• Andersson, A., K. Fennig, C. Klepp, S. Bakan, H. Graßl, and J. Schulz, 2010: The Hamburg Ocean Atmosphere Parameters and Fluxes from Satellite Data—HOAPS-3. Earth Syst. Sci. Data, 2, 215234, https://doi.org/10.5194/essd-2-215-2010.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Arck, M., and D. Scherer, 2002: Problems in the determination of sensible heat flux over snow. Geogr. Ann., 84A, 157169, https://doi.org/10.1111/j.0435-3676.2002.00170.x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Auad, G., A. J. Miller, J. O. Roads, and D. Cayan, 2001: Pacific Ocean wind stress and surface heat flux anomalies from NCEP reanalysis and observations: Cross-statistics and ocean model responses. J. Geophys. Res., 106, 22 24922 265, https://doi.org/10.1029/2000JC000264.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bonan, G. B., F. S. Chapin, and S. L. Thompson, 1995: Boreal forest and tundra ecosystems as components of the climate system. Climatic Change, 29, 145167, https://doi.org/10.1007/BF01094014.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bonan, G. B., S. Levis, S. Sitch, M. Vertenstein, and K. W. Oleson, 2003: A dynamic global vegetation model for use with climate models: Concepts and description of simulated vegetation dynamics. Global Change Biol., 9, 15431566, https://doi.org/10.1046/j.1365-2486.2003.00681.x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bondeau, A., and et al. , 2007: Modelling the role of agriculture for the 20th century global terrestrial carbon balance. Global Change Biol., 13, 679706, https://doi.org/10.1111/j.1365-2486.2006.01305.x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bromwich, D. H., A. B. Wilson, L.-S. Bai, G. W. K. Moore, and P. Bauer, 2016: A comparison of the regional Arctic System Reanalysis and the global ERA-Interim Reanalysis for the Arctic. Quart. J. Roy. Meteor. Soc., 142, 644658, https://doi.org/10.1002/qj.2527.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bromwich, D. H., and et al. , 2018: The Arctic System Reanalysis, version 2. Bull. Amer. Meteor. Soc., 99, 805828, https://doi.org/10.1175/BAMS-D-16-0215.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brown, L., and C. R. Duguay, 2010: The response and role of ice cover in lake-climate interactions. Prog. Phys. Geogr. Earth Environ., 34, 671704, https://doi.org/10.1177/0309133310375653.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Broxton, P. D., X. Zeng, W. Scheftic, and P. A. Troch, 2014: A MODIS-based global 1-km maximum green vegetation fraction dataset. J. Appl. Meteor. Climatol., 53, 19962004, https://doi.org/10.1175/JAMC-D-13-0356.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Budyko, M. I., 1969: The effect of solar radiation variations on the climate of the Earth. Tellus, 21, 611619, https://doi.org/10.3402/tellusa.v21i5.10109.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cayan, D. R., 1992a: Latent and sensible heat flux anomalies over the northern oceans: Driving the sea surface temperature. J. Phys. Oceanogr., 22, 859881, https://doi.org/10.1175/1520-0485(1992)022<0859:LASHFA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cayan, D. R., 1992b: Latent and sensible heat flux anomalies over the northern oceans: The connection to monthly atmospheric circulation. J. Climate, 5, 354369, https://doi.org/10.1175/1520-0442(1992)005<0354:LASHFA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Checkley, D. M., and J. A. Barth, 2009: Patterns and processes in the California Current system. Prog. Oceanogr., 83, 4964, https://doi.org/10.1016/j.pocean.2009.07.028.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Decker, M., M. A. Brunke, Z. Wang, K. Sakaguchi, X. Zeng, and M. G. Bosilovich, 2012: Evaluation of the reanalysis products from GSFC, NCEP, and ECMWF using flux tower observations. J. Cilmate, 25, 19161944, https://doi.org/10.1175/JCLI-D-11-00004.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dee, D. P., and et al. , 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553597, https://doi.org/10.1002/qj.828.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dong, C., J. C. McWilliams, Y. Liu, and D. Chen, 2014: Global heat and salt transports by eddy movement. Nat. Commun., 5, 3294, https://doi.org/10.1038/ncomms4294.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Feldstein, S. B., 2000: Teleconnections and ENSO: The timescale, power spectra, and climate noise properties. J. Climate, 13, 44304440, https://doi.org/10.1175/1520-0442(2000)013<4430:TTPSAC>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fenning, K., A. Andersson, S. Bakan, C. Klepp, and M. Schroeder, 2012: Hamburg Ocean Atmosphere Parameters and fluxes from Satellite data - HOAPS 3.2 - monthly means/6-hourly composites. Satellite Application Facility on Climate Monitoring, accessed 22 February 2019, https://doi.org/10.5676/EUM_SAF_CM/HOAPS/V001.

    • Crossref
    • Export Citation
  • Friedl, M. A., D. Sulla-Menashe, B. Tan, A. Schneider, N. Ramankutty, A. Sibley, and X. Huang, 2010: MODIS collection 5 global land cover: Algorithm refinements and characterization of new datasets. Remote Sens. Environ., 114, 168182, https://doi.org/10.1016/j.rse.2009.08.016.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ganeshan, M., and D. L. Wu, 2016: The open-ocean sensible heat flux and its significance for Arctic boundary layer mixing during early fall. Atmos. Chem. Phys., 16, 13 17313 184, https://doi.org/10.5194/acp-16-13173-2016.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Geiselman, J., A. R. DeGange, K. Oakley, D. Derksen, and M. Whalen, 2012: Changing Arctic ecosystems: Research to understand and project changes in marine and terrestrial ecosystems of the Arctic. USGS Fact Sheet 2011-3136, http://pubs.usgs.gov/fs/2011/3136/.

    • Crossref
    • Export Citation
  • Gelaro, R., and et al. , 2017: The Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2). J. Climate, 30, 54195454, https://doi.org/10.1175/JCLI-D-16-0758.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gulev, S. K., and K. P. Baylev, 2012: Probability distribution characteristics for surface air-sea turbulent heat fluxes over the global ocean. J. Climate, 25, 184206, https://doi.org/10.1175/2011JCLI4211.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gulev, S. K., M. Latif, N. Keenlyside, W. Park, and K. P. Koltermann, 2013: North Atlantic Ocean control on surface heat flux on multidecadal timescales. Nature, 499, 464467, https://doi.org/10.1038/nature12268.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Harder, P., J. W. Pomeroy, and W. Helgason, 2017: Local-scale advection of sensible and latent heat during snowmelt. Geophys. Res. Lett., 44, 97699777, https://doi.org/10.1002/2017GL074394.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hurrell, J. W., and C. Deser, 2009: North Atlantic climate variability: The role of the North Atlantic Oscillation. J. Mar. Syst., 78, 2841, https://doi.org/10.1016/j.jmarsys.2008.11.026.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • IPCC, 2013: Climate Change 2013: The Physical Science Basis. Cambridge University Press, 1535 pp., https://doi.org/10.1017/CBO9781107415324.

    • Crossref
    • Export Citation
  • Jiménez, C., 2011: Global intercomparison of 12 land surface heat flux estimates. J. Geophys. Res., 116, D02102, https://doi.org/10.1029/2010JD014545.

    • Search Google Scholar
    • Export Citation
  • Jin, X., and L. Yu, 2013: Assessing high-resolution analysis of surface heat fluxes in the Gulf Stream region. J. Geophys. Res. Oceans, 118, 53535375, https://doi.org/10.1002/jgrc.20386.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jung, M., M. Reichstein, and A. Bondeau, 2009: Towards global empirical upscaling of FLUXNET eddy covariance observations: Validation of a model tree ensemble approach using a biosphere model. Biogeosciences, 6, 20012013, https://doi.org/10.5194/bg-6-2001-2009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jung, M., and et al. , 2011: Global patterns of land-atmosphere fluxes of carbon dioxide, latent heat, and sensible heat derived from eddy covariance, satellite, and meteorological observations. J. Geophys. Res., 116, G00J07, https://doi.org/10.1029/2010JG001566.

    • Search Google Scholar
    • Export Citation
  • Kent, E. C., D. I. Berry, J. Prytherch, and J. B. Roberts, 2014: A comparison of global marine surface-specific humidity datasets from in situ observations and atmospheric reanalysis. Int. J. Climatol., 34, 355376, https://doi.org/10.1002/joc.3691.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • King, A. D., G. J. van Oldenborgh, D. J. Karoly, S. C. Lewis, and H. Cullen, 2015: Attribution of the record high central England temperature of 2014 to anthropogenic influences. Environ. Res. Lett., 10, 054002, https://doi.org/10.1088/1748-9326/10/5/054002.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kobayashi, S., and et al. , 2015: The JRA-55 Reanalysis: General specifications and basic characteristics. J. Meteor. Soc. Japan, 93, 548, https://doi.org/10.2151/jmsj.2015-001.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Legates, D. R., and G. J. McCabe, 1999: Evaluating the use of “goodness-of-fit” measures in hydrologic and hydroclimatic model validation. Water Resour. Res., 35, 233241, https://doi.org/10.1029/1998WR900018.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lembo, V., I. Bordi, and A. Speranza, 2017: Annual and semiannual cycles of midlatitude near-surface temperature and tropospheric baroclinicity: Reanalysis data and AOGCM simulations. Earth Syst. Dyn., 8, 295312, https://doi.org/10.5194/esd-8-295-2017.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Li, D., T. Sun, M. Liu, L. Yang, L. Wang, and Z. Gao, 2015: Contrasting responses of urban and rural surface energy budgets to heat waves explain synergies between urban heat islands and heat waves. Environ. Res. Lett., 10, 054009, https://doi.org/10.1088/1748-9326/10/5/054009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lindsay, R., M. Wensnahan, A. Schweiger, and J. Zhang, 2014: Evaluation of seven different atmospheric reanalysis products in the Arctic. J. Climate, 27, 25882606, https://doi.org/10.1175/JCLI-D-13-00014.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Loranty, M. M., L. T. Berner, S. J. Goetz, Y. Jin, and J. T. Randerson, 2014: Vegetation controls on northern high latitude snow-albedo feedback: Observations and CMIP5 model simulations. Global Change Biol., 20, 594606, https://doi.org/10.1111/gcb.12391.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lu, Y., E. Wand, X. Yang, and J. Wang, 2010: Contributions of climatic and crop varietal changes to crop production in the North China Plain, since 1980s. Global Change Biol., 16, 22872299, https://doi.org/10.1111/j.1365-2486.2009.02077.x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Marks, D., A. Winstral, G. Flerchinger, M. Reba, J. Pomeroy, T. Link, and K. Elder, 2008: Comparing simulated and measured sensible and latent heat fluxes over snow under a pine canopy to improve an energy balance snowmelt model. J. Hydrometeor., 9, 15061522, https://doi.org/10.1175/2008JHM874.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Marzocchi, A., J. J.-M. Hirschi, N. P. Holliday, S. A. Cunningham, A. T. Blaker, and A. C. Coward, 2015: The North Atlantic subpolar circulation in an eddy-resolving global ocean model. J. Mar. Syst., 142, 126143, https://doi.org/10.1016/j.jmarsys.2014.10.007.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Monteith, J. L., 1965: Radiation and crops. Exp. Agric., 1, 241251, https://doi.org/10.1017/S0014479700021529.

