A satellite temperature profile product correction method and device

Through the probability density matching method and the cumulative distribution lookup table of experience, the systematic error problem of satellite inversion atmospheric temperature profile data is solved, and efficient correction of satellite inversion atmospheric temperature profile is achieved, and the accuracy and credibility of the product are improved. Spend.

CN118520314BActive Publication Date: 2024-11-05ZHEJIANG INST OF METEOROLOGICAL SCI
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410970646.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-11-05
Estimated Expiration
2044-07-19

AI Technical Summary

Technical Problem

The prior art is difficult to effectively eliminate or reduce the systematic error of satellite inversion atmospheric temperature profile data, especially in satellite inversion data with higher spatial and temporal resolution, the error varies greatly with time and space, and the usual linear methods are difficult to correct. .

Method used

By using the probability density matching method (PDF method), by constructing an empirical cumulative distribution lookup table, select the corresponding sub-table according to the mass mark and cloud condition information of the satellite inversion atmospheric temperature profile, obtain the temperature to be corrected for each height layer, and The temperature profile is corrected using the sounding observation temperature as the correction value.

Benefits of technology

Effectively eliminate or reduce the systematic error of satellite inversion atmospheric profile data, improve the accuracy and credibility of the product, clear process, simple operation, and high implementation and general applicability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118520314B_ABST
    Figure CN118520314B_ABST
Patent Text Reader

Abstract

The present invention provides a satellite temperature profile product correction method and device, belonging to the field of satellite remote sensing. The method includes selecting a corresponding subtable in a pre-constructed empirical cumulative distribution lookup table according to the quality mark and cloud condition information of the satellite-inverted atmospheric temperature profile; obtaining the satellite inverted temperature to be corrected for each altitude layer according to the satellite-inverted atmospheric temperature profile; obtaining the satellite cumulative probability distribution corresponding to the satellite inverted temperature to be corrected for each altitude layer according to the satellite-inverted atmospheric temperature profile; obtaining the sounding observation temperature corresponding to the sounding cumulative probability distribution of each altitude layer that is the same as the satellite cumulative probability distribution based on the satellite cumulative probability distribution; using the sounding observation temperature as the temperature correction value of the corresponding satellite inverted temperature to be corrected, and correcting the satellite-inverted atmospheric temperature profile. The method and device of the present invention can realize the correction of satellite atmospheric temperature profile products based on the probability density matching method.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of satellite remote sensing, and more specifically, to a satellite temperature profile product correction method and device. Background Art

[0002] Temperature Profile is a curve that describes the distribution of temperature in the atmosphere with height. Atmospheric temperature and humidity profile is an important parameter of atmospheric thermodynamics. Obtaining high-resolution vertical distribution of atmospheric temperature and humidity helps to reveal the laws of internal atmospheric movement and understand the evolution mechanism of weather at different scales. At the same time, as a basic variable of numerical forecast models and a necessary input of radiation transfer models, atmospheric temperature and humidity profile plays an important role in improving the accuracy of numerical weather forecasts. Radiosonde is the main means of obtaining vertical atmospheric temperature and humidity information. It has high accuracy and representativeness, but it is limited by observation time and site distribution and cannot meet the needs of refined and seamless meteorological services. Satellite observations have the characteristics of high temporal and spatial resolution and wide coverage. Atmospheric information in remote areas or areas with complex geographical conditions (such as oceans, mountains, etc.) can be obtained from them. Infrared hyperspectral detectors have the advantages of high spectral resolution and can continuously obtain high-resolution atmospheric vertical elements. With the development of remote sensing technology, many satellite hyperspectral payloads have been developed at home and abroad to detect high-precision atmospheric temperature and humidity information, such as Atmospheric Infrared Sounder (AIRS), Infrared Atmospheric Sounding Interferometer (IASI), Cross-track Infrared Sounder (CrIS), and Geostationary Interferometric Infrared Sounder (GIIRS). Among them, the GIIRS carried by the Fengyun-4 geostationary satellite has the advantages of large range and high frequency of observation, and can obtain atmospheric temperature and humidity profiles with high temporal and spatial resolution, making up for the shortcomings of conventional sounding data.

[0003] Mastering the detection accuracy of satellite products is the basis for conducting satellite data research and business applications. There are many research results at home and abroad around the inversion performance and accuracy verification of the GIIRS of Fengyun-4A (FY-4A). Some scholars used GIIRS to simulate brightness temperature and invert atmospheric parameters based on the artificial neural network method, and compared it with IASI. They found that the temperature inversion accuracy of GIIRS is better than IASI in the lower atmosphere, and the error is greater than IASI in the upper atmosphere, while the atmospheric humidity inversion accuracy of GIIRS is significantly better than IASI. Ren et al. used the 2020-2021 winter sounding to verify the GIIRS temperature profile product and reconstructed the missing data. The results showed that the mean bias (MB) of GIIRS temperature was 0.07 K, and the root mean square error (RMSE) was 2.546 K. The error was relatively large in the upper and lower troposphere, and small in the middle troposphere. Some scholars have conducted a systematic evaluation of the long series of GIIRS temperature profiles in 2020 based on soundings. The study shows that the product is greatly affected by cloud activity in terms of altitude and monthly statistics; the accuracy in summer and autumn is significantly better than that in winter and spring, and the quality of temperature profile products in low-altitude areas is generally better than that in high-altitude areas. In addition, some scholars have further tested and applied the accuracy of FY-4A / GIIRS profile products for different regions and weather process types.

[0004] The above studies on the accuracy of GIIRS profile products have shown that there are systematic errors that vary with altitude, but there are few reports on how to further deal with these errors, which is the key to improving the credibility and practicality of the products. Systematic errors can lead to negative effects of profile products in actual business applications, such as incorrect estimation of strong convective potential and failure to meet the unbiased assumption required by variational assimilation. Therefore, correcting the systematic errors of profile products can better play the role of the products.

[0005] At present, the correction technology for atmospheric temperature profile products is mainly linear correction. Some scholars use linear regression to fit a univariate equation to correct the deviation of the measured brightness temperature data of ground-based microwave radiometers, making it closer to the simulated brightness temperature, thereby improving the inversion accuracy of ground-based microwave radiometers. Some scholars use linear correction, take sounding and ERA5 reanalysis data as the standard, perform regression analysis based on the least squares principle, obtain the coefficients of the linear regression equation, and then correct the deviation of the secondary data of the ground-based microwave radiometer (that is, the atmospheric temperature and humidity profile). The results show that the linear deviation correction effect of sounding data on rainy days for radiometer temperature and humidity is better than that of sunny and cloudy days. The ERA5 data corrected by sounding data can better improve the accuracy of the radiometer temperature and humidity profile, and the effect on rainy days is better than that on sunny and cloudy days.

[0006] In general, existing correction research on atmospheric temperature profile products is mainly based on ground-based microwave radiometers, while there are fewer reports on correction methods for satellite temperature profile products. The most common mainstream correction method is linear correction.

