Long sequence polar orbit satellite infrared sea surface temperature product reprocessing method and system

Abstract

The application provides a long-sequence polar-orbit satellite infrared sea surface temperature product reprocessing method and system, and relates to the technical field of satellite remote sensing, wherein the method comprises the following steps: S1, sea surface temperature matching, establishing reprocessing sea surface temperature matching data set; S2, sea surface temperature regression, calculating reprocessing sea surface temperature regression coefficient; S3, sea surface temperature inversion, performing per-pixel quality control; S4, sea surface temperature daily synthesis; S5, sea surface temperature monthly synthesis. The application performs sea surface temperature inversion in an iterative manner, replaces the business cloud detection product with the inversion sea surface temperature quality level to inhibit cloud pollution; replaces the relatively fixed regression coefficient with the monthly sea surface temperature regression coefficient to take the daily analysis field sea surface temperature as a background field to perform sea surface temperature quality control, and gives the per-pixel sea surface temperature quality level, so that the precision and consistency of long-time sequence sea surface temperature inversion can be improved; the sea surface temperature daily synthesis is performed according to the principle of quality priority, and a blacklist is established for monthly synthesis, so that the precision of daily / monthly synthesis sea surface temperature can be improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of satellite remote sensing applications, and in particular to a long-sequence polar-orbiting satellite infrared sea surface temperature product reprocessing method and system. BACKGROUND

[0002] Sea surface temperature is an important input parameter for weather forecasting, climate monitoring, diagnosis and prediction, and climate numerical simulation research. The Fengyun meteorological satellite has accumulated visible and infrared scanning radiometer (VIRR) observation data since the launch of the FY-1C satellite on May 10, 1999. The long-wave infrared channel of the polar-orbiting satellite, i.e. the split window channel (10.3-11.3 μm and 11.5-12.5 μm), can be used to estimate sea surface temperature (hereinafter referred to as sea surface temperature). However, due to the limitations of early Fengyun satellite instrument calibration, positioning and sea surface temperature retrieval technology, the Fengyun satellite sea surface temperature product has poor precision and cannot meet the application needs of users.

[0003] The Fengyun-3 (FY-3) meteorological satellite is the second generation of Fengyun polar-orbiting meteorological satellites in China. FY-3A and FY-3B are two test stars of FY-3 (01) batch, which were launched on May 27, 2008 and November 5, 2010, respectively. FY-3C was launched on September 23, 2013, and is the first operational star of FY-3 (02) batch. The National Satellite Meteorological Center has been providing global Fengyun satellite sea surface temperature products to users since May 2014. The Fengyun satellite sea surface temperature has undergone a series of algorithm improvements. The FY-3A / VIRR operational sea surface temperature product uses a multi-channel sea surface temperature (MCSST) retrieval algorithm in different latitude bands during the day. The FY-3B / VIRR used the MCSST retrieval algorithm during the day before 2017, and the nonlinear sea surface temperature (NLSST) retrieval algorithm after 2018, with the first guess sea surface temperature being the 30-year monthly average sea surface temperature. The FY-3C / VIRR sea surface temperature product uses the MCSST retrieval algorithm.

[0004] FY3(01) batch product generation system (PGS) product is limited by the network transmission at that time, the storage mode is designed for 10°x10° block storage, and the global 1 km equal latitude and longitude daily and monthly synthetic sea surface temperature is divided into 648 blocks (the user needs to splice before application to obtain the global sea surface temperature, and the product is not easy to use). Affected by cloud pollution, the sea surface temperature product error is large. Before 2013, the FY-3A / VIRR sea surface temperature inversion deviation is-1.38℃, and the root mean square error is 3.8℃. After 2013, although the algorithm is improved, the sea surface temperature inversion precision is improved, the precision is improved to deviation-0.23℃, and the root mean square error is 1.74℃, but it is still difficult to apply to the monitoring of marine ENSO (El The FY3(02) batch PGS sea surface temperature is improved in design, adds the per-pixel sea surface temperature quality identifier, generates the global 5km equal latitude and longitude sea surface temperature product, and the precision and easy-to-use of the business sea surface temperature product are improved. However, due to the influence of satellite life period calibration update, sea surface temperature regression coefficient update (three sets of sea surface temperature regression coefficients are used), black body temperature fluctuation, satellite battery failure and other satellite working conditions, the precision and consistency of the FY-3C / VIRR business sea surface temperature product are poor. Compared with the analysis field sea surface temperature, the root mean square error is between 0.8 and 1.2℃, the sea surface temperature deviation fluctuates seasonally, which is not conducive to climate application. The business sea surface temperature can only use the sea surface temperature matching sample of the previous months to calculate the sea surface temperature regression coefficient, and the relative fixed sea surface temperature regression coefficient, and the error of the business sea surface temperature with the relative fixed sea surface temperature regression coefficient has seasonal fluctuations, which cannot meet the application requirements of climate users.

[0005] Affected by the business flow operation of the Fengyun satellite ground application system, the change of satellite working conditions, the update of calibration coefficient version, the update of sea surface temperature regression coefficient, the lag of business sea surface temperature regression coefficient behind the generation of business sea surface temperature product, and the difference of background field sea surface temperature, although the Fengyun business sea surface temperature product has a time length of 9 years, its precision and consistency of long sequence sea surface temperature product cannot meet the application requirements of climate research and monitoring. Therefore, it is necessary to improve the Fengyun satellite sea surface temperature inversion algorithm and quality control method, reprocess the long sequence Fengyun satellite sea surface temperature product, improve the precision and consistency of long sequence sea surface temperature product, and fully play the application value of long sequence historical data of Fengyun satellite. SUMMARY

[0006] In order to overcome the deficiencies that the precision of the existing wind and cloud business sea temperature product and the consistency of the long sequence sea temperature product cannot meet the application requirements, the embodiment of the present application provides a long sequence polar orbit satellite infrared sea temperature product reprocessing method and system, which iteratively performs sea temperature inversion, replaces the business cloud detection product with the sea temperature quality level, suppresses cloud pollution, replaces the relatively fixed regression coefficient with the sea temperature regression coefficient of the current month, takes the sea temperature of the current day as the background field to perform sea temperature quality control, and assigns the per-pixel sea temperature quality level, so as to improve the precision and consistency of the long time sequence sea temperature inversion, perform daily synthesis of the sea temperature according to the quality priority principle, and establish a blacklist for monthly synthesis, which can improve the precision of the daily / monthly synthesized sea temperature.