  • Moore, G. W. K., I. Renfrew, and R. S. Pickart, 2012: Spatial distribution of air-sea heat fluxes over the sub-polar North Atlantic Ocean. Geophys. Res. Lett., 39, L18806, https://doi.org/10.1029/2012GL053097.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Moore, G. W. K., K. Vage, R. S. Pickart, and I. Renfrew, 2015: Decreasing intensity of open-ocean convection in the Greenland and Iceland Seas. Nat. Climate Change, 5, 877882, https://doi.org/10.1038/nclimate2688.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Moore, G. W. K., D. H. Bromwich, A. B. Wilson, I. Renfrew, and L. Bai, 2016: Arctic System Reanalysis improvements in topographically-forced winds near Greenland. Quart. J. Roy. Meteor. Soc., 142, 20332045, https://doi.org/10.1002/qj.2798.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • NASA, 2018: FLUXNET. Oak Ridge National Laboratory Distributed Active Archive Center, http://fluxnet.ornl.gov.

  • NCAR/UCAR and PMG/BPCRC/OSU 2017, 2017: Arctic System Reanalysis version 2. Research Data Archive at the National Center for Atmospheric Research, Computational and Information Systems Laboratory, accessed 5 September 2018, https://doi.org/10.5065/D6X9291B.

    • Crossref
    • Export Citation
  • Overland, J. E., and M. Wang, 2010: Large-scale atmospheric circulation changes are associated with the recent loss of Arctic sea ice. Tellus, 62A, 19, https://doi.org/10.1111/j.1600-0870.2009.00421.x.

    • Search Google Scholar
    • Export Citation
  • Pielke, R. A., G. Marland, R. A. Betts, T. N. Chase, J. L. Eastman, J. O. Niles, D. S. Niyogi, and S. W. Running, 2002:The influence of land-use change and landscape dynamics on the climate system: Relevance to climate-change policy beyond the radiative effect of greenhouse gases. Philos. Trans. Roy. Soc. London, 360A, 17051719, https://doi.org/10.1098/rsta.2002.1027.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Saha, S., and et al. , 2010: The NCEP Climate Forecast System Reanalysis. Bull. Amer. Meteor. Soc., 91, 10151057, https://doi.org/10.1175/2010BAMS3001.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Santorelli, A., R. T. Pinker, A. Bentamy, K. B. Katsaros, W. M. Drennan, A. M. Mestas-Nuñez, and J. A. Carton, 2011: Differences between two estimates of air-sea turbulent heat fluxes over the Atlantic Ocean. J. Geophys. Res., 116, C09028, https://doi.org/10.1029/2010JC006927.

    • Search Google Scholar
    • Export Citation
  • Shaman, J., R. Samelson, and E. Skyllingstad, 2010: Air–sea fluxes over the Gulf Stream region: Atmospheric controls and trends. J. Climate, 23, 26512670, https://doi.org/10.1175/2010JCLI3269.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Shi, Q., and S. Liang, 2014: Surface-sensible and latent heat fluxes over the Tibetan Plateau from ground measurements, reanalysis, and satellite data. Atmos. Chem. Phys., 14, 56595677, https://doi.org/10.5194/acp-14-5659-2014.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Shukla, J., C. Nobre, and P. Sellers, 1990: Amazon deforestation and climate change. Science, 247, 13221325, https://doi.org/10.1126/science.247.4948.1322.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Simmons, A. J., K. M. Willett, P. D. Jones, P. W. Thorne, and D. P. Dee, 2010: Low-frequency variations in surface atmospheric humidity, temperature, and precipitation: Inferences from reanalyses and monthly gridded observational data sets. J. Geophys. Res., 115, D01110, https://doi.org/10.1029/2009JD012442.

    • Search Google Scholar
    • Export Citation
  • Taylor, K. E., R. J. Stouffer, and G. A. Meehl, 2012: An overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc., 93, 485498, https://doi.org/10.1175/BAMS-D-11-00094.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Taylor, S. V., D. R. Cayan, N. E. Graham, and K. P. Georgakakos, 2008: Northerly surface winds over the eastern North Pacific Ocean in spring and summer. J. Geophys. Res., 113, D02110, https://doi.org/10.1029/2006JD008053.

    • Search Google Scholar
    • Export Citation
  • Tomita, H., M. Kubota, M. F. Cronin, S. Iwasaki, M. Konda, and H. Ichikawa, 2010: An assessment of surface heat fluxes from J-OFURO2 at the KEO and JKEO sites. J. Geophys. Res., 115, C03018, https://doi.org/10.1029/2009JC005545.

    • Search Google Scholar
    • Export Citation
  • Uppala, S. M., and et al. , 2005: The ERA-40 Re-Analysis. Quart. J. Roy. Meteor. Soc., 131, 29613012, https://doi.org/10.1256/qj.04.176.

  • Wild, M., and et al. , 2015: The energy balance over land and oceans: An assessment based on direct observations and CMIP5climate models. Climate Dyn., 44, 33933429, https://doi.org/10.1007/s00382-014-2430-z.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Willmott, C. J., 1981: On the validation of models. Phys. Geogr., 2, 184194, https://doi.org/10.1080/02723646.1981.10642213.

  • Wilson, K. B., and et al. , 2002: Energy partitioning between latent and sensible heat flux during the warm season at FLUXNET sites. Water Resour. Res., 38, 1294, https://doi.org/10.1029/2001WR000989.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wunsch, C., 2002: Ocean observations and the climate forecast problem. Int. Geophys., 83, 233245, https://doi.org/10.1016/S0074-6142(02)80170-X.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wyrtki, K., L. Magaard, and J. Hager, 1976: Eddy energy in the oceans. J. Geophys. Res., 81, 26412646, https://doi.org/10.1029/JC081i015p02641.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yashayaev, I. M., and I. I. Zveryaev, 2001: Climate of the seasonal cycle in the North Pacific and the North Atlantic Oceans. Int. J. Climatol., 21, 401417, https://doi.org/10.1002/joc.585.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, L., and J. Xiangze, 2014: Confidence and sensitivity study of the OAFlux multisensor synthesis of the global ocean surface vector wind from 1987 onward. J. Geophys. Res. Oceans, 119, 68426862, https://doi.org/10.1002/2014JC010194.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, L., X. Jin, and R. A. Weller, 2008: Multidecade global flux datasets from the Objectively Analyzed Air-sea Fluxes (OAFLUX) project: Latent and sensible heat fluxes, ocean evaporation, and related surface meteorological variables. OAFlux Project Tech. Rep. OA-2008-01, Woods Hole Oceanographic Institution, 64 pp., http://oaflux.whoi.edu/pdfs/OAFlux_TechReport_3rd_release.pdf.

  • Yu, L., Z. Zhang, S. Zhong, M. Zhou, Z. Gao, H. Wu, and B. Sun, 2011: An inter-comparison of six latent and sensible heat flux products over the Southern Ocean. Polar Res., 30, 10167, https://10.3402/polar.v30i0.10167.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhou, C., and K. Wang, 2016: Evaluation of surface fluxes in ERA-Interim using flux tower data. J. Climate, 29, 15731582, https://doi.org/10.1175/JCLI-D-15-0523.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • View in gallery

    Dominant land and vegetation types between 40° and 90°N according to Broxton et al. (2014). Colors for vegetation type match numbers and descriptions below color bar. Numbers on map indicate tower locations matched to their descriptions in Table 2.

  • View in gallery

    Time-averaged sensible heat flux (H; red) and latent heat flux (LE; black) for each of the towers (Fig. 1) over (a),(b) North America and (c),(d) Eurasia. Error bars display the 2-sigma spread of monthly values about the mean and give the seasonal variation and variation between years. (e)–(j) Boxplots show tower-averaged RMSE, coefficient of determination R2, and the Willmott index of agreement d for all reanalyses and BGI. The whiskers display error bars below and above the box indicating the 10th and 90th percentiles, and black dots characterize maximum and minimum values. The lower and upper boundaries of the box indicate the 25th and 75th percentile. The line within the box marks the 50th percentile (median).

  • View in gallery

    Time-averaged sensible heat flux H for (a) OAFLUX, (b) HOAPS, and (c) BGI. (d)–(f) As in (a)–(c), but for latent heat flux (LE; W m−2). Differences (OAFLUX minus HOAPS) are displayed for (g) H and (h) LE, where stippling indicates where differences are significant at the 99% confidence level. OAFLUX and HOAPS (BGI) only provide H and LE over oceanic (land) regions.

  • View in gallery

    Time-averaged sensible heat (H) differences (W m−2) between the reanalyses and OAFLUX for (a) ASRv2, (b) CFSR, (c) MERRA2, and (d) JRA-55 and HOAPS/BGI for (e) ASRv2, (f), CFSR, (g) MERRA2, and (h) JRA-55. Stippling indicates where differences are significant at the 99% confidence level.

  • View in gallery

    Time-averaged zonal mean 10-m wind speed over the (a) North Atlantic and (b) North Pacific for OAFLUX (black), HOAPS (purple), ASRv2 (red), CFSR (green), MERRA2 (dark blue), and JRA-55 (light blue). Differences between reanalyses and OAFLUX are shown in (c) and (d) for each ocean basin, respectively.

  • View in gallery

    Time-averaged latent heat flux (LE) differences (W m−2) between the reanalyses and OAFLUX for (a) ASRv2, (b) CFSR, (c) MERRA2, and (d) JRA-55 and HOAPS/BGI for (e) ASRv2, (f) CFSR, (g) MERRA2, and (h) JRA-55. Stippling indicates where differences are significant at the 99% confidence level.

  • View in gallery

    Amount of variance explained (percentage) by the first harmonic of sensible heat (H) for (a) OAFLUX, (b) HOAPS, and (c) BGI. Differences with respect to OAFLUX/BGI are shown for (d) ASRv2, (e) CFSR, (f) MERRA2, and (g) JRA-55.

  • View in gallery

    As in Fig. 7, but for the first harmonic amplitude of sensible heat (H; W m−2).

  • View in gallery

    As in Fig. 7, but for latent heat (LE).

  • View in gallery

    As in Fig. 8, but for latent heat (LE).

  • View in gallery

    Spatially averaged monthly evaporative fraction (EF) related to individual dominant vegetation patterns (indicated above each panel) for BGI (black), ASRv2 (red), CFSR (green), MERRA2 (light blue), and JRA-55 (purple).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 43 43 4
PDF Downloads 50 50 6

Northern Hemisphere Extratropical Turbulent Heat Fluxes in ASRv2 and Global Reanalyses

View More View Less
  • 1 Universidade Federal de Viçosa, Departamento de Engenharia Agricola, Viçosa, Minas Gerais, Brazil
  • | 2 Polar Meteorology Group, Byrd Polar and Climate Research Center, The Ohio State University, Columbus, Ohio
  • | 3 Universidade Federal de Viçosa, Departamento de Engenharia Agricola, Viçosa, Minas Gerais, Brazil
  • | 4 Polar Meteorology Group, Byrd Polar and Climate Research Center, The Ohio State University, Columbus, Ohio
© Get Permissions
Full access

Abstract

Large-scale objectively analyzed gridded products and satellite estimates of sensible (H) and latent (LE) heat fluxes over the extratropical Northern Hemisphere are compared to those derived from the regional Arctic System Reanalysis version 2 (ASRv2) and a selection of current-generation global reanalyses. Differences in H and LE among the reanalyses are strongly linked to the wind speed magnitudes and vegetation cover. Specifically, ASRv2 wind speeds match closely with observations over the northern oceans, leading to an improved representation of H compared to the global reanalyses. Comparison of evaporative fraction shows that the global reanalyses are characterized by a similar H and LE partitioning from April through September, and therefore exhibit weak intraseasonal variability. However, the higher horizontal resolution and weekly modification of the vegetation cover based on satellite data in ASRv2 provides an improved snow–albedo feedback related to changes in the leaf area index. Hence, ASRv2 better captures the small-scale processes associated with day-to-day vegetation feedbacks with particular improvements to the H over land. All of the reanalyses provide realistic dominant hemispheric patterns of H and LE and the locations of maximum and minimum fluxes, but they differ greatly with respect to magnitude. This is especially true for LE over oceanic regions. Therefore, uncertainties in heat fluxes remain that may be alleviated in reanalyses through improved representation of physical processes and enhanced assimilation of observations.