[0007] Theoretically, any measurement (or estimation) inevitably has errors, which can be divided into two types according to their different statistical characteristics: systematic errors and random errors. Random errors are inherent characteristics that are commonly found in observational data, and are often statistically normally distributed with a mean value of zero. Therefore, the impact of random errors can be basically eliminated in climate averaging. The main difference between systematic errors and random errors is that their mean values ​​are not zero, and there will be obvious deviations from the actual situation when the data is averaged. When the systematic errors are inconsistent in time series or spatial distribution, data inhomogeneity will occur, which will seriously affect the research results. Systematic errors can be divided into independent errors and non-independent errors. The so-called independent errors refer to errors that do not change with region, time or the size of the observed value. They can be eliminated by subtracting the same constant from the entire meteorological field, and the correction method is relatively simple. In actual meteorological data fields, the most common non-independent errors are non-independent errors, which will change with the observed value and are more difficult to correct. For satellite inversion data with high temporal and spatial resolution, it not only has obvious non-independent systematic errors, but also the errors vary greatly with time and space, which are difficult to correct with conventional linear methods. It can be seen that how to effectively eliminate or reduce the systematic errors of satellite-inverted atmospheric temperature profile data is a challenging problem. Summary of the invention

[0008] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a satellite temperature profile product correction method and device to effectively eliminate or reduce the systematic errors of satellite inverted atmospheric temperature profile data, and solve the problem that the satellite inversion data with high temporal and spatial resolution is difficult to correct using conventional linear methods because the errors vary greatly with time and space.

[0009] To achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0010] In a first aspect, the present invention provides an atmospheric temperature profile correction method, comprising the following steps:

[0011] Obtain satellite-derived atmospheric temperature profiles and their quality marks and cloud condition information from satellite temperature profile products;

[0012] According to the quality mark and cloud condition information of the satellite-inverted atmospheric temperature profile, a corresponding subtable is selected in a pre-constructed empirical cumulative distribution lookup table; the empirical cumulative distribution lookup table includes a plurality of subtables corresponding to the quality mark and cloud condition information of the satellite-inverted atmospheric temperature profile; the subtable records the corresponding relationship between the satellite-inverted temperature and the satellite cumulative probability distribution at each altitude layer under the corresponding quality mark and cloud condition information, and the corresponding relationship between the sounding cumulative probability distribution and the sounding observed temperature;

[0013] According to the satellite inverted atmospheric temperature profile, the satellite inverted temperature to be corrected of each altitude layer is obtained; according to the satellite inverted temperature to be corrected, based on the sub-table, the satellite cumulative probability distribution corresponding to the satellite inverted temperature to be corrected of each altitude layer is obtained; and based on the satellite cumulative probability distribution, the sounding observation temperature corresponding to the sounding cumulative probability distribution closest to the satellite cumulative probability distribution of each altitude layer is obtained;

[0014] The sounding observed temperature is used as the temperature correction value of the corresponding satellite inverted temperature to be corrected, and the satellite inverted atmospheric temperature profile is corrected.

[0015] Furthermore, the method for pre-constructing the empirical cumulative distribution lookup table includes:

[0016] Obtain satellite inverted temperature samples and sounding observed temperature samples based on satellite temperature profile products and sounding data;

[0017] Preprocessing the satellite inverted temperature samples and the sounding observed temperature samples to obtain a sample pair; the sample pair includes the satellite inverted temperature samples and the sounding observed temperature samples that are matched simultaneously in time and space;

[0018] According to the quality mark and cloud condition information of the satellite-retrieved atmospheric temperature profile, respectively statistically analyze the empirical cumulative distribution functions of the satellite-retrieved temperature samples and the sounding-observed temperature samples at different altitudes under the same cloud condition information and quality mark in the sample pair, and generate an empirical cumulative distribution lookup table of the corresponding altitudes of the satellite-retrieved temperature samples and the sounding-observed temperature samples with a temperature range of 200-340K and a resolution of 0.1 K by interpolation;

[0019] The empirical cumulative distribution lookup table is divided into multiple sub-tables according to the quality mark and cloud condition information of the satellite-inverted atmospheric temperature profile; the sub-table records the correspondence between the satellite-inverted temperature and the satellite cumulative probability distribution of each altitude layer under the corresponding quality mark and cloud condition information, as well as the correspondence between the sounding cumulative probability distribution and the sounding observed temperature.

[0020] Furthermore, the satellite inversion temperature samples and the sounding observation temperature samples are preprocessed to obtain sample pairs, including:

[0021] Obtain satellite-derived temperature samples and sounding-observed temperature samples at altitudes of 100-1000 hPa;

[0022] Perform quality control on the sounding observation temperature samples, including removing duplicate pressure layer records, checking for abnormal values ​​and sorting the pressure layers, to obtain the sounding observation temperature samples after quality control;

[0023] The sounding observation temperature samples after quality control are interpolated according to the fixed isobaric surface of the altitude layer of the satellite inversion temperature samples;

[0024] The interpolated sounding observation temperature samples and satellite inversion temperature samples are matched in time and space to obtain sample pairs including satellite inversion temperature samples and sounding observation temperature samples that are matched simultaneously in time and space.

[0025] Furthermore, the sounding observation temperature samples and satellite inversion temperature samples after quality control are matched in time and space, including:

[0026] Select the satellite inversion temperature sample within 60 minutes from the sounding observation time of each sounding observation temperature sample to obtain the time matching inversion sample group corresponding to the sounding observation temperature sample; the sounding observation time is the take-off time of the sounding balloon;

[0027] In the time-matched inversion sample group, spatial matching is performed based on the nearest neighbor principle, and the proximity distance between each sounding site of the sounding observation temperature sample and the satellite observation point of the surrounding corresponding satellite inversion temperature sample is calculated in turn;

[0028] If the proximity distance between the sounding site of the sounding observation temperature sample and the nearest satellite observation point is less than the preset distance value, the satellite inversion temperature sample corresponding to the satellite observation point is successfully matched with the sounding observation temperature sample, and the sounding observation temperature sample will not continue to be matched;

[0029] If the proximity distance between the sounding site of the sounding observation temperature sample and the nearest satellite observation point is greater than or equal to the preset distance value, the sounding observation temperature sample is discarded;

[0030] Repeat the above steps until there is a satellite-derived temperature sample matching each sounding observation temperature sample.

[0031] Furthermore, the method for calculating the proximity distance between each sounding station and the surrounding satellite profile products includes: calculating based on longitude and latitude, adding the square of the longitude difference between the sounding station and the satellite observation point to the square of the latitude difference and then taking the root;

[0032] The preset distance value is 0.2 degrees.

[0033] Furthermore, according to the quality mark and cloud condition information of the satellite-retrieved atmospheric temperature profile, the empirical cumulative distribution functions of the satellite-retrieved temperature samples and the sounding-observed temperature samples at different altitudes in the sample pairs under the same cloud condition information and quality mark are respectively counted, including:

[0034] All sample pairs are divided into corresponding sub-sample sets according to cloud condition information and quality marks; the types of cloud condition information include cloud and no cloud; the types of quality marks include 0 and 1;

[0035] The temperature probability density of satellite inversion temperature samples at each altitude layer in the sub-sample set is counted, and the satellite cumulative probability distribution of each temperature is calculated to obtain the empirical cumulative distribution function of discrete satellite inversion temperature at different altitude layers;

[0036] The temperature probability density of the sounding observation temperature samples at each altitude layer in the sub-sample set is statistically analyzed, and the sounding cumulative probability distribution of each temperature is calculated to obtain the empirical cumulative distribution function of discrete sounding observation temperature at different altitude layers.

[0037] Furthermore, the empirical cumulative distribution functions of the satellite inverted temperature samples and the sounding observed temperature samples at different altitudes under the same cloud condition information and quality mark in the sample pairs are respectively counted, and also include:

[0038] When making statistics, no spatial window in the horizontal direction is defined. Instead, all the sounding observation temperature samples and satellite inversion temperature samples at each altitude layer are counted.

