• Arino, O., P. Bicheron, F. Achard, J. Latham, R. Witt, and J. L. Weber, 2008: GlobCover—The most detailed portrait of Earth. ESA Bulletin, No. 136, European Space Agency, Paris, France, 21–34, http://www.esa.int/esapub/bulletin/bulletin136/bul136d_arino.pdf.

  • Arino, O., V. Kalogirou, J. R. Perez, S. Bontemps, P. Defourny, and E. V. Bogaert, 2011: GlobCover 2009: Products description and validation report. European Space Agency, 53 pp., http://dup.esrin.esa.int/files/p68/GLOBCOVER2009_Validation_Report_2.2.pdf.

  • Armstrong, R. L., and E. Brun, Eds., 2008: Snow and Climate: Physical Processes, Surface Energy Exchange and Modeling. Cambridge University Press, 222 pp.

  • Ban, Y., P. Gong, and C. Giri, 2015: Global land cover mapping using Earth observation satellite data: Recent progresses and challenges. ISPRS J. Photogramm. Remote Sens., 103, 16, https://doi.org/10.1016/j.isprsjprs.2015.01.001.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bartholomé, E., and A. S. Belward, 2005: GLC2000: A new approach to global land cover mapping from Earth observation data. Int. J. Remote Sens., 26, 19591977, https://doi.org/10.1080/01431160412331291297.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Boike, J., K. Roth, and O. Ippisch, 2003: Seasonal snow cover on frozen ground: Energy balance calculations of a permafrost site near Ny-Alesund, Spitsbergen. J. Geophys. Res., 108, 8163, https://doi.org/10.1029/2001JD000939.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bontemps, S., and et al. , 2015: Multi-year global land cover mapping at 300 M and characterization for climate modelling: Achievements of the land cover component of the ESA climate change initiative. ISPRS Arch., 40-7W3, 323–328, https://doi.org/10.5194/isprsarchives-XL-7-W3-323-2015.

    • Crossref
    • Export Citation
  • Brasnett, B., 1999: A global analysis of snow depth for numerical weather prediction. J. Appl. Meteor., 38, 726740, https://doi.org/10.1175/1520-0450(1999)038<0726:AGAOSD>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Burt, M. A., D. A. Randall, and M. D. Branson, 2016: Dark warming. J. Climate, 29, 705719, https://doi.org/10.1175/JCLI-D-15-0147.1.

  • Carroll, T., D. Cline, G. Fall, A. Nilsson, L. Li, and A. Rost, 2001: NOHRSC operations and the simulation of snow cover properties for the coterminous U.S. Proc. 69th Western Snow Conf., Sun Valley, ID, Western Snow Conference, https://www.nohrsc.noaa.gov/technology/pdf/wsc2001.pdf.

  • CDED, 2007: Canadian Digital Elevation Data, Level 1: Product specifications, edition 3.0, 2007-06-01. Natural Resources Canada, Canada Centre for Mapping and Earth Observation, accessed 3 October 2016, http://ftp.geogratis.gc.ca/pub/nrcan_rncan/archive/elevation/geobase_cded_dnec/.

  • CERES Science Team, 2017: EBAF-Surface Ed2.8 data. NASA Atmospheric Science Data Center, accessed 12 January 2017, doi:10.5067/Terra+Aqua/CERES/EBAF-Surface_L3B.002.8.

    • Crossref
    • Export Citation
  • Channan, S., K. Collins, and W. R. Emanuel, 2014: Global mosaics of the standard MODIS land cover type data. University of Maryland, College Park, and the Pacific Northwest National Laboratory, http://glcf.umd.edu/data/lc/.

  • Che, T., L. Dai, X. Zheng, X. Li, and K. Zhao, 2016: Estimation of snow depth from passive microwave brightness temperature data in forest regions of northeast China. Remote Sens. Environ., 183, 334349, https://doi.org/10.1016/j.rse.2016.06.005.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ciais, P., and et al. , 2013: Carbon and other biochemical cycles. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 465–570.

  • 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
  • Defourny, P., and et al. , 2014: ESA Climate Change Initiative - Land Cover project. Product User Guide, version 2, 87 pp., https://www.esa-landcover-cci.org/?q=webfm_send/84.

  • Derksen, C., R. Brown, L. Mudryk, and K. Luojus, 2015: Terrestrial snow cover. Arctic Report Card, update for 2015, accessed 6 September 2016, http://www.arctic.noaa.gov/Report-Card/Report-Card-2015/ArtMID/5037/ArticleID/216/Terrestrial-Snow-Cover.

  • Estilow, T. W., A. H. Young, and D. A. Robinson, 2015: A long-term Northern Hemisphere snow cover extent data record for climate studies and monitoring. Earth Syst. Sci. Data, 7, 137142, https://doi.org/10.5194/essd-7-137-2015.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fontana, F. M. A., A. P. Trishchenko, Y. Luo, K. V. Khlopenkov, S. U. Nussbaumer, and S. Wunderle, 2010: Perennial snow and ice variations (2000–2008) in the Arctic circumpolar land area from satellite observations. J. Geophys. Res., 115, F04020, doi:10.1029/2010JF001664.

    • Search Google Scholar
    • Export Citation
  • Frei, A., M. Tedesco, S. Lee, J. Foster, D. K. Hall, R. Kelly, and D. A. Robinson, 2012: A review of global satellite-derived snow products. Adv. Space Res., 50, 10071029, https://doi.org/10.1016/j.asr.2011.12.021.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Friedl, M. A., and et al. , 2002: Global land cover mapping from MODIS: Algorithms and early results. Remote Sens. Environ., 83, 287302, https://doi.org/10.1016/S0034-4257(02)00078-0.

    • Crossref
    • Search Google Scholar
    • 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, 2001–2012, Collection 5.1, IGBP Land Cover, Boston University, Boston, MA, http://glcf.umd.edu/data/lc/.

    • Crossref
    • Export Citation
  • GCOS, 2006: Systematic observation requirements for satellite-based products for climate: Supplemental details to the satellite-based component of the “Implementation Plan for the Global Observing System for Climate in Support of the UNFCC (GCOS-92).” Vol. GCOS-107, WMO/TD-1338, 103 pp., https://library.wmo.int/opac/doc_num.php?explnum_id=3813.

  • GCOS, 2011: Systematic observation requirements for satellite-based data products for climate: 2011 update, supplemental details to the satellite-based component of the “Implementation plan for the Global Observing System for Climate in support of the UNFCCC (2010 update).” GCOS-154, 138 pp., https://library.wmo.int/opac/doc_num.php?explnum_id=3710.

  • GCOS, 2016: 18th Session of the GCOS/WCRP Terrestrial Observation Panel for Climate (TOPC-18). GCOS-202, 38 pp., https://library.wmo.int/opac/doc_num.php?explnum_id=3215.

  • Gómez, C., J. C. White, and M. A. Wulder, 2016: Optical remotely sensed time series data for land cover classification: A review. ISPRS J. Photogramm. Remote Sens., 116, 5572, https://doi.org/10.1016/j.isprsjprs.2016.03.008.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gray, D. M., and D. H. Male, Eds., 1981: Handbook of Snow: Principles, Processes, Management and Use. Pergamon Press, 776 pp.

  • Gregory, J. M., and et al. , 2013: Twentieth-century global-mean sea level rise: Is the whole greater than the sum of the parts? J. Climate, 26, 44764499, https://doi.org/10.1175/JCLI-D-12-00319.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Grekousis, G., G. Mountrakis, and M. Kavouras, 2015: An overview of 21 global and 43 regional land-cover mapping products. Int. J. Remote Sens., 36, 53095335, https://doi.org/10.1080/01431161.2015.1093195.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hall, D. K., G. A. Riggs, V. V. Salomonson, N. E. DiGirolamo, and K. J. Bayr, 2002: MODIS snow-cover products. Remote Sens. Environ., 83, 181194, https://doi.org/10.1016/S0034-4257(02)00095-0.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jung, T., M. A. Kasper, T. Semmler, and S. Serrar, 2014: Arctic influence on subseasonal midlatitude prediction. Geophys. Res. Lett., 41, 36763680, https://doi.org/10.1002/2014GL059961.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jung, T., and et al. , 2016: Advancing polar prediction capabilities on daily to seasonal time scales. Bull. Amer. Meteor. Soc., 97, 16311647, https://doi.org/10.1175/BAMS-D-14-00246.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Khlopenkov, K. V., and A. P. Trishchenko, 2008: Implementation and evaluation of concurrent gradient search method for reprojection of MODIS level 1B imagery. IEEE Trans. Geosci. Remote Sens., 46, 20162027, https://doi.org/10.1109/TGRS.2008.916633.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Luo, Y., A. P. Trishchenko, and K. V. Khlopenkov, 2008: Developing clear-sky, cloud and cloud shadow mask for producing clear-sky composites at 250-meter spatial resolution for the seven MODIS land bands over Canada and North America. Remote Sens. Environ., 112, 41674185, https://doi.org/10.1016/j.rse.2008.06.010.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Luojus, K., and et al. , 2013: GlobSnow-2 Product User Guide Version 1.0. 24 pp., http://www.globsnow.info/se/GlobSnow2_SE_SWE_Product_User_Guide_v1_r1.pdf.

  • Malnes, E., and et al. , 2015: User requirements for the snow and land ice services—CryoLand. Cryosphere, 9, 11911202, https://doi.org/10.5194/tc-9-1191-2015.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mayaux, P., and et al. , 2006: Validation of the global land cover 2000 map. IEEE Trans. Geosci. Remote Sens., 44, 17281737, https://doi.org/10.1109/TGRS.2006.864370.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mernild, S. H., G. E. Liston, and C. A. Hiemstra, 2014: Northern Hemisphere glacier and ice cap surface mass balance and contribution to sea level rise. J. Climate, 27, 60516073, https://doi.org/10.1175/JCLI-D-13-00669.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mesinger, F., and et al. , 2006: North American Regional Reanalysis. Bull. Amer. Meteor. Soc., 87, 343360, https://doi.org/10.1175/BAMS-87-3-343.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Millan, R., J. Mouginot, and E. Rignot, 2017: Mass budget of the glaciers and ice caps of the Queen Elizabeth Islands, Canada, from 1991 to 2015. Environ. Res. Lett., 12, 024016, doi:10.1088/1748-9326/aa5b04.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mudryk, L. R., C. Derksen, P. J. Kushner, and R. Brown, 2015: Characterization of Northern Hemisphere snow water equivalent datasets, 1981–2010. J. Climate, 28, 80378051, https://doi.org/10.1175/JCLI-D-15-0229.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Overland, J., E. Hanna, I. Hanssen-Bauer, S.-J. Kim, J. E. Walsh, M. Wang, U. S. Bhatt, and R. L. Thoman, 2015a: Surface air temperature. Arctic Report Card, update for 2015, http://www.arctic.noaa.gov/Report-Card/Report-Card-2015/ArtMID/5037/ArticleID/210/Surface-Air-Temperature.

