Evaluation of FY-3/VIRR Sea Surface Temperature Data for Climate Applications

风云三号卫星可见光红外辐射计海表温度资料的气候应用评估

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

    Supported by the Guangdong Major Project of Basic and Applied Basic Research (2020B0301030004) and National Key Reseach and Development Program of China (2018YFB0504905 and 2018YFB0504900)

  • doi: 10.1007/s13351-021-1055-5
  • Note: This paper will appear in the forthcoming issue. It is not the finalized version yet. Please use with caution.

PDF

  • We evaluated the sea surface temperature (SST) products derived from the visible infrared radiometer on board the Fengyun-3 satellites (FY-3/VIRR) during 2016–2018 from the perspective of climate applications. The data had previously been reprocessed by the National Satellite Meteorological Center of China Meteorological Administration based on an updated SST retrieval algorithm. The overall consistency between the FY-3/VIRR SST data and the optimum interpolation SST version 2 (OIv2.1) was better for monthly means than for pentad means, and showed a clear dependence on the season and location. There was better consistency in winter than in summer, and in the tropical central–eastern Pacific than in the western Pacific warm pool, tropical North Indian Ocean, and tropical Atlantic Ocean. The monthly deviation of the global average SST anomaly was −0.03 ± 0.07°C and the average root-mean-square errors (RMSEs) presented clear seasonal fluctuations with a maximum of approximately 0.5°C in summer. The poor consistency of the FY-3/VIRR SST in summer may be partially attributed to the bias of the OIv2.1 data in global oceans (especially the Indian Ocean) as a result of the spatially heterogeneous in situ measurements from ships, buoys, and Argo floats. Convective activities and clouds in the tropics may also influence the accuracy of the FY-3/VIRR SST retrievals. The Niño SST indices based on both FY-3/VIRR and OIv2.1 SST data displayed a generally similar evolution, including the start and end of El Niño and La Niña events and their amplitudes, although the deviations were slightly larger when the Pacific SST anomaly was in the neutral state of the El Niño–Southern Oscillation (ENSO). The deviations varied greatly with season in the tropical Indian and Atlantic oceans, suggesting the need to perform further analyses and validation of the FY-3/VIRR SST products in these two basins.
    本文基于国家卫星气象中心提供的一套利用最新的海温反演算法再处理的2016–2018年风云三号卫星可见光红外辐射计(FY-3/VIRR)海表温度产品,以最优插值海表温度数据OIv2.1作为标准,从气候应用的角度对其侯平均和月平均尺度产品进行了质量评估。FY-3/VIRR的月平均产品总体优于候平均产品,与OIv2.1的一致性表现出季节和空间差异性:冬季较好,夏季较差;热带中东太平洋较好,暖池、热带北印度洋和大西洋较差。全球平均的海温偏差为−0.03 ± 0.07°C,均方根误差呈现季节性波动,其中夏季最大,约0.5°C。夏季FY-3/VIRR海温在全球平均和印度洋区域与OIv2.1存在较大偏差,部分原因可能来自于船舶、固定浮标和Argo浮标的空间非均匀测量,而热带地区较多的对流活动和云覆盖也可能影响FY-3/VIRR海温反演的精度。基于FY-3/VIRR得到的Niño海温指数总体上与OIv2.1接近,能够反映ENSO事件的开始、结束和强度。在热带印度洋和大西洋两种海温资料之间的偏差随季节变化很大。因此,有必要对这两个区域的FY-3/VIRR海温产品做进一步的质量分析和验证。
  • 加载中
  • Fig. 1.  Time series of the monthly Niño3.4 index (°C) based on the OIv2.1 SST data during 2016–2018. The labels on the x-axis refers to month and year; for example, Jan-16 denotes January 2016; same applies to other relevant figure labels in this paper.

    Fig. 2.  Time series of the anomaly correlation coefficient (ACC) of (a) monthly and (b–d) pentad SSTAs averaged over the region 70°S–70°N, 0°–360° between the FY-3/VIRR and OIv2.1 data in (a) 2016–2018, (b) 2016, (c) 2017, and (d) 2018, respectively. The gray line indicates the average ACC. The x-axis labels in (a) denote the same as in Fig. 1, while those in (b–d) denote pentad-month-year.

    Fig. 3.  Distribution of temporal correlation coefficient of the monthly SSTAs between the FY-3/VIRR and OIv2.1 data from 2016 to 2018 (annual cycles were subtracted from the original data).

    Fig. 4.  Distributions of temporal correlation coefficient of the pentad SSTAs between the FY-3/VIRR and OIv2.1 data in (a) 2016, (b) 2017, and (c) 2018, respectively (annual cycles were subtracted from the original data).

    Fig. 5.  Time series of the bias (bar) and RMSE (line) of the monthly SSTA (°C) averaged over the region 70°S–70°N, 0°–360° between the FY-3/VIRR and OIv2.1 data from 2016 to 2018. The x-axis labels denote the same as in Fig. 1.

    Fig. 6.  SSTAs (°C) in (a) OIv2.1 and (b) FY-3/VIRR, and (c) SSTA differences between the FY-3/VIRR and OIv2.1 data, in January 2016.

    Fig. 7.  SSTA differences (°C) between the FY-3/VIRR and OIv2.1 data in July (a) 2016, (b) 2017, and (c) 2018, respectively.

    Fig. 8.  Scatter plots of the deviation/bias in SSTA between FY-3/VIRR and OIv2.1 in the tropical Indian Ocean (y-axis) versus the OIv2.1 SSTA (x-axis) in the same region in (a) January 2016, (b) July 2016, (c) July 2017, and (d) July 2018, respectively.

    Fig. 9.  Time series of the monthly mean tropical Pacific Ocean SST indices: (a) Niño3.4, (b) Niño3, (c) Niño4, and (d) NiñoW based on the FY-3/VIRR (red line) and OIv2.1 (blue line) data and related biases (bar; °C), respectively. The x-axis labels denote the same as in Fig. 1.

    Fig. 10.  As in Fig. 9, but for the tropical Indian Ocean SST indices (a) IOBM and (b) IOD.

    Fig. 11.  As in Fig. 9, but for the tropical Atlantic Ocean SST indices (a) equatorial Atlantic Niño (ATL3), (b) tropical North Atlantic (TNA), (c) tropical South Atlantic (TSA), and (d) Tropical Atlantic Dipole (TAD).