[0007] The technical solution adopted by the embodiment of the present application to solve the technical problems is:

[0008] The long sequence polar orbit satellite infrared sea temperature product reprocessing method specifically includes the following steps:

[0009] Step S1, sea temperature matching, calculating the monthly nonlinear sea temperature (NLSST) regression coefficient V0 by using the business sea temperature matching data set, performing sea temperature inversion to generate NLSST V0, and performing sea temperature matching based on the satellite observation data processed by positioning and scaling, NLSST V0, field sea temperature and daily analysis field sea temperature, to establish a monthly reprocessed sea temperature matching data set;

[0010] Step S2, sea temperature regression, based on the monthly reprocessed sea temperature matching data set, performing quadratic least squares regression calculation to obtain the sea temperature regression coefficient V1;

[0011] Step S3, sea temperature inversion, performing original orbit sea temperature inversion by using the sea temperature regression coefficient V1 of the current month, taking the daily analysis field sea temperature as the background field to perform per-pixel sea temperature quality control, and assigning the per-pixel sea temperature quality level;

[0012] Step S4, daily synthesis of sea temperature, performing daily synthesis according to the quality priority principle based on the original orbit sea temperature and the per-pixel sea temperature quality level within a day, taking the daily analysis field sea temperature as the background field to perform sea temperature quality control, and assigning the per-pixel sea temperature quality level;

[0013] Step S5, monthly synthesis of sea temperature, performing monthly synthesis according to the quality priority principle based on the daily synthesized sea temperature and the per-pixel sea temperature quality level within a month, taking the monthly analysis field sea temperature as the background field to perform sea temperature quality control, and assigning the per-pixel sea temperature quality level.

[0014] Preferably, in the step S1, FY-3C / VIRR LIB data, NLSST V0, in-situ sea temperature and daily analysis sea temperature are taken as input, matching is performed in a matching window with a spatial resolution of 3 km and a time resolution of 1 hour, and a matching sample is obtained from a pixel meeting the matching condition in a clear sky sea area.

[0015] Preferably, in the step S2, least square regression is performed according to a sea temperature regression formula on a monthly basis, a first guessed sea temperature regression coefficient is calculated, the regression coefficient is substituted into the sea temperature matching data set, the inverted sea temperature is calculated, a median and a standard deviation of the matching sample are obtained, the matching sample whose difference between the inverted sea temperature and the buoy sea temperature exceeds the median ± 2 standard deviations is removed, secondary least square regression is performed, and a monthly sea temperature regression coefficient is calculated and used for subsequent sea temperature reprocessing inversion.

[0016] Preferably, in the step S3, the sea temperature inversion specifically includes the following steps:

[0017] Step S31, reading in satellite observation data after positioning and scaling, daily analysis sea temperature and monthly sea temperature regression coefficient, identifying the read-in satellite observation data and corresponding land and sea template data to obtain marine observation data, calculating split window channel brightness temperature T11 and T12, and assigning -999 to land pixels;

[0018] Step S32, performing threshold test on T11 and T12 of marine pixels, and entering step S33 for the pixels meeting the brightness temperature threshold, and assigning a default value -888 to the pixels exceeding the brightness temperature threshold and entering step S37;

[0019] Step S33, constructing a 3*3 data block with the current marine pixel as the center, performing spatial consistency test, assigning a default value -888 to the pixels not meeting the spatial consistency and entering step S37, and entering step S34 for the pixels meeting the spatial consistency;

[0020] Step S34, performing bilinear interpolation on the daily analysis sea temperature according to the latitude and longitude of the center pixel of the 3*3 data block to obtain a first guessed sea temperature, performing sea temperature effective value test, entering step S35 for the pixels not passing the test, and entering step S36 for the pixels passing the test;

[0021] Step S35, calculating the inverted sea temperature according to the MCSST regression formula as the first guessed sea temperature;

[0022] Step S36, calculating the inverted sea temperature according to the NLSST regression formula, performing sea temperature quality control, and assigning a pixel-by-pixel sea temperature quality grade;

[0023] Step S37, write the sea surface temperature scientific data set and the per-pixel sea surface temperature quality identification into a sea surface temperature data file;

[0024] The MCSST regression formula is: T s =a0+a1T 11 +a2(T 11 -T 12 )+a3(T 11 -T 12 )(secθ-1)

[0025] The NLSST regression formula is: T s =a0+a1T 11 +a2T FG (T 11 -T 12 )+a3(T 11 -T 12 )(secθ-1)

[0026] Wherein, T S represents the inverted sea surface temperature, T 11 and T 12 represent the 10.8 μm and 12 μm channel brightness temperatures respectively, a0-a3 represent the regression coefficients, θ represents the satellite zenith angle, and T FG represents the first guessed sea surface temperature.

[0027] Preferably, the sea surface temperature quality control specifically includes 7 tests, which are: (1) sea-land template test; (2) satellite zenith angle threshold test; (3) spatial consistency test; (4) background field sea surface temperature threshold test; (5) solar flare region test; (6) sea ice test; and (7) inverted sea surface temperature effective value range test.

[0028] Preferably, in the step S4, on the basis of about 288 5-minute segment sea surface temperature products per day, the original 1.1 km nadir orbit sea surface temperature products are respectively projected and converted into daytime and nighttime global 1 km equirectangular projection intermediate data according to the daytime and nighttime identification, on the basis of the global 1 km equirectangular projection intermediate data, a 5×5 data block is constructed, the 1 km resolution equirectangular projection data is down-sampled to global 5 km equirectangular projection with the quality first principle, and the daytime and nighttime are stored separately, and per-pixel quality control is performed, and the HDF5 format is stored.

[0029] Preferably, the quality priority principle is that in a 5*5 data block, if there is a pixel with quality of excellent, the excellent pixels are averaged to assign an excellent quality level; if there is no excellent pixel, the pixels with quality of good are averaged to assign a good quality level; if there is no good pixel, the pixels with quality of poor are averaged to assign a poor quality level; if there is no poor pixel, a default value is assigned, and on this basis, five types of tests are performed, which are: (1) sea-land template test; (2) sea temperature effective value range test; (3) satellite zenith angle threshold test; (4) background field sea temperature threshold test; and (5) sea ice test.

[0030] Preferably, in the step S5, a blacklist is established according to the quality test result of the daily synthetic sea temperature, and on the basis of 28-31 daily synthetic sea temperature data of each month, monthly synthesis is performed according to the day / night identifier by using the quality priority principle, and the monthly synthesis is stored in day / night and subjected to pixel-by-pixel quality control.

[0031] Preferably, the quality control principle is that in 28-31 daily synthetic sea temperature data, if there is a pixel with quality of excellent, the excellent pixels are averaged to assign an excellent quality level; if there is no excellent pixel, the pixels with quality of good are averaged to assign a good quality level; if there is no good pixel, the pixels with quality of poor are averaged to assign a poor quality level; if there is no poor pixel, a default value is assigned, and on this basis, four types of tests are performed, which are: (1) blacklist test; (2) sea-land template test; (3) sea temperature effective value range test; and (4) climate sea temperature threshold test.

[0032] The long-sequence polar-orbiting satellite infrared sea temperature product reprocessing system comprises a processor and a storage unit, the storage unit is used for storing a computer program, and the processor can realize the long-sequence polar-orbiting satellite infrared sea temperature product reprocessing method by executing the computer program.

