• Anderson, J. L., 2001: An ensemble adjustment Kalman filter for data assimilation. Mon. Wea. Rev., 129, 28842903, doi:10.1175/1520-0493(2001)129<2884:AEAKFF>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Anderson, J. L., 2007: Exploring the need for localization in ensemble data assimilation using a hierarchical ensemble filter. Physica D, 230, 99111, doi:10.1016/j.physd.2006.02.011.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Anderson, J. L., T. Hoar, K. Raeder, H. Liu, N. Collins, R. Torn, and A. Arellano, 2009: The data assimilation research testbed: A community facility. Bull. Amer. Meteor. Soc., 90, 12831296, doi:10.1175/2009BAMS2618.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Andreadis, K. M., and D. P. Lettenmaier, 2012: Implications of representing snowpack stratigraphy for the assimilation of passive microwave satellite observations. J. Hydrometeor., 13, 14931506, doi:10.1175/JHM-D-11-056.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bateni, S. M., S. A. Margulis, E. Podest, and K. C. McDonald, 2015: Characterizing snowpack and the freeze–thaw state of underlying soil via assimilation of multifrequency passive/active microwave data: A case study (NASA CLPX 2003). IEEE Trans. Geosci. Remote Sens., 53, 173189, doi:10.1109/TGRS.2014.2320264.

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

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brown, R. D., and B. Brasnett, 2010: Canadian Meteorological Centre (CMC) Daily Snow Depth Analysis Data (updated annually). National Snow and Ice Data Center, Boulder, CO, digital media, accessed 26 April 2016. [Available online at https://nsidc.org/data/docs/daac/nsidc0447_CMC_snow_depth/.]

  • Brown, R. D., C. Derksen, and L. Wang, 2010: A multi-dataset analysis of variability and change in Arctic spring snow cover extent, 1967–2008. J. Geophys. Res., 115, D16111, doi:10.1029/2010JD013975.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brucker, L., G. Picard, and M. Fily, 2010: Snow grain size profiles deduced from microwave snow emissivities in Antarctica. J. Glaciol., 56, 514526, doi:10.3189/002214310792447806.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brucker, L., A. Royer, G. Picard, A. Langlois, and M. Fily, 2011: Hourly simulations of the microwave brightness temperature of seasonal snow in Quebec, Canada, using a coupled snow evolution–emission model. Remote Sens. Environ., 115, 19661977, doi:10.1016/j.rse.2011.03.019.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Burgers, G., P. J. van Leeuwen, and G. Evensen, 1998: Analysis scheme in the ensemble Kalman filter. Mon. Wea. Rev., 126, 17191724, doi:10.1175/1520-0493(1998)126<1719:ASITEK>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Champollion, N., G. Picard, L. Arnaud, E. Lefebvre, and M. Fily, 2013: Hoar crystal development and disappearance at Dome C, Antarctica: Observation by near-infrared photography and passive microwave satellite. Cryosphere Discuss., 7, 175217, doi:10.5194/tcd-7-175-2013.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chang, A. T. C., J. L. Foster, and D. K. Hall, 1996: Effects of forest on the snow parameters derived from microwave measurements during the BOREAS winter field experiment. Hydrol. Processes, 10, 15651574, doi:10.1002/(SICI)1099-1085(199612)10:12<1565::AID-HYP501>3.0.CO;2-5.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chapin, F. S., and et al. , 2005: Role of land-surface changes in Arctic summer warming. Science, 310, 657660, doi:10.1126/science.1117368.

  • Che, T., X. Li, R. Jin, and C. Huang, 2014: Assimilating passive microwave remote sensing data into a land surface model to improve the estimation of snow depth. Remote Sens. Environ., 143, 5463, doi:10.1016/j.rse.2013.12.009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cohen, J., and D. Entekhabi, 1999: Eurasian snow cover variability and Northern Hemisphere climate predictability. Geophys. Res. Lett., 26, 345348, doi:10.1029/1998GL900321.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Debye, P., H. R. Anderson, and H. Brumberger, 1957: Scattering by an inhomogeneous solid. II. The correlation function and its application. J. Appl. Phys., 28, 679683, doi:10.1063/1.1722830.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dechant, C., and M. Moradkhani, 2011: Radiance data assimilation for operational snow and streamflow forecasting. Adv. Water Resour., 34, 351364, doi:10.1016/j.advwatres.2010.12.009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Durand, M., and S. A. Margulis, 2006: Feasibility test of multifrequency radiometric data assimilation to estimate snow water equivalent. J. Hydrometeor., 7, 443457, doi:10.1175/JHM502.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Durand, M., and S. A. Margulis, 2007: Correcting first-order errors in snow water equivalent estimates using a multifrequency, multiscale radiometric data assimilation scheme. J. Geophys. Res., 112, D13121, doi:10.1029/2006JD008067.