[0039] If the total number of satellite-inverted temperature samples and sounding observation temperature samples for a certain altitude layer is less than 300, it is considered that there are too few valid data for the current altitude layer, and the satellite-inverted temperature for that altitude layer will not be corrected.

[0040] Furthermore, the satellite temperature profile product is a FY-4B / GIIRS atmospheric vertical detection regional synthetic product from the official website of the National Satellite Meteorological Center.

[0041] The sounding data are from the meteorological information comprehensive processing system.

[0042] Furthermore, the method further comprises:

[0043] The statistical parameters MB and RMSE are introduced to quantitatively analyze the accuracy of FY-4B / GIIRS temperature profiles before and after correction.

[0044] The normalized contour frequency map is introduced to statistically analyze the vertical error characteristics of the FY-4B / GIIRS temperature profile, including:

[0045] First, the error value is obtained by subtracting the satellite from the sounding value and the number of samples in each layer is counted. The frequency of the error value in each value range and each altitude layer is calculated. The frequency is normalized based on the total number of samples in the corresponding pressure layer to obtain the percentage of the value range in the total number of samples in the pressure layer. Finally, the frequency of occurrence of different error values ​​in all layers is displayed using a color code.

[0046] In a second aspect, the present invention provides a satellite temperature profile product correction device, comprising:

[0047] Acquisition module: used to obtain the satellite-retrieved atmospheric temperature profile, its quality mark and cloud condition information of satellite temperature profile products;

[0048] Table lookup module: used to select a corresponding subtable in a pre-constructed empirical cumulative distribution lookup table according to the quality mark and cloud condition information of the satellite-inverted atmospheric temperature profile; the empirical cumulative distribution lookup table includes a plurality of subtables corresponding to the quality mark and cloud condition information of the satellite-inverted atmospheric temperature profile; the subtable records the corresponding relationship between the satellite-inverted temperature and the satellite cumulative probability distribution at each altitude layer under the corresponding quality mark and cloud condition information, and the corresponding relationship between the sounding cumulative probability distribution and the sounding observed temperature;

[0049] The sounding retrieval module is used to obtain the satellite inversion temperature to be corrected for each altitude layer according to the satellite inversion atmospheric temperature profile, obtain the satellite cumulative probability distribution corresponding to the satellite inversion temperature to be corrected for each altitude layer based on the sub-table according to the satellite inversion temperature to be corrected, and obtain the sounding observation temperature corresponding to the sounding cumulative probability distribution that is the same as the satellite cumulative probability distribution for each altitude layer based on the satellite cumulative probability distribution;

[0050] Correction module: used to use the sounding observation temperature as the temperature correction value of the corresponding satellite inversion temperature to be corrected, and correct the satellite inverted atmospheric temperature profile.

[0051] In a third aspect, the present invention provides a satellite temperature profile product correction device, including a processor and a storage medium;

[0052] The storage medium is used to store instructions;

[0053] The processor is used to operate according to the instructions to execute the steps of the method described in the first aspect.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] 1. The present invention verifies satellite-inverted atmospheric products based on sounding data, analyzes the vertical distribution characteristics of errors under different conditions, and on this basis, statistically calculates the cumulative probability density of the two. The product with obvious systematic errors is corrected by using the Probability Density Function matching method (PDF method). The correction effect is good, the process is clear, and the feasibility is strong.

[0056] 2. At present, the accuracy research on GIIRS atmospheric temperature profile products has shown that there are systematic errors that vary with altitude, but how to further deal with these errors is the key to improving the credibility and practicality of the product. The present invention will first test the FY-4B / GIIRS temperature profile product based on sounding data, analyze the vertical distribution characteristics of errors under different conditions, and then use the PDF method to correct the products with obvious systematic errors, which can effectively eliminate or reduce the systematic errors of satellite-inverted atmospheric profile data, with high accuracy, simple operation and good feasibility.

[0057] 3. The present invention also uses a time-independent validation set to evaluate the correction effect of the PDF method, which reflects the temporal versatility of the method.

[0058] 4. After the empirical cumulative distribution lookup table of the present invention is made through sounding data and satellite data, it can be used directly under the same satellite and region as long as the satellite and instrument are in good operating condition and the inversion algorithm accuracy remains stable, without the need for statistics and training again. It has a wide range of use and stable performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a schematic diagram of the technical route for correcting satellite temperature profile products;

[0060] Figure 2 This is a schematic diagram of the preprocessing of sounding data and satellite temperature profile products;

[0061] Figure 3 The normalized iso-frequency diagram of the satellite temperature profile under clear and cloudy conditions and the schematic diagram of the vertical distribution of MB and RMSE; (a) represents the clear sky condition, (b) represents the cloudy condition; (c) represents the schematic diagram of the vertical distribution of MB and RMSE;

[0062] Figure 4 The error normalized contour frequency diagram of the satellite temperature profile with mass 0 and 1, as well as the vertical distribution diagram of MB and RMSE; (a) represents the case of mass 0, (b) represents the case of mass 1; (c) represents the vertical distribution diagram of MB and RMSE;

[0063] Figure 5The error normalized contour frequency diagram of the satellite temperature profile with a mass of 0 before and after correction, as well as the vertical distribution diagram of MB and RMSE; (a) represents the situation before correction, (b) represents the situation after correction; (c) represents the vertical distribution diagram of MB and RMSE;

[0064] Figure 6 The error normalized contour frequency diagram of the satellite temperature profile with a mass of 1 before and after correction, as well as the vertical distribution diagram of MB and RMSE; (a) represents the situation before correction, (b) represents the situation after correction; (c) represents the vertical distribution diagram of MB and RMSE;

[0065] Figure 7 The figure shows the scatter diagram of the temperature profile of the satellite with mass 0 and 1 before and after correction. (a) shows the scatter diagram of the temperature profile when the mass is 0 before correction, (b) shows the scatter diagram of the temperature profile when the mass is 0 after correction, (c) shows the scatter diagram of the temperature profile when the mass is 1 before correction, and (d) shows the scatter diagram of the temperature profile when the mass is 1 after correction. DETAILED DESCRIPTION

[0066] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0067] Embodiment 1 provided by the present invention:

[0068] The present invention realizes the correction of satellite temperature profile products based on the probability density function matching method (PDF method). In the present invention application, the temperature profile product of the FY-4B interferometric atmospheric vertical sounder (GIIRS) is used as an example to illustrate the specific ideas including the preprocessing of satellite and sounding data, the verification and evaluation of satellite inversion profiles, and the correction test of satellite inversion profiles. The technical route is as follows: Figure 1 As shown, the detailed technical solutions of each part are described in detail below.

[0069] Data and information:

[0070] GIIRS Instrument Introduction

[0071] GIIRS is the world's first precision remote sensing instrument that detects three-dimensional atmospheric vertical structure in geostationary orbit by infrared hyperspectral interferometry. It is groundbreaking and moves from two-dimensional observation to three-dimensional comprehensive observation in geostationary orbit. The second batch of FY-4 satellites, FY-4B, was launched on June 3, 2021. As a business satellite, the GIIRS on board has improved spectral, temporal, spatial resolution and calibration accuracy compared with the A satellite. GIIRS uses a 32×4 array detector on FY-4A and a 16×8 array detector on FY-4B. It observes infrared radiation in different spectral bands by Michelson interferometry, obtains the vertical distribution of atmospheric temperature and humidity, and provides large-scale, continuous, fast and accurate remote sensing information for weather forecasting. Table 1 gives the instrument parameters of GIIRS on FY-4A and FY-4B.