[0033] The embodiment of the application has the following advantages:

[0034] 1. Compared with the business sea temperature inversion which uses a cloud detection product as input and adopts fixed sea temperature regression coefficients, the application does not involve the input of a cloud detection product, avoids introducing cloud pollution in the business cloud detection, uses daily analysis field sea temperature as the first guessed sea temperature, improves the sea temperature background field, uses the sea temperature regression coefficients of the current month to replace the relatively fixed regression coefficients of the business, and performs sea temperature quality control to improve the sea temperature inversion precision.

[0035] 2、In the sea temperature inversion, 7 kinds of sea temperature quality control are carried out, including sea-land template test, satellite zenith angle threshold test, spatial consistency test, background field sea temperature threshold test, solar flare area test, sea ice test and sea temperature effective value range test, 1 quality grade is given to each sea temperature pixel, the sea temperature inversion precision and the consistency of long sequence sea temperature are improved, and the inconsistency of long sequence sea temperature caused by factors such as business sea temperature regression coefficient lagging behind updating due to satellite working condition change and calibration coefficient updating is made up.

[0036] 3、In the sea temperature daily synthesis, 5 kinds of sea temperature quality control are carried out, including sea-land template test, sea temperature effective value range test, satellite zenith angle threshold test, background field sea temperature threshold test and sea ice test, 1 quality grade is given to each pixel, and the precision of daily synthesis sea temperature and the consistency of long sequence sea temperature are improved.

[0037] 4、In the sea temperature monthly synthesis, a blacklist is established according to the quality test results of daily synthesis sea temperature, the daily synthesis sea temperature on the blacklist is removed in monthly synthesis, so that it does not participate in sea temperature monthly synthesis, the reliability of monthly synthesis sea temperature is improved, and 4 kinds of sea temperature quality control are carried out, including blacklist test, sea-land template test, sea temperature effective value range test and climate sea temperature threshold test, 1 quality grade is given to each pixel, and the precision and reliability of sea temperature monthly synthesis sea temperature are improved. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 It is a flow chart of the long sequence polar orbit satellite infrared sea temperature product reprocessing method of the application;

[0039] Figure 2 It is a flow chart of the sea temperature inversion of the application;

[0040] Figure 3 It is an effect comparison chart of the sea-land template test of the application;

[0041] Figure 4 It is an effect comparison chart of the sea temperature daily synthesis of the application;

[0042] Figure 5 It is a sea temperature precision comparison chart of the sea temperature daily synthesis of the application;

[0043] Figure 6 It is an effect comparison chart of the sea temperature monthly synthesis of the application;

[0044] Figure 7 It is an effect comparison chart of the sea temperature monthly synthesis of the application;

[0045] Figure 8 It is a sea temperature anomaly comparison chart of the sea temperature monthly synthesis of the application. DETAILED DESCRIPTION

[0046] The embodiment of the application provides a long-sequence polar orbit satellite infrared sea surface temperature product reprocessing method and system, solves the problem that the precision of a wind cloud business sea surface temperature product and the consistency of a long-sequence sea surface temperature product cannot meet application requirements in the prior art, performs sea surface temperature inversion in an iterative manner, replaces a business cloud detection product with a sea surface temperature quality level, suppresses cloud pollution, replaces a business relatively fixed regression coefficient with a sea surface temperature regression coefficient of the month, performs sea surface temperature quality control with a daily analysis field sea surface temperature as a background field, gives a per-pixel sea surface temperature quality level, improves the precision and consistency of long-time-sequence sea surface temperature inversion, performs daily sea surface temperature synthesis according to the principle of quality priority, and establishes a blacklist for monthly synthesis, thereby improving the precision of daily / monthly sea surface temperature synthesis.

[0047] The technical solution in the embodiment of the application is as follows to solve the above problems:

[0048] In order to facilitate understanding of the application, the data used in the embodiment of the application is explained and described as follows:

[0049] (1) FY-3C visible light infrared scanning radiometer data

[0050] The instrument used for infrared sea surface temperature inversion in the embodiment of the application is VIRR of FY-3C, and table 1 shows the instrument channel information of VIRR.

[0051]

[0052] Table 1 VIRR spectral characteristics

[0053] The satellite observation data used in the application is FY-3C satellite VIRR observation data of the national satellite meteorological center resource pool, is L1B data in HDF5 format, is stored in 5min segments, each 5min segment L1B data contains 1000M and GEO two data files, and the nadir point spatial resolution is 1.1km. FY-3C performs global observation twice at about 10:00 and 22:00 every day.

[0054] (2) in situ sea surface temperature iQUAM (in situ SST QUAlity Monitor)

[0055] The in-situ sea temperature used in the embodiment of the application is from the NOAA (National Oceanic and Atmospheric Administration) Satellite Applications and Research (STAR) center, stored in a month unit, in a netCDF (network Common Data Form) storage format, and containing information such as sea temperatures of drift buoys, anchored buoys, high-resolution buoys and ships, and longitude, latitude and quality marks thereof, and is used for establishing a FY-3C / VIRR sea temperature matchup database (MDB).

[0056] (3) Daily analysis field sea temperature

[0057] The daily analysis field sea surface temperature used in the embodiments of the present application is a daily reanalysis sea surface temperature normalized to the same depth (20 cm) as the buoy observation, with a resolution of 0.05°x0.05° and stored in netCDF format, wherein 1981-2016 is ESA (European Space Agency) SST CCI (Climate Change Initiative) V2.1, and after 2017 is Copernicus Climate Change Service (C3S) V2.0. ESA SST CCI V2.1 (1981-2016) takes NOAA series and MetOp-A AVHRR (Advanced Very High Resolution Radiometer), European ATSR (Along Track Scanning Radiometer), AATSR (Advanced Along Track Scanning Radiometer) satellite data and OSI-SAF (Ocean and Sea Ice Satellite Application Facilities) sea ice data as input, reanalyzes satellite observed sea surface temperature with long sequence consistency re-scaled and reprocessed based on ATSR and AASTR, and normalizes the reanalysis sea surface temperature to the same depth (20 cm) as the buoy, which has higher resolution, higher accuracy and better long sequence consistency than other long sequence reanalysis sea surface temperature. After 2017, the same processing procedure as ESA SST CCI V2.1 analysis field sea surface temperature is used to take Sentinel-3A SLSTR (Sea and Land Surface Temperature Radiometer), MetOp-A AVHRR and OSI-SAF sea ice data as input, aiming to extend the reanalysis sea surface temperature to near real time (lag 28 days). For ease of description, ESA CCI is collectively referred to.

[0058] (4) Climate sea surface temperature

[0059] The climate sea surface temperature used in the present application is 30-year (1982-2011) monthly mean sea surface temperature OISST (Optimum Interpolation SST), from NOAA Earth System Research Laboratories (ESRL), with a resolution of 1°x1° and stored in netCDF format.