© 2019 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Aaron B. Wilson, wilson.1010@osu.edu

Abstract

Large-scale objectively analyzed gridded products and satellite estimates of sensible (H) and latent (LE) heat fluxes over the extratropical Northern Hemisphere are compared to those derived from the regional Arctic System Reanalysis version 2 (ASRv2) and a selection of current-generation global reanalyses. Differences in H and LE among the reanalyses are strongly linked to the wind speed magnitudes and vegetation cover. Specifically, ASRv2 wind speeds match closely with observations over the northern oceans, leading to an improved representation of H compared to the global reanalyses. Comparison of evaporative fraction shows that the global reanalyses are characterized by a similar H and LE partitioning from April through September, and therefore exhibit weak intraseasonal variability. However, the higher horizontal resolution and weekly modification of the vegetation cover based on satellite data in ASRv2 provides an improved snow–albedo feedback related to changes in the leaf area index. Hence, ASRv2 better captures the small-scale processes associated with day-to-day vegetation feedbacks with particular improvements to the H over land. All of the reanalyses provide realistic dominant hemispheric patterns of H and LE and the locations of maximum and minimum fluxes, but they differ greatly with respect to magnitude. This is especially true for LE over oceanic regions. Therefore, uncertainties in heat fluxes remain that may be alleviated in reanalyses through improved representation of physical processes and enhanced assimilation of observations.

© 2019 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Aaron B. Wilson, wilson.1010@osu.edu

1. Introduction

Earth’s climate and its seasonal cycle are strongly dependent on the interchange of water, momentum, and heat between the troposphere and the underlying surface (Monteith 1965; Budyko 1969). Modulation of the heat exchange, which may be largely represented by the surface net radiation and turbulent heat fluxes [sensible (H) and latent (LE)], is tightly linked to the intensity of the air–surface coupling. Moreover, this two way interaction, which over land involves the biosphere–atmosphere interaction, also varies substantially in the space–time domain (Wyrtki et al. 1976; Bonan et al. 1995).

As extensively discussed in previous modeling and observational studies, the driving force for this interchange insofar as heat and water are concerned, is the available energy which is linked to water vapor pressure and the temperature gradient (e.g., Monteith 1965; Pielke et al. 2002). This complex interaction is connected to the thermal and dynamic behavior of the troposphere induced by sea ice and surface temperature conditions (e.g., Hurrell and Deser 2009). For instance, changes in climate, other environmental conditions, and population dynamics of endangered species in the Northern Hemisphere (NH) polar region are all related to the surface–air coupling (Overland and Wang 2010; Geiselman et al. 2012; IPCC 2013).

On larger scales, accelerated human emissions of greenhouse gases have altered the planetary energy budget and are linked to recent climate change (e.g., King et al. 2015). However, the surface energy budget includes net shortwave and longwave radiation, LE, H, and ground heat fluxes. Among these terms, the H and LE play a substantial role in driving local and global temperatures on monthly to interannual time scales (Pielke et al. 2002).

Thus, assessing the temporal and spatial variability of H and LE in the extratropical regions is essential to understanding both the local and global processes governing climate. Such comparisons elucidate possible drawbacks with respect to vegetation and surface oceanic conditions in climate models, reanalyses, and satellite observations (Cayan 1992b; Lindsay et al. 2014; Zhou and Wang 2016). Evaluations of H and LE have been conducted using a variety of observational networks (Jung et al. 2009; Decker et al. 2012), reanalyses (Jiménez 2011; Zhou and Wang 2016), and global climate models (Wild et al. 2015). For instance, Cayan (1992b) and Hurrell and Deser (2009) demonstrate that the interactions between atmospheric circulation and oceanic H and LE are associated with wind changes on the basin scale. The majority of studies concerning turbulent fluxes are however, conducted for key regions such as those over the oceanic western boundary currents (Jin and Yu 2013), open water (Cayan 1992a,b; Auad et al. 2001; Jin and Yu 2013), or dominant vegetation patterns such as tropical and boreal forests (Shukla et al. 1990; Loranty et al. 2014).

Wild et al. (2015) performed a large-scale intercomparison of the energy budget for 43 models from phase 5 of the Coupled Model Intercomparison Project (CMIP5; Taylor et al. 2012), European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis products, and surface and satellite observations. They showed that the CMIP5 H and LE vary greatly over land among the models, which may result in large biases in the water cycle. Despite differences among the reanalyses and observations, Jiménez (2011) demonstrated that these products correlate well with each other due to similarities related to dominant climatic regimes and geographical features. Indeed, heat flux differences arise primarily from limitations in simulating the contribution of small-scale land/ocean–atmospheric processes that are not resolved in low-resolution latitude–longitude domains (Cayan 1992b; Tomita et al. 2010; Jin and Yu 2013). Simplifications in intrinsic parameterizations of the land surface schemes involving aerodynamic and canopy resistance as well as the wind field have also been associated with the misrepresentations of H and LE (Santorelli et al. 2011; Lindsay et al. 2014; Zhou and Wang 2016). Unfortunately, there are few choices when it comes to accurately comparing turbulent heat fluxes across the Arctic.

The need to understand and properly represent turbulent heat fluxes in the Arctic region is crucial for many reasons. Heat fluxes are important components of the snow-cover energy balance (Marks et al. 2008). They subsequently affect the rate of snow melting and modify the land albedo and atmosphere–biosphere interaction. The influence of H in the Arctic dynamic boundary layer clouds has been discussed by Ganeshan and Wu (2016) who argue that H increases turbulence in the boundary layer in the presence of low pressure systems. This induces changes in the open-ocean boundary layer height and vertical mixing. Moore et al. (2012, 2015) highlight the particular importance of heat fluxes and changes that have occurred due to retreating wintertime sea ice extent over the North Atlantic that are vital to the Atlantic meridional overturning circulation. In this sense, improvements in the quality of global climate models (GCMs) simulations in this highly sensitive region are tightly dependent on reasonable parameterizations of surface conditions (Arck and Scherer 2002).

A newer, higher-resolution reanalysis of the greater Arctic region, the Arctic System Reanalysis (ASR; Bromwich et al. 2016, 2018) has demonstrated promise for improving surface heat fluxes. Version 2 of ASR (herein ASRv2; Bromwich et al. 2018) has a high resolution in space (15 km), with its first full-model level at 4 m above ground and over 25 levels below 850 hPa. ASRv2 shows high skill in representing near-surface and upper-air analysis fields (temperature, humidity, winds), as the high horizontal resolution provides a more realistic topography that has been shown to improve near-surface winds and topographically forced wind events throughout the Arctic (Bromwich et al. 2018; Moore et al. 2016). Dense vertical levels in the boundary layer help capture the surface–atmosphere interactions. Snow cover is updated daily based on the National Centers for Environmental Prediction Final Analysis, and ASRv2 employs a weekly modification of vegetation cover and snow albedo based on Moderate Resolution Imaging Spectroradiometer (MODIS) data. These carefully constructed parameters improve processes related to vegetation changes and snow–albedo feedbacks, making ASRv2 particularly useful in the study of heat fluxes (Bromwich et al. 2018).

The purpose of this investigation is to detail differences in the broad-scale spatial–temporal variability of H and LE over Arctic and extratropical land and ice-free ocean areas as reproduced by reanalyses. Specifically, we investigate the potential mechanisms related to differences in H and LE among these datasets. By evaluating the newest generation of reanalyses including ASRv2, we verify the capabilities and caveats of these new methods and model parameterizations for land surface and air–sea interactions on a large-scale domain. Section 2 describes the data used and the methodological design. It also includes a description of long-term means as represented by the blended and satellite datasets. Section 3 demonstrates the representation of the annual cycle of H and LE, in terms of amplitude and variance, and identifies key relationships between the turbulent fluxes and surface conditions in the space–time domain. Finally, section 4 summarizes the results and discusses the implications of these findings.

2. Data and methods

a. Data

This study focuses on NH Arctic and extratropical regions over both continental and oceanic regions (Fig. 1). One goal is to further understand the influence of distinct vegetation types on H and LE across a hemispheric domain. Evaluation of interannual and seasonal variability of H and LE relies on estimates from eddy covariance methods using FLUXNET (NASA 2018), a global network of meteorological tower sites. Figure 1 shows that the sites selected for this investigation include a variety of land and vegetation types, from barren or snow-ice covered to woodlands, grasslands, and croplands.

Fig. 1.
Fig. 1.

Dominant land and vegetation types between 40° and 90°N according to Broxton et al. (2014). Colors for vegetation type match numbers and descriptions below color bar. Numbers on map indicate tower locations matched to their descriptions in Table 2.

Citation: Journal of Climate 32, 7; 10.1175/JCLI-D-18-0535.1

Turbulent fluxes over land are based on blended data (satellite, observations, and reanalyses) provided by the Department for Biogeochemical Integration of the Max-Planck Institute for Biogeochemistry (BGI; Jung et al. 2009, 2011), and flux data over oceans are from the Objectively Analyzed Air–Sea Fluxes (OAFLUX) for the Global Ocean Project (Yu et al. 2008) from the Woods Hole Oceanographic Institution. The BGI dataset was created by upscaling eddy covariance measurements from the current global network of eddy covariance towers (FLUXNET) coupled to a series of mathematical algorithms and modeling output derived from Lund–Potsdam–Jena managed Land (LPJmL) biosphere model (Bondeau et al. 2007). This approach provides a global distribution of H and LE on a regular 0.5° × 0.5° latitude and longitude (lat–lon) grid over land. The OAFLUX ocean fluxes are based on satellite information, surface observations, and an optimal blending of this data with reanalyses from the National Centers for Environmental Prediction (NCEP) and ECMWF.

The satellite data used in this study is the Hamburg Ocean Atmosphere Parameters and Fluxes (HOAPS; Andersson et al. 2010; Fenning et al. 2012). The HOAPS dataset is based on SSM/I passive microwave radiometers over the ice-free ocean surface on a regular lat–lon grid with a spatial resolution of 0.5° × 0.5°. Yu et al. (2011) have provided an in-depth description of the above datasets as well as discussed their performance in Southern Hemisphere (SH) oceanic regions.

Reanalysis data include ASRv2 (NCAR/UCAR and PMG/BPCRC/OSU 2017; Bromwich et al. 2018), the NCEP Climate Forecast System Reanalysis (CFSR; Saha et al. 2010), the ECMWF interim reanalysis (ERA-I; Dee et al. 2011), the Japanese 55-yr Reanalysis (JRA-55; Kobayashi et al. 2015), and the National Aeronautics and Space Administration (NASA) Modern-Era Retrospective Analysis for Research and Applications version 2 (MERRA2; Gelaro et al. 2017). See Table 1 for more details. Noteworthy, there has been no investigation regarding the large-scale spatial distribution of H and LE as depicted by ASRv2, CFSR, MERRA2, or JRA-55.

Table 1.

List of datasets used in this study.

Table 1.

