificial weather modification, etc. Ground-based microwave radiometers (MWR) can observe the atmospheric radiation brightness temperature (brightness temperature, the brightness temperature value reflects the radiation characteristics of oxygen, water vapor and cloud liquid water in the vertical path of the atmosphere) 10 kilometers above the surface. Based on the brightness temperature, the atmospheric temperature and humidity profile can be effectively inverted (Turner et al., 2007; Forkman et al., 2012; and Maier, 2012). As a passive remote sensing atmospheric equipment, MWR has the application advantage of all-weather continuous measurement, which can not only make up for the shortcomings of conventional sounding due to the long observation time interval to obtain insufficient atmospheric information, and can realize unattended detection, but also make up for the shortcomings of satellite remote sensing for low-altitude measurement performance.

[0003] Based on the above comparison, it can be seen that MWR does have some advantages over sounding observations and satellite remote sensing observations (MSU) in detecting atmospheric temperature and humidity profiles. Improper neural network inversion coefficients in the product inversion process, water accumulation on the antenna cover in the observation environment, and radio interference will lead to obvious errors in the secondary inversion data products (Fu Xinshu et al., 2017; Mao Jiajia et al., 2022).

[0004] Therefore, it is necessary to carry out technical upgrades on the quality control scheme of the MWR secondary inversion data products to fully play the important foundation and guarantee of the value of microwave radiometer detection data. Summary of the invention

[0005] To this end, the technical problem to be solved by the present invention is to provide a more accurate quality control optimization method for secondary products of ground-based microwave radiometer inversion by accurately designing quality control threshold parameters.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] The quality control optimization method of the secondary product inversion of ground-based microwave radiometer includes the following steps:

[0008] Step P1: allowable value check, checking whether the inspected data is within the physical range allowed by the observed value or statistical value of the inspection element;

[0009] Step P2: for the temperature profile, a temperature limit value layered setting method is adopted;

[0010] Step P3: Conduct collaborative quality control on the temperature profile using sounding data;

[0011] Step P4: Check the main range of variation, check whether the data under test is within the climatological variation range allowed for the element and level of the station to which it belongs;

[0012] Step P5, vertical consistency check: by comparing the difference between a certain point in the vertical direction and its adjacent data before and after, check whether there is an abnormal value outside the overall profile;

[0013] Step P6: Perform data evaluation using mean deviation, root mean square error, standard deviation of deviation distribution, and Euclidean distance.

[0014] In the above-mentioned method for optimizing the quality control of the secondary product inverted by the ground-based microwave radiometer, in step P1, the allowable value check is carried out layer by layer for each element. If one or more layers of data of a certain element value or a certain profile fail to pass the check, the check element or the profile is marked as an error, and the quality control code is assigned to 2; if it passes the check, the quality control code is assigned to 0;

[0015] The elements to be checked include the following: ground air temperature, ground relative humidity, ground air pressure, infrared temperature, cloud base height, total integrated water vapor, total integrated liquid water, temperature profile, water vapor density profile, and relative humidity profile.

[0016] In the above-mentioned ground-based microwave radiometer inversion secondary product quality control optimization method, in step P2, the temperature limit value layered setting method is:

[0017] Step P2-1, the entire 0-10km altitude layer is divided into 5 thickness layers according to the pressure altitude layer, namely, the near-surface layer, 775hPa-1000hPa, 550hPa-750hPa, 200hPa-500hPa and the near-top layer, and is respectively divided into 5 thickness layers according to the number of temperature profile layers, namely, 1-9 layers, 10-53 layers, 54-67 layers, 68-77 layers, and 78-83 layers;

[0018] Step P2-2: Based on the long series data of ERA5 reanalysis, calculate the climatological upper and lower limits of the mean temperature of each thickness layer, that is, the highest average temperature in July of each year is recorded as T ymax The lowest average temperature is in January, denoted as T ymin , the standard deviation of the average temperature of each month is not much changed and is recorded as σ. Using the 3σ principle, the upper and lower limits of the temperature of each thickness layer are T ymax +3σ and T ymin -3σ, the allowable temperature variation range of this thickness layer is: [Tymin -3σ,T ymax +3σ].

[0019] The above-mentioned method for optimizing the quality control of the secondary products inverted by the ground-based microwave radiometer, in step P3, carries out a method for collaborative quality control with the sounding data of the same time:

[0020] Step P3-1, select the data of the microwave radiometer of a certain station and the temperature data of a certain altitude layer of the sounding data of the same period, and calculate and determine the deviation threshold;

[0021] Step P3-2: When the difference between the temperature of a certain altitude layer inverted by the microwave radiometer and the sounding temperature of the same altitude layer at the same time exceeds the deviation threshold, it means that the inverted temperature profile fails to pass the collaborative quality control link of this step, and the quality control code of the microwave radiometer temperature profile is set to 1, which means it is suspicious.