[0060] Embodiment 1:

[0061] AsFigure 1 The long-sequence polar orbit satellite infrared sea surface temperature product reprocessing method specifically includes the following steps as shown in the figure:

[0062] Step S1, sea surface temperature matching, calculating monthly sea surface temperature regression coefficients V0 with a business sea surface temperature matching dataset, performing NLSST sea surface temperature inversion to generate NLSST V0, and performing sea surface temperature matching based on satellite observation data after positioning and scaling, NLSST V0, in-situ sea surface temperature and daily analysis field sea surface temperature to establish a sea surface temperature matching dataset. Specifically, 5-minute segment FY-3C / VIRR LIB data, NLSST V0, in-situ sea surface temperature and daily analysis field sea surface temperature are input, and matching is performed according to a matching window of 3 km in space and 1 hour in time, and a pixel that meets the clear sky ocean and matching conditions is a matching sample. According to the latitude and longitude information of the satellite observation, the corresponding ESA CCI sea surface temperature of the matching sample is calculated by bilinear interpolation, and a global monthly FY-3C / VIRR sea surface temperature matching dataset of a 3×3 data block centered on the current pixel is established. The sea surface temperature quality level QL in NLSST V0 is used to replace the cloud detection product in the business sea surface temperature matching algorithm, wherein the quality code QL=5 (excellent) is a confident clear sky pixel, QL=4 (good) is a possible clear sky pixel, and QL=3 (poor) is a possible cloud pixel. The iQUAM buoy type is selected as a drifting buoy, a high-resolution buoy and a tropical anchored buoy. The matching conditions are: (1) there is no land in the 3×3 data block (the number of land pixels is equal to 0); (2) the matching distance between the satellite observation of the 3×3 data block center pixel and the buoy is less than or equal to 1.1 km; (3) the matching time difference between the satellite observation time of the 3×3 data block center pixel and the buoy is less than or equal to 60 minutes; (4) the buoy type of the matching sample is a drifting buoy (Drifter), a high-resolution buoy (Hight-Resolution Drifter) and a tropical anchored buoy (Tropical Mooring); (5) the buoy quality level QualityLevel of the matching sample is 5; (6) the satellite sea surface temperature quality level QualityLevel of the matching sample is 5. Each matching sample includes satellite observation time, latitude and longitude, split window brightness temperature, medium wave infrared channel brightness temperature, satellite zenith angle, satellite zenith angle, number of land pixels in the 3×3 data block, buoy sea surface temperature, buoy observation time, buoy latitude and longitude, buoy type, buoy quality level and ESA CCI SST.

[0063] Step S2, sea surface temperature regression, based on the monthly reprocessed sea surface temperature dataset, according to the sea surface temperature regression formula, the sea surface temperature regression calculation is performed. Specifically, the candidate sea surface temperature regression formula includes:

[0064] MCSST(D / N)T s =a0+a1T 11 +a2(T 11 -T12 )+a3(T 11 -T 12 (secθ-1) (1)

[0065] NLSST(D / N)T s =a0+a1T 11 +a2T FG (T 11 -T 12 )+a3(T 11 -T 12 (secθ-1) (2)

[0066] Among them, T S Indicates the inversion of sea surface temperature, T 11 T 12 These represent the brightness temperatures of the 10.8μm and 12μm channels, respectively; a0-a3 represent the regression coefficients; θ represents the satellite zenith angle; and T FG Let S represent the first guess SST. In this paper, ESA CCI is selected as the first guess SST. N represents the nighttime algorithm and D represents the daytime algorithm.

[0067] Daytime and nighttime are distinguished by a solar zenith angle of 90°. A solar zenith angle less than 90° is considered daytime, and greater than or equal to 90° is considered nighttime. Statistical regressions are performed separately for each. First, a least squares regression is performed monthly using the candidate sea surface temperature (SST) regression formulas to calculate the first guessed SST regression coefficient. These coefficients are then substituted into the SST matching dataset to calculate the retrieved SST. The median and standard deviation (STD) of the matching samples are calculated. Matching samples where the difference between the retrieved SST and the buoy SST exceeds Median ± 2STD are removed. A second least squares regression is then performed to calculate the monthly SST regression coefficients for subsequent SST reprocessing and retrieval.

[0068] Step S3: Sea surface temperature (SST) inversion. SST inversion uses the daily analysis field SST as the first guess to improve the SST background field. The monthly SST regression coefficients replace the relatively fixed operational regression coefficients, and SST quality control is performed to improve SST inversion accuracy. Specifically, Table 2 shows the regression coefficients for each SST regression formula in FY-3C / VIRR. The FY-3C / VIRR SST reprocessing uses the NLSST algorithm. When the first guess SST ESA CCI is invalid, the SST inverted using the MCSST algorithm is used as the first guess SST.

[0069] Algorithm D / N [a0] [a1] [a2] [a3 <!-- 6 -->]]> MCSST N -274.55683286 1.00669110 3.38825980 0.80469184 NLSST N -245.66389369 0.90986767 0.11513046 1.01864190 MCSST D -274.41447791 1.00796995 3.42691036 0.87610920 NLSST D -244.08701237 0.90599464 0.11792040 1.00485258

[0070] Table 2 FY-3C / VIRR retreated sea surface temperature regression coefficients

[0071] Sea surface temperature inversion specifically includes the following steps:

[0072] Step S31, read in the FY-3C / VIRR L1B data after positioning and scaling, ESA CCI, and the regression coefficient of the monthly sea temperature, identify the read-in L1B and its corresponding land-sea template data, obtain the marine observation data, calculate the split window channel brightness temperature T11 and T12 of the marine pixels, and assign -999 to the land pixels.

[0073] Step S32, perform threshold test on T11 and T12 of the marine pixels, the pixels meeting the brightness temperature threshold enter step S33, and the pixels exceeding the brightness temperature threshold are assigned a default value -888 and enter step S37.

[0074] Step S33, construct a 3x3 data block centered on the current marine pixel, perform spatial consistency test, if the spatial consistency is not met, assign a default value -888 and enter step S37, and if the spatial consistency is met, enter step S34.

[0075] Step S34, perform bilinear interpolation on the ESA CCI according to the longitude and latitude of the center pixel of the 3x3 data block to obtain the first guessed sea temperature Tfg, perform sea temperature effective value test, if the test is not passed, enter step S35, and if the test is passed, enter step S36.

[0076] Step S35, calculate the inversion sea temperature according to the MCSST regression formula as the first guessed sea temperature Tfg.