  • Overland, J., J. A. Francis, R. Hall, E. Hanna, S.-J. Kim, and T. Vihma, 2015b: The melting arctic and midlatitude weather patterns: Are they connected? J. Climate, 28, 79177932, https://doi.org/10.1175/JCLI-D-14-00822.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Radic, V., and R. Hock, 2010: Regional and global volumes of glaciers derived from statistical upscaling of glacier inventory data. J. Geophys. Res., 115, F01010, https://doi.org/10.1029/2009JF001373.

    • Search Google Scholar
    • Export Citation
  • Ramsay, B., 1998: The interactive multisensor snow and ice mapping system. Hydrol. Processes, 12, 15371546, https://doi.org/10.1002/(SICI)1099-1085(199808/09)12:10/11<1537::AID-HYP679>3.0.CO;2-A.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • RGI, 2015: Randolph Glacier Inventory—A dataset of global glacier outlines: version 5.0: Global Land Ice Measurements from Space (GLIMS), Boulder, CO, accessed 15 August 2016, https://doi.org/10.7265/N5-RGI-50.

    • Crossref
    • Export Citation
  • RGI, 2017: Randolph Glacier Inventory—A dataset of global glacier outlines, version 6.0: Global Land Ice Measurements from Space (GLIMS), Boulder, CO, accessed 15 August 2017, https://doi.org/10.7265/N5-RGI-60.

    • Crossref
    • Export Citation
  • Robinson, D. A., and et al. , 2012: NOAA Climate Date Record (CDR) of Northern Hemisphere (NH) Snow Cover Extent (SCE), Version 1.0. NOAA National Climatic Data Center, accessed 20 September 2016, doi:10.7289/V5N014G9.

    • Crossref
    • Export Citation
  • Romanov, P., 2017: Global Multisensor Automated Satellite-Based Snow and Ice Mapping System (GMASI) for cryosphere monitoring. Remote Sens. Environ., 196, 4255, https://doi.org/10.1016/j.rse.2017.04.023.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Romanov, P., G. Gutman, and I. Csiszar, 2000: Automated monitoring of snow cover over North America using multispectral satellite data. J. Appl. Meteor., 39, 18661880, https://doi.org/10.1175/1520-0450(2000)039<1866:AMOSCO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Shi, X. G., M. Wild, and D. P. Lettenmaier, 2010: Surface radiative fluxes over the pan-Arctic land region: Variability and trends. J. Geophys. Res., 115, D22104, https://doi.org/10.1029/2010JD014402.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Takala, M., K. Luojus, J. Pulliainen, C. Derksen, J. Lemmetyinen, J. P. Kärnä, J. Koskinen, and B. Bojkov, 2011: Estimating Northern Hemisphere snow water equivalent for climate research through assimilation of space-borne radiometer data and ground-based measurements. Remote Sens. Environ., 115, 35173529, https://doi.org/10.1016/j.rse.2011.08.014.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tedesco, M., R. Kelly, J. L. Foster, and A. T. Chang, 2004: AMSR-E/Aqua Monthly L3 Global Snow Water Equivalent EASE-Grids, version 10. NASA National Snow and Ice Data Center Distributed Active Archive Center, accessed 30 August 2016, https://search.earthdata.nasa.gov/search/granules?p=C179014696-NSIDC_ECS&q=AE_MoSno.

  • Trishchenko, A. P., 2016: Probability maps of the annual minimum snow and ice (MSI) presence over April 1 to September 20 period since 2000 derived from MODIS 250m imagery over Canada and neighboring regions. Accessed 15 June 2017, http://open.canada.ca/data/en/dataset/808b84a1-6356-4103-a8e9-db46d5c20fcf.

  • Trishchenko, A. P., Y. Luo, and K. V. Khlopenkov, 2006: A method for downscaling MODIS land channels to 250-m spatial resolution using adaptive regression and normalization. Remote Sensing for Environmental Monitoring, GIS Applications, and Geology VI, International Society for Optics and Photonics (SPIE Proceedings, Vol. 6366), M. Ehlers and U. Michel, Eds., 636607, doi:10.1117/12.689157.

    • Crossref
    • Export Citation
  • Trishchenko, A. P., Y. Luo, K. V. Khlopenkov, W. M. Park, and S. Wang, 2009: Arctic circumpolar mosaic at 250 m spatial resolution for IPY by fusion of MODIS/TERRA land bands B1–B7. Int. J. Remote Sens., 30, 16351641, https://doi.org/10.1080/01431160802348119.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Trishchenko, A. P., S. Leblanc, S. Wang, J. Li, C. Ungureanu, Y. Lu, K. V. Khlopenkov, and F. Fontana, 2016: Variations of annual minimum snow and ice extent over Canada and neighboring landmass derived from MODIS 250-m imagery for 2000–2014. Can. J. Remote Sens., 42, 214242, https://doi.org/10.1080/07038992.2016.1166043.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wessel, P., and W. H. F. Smith, 1996: A global self-consistent, hierarchical, high-resolution shoreline database. J. Geophys. Res., 101, 87418743, https://doi.org/10.1029/96JB00104.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Woo, M., and K. L. Young, 2014: Disappearing semi-permanent snow in the High Arctic and its consequences. J. Glaciol., 60, 192200, https://doi.org/10.3189/2014JoG13J150.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Young, K. L., and A. G. Lewkowicz, 1990: Surface energy balance of a perennial snowbank, Melville Island, Northwest Territories, Canada. Arct. Alp. Res., 22, 290301, https://doi.org/10.2307/1551592.

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

    Example of snow and ice map for the Canadian Arctic region. (a) MODIS land cover product (MCD12Q1) for 2012 (Friedl et al. 2010). (b) RGI 5.0 data, excluding Greenland (RGI 2015). (c) CCRS MSI false color map for 2012. Turquoise and blue colors correspond to snow/ice type. Very significant differences in land snow/ice cover between the MODIS product in (a) and results in (b) and (c) can be seen.

  • View in gallery

    Surface elevation over the Canadian Arctic region: (a) map and (b) statistical distributions.

  • View in gallery

    The altitude distribution of land ice over the Canadian Arctic from RGI 5.0 dataset.

  • View in gallery

    Time series of the minimum snow and ice (MSI) extent over the Canadian Arctic region derived from CCRS MODIS processing. Results for permanent snow/ice land cover type are also plotted for several land cover datasets.

  • View in gallery

    The MSI and RGI5.0 spatial extent distribution with altitude (at 100-m steps) for each year since 2000: (a) spatial extent and (b) difference MSI − RGI5.0.

  • View in gallery

    Snow/ice probability maps for April–September 2012 and 2013 derived from (left) the NARR dataset and (right) the CCRS MODIS product.

  • View in gallery

    Comparison of warm season MSI time series from MODIS against MSI derived from the NARR snow cover variable SNOWC.

  • View in gallery

    (a) The spatial distribution of the monthly mean temperature (2 m) between May and September for years with minimum (2012) and maximum (2013) MSI spatial extents in the Canadian Arctic from the NARR and ERA-Interim datasets.

  • View in gallery

    (b) As in Fig. 8a, but for the spatial distribution of the duration of time interval T > 0°C.

  • View in gallery

    Time series of regional mean characteristics for the altitudes <1400 m over the Canadian Arctic region from the ERA-Interim and NARR datasets over May–September: (a) air temperature at 2-m height and (b) duration of the melt season L. (c) The time series of the MSI extent are also shown for reference. The orientation of the y axes in (a) and (b) is inverted to show the synchrony with the MSI extent.

  • View in gallery

    Region average surface flux components for the altitudes <1400 m over the Canadian Arctic region from ERA-Interim and NARR datasets over the May–September period.

  • View in gallery

    Warm season (May–September) average maps of the total net surface fluxes from the NARR and ERA-Interim reanalyses.

  • View in gallery

    Region average surface (a) total and (b) radiative net fluxes for the altitudes <1400 m over the Canadian Arctic region from the ERA-Interim, NARR, and CERES datasets over the May–September period. (c) The time series of the MSI extent are also shown for the reference.

  • View in gallery

    Areal average monthly mean time series of SWE for the altitudes <1400 m over the Canadian Arctic archipelago land pixels from ERA-Interim, NARR, and GlobSnow.

  • View in gallery

    Monthly mean values of snowfall amounts (SWE) for 2000–16 from the NARR and ERA-Interim datasets over the Canadian Arctic region. The error bars depict standard deviations. The mean snowfall for May–September period is 95% (77%) of the long-term annual mean for the NARR (ERA-Interim) dataset. Corresponding numbers for annual minimum (July) are 37% and 21%.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 93 93 17
PDF Downloads 66 66 12

Variations of Climate, Surface Energy Budget, and Minimum Snow/Ice Extent over Canadian Arctic Landmass for 2000–16

View More View Less
  • 1 Canada Centre for Remote Sensing, Natural Resources Canada, Ottawa, Canada
© Get Permissions
Open access

Abstract

Snow and ice over land are important hydrological resources and sensitive indicators of climate change. The Moderate Resolution Imaging Spectroradiometer (MODIS) dataset at 250-m spatial resolution generated at the Canada Centre for Remote Sensing (CCRS) is used to derive the annual minimum snow and ice (MSI) extent over the Canadian Arctic landmass over a 17-yr time span (2000–16). The smallest MSI extent (1.53 × 105 km2) was observed in 2012, the largest (2.09 × 105 km2) was observed in 2013; the average value was 1.70 × 105 km2. Several reanalyses and observational datasets are assessed to explain the derived MSI variations: the ERA-Interim reanalysis, North American Regional Reanalysis (NARR), Clouds and the Earth’s Radiant Energy System (CERES) radiative fluxes, and European Space Agency’s GlobSnow dataset. Comparison with the Randolph Glacier Inventory (RGI) showed two important facts: 1) the semipermanent snowpack in the Canadian Arctic that persists through the entire melting season is a significant component relative to the ice caps and glacier-covered areas (up to 36% or 5.58 × 104 km2), and 2) the MSI variations are related to variations in the local climate dynamics such as warm season average temperature, energy fluxes, and snow cover. The correlation coefficients (absolute values) can be as high as 0.77. The reanalysis-based MSI estimates agree with satellite MSI results (average bias of 2.2 × 103 km2 or 1.3% of the mean value).

Denotes content that is immediately available upon publication as open access.