    • Search Google Scholar
    • Export Citation
  • Durand, M., E. J. Kim, and S. A. Margulis, 2008: Quantifying uncertainty in modeling snow microwave radiance for a mountain snowpack at the point-scale, including stratigraphic effects. IEEE Trans. Geosci. Remote Sens., 46, 17531767, doi:10.1109/TGRS.2008.916221.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Durand, M., E. J. Kim, and S. A. Margulis, 2009: Radiance assimilation shows promise for snowpack characterization. Geophys. Res. Lett., 36, L02503, doi:10.1029/2008GL035214.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Evensen, G., 1994: Sequential data assimilation with a nonlinear quasi-geostrophic model using Monte Carlo methods to forecast error statistics. J. Geophys. Res., 99, 10 14310 162, doi:10.1029/94JC00572.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ferrazzoli, P., L. Guerriero, and J.-P. Wigneron, 2002: Simulating L-band emission of forests in view of future satellite applications. IEEE Trans. Geosci. Remote Sens., 40, 27002708, doi:10.1109/TGRS.2002.807577.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Forman, B. A., R. H. Reichle, and M. Rodell, 2012: Assimilation of terrestrial water storage from GRACE in a snow-dominated basin. Water Resour. Res., 48, W01507, doi:10.1029/2011WR011239.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Foster, J. L., A. T. C. Chang, D. K. Hall, and A. Rango, 1991: Derivation of snow water equivalent in boreal forests using microwave radiometry. Arctic, 44, 147152, doi:10.14430/arctic1581.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Foster, J. L., C. Sun, J. P. Walker, R. Kelly, A. Chang, J. Dong, and H. Powell, 2005: Quantifying the uncertainty in passive microwave snow water equivalent observations. Remote Sens. Environ., 94, 187203, doi:10.1016/j.rse.2004.09.012.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gent, P. R., and et al. , 2011: The Community Climate System Model version 4. J. Climate, 24, 49734991, doi:10.1175/2011JCLI4083.1.

  • Gong, G., D. Entekhabi, and J. Cohen, 2003: Modeled Northern Hemisphere winter climate response to realistic Siberian snow anomalies. J. Climate, 16, 39173931, doi:10.1175/1520-0442(2003)016<3917:MNHWCR>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Grant, J. P., K. Saleh, J.-P. Wigneron, M. Guglielmetti, Y. H. Kerr, M. Schwank, N. Skou, and A. Van de Griend, 2008: Calibration of the L-MEB model over a coniferous and a deciduous forest. IEEE Trans. Geosci. Remote Sens., 46, 808818, doi:10.1109/TGRS.2007.914801.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hall, D. K., J. L. Foster, and A. T. C. Chang, 1982: Measurement and modeling of microwave emission from forested snowfields in Michigan. Nord. Hydrol., 13, 129138.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hall, D. K., V. V. Salomonson, and G. A. Riggs, 2006: MODIS/Terra Snow Cover Daily L3 Global 0.05deg CMG, version 5 (updated daily). National Snow and Ice Data Center, Boulder, CO, digital media, accessed 26 April 2016. [Available online at http://nsidc.org/data/docs/daac/modis_v5/mod10c1_modis_terra_snow_daily_global_0.05deg_cmg.gd.html.]

  • Hallikainen, M. T., and P. A. Jolma, 1992: Comparison of algorithms for retrieval of snow water equivalent from Nimbus-7 SMMR data in Finland. IEEE Trans. Geosci. Remote Sens., 30, 124131, doi:10.1109/36.124222.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jackson, T. J., and T. J. Schmugge, 1991: Vegetation effects on the microwave emission of soils. Remote Sens. Environ., 36, 203212, doi:10.1016/0034-4257(91)90057-D.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jin, Y. Q., 1994: Electromagnetic Scattering Modelling for Quantitative Remote Sensing. World Scientific, 348 pp.