[0077] Step S36, calculate the inversion sea temperature Ts according to the NLSST regression formula, perform sea temperature quality control, and assign the pixel-by-pixel sea temperature quality level QL (the definition of the quality level is shown in Table 3). The sea temperature quality control specifically includes 7 tests, which are as follows: (1) land-sea template test, used to correct the wrong land-sea template, the early business land-sea template has errors, and part of the marine pixels are assigned with land identification, the marine pixels that should be marine but are assigned with land identification are assigned with marine pixel identification through the land-sea template test, and then the sea temperature inversion is performed. If the longitude and latitude of the pixel are between west longitude 62° and west longitude 55°, and the latitude is between south latitude 45° and south latitude 39°, the land-sea template is assigned with marine identification LandSeaMask=0, that is, the abnormal land pixels in the longitude range [-62-55] and the latitude range [-45-39] are assigned with marine pixel identification LandSeaMask=0; if the land-sea template LandSeaMask=1 of the pixel is a land pixel, TS=-999, and QL=0, which is determined by the sea temperature quality control. Figure 3It can be seen that after the land-sea template quality test, the land pixels that are extra due to abnormal land-sea template in the business sea surface temperature are controlled by the land-sea template quality control in the reprocessed sea surface temperature product, and the effective sea surface temperature inversion is obtained by assigning the marine pixel identifier to the region, which increases the number of effective sea surface temperature pixels;(2) Satellite zenith angle threshold test: the quality of the pixels within 50° of satellite zenith angle is excellent, QL=5, and the quality of the other pixels is good, QL=4;(3) Spatial consistency test: a 3*3 data block is constructed, and the standard deviation of the split window channel brightness temperature of the clear sky pixels in the data block is counted. The pixels with a standard deviation greater than one are considered as cloud pixels, TS=-888, QL=1. The split window channel brightness temperature of the clear sky pixels in the 3*3 data block is counted. When TB11_Max-Min≤1℃ and TB12_Max-Min≤1℃, the quality level is excellent, QL=5. When TB11_Max-Min≤2℃ and TB12_Max-Min≤2℃, the quality level is good, QL=4, and the others are poor, QL=3. Wherein: TB11, TB12 are 11 and 12 μm channel brightness temperature, Max is the maximum value, and Min is the minimum value;(4) Background field sea surface temperature threshold test: the ESA CCI of the daily analysis field sea surface temperature of the day is used as the reference sea surface temperature to perform the background field sea surface temperature threshold test. The pixels exceeding the threshold are considered as default values TS=-888, QL=1. When |TS-TCCI|≤T1, the quality is excellent, QL=5. When |TS-TCCI|≤T2, the quality is good, QL=4, and the others are poor, QL=3. Wherein: TS is the inversion sea surface temperature, TCCI is the background field sea surface temperature, T1, T2 are empirical thresholds (for example: T1 is 1.2℃, T2 is 2.5℃);(5) Solar flare area test: according to the solar zenith angle, satellite zenith angle and relative azimuth angle, the solar flare angle GlintAngle is calculated according to the following formula. When GlintAngle<=A, the quality is excellent, QL=5, and the others are good, QL=4. GlintAngle=acos(cos(SatZen)*cos(SolarZen)-sin(SatZen)*sin(SolarZen)*cos(RaZen))*180 / PAI, wherein SatZen is the satellite zenith angle, SolarZen is the solar zenith angle, and RaZen is the relative azimuth angle, all in radians.PAI = 3.1415926535897932384626, A is an empirical threshold (e.g. A = 35°); (6) sea ice test, according to the sea ice fraction in ESA CCI, when the sea ice fraction is less than 15%, the quality is excellent, QL = 5, when the sea ice fraction is greater than 60%, the quality is good, QL = 4, otherwise the quality is poor, QL = 3; (7) effective value range test of retrieved sea surface temperature, when the absolute temperature difference between the retrieved sea surface temperature and the brightness temperature of the 11 μm channel is greater than 10 °C, the cloud pixel is considered, TS = -888, QL = 1, when the retrieved sea surface temperature is less than -2 °C, the cloud pixel is considered, TS = -888, QL = 1. When the sea surface temperature quality control is performed, firstly, the background field sea surface temperature threshold test and the spatial consistency test are performed, and then the remaining tests are performed.

[0078] QL Meaning 0 Land 1 Cloud or default value 2 Standby 3 Sea temperature quality is poor 4 Sea temperature quality is good 5 Sea temperature quality is excellent

[0079] Table 3 FY satellite sea surface temperature quality grade definition

[0080] Step S37, write the sea surface temperature scientific data set and the per-pixel sea surface temperature quality identifier into the sea surface temperature data file.

[0081] Table 4 gives the quality test result comparison information of the FY-3C VIRR 5-minute segment sea surface temperature on June 1, 2014 13:45 UTC, with the reanalysis sea surface temperature CMC of the Canadian Meteorological Administration as the reference sea surface temperature. It can be seen that compared with the operational sea surface temperature, the number of samples with excellent quality of the reprocessed sea surface temperature increases from 231389 to 344600, the sea surface temperature bias changes from -0.85 °C to 0.35 °C, and the sea surface temperature root mean square error changes from 1.35 °C to 0.81 °C. The 5-minute segment is observed in the daytime, and due to the solar radiation warming, the satellite observed sea surface temperature is 0.35 °C higher than the reanalysis sea surface temperature CMC which removes the diurnal variation, which is reasonable, and the bias between the operational sea surface temperature and CMC is -0.85 °C, which indicates that there is cloud pollution phenomenon.

[0082] Compared with the operational sea surface temperature, the number of samples with good quality of the reprocessed sea surface temperature increases from 167163 to 226170, the sea surface temperature bias changes from -1.18 °C to -0.93 °C, and the sea surface temperature root mean square error changes from 1.93 °C to 1.47 °C; compared with the operational sea surface temperature, the number of sea surface temperature pixels of the whole sample of the reprocessed sea surface temperature increases from 398552 to 648703, the sea surface temperature bias changes from -0.99 °C to -0.5 °C, and the sea surface temperature root mean square error changes from 1.62 °C to 1.5 °C. The number of samples with poor quality of the operational sea surface temperature is 0, and the number of samples with poor quality of the reprocessed sea surface temperature is 77913, which indicates that the quality control and quality grading of the reprocessed sea surface temperature are superior to the operational sea surface temperature.

[0083] From Figure 3As can be seen from Table 4, the quality control and grading of the sea surface temperature reprocessing algorithm is more reasonable. Not only is the abnormality of the operational sea surface temperature template revised through the sea-land template quality test, but also the reprocessing precision and quality of the sea surface temperature are improved through the application of the sea surface temperature regression coefficient of the month and the more optimized sea surface temperature quality control scheme, and the number of sea surface temperature samples with excellent quality is increased.

[0084]

[0085]

[0086] Table 4 Comparison of 5-minute segment sea surface temperature quality inspection results on June 1, 2014, 13:45 UTC

[0087] Step S4, sea surface temperature daily synthesis, on the basis of about 288 5-minute segment sea surface temperature products per day, according to the day / night identifier, the original orbit sea surface temperature product of the subsolar point 1.1 km is respectively projected and converted into global 1 km equirectangular projection intermediate data in the day and night. On the basis of the global 1 km equirectangular projection intermediate data, a 5*5 data block is constructed, and the 1 km resolution equirectangular projection data is down-sampled to global 5 km equirectangular projection with the principle of quality first, stored separately in the day and night, and subjected to per-pixel quality control and stored in HDF5 format. The principle of quality control is as follows: in the 5*5 data block, if there is a pixel with excellent quality, the excellent pixel is averaged to give an excellent quality level; if there is no excellent pixel, the good pixel is averaged to give a good quality level; if there is no good pixel, the poor pixel is averaged to give a poor quality level; if there is no poor pixel, it is assigned as a default value (cloud or missing line, etc.), and on this basis, the following five types of tests are performed to give each pixel a quality level (the definition of the quality level is shown in Table 3).