    Table 1.  Definitions of the selected tropical SSTA indices. Note: TNA—Tropical North Atlantic, TSA—Tropical South Atlantic, TAD—Tropical Atlantic Dipole; search text for other accronyms

    SST indexDefinition
    Niño3.4SSTA (5°S–5°N, 170°–120°W)
    Niño3SSTA (5°S–5°N, 150°–90°W)
    Niño4SSTA (5°S–5°N, 160°E–150°W)
    NiñoWSSTA (0°–10°N, 120°–140°E)
    IODSSTA (10°S–10°N, 50°–70°E)
    − SSTA (10°S–0°, 90°–110°E)
    IOBMSSTA (20°S–20°N, 40°–110°E)
    ATL3SSTA (3°S–3°N, 20°W–0°)
    TNASSTA (5°–20°N, 60°–30°W)
    TSASSTA (20°S–0°, 30°W–10°E)
    TADSSTATNA − SSTATSA
    Download: Download as CSV
  • [1]

    Annamalai, H., S.-P. Xie, J. P. McCreary, et al, 2005: Impact of Indian Ocean sea surface temperature on developing El Niño. J. Climate, 18, 302–319. doi: 10.1175/JCLI-3268.1.
    [2]

    Brasnett, B., and D. S. Colan, 2016: Assimilating retrievals of sea surface temperature from VIIRS and AMSR2. J. Atmos. Oceanic Technol., 33, 361–375. doi: 10.1175/JTECH-D-15-0093.1.
    [3]

    Chambers, D. P., B. D. Tapley, and R. H. Stewart, 1999: Anomalous warming in the Indian Ocean coincident with El Niño. J. Geophys. Res. Oceans, 104, 3035–3047. doi: 10.1029/1998JC900085.
    [4]

    Ding, Y. H., Y. Y. Liu, and Z.-Z. Hu, 2021: The record-breaking meiyu in 2020 and associated atmospheric circulation and tropical SST anomalies. Adv. Atmos. Sci. . doi: 10.1007/s00376-021-0361-2.
    [5]

    Donlon, C. J., M. Martin, J. Stark, et al., 2012: The Operational Sea Surface Temperature and Sea Ice Analysis (OSTIA) system. Remote Sens. Environ., 116, 140–158. doi: 10.1016/j.rse.2010.10.017.
    [6]

    Du, Y., L. Yang, and S. P. Xie, 2011: Tropical Indian Ocean influence on Northwest Pacific tropical cyclones in summer following strong El Niño. J. Climate, 24, 315–322. doi: 10.1175/2010JCLI3890.1.
    [7]

    Enfield, D. B., A. M. Mestas-Nuñez, D. A., Mayer, et al., 1999: How ubiquitous is the dipole relationship in tropical Atlantic sea surface temperatures? J. Geophys. Res. Oceans, 104, 7841–7848. doi: 10.1029/1998JC900109.
    [8]

    EPA (U.S. Environmental Protection Agency), 2014: Climate Change Indicators in the United States, 2014. Third edition. EPA 430-R-14-004, 112 pp. Available at www.epa.gov/climatechange/indicators. Accessed on 11 August 2021.
    [9]

    GCOS (Global Climate Observing System), 2011: Systematic Observation Requirements for Satellite-Based Data Products for Climate. 2011 update, GCOS-154. World Meteorological Organization, Geneva, Switzerland, 103 pp.
    [10]

    Guo, Y. J., and Y. Q. Ni, 1998: Effects of the tropical Pacific convective activities on China’s winter monsoon. Meteor. Mon., 24, 3–7. (in Chinese)
    [11]

    Han, Z., S. L. Li, and M. Mu, 2011: The role of warm North Atlantic SST in the formation of positive height anomalies over the Ural Mountains during January 2008. Adv. Atmos. Sci., 28, 246–256. doi: 10.1007/s00376-010-0069-1.
    [12]

    Hu, Z.-Z., R. G. Wu, J. L. Kinter III, et al., 2005: Connection of summer rainfall variations in South and East Asia: role of El Niño–southern oscillation. Int. J. Climatol., 25, 1279–1289. doi: 10.1002/joc.1159.
    [13]

    Hu, Z.-Z., A. Kumar, B. Jha, et al., 2012: An analysis of warm pool and cold tongue El Niños: air–sea coupling processes, global influences, and recent trends. Climate Dyn., 38, 2017–2035. doi: 10.1007/s00382-011-1224-9.
    [14]

    Hu, Z.-Z., A. Kumar, J. S. Zhu, et al., 2019: On the challenge for ENSO cycle prediction: An example from NCEP Climate Forecast System, version 2. J. Climate, 32, 183–194. doi: 10.1175/JCLI-D-18-0285.1.
    [15]

    Hu, Z.-Z., A. Kumar, B. Jha, et al., 2020: How much of monthly mean precipitation variability over global land is associated with SST anomalies? Climate Dyn., 54, 701–712. doi: 10.1007/s00382-019-05023-5.
    [16]

    Huang, B. Y., M. L’Heureux, J. Lawrimore, et al., 2013: Why did large differences arise in the sea surface temperature datasets across the tropical Pacific during 2012? J. Atmos. Oceanic Technol., 30, 2944–2953. doi: 10.1175/JTECH-D-13-00034.1.
    [17]

    Huang, B. Y., M. L’Heureux, Z.-Z. Hu, et al., 2016: Ranking the strongest ENSO events while incorporating SST uncertainty. Geophys. Res. Lett., 43, 9165–9172. doi: 10.1002/2016GL070888.
    [18]

    Huang, B. Y., C. Y. Liu, V. Banzon, et al., 2021: Improvements of the Daily Optimum Interpolation Sea Surface Temperature (DOISST) version 2.1. J. Climate, 34, 2923–2939. doi: 10.1175/JCLI-D-20-0166.1.
    [19]

    Huang, G., and K. M. Hu, 2008: Impact of North Indian Ocean SSTA on Northwest Pacific lower layer anomalous anticy-clone in summer. J. Nanjing Inst. Meteor., 31, 749–757. doi: 10.3969/j.issn.1674-7097.2008.06.001. (in Chinese)
    [20]

    Ignatov, A., 2010: GOES-R Advanced Baseline Imager (ABI) Algorithm Theoretical Basis Document for Sea Surface Temperature. NOAA/NESDIS/STAR, 90 pp. Available at https://www.star.nesdis.noaa.gov/goesr/documents/ATBDs/Baseline/ATBD_GOES-R_SST-v2.0_Aug2010.pdf. Accessed on 11 August 2021.
    [21]

    IPCC, 2018: Special Report: Global Warming of 1.5°C. IPCC, 630 pp. Available at https://www.ipcc.ch/site/assets/uploads/sites/2/2019/06/SR15_Full_Report_Low_Res.pdf. Accessed on 11 August 2021.
    [22]