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

Corresponding author: Alexander Trishchenko, alexander.trichtchenko@canada.ca

Abstract

Snow and ice over land are important hydrological resources and sensitive indicators of climate change. The Moderate Resolution Imaging Spectroradiometer (MODIS) dataset at 250-m spatial resolution generated at the Canada Centre for Remote Sensing (CCRS) is used to derive the annual minimum snow and ice (MSI) extent over the Canadian Arctic landmass over a 17-yr time span (2000–16). The smallest MSI extent (1.53 × 105 km2) was observed in 2012, the largest (2.09 × 105 km2) was observed in 2013; the average value was 1.70 × 105 km2. Several reanalyses and observational datasets are assessed to explain the derived MSI variations: the ERA-Interim reanalysis, North American Regional Reanalysis (NARR), Clouds and the Earth’s Radiant Energy System (CERES) radiative fluxes, and European Space Agency’s GlobSnow dataset. Comparison with the Randolph Glacier Inventory (RGI) showed two important facts: 1) the semipermanent snowpack in the Canadian Arctic that persists through the entire melting season is a significant component relative to the ice caps and glacier-covered areas (up to 36% or 5.58 × 104 km2), and 2) the MSI variations are related to variations in the local climate dynamics such as warm season average temperature, energy fluxes, and snow cover. The correlation coefficients (absolute values) can be as high as 0.77. The reanalysis-based MSI estimates agree with satellite MSI results (average bias of 2.2 × 103 km2 or 1.3% of the mean value).

Denotes content that is immediately available upon publication as open access.

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

Corresponding author: Alexander Trishchenko, alexander.trichtchenko@canada.ca

1. Introduction

The Arctic plays an important role in the entire climate system, undergoing significant changes with a warming rate of about twice the global average value (Overland et al. 2015a). The strong seasonal cycle in the Arctic is accompanied by important changes in the state of the cryosphere, when the melting snow and ice cause large variations in albedo and freshwater fluxes from land to ocean. The “dark warming” is associated with positive feedback between melting snow and ice (i.e., darker surface) and increasing amounts of surface absorbed solar radiation available for melt. The surface warming is enhanced by an increase in surface downwelling longwave radiation due to an increase in the water vapor and cloudiness in the warmer Arctic system (Burt et al. 2016). There are concerns about the growth of greenhouse gas emissions in a warming climate related to changes in the Arctic vegetation and permafrost thaw, especially about potentially large emissions of methane (Ciais et al. 2013). Using regional boundaries defined by Radic and Hock (2010), Mernild et al. (2014) concluded that the melt of Canadian Arctic glaciers and ice caps contributed 19% (8% and 11% for the northern and southern Arctic regions, respectively) of the total sea level rise from the world glaciers and ice caps for 1979–2009 period. This contribution has increased to 25% (14% and 11%) in the last decade (1999–2009). The uncertainties in estimates range from 10% to 50% and lead to a difficulty in closing the total sea level rise budget (Gregory et al. 2013). Accurate spatially complete information is required to improve the accuracy of mass balance estimates for climate monitoring and applications (Millan et al. 2017).

The Arctic influences weather in midlatitude regions over long time intervals, potentially extending to a subseasonal scale (Jung et al. 2014; Overland et al. 2015b). In recognition of its impact on the climate system and the need for advanced environmental prediction capabilities in the polar regions, the World Meteorological Organization (WMO) has recently established the Polar Prediction Project as a component of the World Weather Research Programme (see http://polarprediction.net). One of the key goals of the Polar Prediction Project is to improve understanding and representation of key processes in the polar cryosphere (Jung et al. 2016). To coordinate the international activities in the area of key cryospheric in situ and remote sensing observations, the WMO has recently established the Global Cryosphere Watch (GCW) program (http://globalcryospherewatch.org).

The key components of the Arctic polar cryosphere are snow and ice. However, there is significant inconsistency in land snow and ice mapping results. An example is provided for the Canadian Arctic archipelago in Fig. 1, which displays the MODIS land cover product MCD12Q1 for 2012 (Fig. 1a; Friedl et al. 2010), a land ice map derived from the Randolph Glacier Inventory (RGI) (Fig. 1b; RGI 2015, 2017), and the Canada Centre for Remote Sensing (CCRS) minimum snow/ice (MSI) extent (Fig. 1c; Trishchenko et al. 2016). The very significant difference in land snow/ice cover between Figs. 1a and 1b over the Queen Elizabeth Islands and Baffin Island is striking. Although the MODIS land cover represents the “snow and ice” class while RGI data correspond strictly to land ice, the very large difference clearly seen in these two maps raises a question: Is such a large difference real, and can it be attributed to the snow cover extent? The study by Trishchenko et al. (2016) found that the summer MSI extent over land for the Canadian Arctic archipelago in 2012 was the smallest since 2000, being very close to the RGI glaciers’ and ice caps’ total spatial extent. Such a large difference depicted in Fig. 1 creates the need to conduct in-depth analysis of snow/ice cover over the Canadian Arctic land area.

Fig. 1.
Fig. 1.

Example of snow and ice map for the Canadian Arctic region. (a) MODIS land cover product (MCD12Q1) for 2012 (Friedl et al. 2010). (b) RGI 5.0 data, excluding Greenland (RGI 2015). (c) CCRS MSI false color map for 2012. Turquoise and blue colors correspond to snow/ice type. Very significant differences in land snow/ice cover between the MODIS product in (a) and results in (b) and (c) can be seen.

Citation: Journal of Climate 31, 3; 10.1175/JCLI-D-17-0198.1

Our previous research showed that the key climatic parameter of terrestrial snow and ice in this region—the spatial extent during the summer melt season—suffers from very significant inconsistency among various sources (Trishchenko et al. 2016; Fontana et al. 2010). For example, the differences in permanent land snow/ice extents between several land cover datasets can reach 100% or more. Very large differences were identified between the Global Land Cover 2000 (GLC-2000) dataset compiled by the Joint Research Centre (JRC) of the European Commission (EC) (Bartholomé and Belward 2005; Mayaux et al. 2006) and other datasets produced in the framework of the ESA Climate Change Initiative (CCI) and ESA GlobCover projects, as well as the CCRS MSI products generated by Trishchenko et al. (2016). For example, the GLC-2000 data overestimated snow/ice extent by 194% (325 400 km2) for the Canadian Arctic relative to CCRS results.

The ultimate goal of this study is to shed some additional light on the consistency of snow and ice mapping over the Canadian Arctic landmass during the summer period when the minimum extent of snow and ice cover is reached (Woo and Young 2014; Young and Lewkowicz 1990). Specifically, this study will address the following research questions: 1) What is the year-to-year variability of the MSI over the Canadian Arctic land area? 2) How consistent are our new data derived from the CCRS MODIS 250-m datasets with other sources of information? 3) How consistent are the year-to-year variations in MSI with variations in climate conditions in the Canadian Arctic region?

The paper is structured as follows. The next section summarizes details of land snow/ice mapping requirements and available data. Section 3 describes in detail the distribution of land ice in the Canadian Arctic region derived from RGI and digital elevation data. Section 4 provides analysis of CCRS MSI dataset and comparison with other sources. Section 5 deals with analysis of climate variations over our study region since 2000 and their relation to MSI variations. Section 6 concludes the study.

2. Review of land snow/ice mapping requirements and available data

A good summary of user requirements for the snow and land ice services was produced by Malnes et al. (2015) in the framework of the CryoLand project carried out within the 7th Framework Programme of the European Commission. The requirements for climate monitoring of land ice and snow have been developed by the Global Climate Observing System (GCOS; GCOS 2006, 2011, 2016). The characteristics of snow and ice over land are included in the list of essential climate variables (ECVs) for the terrestrial domain. The up-to-date requirements for the snow and land ice observations are also available from the World Meteorological Organization Terrestrial Observational Panel for Climate (TOPC) through the Observing Systems Capability Analysis and Review (OSCAR) tool (https://www.wmo-sat.info/oscar/applicationareas/view/13). The requirements differ between snow and ice, which reflects the natural dynamics and basic physical processes associated with these components. They are briefly summarized below.

a. Glacier and ice cap mapping requirements

The information about spatial extent of glaciers and ice caps in the WMO TOPC OSCAR database is required as two-dimensional vector outlines delineating the glaciated areas with spatial resolutions of 30, 45, and 100 m for the goal, breakthrough, and threshold options, respectively. Uncertainty requirements are 5%, 7%, and 10% and temporal resolutions (frequency of update or observing cycle) are 1 yr, 1 yr, and 5 yr, respectively. The requirements for uncertainty in the surface elevation (topography) of glaciated area are 10, 22, and 100 cm respectively at the same spatial and temporal resolutions as spatial extent. The above numbers are close to those provided in GCOS-154 report for ECV related to glaciers and ice caps except that the goal (or target) spatial resolution is set to be from 15 to 30 m for spatial extent (GCOS 2011). The target requirements for topography in GCOS-154 are 30–100m for spatial resolution, 1 m for vertical resolution (vertical uncertainty), and 10 years for observational cycle.

b. Terrestrial snow mapping requirements

The requirements for snow cover spatial extent [or snow extent (SE)] in the WMO/TOPC OSCAR database are 0.1, 0.45, and 10 km for the goal, breakthrough, and threshold options, respectively. The temporal resolution is 24 h, 3 days, and 30 days with an uncertainty of 5%, 7%, and 10% for the goal, breakthrough, and threshold options, respectively. The target requirement in the GCOS-154 document for the snow cover spatial extent ECV is 1 km except in complex terrain conditions where it is 100 m. The temporal resolution is 24 h; accuracy is 5%. The requirements for the snow water equivalent (SWE) in the OSCAR system are described not under the TOPC domain, but under the Atmospheric Observational Panel for Climate (AOPC) domain. The requirements in the OSCAR database for SWE spatial resolution are 100, 200, and 500 km for the goal, breakthrough, and threshold options, respectively; temporal resolution is 24 h, 2 days, and 7 days, and uncertainty is 5, 6.5, and 10 mm, respectively. The SWE ECV is listed under terrestrial domain in the GCOS-154 document (GCOS 2011), which defines the SWE ECV requirements as 1 km, 24 h, and 10 mm for spatial resolution, temporal resolution, and accuracy, correspondingly.

c. Mapping requirements for snow/ice as land cover category

An important type of snow and ice information over the land comes from land cover products. The land cover classification usually includes a special class called “permanent snow and ice” or “snow/ice” (Arino et al. 2008, 2011; Bartholomé and Belward 2005; Bontemps et al. 2015; Friedl et al. 2002). This class is normally defined as “lands under snow/ice cover throughout the year” (Friedl et al. 2002). The WMO OSCAR TOPC spatial resolution requirements for land cover products are 250, 400, and 1000 m for the goal, breakthrough, and threshold options, respectively. The accuracy and observing cycle specifications are 2%, 4%, and 20% and 1, 2, and 5 yr, respectively. The GCOS requirements for land cover products depend on spatial resolution. The moderate resolution (250 m) land cover products are expected to be updated annually with an accuracy of 15%. The high-resolution (10–30 m) land cover products are expected to be updated every 5 years with an accuracy of 5%.