[0088] The five types of tests are as follows: (1) land-sea template test. If the land-sea template LandSeaMask of a pixel is 1, the pixel is assigned a value of -999 and QL=0; (2) sea surface temperature (SST) effective value range test. The effective SST value is set to -2℃-45℃. If SST is less than -2℃, the default value -888 is assigned and QL=1. If SST is greater than 45℃, the value 45℃ is assigned and QL=3; (3) satellite zenith angle threshold test. The quality of a pixel is excellent if the satellite zenith angle is less than or equal to 50°, and QL=5. The quality of other pixels is good, and QL=4; (4) background field SST threshold test. The ESA CCI daily analysis field SST is used as the reference SST to perform the background field SST threshold test. The quality of a pixel is excellent if |T-TCCI|≤T1, and QL=5. The quality of a pixel is good if |T-TCCI|≤T2, and QL=4. The quality of a pixel is poor if |T-TCCI|≤T3, and QL=3. A pixel is considered to be a default value TS=-888 if it exceeds the threshold T3, and QL=1, where T is the daily synthetic SST, TCCI is the background field SST, and T1 and T2 are empirical thresholds (for example, T1 is 2℃, T2 is 3℃, and T3 is 5℃); (5) sea ice test. The quality of a pixel is excellent if the sea ice fraction is less than 15%, and QL=5. The quality of a pixel is good if the sea ice fraction is greater than 60%, and QL=4. The quality of other pixels is poor, and QL=3.

[0089] Figure 4 Fig. 2 shows a comparison between the FY-3C VIRR daytime 5km global operational SST and the repro SST on November 10, 2015. The left panel is the bias comparison, and the right panel is the root mean square error (RMSE) comparison. The gray color represents the operational SST, and the black color represents the repro SST. Figure 4 It can be seen that, after the high-precision background field ESA CCI SST threshold test, the abnormal strip (Pacific Ocean in eastern Australia) in the operational SST is effectively controlled in the repro SST product, and the reliability of the repro SST is improved.

[0090] Figure 5 Fig. 3 shows the error comparison between the FY-3C VIRR nighttime 5km global operational daily synthetic SST and the repro daily synthetic SST from 2014 to 2018, with the reanalysis field SST CMC as the reference SST. The left panel is the bias comparison, and the right panel is the RMSE comparison. The gray color represents the operational SST, and the black color represents the repro SST. The long-term bias of the operational SST is -0.16℃, and the RMSE is 0.73℃. The bias of the repro SST is 0.06℃, and the RMSE is 0.57℃. It can be seen that the repro SST is superior to the operational SST in terms of long-term consistency and accuracy.

[0091] Step S5, monthly synthesis of sea surface temperature, according to the quality inspection results of daily synthesized sea surface temperature, a blacklist is established (such as: QL=5 of sea surface temperature, root mean square error greater than 1℃ of daily synthesized sea surface temperature), on the basis of 28-31 daily synthesized sea surface temperature data of each month, according to the day and night identification, the principle of quality priority is adopted to carry out monthly synthesis, and is stored separately in day and night, and pixel-by-pixel quality control is carried out. The principle of quality control is: in 28-31 daily synthesized sea surface temperature data, if there is a pixel with excellent quality, then the average of the excellent pixels is taken and the excellent quality level is assigned; if there is no excellent pixel, then the average of the good pixels is taken and the good quality level is assigned; if there is no good pixel, then the average of the poor pixels is taken and the poor quality level is assigned; if there is no poor pixel, then the default value (cloud or data anomaly, etc.) is assigned, and on this basis, four types of verification are carried out, and a quality level is assigned to each pixel (the definition of the quality level is shown in Table 3).

[0092] The four types of verification are respectively: (1) blacklist verification, searching for the blacklist, if the date is on the blacklist, then it is indicated that the daily synthesized sea surface temperature data of the day is abnormal, and is not involved in monthly synthesis; (2) sea-land template verification, if the sea-land template LandSeaMask of the pixel is 1, then it is a land pixel, and -999 is assigned and QL=0; (3) sea surface temperature effective value range verification, the effective value of sea surface temperature is set to -2℃-45℃, if SST is less than -2℃, then -888 is assigned as the default value and QL=1, if SST is greater than 45℃, then 45℃ is assigned and QL=3; (4) climate sea surface temperature threshold verification, taking 30-year monthly average OISST as the reference sea surface temperature, when |T-TOISST|≤T1, the quality is excellent and QL=5, when |T-TOISST|≤T2, the quality is good and QL=4, when |T-TOISST|≤T3, QL=3, and the pixel exceeding the threshold T3 is regarded as a default value TS=-888 and QL=1, wherein: T is daily synthesized sea surface temperature, TOISST is climate sea surface temperature, T1, T2 are empirical thresholds (such as: T1 is 4℃, T2 is 5℃, and T3 is 6℃).

[0093] Figure 6 The comparison chart of the FY-3C / VIRR night operation monthly synthesized sea surface temperature and the repro monthly synthesized sea surface temperature in January 2017 is shown in FIG. 3. Figure 6 It can be seen that after the blacklist verification, the abnormal high temperature in the polar region in the monthly synthesized sea surface temperature in January 2017 in the North Pole sea area caused by the positioning anomaly on January 1, 2017 in the operation sea surface temperature is effectively removed in the repro monthly synthesized sea surface temperature, and the reliability and accuracy of the monthly synthesized sea surface temperature are ensured.

[0094] Figure 7 The comparison chart of the FY-3C / VIRR night operation monthly synthesized sea surface temperature and the repro monthly synthesized sea surface temperature in November 2015 is shown in FIG. 4. FromFigure 7 It can be seen that the reprocessed sea surface temperature more finely depicts the distribution of global sea surface temperature than the operational sea surface temperature. By performing full-disk projection on FY-3C / VIRR and OSTIA (Operational Sea Surface Temperature and Ice Analysis) global sea surface temperature anomaly, the full-disk sea surface temperature anomaly data and images of the Pacific Ocean are generated to focus on the key area of ENSO, so as to illustrate the improvement effect of the FY-3C / VIRR sea surface temperature reprocessing algorithm by comparing the difference of the Pacific Ocean sea surface temperature anomaly before and after the FY-3C / VIRR sea surface temperature reprocessing. Figure 8 The following is a comparison of the FY-3C VIRR sea surface temperature anomaly in November 2015, wherein the left image is the FY-3C VIRR operational monthly synthetic sea surface temperature anomaly, the middle image is the FY-3C VIRR reprocessed monthly synthetic sea surface temperature anomaly, and the right image is the OSTIA monthly sea surface temperature anomaly. It can be seen that:

[0095] ① Although the operational monthly synthetic sea surface temperature anomaly can monitor the abnormal warming of the equatorial mid-east Pacific sea surface temperature, the intensity is far less than that of the reprocessed monthly synthetic sea surface temperature;

[0096] ② The FY-3C VIRR reprocessed monthly synthetic sea surface temperature anomaly is basically consistent with the OSTIA monthly sea surface temperature anomaly, and the abnormal warm sea surface temperature anomaly center in the equatorial mid-east Pacific in November 2015 exceeds 3℃;

[0097] ③ The reprocessing of the sea surface temperature product improves the application ability of the domestic satellite historical data in the marine ENSO climate monitoring.