    • Crossref
    • Export Citation
  • Knowles, K., M. Savoie, R. Armstrong, and M. Brodzik. 2006a: AMSR-E/Aqua daily global quarter-degree gridded brightness temperatures, version 1. National Snow and Ice Data Center, Boulder, CO, digital media, accessed 26 April 2016, doi:10.5067/RRR4WWORG070.

    • Crossref
    • Export Citation
  • Knowles, K., M. Savoie, R. Armstrong, and M. Brodzik, 2006b: AMSR-E/Aqua daily EASE-grid brightness temperatures. National Snow and Ice Data Center, Boulder, CO, digital media, accessed 26 April 2016. [Available online at https://nsidc.org/data/docs/daac/nsidc0301_amsre_gridded_tb.gd.html.]

  • Kruopis, N., J. Praks, A. N. Arslan, H. M. Alasalmi, J. T. Koskinen, and M. T. Hallikainen, 1999: Passive microwave measurements of snow-covered forest area in EMAC’95. IEEE Trans. Geosci. Remote Sens., 37, 26992705, doi:10.1109/36.803417.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kurum, M., P. E. O’Neill, R. H. Lang, A. T. Joseph, M. H. Cosh, and T. J. Jackson, 2012: Effective tree scattering and opacity at L-band. Remote Sens. Environ., 118, 19, doi:10.1016/j.rse.2011.10.024.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kwon, Y., A. M. Toure, Z.-L. Yang, M. Rodell, and G. Picard, 2015: Error characterization of coupled land surface–radiative transfer models for snow microwave radiance assimilation. IEEE Trans. Geosci. Remote Sens., 53, 52475268, doi:10.1109/TGRS.2015.2419977.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kwon, Y., Z.-L. Yang, L. Zhao, T. J. Hoar, A. M. Toure, and M. Rodell, 2016: Estimating snow water storage in North America using CLM4, DART, and snow radiance data assimilation. J. Hydrometeor., 17, 28532874, doi:10.1175/JHM-D-16-0028.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Langlois, A., A. Royer, F. Dupont, A. Roy, K. Goïta, and G. Picard, 2011: Improved corrections of forests effects on passive microwave satellite remote sensing of snow over boreal and subarctic regions. IEEE Trans. Geosci. Remote Sens., 49, 38243837, doi:10.1109/TGRS.2011.2138145.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Langlois, A., A. Royer, C. Derksen, B. Montpetit, F. Dupont, and K. Goïta, 2012: Coupling the snow thermodynamic model SNOWPACK with the microwave emission model of layered snowpacks for subarctic and arctic snow water equivalent retrievals. Water Resour. Res., 48, W12524, doi:10.1029/2012WR012133.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lawrence, D., and et al. , 2011: Parameterization improvements and functional and structural advances in version 4 of the Community Land Model. J. Adv. Model. Earth Syst., 3, M03001, doi:10.1029/2011MS00045.

    • Search Google Scholar
    • Export Citation
  • Lobl, E., 2001: Joint Advanced Microwave Scanning Radiometer (AMSR) Science Team meeting. Earth Observer, Vol. 13, Issue 3, NASA Goddard Space Flight Center, Greenbelt, Maryland, 3–9. [Available online at https://eospso.nasa.gov/sites/default/files/eo_pdfs/may_jun01.pdf.]

  • Löwe, H., and G. Picard, 2015: Microwave scattering coefficient of snow in MEMLS and DMRT-ML revisited: The relevance of sticky hard spheres and tomography-based estimates of stickiness. Cryosphere, 9, 21012117, doi:10.5194/tc-9-2101-2015.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mätzler, C., 1987: Applications of the interaction of microwaves with the natural snow cover. Remote Sens. Rev., 2, 259387, doi:10.1080/02757258709532086.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mätzler, C., 1998: Improved Born approximation for scattering of radiation in a granular medium. J. Appl. Phys., 83, 61116117, doi:10.1063/1.367496.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mätzler, C., 2002: Relation between grain-size and correlation length of snow. J. Glaciol., 48, 461466, doi:10.3189/172756502781831287.