How consistent are these requirements with the available data?

d. Land ice data

Main developments of a global glacier and ice cap inventory are coordinated via the World Glacier Monitoring Service (WGMS; http://wgms.ch). The fifth version of the Randolph Glacier Inventory (RGI 5.0) was published in 2015 (RGI 2015) and updated in August 2017 as version RGI 6.0 (RGI 2017). The baseline period for the mapping of Canadian Arctic region in RGI corresponds to 1999–2003 with many polygons derived from satellite imagery and airborne photographs over the time interval between 1958 and 2010 (RGI 2017). The work is conducted in close collaboration with the U.S. National Snow and Ice Data Center (NSIDC). The majority of glacier outlines were obtained from Landsat (30–60-m resolution), ASTER (15 m), and SPOT (10–20 m) satellite imagery. Synthetic Aperture Radar (SAR) imagery, airborne photography, field observations, and various cartographic maps have been also used to derive land ice maps at different spatial scales (RGI 2015, 2017).

e. Snow data

The snow mapping activities are spread very wide among various countries, agencies, regions, and programs, depending on their objectives. The National Oceanic and Atmospheric Administration (NOAA) National Centers for Environmental Information (NCEI) [formerly the National Climatic Data Center (NCDC)] under the satellite Climate Data Record (CDR) Program has recently produced the long-term Northern Hemisphere snow cover extent data record for climate studies and monitoring (Robinson et al. 2012; Estilow et al. 2015). This record spans a long period of time starting in October 1966 (i.e., nearly 50 years) and has a temporal resolution of one week. The spatial resolution is quite coarse with the cell size varying from ~100 to ~200 km depending on location. Although undoubtedly useful for large-scale climatological analysis (Derksen et al. 2015), this dataset has somewhat limited value for detailed local applications and is not quite consistent with GCOS and WMO TOPC requirements for climate. Spatially and temporally detailed snow cover extent information is available from snow products derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) operated by NASA (Hall et al. 2002). The MODIS snow product is generated as daily and 8-day composite tiles of 10° × 10° at approximately 500-m spatial resolution. These datasets are aggregated into a climate-modeling grid (CMG) product at 0.05° spatial resolution with the same temporal resolution.

Many, if not all, national weather services, are producing some type of operational and/or climatic snow and ice products. The notable products to mention are the recent SE and SWE products derived in the framework of the GlobSnow project established by the European Space Agency (ESA) through the Data User Element (DUE) program (Luojus et al. 2013). The SE products are available in a latitude/longitude coordinate system with a spatial resolution of 0.01° × 0.01°. The product covers part of the Northern Hemisphere between 25° and 84°N and has daily, weekly, and monthly aggregation levels. The GlobSnow fractional snow cover (FSC) data go back to 1995. The GlobSnow SWE product is derived for the Northern Hemisphere on Equal-Area Scalable Earth Grid (EASE-Grid) at a nominal resolution of 25 km × 25 km and temporal aggregation similar to the SE product (Takala et al. 2011). The product spatial coverage is limited between latitudes 35° and 85°N. The SWE product is not generated in mountainous regions or for wet snow conditions. The SWE product is also produced by the NSIDC (Tedesco et al. 2004). Despite a certain progress in satellite technology, the remote sensing retrievals of SWE still have a substantial uncertainty, especially in complex terrain conditions and forested areas and for wet snow conditions (Tedesco et al. 2004; Frei et al. 2012; Mudryk et al. 2015; Che et al. 2016).

There are a number of operational snow products produced by NOAA. The National Environmental Satellite, Data and Information Service (NESDIS) of NOAA produces daily snow products using the Interactive Multisensor Snow and Ice Mapping System (IMS) (Ramsay 1998) and the Global Multisensor Automated Satellite-Based Snow and Ice Mapping System (GMASI) (Romanov et al. 2000; Romanov 2017). The IMS snow maps are generated at 24-km spatial resolution, while the automated system produces 4-km products. A variety of snow products (SWE, snow depth, snow temperature, density, and so on) are produced by the National Operational Hydrologic Remote Sensing Center (NOHRSC) over U.S. territory at 1-km spatial resolution and hourly time steps (Carroll et al. 2001). The operational global snow product (snow depth) is also generated by the Canadian Meteorological Centre (CMC) on a 1/3° latitude–longitude grid every 6 h (Brasnett 1999; http://weather.gc.ca/analysis/index_e.html).

f. Snow and ice as land cover data

An overview of recent developments regarding global and regional land cover mapping products and baseline remote sensing data can be found in Ban et al. (2015), Gómez et al. (2016), and Grekousis et al. (2015). Spatial resolution for land cover data products ranges from 30 m to 1° (i.e., grid box size of about 100 km). The most popular spatial resolution is 1 km. Several land cover products were produced at 250-, 300-, and 500-m spatial resolutions, such as those from the ESA GlobCover and ESA Climate Change Initiative projects (Arino et al. 2008, 2011; Bontemps et al. 2015) and MODIS Land Cover project (Friedl et al. 2010, Channan et al. 2014).

3. Study area and altitude distribution of land ice

The map of surface elevation over the study area is shown in Fig. 2. It encompasses the entire Canadian Arctic region with a boundary defined in Trishchenko et al. (2016) and Fontana et al. (2010). The elevation data were taken from the Canadian Digital Elevation Database available online (CDED 2007). The data were aggregated to the 250-m spatial resolution consistent with MODIS imagery (Luo et al. 2008; Trishchenko et al. 2016). The RGI 5.0 dataset was used as the main source of land ice information for our study region. No major changes between v5.0 and v6.0 were reported for our region except minor data regrouping between subregions, which has no impact on our study (RGI 2017).

Fig. 2.
Fig. 2.

Surface elevation over the Canadian Arctic region: (a) map and (b) statistical distributions.

Citation: Journal of Climate 31, 3; 10.1175/JCLI-D-17-0198.1

Figure 2a shows the digital elevation map, and Fig. 2b shows two probability distributions. The black curve corresponds to the differential relative probability distribution defined as a percentage of the total area lying within the 100-m altitude interval around the altitude H. The red curve shows the integral probability distribution function defined as (i.e., the probability of the elevation to be located above the altitude H). It can be seen that nearly 30% of the total area of the region is located within 100-m altitude, and 80% of the region is located below 500 m.

Figure 3 shows analysis of the RGI data in the Canadian Arctic region with respect to the surface elevation. This information is useful to understand the altitude distribution of land ice. Figure 3a displays the total extent of the glaciated area above the altitude H. It is expressed on the left scale in absolute units (km2) and on the right scale as a percentage of the total extent of the area defined on Fig. 2a. It is seen from Fig. 3a that the entire glaciated area extent in Canadian Arctic region is close to 150 710 km2 or 9.8% of its total area. Figure 3b shows the absolute value of the differential (i.e., within the 100-m altitude interval) glaciated area extent in km2 as a function of altitude. This function has a peak of 15.7 × 103 km2 at around H = 1000 m. The glaciated area fraction (i.e., relative area) (i.e., within the 100-m altitude interval) occupied by land ice is shown in Fig. 3c. This function is defined as , where is the differential surface area within a 100-m interval around the altitude H, so that the total area . As expected, the glaciated area fraction grows with altitude. At low altitudes below 200 m it is less than 1%. At altitudes greater that 1500 m the glaciated area fraction exceeds 95%, which reflects the air temperature vertical lapse rate and climatically low overall air temperatures for this region. For the altitude region around H = 1000 m where the peak of the differential glaciated area extent is observed, the glaciated area fraction normally exceeds 75%.

Fig. 3.
Fig. 3.

The altitude distribution of land ice over the Canadian Arctic from RGI 5.0 dataset.

Citation: Journal of Climate 31, 3; 10.1175/JCLI-D-17-0198.1

4. CCRS MSI data from MODIS

To derive a homogeneous time series representing the land ice and snow cover, we employed MODIS imagery processed at CCRS as described by Trishchenko et al. (2016). This approach leads to derivation of the MSI extent at 250-m spatial resolution. The spatial resolution of 250 m serves as a very good compromise to GCOS requirements because of the large disparity in the spatial resolution requirements between land ice and snow extent mapping, and ensures the consistency of data for the land ice and snow. We briefly summarize below the details of our processing methodology.

The MSI extent is produced from the snow/ice probability maps derived from MODIS/Terra imagery. The MODIS imagery for the five bands (B3 to B7) designed for land applications is downscaled to 250-m spatial resolution collocated with imagery bands B1 and B2 acquired at 250-m resolution. The downscaling procedure utilizes nonlinear adaptive regression and radiometric normalization to exploit the properties of simultaneously acquired imagery at 500-m resolution as described by Trishchenko et al. (2006, 2009), Luo et al. (2008), and Khlopenkov and Trishchenko (2008). The downscaled imagery for seven MODIS bands B1–B7 is then aggregated into 10-day clear-sky composites, atmospherically corrected, normalized to a standard geometry, and then used to derive the sequence of 10-day snow/ice flags for the warm season period 1 April to 20 September each year (maximum of 17 values). The sequence of flags is converted into the probability of snow/ice detection for each pixel computed as a ratio of number of snow/ice pixels to the total number of valid observations per pixel during the warm season. The probability value 100% is used to identify the permanent snow/ice presence during the entire warm season. The time series of probability maps used to derive the MSI extent are publicly available from the Government of Canada Federal Geospatial Platform (FGP) data archive at http://open.canada.ca/data/en/dataset/808b84a1-6356-4103-a8e9-db46d5c20fcf (Trishchenko 2016). The land/water masking was based on the Global Self-Consistent, Hierarchical, High-Resolution Geography Database with full resolution (GSHHG_f) (Wessel and Smith 1996) that was rasterized and remapped at 250 spatial solution for consistency with MODIS imagery.

The smallest MSI extent (1.53 × 105 km2) in the CCRS data was observed in 2012, and the largest extent (2.09 × 105 km2) was observed in 2013. The average MSI value over the 2000–16 period was 1.70 × 105 km2. Comparison with the RGI data showed that semipermanent snowpack in the Canadian Arctic that persists through the entire melting season is a significant component relative to the ice caps and glacier-covered areas. It can occupy up to 36% or 5.58 × 104 km2 of the total MSI area.