Previous investigations and comparisons among reanalyses have been conducted for near-surface and upper-level temperature, dewpoint, relative humidity, pressure (geopotential height), and wind (e.g., Simmons et al. 2010; Kent et al. 2014; Bromwich et al. 2016, 2018, and references therein). Others have provided a similar evaluation between the OAFLUX and BGI data to reanalyses (Yu et al. 2008; Jung et al. 2009; Jiménez 2011; Jin and Yu 2013; Kent et al. 2014). Decker et al. (2012) compared flux tower observations to global reanalyses, concluding that the ERA-40 (Uppala et al. 2005) has the lowest overall bias in H and LE. They also argue that the CFSR data have discontinuities associated with a mismatch between the oceanic and atmospheric assimilation cycles. Zhou and Wang (2016) compare H and LE in ERA-I with flux tower data from mostly over North America and demonstrate that these fluxes in ERA-I are tightly linked to the model-represented vegetation cover. They also found that due to misrepresentation of surface winds, H is underestimated along with the evaporative fraction. This further affects the quality of basic variables such as surface and air temperatures and precipitation in the reanalyses.

Over oceanic regions, such as the Gulf Stream region (western boundary current in the North Atlantic Ocean), studies have indicated that the high-resolution OAFLUX data accurately reproduce the H and LE associated with the oceanic eddies sampled by buoy measurements (Jin and Yu 2013). These authors also show that the CFSR exhibits the smallest bias when compared to the NCEP–NCAR, MERRA, and ERA-I. Evaluation of the Kuroshio region (western boundary current in the North Pacific Ocean) demonstrated that the satellite-derived HOAPS reasonably reproduces the H and LE estimates, but seasonal biases were identified (Tomita et al. 2010). Most of these caveats in the oceanic turbulent fluxes arise from the misrepresentation of wind speed and sea surface temperature (SST) as demonstrated by Santorelli et al. (2011). Our discussion incorporates previous intercomparisons of H and LE based on modeling, reanalyses, and flux tower datasets.

b. Methods

Tower fluxes are compared to reanalyses and blended datasets on their native grids. We compare simulated H and LE to the FLUXNET tower measurements based on eddy covariance calculations, northward of 40° (Fig. 1). The root-mean-square error (RMSE), the coefficient of determination (R2), and the Willmott index of agreement (d) are used. The Willmott index of agreement (Willmott 1981) is a standardized measure of the degree of model temporal-prediction error and varies between 0 and 1. A value of 1 indicates a perfect match, and 0 indicates no agreement at al. This particular statistic detects additive and proportional differences in means and variances but can be overly sensitive to extreme values due to the squared differences (Legates and McCabe 1999).

To compute spatial differences among the reanalyses and OAFLUX, HOAPS, and BGI datasets, data have been interpolated to a 0.5° × 1° lat–lon grid. Finer resolution in the meridional direction was chosen as global H and LE are strongly governed by processes directed by the annual march of the sun (e.g., Budyko 1969; Wunsch 2002). Over land and at monthly time scales, these fluxes are controlled by surface processes associated with vegetation–atmosphere interactions that vary in response to the seasonal radiative budget (Bonan et al. 2003). Heat transport due to mesoscale oceanic eddies, which are mostly meridional, may also be captured more accurately on a 0.5° × 1° lat–lon grid (e.g., Dong et al. 2014). Our results reveal no significant differences with other resolutions (0.5° × 0.5° or 1° × 1°); thus, any of these resolutions may be used insofar as evaluations for the hemispheric domain are concerned. Although reanalyses are compared to site-specific data such as those provided by flux towers, our primary goal is to provide a hemispheric evaluation of the differences between reanalyses and their ability to reproduce large-scale blended H and LE fluxes.

The period of study is based on 11 years (2000–10). This interval is used for the reanalyses, OAFLUX, and BGI, but data from HOAPS are only available from 2000 to 2008. Results are presented as long-term means and amplitudes of the seasonal cycle (the first harmonic of Fourier transform). The first-order harmonic shows different climate regimes and transition regions that are important for understanding the changes in H and LE in more detail by tracing their linkages to land and ocean characteristics (Jiménez 2011).

Additional analyses are conducted to evaluate the effect of vegetation cover on the ratio between H and LE through the calculation of the evaporative fraction [EF = LE/(LE + H)]. Monthly mean changes in EF for major vegetation classes in the NH extratropics are computed according to the MODIS Land Cover from 2001 to 2010 (Friedl et al. 2010). We chose the EF instead of the β = H/LE, because as LE decreases to zero, β is not appropriate. Furthermore, by evaluating the EF, potential problems associated with the energy balance closure observed in the eddy covariance method may be partially alleviated (Jiménez 2011).

3. Results

a. Tower flux analysis

As shown in Table 2, the FLUXNET dataset spans different time intervals and vegetation types. Some of these towers have vegetation types (microclimates) that are different than surrounding areas (e.g., site 7–USOho is over a deciduous broadleaf forest surrounded mainly by cropland; Fig. 1). Figure 2 shows the FLUXNET means and the monthly amplitudes of H and LE based on eight (nine) towers across North America (Eurasia). Monthly average H varies 8–44 W m−2 in North America and 0–28 W m−2 in Eurasia, while LE ranges are 6–85 W m−2 for North America and 10–65 W m−2 for Eurasia.

Table 2.

Descriptions of meteorological towers used in this study. Number (No.) indicates the tower locations displayed in Fig 1.

Table 2.
Fig. 2.
Fig. 2.

Time-averaged sensible heat flux (H; red) and latent heat flux (LE; black) for each of the towers (Fig. 1) over (a),(b) North America and (c),(d) Eurasia. Error bars display the 2-sigma spread of monthly values about the mean and give the seasonal variation and variation between years. (e)–(j) Boxplots show tower-averaged RMSE, coefficient of determination R2, and the Willmott index of agreement d for all reanalyses and BGI. The whiskers display error bars below and above the box indicating the 10th and 90th percentiles, and black dots characterize maximum and minimum values. The lower and upper boundaries of the box indicate the 25th and 75th percentile. The line within the box marks the 50th percentile (median).

Citation: Journal of Climate 32, 7; 10.1175/JCLI-D-18-0535.1

Figure 2 demonstrates moderate spatial and temporal variability for H and LE, and specific sites highlight the importance of land surface and vegetation types on heat fluxes. In North America, the ATQ tower in northern Alaska is located in a tundra climate over a grassland land cover and shows the lowest values of H (Fig. 2a) and LE (Fig. 2b). Colder temperatures and a short “warm” season limit evapotranspiration rates at this Arctic site. The greatest annual values and variability of H generally occur over forests (e.g., USHa1 and USHo1) compared to other types in North America.

Geographic location plays an important role in heat flux variability as well. The CATP1 tower for instance, which has a cropland land cover in an area of primarily forested land, exhibits the largest amplitude of H but one of the lowest means (Fig. 2a). This site is situated just north of Lake Erie in North America, which drives local and regional weather variability due to its relatively large size but shallow depth. Cold, Canadian air masses repeatedly moving over Lake Erie during the autumn and winter lead to lake freeze-up (Brown and Duguay 2010), and lingering ice cover in the early part of spring despite frequent warm spells impacts temperatures at the nearby tower site. Additionally, the cooler water body has a mitigating impact on summer heat over land. This results in large month-to-month differences in temperature between the canopy and the overlying air, further increasing H variability. Although some tower sites are located over similar land cover [e.g., CATP1 and USNe3 (cropland) or CAQfo and USHo1 (forest types)], they individually deliver distinct temporal variability. Likewise, some may be impacted by changes in heat and moisture advection from surrounding types (e.g., USNe3—a cropland surrounded by grassland; Fig. 1).

Mean values of H for most of stations across Eurasia are much smaller than North America (6–20 W m−2; Fig. 2c), while LE reflects similar magnitudes and variability between the two regions (Fig. 2d). Eurasian sites also demonstrate less spatial variation of H compared to North America. Unlike the wide vegetation diversity of selected sites across North America (Fig. 1), the vegetation distribution across Eurasia is limited with sites primarily located over croplands and forests (central European) environments. In both regions, the largest mean values of LE occur over the southern locations where evaporation is likely to be strongest (e.g., USNe3, USOho, USSdh, ATNeu, and FRLq1).

Reanalysis heat fluxes for each corresponding tower site based on the closest latitude and longitude grid point are compared to the tower data. Figures 2e–j show the tower-averaged RMSE, R2, and d. BGI and ERA-I are expected to deliver similar results as the former is forced by the latter to improve the interpolation over larger areas lacking observations. These two datasets generally show the greatest precision in H and LE revealed by the R2 and d indices, and the lowest RMSE (except for JRA-55 which shows lower RMSE for North America). The CFSR and MERRA2 have the lowest R2 and d and the highest RMSE; but overall, there is not much statistical difference among the reanalyses (especially for the Eurasian stations).

In general, the interannual variabilities of H and LE are well captured by BGI, ERA-I, ASRv2, and JRA-55. Reanalyses show smaller deviations from observations for H in both North America and Eurasia as demonstrated by the RMSE, but LE has a larger spread (Figs. 2e,f). Good matches between reanalyses and towers are found for LE with greater R2 and d (generally >0.8). Reanalyses show a large spatial variability in H compared to observed values over North America resulting in lower R2 and d (Figs. 2g,i). Across Eurasia, both heat fluxes are well reproduced, with lower RMSE and higher R2 and d compared to North America (Figs. 2f,h,j).

It is important to remember that the values extracted from reanalyses represent mean grid values which are associated with the dominant vegetation type within each grid box (though tiling is used in some cases). These grid boxes may not exhibit the specific surface cover as the tower observations, and procedures vary among reanalyses. This can have significant implications for the comparison between the reanalyses and the heat flux observations. Other potential sources for discrepancies are related to the height at which the fluxes are computed in the reanalysis data. In most cases, H and LE produced by reanalyses are computed at levels that differ from tower heights that also depend on the vegetation type. However, this comparison has demonstrated that ASRv2 compares similarly among other reanalyses with respect to the tower sites. The following comparison utilizes data that blend out these small microclimates, which are not well represented by model grid boxes.

b. Annual means

1) Blended data

As previously mentioned, HOAPS and OAFLUX (BGI) provide H and LE fluxes over ocean (land) areas. The spatial distribution of the time mean H and LE show that the OAFLUX and HOAPS demonstrate similar patterns over both the North Pacific and Atlantic Oceans (Figs. 3a,b,d,e); H and LE maxima are found northward of the Kuroshio, Gulf Stream, and North Atlantic Drift, in agreement with buoy estimates (e.g., Tomita et al. 2010; Jin and Yu 2013). It is important to note that the inner domain of ASRv2 does not include the southern extents of the Gulf Stream or the Kuroshio, but they are present in the outer domain.

Fig. 3.
Fig. 3.

Time-averaged sensible heat flux H for (a) OAFLUX, (b) HOAPS, and (c) BGI. (d)–(f) As in (a)–(c), but for latent heat flux (LE; W m−2). Differences (OAFLUX minus HOAPS) are displayed for (g) H and (h) LE, where stippling indicates where differences are significant at the 99% confidence level. OAFLUX and HOAPS (BGI) only provide H and LE over oceanic (land) regions.

Citation: Journal of Climate 32, 7; 10.1175/JCLI-D-18-0535.1

Comparing five heat flux products, Tomita et al. (2010) argue that HOAPS reproduces annual conditions in good agreement with Kuroshio Extension Observatory (KEO buoy) with the lowest summer bias among all datasets (although a negative bias is found for the winter season). The OAFLUX also compares well with observations, indicated by correlations greater than 0.70 in summer and 0.93 in winter (despite a larger positive bias in summer). For the eddy-rich Gulf Stream region, Jin and Yu (2013) demonstrate that the OAFLUX accurately reproduces H based on estimates from moored buoys, with differences of about 8%.