    Izumo, T., C. D. B. Montégut, J. J. Luo, et al., 2008: The role of the western Arabian Sea upwelling in Indian monsoon rainfall variability. J. Climate, 21, 5603–5623. doi: 10.1175/2008JCLI2158.1.
    [23]

    Kanamitsu, M., W. Ebisuzaki, J. Woollen, et al., 2002: NCEP–DOE AMIP-II reanalysis (R-2). Bull. Amer. Meteor. Soc., 83, 1631–1644. doi: 10.1175/BAMS-83-11-1631.
    [24]

    Kao, H.-Y., and J.-Y. Yu, 2009: Contrasting eastern-Pacific and central-Pacific types of ENSO. J. Climate, 22, 615–632. doi: 10.1175/2008JCLI2309.1.
    [25]

    Liu, Y. Y., Y. H. Ding, H. Gao, et al., 2013: Tropospheric biennial oscillation of the western Pacific subtropical high and its relationships with the tropical SST and atmospheric circulation anomalies. Chinese Sci. Bull., 58, 3664–3672. doi: 10.1007/s11434-013-5854-7.
    [26]

    Liu, Y. Y., Z.-Z. Hu, A. Kumar, et al., 2015: Tropospheric bien-nial oscillation of summer monsoon rainfall over East Asia and its association with ENSO. Climate Dyn., 45, 1747–1759. doi: 10.1007/s00382-014-2429-5.
    [27]

    Liu, Y. Y., Z. J. Ke, and Y. H. Ding, 2019a: Predictability of East Asian summer monsoon in seasonal climate forecast models. Int. J. Climatol., 39, 5688–5701, doi: 10.1002/joc.6180.
    [28]

    Liu, Y. Y., P. Liang, and Y. Sun, 2019b: The Asian Summer Monsoon: Characteristics, Variability, Teleconnections and Projection. Elsevier, Cambridge, 237 pp, doi: 10.1016/C2017-0-04074-0.
    [29]

    National Research Council, 2010: Assessment of Intraseasonal to Interannual Climate Prediction and Predictability. National Academies Press, Washington, 192 pp.
    [30]

    Nitta, T., and Z.-Z. Hu, 1996: Summer climate variability in China and its association with 500 hPa height and tropical convection. J. Meteor. Soc. Japan, 74, 425–445. doi: 10.2151/jmsj1965.74.4_425.
    [31]

    Ohring, G., B. Wielicki, R. Spencer, et al., 2005: Satellite instrument calibration for measuring global climate change: Report of a workshop. Bull. Amer. Meteor. Soc., 86, 1303–1314. doi: 10.1175/BAMS-86-9-1303.
    [32]

    Reynolds, R. W., N. A. Rayner, T. M. Smith, et al., 2002: An improved in situ and satellite SST analysis for climate. J. Climate, 15, 1609–1625. doi: 10.1175/1520-0442(2002)015<1609:AIISAS>2.0.CO;2.
    [33]

    Reynolds, R. W., T. M. Smith, C. Y. Liu, et al., 2007: Daily high-resolution-blended analyses for sea surface temperature. J. Climate, 20, 5473–5496. doi: 10.1175/2007JCLI1824.1.
    [34]

    Saji, N. H., B. N. Goswami, P. N. Vinayachandran, et al., 1999: A dipole mode in the tropical Indian Ocean. Nature, 401, 360–363. doi: 10.1038/43854.
    [35]

    Wang, B., R. G. Wu, and X. H. Fu, 2000: Pacific–East Asian teleconnection: How does ENSO affect East Asian climate? J. Climate, 13, 1517–1536. doi: 10.1175/1520-0442(2000)013<1517:PEATHD>2.0.CO;2.
    [36]

    Wang, S. J., P. Cui, M. N. Ran, et al., 2014a: Algorithm improvement and quality validation of operational FY3A/VIRR SST product. Meteor. Sci. Technol., 42, 748–752. doi: 10.3969/j.issn.1671-6345.2014.05.004. (in Chinese)
    [37]

    Wang, S. J., P. Cui, P. Zhang, et al., 2014b: The improvement of FY-3B/VIRR SST algorithm and its accuracy. J. Appl. Meteor. Sci., 25, 701–710. (in Chinese)
    [38]

    Wang, S. J., P. Cui, P. Zhang, et al., 2014c: FY-3C/VIRR SST algorithm and cal/val activities at NSMC/CMA. Proc. Volume 9261, Ocean Remote Sensing and Monitoring from Space, SPIE, Beijing, 92610G, doi: 10.1117/12.2068773.
    [39]

    Wang, S. J., P. Cui, P. Zhang, et al., 2020: FY-3C/VIRR sea surface temperature products and quality validation. J. Appl. Meteor. Sci., 31, 729–739. doi: 10.11898/1001-7313.20200608. (in Chinese)
    [40]

    Wu, G. X., P. Liu, Y. M. Liu, et al., 2000: Impacts of the sea surface temperature anomaly in the Indian Ocean on the subtropical anticyclone over the western Pacific—Two-stage thermal adaptation in the atmosphere. Acta Meteor. Sinica, 58, 513–522. doi: 10.11676/qxxb2000.054. (in Chinese)
    [41]

    Wu, R. G., Z.-Z. Hu, and B. P. Kirtman, 2003: Evolution of ENSO-related rainfall anomalies in East Asia. J. Climate, 16, 3742–3758. doi: 10.1175/1520-0442(2003)016<3742:EOERAI>2.0.CO;2.
    [42]

    Wu, R. G., and B. P. Kirtman, 2004: Understanding the impacts of the Indian Ocean on ENSO variability in a coupled GCM. J. Climate, 17, 4019–4031. doi: 10.1175/1520-0442(2004)017<4019:UTIOTI>2.0.CO;2.
    [43]

    Xu, F., and A. Ignatov, 2014: In situ SST Quality Monitor (iQuam). J. Atmos. Oceanic Technol., 31, 164–180. doi: 10.1175/JTECH-D-13-00121.1.
    [44]

    Xie, S.-P., K. Hu, J. Hafner, et al, 2009: Indain Ocean capacitor effect on Indo-western Pacific climate during the summer following El Niño. J. Climate, 22, 730–747. doi: 10.1175/2008JCLI2544.1.
    [45]

    Yang, J., Q. Liu, S.-P. Xie, et al, 2007: Impact of the Indian Ocean SST basin mode on the Asian summer monsoon. Geophys. Res. Lett., 34, L02708. doi: 10.1029/2006GL028571.
    [46]

    Yang, J., P. Zhang, N. M. Lu, et al., 2012: Improvements on glo-bal meteorological observations from the current Fengyun 3 satellites and beyond. Int. J. Digital Earth, 5, 251–265. doi: 10.1080/17538947.2012.658666.
    [47]