Time series of the derived MSI extent over Canadian Arctic region since 2000 are shown in Fig. 4. The data for several global land cover classifications are also plotted for comparison, including 1) the Global Land Cover 2000 dataset (GLC-2000) produced by the European Joint Research Centre (Bartholomé and Belward 2005; Mayaux et al. 2006), 2) the ESA GlobCover (GLC) 2005 and 2009 datasets (Arino et al. 2008, 2011), and 3) the ESA Climate Change Initiative land cover datasets representative for the 1998–2002 (~2000), 2003–07 (~2005), and 2008–12 (~2010) epochs (Defourny et al. 2014). One can draw three important conclusions from Fig. 4: 1) a noticeable year-to-year variability of MSI extent (up to 36%) derived from CCRS MODIS processing, 2) significant differences between CCRS MSI and some global land cover classifications that can reach 194% (absolute bias of 3.25 × 105 km2) for GCL 2000 data, and 3) comparatively close agreement between land ice inventory RGI and CCRS MSI extent.

Fig. 4.
Fig. 4.

Time series of the minimum snow and ice (MSI) extent over the Canadian Arctic region derived from CCRS MODIS processing. Results for permanent snow/ice land cover type are also plotted for several land cover datasets.

Citation: Journal of Climate 31, 3; 10.1175/JCLI-D-17-0198.1

The altitudinal distribution of the MSI extent computed for 100-m steps is presented in Fig. 5 for each year from 2000 to 2016. The RGI results are also plotted for reference. The MSI and RGI distributions above approximately 1400-m elevation are very close to each other. This reflects the fact that high-elevation areas in the Arctic region are glaciated, and the extent of glaciation is close to 100% (see Fig. 3c). As such, the main year-to year variability of the MSI extent occurs over lower-altitude regions below 1400-m elevation. Two distributions, for 2012 and 2013, when minimum and maximum MSI extents were observed over the region, are plotted in thick brown (2012) and green (2013) lines. The MSI line for 2012 follows RGI results very closely for all altitudes with the exception of the lowest 100-m level. This indicates nearly complete melt of snow covered areas in 2012. Some differences within the lowest 100-m elevation level can be attributed to uncertainties in the coastline database and land/water masking. One can also notice the small negative difference (MSI − RGI) in the areal extent for some altitudes in Fig. 5b. It can be in the range of 5%–10% (relative to average values at specific altitude). This probably indicates the level of uncertainty in our MSI results and RGI mapping. The accuracy of snow/ice mapping during the warm season using optical data is affected by uncertainties in reflective properties of snow/ice over land during possible melt–freeze cycling and exposure of snow impurities and debris as a result of the melting process. Despite these uncertainties, the results presented in Figs. 4 and 5 demonstrate substantial year-to-year variability that needs to be assessed for consistency with climate variations and other observational sources of information.

Fig. 5.
Fig. 5.

The MSI and RGI5.0 spatial extent distribution with altitude (at 100-m steps) for each year since 2000: (a) spatial extent and (b) difference MSI − RGI5.0.

Citation: Journal of Climate 31, 3; 10.1175/JCLI-D-17-0198.1

5. Variations of climate over Canadian Arctic region since 2000

a. Data sources

The data from the ERA-Interim reanalysis project (Dee et al. 2011) and the North American Regional Reanalysis (NARR) project (Mesinger et al. 2006) are employed to analyze the variation of climate over the Canadian Arctic region. We used the ERA-Interim dataset with 6-h temporal resolution and 0.125° × 0.125° spatial resolution. The monthly mean data are also utilized at the same spatial resolution. The ERA-Interim data distributed at 0.125° × 0.125° spatial resolution show signs of interpolation from the coarser-resolution data originally archived at 0.75° × 0.75°. The NARR dataset has spatial resolution of 32 km × 32 km and 3-h temporal resolution. The monthly mean NARR data are also used. The surface level satellite radiation energy budget datasets from NASA’s Clouds and the Earth’s Radiant Energy System (CERES) project are also analyzed for comparison with reanalyzed data. The Energy Balanced and Filled (EBAF) version 2.8 of CERES data was used (CERES Science Team 2017). The CERES EBAF data are available as of time of writing this paper as monthly mean fluxes since March 2000 to February 2016 at 1° × 1° spatial resolution. Details about the GlobSnow SWE dataset used in this paper are provided in section 2 (Luojus et al. 2013).

All reanalyses and satellite data fields were remapped into the map projection compatible with our MODIS imagery at a spatial resolution of 5 km × 5 km. The land/water mask derived from GSHHG_f dataset (Wessel and Smith 1996) was applied as described above in section 4 and by Trishchenko et al. (2016).

b. Variations of summer snow cover

The most direct comparison with the CCRS MSI product can be achieved through comparison of similar product generated from the reanalysis data. This can be done for the NARR snow cover data (variable SNOWC). There is no similar variable in the ERA-Interim dataset that mixes together the snow and land ice (glaciers) in the snow physics accumulation and melt processes. Glaciers in the ERA-Interim modeling scheme are represented by 10-m liquid water equivalent over the nominal glacier mask generated at the coarse spatial resolution 0.75° × 0.75° (ECMWF 2016). A very coarse and imprecise ECMWF glacier mask over the Canadian Arctic region leads to systematic overestimation of glaciated area by a factor of ~2.8 and to very small interannual variability (~1%) in summer MSI. For this reason we found that ERA-Interim snow modeling results are not suitable for our purpose.

Unlike the ERA-Interim, the NARR snow cover (SNOWC) provides a very attractive choice to estimate the summer snow cover over our study region. It contains a binary (0/1) mask of snow cover at the surface level. During the summer season, snow cover may disappear over the glaciated areas if thermodynamics and water budget conditions are favorable for complete melt of the snowpack. For compatibility with MODIS MSI data, the constant spatial mask of land ice generated from RGI and aggregated to 5-km spatial scale was merged with the NARR snow cover data. To exclude the impact of sea ice that could affect the snow/ice mapping over the land during a remapping step, the land/water mask derived from the NARR time-invariant geopotential height data at the NARR native resolution and projection was applied before the remapping process.

The NARR snow cover with 3-h temporal resolution mapped onto a 5 km × 5 km grid was analyzed for April–September to determine the probability of snow presence for each point in a way similar to CCRS MSI processing. The probability was then computed as a percentage of observations with snow cover with respect to the total number of points in a temporal sequence. Examples of the NARR and CCRS MODIS snow probability maps for the 2012 and 2013 April–September seasons are presented in Fig. 6 to show two extremes (minimum and maximum) in CCRS MSI annual time series. There is an overall good agreement between satellite and NARR distributions especially for the areas of high snow/ice probability (i.e., persistent snow/ice presence). Significantly larger area of snow/ice-covered regions (red colors) is clearly seen in 2013 maps for both datasets with respect to 2012. Time series of NARR MSI extent (>98% probability) are plotted in Fig. 7 together with CCRS MSI data. The average difference between two datasets is about 2.2 × 103 km2 or 1.3% of the mean value. The correlation coefficient is equal to 0.56. In the beginning of the period, especially in 2002, the NARR MSI shows slightly lower values, and underestimation of peak MSI value in 2013 can be also noted. The range of variations (maximum–minimum) for NARR is 3.82 × 104 km2 versus 5.58 × 104 km2 for CCRS MSI; the standard deviations are 9.8 × 103 km2 versus 13.8 × 103 km2, correspondingly. Overall, the results display very good agreement and demonstrate the presence of year-to-year variations associated with the summer snowpack that persists over the entire summer period.

Fig. 6.
Fig. 6.

Snow/ice probability maps for April–September 2012 and 2013 derived from (left) the NARR dataset and (right) the CCRS MODIS product.

Citation: Journal of Climate 31, 3; 10.1175/JCLI-D-17-0198.1

Fig. 7.
Fig. 7.

Comparison of warm season MSI time series from MODIS against MSI derived from the NARR snow cover variable SNOWC.

Citation: Journal of Climate 31, 3; 10.1175/JCLI-D-17-0198.1

Additional analyses of temperature, energy fluxes, SWE, and snowfall amounts are presented in the following sections to understand better the local climate dynamics leading to the existence of summer snowpack in the Canadian Arctic region.

c. Temperature field variations

The time series of air temperature T at 2-m height from the ERA-Interim and NARR datasets are analyzed first. In addition to monthly mean temperatures, we also evaluated the length of the melt period (i.e., when T > 0°C) on a monthly scale and over the duration of the warm season. Unlike the MSI analysis, we defined the warm season for climate analysis as period from 1 May to 30 September, because the actual snowmelt over our study region starts in May and continues till September. Including April in our analysis of climate variations would lead, therefore, to inconsistencies and possible biases. Variable is defined as follows:
e1
where is the temporal resolution (time steps) of the reanalysis datasets (i.e., 3 h for NARR and 6 h for ERA-Interim).

The spatial maps of monthly mean values of temperature and parameter are shown in Figs. 8a and 8b for two years corresponding to the smallest (2012) and highest (2013) MSI extents detected in CCRS MSI time series. The data for NARR and ERA-Interim datasets are plotted side by side for easy comparison. The comparison of year 2012 versus 2013 clearly reveals generally warmer conditions over the region in 2012, with the melting season lasting longer and being more intensive. It follows from Figs. 8a and 8b that the month of July is the warmest month with highest average temperatures and the longest duration of melt period. Thermal conditions for summer 2012 confirmed that melting conditions lasted during the entire month of July (except very small areas of high elevation on Ellesmere Island) and therefore could lead to a minimum value of the MSI in 2012 that is close to glaciated area extent. The situation for 2013 shows much colder temperatures and shorter melting periods over the high Arctic region (Ellesmere Island and Baffin Island) in July for both datasets NARR and ERA-Interim, which may lead to incomplete melt and/or accumulation of snowfall that maintains the snowpack over the entire warm season.

Fig. 8.
Fig. 8.

(a) The spatial distribution of the monthly mean temperature (2 m) between May and September for years with minimum (2012) and maximum (2013) MSI spatial extents in the Canadian Arctic from the NARR and ERA-Interim datasets.

Citation: Journal of Climate 31, 3; 10.1175/JCLI-D-17-0198.1

Fig. 8.
Fig. 8.

(b) As in Fig. 8a, but for the spatial distribution of the duration of time interval T > 0°C.

Citation: Journal of Climate 31, 3; 10.1175/JCLI-D-17-0198.1

It follows from Fig. 5 that the major variability of the MSI extent relative to RGI occurs at altitudes <1400 m. As such, we computed the May–September average values over the region areas where altitudes were <1400 m. The time series of regional average values (<1400 m) for 2000–16 period are plotted in Fig. 9 for T and L. Significant negative correlation between temperature characteristics and the MSI extent can be clearly observed. The correlation coefficients and their significance (p values) are shown in Table 1. The absolute value of the negative correlation coefficient can be as high as 0.77. Despite some small differences that could be observed between the NARR and ERA-Interim datasets, they demonstrate a very high degree of consistency in time series shown in Fig. 9.

Fig. 9.
Fig. 9.