[0072] Table 1 GIIRS instrument parameters

[0073]

[0074] GIIRS atmospheric temperature profile product introduction:

[0075] The present invention uses the FY-4B / GIIRS atmospheric vertical detection regional synthesis product from the official website of the National Satellite Meteorological Center. The product completes regional splicing every 2 hours, with a horizontal resolution of 12 km and data in NETCDF format. The product contains latitude and longitude, cloud detection, and atmospheric temperature, humidity, ozone and other information at an altitude of 0.005~1100 hPa, totaling 101 layers. Based on the L1-level data quality and the difference between the inversion results and the reference value numerical model field, each layer of profile has a corresponding quality mark, where 0 is "perfect", that is, the data quality is the best; 1 is "good", that is, the data quality is good; 2 is "bad", that is, the data quality is poor; 3 is "do not use", that is, the data quality is poor and cannot be used, and "-999999" is a system invalid value. The research of the present invention is mainly aimed at data with quality marks of 0 and 1, and the data of other qualities are eliminated and will not be discussed.

[0076] Sounding data introduction:

[0077] The L-band radarsonde system can determine the spatial position of the sonde at every moment and infer the upper wind direction, wind speed, temperature and humidity. Conventional radiosonde observations come from the Meteorological Information Integrated Processing System (MICAPS), which selects 120 stations across the country. Sonde observations are made twice a day, at UTC 00 and 12 (universal time, the same below), representing daytime and nighttime atmospheric information.

[0078] The deviation correction method adopts the probability density matching method:

[0079] Probability Density Function matching method (PDF method) principle:

[0080] Although the non-independent systematic errors of satellite inversion data vary greatly with time and space, the probability density distribution of inversion elements within a certain time and space range is relatively stable. Therefore, the satellite inversion value can be adjusted to make the probability density distribution of the satellite inversion value consistent with the observed value, so as to achieve the purpose of correcting the systematic errors of satellite inversion data. The above idea is the main idea of ​​the PDF method to correct the systematic errors of data. Studies have shown that the PDF method is superior in correcting non-independent systematic errors. It does not need to analyze the complex sources of systematic errors in the data. The calculation idea is simple and the correction effect is good. In recent years, it has gradually become the mainstream satellite data correction method in the world. There have been many application results in correcting satellite inversion precipitation, satellite inversion of sea surface temperature, satellite inversion of surface solar radiation, etc.

[0081] The PDF method was originally used to correct precipitation forecasts and inversion products. This method can effectively correct the systematic deviations of precipitation forecast rainfall and rain area, but it cannot be used to correct the deviation of the rain area location. Therefore, it has a better correction effect on systematic precipitation or large-scale precipitation, but a poor correction effect on local precipitation or small-scale precipitation. In addition, it can also be extended to other forecast quantities. Unlike precipitation, the atmospheric profile covers a very wide range, so there is no need to worry about the falling area when correcting, which is very suitable for the PDF method.

[0082] The core of this method is to construct an empirical cumulative distribution lookup table; the empirical cumulative distribution lookup table includes multiple sub-tables corresponding to the quality marks and cloud condition information of the satellite-inverted atmospheric temperature profile; the sub-table records the correspondence between the satellite-inverted temperature and the satellite cumulative probability distribution of each altitude layer under the corresponding quality marks and cloud condition information, as well as the correspondence between the sounding cumulative probability distribution and the sounding observed temperature.

[0083] According to satellite data, the altitude layers are divided into 101 layers, which are divided according to the corresponding air pressure.

[0084] The empirical cumulative distribution lookup table (part) is shown in Table 2 and Table 3, where Table 2 is the correspondence table between the satellite inverted temperature and the satellite cumulative probability distribution at each altitude layer; Table 3 is the correspondence table between the sounding cumulative probability distribution and the sounding observed temperature at each altitude layer.

[0085] Table 2 Correspondence between satellite-derived temperature and satellite cumulative probability distribution at each altitude layer

[0086]

[0087] Table 3 Corresponding relationship between the sounding cumulative probability distribution and the sounding observed temperature at each altitude layer

[0088]

[0089] One column in the table represents the same altitude layer number, and one row represents the same temperature value (unit 0.1K); the data in the table represent the cumulative probability distribution; note that the cumulative probability distribution between 0-1 represents the corresponding cumulative probability distribution; the data in the table is 0, which means that the altitude layer is less than 100hPa or higher than 1000hPa, and no correction is made; the data in the table is -999, which means that there are less than 300 samples of the temperature in the altitude group, and no correction is made. The corresponding relationship table between satellite inversion temperature and satellite cumulative probability distribution and the corresponding relationship table between sounding cumulative probability distribution and sounding observed temperature both have 101 columns and 1401 rows, representing the discrete empirical cumulative distribution function of 101 layers, 200-340K, and 0.1K resolution, respectively, which overall represents the corresponding relationship between satellite inversion temperature and satellite cumulative probability distribution, and sounding cumulative probability distribution and sounding observed temperature.

[0090] When in use, first select the corresponding subtable in the pre-constructed empirical cumulative distribution lookup table according to the quality mark and cloud condition information of the satellite-inverted atmospheric temperature profile; obtain the satellite inversion temperature to be corrected for each altitude layer according to the satellite-inverted atmospheric temperature profile, and obtain the satellite cumulative probability distribution corresponding to the satellite inversion temperature to be corrected for each altitude layer based on the altitude layer information and the satellite inversion temperature to be corrected and the corresponding relationship table between the satellite inversion temperature and the satellite cumulative probability distribution; finally, based on the corresponding relationship table between the sounding cumulative probability distribution and the sounding observation temperature, obtain the sounding observation temperature corresponding to the sounding cumulative probability distribution that is the same as the satellite cumulative probability distribution for each altitude layer through the altitude layer information and the satellite cumulative probability distribution.

[0091] Note that the accuracy of the satellite inverted temperature to be corrected is 0.1K. If the satellite cumulative probability distribution corresponding to the satellite inverted temperature to be corrected does not have a sounding cumulative probability that is exactly equal to it in the corresponding relationship table between sounding cumulative probability distribution and sounding observed temperature, the sounding observed temperature corresponding to the sounding cumulative probability closest to the satellite cumulative probability distribution is selected as the temperature correction value for correction. Directly selecting the closest one is simpler in operation, and the accuracy can also reach 0.1K, which can meet the use under general conditions.

[0092] After the empirical cumulative distribution lookup table of the present invention is made by using sounding data and satellite data, it can be directly used under the same satellite and region as long as the satellite and instrument are in good operating condition and the accuracy of the inversion algorithm remains stable, without the need for statistics and training again, and has a wide range of use and stable performance. At the same time, even if the satellite product is replaced, it only needs to obtain the satellite data and sounding data of the satellite product to regenerate, which is convenient and fast.

[0093] Prebuilt methods for empirical cumulative distribution lookup tables include:

[0094] Obtain satellite inverted temperature samples and sounding observed temperature samples based on satellite temperature profile products and sounding data;

[0095] Preprocessing the satellite inverted temperature samples and the sounding observed temperature samples to obtain a sample pair; the sample pair includes the satellite inverted temperature samples and the sounding observed temperature samples that are matched simultaneously in time and space;

[0096] According to the quality mark and cloud condition information of the satellite-inverted atmospheric temperature profile, the empirical cumulative distribution functions of the satellite-inverted temperature samples and the sounding observation temperature samples at different altitudes under the same cloud condition information and quality mark in the sample pairs are respectively counted, and by interpolation, an empirical cumulative distribution lookup table of the corresponding altitudes with a temperature range of 200-340K and a resolution of 0.1K for the satellite-inverted temperature samples and the sounding observation temperature samples is generated.

