Evaluation of Reprocessed Fengyun-3B Global Outgoing Longwave Radiation Data: Comparison with CERES OLR

FY-3B全球OLR再处理数据集评估:与CERES OLR数据集对比

+ Author Affiliations + Find other works by these authors
  • Corresponding author: Jian LIU, liujian@cma.gov.cn
  • Funds:

    Supported by the National Key Research and Development Program of China (2018YFB0504900 and 2018YFB0504905) and National Natural Science Foundation of China (41801278)

  • doi: 10.1007/s13351-022-1132-4

PDF

  • Outgoing longwave radiation (OLR) at the top of the atmosphere (TOA) is a key parameter for understanding and interpreting the relationship between clouds, radiation, and climate interactions. It has been one of the operational products of the Fengyun (FY) meteorological satellites. OLR accuracy has gradually improved with advancements in satellite payload performance and the OLR retrieval algorithm. Supported by the National Key R&D Program Retrospective Calibration of Historical Chinese Earth Observation Satellite data (Richceos) project, a long-term OLR climate data record (CDR) was reprocessed based on the recalibrated Level 1 data of FY series satellites using the latest OLR retrieval algorithm. In this study, Fengyun-3B (FY-3B)’s reprocessed global OLR data from 2010 to 2018 were evaluated by using the Clouds and the Earth’s Radiant Energy System (CERES) global daily OLR data. The results showed that there was a high consistency between the FY-3B instantaneous OLR and CERES Single Scanner Footprint (SSF) OLR. Globally, between the two CDR datasets, the correlation coefficient reached 0.98, and the root-mean-square error (RMSE) was approximately 8–9 W m−2. The bias mainly came from the edge regions of the satellite orbit, which may be related to the satellite zenith angle and cloud cover distribution. It was shown that the long-term FY-3B OLR had temporal stability compared to CERES OLR long-term data. In terms of spatial distribution, the mean deviations showed zonal and seasonal characteristics, although seasonal fluctuations were observed in the differences between the two datasets. Effects of FY-3B OLR application to the South China Sea monsoon region and ENSO were demonstrated and analyzed, and the results showed that the seasonal deviation of FY-3B’s OLR comes mainly from the retrieval algorithm. However, it has little effect on the analysis of climate events.
    大气顶出射长波辐射(OLR)是研究云、辐射和气候相互作用关系的重要参量,也是风云气象卫星的业务产品之一。随着卫星有效载荷性能和OLR算法的更新迭代,风云卫星OLR产品精度也逐渐提高。在国家重点研发计划项目的支持下,利用最新OLR算法,基于风云卫星再定标L1级数据,对OLR产品进行一致性处理,生产OLR长时间气候数据集。本研究采用云和地球辐射能量系统(CERES)全球OLR日产品评估FY-3B卫星2010年至2018年的再处理全球OLR数据。结果显示,FY-3B的瞬时OLR和CERES OLR之间有很高的一致性。在全球范围内,两个数据集之间的相关系数达到0.98,均方根误差(RMSE)约为8–9 W m−2。偏差主要来自于卫星轨道的边缘区域,这可能与卫星天顶角和云量分布有关。长序列数据对比显示,FY-3B的长序列OLR产品具有较好的时间稳定性;在空间分布方面,与CERES OLR相比,平均偏差呈现出纬度带和季节性特征。对南海季风区和ENSO区分析表明,FY-3B OLR数据集的季节偏差主要来自于算法;然而,它对区域气候事件的分析影响不大。
  • 加载中
  • Fig. 1.  Instantaneous deviations between FY-3B OLR and CERES SSF OLR during (a) the day and (b) the night on 22 August 2018.

    Fig. 2.  Factors influencing the spatial distribution of the instantaneous deviations between FY-3B OLR and CERES SSF OLR. (a) Satellite zenith angle influence and (b) cloud cover influence. The blue dotted line represents the average bias, the red dotted line indicates the average RMSE, and the gray bar indicates the number of samples considered in the statistical analysis.

    Fig. 3.  The probability density distribution of the daily global mean (a) CERES OLR and (b) FY-3B OLR.

    Fig. 4.  Time series of the (a) RMSE, (b) bias, (c) correlation coefficients, and (d) daily mean OLR of the global daily FY-3B OLR and CERES OLR data from 2010 to 2018.

    Fig. 5.  Climate analysis through evaluation of annual mean OLR over the globe during 2010–2018 in the (a) CERES OLR dataset and (b) FY-3B OLR dataset, and (c) OLR difference between the OLR means of CERES and FY-3B.

    Fig. 6.  Influence of TCW on the daily OLR comparisons. The color bars indicate the date.

    Fig. 7.  (a) The time coefficient and (b) the spatial correlation of the first mode of the SVD.

    Fig. 8.  The difference (OLR2 − OLR24) in the daily mean OLRs calculated in each month of 2020 using FY-4A OLR. FY-4A OLR data were recorded on an hourly basis (OLR24) as well as at two-time points (OLR2; 0130 and 1330 local time).

    Fig. 9.  Spatial distributions of the statistical characteristics in different seasons: (a–d) RMSE and (e–h) bias.

    Fig. 10.  Time series of (a) the daily average OLR and (b) OLR anomalies in key region (10°–20°N, 110°–120°E) of SCSSM monitoring from 20 April to 10 June 2016. The red line and blue line represent FY-3B OLR and CERES OLR, respectively.

    Fig. 11.  Time series of the monthly average FY-3B OLR (red) and CERES OLR (blue) in the (a) Niño-3 region and (b) Niño-4 region from January 2011 to November 2018.

    Fig. 12.  Time series of the monthly FY-3B OLR (red) and CERES OLR (blue) anomaly in the (a) Niño-3 region and (b) Niño-4 region from January 2011 to November 2018.