[0022] In the above-mentioned ground-based microwave radiometer inversion secondary product quality control optimization method, in step P4, this step is for the ground temperature element. If any layer of the checked element exceeds the upper and lower limits of the climatological boundary value of the automatic observation station, the quality control code is assigned to 2, otherwise it is assigned to 0.

[0023] In the above-mentioned ground-based microwave radiometer inversion secondary product quality control optimization method, in step P5, the vertical consistency check includes a stiffness check and a standard deviation check of the vertical temperature change rate.

[0024] Step P5-1: During the rigid value check, if the same ground temperature value and ground relative humidity value in the vertical direction appear continuously for 15 or more times at any time, the temperature profile and relative humidity profile are judged to be wrong, and the quality control code is assigned to 2;

[0025] In step P5-2, the standard deviation of the vertical temperature change rate is checked. The vertical temperature change rate indicates the vertical change amplitude of the temperature in °C / 100m and is defined as

[0026]

[0027] Among them, dT h is the vertical temperature change rate of the hth layer, T h and T h+1 are the temperatures of the hth layer and (h+1)th layer, respectively, H h and H h+1 is the corresponding height;

[0028] The standard deviation of the vertical temperature change rate is defined as follows:

[0029]

[0030] in, is the height dT of the same profileh When δ exceeds a certain range, the temperature profile is judged as an error profile and the quality control code is assigned 2, otherwise the quality control code is assigned 0.

[0031] In the above-mentioned ground-based microwave radiometer inversion secondary product quality control optimization method, in step P5-2, for different models of microwave radiometers, the value of the standard deviation δ parameter of the vertical temperature change rate is different: the value of the δ parameter is dynamically adjusted to ensure that the data quality control pass rate is above 95%.

[0032] The above-mentioned ground-based microwave radiometer inversion secondary product quality control optimization method, TQ967 model: δ parameter value is 2.4℃ / 100m;

[0033] YKW2 model: δ parameter value is 0.8℃ / 100m;

[0034] ZP model: The δ parameter value is 1.6℃ / 100m.

[0035] The method for optimizing the quality control of secondary products inverted by a ground-based microwave radiometer according to claim 1, characterized in that in step P6,

[0036] Mean Deviation:

[0037] Standard Deviation:

[0038] Root Mean Square Error:

[0039] Euclidean distance:

[0040] Among them, Xi and Yi represent the temperature data of different data at different times in the i-th layer of the atmosphere, N represents the sample size, and L represents the total number of profile layers.

[0041] Using formulas (3) and (5), the average bias (Bias) and root mean square error (RMSE) of each layer between different data can be calculated, which can intuitively reflect the difference between data. Sta in formula (4) represents the standard deviation of the bias between different data, which indicates the variability of temperature bias between different data. The closer the results of the three indicators are to 0, the more consistent the temperature of the two sets of data is, and vice versa, the greater the difference between the two sets of data is. In order to comprehensively characterize the influence of RMSE and Bias, the evaluation index ED in formula (6) is designed, which represents the weighted Euclidean distance of atmospheric temperature data of each layer between different data, and can also indicate the degree of deviation of the temperature profile between two data.

[0042] The technical solution of the present invention achieves the following beneficial technical effects:

[0043] 1. Based on the existing rapid quality control scheme, in view of the deficiencies in the setting of quality control threshold parameters, the existing quality control scheme was upgraded and optimized in three aspects: layered setting of temperature limit values, coordinated quality control with simultaneous sounding data, and dynamic adjustment of standard deviation threshold parameters of temperature vertical variability, and the improvement effect was analyzed. The test results show that the improvements in the above three aspects have improved the effect of the original quality control scheme to a certain extent.

[0044] 2. Considering that the atmospheric temperature decreases vertically at a rate of approximately 0.65℃ / 100m with altitude, a layered design method for temperature limit values ​​is proposed. The entire 0-10km altitude layer is roughly divided into five thickness layers according to the pressure altitude layer. Based on the long sequence data of ERA5 reanalysis, the climatological upper and lower limits of the average temperature of each thickness layer can be calculated.

[0045] 3. There is a significant inversion layer in the 1800m-3800m altitude layer. The microwave radiometer inverted temperature profile has a poor ability to capture the inversion layer. The microwave radiometer temperature profile is completely unable to depict the appearance of the inversion layer. Therefore, it is proposed to increase the collaborative quality control with the simultaneous sounding temperature profile, calculate and determine the deviation threshold. When the difference between the temperature of a certain altitude layer inverted by the microwave radiometer and the simultaneous sounding temperature of the same altitude layer exceeds the threshold, it means that the inverted temperature profile has failed to pass the collaborative quality control link of this step.