Despite similarities in the spatial distribution between the H and LE fluxes in all datasets (Figs. 3a,b,d,e), LE values are in general 3 times larger than H, reaching up to 120 W m−2 along the North American and Asian coastal margins. The North Atlantic contributes more to the NH heat fluxes than the North Pacific. However, differences arise between OAFLUX and HOAPS over the North Pacific basin (40°–80°N, Figs. 3g,h), which are related to larger SST–air temperature and wind speed differences in the HOAPS dataset (stippled regions in Figs. 3g,h). It does not apply to the North Atlantic open water, northward of 40°N, where most of differences are not statistically significant at the 99% confidence level (based on Student’s t-test calculation).

Figures 3c and 3f show annual mean conditions over land as depicted by the BGI dataset. The H and LE fluxes are smaller than their oceanic counterparts. Southward of 60°N, the H and LE show dominant patterns where maximum LE is balanced by minimum H and vice versa, in line with what is expected by the H/LE ratio. This is clear over the North America midlatitudes and central Asia. Indeed, this primarily reproduces biophysical aspects of vegetation cover. Over western North America and central Asia, higher H is associated with grassland, semidesert, and shrubs that have limited capacity for evapotranspiration. In the presence of these vegetation types, less energy will be delivered to evapotranspiration as compared to the energy spent to warm the adjacent air.

On the other hand, lower H (higher LE) is positively correlated with evergreen forests such as needleleaf and broadleaf types in the eastern and northern sectors of North America and Eurasia (vegetation types 1 to 5 in Fig. 1). Polar land surface types such as tundra and taiga, which are dominant over the northern extents of Canada and Siberia, do not show substantial differences regarding the magnitudes of H and LE fluxes.

2) Reanalyses

(i) Sensible heat flux

As previously mentioned, because BGI uses ERA-I to adjust its local mean seasonal cycles, ERA-I spatial comparisons will not be included in further discussion. Comparing H in the reanalyses to OAFLUX (Fig. 4), a similar bias pattern is found in the North Pacific for ASRv2, MERRA2, and JRA-55, though all of the reanalyses differ strongly with respect to their magnitudes. The ASRv2 is dominated by greater H compared to other reanalyses (except JRA-55) and OAFLUX (Fig. 4a). MERRA2, and to a lesser extent JRA-55, show less H along the coastal margins, in particular northward of the Kuroshio region (Figs. 4c,d). Differences among reanalyses are also highlighted by the CFSR pattern in both the Atlantic and Pacific (Fig. 4b). However, the CFSR’s anomalous distribution shows statistically significant differences at 99% level (dotted region in Fig. 4b) in the eastern Atlantic and Pacific Oceans only. In the northeastern North Atlantic, ASRv2, MERRA2, and JRA-55 exhibit larger H compared to OAFLUX, more so in ASRv2 northward of the Gulf Stream and the Labrador Current regions (Fig. 4a). ASRv2 and JRA-55 also show larger differences along the coastal regions of North America, with statistically significant (99%) differences across the entire basin.

Fig. 4.
Fig. 4.

Time-averaged sensible heat (H) differences (W m−2) between the reanalyses and OAFLUX for (a) ASRv2, (b) CFSR, (c) MERRA2, and (d) JRA-55 and HOAPS/BGI for (e) ASRv2, (f), CFSR, (g) MERRA2, and (h) JRA-55. Stippling indicates where differences are significant at the 99% confidence level.

Citation: Journal of Climate 32, 7; 10.1175/JCLI-D-18-0535.1

HOAPS generally depicts greater H over the Pacific Ocean compared to OAFLUX (except in the Bering Sea and Sea of Okhotsk; Fig. 3g). Figures 4e–h reveal that ASRv2, MERRA2, and JRA-55 agree closely with HOAPS in the North Pacific. Differences are generally not statistically significant between HOAPS and ASRv2 (Fig. 4e), with values between ±4 W m−2. Across the North Atlantic, however, reanalyses overestimate H in the western boundary currents (the Gulf Stream and Labrador). ASRv2 and JRA-55 show similar patterns, whereas CFSR and MERRA2 exhibit less H comparatively.

The anomalous patterns of H in the reanalyses may be understood from the perspective of the near-surface air–sea temperature gradient (dT; not shown), and the capability of reanalyses to reproduce the near-surface wind. Across the Arctic Ocean, reanalyses show less H compared to the OAFLUX. (Comparison between dT in the HOAPS dataset was not conducted as air temperature is not directly available.) In particular, differences in ASRv2 and CFSR (other reanalyses) are associated with greater (less) air–sea thermal contrast over the seasonally free sea ice region (not shown). This relationship is not straightforward over open water in the midlatitudes where the atmospheric circulation is dominant, especially in the North Atlantic (Cayan 1992a,b; Gulev et al. 2013). Over the North Pacific, where intraseasonal variability is weaker, good correlations between dT and H are found for ASRv2. This cannot be said for CFSR, MERRA2, or JRA-55.

Likewise, differences in the wind speed may be an important source for error in H. Cayan (1992b) and Lindsay et al. (2014) demonstrate that wind speed biases in the global reanalyses are 2–3.5 m s−1 over most of the Arctic. Figure 5 supports these results, as 10-m wind speeds over the ocean in the reanalyses capture the zonal profile, but differ from OAFLUX in the North Atlantic by 0.5 m s−1 between 40° and 65°N (Figs. 5a,c) but 1–2.0 m s−1 northward of 65°N.

Fig. 5.
Fig. 5.

Time-averaged zonal mean 10-m wind speed over the (a) North Atlantic and (b) North Pacific for OAFLUX (black), HOAPS (purple), ASRv2 (red), CFSR (green), MERRA2 (dark blue), and JRA-55 (light blue). Differences between reanalyses and OAFLUX are shown in (c) and (d) for each ocean basin, respectively.

Citation: Journal of Climate 32, 7; 10.1175/JCLI-D-18-0535.1

CFSR and JRA-55 are dominated by stronger zonally averaged 10-m wind speed, whereas MERRA2 is weaker with respect to OAFLUX and HOAPS (Fig. 5c). The close agreement in wind speed between the ASRv2 and OAFLUX should be emphasized, particularly in midlatitudes. Similar results among the reanalyses are observed over the North Pacific, including the peak near 65°N, where the zonal mean consists mostly of 10-m wind speeds over the continental areas of Siberia and Alaska. All of the reanalyses are stronger than OAFLUX in this region. As suggested by Yu and Xiangze (2014), OAFLUX may exhibit difficulties in reproducing high wind magnitudes or winds during heavy rain.

Correlations between 10-m wind speed and H demonstrate values as high as 0.80 in open water and 0.90 over the North America and Asia continental margins (not shown). On the basinwide scale, correlations between 10-m wind speed and the heat flux anomalies are not local but are highly dependent on the North Atlantic and North Pacific atmospheric circulation, and therefore may be influenced by the Arctic Oscillation and the North Atlantic Oscillation (NAO; Gulev et al. 2013).

Thus, differences relative to OAFLUX as demonstrated by the global reanalyses may arise from both surface wind misrepresentation itself and coupled to distinct patterns of dT. While ASRv2’s depiction of dT is reasonable, the near-surface wind field has been demonstrated to be superior to other global reanalyses (Moore et al. 2016; Bromwich et al. 2018), resulting in greater H compared to OAFLUX. This indicates that nonlinearities in the ocean–atmosphere coupling strongly determine the magnitude of H over open water in ASRv2.

Sensible heat flux is an important driver in defining the regional and global climate due to its close relationship with vegetation cover (e.g., Bonan et al. 1995). On long time scales, boreal forests and tundra are crucial in determining the global atmospheric carbon dioxide and methane levels. On seasonal time scales, they control the snow–vegetation albedo feedback that may lead to modifications in the partitioning of H and LE (Loranty et al. 2014). Figure 4e shows that over land northward of about 55°N, differences between BGI and ASRv2 are generally not statistically significant, with values between ±4 W m−2. Bromwich et al. (2018) found that ASRv2 wind speed and surface temperatures exhibit significantly smaller biases compared to ERA-I, attributed to the high-resolution terrain and a detailed land surface representation in ASRv2.

Larger differences in H are found across Asia, Europe, and North America as shown by other reanalyses (Figs. 4f–h), with statistically significant values as low as −30 W m−2. The global reanalyses 10-m wind speed (surface temperatures) is negatively (positively) correlated with H, with values less (greater) than −0.7 (0.8). In ASRv2, however, over most of Eurasia and North America, 10-m wind speed and H are in phase (correlations up to 0.8), which demonstrates the important role of the wind field as a driver of the H distribution.

(ii) Latent heat flux

Turning to the LE (Fig. 6), it is demonstrated that insofar as annual mean conditions are concerned, the reanalyses overestimate LE compared to OAFLUX over open water (Fig. 6). In ASRv2, this feature is likely related to water vapor differences (though temperature is a factor), because 10-m wind speeds in ASRv2 and OAFLUX are similar. It should be noted that the pattern of differences in ASRv2 resembles that of other reanalyses but differs in magnitude. In the North Atlantic, differences are generally most pronounced east of 30°W.

Fig. 6.
Fig. 6.

Time-averaged latent heat flux (LE) differences (W m−2) between the reanalyses and OAFLUX for (a) ASRv2, (b) CFSR, (c) MERRA2, and (d) JRA-55 and HOAPS/BGI for (e) ASRv2, (f) CFSR, (g) MERRA2, and (h) JRA-55. Stippling indicates where differences are significant at the 99% confidence level.

Citation: Journal of Climate 32, 7; 10.1175/JCLI-D-18-0535.1

Differences between reanalyses and HOAPS demonstrate that ASRv2 and JRA-55 show stronger correspondence with HOAPS over the North Pacific than CFSR and MERRA2. ASRv2 differences are not statistically significant. Differences in the North Atlantic reach values as high as 20 W m−2 but are in close agreement with dT differences, and thus changes in water vapor. Northward of 65°N, changes are also likely related to the wind speed magnitudes (Fig. 4).

Over the continents, good correspondence is found among the reanalyses in particular over northern Eurasia and Canada (Fig. 6). In fact, the reanalyses show strong individual skill for some regions. MERRA2 shows larger departures from BGI northward of 65°N, whereas all other reanalyses show larger differences in midlatitudes. With the accurate representation of surface air and dewpoint temperatures and surface winds compared to ERA-I (Moore et al. 2016; Bromwich et al. 2018), and the weekly updates to snow cover and vegetation fraction allowing for more accurate biosphere–atmosphere interchange, the larger differences in LE over continental regions in ASRv2 are surprising. It has to be mentioned, however, that both H and LE are tightly dependent on the transfer coefficients, thus adding some uncertainty to the ASRv2 results.

c. Seasonal cycle

Previous investigations of H and LE have demonstrated that most disagreement between observations, OAFLUX, HOAPS, and BGI are associated with the misrepresentation of the seasonal cycle (Yu et al. 2008; Andersson et al. 2010; Tomita et al. 2010; Jiménez 2011; Jung et al. 2011; Santorelli et al. 2011; Jin and Yu 2013). This primarily arises from limitations in simulating seasonal changes in mesoscale surface winds. In most cases, this is associated with the land–sea breeze or influenced by topographically driven wind patterns (Shi and Liang 2014). For instance, reduced thermal contrast between the land and the oceanic western boundary in both Pacific and the Atlantic regions has been assumed to be the cause of H and LE errors in reanalyses and OAFLUX (Yu et al. 2008; Jin and Yu 2013). Seasonal changes in vegetation have also been demonstrated to result in an amplification of the H and LE annual variability (e.g., Loranty et al. 2014; Shi and Liang 2014).

The dominance of the seasonal cycle in a time series is defined by its explained variance. For example, large-scale atmospheric or oceanic processes are primarily influenced by Earth’s march around the sun, and the first-order harmonic of H/LE reflects long-term effects. Higher-order harmonics show the effects of short-term fluctuations.