    Table 1.  Specifications of the data used in this study

    DataSpatial resolutionTemporal coverage
    Satellite dataOLRFY-3B VIRR0.05° × 0.05°2010-12-05 to 2019-12-31
    OLRFY-4A Advanced Geostationary Radiation Imager (AGRI)0.05° × 0.05°2017-01-08 to present
    SSF OLRAqua CERES~20 km (FOV)2002-07-03 to present
    SYN1deg OLRAqua/Terra CERES1° × 1°2000-03-01 to present
    Total cloud area fractionAqua CERES1° × 1°2002-07-01 to present
    Reanalysis dataTemperature profileECMWF ERA50.25° × 0.25°1979-01-01 to present
    Humidity profileECMWF ERA50.25° × 0.25°1979-01-01 to present
    Column total waterECMWF ERA50.25° × 0.25°1979-01-01 to present
    Download: Download as CSV

    Table 2.  The annual average statistics of FY-3B OLR compared to those of CERES SYN1deg OLR

    YearMEAN (Mean/Std)MAX (value)MIN (value)RMSE (Mean/Std)Bias (Mean/Std)R (Mean/Std)STD (Mean/Std)
    2010223.42/1.11225.53221.1212.07/0.281.49/0.370.968/0.0039.00/0.23
    2011225.56/3.34232.73214.0012.57/0.762.29/0.650.963/0.0069.41/0.53
    2012225.44/2.88231.85219.9812.67/0.752.23/0.730.962/0.0069.44/0.51
    2013225.25/2.88231.73218.9612.44/0.821.88/0.760.963/0.0069.32/0.63
    2014225.08/2.85230.68219.7512.45/0.811.66/0.680.963/0.0069.32/0.59
    2015224.64/2.79230.72219.6912.43/0.831.48/0.700.964/0.0069.30/0.59
    2016225.40/2.62231.43220.4312.60/0.891.28/0.720.962/0.0069.45/0.59
    2017224.92/2.49230.59220.3912.44/0.991.20/0.640.963/0.0069.32/0.70
    2018224.59/2.92230.68219.5212.60/1.161.14/0.600.962/0.0069.39/0.82
    Note: MEAN (Mean/Std) indicates the mean value and mean deviation (W m−2) of the FY-3B daily mean OLR time series; MAX/MIN (value) indicates the maximum/minimum value (W m−2) of the FY-3B daily mean OLR time series; and RMSE/Bias/R/STD indicates the root-mean-square error (W m−2)/the mean deviation (W m−2)/the correlation coefficient/the standard deviation (W m−2) of FY-3B OLR compared to CERES OLR, respectively.
    Download: Download as CSV

    Table A1.  The reprocessed FY-3B OLR dataset

    FieldContent
    Title of datasetFY-3B VIRR Reprocessed Daily Outgoing Longwave Radiation (OLR) Dataset
    AbbreviationFY3VIRRReproDailyOLR
    Retrieval and rangeData are retrieved by a unified inversion algorithm based on the FY-3A/B/C recalibrated VIRR L1 data, with values ranging from 0 to 500 W m−2
    Main data producerWanchun ZHANG
    Email: zhangwc@cma.gov.cn
    Tel: +86-10-68407237
    Spatial coverage0°–360°, 90°S–90°N
    Temporal coverageFrom 5 December 2010 to 31 December 2019
    ResolutionsSpatial resolution: 0.05° × 0.05°;
    Temporal resolution: daily/monthly
    Format and sizeHDF; approximately 150 M every day
    Data publishing unitNational Satellite Meteorological Center, China Meteorological Administration
    Funding projectNational Key Research and Development Program of China (2018YFB0504905)
    DOI (accessible at) 10.12185/NSMC.RICHCEOS.FCDR.FY3VIRRReproDailyOLR.FY3.VIRR.L2.GBAL.POAD.GLL.5000M.HDF.2021.3.V2
    Download: Download as CSV
  • [1]

    Bowman, K. W., D. T. Shindell, H. M. Worden, et al., 2013: Evaluation of ACCMIP outgoing longwave radiation from tropospheric ozone using TES satellite observations. Atmos. Chem. Phys., 13, 4057–4072. doi: 10.5194/acp-13-4057-2013.
    [2]

    Chen, Y., 2006: A study of the southeast Asian summer monsoon onset, evolution and its influence on the weather and climate over the southwest of China. Ph.D. dissertation, Nanjing University of Information Science & Technology, Nanjing, 152 pp. (in Chinese)
    [3]

    Chiodi, A. M., and D. E. Harrison, 2013: El Niño impacts on seasonal U.S. atmospheric circulation, temperature, and precipitation anomalies: The OLR-event perspective. J. Climate, 26, 822–837. doi: 10.1175/JCLI-D-12-00097.1.
    [4]

    Chiodi, A. M., and D. E. Harrison, 2015: Global seasonal precipitation anomalies robustly associated with El Niño and La Niña events—An OLR perspective. J. Climate, 28, 6133–6159. doi: 10.1175/JCLI-D-14-00387.1.
    [5]

    Clerbaux, N., S. Dewitte, L. Gonzalez, et al., 2003: Outgoing longwave flux estimation: improvement of angular modelling using spectral information. Remote Sens. Environ., 85, 389–395. doi: 10.1016/S0034-4257(03)00015-4.
    [6]

    Clerbaux, N., T. Akkermans, E. Baudrez, et al., 2020: The climate monitoring SAF outgoing longwave radiation from AVHRR. Remote Sens., 12, 929. doi: 10.3390/rs12060929.
    [7]

    Gruber, A., and A. F. Krueger, 1984: The status of the NOAA outgoing longwave radiation data set. Bull. Amer. Meteor. Soc., 65, 958–962. doi: 10.1175/1520-0477(1984)065<0958:TSOTNO>2.0.CO;2.
    [8]

    Hu, X. Q., 2012: Unified radiometric recalibration study on long-term historical data record of meteorological satellite sensors. Ph.D. dissertation, Institute of Remote Sensing Applications, Chinese Academy of Sciences, Beijing, 262 pp. (in Chinese)
    [9]

    Huang, M., J. D. Li, G. Zeng, et al., 2020: Regional characteristics of cloud radiative effects before and after the South China Sea summer monsoon onset. J. Meteor. Res., 34, 1167–1182. doi: 10.1007/s13351-020-0018-6.
    [10]

    Inoue, T., and S. A. Ackerman, 2002: Radiative effects of various cloud types as classified by the split window technique over the eastern sub-tropical Pacific derived from collocated ERBE and AVHRR data. J. Meteor. Soc. Japan, 80, 1383–1394. doi: 10.2151/jmsj.80.1383.
    [11]

    Kiladis, G. N., J. Dias, K. H. Straub, et al., 2014: A comparison of OLR and circulation-based indices for tracking the MJO. Mon. Wea. Rev., 142, 1697–1715. doi: 10.1175/MWR-D-13-00301.1.
    [12]

    Kim, B.-Y., and K.-T. Lee, 2019: Using the Himawari-8 AHI multi-channel to improve the calculation accuracy of outgoing longwave radiation at the top of the atmosphere. Remote Sens., 11, 589. doi: 10.3390/rs11050589.
    [13]