[0098] Embodiment 2:

[0099] A long-sequence polar orbit satellite infrared sea surface temperature product reprocessing system, the system comprises a processor and a storage unit, the storage unit is used for storing a computer program, and the processor can realize the long-sequence polar orbit satellite infrared sea surface temperature product reprocessing method by executing the computer program.

[0100] Although the embodiments of the present application are based on the FY-3C visible and infrared radiometer (VIRR) sea surface temperature product for reprocessing, it should be noted that the method of the embodiments of the present application is suitable for the sea surface temperature retrieval and reprocessing of most infrared remote sensing instruments on satellites, and the data of the FY-3C is only used for illustration, and should not limit the protection scope of the present application.

[0101] The reprocessed sea surface temperature adopts a unified product design, on the basis of covering the FY3 (02) batch PGS sea surface temperature scientific data set, adding a per-pixel sea surface temperature quality level (Quality Level, QL) and sea ice auxiliary data, providing more references for sea surface temperature quality control and user application, and can be expanded to FY-1C / D and FY-3A / B in the future, so that the long sequence of Fengyun satellite climate sea surface temperature products has a unified data specification, and the long sequence of Fengyun satellite sea surface temperature products obtained by using the improved sea surface temperature reprocessing algorithm to retrieve the historical observation data of the Fengyun satellite have good ENSO event monitoring capability.

[0102] Compared with the prior art, the present application has the following beneficial effects:

[0103] 1、In the present application, the reanalysis sea surface temperature ESA CCI V2.1 based on the 40-year long sequence satellite sea surface temperature of the European reprocessing normalized to the depth of 20 cm of the sea surface is taken as a background field, which provides a high-precision and long-sequence consistent reference sea surface temperature for the long-sequence Fengyun satellite sea surface temperature retrieval, and the depth of the sea surface temperature of the drifting buoy participating in the regression calculation is the same, and the long-sequence Fengyun satellite sea surface temperature retrieved is equivalent to the sea surface temperature of 20 cm depth, thereby laying a good long-sequence consistency foundation for the long-sequence Fengyun satellite sea surface temperature retrieval.

[0104] 2、In the present application, the sea surface temperature retrieval reprocessing directly takes FY L1 in the resource pool of the National Satellite Meteorological Center as input, which is different from the business sea surface temperature retrieval taking FY L1 and L2 cloud detection products as input and adopting fixed sea surface temperature regression coefficients, thereby not only avoiding the introduction of cloud pollution in the business cloud detection, but also adopting an iterative method, first taking the business sea surface temperature regression coefficient of the current month, taking the high-precision daily analysis field ESA CCI as the first guessed sea surface temperature to perform nonlinear NLSST sea surface temperature retrieval, correcting the abnormal sea-land template in the FY L1, adding high-precision background field sea surface temperature quality control, flare area quality control and sea ice quality control on the basis of the business sea surface temperature quality control, and assigning a quality level (QL=0-5) to each sea surface temperature pixel, and on this basis, taking QL=5 as certain clear sky, QL=4 as possible clear sky and QL=3 as possible cloud to perform sea surface temperature matching, and establishing a long-sequence sea surface temperature reprocessing matching data set MDB.

[0105] 3、The application is based on long sequence sea surface temperature reprocessing matching data set MDB, and monthly least square based quadratic regression is carried out to calculate monthly sea surface temperature regression coefficient, and 7 kinds of sea surface temperature quality control are carried out, i.e., sea-land template test, satellite zenith angle threshold test, spatial consistency test, background field sea surface temperature threshold test, solar flare area test, sea ice test and inversion sea surface temperature effective value range test, so as to give each sea surface temperature pixel a quality grade and improve sea surface temperature inversion precision and long sequence sea surface temperature consistency.

[0106] 4、The application is based on about 288 5min segment sea surface temperature products per day, and according to day and night identification, global 5km sea surface temperature daily synthesis is carried out according to quality priority principle, and is stored in day and night, and 5 kinds of sea surface temperature quality control are carried out, i.e., sea-land template test, sea surface temperature effective value range test, satellite zenith angle threshold test, background field sea surface temperature threshold test and sea ice test, so as to give each pixel a quality grade and improve sea surface temperature daily synthesis precision and long sequence sea surface temperature consistency.

[0107] 5、The application is based on daily synthesis sea surface temperature quality test result to establish a blacklist, and sea surface temperature on the blacklist is removed in monthly synthesis, so as to participate in sea surface temperature monthly synthesis and improve monthly synthesis sea surface temperature reliability. According to day and night identification, monthly synthesis is carried out according to quality priority principle, and is stored in day and night, and 4 kinds of sea surface temperature quality control are carried out, i.e., blacklist test, sea-land template test, sea surface temperature effective value range test and climate sea surface temperature threshold test, so as to give each pixel a quality grade and improve sea surface temperature monthly synthesis precision and reliability.

[0108] Finally, it should be noted that: obviously, the above embodiments are only examples for clearly illustrating the present application, and are not limited to the implementation. For ordinary skilled in the art, other different forms of changes or variations can be made on the basis of the above description. Here, all the implementation is not required or can not be exhausted. The obvious changes or variations derived from the present application are still within the protection scope of the present application.