    Zebiak, S. E., 1993: Air–sea interaction in the equatorial Atlantic region. J. Climate, 6, 1567–1586. doi: 10.1175/1520-0442(1993)006<1567:AIITEA>2.0.CO;2.
    [48]

    Zhang, R. H., and A. Sumi, 2002: Moisture circulation over East Asia during El Niño episode in northern winter, spring and autumn. J. Meteor. Soc. Japan, 80, 213–227. doi: 10.2151/jmsj.80.213.
    [49]

    Zhang, R. H., A. Sumi, and M. Kimoto, 1999: A diagnostic study of the impact of El Niño on the precipitation in China. Adv. Atmos. Sci., 16, 229–241. doi: 10.1007/BF02973084.
    [50]

    Zhang, Z. Y., D. Y. Gong, D. Guo, et al., 2008: Anomalous winter temperature and precipitation events in southern China. Acta Geogra. Sinica, 63, 899–912. doi: 10.3321/j.issn:0375-5444.2008.09.001. (in Chinese)
    [51]

    Zhou, L. T., and R. G. Wu, 2010: Respective impacts of the East Asian winter monsoon and ENSO on winter rainfall in China. J. Geophy. Res. Atmos., 115, D02107. doi: 10.1029/2009JD012502.
  • 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 FY-3/VIRR Sea Surface Temperature Data for Climate Applications

    Corresponding author: Jian LIU, liujian@cma.gov.cn
  • 1. Laboratory of Climate Studies, National Climate Center, China Meteorological Administration, Beijing 100081
  • 2. Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters, Nanjing University of Information Science & Technology, Nanjing 210044
  • 3. National Satellite Meteorological Center, China Meteorological Administration, Beijing 100081
  • 4. Key Laboratory of Radiometric Calibration and Validation for Environmental Satellites, National Satellite Meteorological Center, China Meteorological Administration, Beijing 100081
Funds: Supported by the Guangdong Major Project of Basic and Applied Basic Research (2020B0301030004) and National Key Reseach and Development Program of China (2018YFB0504905 and 2018YFB0504900)

Abstract: We evaluated the sea surface temperature (SST) products derived from the visible infrared radiometer on board the Fengyun-3 satellites (FY-3/VIRR) during 2016–2018 from the perspective of climate applications. The data had previously been reprocessed by the National Satellite Meteorological Center of China Meteorological Administration based on an updated SST retrieval algorithm. The overall consistency between the FY-3/VIRR SST data and the optimum interpolation SST version 2 (OIv2.1) was better for monthly means than for pentad means, and showed a clear dependence on the season and location. There was better consistency in winter than in summer, and in the tropical central–eastern Pacific than in the western Pacific warm pool, tropical North Indian Ocean, and tropical Atlantic Ocean. The monthly deviation of the global average SST anomaly was −0.03 ± 0.07°C and the average root-mean-square errors (RMSEs) presented clear seasonal fluctuations with a maximum of approximately 0.5°C in summer. The poor consistency of the FY-3/VIRR SST in summer may be partially attributed to the bias of the OIv2.1 data in global oceans (especially the Indian Ocean) as a result of the spatially heterogeneous in situ measurements from ships, buoys, and Argo floats. Convective activities and clouds in the tropics may also influence the accuracy of the FY-3/VIRR SST retrievals. The Niño SST indices based on both FY-3/VIRR and OIv2.1 SST data displayed a generally similar evolution, including the start and end of El Niño and La Niña events and their amplitudes, although the deviations were slightly larger when the Pacific SST anomaly was in the neutral state of the El Niño–Southern Oscillation (ENSO). The deviations varied greatly with season in the tropical Indian and Atlantic oceans, suggesting the need to perform further analyses and validation of the FY-3/VIRR SST products in these two basins.

风云三号卫星可见光红外辐射计海表温度资料的气候应用评估

本文基于国家卫星气象中心提供的一套利用最新的海温反演算法再处理的2016–2018年风云三号卫星可见光红外辐射计(FY-3/VIRR)海表温度产品,以最优插值海表温度数据OIv2.1作为标准,从气候应用的角度对其侯平均和月平均尺度产品进行了质量评估。FY-3/VIRR的月平均产品总体优于候平均产品,与OIv2.1的一致性表现出季节和空间差异性:冬季较好,夏季较差;热带中东太平洋较好,暖池、热带北印度洋和大西洋较差。全球平均的海温偏差为−0.03 ± 0.07°C,均方根误差呈现季节性波动,其中夏季最大,约0.5°C。夏季FY-3/VIRR海温在全球平均和印度洋区域与OIv2.1存在较大偏差,部分原因可能来自于船舶、固定浮标和Argo浮标的空间非均匀测量,而热带地区较多的对流活动和云覆盖也可能影响FY-3/VIRR海温反演的精度。基于FY-3/VIRR得到的Niño海温指数总体上与OIv2.1接近,能够反映ENSO事件的开始、结束和强度。在热带印度洋和大西洋两种海温资料之间的偏差随季节变化很大。因此,有必要对这两个区域的FY-3/VIRR海温产品做进一步的质量分析和验证。
1.   Introduction
  • Sea surface temperature (SST) is an important climate variable that describes the large heat capacity of wide-coverage global oceans. Through modulating geopotential height and wind fields in the lower–upper troposphere, the anomalies of SST (SSTAs) may induce anomalies of the earth’s climate over both the land and oceans. SST has been widely used in simulation, prediction, monitoring, and assessment of climate and climate change, as well as in the applications related to environment, agriculture, and industry (EPA, 2014; IPCC, 2018).

    The El Niño–Southern Oscillation (ENSO) in the tropical central–eastern Pacific Ocean represents the strongest SSTA forcing to the overlying atmosphere on seasonal–interannual timescales, which relates intimately to severe flooding, drought, heatwave, landslide, and many other natural disasters affecting the safety and economic development of the world (National Research Council, 2010). For example, the ENSO in different phases exerts distinct impacts on the variability of the East Asian climate (Zhang et al., 1999; Wang et al., 2000; Zhang and Sumi, 2002; Wu et al., 2003; Hu et al., 2005; Zhou and Wu, 2010; Liu et al., 2013, 2015, 2019b). The SSTAs in other oceans, including the tropi-cal Indian Ocean (Wu et al., 2000, 2004; Liu et al., 2019a; Ding et al., 2021), Atlantic Ocean (Han et al., 2011; Liu et al., 2019a, b), and western Pacific warm pool (Nitta and Hu, 1996; Guo and Ni, 1998), also affect the atmospheric circulation and precipitation anomalies over Asia (Zhang et al., 2008; Liu et al., 2019b). A high-quality SST dataset for understanding and quantifying the variation of SSTs and their global impacts is therefore crucial and in high demand.