    Kim, B.-Y., K.-T. Lee, J.-B. Jee, et al., 2018: Retrieval of outgoing longwave radiation at top-of-atmosphere using Himawari-8 AHI data. Remote Sens. Environ., 204, 498–508. doi: 10.1016/j.rse.2017.10.006.
    [14]

    Knapp, K. R., S. Ansari, C. L. Bain, et al., 2011: Globally gridded satellite observations for climate studies. Bull. Amer. Meteor. Soc., 92, 893–907. doi: 10.1175/2011BAMS3039.1.
    [15]

    Lee, H.-T., 2014: Climate Algorithm Theoretical Basis Document (C-ATBD): Outgoing Longwave Radiation (OLR)-Daily. NOAA’s Climate Data Record (CDR) Program, CDRP-ATBD-0526, 46 pp. Available online at https://www.ncei.noaa.gov/pub/data/sds/cdr/CDRs/Outgoing%20Longwave%20Radiation%20-%20Daily/AlgorithmDescription_01B-21.pdf. Accessed on 16 May 2022.
    [16]

    Lee, H.-T., A. Gruber, R. G. Ellingson, et al., 2007: Development of the HIRS outgoing longwave radiation climate dataset. J. Atmos. Oceanic Technol., 24, 2029–2047. doi: 10.1175/2007JTECHA989.1.
    [17]

    Liebmann, B., and C. A. Smith, 1996: Description of a complete (interpolated) outgoing longwave radiation dataset. Bull. Amer. Meteor. Soc., 77, 1275–1277.
    [18]

    Liu, J., Y. Q. Da, T. Li, et al., 2020: Impact of ENSO on MJO pattern evolution over the Maritime Continent. J. Meteor. Res., 34, 1151–1166. doi: 10.1007/s13351-020-0046-2.
    [19]

    Liu, L., W. C. Zhang, W. Chen, et al., 2021: Evaluation of FY-3B reprocessed OLR data in the Asian–Australian monsoon region during 2011–2019: Comparison with NOAA OLR. J. Meteor. Res., 35, 964–974. doi: 10.1007/s13351-021-1086-y.
    [20]

    Matthews, A. J., 2008: Primary and successive events in the Madden–Julian Oscillation. Quart. J. Roy. Meteor. Soc., 134, 439–453. doi: 10.1002/qj.224.
    [21]

    Nakazawa, T., 2006: Madden-Julian Oscillation activity and typhoon landfall on Japan in 2004. SOLA, 2, 136–139. doi: 10.2151/sola.2006-035.
    [22]

    Ohring, G., A. Gruber, and R. Ellingson, 1984: Satellite determinations of the relationship between total longwave radiation flux and infrared window radiance. J. Appl. Meteor. Climatol., 23, 416–425. doi: 10.1175/1520-0450(1984)023<0416:SDOTRB>2.0.CO;2.
    [23]

    Priestley, K. J., G. L. Smith, S. Thomas, et al., 2007: Validation protocol for climate quality CERES measurements. Proceedings of SPIE 6678, Infrared Spaceborne Remote Sensing and Instrumentation XV, SPIE, San Diego, USA, 66781I, doi: 10.1117/12.735312.
    [24]

    Ren, S. L., Y. Li, X. Fang, et al., 2018: The South China Sea summer monsoon onset index using FY satellite derived data. J. Trop. Meteor., 34, 587–597. doi: 10.16032/j.issn.1004-4965.2018.05.002. (in Chinese)
    [25]

    Schmetz, J., and Q. H. Liu, 1988: Outgoing longwave radiation and its diurnal variation at regional scales derived from Meteosat. J. Geophys. Res. Atmos., 93, 11,192–11,204. doi: 10.1029/JD093iD09p11192.
    [26]

    Short, D. A., and R. F. Cahalan, 1983: Interannual variability and climatic noise in satellite-observed outgoing longwave radiation. Mon. Wea. Rev., 111, 572–577. doi: 10.1175/1520-0493(1983)111<0572:IVACNI>2.0.CO;2.
    [27]

    Singh, A., U. C. Mohanty, and G. Mishra, 2014: Long-lead prediction skill of Indian summer monsoon rainfall using outgoing longwave radiation (OLR): an application of canonical correlation analysis. Pure Appl. Geophys., 171, 1519–1530. doi: 10.1007/s00024-013-0689-3.
    [28]

    Stechmann, S. N., and H. R. Ogrosky, 2014: The Walker circulation, diabatic heating, and outgoing longwave radiation. Geophys. Res. Lett., 41, 9097–9105. doi: 10.1002/2014GL062257.
    [29]

    Stowe, L. L., H. Jacobowitz, G. Ohring, et al., 2002: The Advanced Very High Resolution Radiometer (AVHRR) Pathfinder Atmosphere (PATMOS) climate dataset: Initial analyses and evaluations. J. Climate, 15, 1243–1260. doi: 10.1175/1520-0442(2002)015<1243:TAVHRR>2.0.CO;2.
    [30]

    Susskind, J., G. Molnar, L. Iredell, et al., 2012: Interannual variability of outgoing longwave radiation as observed by AIRS and CERES. J. Geophys. Res. Atmos., 117, D23107. doi: 10.1029/2012JD017997.
    [31]

    Taylor, P. C., 2012: Tropical outgoing longwave radiation and longwave cloud forcing diurnal cycles from CERES. J. Atmos. Sci., 69, 3652–3669. doi: 10.1175/JAS-D-12-088.1.
    [32]

    Wang, C. Z., R. H. Weisberg, and J. I. Virmani, 1999: Western Pacific interannual variability associated with the El Niño–Southern Oscillation. J. Geophys. Res. Oceans, 104, 5131–5149. doi: 10.1029/1998JC900090.
    [33]

    Weickmann, K. M., G. R. Lussky, and J. E. Kutzbach, 1985: Intraseasonal (30–60 day) fluctuations of outgoing longwave radiation and 250 mb streamfunction during northern winter. Mon. Wea. Rev., 113, 941–961. doi: 10.1175/1520-0493(1985)113<0941:IDFOOL>2.0.CO;2.
    [34]

    Whitburn, S., L. Clarisse, S. Bauduin, et al., 2020: Spectrally resolved fluxes from IASI data: Retrieval algorithm for clear-sky measurements. J. Climate, 33, 6971–6988. doi: 10.1175/JCLI-D-19-0523.1.
    [35]

    Wu, X., and J. J. Yan, 2011: Estimating the outgoing longwave radiation from the FY-3B satellite visible infrared radiometer Channel 5 radiance observations. Chinese Sci. Bull., 56, 3480–3485. doi: 10.1007/s11434-011-4686-6.
    [36]