Figure 7 demonstrates that over most of the ocean, the first harmonic explains more than 80% of the variance, highlighting the importance of the seasonal cycle. This is reasonable considering the seasonal atmospheric westerly flow is highly correlated with open water heat fluxes in the extratropics (e.g., Auad et al. 2001). However, over the eastern North Atlantic and Pacific, other harmonics also influence the time variability (Figs. 7a,b). These regions are largely influenced by regional wind patterns associated with the Azores high and the Aleutian low, which experience significant day-to-day variability. In the North Pacific, this occurs through Ekman dynamics changes in the SST-driven upwelling flow, which is embedded in both offshore and nearshore variability (Checkley and Barth 2009). In turn, this modifies the near-surface atmospheric stability resulting in differences between the surface temperature and the near-surface air temperature on shorter time scales (Taylor et al. 2008).

Fig. 7.
Fig. 7.

Amount of variance explained (percentage) by the first harmonic of sensible heat (H) for (a) OAFLUX, (b) HOAPS, and (c) BGI. Differences with respect to OAFLUX/BGI are shown for (d) ASRv2, (e) CFSR, (f) MERRA2, and (g) JRA-55.

Citation: Journal of Climate 32, 7; 10.1175/JCLI-D-18-0535.1

Comparing the OAFLUX to the reanalyses, it is demonstrated that over the eastern Pacific, the first harmonic is different among the reanalyses (Fig. 7), reflecting that climate mechanisms related to short-term variability are reproduced differently. More similarity is noted between ASRv2 and JRA-55, whereas CFSR and MERRA2 exhibit individual patterns. Higher seasonal-cycle variance over mostly eastern boundary currents in ASRv2, MERRA2, and JRA-55 indicates that climate mechanisms related to long-term fluctuations are more influential in characterizing the time variability in these reanalyses (Figs. 7d,f,g). The opposite applies for CFSR (Fig. 7e).

For the eastern North Atlantic, the seasonal cycle explains less variance in the reanalyses (Fig. 7). Lower variance delivered by the first harmonic in ASRv2, and very likely in MERRA2 has been found to be highly correlated to the NAO and surface winds (not shown). Indeed, the regression pattern between the NAO first principal component and the H anomaly field in ASRv2 matches closely the features depicted in Fig. 7d. The NAO has been primarily characterized by a white spectrum with high day-to-day variability associated with short-time-scale atmospheric stochastic processes (Feldstein 2000; Hurrell and Deser 2009). Therefore, higher-order harmonics are expected to dominate the air–sea interaction over the northeastern North Atlantic in ASRv2. However, the H response to the NAO and other weather patterns seems to be stronger in the other reanalyses, which may result in a weaker seasonal cycle/stronger daily variability as shown in Figs. 7e–g.

Over land, the seasonal variances of heat fluxes are compared to BGI. To calculate H and LE, BGI applies a static IGBP vegetation characteristic of the 1982–2008 period. According to Jung et al. (2011), this can lead to errors in the fluxes associated with the absence of intraseasonal and interannual variability of the land cover. BGI also uses monthly satellite-based Fraction of Absorbed Photosynthetically Active Radiation (fAPAR) that is reflective of the monthly changes in vegetation. In some cases however, the satellite pixel is not representative of the surface conditions encountered in the original flux tower dataset.

In ASRv2, vegetation is updated weekly, resulting in a better representation of surface processes such as land albedo and leaf area index (LAI), and subsequently more accurate heat flux estimates. Large differences are noted between the reanalyses and BGI as well as among reanalyses (Figs. 7d–g). The largest differences occur over Asia and northern Canada in ASRv2, CFSR, and to some extent MERRA2. This is reasonable over the central part of North America, where primarily grassland and cropland are dominant (see vegetation type distributions in Fig. 1). Reduced variance is also noted in ASRv2 in the heavily cultivated northeast and eastern seaboard regions of the Chinese Great Plains (40°–60°N, 120°–140°E; Lu et al. 2010). This ASRv2 feature differs from the pattern delivered by the other reanalyses. The daily fluctuations of H over North America are shown by the meteorological towers located at the U.S.–Canada border as discussed for Fig. 2. It is clear that the monthly variability of H can be as large as the seasonal cycle variance.

Grasslands and croplands experience substantial changes of LAI during their growth stages, and therefore, modifications in the absorption of shortwave radiation are expected (Monteith 1965). Changes in the shortwave distribution further alter the net radiative balance and the near surface microclimate (Wilson et al. 2002). It is important to note that the majority of crops in these areas have a life cycle that ranges between 90 and 120 days (e.g., maize, wheat, and soybeans), and crop rotation practices result in different rates of H and LE on the intraseasonal time scale.

This feature highlights the importance of proper specification of the lower boundary conditions (such as weekly vegetation in ASRv2) as well the importance of reasonable spatial resolution in the land surface scheme. Processes associated with modifications of H over grassland, cropland, and sparse vegetation are more tightly connected to temperature and vapor pressure deficit compared to dense vegetation, which can rely on a deeper root system (Wilson et al. 2002). In addition, surface resistance to water vapor transport and heat transfer coefficients should be properly parameterized in order to take into account incoming energy and energy loss by changes in the LAI.

Figure 8 shows the first harmonic amplitude of H. With the exception of OAFLUX, larger seasonal contrasts dominate over regions where the variance is greater than 70% (Figs. 7a–c), such as over open water. Overall, both oceanic basins are characterized by an east–west dipole in terms of the amplitudes of the seasonal cycle: higher in the west and lower in the east. The seasonal variability related to the Labrador Sea current, in particular for OAFLUX, is clearly depicted (Fig. 8a). However, OAFLUX does not show this seasonal pattern insofar as H variance is concerned (Fig. 7a). HOAPS, primarily based on satellite retrievals and atmospheric reanalysis data, shows broader agreement with fluxes over the Gulf Stream region (Shaman et al. 2010). Over land, higher amplitudes are noted northward of 50°N, particularly over the northwestern part of North America (Fig. 8c). This part of North America is covered by a mix of evergreen needle-leaf forest, woody savannas, and open shrublands, each with varying sensitivities to summertime conditions that modify H and LE. These heterogeneous surfaces also play a role in the advection of heat and water vapor across the area (Arck and Scherer 2002; Harder et al. 2017). However, during winter, homogeneous conditions related to snow cover reduces the zonal advection and the biosphere–atmosphere interaction resulting in large seasonal differences.

Fig. 8.
Fig. 8.

As in Fig. 7, but for the first harmonic amplitude of sensible heat (H; W m−2).

Citation: Journal of Climate 32, 7; 10.1175/JCLI-D-18-0535.1

Compared to the blended data (Figs. 8d–g), ASRv2 shows amplitudes that closely match over most of the domain, highlighted by small differences over oceanic areas. ASRv2 also represents the eddy activity in the Labrador Current (Fig. 8d; Wyrtki et al. 1976; Marzocchi et al. 2015). This indicates that ASRv2 surface scheme properly assimilates small-scale processes at its lower boundary over open water. This eddy-driven air–sea interaction that is embedded in the seasonal cycle contributes to the increase in the variance and the amplitude as well. Departure from OAFLUX is larger in the other reanalyses (Figs. 7e–g).

Substantial differences in H occur over land. With the exception of ASRv2, all reanalyses exhibit smaller amplitude north of 60°N compared to BGI, suggesting a weaker seasonal cycle (Fig. 8). These differences are greater over northern Eurasia and North America indicating that the snow–vegetation feedback and the vegetation response to seasonality are similarly incorporated in CFSR, MERRA2, and JRA-55. Over regions dominated by cropland in North America and East Asia, the ASRv2 amplitude differences with respect to BGI show coherence, with the lowest variance in H (Figs. 7d,g).

For the seasonal cycle of LE, we discover that the amount of variance explained by OAFLUX and HOAPS first harmonic (Fig. 9) is very similar to H (Fig. 7). However, there are differences between the reanalyses and BGI over land. The seasonal cycle is enhanced over most of North America and northeastern Asia in CFSR and JRA-55. This is not the case for ASRv2. Due to its higher sensitivity for vegetation changes related to precipitation and temperature and therefore evapotranspiration, the overall differences between the ASRv2 and the blended data are minimal.

Fig. 9.
Fig. 9.

As in Fig. 7, but for latent heat (LE).

Citation: Journal of Climate 32, 7; 10.1175/JCLI-D-18-0535.1

Figure 10 shows the amplitude of the seasonal cycle for LE. Over the ocean, the large LE response over the western boundary currents is clearly identified. This is the result of stronger seasonality of the atmospheric flow, and over North America especially, to cold and dry air outbreaks (Figs. 10a,b). These outbreaks induce changes in the vertical gradient of temperature and humidity (Shaman et al. 2010).

Fig. 10.
Fig. 10.

As in Fig. 8, but for latent heat (LE).

Citation: Journal of Climate 32, 7; 10.1175/JCLI-D-18-0535.1

For reanalyses compared to the OAFLUX amplitudes, changes are noted. The anomalous pattern in ASRv2 departs from the other reanalyses across the North Pacific (Fig. 10d), where lower amplitudes in ASRv2 are related to fluctuations of the LE in response to daily variability of weather patterns. Better agreement among reanalyses is found in midlatitudes over the North Atlantic open water, where the atmospheric flow exhibits a much weaker intraseasonal variability compared to the North Pacific (Yashayaev and Zveryaev 2001; Lembo et al. 2017).

Over the continents, zonal and meridional patterns are depicted (Fig. 10c). Differences across North America show the influence of much denser vegetation in the eastern portion of the continent with savanna, grassland, and open shrubland in the west (Fig. 10c). This is reasonable considering that small fluctuations of LE are expected in desert and semiarid climates as well as in regions with sparse vegetation, such as the North American west coast and southern Asia.

Differences are apparent among the reanalyses in North America and Asia over midlatitudes, but similarities are found for ASRv2 and MERRA2 over Europe and ASRv2 and JRA-55 in western North America (Figs. 10d,f,g). In ASRv2, lower amplitudes in the Canadian Arctic and Siberia are the result of sparse vegetation and its limited evapotranspiration capacity. The CFSR LE amplitude pattern is dominated by regional features, whereas the other reanalyses are characterized by large-scale processes and enhanced seasonality.

To evaluate the partitioning of H and LE related to specific land cover, Fig. 11 shows the spatially averaged monthly mean values of the evaporative fraction (EF), over 11 dominant vegetation types. The vegetation (land cover) distribution between 40° and 75°N is from the MODIS global 1-km maximum vegetation fraction (Broxton et al. 2014). The dominant vegetation types in the NH extratropics are evergreen (needleleaf and broadleaf), deciduous (needleleaf and broadleaf), open shrubland, grassland/savannas, and croplands.

Fig. 11.
Fig. 11.

Spatially averaged monthly evaporative fraction (EF) related to individual dominant vegetation patterns (indicated above each panel) for BGI (black), ASRv2 (red), CFSR (green), MERRA2 (light blue), and JRA-55 (purple).

Citation: Journal of Climate 32, 7; 10.1175/JCLI-D-18-0535.1

Based on Fig. 11, with the exception of ASRv2, the reanalyses are quite different compared to BGI in terms of the EF. Discrepancies are greater during the winter/early spring when LE is very small over snow-covered evergreen needleleaf, deciduous broadleaf, and open shrubland (Jiménez 2011). The maximum value of both H and LE occur in July/August revealing that in the summer, both H and LE have similar magnitudes. The reanalyses strongly disagree, particularly over deciduous needleleaf (Fig. 11b), open shrubland (Fig. 11f), woody savannas (Fig. 11g) and permanent wetlands (Fig. 11i). This indicates that the land surface scheme, and therefore, H and LE are tightly linked to the vegetation properties forced by atmospheric conditions and vice versa.