[0046] 4. As the deviation limit value decreases and the quality control conditions are tightened, the number of profiles detected as abnormal increases significantly, and most of the abnormal temperature profiles with ED values ​​are detected by quality control. By comparing the microwave radiometer and sounding data with longer time periods and matching each other, the long-term series change characteristics of evaluation indicators such as RMSE, Bias, and ED can be used to detect data anomalies over a period of time.

[0047] 5. According to different models of microwave radiometers, the value of the standard deviation δ parameter of the vertical temperature change rate is dynamically adjusted to ensure that the quality control pass rate of the temperature profile data is above 96%. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a photo of the appearance of the ground-based microwave radiometer instrument;

[0049] Figure 2 The time series diagram of the mean and variance changes of atmospheric temperature in three thickness layers obtained based on ERA5 data, a is for Aksu Station in Xinjiang, b is for Beijing Station, and c is for Qingyuan Station in Guangdong;

[0050] Figure 3Time series diagram of temperature deviation between microwave radiometers (after stratified quality control) at different stations and sounding data and ERA5 reanalysis data at 200-750hPa and 775-1000hPa from January to March 2024: a) Aksu, Xinjiang; c) Beijing; e) Qingyuan, Guangdong; Time series diagram of standard deviation: b) Aksu, Xinjiang; d) Beijing; f) Qingyuan, Guangdong;

[0051] Figure 4 It is the weighted Euclidean distance time series of the upper and lower layers of the atmosphere between the microwave radiometer at Beijing 54511 station and the ERA5 reanalysis data before and after quality control from January to March 2024; a) is the upper layer of the atmosphere, b) is the lower layer of the atmosphere;

[0052] Figure 5 This is a comparison chart of the temperature profile of the microwave radiometer at Beijing 54511 station at 2024-02-20 and the temperature profile of the second-level sounding data;

[0053] Figure 6 It is the weighted Euclidean distance time series of the upper and lower layers of the atmosphere before and after quality control of the microwave radiometer and sounding data at the Beijing station from January to March 2024 when the deviation limit value is set to 12°C. a) is the upper layer of the atmosphere, and b) is the lower layer of the atmosphere;

[0054] Figure 7 It is the weighted Euclidean distance time series of the upper and lower layers of the atmosphere before and after quality control of the microwave radiometer and sounding data of Beijing Station 54511 from January to March 2024 when the deviation limit value is set to 10℃, a) is the upper layer of the atmosphere, b) is the lower layer of the atmosphere;

[0055] Figure 8 is the frequency probability distribution of the standard deviation of the vertical temperature change rate of microwave radiometers of different models. DETAILED DESCRIPTION

[0056] 1. Introduction of Microwave Radiometer Data Products

[0057] Microwave radiometer (eg Figure 1 The microwave radiometer (shown in Figure 1) is a passive microwave remote sensing electronic instrument used to receive microwave noise energy radiated, scattered or reflected by various objects within the antenna field of view, and convert it into black body temperature. The microwave radiometer uses atmospheric radiation brightness temperature data in the V-band atmospheric oxygen absorption band (51-59GHz) and the K-band atmospheric water vapor absorption band (22-31GHz), combined with ground meteorological data and localized regional model algorithms, to continuously and automatically obtain minute-level temperature, relative humidity, water vapor density and liquid water profile data within the range of 10 kilometers from the ground to the upper air around the clock.

[0058] The ground-based microwave radiometer can generate three levels of data, namely LV0, LV1, and LV2, which represent zero-level data, first-level data, and second-level data. LV0 data is the voltage value received by each channel, which is unprocessed rough data. LV1 data is the brightness temperature value observed by each channel, which comes from the calibration information in the LV0 data and configuration file of the same period. LV2 data mainly includes the atmospheric temperature, water vapor density, relative humidity, and liquid water density profile obtained by inverting the radiation brightness temperature data (LV1) of the same period and the neural network method.

[0059] Based on the LV1 data, the brightness temperature is observed, and statistical inversion methods such as multivariate linear regression or nonlinear neural network methods can be used to invert the atmospheric temperature and humidity profile with high temporal resolution. At present, there are still some uncertainties in the accuracy of the atmospheric temperature and humidity profile (LV2 data) inverted from the brightness temperature (LV1 data) observed by ground-based microwave radiometers, especially the quality of the brightness temperature data itself will directly affect the inverted atmospheric profile.

[0060] Usually, meteorological business departments directly use the atmospheric temperature and humidity profiles provided by ground-based microwave radiometers. Therefore, this application focuses on the data quality assessment method and quality control scheme for the real-time collection of microwave radiometer inversion secondary products (temperature and humidity profiles, etc.) by the information center. The vertical resolution of the microwave radiometer inversion secondary product varies with altitude, ranging from 25m to 250m, and the time resolution is 2min [≤25 / 50m (0~500m); ≤50 / 100m (500~2000m); ≤250m (2~10km) sampling rate ≤2min].