    Yang, G.-Y., and J. Slingo, 2001: The diurnal cycle in the tropics. Mon. Wea. Rev., 129, 784–801. doi: 10.1175/1520-0493(2001)129<0784:TDCITT>2.0.CO;2.
    [37]

    Yang, J., C. H. Dong, N. M. Lu, et al., 2009: FY-3A: The new generation polar-orbiting meteorological satellite of China. Acta Meteor. Sinica, 67, 501–509. doi: 10.11676/qxxb2009.050. (in Chinese)
    [38]

    Yang, J., Z. Q. Zhang, C. Y. Wei, et al., 2017: Introducing the new generation of Chinese geostationary weather satellites, Fengyun-4. Bull. Amer. Meteor. Soc., 98, 1637–1658. doi: 10.1175/BAMS-D-16-0065.1.
    [39]

    Zhang, P., Q. F. Lu, X. Q. Hu, et al., 2019: Latest progress of the Chinese meteorological satellite program and core data processing technologies. Adv. Atmos. Sci., 36, 1027–1045. doi: 10.1007/s00376-019-8215-x.
  • Jian LIU and Wanchun ZHANG.pdf

  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Evaluation of Reprocessed Fengyun-3B Global Outgoing Longwave Radiation Data: Comparison with CERES OLR

    Corresponding author: Jian LIU, liujian@cma.gov.cn
  • 1. Key Laboratory of Radiometric Calibration and Validation for Environmental Satellites, National Satellite Meteorological Center, China Meteorological Administration, Beijing 100081
  • 2. Fengyun Meteorological Satellite Innovation Center, China Meteorological Administration, Beijing 100081
  • 3. National Climate Center, China Meteorological Administration, Beijing 100081
Funds: Supported by the National Key Research and Development Program of China (2018YFB0504900 and 2018YFB0504905) and National Natural Science Foundation of China (41801278)

Abstract: Outgoing longwave radiation (OLR) at the top of the atmosphere (TOA) is a key parameter for understanding and interpreting the relationship between clouds, radiation, and climate interactions. It has been one of the operational products of the Fengyun (FY) meteorological satellites. OLR accuracy has gradually improved with advancements in satellite payload performance and the OLR retrieval algorithm. Supported by the National Key R&D Program Retrospective Calibration of Historical Chinese Earth Observation Satellite data (Richceos) project, a long-term OLR climate data record (CDR) was reprocessed based on the recalibrated Level 1 data of FY series satellites using the latest OLR retrieval algorithm. In this study, Fengyun-3B (FY-3B)’s reprocessed global OLR data from 2010 to 2018 were evaluated by using the Clouds and the Earth’s Radiant Energy System (CERES) global daily OLR data. The results showed that there was a high consistency between the FY-3B instantaneous OLR and CERES Single Scanner Footprint (SSF) OLR. Globally, between the two CDR datasets, the correlation coefficient reached 0.98, and the root-mean-square error (RMSE) was approximately 8–9 W m−2. The bias mainly came from the edge regions of the satellite orbit, which may be related to the satellite zenith angle and cloud cover distribution. It was shown that the long-term FY-3B OLR had temporal stability compared to CERES OLR long-term data. In terms of spatial distribution, the mean deviations showed zonal and seasonal characteristics, although seasonal fluctuations were observed in the differences between the two datasets. Effects of FY-3B OLR application to the South China Sea monsoon region and ENSO were demonstrated and analyzed, and the results showed that the seasonal deviation of FY-3B’s OLR comes mainly from the retrieval algorithm. However, it has little effect on the analysis of climate events.

FY-3B全球OLR再处理数据集评估:与CERES OLR数据集对比

大气顶出射长波辐射(OLR)是研究云、辐射和气候相互作用关系的重要参量,也是风云气象卫星的业务产品之一。随着卫星有效载荷性能和OLR算法的更新迭代,风云卫星OLR产品精度也逐渐提高。在国家重点研发计划项目的支持下,利用最新OLR算法,基于风云卫星再定标L1级数据,对OLR产品进行一致性处理,生产OLR长时间气候数据集。本研究采用云和地球辐射能量系统(CERES)全球OLR日产品评估FY-3B卫星2010年至2018年的再处理全球OLR数据。结果显示,FY-3B的瞬时OLR和CERES OLR之间有很高的一致性。在全球范围内,两个数据集之间的相关系数达到0.98,均方根误差(RMSE)约为8–9 W m−2。偏差主要来自于卫星轨道的边缘区域,这可能与卫星天顶角和云量分布有关。长序列数据对比显示,FY-3B的长序列OLR产品具有较好的时间稳定性;在空间分布方面,与CERES OLR相比,平均偏差呈现出纬度带和季节性特征。对南海季风区和ENSO区分析表明,FY-3B OLR数据集的季节偏差主要来自于算法;然而,它对区域气候事件的分析影响不大。
    • Outgoing longwave radiation (OLR) at the top of the atmosphere (TOA) is the energy flux density of thermal radiation emitted by the earth’s atmospheric system into outer space. OLR is primarily determined by surface temperature, water vapor, and cloud cover (Gruber and Krueger, 1984; Schmetz and Liu, 1988; Yang and Slingo, 2001; Clerbaux et al., 2003, 2020; Bowman et al., 2013). OLR is an essential parameter of the earth’s radiation budget. It is critical to study interactions between clouds and radiation and improve surface and atmospheric parameterization in models (Inoue and Ackerman, 2002; Kim et al., 2018; Kim and Lee, 2019). It is also one of the key factors to understand deep tropical convection (such as typhoon and precipitation), Walker circulation, El Niño–Southern Oscillation (ENSO), Madden–Julian oscillation, monsoons, and so on (Nakazawa, 2006; Matthews, 2008; Susskind et al., 2012; Chiodi and Harrison, 2013, 2015; Kiladis et al., 2014; Singh et al., 2014; Stechmann and Ogrosky, 2014; Huang et al., 2020; Liu et al., 2020; Liu et al., 2021).