    Climate applications require SST data with an accuracy of 0.1 K and a stability of 0.04 K decade−1 (Ohring et al., 2005). SST can be estimated by retrieval of observations from satellites (GCOS, 2011). The Fengyun-3 (FY-3) series of satellites are the second generation polar-orbiting meteorological satellites launched by China. FY-3B and FY-3C were launched on 5 November 2010 and 23 September 2013, respectively; and operated in a sun-synchronous afternoon orbit with local equator-crossing times of 1400 and 1000, respectively, in the descending node. SST is a key global product of the FY-3 series satellites (Wang et al., 2014a, b, c, 2020).

    The visible infrared radiometer (VIRR) on board the FY-3 satellites is a 10-channel radiometer for multi-purpose imagery with a 1.1-km resolution at nadir. The swath of the VIRR is 2800 km (Yang et al., 2012). The VIRR has one mid-wavelength infrared channel (3.55–3.93 μm) and two long-wavelength infrared channels (the split window channels: 10.3–11.3 and 11.5–12.5 μm), which are used to estimate SST. To meet the needs of climate applications, the National Satellite Meteorological Center of the China Meteorological Administration has produced a set of global daily satellite merged SST products by combining the FY-3B and FY-3C VIRR SST data (hereafter the FY-3/VIRR SST).

    So far, the FY-3/VIRR SST data have not been carefully evaluated. To fill this gap, this study aims to assess the FY-3/VIRR SST products from the perspective of climate applications and analyze the relevant biases and possible causes on both pentad and monthly timescales, with an expectation to provide reference information for the reprocessing of a comprehensive set of historical FY-3/VIRR data. We also select and assess suitable SST indicators for applications such as climate monitoring and diagnosis in real-time operational mode, based on the independent FY-3 satellite products.

    The remainder of this paper is organized as follows. Section 2 describes the FY-3/VIRR SST datasets, SST indices, and data processing methods. Section 3 addresses the deviations of the FY-3/VIRR SST data from the observed SST data and the possible causes based on regional SST indices. A summary and discussion are provided in Section 4.

2.   Datasets and methods
  • We used daily SST version 1 (RV1) data derived from the VIRR on board FY-3B and FY-3C polar-orbiting satellites. The RV1 data were produced through reprocessing the original FY-3B and FY-3C observations and the operational level 1 data with monthly nonlinear SST coefficients over January 2014–December 2019. The dataset is available upon request to liujian@cma.gov.cn. The Operational Sea Surface Temperature and Sea Ice Analysis (OSTIAv2.0; Donlon et al., 2012) was used as the first-guess SST in the nonlinear SST algorithm. The nonlinear SST algorithm was tuned by regression of the SST against the quality-controlled buoy data from the in situ SST quality monitor (iQuam; Xu et al., 2014).

    The FY-3/VIRR satellite SST based on the statistical regression of the buoy SST was defined as the bulk SST using the algorithm theoretical basis document from the NOAA National Environmental Satellite, Data, and Information Service (Ignatov, 2010). This procedure converts the retrieval of temperature from the “skin” to the “bulk” SST and is sensitive to the skin SST (Ignatov, 2010). The granule FY-3B and FY-3C VIRR SST data were resampled to a latitude–longitude grid at 0.05° (5-km) resolution and separated into daytime and nighttime data for each satellite. A quality index was computed on each grid to classify the processing conditions as excellent, good, bad, or unprocessed (cloud, land, or missing data). An introduction of the specific algorithms has been provided by Wang et al. (2020).

    The difference between the nighttime and daytime SSTs was taken into account in the processing of the FY-3/VIRR merged SST data. The daily FY-3/VIRR merged SST of RV1 was processed by using the priority FY-3B nighttime, FY-3C nighttime, and FY-3B and FY-3C daytime SSTs when the wind speed was > 6 m s−1, to account for the diurnal warming caused by strong breezes and sunshine. The global resolution was 5 km. The daily vector wind data were from the NCEP–Department of Energy reanalysis-2 dataset for the period from 1979 to 2020 (Kanamitsu et al., 2002; www.esrl.noaa.gov/psd). We only used the SST data with an “excellent” quality index. We set the global SST spatial range as 70°S–70°N, 0–360° to avoid missing SST data in the polar regions that are usually covered by sea ice.

    The reference data for evaluation of the FY-3/VIRR SST was the daily mean SST from version 2.1 of the optimum interpolation SST (OIv2.1; Reynolds et al., 2002, 2007) on a horizontal resolution of 0.25° × 0.25°. The OIv2.1 is an upgraded version of OIv2 (Reynolds et al., 2002) with reduced biases in the global oceans and especially the Indian Ocean compared with Argo observations (Huang et al., 2021). It has been widely used in climate assessment and monitoring by the NCEP Climate Prediction Center and the China Meteorological Administration National Climate Center. The OIv2.1 SST is a blend of in situ ship and buoy SSTs with satellite SSTs derived from the Advanced Very High Resolution Radiometer (Huang et al., 2021), and this SST product is obtained by optimal interpolation and reflects the bulk SST (Ignatov, 2010). The monthly and pentad climatologies used in this work were averages of the OIv2.1 data from 1991 to 2010. In addition, the daily average Canadian Meteorological Center Level 4 SST data (Brasnett and Surcel Colan, 2016) were also selected to verify the credibility of the FY-3/VIRR SST.

    The FY-3/VIRR SST data were interpolated onto the same spatial resolution as the OIv2.1 data for comparison. Only the FY-3/VIRR data with a quality index of 5 (excellent) were selected for interpolation; this accounts for 51% of the total global FY-3/VIRR SST. The daily FY-3/VIRR SST was processed into pentad and monthly averages according to the demand of operational climate applications. The mean of the first to fifth day of each month represents the first pentad, the mean of the sixth to tenth day is the second pentad, and so on. The last pentad is the average from the 26th to the last day of each month, with a total of 6 pentads in each month. Through this interpolation process, the missing values are thus greatly reduced in both space and time.

    The SSTs in the equatorial central and eastern Pacific associated with ENSO are key signals in the climate monitoring and prediction (Hu et al., 2020). It has been revealed that the variations in some regional mean SSTA indices can be used to represent key climate variability modes including the El Niño and La Niña events. For example, based on the OIv2.1 SST, the variation in the Niño3.4 index indicates that the ENSO evolved from a positive phase to a neutral state and then to a negative phase from 2016 to 2018 (Fig. 1). In addition to the Niño3.4 index, the Niño3 and Niño4 indices have been used to represent different flavors of the ENSO (Kao and Yu, 2009; Hu et al., 2012). The western Pacific warm pool (NiñoW), Indian Ocean basin mode (IOBM; Chambers et al., 1999), Indian Ocean Dipole (IOD; Saji et al., 1999), and equatorial Atlantic Niño (ATL3; Zebiak, 1993) indices were selected to represent tropical SSTA forcing in different ocean basins (Enfield et al., 1999). Table 1 gives the definitions of these SSTA indices.