[0061] At present, in real-time business, the Information Center receives about 94 microwave radiometer data upload sites, and the types of instruments from different manufacturers are actually different. Among them, the instrument types with more sites are 6000A, TQ967, HTG4, SMR10, etc. Among them, 6000A and TQ967 are the most, with 27 stations each; ITG4 / 41 has 19 stations; SME10 has 9 stations; YKW2 has 7 stations; TK01 has 5 stations; the number of stations for the other 10 types is less than 3 stations.

[0062] 2. Existing Quality Control Methods

[0063] 2.1. The inspection steps of each element are shown in Table 1. In general, a relatively comprehensive inspection was carried out on the temperature and relative humidity profiles of the microwave radiometer, and only the allowable value inspection was carried out on the remaining 10 elements such as relative humidity, air pressure, infrared temperature, cloud base height, total liquid water integrated amount, precipitation mark, water vapor density and liquid water density.

[0064] Table 1 Quality control steps for different elements in microwave radiometer secondary products

[0065]

[0066] Total Total Allowed value check √ √ √ √ √ √ √ √ √ √ Major changes scope check √ Vertical consistency check √ √

[0067] The specific meaning and control principles of each step inspection are as follows:

[0068] 2.1.1. Allowable value check: check whether the inspected data is within the physical range allowed by the observed value or statistical value of the inspection element;

[0069] Data that exceeds the climate extreme range is marked as an error. The allowable value check is carried out element by element and layer by layer. If one or more layers of data for a certain element value or a certain profile fail the check, the element or the profile is marked as an error and the quality control code is assigned to 2; if it passes the check, the quality control code is assigned to 0. The parameter threshold of the limit value comes from the sounding data quality control plan. The allowable value range of each element is shown in Table 2.

[0070] Table 2 Allowable range of each element in the secondary product of microwave radiometer

[0071] elements Allowable value range unit Ground air temperature -89.2～60 ℃ Ground relative humidity 0～100 ％ Ground pressure 300～1200 h Infrared temperature -50～50 ℃ Precipitation Identifier 0 or 1 - Cloud base height 0.015～22 km Total water vapor integral 0～100 mm Total liquid water integral 0～100 mm Temperature profile -120～60 ℃ Water vapor density profile 0～100 <![CDATA[g / m 3 ]]> Relative humidity profile 0～100 ％ Liquid water density profile 0～100 <![CDATA[g / m 3 ]]>

[0072] 2.1.2. Check the main range of variation: Check whether the data under inspection is within the climatological variation range allowed for the element and level of the station. If any level of the element under inspection exceeds the upper and lower limits of the station climatological limit value, the quality control code is assigned 2, otherwise it is assigned 0.

[0073] 2.1.3. Vertical consistency check: By comparing the difference between a certain point in the vertical direction and its adjacent data before and after, check whether there are abnormal values ​​outside the overall profile. The temperature changes regularly with height, and the temperature difference between two adjacent heights should be within a certain range. Common situations that do not meet vertical consistency include wild points in the vertical direction of the temperature profile, large changes between temperature layers, and rigid values. In response to the above situations, the vertical rigid value check and the standard deviation check of the vertical temperature change rate are designed with the number of unchanged temperature data and the vertical temperature change rate as indicators.

[0074] (a) Dead value check: If the same air temperature and relative humidity values ​​appear consecutively for 15 or more times in the vertical direction at any time, the atmospheric temperature and relative humidity profiles are judged to be wrong.

[0075] (b) Check the standard deviation of the vertical temperature change rate: The vertical temperature change rate indicates the vertical change of the temperature, with the unit of ℃ / 100m, and is defined as:

[0076]

[0077] Among them, dT h is the vertical temperature change rate of the hth layer, T h and Th+1 are the temperatures of the hth layer and (h+1)th layer, respectively, H h and H h+1 is the corresponding height;

[0078] The standard deviation of the vertical temperature change rate is defined as follows:

[0079]

[0080] in, is the height dT of the same profile h The average value of .

[0081] When δ exceeds 0.8℃ / 100m, the temperature profile is judged as an error profile and the quality control code is assigned to 2, otherwise the quality control code is assigned to 0.

[0082] 2.2 Quality Control Code and Its Significance

[0083] In the process of quality control of data, as the quality control process proceeds, it is necessary to continuously set or modify the quality control code, referred to as QC code, for the inspected data. The provisions of QC code are shown in Table 3. If the quality control code of a data is 0, the next quality control step will be entered. If the quality control code is 1 or 2, no subsequent quality control steps will be performed.

[0084] Table 3 Quality Control Code Requirements

[0085] Data attributes QC Code correct 0 suspicious 1 mistake 2

[0086] 2.3. Data evaluation was performed using mean deviation, root mean square error, standard deviation of deviation distribution and Euclidean distance.