      NOAA has been releasing NOAA Advanced Very High Resolution Radiometer (AVHRR) OLR products since 1979. The AVHRR OLR includes daily and monthly data with 2.5° × 2.5° spatial resolution. The retrieval algorithm is mainly based on AVHRR infrared window channels (Gruber and Krueger, 1984; Stowe et al., 2002; Clerbaux et al., 2020). NOAA National Environmental Satellite, Data, and Information Service (NESDIS) developed an OLR retrieval algorithm based on multiple channels [including infrared (IR) window channel, temperature and water vapor channels] in 1998. NESDIS has been releasing daily/monthly average High Resolution Infrared Radiation Sounder (HIRS) OLR since 1979. The spatial resolution of this dataset is 1° × 1° (Lee et al., 2007; Lee, 2014). In general, the OLR derived from broadband observations (3–100 µm) has higher accuracy. The Clouds and the Earth’s Radiant Energy System (CERES) onboard Aqua/Terra uses a wide-band direct observation method to obtain shortwave radiation, longwave radiation, and cloud cover data at the TOA. The longwave radiation observation accuracy of CERES is greater than 0.5% compared to the Earth Radiation Budget Experiment (ERBE) instrument. Twice-daily observations from polar-orbiting satellites do not accurately represent daily average conditions. Thus, the CERES (SYN1deg) product was developed by fusing CERES data with high temporal resolution observation data from geostationary satellites, reducing uncertainty by 20% (Priestley et al., 2007). This dataset is currently considered to be the OLR product with the best quality. It is widely used to verify the accuracy of other OLR data (Short and Cahalan, 1983; Weickmann et al., 1985; Susskind et al., 2012; Bowman et al., 2013; Kiladis et al., 2014). In addition to CERES data, HIRS OLR and AVHRR OLR climate data record (CDR) data are also available for research although they are narrowband estimation data. Compared with CERES OLR, the average deviation of the HIRS OLR is less than 3 W m−2 in most regions (Lee et al., 2007). The deviation between the AVHRR OLR and CERES OLR is approximately −9 W m−2 (Lee et al., 2007).

      Fengyun-3 (FY-3) satellites are second-generation polar-orbiting meteorological satellites developed by China (Zhang et al., 2019). Since the launch of the FY-3A satellite in 2008, five satellites (FY-3A/B/C/D/E) have been launched (Yang et al., 2009). During this period, the FY-3 instrument operational calibration and OLR retrieval algorithms were continuously updated. These factors have resulted in the inconsistent accuracy of operational products in different periods. A unified retrieval algorithm was used to reprocess the OLR datasets based on the recalibration Level 1B data of FY-3 satellites. In this study, we performed a statistical evaluation of the reprocessed FY-3B OLR data (hereafter referred to as FY-3B OLR) from the perspective of long-sequence stability and application. This article is structured in the following manner. Data and algorithm are introduced in Section 2. The evaluation of reprocessed FY-3B OLR datasets and the main characteristics of the error distributions are presented in Section 3. The effects of FY-3B OLR application on climate applications have been discussed in Section 4. The summary and conclusions are presented in Section 5.

    2.   Data and method
    • This study used the visible infrared scanning radiometers (VIRR) instrument boarded on the FY-3B satellite to retrieve the OLR. This instrument has 10 spectral channels with a spatial resolution of 1 km × 1 km. Among these channels, the window channel (11.8 μm) is used to retrieve OLR. The FY-3B OLR data include daily and monthly products with a horizontal resolution of 0.05° × 0.05°. The data from 5 December 2010 to 30 November 2018 were analyzed in this study. Details of the full dataset can be found in Table A1 of the Appendix.

      CERES OLR data were used to assess the accuracy of the FY-3B OLR data. CERES is widely used in radiation budget research and provides global observations of TOA OLR within a 5–100-µm range. It is a 3-channel radiometer measuring reflected solar radiation in the 0.3–5-µm wavelength spectral region, emitted terrestrial radiation in the 8–12-µm bands, and total radiation from 0.3 to beyond 100 µm. The first CERES instrument was launched on the Tropical Rainfall Measuring Mission satellite in November 1997. The second and third CERES instruments were launched on the Terra satellite in December 1999, and the fourth and fifth CERES instruments were launched on the Aqua satellite. Two types of CERES OLR data were used in this study. One is the CERES Single Scanner Footprint Edition4A product (SSF OLR) based on the Aqua satellite orbit observations. These data were constantly observed at a spatial resolution of 20 km within the CERES field of view (FOV). The other type is the CERES SYN1deg Ed4A daily OLR product (SYN1deg OLR), which involved a high level of data fusion with a horizontal resolution of 1° × 1°. CERES SSF OLR and SYN1deg OLR were used to assess FY-3B instantaneous OLR and daily OLR, respectively.

      In the instantaneous orbit data analysis, CERES SSF OLR and FY-3B OLR had different spatial resolutions and observation times. Spatial and temporal matching of the data had to be performed before comparing the two OLR datasets. Data whose observation time difference was within 40 min were matched. Both data were projected onto a grid with a spatial resolution of 0.5° × 0.5°. Matching samples from day and night were selected separately for analysis. In addition, Aqua/CERES total cloud fraction data were used in this study, with a spatial resolution of 1° × 1°. Hourly OLR data with a resolution of 0.05° × 0.05° from a geostationary satellite FY-4A were also used. The data covered 50°S–50°N and 55°–155°E area (Yang et al., 2017). The total column water (TCW) and the atmospheric profiles of temperature and humidity data of the ECMWF Reanalysis version 5 (ERA5) were used in the evaluation and model simulation work, respectively. The properties of all research data are presented in Table 1.

      DataSpatial resolutionTemporal coverage
      Satellite dataOLRFY-3B VIRR0.05° × 0.05°2010-12-05 to 2019-12-31
      OLRFY-4A Advanced Geostationary Radiation Imager (AGRI)0.05° × 0.05°2017-01-08 to present
      SSF OLRAqua CERES~20 km (FOV)2002-07-03 to present
      SYN1deg OLRAqua/Terra CERES1° × 1°2000-03-01 to present
      Total cloud area fractionAqua CERES1° × 1°2002-07-01 to present
      Reanalysis dataTemperature profileECMWF ERA50.25° × 0.25°1979-01-01 to present
      Humidity profileECMWF ERA50.25° × 0.25°1979-01-01 to present
      Column total waterECMWF ERA50.25° × 0.25°1979-01-01 to present