Time series of regional mean characteristics for the altitudes <1400 m over the Canadian Arctic region from the ERA-Interim and NARR datasets over May–September: (a) air temperature at 2-m height and (b) duration of the melt season L. (c) The time series of the MSI extent are also shown for reference. The orientation of the y axes in (a) and (b) is inverted to show the synchrony with the MSI extent.

Citation: Journal of Climate 31, 3; 10.1175/JCLI-D-17-0198.1

Table 1.

Coefficient of correlation (r), between minimum snow/ice (MSI) extent and average T, L, total net surface energy, and radiative fluxes for the May–September period and altitudes <1400 m. The p value is also included.

Table 1.

d. Variations of energy budget components

The surface energy budget is the primary driver that influences the snowmelt conditions. The basic physical model of snowmelt contains the energy and water balance equations formulated for a unit area (Gray and Male 1981; Armstrong and Brun 2008):
e2
where is the rate of change in the internal energy of the snow layer, is the energy flux available for snowmelt, is the net shortwave (SW) or solar energy flux at the snow layer top, is the net longwave (LW) or thermal energy flux at the snow layer top, is the energy flux brought in by precipitation, is the ground heat flux into the snow layer at the bottom, is the sensible (convective) heat flux at the air–snow interface, is the latent heat flux at the air–snow interface, is the snow water equivalent, is the rain fall rate, is the snowfall rate, is the meltwater flow (runoff) from the snowpack, and is the snow sublimation. Equation (2) shows that during the snowmelt process the net energy received by the snow is distributed between the internal energy of the snow (and effective surface layer at the snow layer bottom that is involved in the energy exchange) and the energy available for snowmelt when snow temperature exceeds T = 0°C. The snowmelt process continues until the entire snow layer melts [i.e., SWE (variable W above) becomes 0]. The amount of snowmelt produced by can be calculated as , where is density of water, is latent heat of fusion, and is mass fraction of frozen water (snow and ice) in the wet snow layer.

The surface net shortwave (net SW) and net longwave (net LW) radiative fluxes as well as sensible and latent heat fluxes are analyzed in this section. The ground heat flux and the energy flux from precipitation are not included in the analysis, because they are on average very small relative to other components (Armstrong and Brun 2008; Boike et al. 2003). Similar to temperature, we considered the monthly mean fluxes from May to September only, as the melting season over this region effectively starts in May, as seen from Fig. 8. All fluxes are remapped into Lambert conformal conic (LCC) geographic map projection compatible with CCRS MSI product, but at 5 km × 5 km spatial resolution. No height adjustments were included in the remapping process. It is also understood that the coarse-resolution products may introduce some uncertainties due to mixing of the land and water surfaces (Shi et al. 2010).

Time series for 2000–16 for all flux components are analyzed, but only examples of the total surface net flux maps for 2012 and 2013 are plotted to show the range of variability between two extreme years. The total surface net fluxes are generally obtained as a result of summation of relatively large values for various components with opposite signs (incoming shortwave vs outgoing longwave radiation; sensible and latent heat fluxes) as seen in Fig. 10. The maps of the total surface net fluxes derived from the NARR and ERA-Interim datasets for two extreme years 2012 and 2013 are shown in Fig. 11. In general the average total surface net fluxes for the entire warm season vary between −10 and +60 W m−2. The negative and small positive values are observed over Ellesmere Island in the ERA-Interim dataset. The large values are sometimes observed among the coastlines, which may be partially attributed to artifacts from remapping of coarse spatial resolution maps into higher spatial resolution maps. The overall larger values of the total net fluxes for 2012 relative to 2013 can be clearly seen in Fig. 11 for both datasets. This is consistent with the temperature analysis presented earlier in section 5c.

Fig. 10.
Fig. 10.

Region average surface flux components for the altitudes <1400 m over the Canadian Arctic region from ERA-Interim and NARR datasets over the May–September period.

Citation: Journal of Climate 31, 3; 10.1175/JCLI-D-17-0198.1

Fig. 11.
Fig. 11.

Warm season (May–September) average maps of the total net surface fluxes from the NARR and ERA-Interim reanalyses.

Citation: Journal of Climate 31, 3; 10.1175/JCLI-D-17-0198.1

Regional average values of the net total surface flux and net total radiative fluxes for the May–September period and altitudes below 1400 m are plotted in Fig. 10. It shows all major energy flux components (netSW, netLW, and sensible and latent heat fluxes) separately. Overall, the ERA-Interim total average net flux is larger than the NARR one (up to 12 W m−2). The NARR total net surface flux shows a long-term trend that is not observed in ERA-Interim data (Fig. 12). The ERA-Interim data show much smaller range of variability relative to the NARR dataset. The NARR minimum value of the total surface net flux is 12.9 W m−2, the maximum is 31.0 W m−2, and the range is 18.1 W m−2. The ERA-Interim minimum value of the total surface net flux is 24.6 W m−2, the maximum value is 29.1 W m−2, and the range is 4.5 W m−2. Analysis of flux components in Fig. 10 reveals that the major energy input comes from the net shortwave radiation (~100 W m−2), which also shows high degree of negative correlation with MSI. The warm season region average surface net longwave radiation flux (~ −50 W m−2) and sensible and latent heat fluxes (both are in the range of ~ −15 W m−2) have negative sign. As a result, the surface total net energy flux is a small residue and, therefore, potentially may have relatively large uncertainties.

Fig. 12.
Fig. 12.

Region average surface (a) total and (b) radiative net fluxes for the altitudes <1400 m over the Canadian Arctic region from the ERA-Interim, NARR, and CERES datasets over the May–September period. (c) The time series of the MSI extent are also shown for the reference.

Citation: Journal of Climate 31, 3; 10.1175/JCLI-D-17-0198.1

Time series of the total surface net fluxes and the total net radiative fluxes are plotted separately in Fig. 12. They show region mean values for altitudes below 1400 m averaged over the May–September period. For comparison, Fig. 12c displays also CCRS MSI time series. In addition to the NARR and the ERA-Interim results, the CERES radiative fluxes are also plotted. On average, the CERES total surface radiative net flux (66.4 W m−2) is larger than ERA-Interim (59.0 W m−2) and NARR (54.0 W m−2) fluxes. One can see the generally negative correlation between flux components and MSI. The correlation coefficients shown in Table 1 for MSI are most significant with CERES radiative fluxes (−0.69) and with ERA-Interim total net surface flux (−0.61). Substantial negative correlation between MSI and the total net and radiative fluxes corroborate well with our analysis of the multiyear variations in the Canadian Arctic snow cover.

e. Variations of snow water equivalent SWE and snowfall amounts

The SWE values are analyzed in this section to understand better the climate conditions associated with variations of snow cover for our study region. The time series of monthly mean values of SWE are plotted in Fig. 13 for the NARR, ERA-Interim, and GlobSnow satellite product since January 2000. The SWE for the ERA-Interim dataset was taken from the variable called “snow depth” (SD) expressed in millimeters of SWE. The SWE for NARR reanalysis dataset was taken from the variable called “monthly accumulated snow at surface” (WEASD) also expressed in millimeters of SWE. In computing the regional averages for the ERA-Interim dataset, all SWE values greater than 500 mm were excluded because the SWE values over the regions labeled as land ice with coarse-resolution mask (0.75° × 0.75°) are assigned very large values equal to 10 m. The SWE in GlobSnow over high-elevation areas and wet conditions is labeled as missing data or assigned a nominal small value of 0.01 mm (Luojus et al. 2013). For compatibility with previous statistics we excluded all data for altitudes above 1400 m. The SWE in GlobSnow time series for summer months June, July, and August is usually missing.

Fig. 13.
Fig. 13.

Areal average monthly mean time series of SWE for the altitudes <1400 m over the Canadian Arctic archipelago land pixels from ERA-Interim, NARR, and GlobSnow.

Citation: Journal of Climate 31, 3; 10.1175/JCLI-D-17-0198.1

Time series of monthly mean SWE averaged for all good pixels over our region as explained above are plotted in Fig. 13. This figure shows that the annual minimum SWE values are typically observed in August. The annual maximum values are typically observed in April. At the peak of accumulation season in April, the regional average SWE can reach values up to 80–90 mm (GlobSnow and ERA). The NARR SWE values are on average smaller than ERA by ~10 mm, but the difference for some years can be as high as ~30 mm. There is a distinct minimum in SWE time series observed for 2007–09 period in the ERA-Interim and NARR data. This minimum is not seen in the GlobSnow data, which has significant positive bias by ~30 mm for the large 7-yr period spanning 2004–10 with respect to both reanalyzes. The local annual maximum SWE values can frequently be around 100–150 mm, or even reach 250 mm in some elevated locations (according to NARR data). The annual minimum SWE maps for August 2012 and 2013 (not shown) indeed demonstrate that the areal extent of snow cover (i.e., SWE > 0) is visibly larger for 2013 than for 2012. This is also consistent with maps of warm season snow probability for the 2012 and 2013 warm seasons shown in Fig. 6. The overall conclusion from analysis of SWE time series is that both reanalysis datasets show a consistent presence of snow cover in high Arctic during the entire summer season in agreement with results presented in section 5b.

To understand better the snow dynamics in the summer we evaluated the contribution of frozen precipitation processes in the cold Arctic environment that can contribute toward sustaining the snow cover over entire warm season. These data were extracted from the ERA-Interim dataset containing snowfall amount (variable SF). There is no direct snowfall parameter in the NARR dataset that contains the accumulated total precipitation amount (APCP) supplemented with categorical parameters (yes/no) indicating snow (CSNOW), rain (CRAIN), freezing rain (CFRZR), and ice pellets (CICEP). To determine the frozen precipitation amounts from the NARR data we processed 3-hourly time series for each year 2000–16. Precipitation events with categories of snow, freezing rain, and ice pellets were combined together to get the snowfall amounts. The monthly mean statistics of snowfall amounts derived as described above are presented in Fig. 14. The standard deviations are also plotted. The areal means over the entire region and those computed for altitudes below 1400 m are very close to each other (<1% difference). The maximum snowfall amounts occur in October (27 and 21 mm for the NARR and ERA-Interim, correspondingly). The minimum snowfall amounts occur in July (5 and 3 mm for NARR and ERA-Interim respectively). There is also a pronounced maximum in snowfall amounts at the end of the spring season in May when values can reach 18 mm (NARR) and 13 mm (ERA-Interim). The NARR results are systematically higher than ERA-Interim (by ~2.7 mm on average); however, the correlation between two datasets is very high (0.93). The minimum snowfall in July is 37%, and May–September snowfall is 95% of long-term monthly mean value for the NARR data. Corresponding values for the ERA-Interim dataset are 21% and 77%. The statistics for snowfall amounts presented in Fig. 14 shows significance of frozen precipitation in the warm season (77%–95%). They are significant even in July (21%–37%) when the most intensive melting occurs. One can conclude, therefore, that summer snowfall is a systematic and important feature, which, in combination with the negative anomalies of temperature and net total surface flux, contributes toward sustaining the warm season snow cover and its interannual variability in the Canadian Arctic region.