These changes are present not only with respect to the EF magnitude but also in the monthly variability. Some agreement among the reanalyses and BGI is found for deciduous broadleaf (Fig. 11c), grassland (Fig. 11h), and cropland (Fig. 11j) vegetation types. It is reasonable to argue that the partitioning of the heat budget between H and LE over Arctic vegetation, such as those northward of 60°N, is problematic under snow cover conditions where the reanalyses differ dramatically from BGI.

Indeed, reanalyses reproduce the EF seasonal cycle but differ in magnitude. Results discussed above are in line with those of Zhou and Wang (2016). Comparing H and LE from flux towers at AmeriFlux sites and the ERA-I, it was demonstrated that ERA-I has little capability to reproduce the relationship between these fluxes as represented by the EF, and that most differences between the reanalyses and observations are found over deciduous broadleaf forest, grassland, and cropland vegetation types.

It should be stressed that as far as BGI monthly changes are concerned, ASRv2 exhibits the closest match. Larger differences are found for deciduous needleleaf (Fig. 11b) and closed shrubland (Fig. 11e) throughout the year. For most vegetation types, BGI does not show the semiannual variability over deciduous needleleaf, open shrubland, woody savannas, or permanent wetlands. This characteristic is enhanced in the reanalyses but damped in particular in ASRv2. In this sense, these results complement the previous evaluation by Zhou and Wang (2016) which was confined to North America.

4. Concluding remarks

Based on an evaluation of modern reanalyses, blended data, and satellite estimates of H and LE fluxes, there is some agreement among the datasets. Dominant hemispheric patterns of H and LE are similar with regard to the locations of maxima and minima and particularly over oceanic regions. However, large differences in the magnitudes of H and LE are clearly demonstrated among the reanalyses as well as differences between the two oceanic blended datasets (HOAPS and OAFLUX). Annually averaged and among all reanalyses, ASRv2 and JRA-55 deliver the largest differences compared to OAFLUX, but ASRv2 shows the closest match with respect to HOAPS and BGI. The strongest agreement occurs for H over land.

Insofar as the seasonal cycle is concerned, there are substantial differences in the reanalyses in the first harmonic variance and amplitude. These datasets are characterized by distinct temporal variability differences. Over land, the largest disagreements are based on the effect of vegetation and land cover on H and LE. Indeed, evaluation of the EF shows that with the exception of the ASRv2, the reanalyses are characterized by a similar H and LE partitioning from April through September. Hence, there is smaller month-to-month variability compared to the BGI dataset. On the other hand, ASRv2 reasonably reproduces the seasonal and intraseasonal variability as depicted by the EF in BGI. This is likely the result of the following factors: 1) weekly modification of the vegetation cover based on MODIS data; 2) fine horizontal resolution capable of accurately capturing LAI and the related snow–albedo feedbacks which impact H and LE; and 3) accurate representation of tropospheric winds due to realistic topography (Bromwich et al. 2016; Moore et al. 2016; Bromwich et al. 2018).

In an overall perspective, it may be argued that the reanalyses do a reasonable job of estimating the H and LE when seasonal, large-scale dynamics dominate the atmospheric/oceanic flow. For conditions where small-scale processes associated with cultivated crops or day-to-day vegetation feedbacks are involved, the global reanalyses are limited whereas ASRv2 performs well. This gives confidence in using ASRv2 to further investigate heat flux extremes and interactions over sea ice across the Arctic (e.g., Gulev and Baylev 2012).

Excessive local warming, as documented by recent datasets as well as simulated by global, regional, and weather forecast models (particularly over urban or rural areas), largely depends on the degree of the evaporative cooling (e.g., Li et al. 2015). Therefore, proper representation of the heat partitioning is crucial for simulating climate variables in agreement with local observations. This urges improvements of micro- and mesoscale biosphere–atmosphere surface processes in the reanalyses. For the time being, it is necessary at minimum to deal with spatial resolution; that is to drive down the spatial scale to the level of properly resolving vegetation and land cover changes and their climatic feedbacks, such as temporal changes of land/vegetation albedo and evapotranspiration. Thus, these results support the need for improvement in the representation of the surface fluxes including physical processes, assimilation of observations, and the coupled interface.

Acknowledgments

This work was supported by the National Aeronautics and Space Administration Grant NNX12AI29G, Brazilian National Research Council projects 232718/2014-8, 407681/2013-2 and 306181/2016-9, and the Office of Naval Research Grant N00014-18-1-2361. The authors thank the Ohio Supercomputer Center (http://www.osc.edu) for their use of the Glenn, Oakley, and Ruby clusters in order to conduct the ASRv2. This is Contribution 1576 of the Byrd Polar and Climate Research Center.

REFERENCES

  • Andersson, A., K. Fennig, C. Klepp, S. Bakan, H. Graßl, and J. Schulz, 2010: The Hamburg Ocean Atmosphere Parameters and Fluxes from Satellite Data—HOAPS-3. Earth Syst. Sci. Data, 2, 215234, https://doi.org/10.5194/essd-2-215-2010.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Arck, M., and D. Scherer, 2002: Problems in the determination of sensible heat flux over snow. Geogr. Ann., 84A, 157169, https://doi.org/10.1111/j.0435-3676.2002.00170.x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Auad, G., A. J. Miller, J. O. Roads, and D. Cayan, 2001: Pacific Ocean wind stress and surface heat flux anomalies from NCEP reanalysis and observations: Cross-statistics and ocean model responses. J. Geophys. Res., 106, 22 24922 265, https://doi.org/10.1029/2000JC000264.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bonan, G. B., F. S. Chapin, and S. L. Thompson, 1995: Boreal forest and tundra ecosystems as components of the climate system. Climatic Change, 29, 145167, https://doi.org/10.1007/BF01094014.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bonan, G. B., S. Levis, S. Sitch, M. Vertenstein, and K. W. Oleson, 2003: A dynamic global vegetation model for use with climate models: Concepts and description of simulated vegetation dynamics. Global Change Biol., 9, 15431566, https://doi.org/10.1046/j.1365-2486.2003.00681.x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bondeau, A., and et al. , 2007: Modelling the role of agriculture for the 20th century global terrestrial carbon balance. Global Change Biol., 13, 679706, https://doi.org/10.1111/j.1365-2486.2006.01305.x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bromwich, D. H., A. B. Wilson, L.-S. Bai, G. W. K. Moore, and P. Bauer, 2016: A comparison of the regional Arctic System Reanalysis and the global ERA-Interim Reanalysis for the Arctic. Quart. J. Roy. Meteor. Soc., 142, 644658, https://doi.org/10.1002/qj.2527.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bromwich, D. H., and et al. , 2018: The Arctic System Reanalysis, version 2. Bull. Amer. Meteor. Soc., 99, 805828, https://doi.org/10.1175/BAMS-D-16-0215.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brown, L., and C. R. Duguay, 2010: The response and role of ice cover in lake-climate interactions. Prog. Phys. Geogr. Earth Environ., 34, 671704, https://doi.org/10.1177/0309133310375653.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Broxton, P. D., X. Zeng, W. Scheftic, and P. A. Troch, 2014: A MODIS-based global 1-km maximum green vegetation fraction dataset. J. Appl. Meteor. Climatol., 53, 19962004, https://doi.org/10.1175/JAMC-D-13-0356.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Budyko, M. I., 1969: The effect of solar radiation variations on the climate of the Earth. Tellus, 21, 611619, https://doi.org/10.3402/tellusa.v21i5.10109.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cayan, D. R., 1992a: Latent and sensible heat flux anomalies over the northern oceans: Driving the sea surface temperature. J. Phys. Oceanogr., 22, 859881, https://doi.org/10.1175/1520-0485(1992)022<0859:LASHFA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cayan, D. R., 1992b: Latent and sensible heat flux anomalies over the northern oceans: The connection to monthly atmospheric circulation. J. Climate, 5, 354369, https://doi.org/10.1175/1520-0442(1992)005<0354:LASHFA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Checkley, D. M., and J. A. Barth, 2009: Patterns and processes in the California Current system. Prog. Oceanogr., 83, 4964, https://doi.org/10.1016/j.pocean.2009.07.028.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Decker, M., M. A. Brunke, Z. Wang, K. Sakaguchi, X. Zeng, and M. G. Bosilovich, 2012: Evaluation of the reanalysis products from GSFC, NCEP, and ECMWF using flux tower observations. J. Cilmate, 25, 19161944, https://doi.org/10.1175/JCLI-D-11-00004.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dee, D. P., and et al. , 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553597, https://doi.org/10.1002/qj.828.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dong, C., J. C. McWilliams, Y. Liu, and D. Chen, 2014: Global heat and salt transports by eddy movement. Nat. Commun., 5, 3294, https://doi.org/10.1038/ncomms4294.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Feldstein, S. B., 2000: Teleconnections and ENSO: The timescale, power spectra, and climate noise properties. J. Climate, 13, 44304440, https://doi.org/10.1175/1520-0442(2000)013<4430:TTPSAC>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fenning, K., A. Andersson, S. Bakan, C. Klepp, and M. Schroeder, 2012: Hamburg Ocean Atmosphere Parameters and fluxes from Satellite data - HOAPS 3.2 - monthly means/6-hourly composites. Satellite Application Facility on Climate Monitoring, accessed 22 February 2019, https://doi.org/10.5676/EUM_SAF_CM/HOAPS/V001.

    • Crossref
    • Export Citation
  • Friedl, M. A., D. Sulla-Menashe, B. Tan, A. Schneider, N. Ramankutty, A. Sibley, and X. Huang, 2010: MODIS collection 5 global land cover: Algorithm refinements and characterization of new datasets. Remote Sens. Environ., 114, 168182, https://doi.org/10.1016/j.rse.2009.08.016.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ganeshan, M., and D. L. Wu, 2016: The open-ocean sensible heat flux and its significance for Arctic boundary layer mixing during early fall. Atmos. Chem. Phys., 16, 13 17313 184, https://doi.org/10.5194/acp-16-13173-2016.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Geiselman, J., A. R. DeGange, K. Oakley, D. Derksen, and M. Whalen, 2012: Changing Arctic ecosystems: Research to understand and project changes in marine and terrestrial ecosystems of the Arctic. USGS Fact Sheet 2011-3136, http://pubs.usgs.gov/fs/2011/3136/.

    • Crossref
    • Export Citation
  • Gelaro, R., and et al. , 2017: The Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2). J. Climate, 30, 54195454, https://doi.org/10.1175/JCLI-D-16-0758.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gulev, S. K., and K. P. Baylev, 2012: Probability distribution characteristics for surface air-sea turbulent heat fluxes over the global ocean. J. Climate, 25, 184206, https://doi.org/10.1175/2011JCLI4211.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gulev, S. K., M. Latif, N. Keenlyside, W. Park, and K. P. Koltermann, 2013: North Atlantic Ocean control on surface heat flux on multidecadal timescales. Nature, 499, 464467, https://doi.org/10.1038/nature12268.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Harder, P., J. W. Pomeroy, and W. Helgason, 2017: Local-scale advection of sensible and latent heat during snowmelt. Geophys. Res. Lett., 44, 97699777, https://doi.org/10.1002/2017GL074394.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hurrell, J. W., and C. Deser, 2009: North Atlantic climate variability: The role of the North Atlantic Oscillation. J. Mar. Syst., 78, 2841, https://doi.org/10.1016/j.jmarsys.2008.11.026.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • IPCC, 2013: Climate Change 2013: The Physical Science Basis. Cambridge University Press, 1535 pp., https://doi.org/10.1017/CBO9781107415324.