[0097] Data preprocessing and analysis methods:

[0098] Preprocessing the satellite inverted temperature samples and the sounding observed temperature samples to obtain a sample pair; the sample pair includes the satellite inverted temperature samples and the sounding observed temperature samples that are matched simultaneously in time and space;

[0099] Specifically, the sounding observation temperature samples after quality control and the satellite inversion temperature samples are matched in time and space, including:

[0100] Select the satellite inversion temperature sample within 60 minutes from the sounding observation time of each sounding observation temperature sample to obtain the time matching inversion sample group corresponding to the sounding observation temperature sample; the sounding observation time is the take-off time of the sounding balloon;

[0101] In the time-matched inversion sample group, spatial matching is performed based on the nearest neighbor principle, and the proximity distance between each sounding site of the sounding observation temperature sample and the satellite observation point of the surrounding corresponding satellite inversion temperature sample is calculated in turn;

[0102] If the proximity distance between the sounding site of the sounding observation temperature sample and the nearest satellite observation point is less than the preset distance value, the satellite inversion temperature sample corresponding to the satellite observation point is successfully matched with the sounding observation temperature sample, and the sounding observation temperature sample will not continue to be matched;

[0103] If the proximity distance between the sounding site of the sounding observation temperature sample and the nearest satellite observation point is greater than or equal to the preset distance value, the sounding observation temperature sample will be discarded; the calculation method for the proximity distance between each sounding site and the surrounding satellite profile products includes: calculating based on longitude and latitude, adding the square of the longitude difference and the square root of the latitude difference; the preset distance value is 0.2 degrees.

[0104] Repeat the above steps until there is a satellite-derived temperature sample matching each sounding observation temperature sample.

[0105] Specifically, according to the quality mark and cloud condition information of the satellite-retrieved atmospheric temperature profile, the empirical cumulative distribution functions of the satellite-retrieved temperature samples and the sounding-observed temperature samples at different altitudes in the sample pairs under the same cloud condition information and quality mark are respectively counted, including:

[0106] All sample pairs are divided into corresponding sub-sample sets according to cloud condition information and quality marks; the types of cloud condition information include cloud and no cloud; the types of quality marks include 0 and 1;

[0107] The temperature probability density of satellite inversion temperature samples at each altitude layer in the sub-sample set is counted, and the satellite cumulative probability distribution of each temperature is calculated to obtain the empirical cumulative distribution function of discrete satellite inversion temperature at different altitude layers;

[0108] The temperature probability density of the sounding observation temperature samples at each altitude layer in the sub-sample set is statistically analyzed, and the sounding cumulative probability distribution of each temperature is calculated to obtain the empirical cumulative distribution function of discrete sounding observation temperature at different altitude layers.

[0109] When making statistics, no spatial window in the horizontal direction is defined. Instead, all the sounding observation temperature samples and satellite inversion temperature samples at each altitude layer are counted.

[0110] If the total number of satellite-inverted temperature samples and sounding observation temperature samples for a certain altitude layer is less than 300, it is considered that there are too few valid data for the current altitude layer, and the satellite-inverted temperature for that altitude layer will not be corrected.

[0111] Before making corrections, it is necessary to collect sounding data and FY-4B / GIIRS atmospheric temperature profile products, and obtain sample pairs through preprocessing. These samples can not only be used to verify the inversion accuracy of FY-4B / GIIRS atmospheric temperature profile products and analyze their systematic errors, but also be used to calculate the cumulative distribution functions of the two, providing a basis for subsequent PDF method corrections. The data preprocessing process is as follows: Figure 2 shown.

[0112] The temporal and spatial resolutions of sounding data and satellite data are inconsistent, so it is necessary to match the sounding data and satellite data in time and space. Before matching the sounding data and satellite observations, the sounding data must also be quality controlled, including the removal of duplicate pressure layer records, abnormal value inspection, pressure layer sorting, etc. The sounding data after quality control will be used as the true value for inspection and correction.

[0113] Since many sounding stations have many missing records above 100 hPa, the present invention selects the altitude of 100~1000 hPa for research. In order to ensure accurate comparison of the two in the vertical layer, the sounding is interpolated according to the 101-layer fixed isobaric surface of the inversion product. According to the average ascent speed of the sounding balloon of about 6 m / s, it takes about 50 min (minutes) for the balloon to rise to 100 hPa. Considering the time resolution of the two observations, satellite data within 60 min of the sounding observation time are selected for comparison. Spatial matching is performed based on the nearest neighbor principle, and the distance between each sounding profile and the surrounding satellite inversion profile products is calculated in turn. The profile product closest to the sounding profile is selected and paired with it as the sample observation value and inversion value. The pixel resolution of FY-4B / GIIRS is 12×12 km. Referring to previous studies, the present invention eliminates matching samples where the sounding and inversion products are more than 0.2° apart, thereby reducing the error caused by spatial matching. Data were collected from February 2023 to January 2024, and a total of 61,565 profile samples were obtained after the above processing, including 21,253 clear sky samples and 40,312 cloudy conditions.

[0114] In order to quantitatively analyze the accuracy of FY-4B / GIIRS temperature profiles before and after correction, the statistical parameters mean error MB and root mean square error RMSE are introduced:

[0115] (1)

[0116] (2)

[0117] in, x i is the ith sounding observation value, y i is the i-th satellite inversion value (i.e., temperature in the present invention), N is the total number of samples in each pressure layer, and i is the sample ordinal number.

[0118] In addition, in order to count the vertical error characteristics of the FY-4B / GIIRS temperature profile, the normalized contoured frequency by altitude diagrams (NCFAD) are introduced. The specific method is to first subtract the satellite from the sounding to obtain the error value and count the number of samples in each layer, calculate the frequency of the error value in each value range and each altitude layer, normalize the frequency based on the total number of samples in the corresponding pressure layer (to avoid the influence of the difference in the number of samples), and obtain the percentage of the value range in the total number of samples in the pressure layer. Finally, use a color code to show the frequency of occurrence of different error values ​​in all layers.

[0119] PDF method correction process:

[0120] It can be seen from the principle that the PDF method is based on a stable cumulative probability distribution, and for this purpose, a large number of historical samples need to be counted. The sounding data sites as the true value are sparse and the observation frequency is low. When performing statistics, the present invention no longer defines a spatial window in the horizontal direction, but instead performs statistics on all sounding site data at each altitude layer. The time window is consistent with the inspection and evaluation, which is from February 2023 to January 2024. In order to achieve a statistically stable probability density distribution, the number of selected samples is at least 300. If the requirements are not met, it is considered that there are too few valid data at the current altitude layer and no correction is made. Analysis shows that the probability density distribution of temperature samples at different altitudes is not consistent. Some are close to normal distribution, some are close to Gamma distribution, and some are difficult to express with commonly used distribution functions. Therefore, the present invention does not adopt the method of fitting a fixed probability density function, but uses an empirical cumulative distribution function to describe the distribution probability of temperature samples.

[0121] After statistically analyzing the empirical cumulative distribution functions of different altitude layers based on the quality mark, an empirical cumulative distribution lookup table with a resolution of 0.1 degree, covering the product temperature detection range and applicable to the corresponding altitude layer is generated through interpolation.