  • Mätzler, C., H. Aebischer, and E. Schanda, 1984: Microwave dielectric properties of surface snow. IEEE J. Oceanic Eng., 9, 366371, doi:10.1109/JOE.1984.1145644.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mo, T., B. J. Choudhury, T. J. Schmugge, J. R. Wang, and T. J. Jackson, 1982: A model for microwave emission from vegetation-covered fields. J. Geophys. Res., 87, 11 22911 237, doi:10.1029/JC087iC13p11229.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Namias, J., 1985: Some empirical evidence for the influence of snow cover on temperature and precipitation. Mon. Wea. Rev., 113, 15421553, doi:10.1175/1520-0493(1985)113<1542:SEEFTI>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Niu, G.-Y., and Z.-L. Yang, 2007: An observation-based formulation of snow cover fraction and its evaluation over large North American river basins. J. Geophys. Res., 112, D21101, doi:10.1029/2007JD008674.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Njoku, E. G., and L. Li, 1999: Retrieval of land surface parameters using passive microwave measurements at 6–18 GHz. IEEE Trans. Geosci. Remote Sens., 37, 7993, doi:10.1109/36.739125.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Njoku, E. G., P. Ashcroft, T. K. Chan, and L. Li, 2005: Global survey and statistics of radio-frequency interference in AMSR-E land observations. IEEE Trans. Geosci. Remote Sens., 43, 938947, doi:10.1109/TGRS.2004.837507.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Oleson, K. W., and et al. , 2010: Technical description of version 4.0 of the Community Land Model (CLM). NCAR Tech. Note NCAR/TN-478+STR, 257 pp., doi:10.5065/D6FB50WZ.

    • Crossref
    • Export Citation
  • Paloscia, S., 1995: Microwave emission from vegetation. Passive Microwave Remote Sensing of Land–Atmosphere Interactions, B. J. Choudhury et al., Eds., VSP Press, 357–374.

    • Crossref
    • Export Citation
  • Paloscia, S., and P. Pampaloni, 1988: Microwave polarization index for monitoring vegetation growth. IEEE Trans. Geosci. Remote Sens., 26, 617621, doi:10.1109/36.7687.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pampaloni, P., 2004: Microwave radiometry of forests. Waves Random Media, 14, S275S298, doi:10.1088/0959-7174/14/2/009.