[0087] Mean Deviation:

[0088] Standard Deviation:

[0089] Root Mean Square Error:

[0090] Euclidean distance:

[0091] Among them, Xi and Yi represent the temperature data of different data at different times in the i-th layer of the atmosphere, N represents the sample size, and L represents the total number of profile layers; RMSE j It represents the standard deviation of the j-layer atmosphere. ED represents the weighted Euclidean distance of the atmospheric temperature data of each layer between different data.

[0092] 3. Quality Control Upgrade Plan and Effect

[0093] Earlier, during the development of the “Development Plan and Evaluation Report on China Regional Ground-based Microwave Radiometer Real-time Products (V1.0)”, the Data Research Office of the National Meteorological Information Center carried out the design of a rapid quality control scheme for MWR secondary inversion data products for microwave radiometer real-time data. Please refer to the summary of the second part of this application for details. However, the selection of threshold parameters in the design of existing quality control schemes is mainly based on the existing sounding data quality control schemes, and has not been personalized for microwave radiometer data. In particular, the temperature limit values ​​at different heights and the specific values ​​of the standard deviation of the vertical temperature gradient are not clearly given. At the same time, the purpose of the design of existing quality control schemes is mainly to target the gross errors in real-time data, and to emphasize the creation of relevant methods for microwave radiometer data before the live display and application of Tianqing, and it is difficult to take into account the precise design of some specific quality control threshold parameters.

[0094] Based on this, the third part aims to upgrade and optimize the existing quality control scheme based on the rapid quality control scheme used in the above-mentioned evaluation report, targeting the deficiencies in the setting of quality control threshold parameters, in three aspects: 3.1 layered setting of temperature limit values, 3.2 collaborative quality control with simultaneous sounding data, 3.3 microwave radiometers of different instrument models, and selection of standard deviation threshold parameters of temperature vertical variability, and analyze the improvement effects.

[0095] 3.1、Temperature limit value layer setting

[0096] 3.1.1. Threshold value stratified design method

[0097] In the existing rapid quality control methods, the limit value setting for the temperature profile is in the same range [-120℃, 60℃] throughout the 0-10km altitude layer. In fact, the atmospheric temperature has a vertical decrease rate of approximately 0.65℃ / 100m with altitude, that is, the temperature at different altitudes should be within different limit value ranges. Since the current microwave radiometer temperature profile data generally has 83 layers, it is difficult to set different temperature limit values ​​for each layer. In the actual quality control process, considering the efficiency of the quality control algorithm, it is not necessary to set temperature limit values ​​layer by layer.

[0098] Therefore, this section proposes a layered design method for temperature limit values, and divides the entire 0-10km altitude layer into 5 thickness layers (near-surface layer, 775hPa-1000hPa, 550hPa-750hPa, 200hPa-500hPa, near-top layer) according to the pressure altitude layer, and correspondingly divides it into 5 thickness layers (1-9 layers, 10-53 layers, 54-67 layers, 68-77 layers, 78-83 layers) according to the number of profile layers.

[0099] Based on the long series data of ERA5 reanalysis, the climatological upper and lower limits of the mean temperature of each thickness layer can be calculated.

[0100] like Figure 2 The time series of the temperature mean and standard deviation changes of the ERA5 reanalysis data in three thickness layers (775hPa-1000hPa, 550hPa-750hPa, 200hPa-500hPa) from 2013 to 2023 are given at three observation stations: (a) Aksu Station in Xinjiang (51628), (b) Beijing Station (54511), and (c) Qingyuan Station in Guangdong (99280). The dotted lines in the figure represent the temperature mean of each layer, and the shaded part represents the standard deviation of the temperature change of each layer. It can be seen that for the three stations, the temperature change generally has a significant annual cycle change feature, and the average temperature in July of each year is the highest, denoted as T ymax The lowest average temperature is in January, denoted as T ymin , the standard deviation of the average temperature of each month is not much changed and is recorded as σ. Using the 3σ principle, the upper and lower limits of the temperature of each thickness layer are T ymax +3σ and T ymin -3σ, the allowable temperature variation range of this thickness layer is: [T ymin -3σ,T ymax +3σ]. Thus, the range of air temperature variation in each layer at different stations is obtained in Table 4, which provides a reference for the final setting of the temperature limit values ​​of each thickness layer of the microwave radiometer.

[0101] 【The microwave radiometer model of Xinjiang Aksu Station (station number: 51628, 80.3819°E, 41.1197°N, altitude 1108.3m) is ZP type, and there are 177 hours of data in the statistical period of this article. The model of Beijing Station (station number: 54511, 116.4700°E, 38.8070°N, altitude 34m) is TQ967 type, with a total of 156 hours of data. The model of Guangdong Qingyuan Station (station number: 59280, 113.0847°E, 23.7103°N, altitude 80.4m) is YKW2 type, and there are 178 hours of data in the statistical period. The microwave radiometers at the three sites have 83 layers of data in the vertical direction, with a vertical resolution of 25m below 0.5km, a vertical resolution of 50m from 0.5 to 2km, and a layer of observation data every 250m from 2 to 10km.