Claims

1. A long sequence polar orbiting satellite infrared sea surface temperature product reprocessing method, characterized in that, The long-sequence polar orbit satellite infrared sea surface temperature product reprocessing method specifically comprises the following steps: Step S1, sea surface temperature matching, calculating monthly nonlinear sea surface temperature (NLSST) regression coefficient V0 with a business sea surface temperature matching dataset, performing sea surface temperature inversion to generate NLSST V0, performing sea surface temperature matching based on satellite observation data after positioning and scaling, NLSST V0, in-situ sea surface temperature and daily analysis field sea surface temperature to establish a monthly reprocessing sea surface temperature matching dataset; Step S2, sea surface temperature regression, based on the monthly reprocessing sea surface temperature matching dataset, performing quadratic least squares regression calculation to obtain sea surface temperature regression coefficient V1; Step S3, sea surface temperature inversion, performing original orbit sea surface temperature inversion with the monthly sea surface temperature regression coefficient V1, performing per-pixel sea surface temperature quality control with the daily analysis field sea surface temperature as the background field, and assigning per-pixel sea surface temperature quality grades; Step S4, sea surface temperature daily synthesis, performing daily synthesis according to the quality priority principle based on the original orbit sea surface temperature and the per-pixel sea surface temperature quality grades within a day, performing sea surface temperature quality control with the daily analysis field sea surface temperature as the background field, and assigning per-pixel sea surface temperature quality grades; Step S5, sea surface temperature monthly synthesis, performing monthly synthesis according to the quality priority principle based on the daily synthesis sea surface temperature and the per-pixel sea surface temperature quality grades within a month, performing sea surface temperature quality control with the monthly analysis field sea surface temperature as the background field, and assigning per-pixel sea surface temperature quality grades; In the step S1, 5min segment FY-3C / VIRR LIB data, NLSST V0, in-situ sea surface temperature and daily analysis field sea surface temperature are taken as inputs, matching is performed according to a matching window of 3km in space and 1hr in time, pixels meeting the clear sky ocean and matching conditions are taken as a matching sample, the corresponding daily analysis field sea surface temperature of the matching sample is calculated through bilinear interpolation according to the latitude and longitude information of the satellite observation, and a sea surface temperature matching dataset of a 3*3 data block centered on the current pixel is established; In the step S2, least squares regression is performed according to the sea surface temperature regression formula on a monthly basis, the first guessed sea surface temperature regression coefficient is calculated, the regression coefficient is substituted into the sea surface temperature matching dataset, the inverted sea surface temperature is calculated, the median (Median) and the standard deviation (STD) of the matching sample are calculated, the matching sample whose difference between the inverted sea surface temperature and the buoy sea surface temperature exceeds the Median±2STD is removed, quadratic least squares regression is performed, the monthly sea surface temperature regression coefficient is calculated, and is used for subsequent sea surface temperature reprocessing inversion; Step S31, read in the satellite observation data after positioning and scaling, the daily analysis field sea temperature, and the regression coefficient of the sea temperature in the current month, identify the read-in satellite observation data and the corresponding land-sea template data to obtain marine observation data, calculate the split window channel brightness temperature T 11 and T 12 , and the land pixel is assigned -999; Step S32, T 11 and T 12 Threshold test is performed on the marine pixels, the pixels meeting the brightness temperature threshold enter step S33, the pixels exceeding the brightness temperature threshold are given a default value -888 and enter step S37; Step S33, a 3*3 data block centered on the current ocean pixel is constructed, spatial consistency test is performed, if the spatial consistency is not met, a default value -888 is assigned, and step S37 is entered, if the spatial consistency is met, step S34 is entered; Step S34, the daily analysis field sea surface temperature is bilinearly interpolated according to the latitude and longitude of the center pixel of the 3*3 data block to obtain the first guessed sea surface temperature, sea surface temperature effective value test is performed, if not passed, step S35 is entered, if passed, step S36 is entered; Step S35, the inverted sea surface temperature is calculated according to the MCSST regression formula, as the first guessed sea surface temperature; Step S36, the inverted sea surface temperature is calculated according to the NLSST regression formula, sea surface temperature quality control is performed, and per-pixel sea surface temperature quality grades are assigned; Step S37, write the sea surface temperature scientific data set and the per-pixel sea surface temperature quality mark into a sea surface temperature data file; The MCSST regression formula is: T s = a0 + a1T 11 + a2(T 11 - T 12 ) + a3(T 11 - T 12 )(secθ - 1) The NLSST regression formula is: T s = a0 + a1T 11 + a2T FG (T 11 - T 12 ) + a3(T 11 - T 12 )(sec0 - 1) where T S represents the inverted sea surface temperature, T 11 , T 12 represent the brightness temperatures at 10.8 μm and 12 μm channels, respectively, a0-a3 represent the regression coefficients, and θ represents the satellite zenith angle, T FG represents the first guessed sea surface temperature.

2. The long sequence polar orbiting satellite infrared sea surface temperature product reprocessing method according to claim 1, characterized in that, The sea surface temperature quality control specifically includes seven tests, namely: (1) sea-land template test; (2) satellite zenith angle threshold test; (3) spatial consistency test; (4) background field sea surface temperature threshold test; (5) solar flare region test; (6) sea ice test; and (7) effective sea surface temperature value range test.

3. The long sequence polar orbiting satellite infrared sea surface temperature product reprocessing method of claim 1, wherein, In the step S4, on the basis of about 288 5-min sea surface temperature products per day, the original 1.1 km nadir orbit sea surface temperature products are projected and converted into day and night global 1 km equirectangular projection intermediate data according to day and night marks, 5*5 data blocks are constructed on the basis of the global 1 km equirectangular projection intermediate data, the 1 km equirectangular projection data is down-sampled to global 5 km equirectangular projection with quality priority, stored separately according to day and night, and subjected to per-pixel quality control, and stored in HDF5 format.

4. The long sequence polar orbit satellite infrared sea surface temperature product reprocessing method according to claim 3, characterized in that, The quality priority principle is that, in the 5*5 data block, if there is a pixel with excellent quality, the excellent pixels are averaged to obtain an excellent quality level; if there is no excellent pixel, the good pixels are averaged to obtain a good quality level; if there is no good pixel, the poor pixels are averaged to obtain a poor quality level; and if there is no poor pixel, a default value is assigned, and five tests are performed on the basis thereof, the five tests being: (1) sea-land template test; (2) effective sea surface temperature value range test; (3) satellite zenith angle threshold test; (4) background field sea surface temperature threshold test; and (5) sea ice test.

5. The long sequence polar orbit satellite infrared sea surface temperature product reprocessing method according to claim 1, characterized in that, In the step S5, a black list is established according to the quality test results of the daily synthesized sea surface temperature, and on the basis of 28-31 daily synthesized sea surface temperature data, the monthly synthesis is performed according to day and night marks with quality priority, stored separately according to day and night, and subjected to per-pixel quality control.

6. The long sequence polar orbit satellite infrared sea surface temperature product reprocessing method according to claim 5, characterized in that, The quality control principle is that, in the 28-31 daily synthesized sea surface temperature data, if there is a pixel with excellent quality, the excellent pixels are averaged to obtain an excellent quality level; if there is no excellent pixel, the good pixels are averaged to obtain a good quality level; if there is no good pixel, the poor pixels are averaged to obtain a poor quality level; and if there is no poor pixel, a default value is assigned, and four tests are performed on the basis thereof, the four tests being: (1) black list test; (2) sea-land template test; (3) effective sea surface temperature value range test; and (4) climate sea surface temperature threshold test.

7. A long sequence polar orbiting satellite infrared sea surface temperature product reprocessing system characterized by, The long-sequence polar orbit satellite infrared sea surface temperature product reprocessing system comprises a processor and a storage unit, the storage unit is used for storing a computer program, and the processor can realize the long-sequence polar orbit satellite infrared sea surface temperature product reprocessing method in any one of claims 1-6 by executing the computer program.

Images