      Table 1.  Specifications of the data used in this study

    • Many studies have calculated TOA OLR by converting the observed radiation flux to broadband radiation flux with a narrowband sensor (Ohring et al., 1984; Stowe et al., 2002; Lee et al., 2007). In these studies, the algorithms mainly use the brightness temperature of the satellite window channel (Gruber and Krueger, 1984; Whitburn et al., 2020), and some algorithms also consider the water vapor channels (Lee et al., 2007; Lee, 2014; Clerbaux et al., 2020). The VIRR on board the FY-3B polar-orbiting meteorological satellite is an instrument similar to the NOAA satellite’s AVHRR. VIRR has split-window channels, in which channels 4 (10.8 μm) and 5 (11.8 μm) are similar to the AVHRR split-window channels. The VIRR OLR retrieval algorithm uses the classic AVHRR channel 5 brightness temperature algorithm to calculate the OLR at the TOA (Ohring et al., 1984; Liebmann and Smith, 1996; Wu and Yan, 2011). First, the OLR retrieval model was established based on infrared radiation transfer simulations. Global 2521 ERA5 atmospheric profiles from 2010 were used for regression analysis. Each profile had 101 layers. The parameters include temperature, water vapor mixing ratio, ozone volume mixing ratio, surface temperature, and surface pressure. These data represent various weather conditions, such as clear sky, cloudy, and so on. The flux equivalent brightness temperature (TF) and channel 5 radiance (TB5) of the VIRR were simulated for each sounding profile based on Line-by-Line Radiative Transfer Model (LBLRTM). Then, a nonlinear relationship between the two parameters was established as follows:

      $$ {T}_{{\rm{F}}}=A+B{T}_{{\rm{B}}5}+C{T}_{{\rm{B}}5}^{2}, $$ (1)

      where A, B, and C are coefficients. Finally, the longwave radiation flux was calculated from the TF by using the Planck function.

    3.   Results
    • FY-3B instantaneous OLR data on 22 August 2018 were analyzed. Matching of CERES and FY-3B data revealed 9552 matched day samples. The matched data were concentrated over the polar regions. Among these samples, 89.7% of the samples had an OLR difference less than 10 W m−2, with an average bias of 1.63 W m−2, root-mean-square error (RMSE) of 8.57 W m−2, and correlation coefficient of 0.98. There were 13,025 matched samples at night. Among these samples, 92% of the samples had an OLR difference less than 10 W m−2, with an average bias of 1.38 W m−2, RMSE of 8.45 W m−2, and correlation coefficient of 0.98. These statistics reveal appreciable consistency between the instantaneous FY-3B OLR and CERES OLR.

      The spatial distribution of deviation was analyzed to understand the difference between the FY-3B and CERES instantaneous data. Figure 1 shows the geographical distributions of the instantaneous deviations between FY-3B OLR and CERES SSF OLR during the day and night on 22 August 2018. In general, the difference between these two datasets ranged from −20 to 20 W m−2. The relatively larger deviation areas mainly appeared at the edges of the orbit, which had a large satellite zenith angle. Moreover, the lower retrieval accuracy could have been caused by limb darkening. In addition, cloud cover could also have affected the spatial distribution of the differences. To further explain the influence of cloud cover and satellite zenith angle on deviation, we compared the statistical RMSE and bias of FY-3B OLR with that of CERES SSF OLR as a function of the satellite zenith angle and cloud cover (Fig. 2). When the satellite zenith angle was less than 60° (Fig. 2a), the RMSE ranged from 6 to 9 W m−2, and the mean bias was approximately 1–4 W m−2. However, when the zenith angle was greater than 60°, the RMSE was greater than 11 W m−2, and the mean bias was approximately −3 W m−2. Although the limb darkening correction was carried out when the zenith angle was greater than 60°, deviations may have existed in the corrected model, causing to the bias increase significantly with the increase in the satellite zenith angle. We also considered the influence of cloud cover distribution. Figure 2b shows the corresponding total cloud fraction statistical results. The RMSE was approximately 9 W m−2, and the bias was within 2 W m−2 when the total cloud fraction was larger than 80%. It should be noted that in the case of more cloud cover, the inversion consistency between FY-3B and CERES OLR is higher. However, when the total cloud cover is less than 80%, the consistency between the two OLR datasets is unstable, which may be related to the small sample size.

      Figure 1.  Instantaneous deviations between FY-3B OLR and CERES SSF OLR during (a) the day and (b) the night on 22 August 2018.

      Figure 2.  Factors influencing the spatial distribution of the instantaneous deviations between FY-3B OLR and CERES SSF OLR. (a) Satellite zenith angle influence and (b) cloud cover influence. The blue dotted line represents the average bias, the red dotted line indicates the average RMSE, and the gray bar indicates the number of samples considered in the statistical analysis.

    • In this section, FY-3B daily OLR data from 5 December 2010 to 30 November 2018 were evaluated by using the CERES SYN1deg OLR (hereafter referred to as CERES OLR) dataset. The probability density distributions of the global daily CERES OLR and FY-3B OLR from 2010 to 2018 are shown in Fig. 3. Both datasets show a bimodal structure, with a bimodal dividing point of 225 W m−2. The FY-3B OLR ranged from 216 to 232 W m−2. The CERES OLR ranged from 216 to 230 W m−2. For the FY-3B OLR, the mean value of FY-3B OLR was 223.84 W m−2, the coefficient of skewness was −30.62, and the coefficient of kurtosis was 943.66. For the CERES OLR, the mean was 222.17 W m−2, the coefficient of skewness was −30.67, and the coefficient of kurtosis was 945.82.

      Figure 3.  The probability density distribution of the daily global mean (a) CERES OLR and (b) FY-3B OLR.

      The global daily statistics time series of the FY-3B OLR and CERES OLR is shown in Fig. 4. The RMSE of FY-3B OLR ranged from 11 to 14 W m−2 (Fig. 4a). The global mean bias of the FY-3B OLR was approximately 3 W m−2 (Fig. 4b). The correlation coefficient was approximately 0.96. Table 2 presents the annual mean statistics of the daily FY-3B OLR. Except for 2010 and 2018 (years for which the statistics are not available for the entire year), the statistics were mostly stable.

      Figure 4.  Time series of the (a) RMSE, (b) bias, (c) correlation coefficients, and (d) daily mean OLR of the global daily FY-3B OLR and CERES OLR data from 2010 to 2018.

      YearMEAN (Mean/Std)MAX (value)MIN (value)RMSE (Mean/Std)Bias (Mean/Std)R (Mean/Std)STD (Mean/Std)
      2010223.42/1.11225.53221.1212.07/0.281.49/0.370.968/0.0039.00/0.23
      2011225.56/3.34232.73214.0012.57/0.762.29/0.650.963/0.0069.41/0.53
      2012225.44/2.88231.85219.9812.67/0.752.23/0.730.962/0.0069.44/0.51
      2013225.25/2.88231.73218.9612.44/0.821.88/0.760.963/0.0069.32/0.63
      2014225.08/2.85230.68219.7512.45/0.811.66/0.680.963/0.0069.32/0.59
      2015224.64/2.79230.72219.6912.43/0.831.48/0.700.964/0.0069.30/0.59
      2016225.40/2.62231.43220.4312.60/0.891.28/0.720.962/0.0069.45/0.59
      2017224.92/2.49230.59220.3912.44/0.991.20/0.640.963/0.0069.32/0.70
      2018224.59/2.92230.68219.5212.60/1.161.14/0.600.962/0.0069.39/0.82
      Note: MEAN (Mean/Std) indicates the mean value and mean deviation (W m−2) of the FY-3B daily mean OLR time series; MAX/MIN (value) indicates the maximum/minimum value (W m−2) of the FY-3B daily mean OLR time series; and RMSE/Bias/R/STD indicates the root-mean-square error (W m−2)/the mean deviation (W m−2)/the correlation coefficient/the standard deviation (W m−2) of FY-3B OLR compared to CERES OLR, respectively.