Fig. 14.
Fig. 14.

Monthly mean values of snowfall amounts (SWE) for 2000–16 from the NARR and ERA-Interim datasets over the Canadian Arctic region. The error bars depict standard deviations. The mean snowfall for May–September period is 95% (77%) of the long-term annual mean for the NARR (ERA-Interim) dataset. Corresponding numbers for annual minimum (July) are 37% and 21%.

Citation: Journal of Climate 31, 3; 10.1175/JCLI-D-17-0198.1

6. Conclusions

The Arctic is considered an important region because of its significance for the Earth climate system and sensitivity to climate change. Despite long-term efforts invested by the scientific community to better understand this region, many features of this cold high-latitude environment are known with large uncertainty. To a very significant degree this is due to remoteness and harsh environmental conditions, low population, and a sparse ground observational network over this area. In the current study we focused on multiyear variations of land snow and ice cover extent in the Canadian Arctic Archipelago during the melt season. Snow cover is an important climate parameter included by GCOS in the list of essential climate variables. The major motivation for this research was finding very large discrepancies for snow/ice land cover class (up to nearly 200%) in several major land cover classifications for the Canadian Arctic archipelago. To establish an improved baseline, a new time series of minimum snow/ice (MSI) extent has been produced with a new approach for the MODIS 250-m satellite data processing at the Canada Centre for Remote Sensing based on multitemporal analysis of data for the entire warm season as described by Trishchenko et al. (2016). The MSI data at 250-m spatial resolution represent a very good compromise among GOCS requirements for land ice and snow mapping containing enough details to resolve the main spatial features of glaciers, ice caps, and snowpack.

CCRS MSI time series derived from MODIS shows very good consistency with the Randolph Glacier Inventory (RGI) and are close to ESA Climate Change Initiative (CCI) land cover. However, unlike ESA CCI, CCRS MSI time series display clear year-to-year variability over a 17-yr time span (2000–16). The smallest MSI extent (1.53 × 105 km2) was observed in 2012 and the largest (2.09 × 105 km2) was observed in 2013, with an average value of 1.70 × 105 km2. We employed several reanalyses and observational datasets to explain the derived MSI variations. Among them are the ERA-Interim reanalysis, the North American Regional Reanalysis (NARR), the Clouds and the Earth’s Radiant Energy System (CERES) radiative fluxes, and the European Space Agency’s GlobSnow dataset.

The detailed analysis of altitude distribution of the MSI extent was conducted and compared with RGI results. It showed that all areas above the elevation of 1400 m are nearly 100% glaciated, and this is consistent with the MSI 17-yr record. The interannual variability in the MSI extent is concentrated in low-altitude regions. Analysis and comparison with the Randolph Glacier Inventory showed two important facts: 1) the semipermanent snowpack in the Canadian Arctic that persists through the entire melting season is a significant component relative to the ice caps and glacier-covered areas (up to 36% or 5.58 × 104 km2) and agrees very well with minimum snow cover variations derived from the NARR reanalysis (average difference 1.3%), and 2) the regional MSI variations over the Canadian Arctic region are related to variations in the local climate dynamics, such as warm season average temperature and energy fluxes. The correlation coefficients (absolute values) can be as high as 0.77. The snowfall amounts during the warm season (May–September) are significant (77%–95% of the average long-term monthly mean). They are significant even in July (21%–37%) when the most intensive melting conditions take place. As such, one can conclude that the warm season frozen precipitation events combined with the negative anomalies of temperature and net total surface flux are able to sustain the snowpack during the melt season and contribute to interannual variations of snow cover in the Canadian Arctic that persists over the entire warm season.

The reasons why several major international land cover datasets contain significant positive biases relative to our results cannot be precisely answered in this study. Deficiencies in the clear-sky compositing process, data temporal synthesis, and analysis of seasonal variations, as well as some specific details of the snow/ice cover identification procedure, could potentially contribute to these biases.

Acknowledgments

This work was conducted as part of the Climate Change Geoscience Program and CCRS Long-Term Satellite Data Records (LTSDR) project of Natural Resources Canada (NRCan). Authors are grateful to G. Choma from CCRS/NRCan for useful suggestions and careful editing of the manuscript. The authors are also grateful to the anonymous reviewers for extremely useful and constructive comments that helped to improve this study. The NARR reanalysis data were provided by NOAA ESRL (ftp://ftp.cdc.noaa.gov/Datasets/NARR/monolevel/). The ERA-Interim data were provided by the European Centre for Medium-Range Weather Forecasts (http://apps.ecmwf.int). The GlobSnow dataset was obtained from the ESA GlobSnow project archive (http://www.globsnow.info). Authors gratefully acknowledge the use of MODIS data acquired from the NASA Distributed Archive Center and the satellite CERES radiative fluxes obtained from the NASA’s Langley Research Center archive (https://eosweb.larc.nasa.gov/project/ceres/ceres_table). This manuscript was assigned the NRCan contribution number 20170262.

REFERENCES

  • Arino, O., P. Bicheron, F. Achard, J. Latham, R. Witt, and J. L. Weber, 2008: GlobCover—The most detailed portrait of Earth. ESA Bulletin, No. 136, European Space Agency, Paris, France, 21–34, http://www.esa.int/esapub/bulletin/bulletin136/bul136d_arino.pdf.

  • Arino, O., V. Kalogirou, J. R. Perez, S. Bontemps, P. Defourny, and E. V. Bogaert, 2011: GlobCover 2009: Products description and validation report. European Space Agency, 53 pp., http://dup.esrin.esa.int/files/p68/GLOBCOVER2009_Validation_Report_2.2.pdf.

  • Armstrong, R. L., and E. Brun, Eds., 2008: Snow and Climate: Physical Processes, Surface Energy Exchange and Modeling. Cambridge University Press, 222 pp.

  • Ban, Y., P. Gong, and C. Giri, 2015: Global land cover mapping using Earth observation satellite data: Recent progresses and challenges. ISPRS J. Photogramm. Remote Sens., 103, 16, https://doi.org/10.1016/j.isprsjprs.2015.01.001.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bartholomé, E., and A. S. Belward, 2005: GLC2000: A new approach to global land cover mapping from Earth observation data. Int. J. Remote Sens., 26, 19591977, https://doi.org/10.1080/01431160412331291297.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Boike, J., K. Roth, and O. Ippisch, 2003: Seasonal snow cover on frozen ground: Energy balance calculations of a permafrost site near Ny-Alesund, Spitsbergen. J. Geophys. Res., 108, 8163, https://doi.org/10.1029/2001JD000939.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bontemps, S., and et al. , 2015: Multi-year global land cover mapping at 300 M and characterization for climate modelling: Achievements of the land cover component of the ESA climate change initiative. ISPRS Arch., 40-7W3, 323–328, https://doi.org/10.5194/isprsarchives-XL-7-W3-323-2015.

    • Crossref
    • Export Citation
  • Brasnett, B., 1999: A global analysis of snow depth for numerical weather prediction. J. Appl. Meteor., 38, 726740, https://doi.org/10.1175/1520-0450(1999)038<0726:AGAOSD>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Burt, M. A., D. A. Randall, and M. D. Branson, 2016: Dark warming. J. Climate, 29, 705719, https://doi.org/10.1175/JCLI-D-15-0147.1.

  • Carroll, T., D. Cline, G. Fall, A. Nilsson, L. Li, and A. Rost, 2001: NOHRSC operations and the simulation of snow cover properties for the coterminous U.S. Proc. 69th Western Snow Conf., Sun Valley, ID, Western Snow Conference, https://www.nohrsc.noaa.gov/technology/pdf/wsc2001.pdf.

  • CDED, 2007: Canadian Digital Elevation Data, Level 1: Product specifications, edition 3.0, 2007-06-01. Natural Resources Canada, Canada Centre for Mapping and Earth Observation, accessed 3 October 2016, http://ftp.geogratis.gc.ca/pub/nrcan_rncan/archive/elevation/geobase_cded_dnec/.

  • CERES Science Team, 2017: EBAF-Surface Ed2.8 data. NASA Atmospheric Science Data Center, accessed 12 January 2017, doi:10.5067/Terra+Aqua/CERES/EBAF-Surface_L3B.002.8.

    • Crossref
    • Export Citation
  • Channan, S., K. Collins, and W. R. Emanuel, 2014: Global mosaics of the standard MODIS land cover type data. University of Maryland, College Park, and the Pacific Northwest National Laboratory, http://glcf.umd.edu/data/lc/.

  • Che, T., L. Dai, X. Zheng, X. Li, and K. Zhao, 2016: Estimation of snow depth from passive microwave brightness temperature data in forest regions of northeast China. Remote Sens. Environ., 183, 334349, https://doi.org/10.1016/j.rse.2016.06.005.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ciais, P., and et al. , 2013: Carbon and other biochemical cycles. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 465–570.

  • 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
  • Defourny, P., and et al. , 2014: ESA Climate Change Initiative - Land Cover project. Product User Guide, version 2, 87 pp., https://www.esa-landcover-cci.org/?q=webfm_send/84.

  • Derksen, C., R. Brown, L. Mudryk, and K. Luojus, 2015: Terrestrial snow cover. Arctic Report Card, update for 2015, accessed 6 September 2016, http://www.arctic.noaa.gov/Report-Card/Report-Card-2015/ArtMID/5037/ArticleID/216/Terrestrial-Snow-Cover.

  • Estilow, T. W., A. H. Young, and D. A. Robinson, 2015: A long-term Northern Hemisphere snow cover extent data record for climate studies and monitoring. Earth Syst. Sci. Data, 7, 137142, https://doi.org/10.5194/essd-7-137-2015.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fontana, F. M. A., A. P. Trishchenko, Y. Luo, K. V. Khlopenkov, S. U. Nussbaumer, and S. Wunderle, 2010: Perennial snow and ice variations (2000–2008) in the Arctic circumpolar land area from satellite observations. J. Geophys. Res., 115, F04020, doi:10.1029/2010JF001664.

    • Search Google Scholar
    • Export Citation
  • Frei, A., M. Tedesco, S. Lee, J. Foster, D. K. Hall, R. Kelly, and D. A. Robinson, 2012: A review of global satellite-derived snow products. Adv. Space Res., 50, 10071029, https://doi.org/10.1016/j.asr.2011.12.021.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Friedl, M. A., and et al. , 2002: Global land cover mapping from MODIS: Algorithms and early results. Remote Sens. Environ., 83, 287302, https://doi.org/10.1016/S0034-4257(02)00078-0.

    • Crossref
    • Search Google Scholar
    • 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, 2001–2012, Collection 5.1, IGBP Land Cover, Boston University, Boston, MA, http://glcf.umd.edu/data/lc/.