[0102] Table 4 Temperature variation range of each layer at different stations

[0103]

[0104] With reference to the temperature variation range in Table 4 and the temperature variation range of all levels in the original quality control scheme [-120℃, 60℃], and the empirical rule that the atmospheric temperature decreases at a vertical rate of approximately 0.65℃ / 100m with altitude, the temperature limit values ​​of different altitude layers of the microwave radiometer temperature profile are finally set as shown in Table 5.

[0105] Table 5 Temperature limit values ​​at different altitudes of microwave radiometer temperature profile (℃)

[0106] Reference pressure stratification Approximate number of contour layers Tmin Tmax Near-surface layer 1--9 0km--0.2km -60 60 775~1000 hPa 10--53 0.225km--2.5km -80 40 550~750hPa 54--67 2.75km--6km -90 40 200~500 hPa 68--77 6.25km--8.5km -100 30 Near the top 78--83 8.75km--10km -120 30

[0107] 3.1.2 Effect of quality control scheme upgrade

[0108] like Figure 3 The time series of temperature deviation and standard deviation of microwave radiometers (after stratified quality control) at three stations in Aksu, Xinjiang, Beijing, and Qingyuan, Guangdong, and sounding data and ERA5 reanalysis data at 200-750hPa and 775-1000hPa are shown. It can be seen that at some times with large deviations and standard deviations, the upgraded quality control check failed.

[0109] Among the data evaluation indicators, the weighted Euclidean distance ED in formula (6) is designed to comprehensively characterize the impact of RMSE and Bias. Figure 4 The weighted Euclidean distance between the average upper and lower atmospheric temperatures of the microwave radiometer data of Beijing Station 54511 before and after quality control and the ERA5 and sounding data was used to test the effect of the quality control scheme upgrade. It can be seen that at some times, abnormal temperature profiles with large ED values ​​were identified by quality control.

[0110] However, a careful analysis of the graph shows that at other times, such as the times starting in late February, abnormal temperature profiles with large ED values ​​were not identified by quality control. Obviously, it is difficult to carry out effective quality control of such data anomalies by upgrading the quality control plan by "temperature limit value stratification design" alone. Other methods must be considered, such as the subsequent collaborative quality control with sounding data, to continue to improve the existing quality control plan.

[0111] 3.2. Collaborative quality control with sounding data

[0112] 3.2.1. Collaborative quality control method

[0113] In response to the problems found in 3.1 above, the temperature profile of the microwave radiometer at Beijing 54511 station at 2024-02-20 and the temperature profile of the second-level sounding data were extracted for comparative analysis. Figure 5As shown in the figure, it can be seen that the secondary product of the temperature profile inverted by the microwave radiometer has obvious anomalies compared with the sounding temperature profile observed at the same address and approximately the same time. In other words, although the inverted temperature values ​​at all levels are within the range of "Table 5-Temperature limit values ​​of microwave radiometer temperature profile at different altitudes", the deviation from the actual temperature value observed at the sounding fixed point is large (the temperature difference caused by sounding drift will not be as large as shown in the figure), indicating that the temperature profile inverted by the microwave radiometer is abnormal.

[0114] The sounding temperature profile shows that there is a significant inversion layer at the 1800m-3800m altitude layer at this time. The microwave radiometer inversion temperature profile has a poor ability to capture the inversion layer, and the inversion temperature still decreases with increasing altitude. It can be said that the microwave radiometer temperature profile at this time is completely unable to depict the appearance of the inversion layer.

[0115] Therefore, in the subsequent quality control plan upgrade process, consider adding collaborative quality control with the simultaneous sounding temperature profile. A certain deviation threshold is determined through experiments. When the difference between the temperature of a certain altitude layer inverted by the microwave radiometer and the simultaneous sounding temperature of the same altitude layer exceeds the threshold, it means that the inverted temperature profile fails to pass the collaborative quality control link of this step, and the quality control code of the microwave radiometer temperature profile is set to 1 (suspicious).

[0116] 3.2.2 Effect of quality control scheme upgrade

[0117] like Figure 6 and 7 These are the weighted Euclidean distance time series of the upper and lower layers of the atmosphere before and after quality control when the deviation limits of microwave radiometer and sounding data are set to 12°C and 10°C, respectively. Figure 4 The abnormal temperature profile data in late February that have not been quality controlled can be well quality controlled using the collaborative quality control method with sounding here. Moreover, as the deviation limit value decreases (from 12℃ to 10℃) and the quality control conditions are tightened, the number of profiles detected to be abnormal increases significantly, and most of the abnormal temperature profiles with ED values ​​above 6℃ are quality controlled.