    Figure 1.  Time series of the monthly Niño3.4 index (°C) based on the OIv2.1 SST data during 2016–2018. The labels on the x-axis refers to month and year; for example, Jan-16 denotes January 2016; same applies to other relevant figure labels in this paper.

    SST indexDefinition
    Niño3.4SSTA (5°S–5°N, 170°–120°W)
    Niño3SSTA (5°S–5°N, 150°–90°W)
    Niño4SSTA (5°S–5°N, 160°E–150°W)
    NiñoWSSTA (0°–10°N, 120°–140°E)
    IODSSTA (10°S–10°N, 50°–70°E)
    − SSTA (10°S–0°, 90°–110°E)
    IOBMSSTA (20°S–20°N, 40°–110°E)
    ATL3SSTA (3°S–3°N, 20°W–0°)
    TNASSTA (5°–20°N, 60°–30°W)
    TSASSTA (20°S–0°, 30°W–10°E)
    TADSSTATNA − SSTATSA

    Table 1.  Definitions of the selected tropical SSTA indices. Note: TNA—Tropical North Atlantic, TSA—Tropical South Atlantic, TAD—Tropical Atlantic Dipole; search text for other accronyms

3.   Results
  • Most climate applications focus on the variation of SSTAs. We therefore used the spatial distribution of the anomaly correlation coefficient (ACC) to measure the consistency (difference) between the two SST datasets. Figure 2a presents the variation in the ACC of the monthly global SSTAs (70°S–70°N, 0°–360°) between the FY-3/VIRR and OIv2.1 data from 2016 to 2018. The overall consistency between the two datasets was good, with a mean ACC of 0.895. There was a clear dependence of ACC on the season, with an average ACC value of 0.920 in winter (December–January–February) and that of 0.868 in summer (June–July–August).

    Figure 2.  Time series of the anomaly correlation coefficient (ACC) of (a) monthly and (b–d) pentad SSTAs averaged over the region 70°S–70°N, 0°–360° between the FY-3/VIRR and OIv2.1 data in (a) 2016–2018, (b) 2016, (c) 2017, and (d) 2018, respectively. The gray line indicates the average ACC. The x-axis labels in (a) denote the same as in Fig. 1, while those in (b–d) denote pentad-month-year.

    The variations in the ACCs of the pentad SSTAs in 2016, 2017, and 2018 had similar features (Figs. 2bd) and were better in winter than in summer, although the overall ACCs were lower than the ACCs of the monthly SSTAs. There were also some differences in the ACCs during these three years. In 2016, a decaying year of El Niño, the SST in the central and eastern tropical Pacific was in an abnormally warm to neutral state, with an average ACC of 0.858. In 2018, a decaying year of La Niña, the SST in the equatorial central and eastern Pacific varied from anomalously cold to warm, with an average ACC of 0.856. The SSTA of the tropical central and eastern Pacific (Niño3.4 index) was neutral throughout most of 2017 (Fig. 1) and the average ACC was 0.845, slightly lower than that in 2016 and 2018. The ENSO cycle was therefore associated with the consistency between the two SST datasets, with a higher consistency during both the warm and cold phases of the ENSO than during the neutral phase. This was probably a result of the large amplitude of the SSTA and the signal-to-noise ratio in the tropical central and eastern Pacific in both the warm and cold phases of the ENSO (Hu et al., 2019).

    Figure 3 shows a spatial distribution of point-to-point correlation of the monthly SSTAs between the FY-3/VIRR and OIv2.1 data from 2016 to 2018. The consistency between the two datasets was excellent in most regions, with the correlation coefficient > 0.9, but there was a clear dependence on location. The correlation coefficient was > 0.95 in the tropical central–eastern Pacific and > 0.9 in the North Pacific, central North Atlantic, South Indian, South Pacific, and South Atlantic oceans. By contrast, relatively low correlation coefficients were present in some key tropical areas in SST monitoring operation. For example, the correlation coefficients were generally < 0.8 in the western Pacific warm pool, northern Indian Ocean, and equatorial Atlantic Ocean.

    Figure 3.  Distribution of temporal correlation coefficient of the monthly SSTAs between the FY-3/VIRR and OIv2.1 data from 2016 to 2018 (annual cycles were subtracted from the original data).

    Figure 4 displays similar distributions of correlation coefficient of the pentad SSTAs between FY-3/VIRR and OIv2.1, but the correlation coefficient was generally lower than that of the monthly average. There were also slight differences among the three years. As a result of warming of the tropical central–eastern Pacific and In-dian oceans in 2016 (figure omitted), the correlation coefficients in these regions, as well as in the western Pacific warm pool, were higher than those in 2017 and 2018. The correlation coefficient in the tropical Atlantic, however, showed little difference among the three years.

    Figure 4.  Distributions of temporal correlation coefficient of the pentad SSTAs between the FY-3/VIRR and OIv2.1 data in (a) 2016, (b) 2017, and (c) 2018, respectively (annual cycles were subtracted from the original data).

  • We analyzed the possible reasons for the low consistency between the FY-3/VIRR and OIv2.1 data in some of the study regions and suggested how to reasonably use the FY-3/VIRR SST data products.

    Figure 5 presents the bias and root-mean-square-error (RMSE) of the monthly FY-3/VIRR SSTAs averaged over the region 70°S–70°N, 0–360° compared with the OIv2.1 SSTA from 2016 to 2018. In general, the FY-3/VIRR SSTAs were colder than the OIv2.1 SSTAs, except for during a few months of summer, with a deviation range of −0.03 ± 0.07°C. The RMSE of the monthly FY-3/VIRR SSTA from 2016 to 2018 showed a clear seasonal fluctuation, with a maximum of about 0.5°C in summer. Huang et al. (2021) showed that the OIv2.1 data have a residual cold bias of about −0.04°C over the global oceans and about −0.08°C in the Indian Ocean, which may result from heterogeneous in situ measurements from ships and Argo floats. The biases of the OIv2.1 data in the global oceans (and the Indian Ocean) may contribute to the notable deviation between the two SST datasets. The lower consistency in summer may also be partly attributed to the merge algorithm of the FY-3/VIRR, specifically, the processing method to avoid diurnal warming (Wang et al., 2020).