    • Crossref
    • Export Citation
  • Jiménez, C., 2011: Global intercomparison of 12 land surface heat flux estimates. J. Geophys. Res., 116, D02102, https://doi.org/10.1029/2010JD014545.

    • Search Google Scholar
    • Export Citation
  • Jin, X., and L. Yu, 2013: Assessing high-resolution analysis of surface heat fluxes in the Gulf Stream region. J. Geophys. Res. Oceans, 118, 53535375, https://doi.org/10.1002/jgrc.20386.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jung, M., M. Reichstein, and A. Bondeau, 2009: Towards global empirical upscaling of FLUXNET eddy covariance observations: Validation of a model tree ensemble approach using a biosphere model. Biogeosciences, 6, 20012013, https://doi.org/10.5194/bg-6-2001-2009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jung, M., and et al. , 2011: Global patterns of land-atmosphere fluxes of carbon dioxide, latent heat, and sensible heat derived from eddy covariance, satellite, and meteorological observations. J. Geophys. Res., 116, G00J07, https://doi.org/10.1029/2010JG001566.

    • Search Google Scholar
    • Export Citation
  • Kent, E. C., D. I. Berry, J. Prytherch, and J. B. Roberts, 2014: A comparison of global marine surface-specific humidity datasets from in situ observations and atmospheric reanalysis. Int. J. Climatol., 34, 355376, https://doi.org/10.1002/joc.3691.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • King, A. D., G. J. van Oldenborgh, D. J. Karoly, S. C. Lewis, and H. Cullen, 2015: Attribution of the record high central England temperature of 2014 to anthropogenic influences. Environ. Res. Lett., 10, 054002, https://doi.org/10.1088/1748-9326/10/5/054002.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kobayashi, S., and et al. , 2015: The JRA-55 Reanalysis: General specifications and basic characteristics. J. Meteor. Soc. Japan, 93, 548, https://doi.org/10.2151/jmsj.2015-001.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Legates, D. R., and G. J. McCabe, 1999: Evaluating the use of “goodness-of-fit” measures in hydrologic and hydroclimatic model validation. Water Resour. Res., 35, 233241, https://doi.org/10.1029/1998WR900018.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lembo, V., I. Bordi, and A. Speranza, 2017: Annual and semiannual cycles of midlatitude near-surface temperature and tropospheric baroclinicity: Reanalysis data and AOGCM simulations. Earth Syst. Dyn., 8, 295312, https://doi.org/10.5194/esd-8-295-2017.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Li, D., T. Sun, M. Liu, L. Yang, L. Wang, and Z. Gao, 2015: Contrasting responses of urban and rural surface energy budgets to heat waves explain synergies between urban heat islands and heat waves. Environ. Res. Lett., 10, 054009, https://doi.org/10.1088/1748-9326/10/5/054009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lindsay, R., M. Wensnahan, A. Schweiger, and J. Zhang, 2014: Evaluation of seven different atmospheric reanalysis products in the Arctic. J. Climate, 27, 25882606, https://doi.org/10.1175/JCLI-D-13-00014.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Loranty, M. M., L. T. Berner, S. J. Goetz, Y. Jin, and J. T. Randerson, 2014: Vegetation controls on northern high latitude snow-albedo feedback: Observations and CMIP5 model simulations. Global Change Biol., 20, 594606, https://doi.org/10.1111/gcb.12391.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lu, Y., E. Wand, X. Yang, and J. Wang, 2010: Contributions of climatic and crop varietal changes to crop production in the North China Plain, since 1980s. Global Change Biol., 16, 22872299, https://doi.org/10.1111/j.1365-2486.2009.02077.x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Marks, D., A. Winstral, G. Flerchinger, M. Reba, J. Pomeroy, T. Link, and K. Elder, 2008: Comparing simulated and measured sensible and latent heat fluxes over snow under a pine canopy to improve an energy balance snowmelt model. J. Hydrometeor., 9, 15061522, https://doi.org/10.1175/2008JHM874.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Marzocchi, A., J. J.-M. Hirschi, N. P. Holliday, S. A. Cunningham, A. T. Blaker, and A. C. Coward, 2015: The North Atlantic subpolar circulation in an eddy-resolving global ocean model. J. Mar. Syst., 142, 126143, https://doi.org/10.1016/j.jmarsys.2014.10.007.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Monteith, J. L., 1965: Radiation and crops. Exp. Agric., 1, 241251, https://doi.org/10.1017/S0014479700021529.

  • Moore, G. W. K., I. Renfrew, and R. S. Pickart, 2012: Spatial distribution of air-sea heat fluxes over the sub-polar North Atlantic Ocean. Geophys. Res. Lett., 39, L18806, https://doi.org/10.1029/2012GL053097.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Moore, G. W. K., K. Vage, R. S. Pickart, and I. Renfrew, 2015: Decreasing intensity of open-ocean convection in the Greenland and Iceland Seas. Nat. Climate Change, 5, 877882, https://doi.org/10.1038/nclimate2688.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Moore, G. W. K., D. H. Bromwich, A. B. Wilson, I. Renfrew, and L. Bai, 2016: Arctic System Reanalysis improvements in topographically-forced winds near Greenland. Quart. J. Roy. Meteor. Soc., 142, 20332045, https://doi.org/10.1002/qj.2798.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • NASA, 2018: FLUXNET. Oak Ridge National Laboratory Distributed Active Archive Center, http://fluxnet.ornl.gov.

  • NCAR/UCAR and PMG/BPCRC/OSU 2017, 2017: Arctic System Reanalysis version 2. Research Data Archive at the National Center for Atmospheric Research, Computational and Information Systems Laboratory, accessed 5 September 2018, https://doi.org/10.5065/D6X9291B.

    • Crossref
    • Export Citation
  • Overland, J. E., and M. Wang, 2010: Large-scale atmospheric circulation changes are associated with the recent loss of Arctic sea ice. Tellus, 62A, 19, https://doi.org/10.1111/j.1600-0870.2009.00421.x.

    • Search Google Scholar
    • Export Citation
  • Pielke, R. A., G. Marland, R. A. Betts, T. N. Chase, J. L. Eastman, J. O. Niles, D. S. Niyogi, and S. W. Running, 2002:The influence of land-use change and landscape dynamics on the climate system: Relevance to climate-change policy beyond the radiative effect of greenhouse gases. Philos. Trans. Roy. Soc. London, 360A, 17051719, https://doi.org/10.1098/rsta.2002.1027.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Saha, S., and et al. , 2010: The NCEP Climate Forecast System Reanalysis. Bull. Amer. Meteor. Soc., 91, 10151057, https://doi.org/10.1175/2010BAMS3001.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Santorelli, A., R. T. Pinker, A. Bentamy, K. B. Katsaros, W. M. Drennan, A. M. Mestas-Nuñez, and J. A. Carton, 2011: Differences between two estimates of air-sea turbulent heat fluxes over the Atlantic Ocean. J. Geophys. Res., 116, C09028, https://doi.org/10.1029/2010JC006927.

    • Search Google Scholar
    • Export Citation
  • Shaman, J., R. Samelson, and E. Skyllingstad, 2010: Air–sea fluxes over the Gulf Stream region: Atmospheric controls and trends. J. Climate, 23, 26512670, https://doi.org/10.1175/2010JCLI3269.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Shi, Q., and S. Liang, 2014: Surface-sensible and latent heat fluxes over the Tibetan Plateau from ground measurements, reanalysis, and satellite data. Atmos. Chem. Phys., 14, 56595677, https://doi.org/10.5194/acp-14-5659-2014.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Shukla, J., C. Nobre, and P. Sellers, 1990: Amazon deforestation and climate change. Science, 247, 13221325, https://doi.org/10.1126/science.247.4948.1322.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Simmons, A. J., K. M. Willett, P. D. Jones, P. W. Thorne, and D. P. Dee, 2010: Low-frequency variations in surface atmospheric humidity, temperature, and precipitation: Inferences from reanalyses and monthly gridded observational data sets. J. Geophys. Res., 115, D01110, https://doi.org/10.1029/2009JD012442.

    • Search Google Scholar
    • Export Citation
  • Taylor, K. E., R. J. Stouffer, and G. A. Meehl, 2012: An overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc., 93, 485498, https://doi.org/10.1175/BAMS-D-11-00094.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Taylor, S. V., D. R. Cayan, N. E. Graham, and K. P. Georgakakos, 2008: Northerly surface winds over the eastern North Pacific Ocean in spring and summer. J. Geophys. Res., 113, D02110, https://doi.org/10.1029/2006JD008053.

    • Search Google Scholar
    • Export Citation
  • Tomita, H., M. Kubota, M. F. Cronin, S. Iwasaki, M. Konda, and H. Ichikawa, 2010: An assessment of surface heat fluxes from J-OFURO2 at the KEO and JKEO sites. J. Geophys. Res., 115, C03018, https://doi.org/10.1029/2009JC005545.

    • Search Google Scholar
    • Export Citation
  • Uppala, S. M., and et al. , 2005: The ERA-40 Re-Analysis. Quart. J. Roy. Meteor. Soc., 131, 29613012, https://doi.org/10.1256/qj.04.176.

  • Wild, M., and et al. , 2015: The energy balance over land and oceans: An assessment based on direct observations and CMIP5climate models. Climate Dyn., 44, 33933429, https://doi.org/10.1007/s00382-014-2430-z.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Willmott, C. J., 1981: On the validation of models. Phys. Geogr., 2, 184194, https://doi.org/10.1080/02723646.1981.10642213.

  • Wilson, K. B., and et al. , 2002: Energy partitioning between latent and sensible heat flux during the warm season at FLUXNET sites. Water Resour. Res., 38, 1294, https://doi.org/10.1029/2001WR000989.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wunsch, C., 2002: Ocean observations and the climate forecast problem. Int. Geophys., 83, 233245, https://doi.org/10.1016/S0074-6142(02)80170-X.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wyrtki, K., L. Magaard, and J. Hager, 1976: Eddy energy in the oceans. J. Geophys. Res., 81, 26412646, https://doi.org/10.1029/JC081i015p02641.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yashayaev, I. M., and I. I. Zveryaev, 2001: Climate of the seasonal cycle in the North Pacific and the North Atlantic Oceans. Int. J. Climatol., 21, 401417, https://doi.org/10.1002/joc.585.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, L., and J. Xiangze, 2014: Confidence and sensitivity study of the OAFlux multisensor synthesis of the global ocean surface vector wind from 1987 onward. J. Geophys. Res. Oceans, 119, 68426862, https://doi.org/10.1002/2014JC010194.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, L., X. Jin, and R. A. Weller, 2008: Multidecade global flux datasets from the Objectively Analyzed Air-sea Fluxes (OAFLUX) project: Latent and sensible heat fluxes, ocean evaporation, and related surface meteorological variables. OAFlux Project Tech. Rep. OA-2008-01, Woods Hole Oceanographic Institution, 64 pp., http://oaflux.whoi.edu/pdfs/OAFlux_TechReport_3rd_release.pdf.

  • Yu, L., Z. Zhang, S. Zhong, M. Zhou, Z. Gao, H. Wu, and B. Sun, 2011: An inter-comparison of six latent and sensible heat flux products over the Southern Ocean. Polar Res., 30, 10167, https://10.3402/polar.v30i0.10167.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhou, C., and K. Wang, 2016: Evaluation of surface fluxes in ERA-Interim using flux tower data. J. Climate, 29, 15731582, https://doi.org/10.1175/JCLI-D-15-0523.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
Save