[0118] It is worth mentioning that due to the low frequency of sounding observations and the small number of microwave radiometer stations at the same site as the sounding observations (only about 50), the amount of microwave radiometer data that can be quality controlled by this quality control method is relatively small compared to the microwave radiometer profile observation frequency of once every 2 minutes. Nevertheless, the sounding-coordinated quality control method proposed here can effectively detect abnormal temperature profiles and is still proven to be effective. At the same time, by Figure 6 and 7In the comparison, the microwave radiometer and sounding data with longer time periods and matching each other are selected. Through the long-term series change characteristics of evaluation indicators such as RMSE, Bias, and ED, the data anomalies within a period of time can be checked, and the quality control code of the secondary product data of the microwave radiometer inverted temperature profile at other observation times in the time window should also be set to 1 (suspicious). Specifically, for example, the microwave radiometer data of Beijing Station at 2024022020 (BTC) and 2024022108 (BTC) are coordinated with the sounding data. The data is abnormal (the quality control code is not 0), so the microwave radiometer inverted temperature profile between 2024022020 (BTC) and 2024022108 (BTC) can be set to 1 (suspicious).

[0119] 3.3 Selection of standard deviation threshold parameter for vertical temperature variability

[0120] 3.3.1 Parameter selection method

[0121] In section 2.1.3 (b), the standard deviation check of the vertical temperature change rate is introduced, and it is set that when δ exceeds 0.8℃ / 100m, the temperature profile is judged as an error profile and the quality control code is assigned to 2, otherwise the quality control code is assigned to 0.

[0122] However, the standard deviation of the vertical temperature change rate is 0.8℃ / 100m, which is mainly based on the existing research results (Fu Xinshu, et al. 2017), which was obtained by statistically analyzing the temperature profile data of the MP-3000A microwave radiometer at the Shanghai World Expo Park Station. Whether this parameter threshold is suitable for other types of microwave radiometers requires specific experimental analysis.

[0123] This application proposes that different types of microwave radiometers have different values ​​for the standard deviation parameter of the vertical temperature change rate. Based on the microwave radiometer data from January to March 2024, different threshold parameters are statistically obtained for three different types of instruments, namely Beijing Station (TQ967), Guangdong Qingyuan Station (YKW2), and Xinjiang Aksu Station (ZP), and are used in the quality control upgrade plan. Figure 8, the frequency probability distribution of the standard deviation of the vertical temperature change rate of different types of microwave radiometers is given. It can be seen that the results of the YKW2 microwave radiometer at Qingyuan Station in Guangdong (59280) are closest to those of the MP-3000A at Shanghai World Expo Park Station. When the value is 0.8℃ / 100m, the quality control pass rate of the temperature profile data is above 96%; while for the ZP microwave radiometer at Aksu Station in Xinjiang (51628), the value of 1.6℃ / 100m can ensure that the quality control pass rate of the temperature profile data is above 95%; for the TQ967 microwave radiometer at Beijing Station (54511), the value of 2.4℃ / 100m can ensure that the quality control pass rate of the temperature profile data is above 95%.

[0124] 3.3.2 Effect of quality control scheme upgrade

[0125] As shown in Table 6, the quality control pass rate of microwave radiometer data before and after the threshold parameter adjustment is given. It can be seen that when the δ value is the original fixed value of 0.8, there are significant differences in the quality control pass rate of the data for different types of microwave radiometer data. For the TQ967 instrument at Beijing Station, the data pass rate is only 58%. When the δ value is dynamically adjusted for different instrument types, the data quality control pass rate can be guaranteed to be at the same level, such as the approximately 97% given in the table. At the same time, according to the actual data application requirements, the δ value can be dynamically adjusted to ensure that the required proportion of data passes the quality control inspection in this step, thereby ensuring the quality control effect.

[0126] Table 6 Data passing rate before and after adjustment of quality control threshold parameters

[0127]

[0128] Obviously, the above embodiments are merely examples for the purpose of clear explanation, and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the scope of protection of the claims of this patent application.

Claims

1. Optimization method for quality control of secondary products inverted by ground-based microwave radiometer, It is characterized in that The steps include: Step P1: allowable value check, checking whether the inspected data is within the physical range allowed by the observed value or statistical value of the inspection element; Step P2: for the temperature profile, a temperature limit value layered setting method is adopted; Step P3: Conduct collaborative quality control on the temperature profile using sounding data; Step P4: Check the main range of variation, check whether the data under test is within the climatological variation range allowed for the element and level of the station to which it belongs; Step P5, vertical consistency check: by comparing the difference between a certain point in the vertical direction and its adjacent data before and after, check whether there is an abnormal value outside the overall profile; Step P6: Perform data evaluation using mean deviation, root mean square error, standard deviation of deviation distribution, and Euclidean distance.