[0122] According to the quality mark and cloud condition information of the satellite-retrieved atmospheric temperature profile, respectively statistically analyzing the empirical cumulative distribution functions of the satellite-retrieved temperature samples and the sounding-observed temperature samples at different altitudes under the same cloud condition information and quality mark in the sample pair, and generating an empirical cumulative distribution lookup table of the corresponding altitudes of the satellite-retrieved temperature samples and the sounding-observed temperature samples with a temperature range of 200-340K and a resolution of 0.1K by interpolation;

[0123] When correcting, firstly, according to the quality mark, the satellite inversion temperature of each sample is obtained from the corresponding relationship table of satellite inversion temperature and satellite cumulative probability distribution at each altitude layer. T satellite Cumulative probability distribution of satellites fsatellite (T satellite ) , find the sounding cumulative probability distribution that is closest to the satellite cumulative probability distribution in the corresponding relationship table between the sounding cumulative probability distribution and the sounding observed temperature at each altitude layer f sounding (T sounding ) The corresponding sounding temperature T sounding ,make f sounding (T sounding )≈f satellite (T satellite ) , then at this time T sounding Corrected value for temperature.

[0124] For example, the inverted temperature of a sample satellite with a mass label of 1 is 270 K. The corresponding cumulative probability distribution in the lookup table based on the mass 1 sample statistics is 25%, while the observed temperature corresponding to the cumulative probability distribution of 25% in the lookup table based on the sounding sample statistics is 271 K. The corrected temperature value is 271 K.

[0125] At present, the accuracy studies on GIIRS atmospheric temperature profile products have shown that there are systematic errors that vary with altitude, but there are few reports on how to further deal with these errors, which is the key to improving the credibility and practicality of the products. Systematic errors can lead to negative effects of profile products in actual business applications, such as incorrect estimation of strong convective potential and failure to meet the unbiased assumption required by variational assimilation. Therefore, correcting the systematic errors of profile products can better play the role of the products.

[0126] Before correcting the GIIRS atmospheric temperature profile product, we must first understand the error characteristics of the product itself, so that we can make different treatments for the error characteristics under different conditions. Therefore, this section will first test the FY-4B / GIIRS temperature profile product based on sounding data, analyze the vertical distribution characteristics of errors under different conditions, and then use the PDF method to correct products with obvious systematic errors, and use a time-independent validation set to evaluate the correction effect of the PDF method.

[0127] FY-4B / GIIRS temperature profile product inspection:

[0128] Error characteristics of different cloud conditions

[0129] The data with quality marks of 0 and 1 in the inversion data are selected as valid observations, and based on the Laida criterion, the outlier samples whose difference from the mean value exceeds 3 times of the standard deviation in each layer are eliminated. Figure 3 Figure 1 shows the error NCFAD of the FY-4B / GIIRS temperature profile under clear sky (a) and cloudy (b) conditions, as well as the vertical distribution of MB and RMSE (c). The error frequency, MB and vertical distribution of RMSE of the FY-4B / GIIRS temperature profile under clear sky and cloudy conditions are statistically analyzed. Comparing the error NCFAD under clear sky and cloudy conditions, it can be seen that the errors under clear sky conditions are concentrated around 0 and are distributed more closely ( Figure 3 (a)), while the temperature error increases significantly when there are clouds, which is obviously colder, and the error distribution is more discrete ( Figure 3 (b)). Further analysis of MB and RMSE shows that the temperature MB is within -0.3~1 K under clear sky conditions, negative in the 950~300 hPa altitude layer, positive in the upper troposphere and near the ground layer, and decreases first and then increases with increasing altitude. RMSE is below 2 K at most altitudes, with a minimum of about 1.1 K near 400 hPa. Under cloudy conditions, the temperature profile shows an overall positive deviation, MB is around 0 K at the bottom layer, increases first and then decreases with increasing altitude, and the maximum value is located at 150~450hPa, close to 1.7 K, and RMSE is 2.2~2.7 K. The variation in the troposphere is not large, with the minimum and maximum values ​​at 900 and 170 hPa, respectively ( Figure 3 (c)).

[0130] Based on the above analysis, it can be seen that the quality of FY-4B / GIIRS profile is better under clear sky conditions. Under non-clear sky background, the cloud layer changes greatly in the vertical direction. The precipitation or cloud particles in it will absorb and emit infrared radiation, making it difficult to invert the temperature profile through infrared radiation. In addition, it can be found that the error distribution in the near-surface layer is more discrete than that at other altitudes. On the one hand, this is related to the distribution of the weight function of the GIIRS detection channel. From the Jacobian point of view, the GIIRS has a weak ability to detect the temperature of the lower atmosphere. On the other hand, it is related to the underlying surface. According to the statistics of the effective samples in the altitude layer, the number of samples decreases rapidly when approaching the ground (especially under cloudy conditions), lacking statistical significance, which will also lead to a higher discreteness of the near-surface error.

[0131] Error characteristics of different mass markers:

[0132] The above analysis shows the impact of cloudy conditions on the FY-4B / GIIRS temperature profile detection capability, and further classifies and statistics the error characteristics of samples of different quality under cloudy conditions. Figure 4 The error NCFAD, MB, and RMSE for the temperature samples with quality 0 and 1 are shown by Figure 4It can be seen that the error distribution and discreteness of samples with masses of 0 and 1 are different. When the mass is 0, there is a systematic error of 0~1 K in the temperature, which is concentrated around the average deviation value and is normally distributed. The RMSE fluctuates slightly between 1.7 and 2.3 K. When the mass is 1, the systematic error increases significantly, and the maximum MB reaches 2.8 K. Although the partial error distribution is tighter than that of mass 0, it presents an asymmetric bimodal distribution below 600 hPa and above 400 hPa. The RMSE change trend is similar but increases to 2.9~3.4 K.

[0133] Figure 4 Schematic diagram of the error NCFAD of the FY-4B / GIIRS temperature profile with mass 0 (a) and 1 (b), as well as the vertical distribution of MB and RMSE (c). Comparative analysis shows that there are systematic errors in the FY-4B / GIIRS temperature profile when there are clouds. Compared with the sample with mass 0, the systematic error of the sample with mass 1 is more significant, it is colder than the sounding, and the error distribution is more discrete. For the temperature profile, the error does not even obey the normal distribution at some altitudes. Different systematic errors lead to different error change characteristics.

[0134] FY-4B / GIIRS temperature profile correction:

[0135] In order to objectively verify the temporal generalization ability of the model, in addition to using the profile samples from February 2023 to January 2024 as a training set to establish the empirical cumulative distribution lookup table, data from March to May 2024 were also collected and processed to obtain 8461 cloud profile samples as a validation set to evaluate the correction effect of the PDF method.