  • Pampaloni, P., and S. Paloscia, 1986: Microwave emission and plant water content: A comparison between field measurements and theory. IEEE Trans. Geosci. Remote Sens., 24, 900905, doi:10.1109/TGRS.1986.289705.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pellarin, T., Y. H. Kerr, and J.-P. Wigneron, 2006: Global simulations of brightness temperature at 6.6 and 10.7 GHz over land based on SMMR data set analysis. IEEE Trans. Geosci. Remote Sens., 44, 24922505, doi:10.1109/TGRS.2006.874139.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Picard, G., L. Brucker, A. Roy, F. Dupont, M. Fily, and A. Royer, 2013: Simulation of the microwave emission of multi-layered snowpacks using the dense media radiative transfer theory: The DMRT-ML model. Geosci. Model Dev., 6, 10611078, doi:10.5194/gmd-6-1061-2013.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pulliainen, J., J.-P. Kärnä, and M. Hallikainen, 1993: Development of geophysical retrieval algorithms for the MIMR. IEEE Trans. Geosci. Remote Sens., 31, 268277, doi:10.1109/36.210466.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Raeder, K., J. L. Anderson, N. Collins, T. J. Hoar, J. E. Kay, P. H. Lauritzen, and R. Pincus, 2012: DART/CAM: An ensemble data assimilation system for CESM atmospheric models. J. Climate, 25, 63046317, doi:10.1175/JCLI-D-11-00395.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rees, A., J. Lemmetyinen, C. Derksen, J. Pulliainen, and M. English, 2010: Observed and modelled effects of ice lens formation on passive microwave brightness temperatures over snow covered tundra. Remote Sens. Environ., 114, 116126, doi:10.1016/j.rse.2009.08.013.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reichle, R. H., R. D. Koster, G. J. M. De Lannoy, B. A. Forman, Q. Liu, S. P. P. Mahanama, and A. Toure, 2011: Assessment and enhancement of MERRA land surface hydrology estimates. J. Climate, 24, 63226338, doi:10.1175/JCLI-D-10-05033.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rosenfeld, S., and N. Grody, 2000: Anomalous microwave spectra of snow cover observed from Special Sensor Microwave/Imager measurements. J. Geophys. Res., 105, 14 91314 925, doi:10.1029/1999JD900486.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Roy, A., A. Royer, J.-P. Wigneron, A. Langlois, J. Bergeron, and P. Cliche, 2012: A simple parameterization for a boreal forest radiative transfer model at microwave frequencies. Remote Sens. Environ., 124, 371383, doi:10.1016/j.rse.2012.05.020.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Roy, A., G. Picard, A. Royer, B. Montpetit, F. Dupont, A. Langlois, C. Derksen, and N. Champollion, 2013: Brightness temperature simulations of the Canadian seasonal snowpack driven by measurements of the snow specific surface area. IEEE Trans. Geosci. Remote Sens., 51, 46924704, doi:10.1109/TGRS.2012.2235842.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schmugge, T. J., and T. J. Jackson, 1992: A dielectric model of the vegetation effects on the microwave emission from soils. IEEE Trans. Geosci. Remote Sens., 30, 757760, doi:10.1109/36.158870.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stewart, I. T., D. R. Cayan, and M. D. Dettinger, 2004: Changes in snowmelt runoff timing in western North America under a ‘business as usual’ climate change scenario. Climatic Change, 62, 217232, doi:10.1023/B:CLIM.0000013702.22656.e8.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sturm, M., J. Holmgren, and G. E. Liston, 1995: A seasonal snow cover classification system for local to regional applications. J. Climate, 8, 12611283, doi:10.1175/1520-0442(1995)008<1261:ASSCCS>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sturm, M., B. Taras, G. E. Liston, C. Derksen, T. Jonas, and J. Lea, 2010: Estimating snow water equivalent using snow depth data and climate classes. J. Hydrometeor., 11, 13801394, doi:10.1175/2010JHM1202.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Su, H., Z.-L. Yang, R. E. Dickinson, C. R. Wilson, and G.-Y. Niu, 2010: Multisensor snow data assimilation at the continental scale: The value of Gravity Recovery and Climate Experiment terrestrial water storage information. J. Geophys. Res., 115, D10104, doi:10.1029/2009JD013035.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tedesco, M., and E. J. Kim, 2006: Intercomparison of electromagnetic models for passive microwave remote sensing of snow. IEEE Trans. Geosci. Remote Sens., 44, 26542666, doi:10.1109/TGRS.2006.873182.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tippett, M. K., J. L. Anderson, C. H. Bishop, T. M. Hamill, and J. S. Whitaker, 2003: Ensemble square root filters. Mon. Wea. Rev., 131, 14851490, doi:10.1175/1520-0493(2003)131<1485:ESRF>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Toure, A. M., K. Goïta, A. Royer, E. J. Kim, M. Durand, S. A. Margulis, and H. Lu, 2011: A case study of using a multilayered thermodynamical snow model for radiance assimilation. IEEE Trans. Geosci. Remote Sens., 49, 28282837, doi:10.1109/TGRS.2011.2118761.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Toure, A. M., M. Rodell, Z.-L. Yang, H. Beaudoing, E. Kim, Y. Zhang, and Y. Kwon, 2016: Evaluation of the snow simulations from the Community Land Model, version 4 (CLM4). J. Hydrometeor., 17, 153170, doi:10.1175/JHM-D-14-0165.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tsang, L., and J. A. Kong, 2001: Scattering of Electromagnetic Waves: Advanced Topics. Vol. 3. Wiley-Interscience, 413 pp.