    Figure 5.  Time series of the bias (bar) and RMSE (line) of the monthly SSTA (°C) averaged over the region 70°S–70°N, 0°–360° between the FY-3/VIRR and OIv2.1 data from 2016 to 2018. The x-axis labels denote the same as in Fig. 1.

    We examined the differences in spatial distribution of SSTAs between the two datasets (see Figs. 6, 7) and found good (poor) consistency in winter (summer). January 2016 had the highest correlation for the two sets of SSTA data, with a correlation coefficient of 0.932 and an RMSE of 0.335°C (Fig. 2a). In the OIv2.1 SSTA data, positive SSTAs were seen in the tropical central–eastern Pacific, tropical Indian Ocean, and western Atlantic Ocean, whereas negative SSTAs were seen in both the north and south midlatitudes of the Pacific and Atlantic oceans, typical of the mature phase of extremely strong El Niño events (Fig. 6a).

    Figure 6.  SSTAs (°C) in (a) OIv2.1 and (b) FY-3/VIRR, and (c) SSTA differences between the FY-3/VIRR and OIv2.1 data, in January 2016.

    Figure 7.  SSTA differences (°C) between the FY-3/VIRR and OIv2.1 data in July (a) 2016, (b) 2017, and (c) 2018, respectively.

    The FY-3/VIRR data well captured the spatial pattern of SSTAs in general (Fig. 6b). Detailed differences from the OIv2.1 data could be seen in Fig. 6c, which reveals a clear dependence on latitude with a quasi-symmetrical zonal distribution—That is, the region from the equator to the south and north presented “negative–positive–negative” deviations. An overall negative deviation was observed in the tropical oceans, especially in the equatorial central–eastern Atlantic, equatorial central–eastern Pacific, and equatorial western Indian oceans. The significant SSTA variability in these areas may be associated with large differences between the two SST datasets.

    In contrast to the good consistency in winter months, July had the lowest ACC consistency (Fig. 2). Figure 7 shows spatial distributions of the SSTA deviation in July 2016, 2017, and 2018. The distribution of the deviation in July was different from that in January in that the differences in SSTA between the two SST datasets were more significant in the Northern Hemisphere than in the Southern Hemisphere. The FY-3/VIRR SSTAs were colder than the OIv2.1 SSTAs in the North Indian as well as equatorial eastern Pacific and North Atlantic oceans. Large SSTA appeared in the North Pacific region around 40°N with a “warm in the west and cold in the east” pattern, which did not occur in the SSTA deviation field in January (Fig. 6c). The differences in SSTA in these oceanic regions in summer need further verification before the data being applied to climate monitoring.

    There were obvious differences in SSTA between the FY-3/VIRR and OIv2.1 data in the tropical Indian Ocean in both winter and summer, when the evolution of the SSTA had a significant impact on precipitation over Asia, especially in China during the flood season (Wu and Kirtman, 2004; Annamalai et al., 2005; Yang et al., 2007; Xie et al., 2009; Ding et al., 2021). This may be related to more clouds and convective activities in the tropical ocean area. To rectify the FY-3/VIRR SST biases in the tropical Indian Ocean, we further analyzed whether the bias distribution was somehow related to SSTA itself in this region, with the results given in Fig. 8.

    Figure 8.  Scatter plots of the deviation/bias in SSTA between FY-3/VIRR and OIv2.1 in the tropical Indian Ocean (y-axis) versus the OIv2.1 SSTA (x-axis) in the same region in (a) January 2016, (b) July 2016, (c) July 2017, and (d) July 2018, respectively.

    In January 2016, the SSTA in OIv2.1 over the Indian Ocean was averagely > 1°C, the spread of the FY-3/VIRR SSTA bias/deviation was relatively small, but the linear fitting coefficient between the two was −0.1698 without reaching the 0.1 significance level (Fig. 8a). In July 2016, 2017, and 2018, the spreads of the FY-3/VIRR SSTA deviation were obviously larger (Fig. 8b–d), without a significant relationship with the OIv2.1 SSTA either. Nevertheless, it is found from Fig. 8b–d that more cold than warm biases of FY-3/VIRR SSTA occurred during the three summers, corresponding to more warm than cold SSTAs of OIv2.1. Overall, no definite relationship was found between the bias of FY-3/VIRR SSTA and the OIv2.1 SSTA over the tropical Indian Ocean in winter and summer, except for a rough association of warmer OIv2.1 SSTA with colder FY-3/VIRR SSTA bias, which may be useful for real-time climate applications.

    As shown in Fig. 7, the coldest biases of FY-3/VIRR SSTA was located in the coastal northern Indian Ocean in summer, where the seasonal variability of SSTA is large and has a significant impact on atmospheric heating and climate anomalies (Huang and Hu, 2008; Izumo et al., 2008; Du et al., 2011). The accuracy of the SSTAs in this region and in the North Pacific region (Fig. 7) in the FY-3/VIRR SST dataset needs further improvement by cross validation with different SST datasets and in situ observations because of their important role in climate variability and predictability (i.e., Huang et al., 2013, 2016).

  • Various SST monitoring indices have been designed for key regions to describe the evolution of the external forcing signal of SSTAs for use in climate operations and research, including the Niño3.4, NiñoW, IOBM, IOD, and ATL3 indices (i.e., Zebiak, 1993; Chambers et al., 1999; Saji et al., 1999; see Table 1). We selected several representative SST monitoring indices in the tropical Pacific, Indian, and Atlantic oceans to evaluate the quality of the FY-3/VIRR SST data from the monthly evolution of these indicators (Figs. 911).

    Figure 9.  Time series of the monthly mean tropical Pacific Ocean SST indices: (a) Niño3.4, (b) Niño3, (c) Niño4, and (d) NiñoW based on the FY-3/VIRR (red line) and OIv2.1 (blue line) data and related biases (bar; °C), respectively. The x-axis labels denote the same as in Fig. 1.

    Figure 10.  As in Fig. 9, but for the tropical Indian Ocean SST indices (a) IOBM and (b) IOD.

    Figure 9 shows the four monitoring indices (Niño3.4, Niño3, Niño4, and NiñoW) for ENSO events in the tropical Pacific Ocean. All the indices show consistent evolution characteristics as a result of the relatively good consistency of the two datasets in the equatorial central and eastern Pacific (Figs. 9ac). A transition from El Niño (Niño3.4 index > 0.5°C) to La Niña (Niño3.4 index < −0.5°C) events occurred from 2016 to 2018. The Niño indices based on both the FY-3/VIRR and OIv2.1 SSTAs and the amplitudes of positive and negative SSTAs were consistent at the beginning and end of El Niño and La Niña events. However, the differences in the Niño3.4 index between the two datasets were slightly larger in the neutral ENSO state (Niño3.4 index between −0.5 and 0.5°C) than during El Niño and La Niña events (Fig. 9a).