[0136] Figure 5 Schematic diagram of the error NCFAD and vertical distribution of MB and RMSE (c) of the FY-4B / GIIRS temperature profile with mass 0 before (a) and after (b) correction; Figure 6 Schematic diagram of the error NCFAD and vertical distribution of MB and RMSE (c) of the FY-4B / GIIRS temperature profile with a mass of 1 before (a) and after (b) correction; Figure 5 and Figure 6 The vertical characteristics of the error of the mass 0 and 1 temperature samples in the validation set before and after the PDF method correction. Figure 5 It can be seen that for the mass 0 sample, the error before correction is normally distributed as a whole, and the systematic error gradually increases with altitude. Due to the small number of samples in the validation set, the error distribution is relatively discrete. After correction using the PDF method, the error distribution is near 0, the MB decreases with altitude, and the RMSE also decreases in the middle and upper atmospheres, and the systematic error is basically eliminated. For the mass 1 sample, the error before correction is unevenly bimodal ( Figure 6), the two peaks are located at -3 K and 3 K respectively (hereinafter referred to as the left peak and the right peak), the bimodal distribution is more obvious below 100 hPa and 700 hPa, the proportion of the left peak in other altitude layers is small, and the left peak near 500 hPa basically disappears and only the right peak remains. After correction using the PDF method, the error is close to 0 as a whole, but the situation of the bimodal distribution has not changed. The right peak where the 500 hPa error is located is reduced to around 0.5 K, while the value of the right peak of the error in the upper and lower altitude layers increases with the increase of the proportion of the left peak. At the altitude below 700 hPa where the bimodal distribution is more obvious, the value of the right peak gradually returns to around 2 K. Corresponding to NCFAD, MB and RMSE do not change significantly in the near-surface layer of the bimodal distribution, but with the increase of altitude, the proportion of the left peak of the error decreases, gradually approaching the unimodal normal distribution, MB and RMSE are greatly reduced, and the systematic error is partially eliminated.

[0137] Figure 7 The scatter point distribution of the temperature samples of mass 0 and 1 in the validation set before and after correction, where (a) is a schematic diagram of the temperature profile scatter point when the mass is 0 before correction, (b) is a schematic diagram of the temperature profile scatter point when the mass is 0 after correction, (c) is a schematic diagram of the temperature profile scatter point when the mass is 1 before correction, and (d) is a schematic diagram of the temperature profile scatter point when the mass is 1 after correction. The color represents the normalized frequency, and the overall MB, RMSE and correlation coefficient (R) of the sample are also given. Compared with before correction, the MB of the mass 0 and 1 temperature samples after correction decreased from 0.7365 K and 2.0728 K to -0.0320 K and 0.0115 K, respectively, the RMSE decreased from 1.8854 K and 3.2043 K to 1.7279 K and 2.3371 K, respectively, and the R increased slightly from 0.9975 and 0.9948 to 0.9976 and 0.9952, respectively, indicating that the PDF method correction of this application has an overall positive effect.

[0138] Based on the above analysis, the PDF method can change the range of error distribution, making it tend to be unbiased, thereby eliminating systematic errors. However, it is difficult to play a role when the overall error distribution is already unbiased (mean is 0). This explains the correction results of mass 0 and 1 samples: for the mass 0 sample of the normal distribution, the positive deviation increases with height, and after correction, each altitude layer is unbiased; for the mass 1 sample of the bimodal distribution, the PDF method can also make some improvements in the altitude layer where the left peak accounts for a small proportion (that is, biased), making the right peak tend to 0, but when the proportion of the left peak gradually increases to the same as the right peak and the overall error distribution tends to be unbiased, the influence of the PDF method gradually becomes invalid. In addition, the PDF method cannot change the discreteness of the error, because this parameter is determined by the detection accuracy and the accuracy of the inversion algorithm, reflecting the random error of the inversion product rather than the systematic error. This is reflected in the error evaluation as follows: for altitude layers where MB is not 0, the PDF method can reduce RMSE by making MB tend to 0, but when MB is already close to 0, RMSE cannot be effectively reduced. The improvement comes from the elimination of systematic errors rather than the reduction of random errors.

[0139] The present invention adopts the PDF method to realize the correction of the satellite-inverted atmospheric temperature profile, but the available methods are not limited to this, and can also be machine learning methods represented by random forests, neural networks, etc. In terms of application scenarios, this method can be used not only for the correction of temperature profiles, but also for the correction of humidity profiles, and can be used not only in cloudy conditions but also in sunny conditions. The satellite to which the atmospheric temperature profile product belongs can be a polar-orbiting satellite or a geostationary satellite, and the detection instrument can be an infrared hyperspectral detector or a microwave vertical detector.

[0140] The key innovation of the present invention is: to verify the satellite inverted atmospheric products based on sounding data, analyze the vertical distribution characteristics of errors under different conditions, and on this basis, statistically calculate the cumulative probability density of the two, and use the PDF method to correct products with obvious systematic errors.

[0141] Based on the same technical concept, the method of the present invention can be extended to the correction of satellite humidity profile products. Satellite inversion humidity samples and sounding observation humidity samples are processed based on the same method to obtain an empirical cumulative distribution lookup table of humidity, and the satellite humidity profile product is corrected based on the empirical cumulative distribution lookup table of humidity.

[0142] Embodiment 2 provided by the present invention:

[0143] This embodiment provides a satellite temperature profile product correction device, including:

[0144] Acquisition module: used to obtain the satellite-retrieved atmospheric temperature profile, its quality mark and cloud condition information of satellite temperature profile products;

[0145] Table lookup module: used to select a corresponding subtable in a pre-constructed empirical cumulative distribution lookup table according to the quality mark and cloud condition information of the satellite-inverted atmospheric temperature profile; the empirical cumulative distribution lookup table includes a plurality of subtables corresponding to the quality mark and cloud condition information of the satellite-inverted atmospheric temperature profile; the subtable records the corresponding relationship between the satellite-inverted temperature and the satellite cumulative probability distribution at each altitude layer under the corresponding quality mark and cloud condition information, and the corresponding relationship between the sounding cumulative probability distribution and the sounding observed temperature;

[0146] The sounding retrieval module is used to obtain the satellite inversion temperature to be corrected for each altitude layer according to the satellite inversion atmospheric temperature profile, obtain the satellite cumulative probability distribution corresponding to the satellite inversion temperature to be corrected for each altitude layer based on the sub-table according to the satellite inversion temperature to be corrected, and obtain the sounding observation temperature corresponding to the sounding cumulative probability distribution that is the same as the satellite cumulative probability distribution for each altitude layer based on the satellite cumulative probability distribution;

[0147] Correction module: used to use the sounding observation temperature as the temperature correction value of the corresponding satellite inversion temperature to be corrected, and correct the satellite inverted atmospheric temperature profile.

[0148] The device of this embodiment can be used to execute the steps of the method described in Example 1.

[0149] Embodiment 3 provided by the present invention:

[0150] This embodiment provides a satellite temperature profile product correction device, including a processor and a storage medium;

[0151] The storage medium is used to store instructions;

[0152] The processor is used to operate according to the instructions to execute the steps of the method described in Example 1.

[0153] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0154] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0155] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0156] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0157] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A satellite temperature profile product correction method. It is characterized in that The following steps are involved: Obtain satellite-derived atmospheric temperature profiles and their quality marks and cloud condition information from satellite temperature profile products; According to the quality mark and cloud condition information of the satellite-retrieved atmospheric temperature profile, the corresponding subtable is selected in the pre-built empirical cumulative distribution lookup table; The empirical cumulative distribution lookup table includes a plurality of sub-tables corresponding to the quality marks and cloud condition information of the satellite-inverted atmospheric temperature profile; the sub-tables record the corresponding relationship between the satellite-inverted temperature and the satellite cumulative probability distribution at each altitude layer under the corresponding quality marks and cloud condition information, and the corresponding relationship between the sounding cumulative probability distribution and the sounding observed temperature; According to the satellite inverted atmospheric temperature profile, the satellite inverted temperature to be corrected of each altitude layer is obtained; according to the satellite inverted temperature to be corrected, based on the sub-table, the satellite cumulative probability distribution corresponding to the satellite inverted temperature to be corrected of each altitude layer is obtained; and based on the satellite cumulative probability distribution, the sounding observation temperature corresponding to the sounding cumulative probability distribution closest to the satellite cumulative probability distribution of each altitude layer is obtained; The sounding observed temperature is used as the temperature correction value of the corresponding satellite inverted temperature to be corrected, and the satellite inverted atmospheric temperature profile is corrected.