    • Crossref
    • Export Citation
  • Tsang, L., P. Xu, and K. S. Chen, 2008: Third and fourth Stokes parameters in polarimetric passive microwave remote sensing of rough surfaces over layered media. Microwave Opt. Technol. Lett., 50, 30633069, doi:10.1002/mop.23892.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Turner, D. D., S. A. Clough, J. C. Liljegren, E. E. Clothiaux, K. E. Cady-Pereira, and K. L. Gaustad, 2007: Retrieving liquid water path and precipitable water vapor from the Atmospheric Radiation Measurement (ARM) microwave radiometers. IEEE Trans. Geosci. Remote Sens., 45, 36803690, doi:10.1109/TGRS.2007.903703.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ulaby, F. T., R. K. Moore, and A. K. Fung, 1981: Microwave Remote Sensing: Active and Passive. Vol. 1. Addison-Wesley, 456 pp.

  • Vernekar, A. D., J. Zhou, and J. Shukla, 1995: The effect of Eurasian snow cover on the Indian monsoon. J. Climate, 8, 248266, doi:10.1175/1520-0442(1995)008<0248:TEOESC>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wegmüller, U., and C. Mätzler, 1999: Rough bare soil reflectivity model. IEEE Trans. Geosci. Remote Sens., 37, 13911395, doi:10.1109/36.763303.

  • Wiesmann, A., and C. Mätzler, 1999: Microwave emission model of layered snowpacks. Remote Sens. Environ., 70, 307316, doi:10.1016/S0034-4257(99)00046-2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • WMO, 2007: Integrated Global Observing Strategy for the monitoring of our environment from space and from Earth. WMO/TD-1405, 100 pp. [Available online at http://cryos.ssec.wisc.edu/docs/cryos_theme_report.pdf.]

  • Zhang, Y.-F., T. J. Hoar, Z.-L. Yang, J. L. Anderson, A. M. Toure, and M. Rodell, 2014: Assimilation of MODIS snow cover through the Data Assimilation Research Testbed and the Community Land Model version 4. J. Geophys. Res. Atmos., 119, 70917103, doi:10.1002/2013JD021329.

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

    Schematic diagram of the coupled RA system. Atmospheric ensemble forcing used by the atmospheric RTM includes atmospheric pressure, air temperature, and humidity. The CLM4 surface input data read by the RTMs are sand and clay fractions and vegetation fraction.

  • View in gallery

    (a) Snow and (b) land-cover types in North America: snow class are 1) tundra, 2) taiga, 3) maritime, 4) ephemeral, 5) prairie, and 6) alpine; land-cover types are 1) bare soil, 2) forest, 3) shrub, 4) grass, and 5) crop.

  • View in gallery

    The snow depth RMSE (m) in the RA experimental cases, using different snowpack RTMs (DMRT-ML and MEMLS) and assimilating different AMSR-E frequency channels, for North America. The horizontal dotted line represents the RMSE of the open-loop run (without assimilation).

  • View in gallery

    The snow depth RMSE (m) in the RA experimental cases, using different snowpack RTMs (DMRT-ML and MEMLS) and assimilating different AMSR-E frequency channels, for snow classes as defined in Sturm et al. (1995): (a) tundra, (b) taiga, (c) maritime, (d) ephemeral, (e) prairie, and (f) alpine. The horizontal dotted line represents the RMSE of the open-loop run (without assimilation).

  • View in gallery

    The snow depth RMSE (m) in the RA experimental cases, using different snowpack RTMs (DMRT-ML and MEMLS) and assimilating different AMSR-E frequency channels, for land-cover types: (a) bare soil, (b) forest, (c) shrub, (d) grass, and (e) crop. The horizontal dotted line represents the RMSE of the open-loop run (without assimilation).

  • View in gallery

    The snow depth RMSE (m) in the RA experimental cases, using different snowpack RTMs (DMRT-ML and MEMLS) and assimilating two frequency channels simultaneously (i.e., 18.7 and 23.8 GHz or 18.7 and 36.5 GHz) for North America. The vegetation single-scattering albedo was neglected (in the D1823, D1836, M1823, and M1836 cases), was set to 0.064 [in the D1823(ω), D1836(ω), M1823(ω), and M1836(ω) cases], or was updated during the assimilation [in the D1823(ω_up), D1836(ω_up), M1823(ω_up), and M1836(ω_up) cases]. The horizontal dotted line represents the RMSE of the open-loop run (without assimilation).