      Table 2.  The annual average statistics of FY-3B OLR compared to those of CERES SYN1deg OLR

      Figure 5 shows the spatial distributions of the annual mean CERES OLR and FY-3B OLR from 2010 to 2018. The spatial characteristics of the two datasets have similar distribution patterns. The difference in OLR between CERES and FY-3B (Fig. 5c) indicates that FY-3B OLR was higher in northern Africa and lower in the South Pacific and South Atlantic (compared to CERES OLR). We used the ERA5 TCW data to further analyze the reasons for the distribution pattern. Considering the small threshold range of this parameter in the polar region (all less than 30 kg m−2), we examined the daily average global OLR (two days in winter and two days in summer) with TCW as a bin (Fig. 6). The results show that water vapor affects the accuracy of the OLR. The bias of the FY-3B OLR compared to CERES OLR was smaller when the TCW was between 10 and 40 kg m−2. That was larger when the TCW was less than 10 kg m−2 and more than 40 kg m−2. Here, TCW includes water vapor, clouds, and precipitation.

      Figure 5.  Climate analysis through evaluation of annual mean OLR over the globe during 2010–2018 in the (a) CERES OLR dataset and (b) FY-3B OLR dataset, and (c) OLR difference between the OLR means of CERES and FY-3B.

      Figure 6.  Influence of TCW on the daily OLR comparisons. The color bars indicate the date.

      Singular value decomposition (SVD) technology was used to analyze the climatic correlation between the two datasets. CERES daily OLR was set as the left field, and FY-3B OLR was set as the right field. The cumulative variance of the first mode reached 98.2%, indicating that the first mode of the SVD result could represent the OLR distribution. Figure 7 shows the time coefficient of the first mode of SVD and its spatial correlation. The time coefficient correlation was 0.93, and the divergence was small, indicating that FY-3B OLR is strongly consistent with CERES OLR in climatology.

      Figure 7.  (a) The time coefficient and (b) the spatial correlation of the first mode of the SVD.

      Figure 4 shows seasonal fluctuations in FY-3B OLR (compared to CERES OLR). In January and July, the RMSE was the largest, while the correlation coefficient was the lowest. In April and October, the RMSE was the smallest, while the correlation coefficient was the highest. The fact that the CERES OLR is designed to provide the highest temporal resolution TOA flux dataset by incorporating hourly GEO (Geostationary Earth Orbits) imager data is a possible reason for the fluctuations (Knapp et al., 2011; Taylor, 2012). However, FY-3B OLR only uses two observations per day to produce daily mean OLR, which is not sufficient to represent diurnal variation. To illustrate the influence of the number of observations on the daily average data, we used the FY-4A geostationary meteorological satellite data. Figure 8 shows the difference in the daily mean OLR for each month of 2020 calculated using the FY-4A OLR at different observation frequencies. The observation frequency is 1 and 12 h, respectively. The two-time points of 12 h correspond to the FY-3B observation time, which is about 0130 and 1330 local time. There are differences in the daily average OLR calculated by the two different sampling frequencies in different months. The magnitude of the difference may be related to changes in weather systems in the region.

      Figure 8.  The difference (OLR2 − OLR24) in the daily mean OLRs calculated in each month of 2020 using FY-4A OLR. FY-4A OLR data were recorded on an hourly basis (OLR24) as well as at two-time points (OLR2; 0130 and 1330 local time).

      To further evaluate the FY-3B OLR, the spatial distributions of the deviations were analyzed for different seasons. The results are presented in Fig. 9. The statistical results show that the global average RMSE was 11.61 W m−2 in December–January–February (DJF), 12.03 W m−2 in March–April–May (MAM), 13.10 W m−2 in June–July–August (JJA), and 11.69 W m−2 in September–October–November (SON). The global mean biases were 0.86 W m−2 (DJF), 1.90 W m−2 (MAM), 2.06 W m−2 (JJA), and 1.82 W m−2 (SON). Figures 9a–d show the spatial distributions of RMSE with 1° × 1° spatial resolution in different seasons. In general, the RMSE had a belt-distribution pattern with latitude. In spring and autumn, the RMSE was relatively lower in most regions; however, it was high mainly over central Africa. In winter and summer, higher RMSE areas appeared over multiple regions, such as regions between 60° and 80°S. Central and northern Africa had relatively higher RMSE in summer, and central Africa had higher RMSE in winter. Figures 9e–h show the spatial distributions of the average bias in different seasons. The belt-distribution pattern of bias with latitude was clearer. Different seasons had different distribution characteristics. In autumn and winter, the Northern Hemisphere was dominated by a positive bias, where the bias of most regions above 30°N was greater than 5 W m−2, and the Southern Hemisphere was dominated by a negative bias. A positive bias in summer occurred mainly over the Southern Hemisphere. In spring, positive bias belts existed between 30° and 60° north and south latitudes. The low-latitude equatorial area was mainly characterized by a negative bias in each season, especially in JJA. Over the high latitudes of the Northern Hemisphere in winter and the Southern Hemisphere in summer, the high positive deviations may have been related to satellite observations (Hu, 2012). The negative deviations in low latitudes, especially in the low latitudes of the Southern Hemisphere, may have been caused by the OLR retrieval algorithm. Lee et al. (2007) found that a single IR window-based OLR retrieval algorithm would cause instability in certain regions, particularly in the form of biases, because of deviations in water vapor content estimation.

      Figure 9.  Spatial distributions of the statistical characteristics in different seasons: (a–d) RMSE and (e–h) bias.

    4.   Intercomparisons for application demonstration
    • OLR is widely used in climate monitoring, i.e., in the South China Sea summer monsoon (SCSSM) and ENSO.