2. The method for optimizing the quality control of secondary products inverted by ground-based microwave radiometer according to claim 1, It is characterized in that In step P1, the allowable value check is carried out element by element and layer by layer. If one or more layers of data of a certain element value or a certain profile fail the check, the check element or the profile is marked as an error and the quality control code is assigned to 2; if it passes the check, the quality control code is assigned to 0; The elements to be checked include the following: ground air temperature, ground relative humidity, ground air pressure, infrared temperature, cloud base height, total integrated water vapor, total integrated liquid water, temperature profile, water vapor density profile, and relative humidity profile.

3. The method for optimizing the quality control of secondary products inverted by ground-based microwave radiometer according to claim 1, It is characterized in that In step P2, the temperature limit value layering setting method is: Step P2-1, the entire 0-10km altitude layer is divided into 5 thickness layers according to the pressure altitude layer, namely, the near-surface layer, 775hPa-1000hPa, 550hPa-750hPa, 200hPa-500hPa and the near-top layer, and is respectively divided into 5 thickness layers according to the number of temperature profile layers, namely, 1-9 layers, 10-53 layers, 54-67 layers, 68-77 layers, and 78-83 layers; Step P2-2: Based on the long series data of ERA5 reanalysis, calculate the climatological upper and lower limits of the mean temperature of each thickness layer, that is, the highest average temperature in July of each year is recorded as T ymax The lowest average temperature is in January, denoted as T ymin , the standard deviation of the average temperature of each month is not much changed and is recorded as σ. Using the 3σ principle, the upper and lower limits of the temperature of each thickness layer are T ymax +3σ and T ymin -3σ, the allowable temperature variation range of this thickness layer is: [T ymin -3σ,T ymax +3σ].

4. The method for optimizing the quality control of secondary products inverted by ground-based microwave radiometer according to claim 1, It is characterized in that In step P3, the method of collaborative quality control with the sounding data of the same time is as follows: Step P3-1, select the data of the microwave radiometer of a certain station and the temperature data of a certain altitude layer of the sounding data of the same period, and calculate and determine the deviation threshold; Step P3-2: When the difference between the temperature of a certain altitude layer inverted by the microwave radiometer and the sounding temperature of the same altitude layer at the same time exceeds the deviation threshold, it means that the inverted temperature profile fails to pass the collaborative quality control link of this step, and the quality control code of the microwave radiometer temperature profile is set to 1, which means it is suspicious.

5. The method for optimizing the quality control of secondary products inverted by ground-based microwave radiometer according to claim 1, It is characterized in that In step P4, this step targets the surface temperature element. If any layer of the element being checked exceeds the upper and lower limits of the climatological boundary values ​​of the automatic observation station, the quality control code is assigned a value of 2, otherwise it is assigned a value of 0.

6. The method for optimizing the quality control of secondary products inverted by ground-based microwave radiometer according to claim 1, It is characterized in that In step P5, the vertical consistency check includes a stiffness check and a standard deviation check of the vertical temperature change rate. Step P5-1: During the rigid value check, if the same ground temperature value and ground relative humidity value in the vertical direction appear continuously for 15 or more times at any time, the temperature profile and relative humidity profile are judged to be wrong, and the quality control code is assigned to 2; In step P5-2, in the standard deviation check of the vertical temperature change rate, the vertical temperature change rate indicates the change amplitude of the temperature in the vertical direction, in units of ℃ / 100m, and is defined as: Among them, dT h is the vertical temperature change rate of the hth layer, T h and T h+1 are the temperatures of the hth layer and (h+1)th layer, respectively, H h and H h+1 is the corresponding height; The standard deviation of the vertical temperature change rate is defined as follows: in, is the height dT of the same profile h When δ exceeds a certain range, the temperature profile is judged as an error profile and the quality control code is assigned to 2, otherwise the quality control code is assigned to 0.

7. The method for optimizing the quality control of secondary products inverted by ground-based microwave radiometer according to claim 6, It is characterized in that In step P5-2, for different types of microwave radiometers, the value of the standard deviation δ parameter of the vertical temperature change rate is different: the value of the δ parameter is dynamically adjusted to ensure that the data quality control pass rate is above 95%.

8. The method for optimizing the quality control of secondary products inverted by ground-based microwave radiometer according to claim 7, It is characterized in that TQ967 model: δ parameter value is 2.4℃ / 100m; YKW2 model: δ parameter value is 0.8℃ / 100m; ZP model: The δ parameter value is 1.6℃ / 100m.

9. The method for optimizing the quality control of secondary products inverted by ground-based microwave radiometer according to claim 1, It is characterized in that In step P6, Mean Deviation: Standard Deviation: Root Mean Square Error: Euclidean distance: Among them, X i and Y i Respectively represent the temperature data of different data at different times in the i-th layer of the atmosphere, N represents the sample size, and L represents the total number of layers of the profile; RMSE j represents the standard deviation of the jth layer of the atmosphere.

Images