  • View in gallery

    The snow depth RMSE (m) in the RA experimental cases, using different snowpack RTMs (DMRT-ML and MEMLS) and assimilating two frequency channels simultaneously (i.e., 18.7 and 23.8 GHz or 18.7 and 36.5 GHz) for six snow classes as defined in Sturm et al. (1995): (a) tundra, (b) taiga, (c) maritime, (d) ephemeral, (e) prairie, and (f) alpine. The vegetation single-scattering albedo was neglected (in the D1823, D1836, M1823, and M1836 cases), was set to 0.064 [in the D1823(ω), D1836(ω), M1823(ω), and M1836(ω) cases], or was updated during the assimilation [in the D1823(ω_up), D1836(ω_up), M1823(ω_up), and M1836(ω_up) cases]. The horizontal dotted line represents the RMSE of the open-loop run (without assimilation).

  • View in gallery

    The snow depth RMSE (m) in the RA experimental cases, using different snowpack RTMs (DMRT-ML and MEMLS) and assimilating two frequency channels simultaneously (i.e., 18.7 and 23.8 GHz or 18.7 and 36.5 GHz) for five land-cover types: (a) bare soil, (b) forest, (c) shrub, (d) grass, and (e) crop. The vegetation single-scattering albedo was neglected (in the D1823, D1836, M1823, and M1836 cases), was set to 0.064 [in the D1823(ω), D1836(ω), M1823(ω), and M1836(ω) cases], or was updated during the assimilation [in the D1823(ω_up), D1836(ω_up), M1823(ω_up), and M1836(ω_up) cases]. The horizontal dotted line represents the RMSE of the open-loop run (without assimilation).

  • View in gallery

    The estimated (ensemble mean) vegetation transmissivity in the (a) D1823, (b) M1823, (c) D1823(ω), and (d) M1823(ω) cases. The values were averaged over two frequency channels (18.7- and 23.8-GHz vertical polarizations) during the assimilation period.

  • View in gallery

    The ratio of the estimated (ensemble mean) to the estimated (ensemble mean) in the cases neglecting ω [(a) D1823 and (b) M1823] and in the cases considering ω [(c) D1823(ω) and (d) M1823(ω)]. The values were averaged over two frequency channels (18.7- and 23.8-GHz vertical polarizations) during the assimilation period. (e),(f) Changes in the estimated ratio by considering ω.

  • View in gallery

    Spatial distributions of the snow depth RMSE difference: (a) D1823(ω) − D1823, (b) M1823(ω) − M1823, (c) D1823(ω) − open loop, and (d) M1823(ω) − open loop. Negative and positive values denote the improvement and degradation of the RA performance, respectively.

  • View in gallery

    The estimated (ensemble mean) (a) snow water volume (103 km3) and (b) snow cover area (106 km2) in North America (asterisk indicates the difference was divided by the value of the open-loop run). The CMC snow water volume was calculated using the SWE estimates obtained by Reichle et al. (2011) from the CMC snow depth and climatological snow densities suggested by Sturm et al. (2010).

  • View in gallery

    The improvement of the RA performance in estimating snow depth for (a) North America, (b) bare soil land cover, and (c) forest land cover by simultaneously updating all model physical states and parameters determining TB based on a rule (RArule); by assimilating the best-performing frequency channels, that is, 18.7 and 23.8 GHz (RA1823); and by considering the vegetation single-scattering albedo [RA1823(ω)].

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 58 58 5
PDF Downloads 17 17 1

Improving the Radiance Assimilation Performance in Estimating Snow Water Storage across Snow and Land-Cover Types in North America

View More View Less
  • 1 Jackson School of Geosciences, Department of Geological Sciences, The University of Texas at Austin, Austin, Texas
  • | 2 Jackson School of Geosciences, Department of Geological Sciences, The University of Texas at Austin, Austin, Texas, and Key Laboratory of Regional Climate-Environment, Research for Temperate East Asia, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China
  • | 3 National Center for Atmospheric Research, Boulder, Colorado
  • | 4 Hydrological Sciences Laboratory, NASA Goddard Space Flight Center, Greenbelt, Maryland
© Get Permissions
Full access