    The two datasets showed a larger discrepancy in the NiñoW index than in the equatorial central and eastern Pacific (Fig. 9d), with July 2016, August 2016, and August 2017 showing large positive deviations as well as November 2016, November 2017, and December 2017 showing large negative deviations. The monthly variability of the NiñoW index was larger in the FY-3/VIRR dataset than in the OIv2.1 dataset. This is consistent with the poor point-to-point correlation in the western Pacific warm pool, which may be related to deep convective activities and clouds in this region (Wang et al., 2020). We need to further quantify the impact of clouds on the SST in the satellite retrieval data in this region.

    Figure 10 displays monthly evolution of the two In-dian Ocean SST indices. There are two typical spatial modes of the SSTAs in the tropical Indian Ocean (Chambers et al., 1999; Saji et al., 1999): one mode is IOBM, during which the SSTA over the whole Indian Ocean basin tends to be warmer or colder; and the other is IOD, during which SSTA is opposite between the tropical western and eastern Indian Ocean. The IOBM index was more consistent between the FY-3/VIRR and OIv2.1 datasets than the IOD index, although there was a clear deviation between the two datasets when the IOBM index reached a maximum or minimum. There were similar deviations in the monthly IOD index, particularly for the cold deviations accompanied by a minimum value of the IOD index. For example, the IOD index from the FY-3/VIRR dataset was colder than that from the OIv2.1 dataset by > 1°C in June 2018. Such a large SSTA deviation may have a crucial impact on operational climate monitoring and forecasts.

    The four SST indices were also compared over the tropical Atlantic Ocean. The FY-3/VIRR data showed remarkable deviations from the OIv2.1 data, especially for the ATL3 index in the equatorial Atlantic (Fig. 11). This is partly linked to the cold SST deviation of the FY-3/VIRR data relative to the OIv2.1 data in the lower latitudes of the Atlantic Ocean (Fig. 5). The TNA and TSA indices of the FY-3/VIRR data were colder than those in the OIv2.1 data and the TAD showed an irregular deviation during this period.

    Figure 11.  As in Fig. 9, but for the tropical Atlantic Ocean SST indices (a) equatorial Atlantic Niño (ATL3), (b) tropical North Atlantic (TNA), (c) tropical South Atlantic (TSA), and (d) Tropical Atlantic Dipole (TAD).

4.   Summary and discussion
  • We evaluated the quality of the FY-3/VIRR SST product in the period of 2016–2018 when the SSTA in the tropical central and eastern Pacific Ocean experienced a typical “warm–neutral–cold” evolution. This FY-3/VIRR SST product had been reprocessed by the National Satellite Meteorological Center of the China Meteorological Administration based on an updated SST retrieval algorithm. We compared the FY-3/VIRR SST product with the OIv2.1 SST, which is commonly used in climate monitoring, assessment, and prediction operations. We only selected FY-3/VIRR global SST data with an “excellent” quality level and smoothed to pentad and monthly timescales to meet the operation needs of the climate monitoring and prediction. Our main conclusions are as follows.

    (1) The consistency of the monthly SSTA between the FY-3/VIRR and OIv2.1 datasets was better than that of the pentad SSTA. The consistency was clearly dependent on the season and location. The consistency was better in winter than in summer and better in the tropical central and eastern Pacific Ocean than in the western Pacific warm pool, tropical North Indian Ocean, and tropical Atlantic Ocean.

    (2) The monthly deviation of the global average SSTA between the two datasets was −0.03 ± 0.07°C and the global average RMSE showed clear seasonal fluctuations with a maximum of around 0.5°C in summer. The SSTA deviations in the Northern Hemisphere were larger than those in the Southern Hemisphere in summer and the cold deviations in the FY-3/VIRR data were mainly present in the North Indian, equatorial eastern Pacific, and North Atlantic oceans. The poor consistency in summer may be partially associated with the biases in the OIv2.1 data in the global ocean, particularly in the Indian Ocean, as a result of heterogeneous in situ measurements from ships, buoys, and Argo floats. Convective activities and clouds in the tropics, including the western Pacific, North Indian, and tropical Atlantic oceans, may affect the accuracy of satellite infrared sensors and, in turn, the SST retrievals.

    (3) Based on the Niño indices, the FY-3/VIRR and OIv2.1 SST data display a similar ENSO evolution, including the beginning and end of El Niño and La Niña events and their amplitudes, although the deviation is slightly larger in the ENSO neutral condition. The deviations of the SST indices between the two datasets vary with the season in the tropical Indian and Atlantic oceans.

    To further verify the validity of the FY3/VIRR SST product, an additional comparative analysis was made of the daily FY3/VIRR SST data and the daily OIv2.1 as well as the daily Canadian Meteorological Center Level 4 SST from 2016 to 2018. The results were generally consistent with the results for the pentad and monthly mean SSTAs. Compared with the daily OIv2.1 data, the deviation and RMSE of the FY3/VIRR SSTA were 0.01 and 0.54°C, respectively. The RMSE also showed a clear seasonal fluctuation, with the largest deviation in summer (figures omitted). The deviation and RMSE of the FY3/VIRR SSTA from the daily Canadian Meteorologi-cal Center SSTA were 0.03 and 0.47°C, respectively; and the seasonal fluctuation of the RMSE was slightly smaller than that relative to the daily OIv2.1. This dependence suggests that we need to further evaluate the FY3/VIRR SST product with other SST reanalysis products in addition to in situ observations (Huang et al., 2013, 2016).

    The accuracy of FY-3/VIRR SST product is highly dependent on the performance of the VIRR instrument, onboard operational status, positioning and calibration accuracy, and historical reprocessing of the SST. The quality assessment of the FY-3/VIRR satellite product reported here is helpful to operational climate monitoring and prediction. However, the historical time series of the FY satellite data products is relatively short and the sample size is limited. Evaluation of the consistency of longer time series of SST products and identification of the improvement or degradation of the FY-3/VIRR SST products compared with the OSTIA SST (Donlon et al., 2012) will be considered in future work. Infrared instruments measure the skin SST at a high spatial resolution, but the surface is obscured by clouds, whereas microwave instruments provide an approximation to the sub-skin SST, including through clouds, thus having a fuller coverage at a reduced spatial resolution and > 50–100 km from land. Microwave SST products could be combined with the infrared SST in multi-payload SST product fusion processes to reduce the impact of clouds.

    Acknowledgments. The authors thank the two reviewers for helpful comments and suggestions.

Reference (51)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return