2. The satellite temperature profile product correction method according to claim 1, It is characterized in that The method for pre-constructing the empirical cumulative distribution lookup table includes: Obtain satellite inverted temperature samples and sounding observed temperature samples based on satellite temperature profile products and sounding data; Preprocessing the satellite inverted temperature samples and the sounding observed temperature samples to obtain a sample pair; the sample pair includes the satellite inverted temperature samples and the sounding observed temperature samples that are matched simultaneously in time and space; According to the quality mark and cloud condition information of the satellite-retrieved atmospheric temperature profile, respectively statistically analyze the empirical cumulative distribution functions of the satellite-retrieved temperature samples and the sounding-observed temperature samples at different altitudes under the same cloud condition information and quality mark in the sample pair, and generate an empirical cumulative distribution lookup table of the corresponding altitudes of the satellite-retrieved temperature samples and the sounding-observed temperature samples with a temperature range of 200-340 K and a resolution of 0.1 K by interpolation; The empirical cumulative distribution lookup table is divided into multiple sub-tables according to the quality mark and cloud condition information of the satellite-inverted atmospheric temperature profile; the sub-table records the correspondence between the satellite-inverted temperature and the satellite cumulative probability distribution of each altitude layer under the corresponding quality mark and cloud condition information, as well as the correspondence between the sounding cumulative probability distribution and the sounding observed temperature.

3. The satellite temperature profile product correction method according to claim 2, It is characterized in that The satellite inversion temperature samples and the sounding observation temperature samples are preprocessed to obtain sample pairs, including: Obtain satellite-derived temperature samples and sounding-observed temperature samples at altitudes of 100-1000 hPa; Perform quality control on the sounding observation temperature samples, including removing duplicate pressure layer records, checking for abnormal values ​​and sorting the pressure layers, to obtain the sounding observation temperature samples after quality control; The sounding observation temperature samples after quality control are interpolated according to the fixed isobaric surface of the altitude layer of the satellite inversion temperature samples; The interpolated sounding observation temperature samples and satellite inversion temperature samples are matched in time and space to obtain sample pairs including satellite inversion temperature samples and sounding observation temperature samples that are matched simultaneously in time and space.

4. The satellite temperature profile product correction method according to claim 3, It is characterized in that The sounding observation temperature samples and satellite inversion temperature samples after quality control and interpolation are matched in time and space, including: Select the satellite inversion temperature sample within 60 minutes from the sounding observation time of each sounding observation temperature sample to obtain the time matching inversion sample group corresponding to the sounding observation temperature sample; the sounding observation time is the take-off time of the sounding balloon; In the time-matched inversion sample group, spatial matching is performed based on the nearest neighbor principle, and the proximity distance between each sounding site of the sounding observation temperature sample and the satellite observation point of the surrounding corresponding satellite inversion temperature sample is calculated in turn; If the proximity distance between the sounding site of the sounding observation temperature sample and the nearest satellite observation point is less than the preset distance value, the satellite inversion temperature sample corresponding to the satellite observation point is successfully matched with the sounding observation temperature sample, and the sounding observation temperature sample will not continue to be matched; If the proximity distance between the sounding site of the sounding observation temperature sample and the nearest satellite observation point is greater than or equal to the preset distance value, the sounding observation temperature sample is discarded; Repeat the above steps until there is a satellite-derived temperature sample matching each sounding observation temperature sample.

5. The satellite temperature profile product correction method according to claim 4, It is characterized in that The calculation method for calculating the proximity distance between each sounding station and the surrounding satellite profile products includes: according to the longitude and latitude, the square of the longitude difference between the sounding station and the satellite observation point is added to the square of the latitude difference and then the root is taken; The preset distance value is 0.2 degrees.

6. The satellite temperature profile product correction method according to claim 2, It is characterized in that According to the quality mark and cloud condition information of the satellite-retrieved atmospheric temperature profile, the empirical cumulative distribution functions of the satellite-retrieved temperature samples and the sounding-observed temperature samples at different altitudes in the sample pairs under the same cloud condition information and quality mark are respectively counted, including: All sample pairs are divided into corresponding sub-sample sets according to cloud condition information and quality marks; the types of cloud condition information include cloud and no cloud; the types of quality marks include 0 and 1; The temperature probability density of satellite inversion temperature samples at each altitude layer in the sub-sample set is counted, and the satellite cumulative probability distribution of each temperature is calculated to obtain the empirical cumulative distribution function of discrete satellite inversion temperature at different altitude layers; The temperature probability density of the sounding observation temperature samples at each altitude layer in the sub-sample set is statistically analyzed, and the sounding cumulative probability distribution of each temperature is calculated to obtain the empirical cumulative distribution function of discrete sounding observation temperature at different altitude layers.

7. The satellite temperature profile product correction method according to claim 6, It is characterized in that The empirical cumulative distribution functions of the satellite inversion temperature samples and the sounding observation temperature samples at different altitudes under the same cloud condition information and quality mark in the sample pairs are respectively counted, and also include: When making statistics, no spatial window in the horizontal direction is defined. Instead, all the sounding observation temperature samples and satellite inversion temperature samples at each altitude layer are counted. If the total number of satellite-inverted temperature samples and sounding observation temperature samples for a certain altitude layer is less than 300, it is considered that there are too few valid data for the current altitude layer, and the satellite-inverted temperature for that altitude layer will not be corrected.

8. The satellite temperature profile product correction method according to claim 1, It is characterized in that The satellite temperature profile product is a regional synthetic product of FY-4B / GIIRS atmospheric vertical detection from the official website of the National Satellite Meteorological Center; The sounding data are from the meteorological information comprehensive processing system.

9. A satellite temperature profile product correction device, It is characterized in that include: Acquisition module: used to obtain the satellite-retrieved atmospheric temperature profile, its quality mark and cloud condition information of satellite temperature profile products; Table lookup module: used to select the corresponding subtable in the pre-built empirical cumulative distribution lookup table according to the quality mark and cloud condition information of the satellite-retrieved atmospheric temperature profile; The empirical cumulative distribution lookup table includes a plurality of sub-tables corresponding to the quality marks and cloud condition information of the satellite-inverted atmospheric temperature profile; the sub-tables record the corresponding relationship between the satellite-inverted temperature and the satellite cumulative probability distribution at each altitude layer under the corresponding quality marks and cloud condition information, and the corresponding relationship between the sounding cumulative probability distribution and the sounding observed temperature; The sounding retrieval module is used to obtain the satellite inversion temperature to be corrected for each altitude layer according to the satellite inversion atmospheric temperature profile, obtain the satellite cumulative probability distribution corresponding to the satellite inversion temperature to be corrected for each altitude layer based on the sub-table according to the satellite inversion temperature to be corrected, and obtain the sounding observation temperature corresponding to the sounding cumulative probability distribution that is the same as the satellite cumulative probability distribution for each altitude layer based on the satellite cumulative probability distribution; Correction module: used to use the sounding observed temperature as the correction value of the corresponding satellite inverted temperature to be corrected, and correct the satellite inverted atmospheric temperature profile.

10. A satellite temperature profile product correction device, It is characterized in that including processor and storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method according to any one of claims 1 to 8.