      The SCSSM, an important system that affects the characteristics of summer weather in China, has been extensively studied by meteorologists for a long time. There is a wide consensus that precipitation begins to increase significantly after the onset of the SCSSM in China. Therefore, the monitoring of the SCSSM is an essential part of the annual climate service at the National Climate Center (NCC) (Ren et al., 2018). The daily regional average OLR dropping below 230 W m−2, its anomaly becoming negative, and the two previously listed conditions lasting for more than five days indicate the onset of SCSSM (Chen, 2006). Figure 10 shows the daily mean OLR and OLR anomaly time series in the SCSSM monitoring key regions (10°–20°N, 110°–120°E) from 20 April to 10 June 2016. As shown in Fig. 10a, the mean OLR in this region was less than 230 W m−2 starting on 20 May and the OLR anomaly over this region became negative (Fig. 10b). These results show that the SCSSM began on 20 May. These dates coincide with the time predicted for SCSSM onset in 2016 by the NCC, suggesting that the FY-3B OLR and CERES OLR can simultaneously and accurately indicate the onset of the SCSSM.

      Figure 10.  Time series of (a) the daily average OLR and (b) OLR anomalies in key region (10°–20°N, 110°–120°E) of SCSSM monitoring from 20 April to 10 June 2016. The red line and blue line represent FY-3B OLR and CERES OLR, respectively.

      ENSO is an important mode of air–sea interaction in the tropical Pacific. OLR is a good indicator of ENSO monitoring. It is well known that during a single El Niño (La Niña) event, large negative or positive OLR anomalies are observed in the equatorial central Pacific (Wang et al., 1999). Figure 11 shows the time series of the monthly mean OLR in the Niño-3 region (Fig. 11a; 5°S–5°N, 90°–150°W) and Niño-4 region (Fig. 11b; 5°S–5°N, 160°E–150°W) from January 2011 to November 2018. The FY-3B OLR and CERES OLR were in agreement with the low value of the OLR oscillation in the Niño-3 region and the overall range in the Niño-4 region. However, the FY-3B OLR had been considerably underestimated in the high value of the OLR oscillation in the Niño-3 region. These results are consistent with the comparison between the AVHRR OLR and CERES OLR in Lee et al. (2007). The FY-3B OLR retrieval algorithm is similar to that of the AVHRR OLR, which uses a single channel to establish the regression coefficients. This method is considered to be insensitive to water vapor changes and atmospheric changes (such as temperature inversion), which may cause large deviations (Gruber and Krueger, 1984). Lee et al. (2007) explained that the Niño-3 region in the El Niño year and Niño-4 region were always overcast, and the single-channel OLR was consistent with the broadband OLR.

      Figure 11.  Time series of the monthly average FY-3B OLR (red) and CERES OLR (blue) in the (a) Niño-3 region and (b) Niño-4 region from January 2011 to November 2018.

      Figure 12 shows the time series of the FY-3B OLR anomaly and CERES OLR anomaly in the Niño-3 region (Fig. 12a) and Niño-4 region (Fig. 12b) from January 2011 to November 2018. The anomaly value was calculated monthly to remove seasonal influence. The FY-3B OLR and CERES OLR were in agreement in these two regions, indicating that the algorithm-determined systematic deviation of the OLR in Fig. 11 was seasonal. This difference was eliminated in the deseasonal anomaly.

      Figure 12.  Time series of the monthly FY-3B OLR (red) and CERES OLR (blue) anomaly in the (a) Niño-3 region and (b) Niño-4 region from January 2011 to November 2018.

    5.   Conclusions
    • Based on the reprocessed Level 1 data, the FY-3 OLR was reprocessed by using an updated retrieval algorithm that used the atmospheric profile data from the ECMWF reanalysis field. This study used CERES OLR to evaluate the reprocessed FY-3B OLR products in terms of product accuracy, long-term stability, and application in climate monitoring.

      The instantaneous comparison results showed that FY-3B OLR was consistent with CERES SSF OLR, and only parts of the observed orbital edge regions have large deviations. Globally, the FY-3B OLR corresponded with CERES OLR possessing an accuracy within 12 W m−2 and precision of approximately 1 W m−2. The correlation coefficient between FY-3B and CERES global daily mean OLR was 0.96. In the temporal and spatial distributions of FY-3B OLR and CERES OLR, the deviation appeared as a zonal belt pattern with seasonal variations, which might be related to the diurnal variation in OLR. By comparing the climate monitoring capabilities of the two datasets, it was determined that the time series of the FY-3B OLR anomaly exhibits excellent consistency with CERES data over the SCSSM monitoring region and the ENSO monitoring regions. The results showed that although FY-3B OLR and CERES OLR have deviations, in regional climate event monitoring, FY-3 OLR and CERES OLR have the same monitoring capabilities.

      A comprehensive analysis of the completed FY-3B single-satellite OLR showed the reliability of the reprocessed data. However, when the long-term dataset is processed in the future, some important questions need to be addressed concerning intersatellite calibration, fluctuations in different seasons, and drift in equator crossing time (ECT). More detailed analyses of long-term FY OLR data will be presented in subsequent papers.

      Acknowledgments. The authors are grateful to the LARC data center (https://ceres.larc.nasa.gov/) and the ECMWF (https://www.ecmwf.int/) for providing the CERES data and the ERA5 data used in this work. The authors wish to acknowledge the Editor Dr. Jun Li and two anonymous reviewers for their comments that have helped improve the manuscript.

    Appendix
    • FieldContent
      Title of datasetFY-3B VIRR Reprocessed Daily Outgoing Longwave Radiation (OLR) Dataset
      AbbreviationFY3VIRRReproDailyOLR
      Retrieval and rangeData are retrieved by a unified inversion algorithm based on the FY-3A/B/C recalibrated VIRR L1 data, with values ranging from 0 to 500 W m−2
      Main data producerWanchun ZHANG
      Email: zhangwc@cma.gov.cn
      Tel: +86-10-68407237
      Spatial coverage0°–360°, 90°S–90°N
      Temporal coverageFrom 5 December 2010 to 31 December 2019
      ResolutionsSpatial resolution: 0.05° × 0.05°;
      Temporal resolution: daily/monthly
      Format and sizeHDF; approximately 150 M every day
      Data publishing unitNational Satellite Meteorological Center, China Meteorological Administration
      Funding projectNational Key Research and Development Program of China (2018YFB0504905)
      DOI (accessible at) 10.12185/NSMC.RICHCEOS.FCDR.FY3VIRRReproDailyOLR.FY3.VIRR.L2.GBAL.POAD.GLL.5000M.HDF.2021.3.V2

      Table A1.  The reprocessed FY-3B OLR dataset

Reference (39)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return