    • Crossref
    • Export Citation
  • GCOS, 2006: Systematic observation requirements for satellite-based products for climate: Supplemental details to the satellite-based component of the “Implementation Plan for the Global Observing System for Climate in Support of the UNFCC (GCOS-92).” Vol. GCOS-107, WMO/TD-1338, 103 pp., https://library.wmo.int/opac/doc_num.php?explnum_id=3813.

  • GCOS, 2011: Systematic observation requirements for satellite-based data products for climate: 2011 update, supplemental details to the satellite-based component of the “Implementation plan for the Global Observing System for Climate in support of the UNFCCC (2010 update).” GCOS-154, 138 pp., https://library.wmo.int/opac/doc_num.php?explnum_id=3710.

  • GCOS, 2016: 18th Session of the GCOS/WCRP Terrestrial Observation Panel for Climate (TOPC-18). GCOS-202, 38 pp., https://library.wmo.int/opac/doc_num.php?explnum_id=3215.

  • Gómez, C., J. C. White, and M. A. Wulder, 2016: Optical remotely sensed time series data for land cover classification: A review. ISPRS J. Photogramm. Remote Sens., 116, 5572, https://doi.org/10.1016/j.isprsjprs.2016.03.008.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gray, D. M., and D. H. Male, Eds., 1981: Handbook of Snow: Principles, Processes, Management and Use. Pergamon Press, 776 pp.

  • Gregory, J. M., and et al. , 2013: Twentieth-century global-mean sea level rise: Is the whole greater than the sum of the parts? J. Climate, 26, 44764499, https://doi.org/10.1175/JCLI-D-12-00319.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Grekousis, G., G. Mountrakis, and M. Kavouras, 2015: An overview of 21 global and 43 regional land-cover mapping products. Int. J. Remote Sens., 36, 53095335, https://doi.org/10.1080/01431161.2015.1093195.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hall, D. K., G. A. Riggs, V. V. Salomonson, N. E. DiGirolamo, and K. J. Bayr, 2002: MODIS snow-cover products. Remote Sens. Environ., 83, 181194, https://doi.org/10.1016/S0034-4257(02)00095-0.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jung, T., M. A. Kasper, T. Semmler, and S. Serrar, 2014: Arctic influence on subseasonal midlatitude prediction. Geophys. Res. Lett., 41, 36763680, https://doi.org/10.1002/2014GL059961.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jung, T., and et al. , 2016: Advancing polar prediction capabilities on daily to seasonal time scales. Bull. Amer. Meteor. Soc., 97, 16311647, https://doi.org/10.1175/BAMS-D-14-00246.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Khlopenkov, K. V., and A. P. Trishchenko, 2008: Implementation and evaluation of concurrent gradient search method for reprojection of MODIS level 1B imagery. IEEE Trans. Geosci. Remote Sens., 46, 20162027, https://doi.org/10.1109/TGRS.2008.916633.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Luo, Y., A. P. Trishchenko, and K. V. Khlopenkov, 2008: Developing clear-sky, cloud and cloud shadow mask for producing clear-sky composites at 250-meter spatial resolution for the seven MODIS land bands over Canada and North America. Remote Sens. Environ., 112, 41674185, https://doi.org/10.1016/j.rse.2008.06.010.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Luojus, K., and et al. , 2013: GlobSnow-2 Product User Guide Version 1.0. 24 pp., http://www.globsnow.info/se/GlobSnow2_SE_SWE_Product_User_Guide_v1_r1.pdf.

  • Malnes, E., and et al. , 2015: User requirements for the snow and land ice services—CryoLand. Cryosphere, 9, 11911202, https://doi.org/10.5194/tc-9-1191-2015.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mayaux, P., and et al. , 2006: Validation of the global land cover 2000 map. IEEE Trans. Geosci. Remote Sens., 44, 17281737, https://doi.org/10.1109/TGRS.2006.864370.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mernild, S. H., G. E. Liston, and C. A. Hiemstra, 2014: Northern Hemisphere glacier and ice cap surface mass balance and contribution to sea level rise. J. Climate, 27, 60516073, https://doi.org/10.1175/JCLI-D-13-00669.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mesinger, F., and et al. , 2006: North American Regional Reanalysis. Bull. Amer. Meteor. Soc., 87, 343360, https://doi.org/10.1175/BAMS-87-3-343.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Millan, R., J. Mouginot, and E. Rignot, 2017: Mass budget of the glaciers and ice caps of the Queen Elizabeth Islands, Canada, from 1991 to 2015. Environ. Res. Lett., 12, 024016, doi:10.1088/1748-9326/aa5b04.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mudryk, L. R., C. Derksen, P. J. Kushner, and R. Brown, 2015: Characterization of Northern Hemisphere snow water equivalent datasets, 1981–2010. J. Climate, 28, 80378051, https://doi.org/10.1175/JCLI-D-15-0229.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Overland, J., E. Hanna, I. Hanssen-Bauer, S.-J. Kim, J. E. Walsh, M. Wang, U. S. Bhatt, and R. L. Thoman, 2015a: Surface air temperature. Arctic Report Card, update for 2015, http://www.arctic.noaa.gov/Report-Card/Report-Card-2015/ArtMID/5037/ArticleID/210/Surface-Air-Temperature.

  • Overland, J., J. A. Francis, R. Hall, E. Hanna, S.-J. Kim, and T. Vihma, 2015b: The melting arctic and midlatitude weather patterns: Are they connected? J. Climate, 28, 79177932, https://doi.org/10.1175/JCLI-D-14-00822.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Radic, V., and R. Hock, 2010: Regional and global volumes of glaciers derived from statistical upscaling of glacier inventory data. J. Geophys. Res., 115, F01010, https://doi.org/10.1029/2009JF001373.

    • Search Google Scholar
    • Export Citation
  • Ramsay, B., 1998: The interactive multisensor snow and ice mapping system. Hydrol. Processes, 12, 15371546, https://doi.org/10.1002/(SICI)1099-1085(199808/09)12:10/11<1537::AID-HYP679>3.0.CO;2-A.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • RGI, 2015: Randolph Glacier Inventory—A dataset of global glacier outlines: version 5.0: Global Land Ice Measurements from Space (GLIMS), Boulder, CO, accessed 15 August 2016, https://doi.org/10.7265/N5-RGI-50.

    • Crossref
    • Export Citation
  • RGI, 2017: Randolph Glacier Inventory—A dataset of global glacier outlines, version 6.0: Global Land Ice Measurements from Space (GLIMS), Boulder, CO, accessed 15 August 2017, https://doi.org/10.7265/N5-RGI-60.

    • Crossref
    • Export Citation
  • Robinson, D. A., and et al. , 2012: NOAA Climate Date Record (CDR) of Northern Hemisphere (NH) Snow Cover Extent (SCE), Version 1.0. NOAA National Climatic Data Center, accessed 20 September 2016, doi:10.7289/V5N014G9.

    • Crossref
    • Export Citation
  • Romanov, P., 2017: Global Multisensor Automated Satellite-Based Snow and Ice Mapping System (GMASI) for cryosphere monitoring. Remote Sens. Environ., 196, 4255, https://doi.org/10.1016/j.rse.2017.04.023.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Romanov, P., G. Gutman, and I. Csiszar, 2000: Automated monitoring of snow cover over North America using multispectral satellite data. J. Appl. Meteor., 39, 18661880, https://doi.org/10.1175/1520-0450(2000)039<1866:AMOSCO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Shi, X. G., M. Wild, and D. P. Lettenmaier, 2010: Surface radiative fluxes over the pan-Arctic land region: Variability and trends. J. Geophys. Res., 115, D22104, https://doi.org/10.1029/2010JD014402.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Takala, M., K. Luojus, J. Pulliainen, C. Derksen, J. Lemmetyinen, J. P. Kärnä, J. Koskinen, and B. Bojkov, 2011: Estimating Northern Hemisphere snow water equivalent for climate research through assimilation of space-borne radiometer data and ground-based measurements. Remote Sens. Environ., 115, 35173529, https://doi.org/10.1016/j.rse.2011.08.014.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tedesco, M., R. Kelly, J. L. Foster, and A. T. Chang, 2004: AMSR-E/Aqua Monthly L3 Global Snow Water Equivalent EASE-Grids, version 10. NASA National Snow and Ice Data Center Distributed Active Archive Center, accessed 30 August 2016, https://search.earthdata.nasa.gov/search/granules?p=C179014696-NSIDC_ECS&q=AE_MoSno.

  • Trishchenko, A. P., 2016: Probability maps of the annual minimum snow and ice (MSI) presence over April 1 to September 20 period since 2000 derived from MODIS 250m imagery over Canada and neighboring regions. Accessed 15 June 2017, http://open.canada.ca/data/en/dataset/808b84a1-6356-4103-a8e9-db46d5c20fcf.

  • Trishchenko, A. P., Y. Luo, and K. V. Khlopenkov, 2006: A method for downscaling MODIS land channels to 250-m spatial resolution using adaptive regression and normalization. Remote Sensing for Environmental Monitoring, GIS Applications, and Geology VI, International Society for Optics and Photonics (SPIE Proceedings, Vol. 6366), M. Ehlers and U. Michel, Eds., 636607, doi:10.1117/12.689157.

    • Crossref
    • Export Citation
  • Trishchenko, A. P., Y. Luo, K. V. Khlopenkov, W. M. Park, and S. Wang, 2009: Arctic circumpolar mosaic at 250 m spatial resolution for IPY by fusion of MODIS/TERRA land bands B1–B7. Int. J. Remote Sens., 30, 16351641, https://doi.org/10.1080/01431160802348119.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Trishchenko, A. P., S. Leblanc, S. Wang, J. Li, C. Ungureanu, Y. Lu, K. V. Khlopenkov, and F. Fontana, 2016: Variations of annual minimum snow and ice extent over Canada and neighboring landmass derived from MODIS 250-m imagery for 2000–2014. Can. J. Remote Sens., 42, 214242, https://doi.org/10.1080/07038992.2016.1166043.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wessel, P., and W. H. F. Smith, 1996: A global self-consistent, hierarchical, high-resolution shoreline database. J. Geophys. Res., 101, 87418743, https://doi.org/10.1029/96JB00104.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Woo, M., and K. L. Young, 2014: Disappearing semi-permanent snow in the High Arctic and its consequences. J. Glaciol., 60, 192200, https://doi.org/10.3189/2014JoG13J150.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Young, K. L., and A. G. Lewkowicz, 1990: Surface energy balance of a perennial snowbank, Melville Island, Northwest Territories, Canada. Arct. Alp. Res., 22, 290301, https://doi.org/10.2307/1551592.

    • Crossref
    • Search Google Scholar
    • Export Citation
Save