Abstract

Continental-scale snow radiance assimilation (RA) experiments are conducted in order to improve snow estimates across snow and land-cover types in North America. In the experiments, the ensemble adjustment Kalman filter is applied and the Advanced Microwave Scanning Radiometer for Earth Observing System (AMSR-E) brightness temperature TB observations are assimilated into an RA system composed of the Community Land Model, version 4 (CLM4); radiative transfer models (RTMs); and the Data Assimilation Research Testbed (DART). The performance of two snowpack RTMs, the Dense Media Radiative Transfer–Multi-Layers model (DMRT-ML), and the Microwave Emission Model of Layered Snowpacks (MEMLS) in improving snow depth estimates through RA is compared. Continental-scale snow estimates are enhanced through RA by using AMSR-E TB at the 18.7- and 23.8-GHz channels [3% (DMRT-ML) and 2% (MEMLS) improvements compared to the cases using the 18.7- and 36.5-GHz channels] and by considering the vegetation single-scattering albedo ω [2.5% (DMRT-ML) and 4.8% (MEMLS) improvements compared to the cases neglecting ω]. The contribution of TB of the vegetation canopy to TB at the top of the atmosphere is better represented by considering ω in the RA system, and improvements in the resulting snow depth are evident for the forest land-cover type (about 5%–11%) and the taiga and alpine snow classes (about 5%–11% and 4%–8%, respectively), especially in the MEMLS case. Compared to the open-loop run (0.171-m snow depth RMSE), about 7% (DMRT-ML) and 10% (MEMLS) overall improvements of the RA performance are achieved.

© 2017 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author e-mail: Zong-Liang Yang, liang@jsg.utexas.edu

Abstract

Continental-scale snow radiance assimilation (RA) experiments are conducted in order to improve snow estimates across snow and land-cover types in North America. In the experiments, the ensemble adjustment Kalman filter is applied and the Advanced Microwave Scanning Radiometer for Earth Observing System (AMSR-E) brightness temperature TB observations are assimilated into an RA system composed of the Community Land Model, version 4 (CLM4); radiative transfer models (RTMs); and the Data Assimilation Research Testbed (DART). The performance of two snowpack RTMs, the Dense Media Radiative Transfer–Multi-Layers model (DMRT-ML), and the Microwave Emission Model of Layered Snowpacks (MEMLS) in improving snow depth estimates through RA is compared. Continental-scale snow estimates are enhanced through RA by using AMSR-E TB at the 18.7- and 23.8-GHz channels [3% (DMRT-ML) and 2% (MEMLS) improvements compared to the cases using the 18.7- and 36.5-GHz channels] and by considering the vegetation single-scattering albedo ω [2.5% (DMRT-ML) and 4.8% (MEMLS) improvements compared to the cases neglecting ω]. The contribution of TB of the vegetation canopy to TB at the top of the atmosphere is better represented by considering ω in the RA system, and improvements in the resulting snow depth are evident for the forest land-cover type (about 5%–11%) and the taiga and alpine snow classes (about 5%–11% and 4%–8%, respectively), especially in the MEMLS case. Compared to the open-loop run (0.171-m snow depth RMSE), about 7% (DMRT-ML) and 10% (MEMLS) overall improvements of the RA performance are achieved.

© 2017 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author e-mail: Zong-Liang Yang, liang@jsg.utexas.edu

1. Introduction

In a host of modeling and observational studies (e.g., Namias 1985; Vernekar et al. 1995; Cohen and Entekhabi 1999; Gong et al. 2003; Chapin et al. 2005), snow’s high albedo, low thermal conductivity, and water-holding capacity have been documented as critical factors affecting Earth’s climate system. In addition, spring and early summer snowmelt runoff is the principal source of water for humans and ecosystems in many of the mid- to high latitudes in the Northern Hemisphere (e.g., Stewart et al. 2004). The climate and hydrological research communities are therefore invested in improving the estimation of spatial and temporal variation in snowpack.

One approach to improving these estimates is the use of snow radiance data assimilation [hereafter, radiance assimilation (RA)] methods, in which microwave brightness temperature TB observations are directly assimilated into a land surface model (LSM). Previous studies have made significant progress in using this method through synthetic tests (e.g., Durand and Margulis 2006, 2007), point-scale applications (e.g., Durand et al. 2009; Toure et al. 2011), and basin-scale applications (e.g., Dechant and Moradkhani 2011). Further applications of the RA method have been presented by Andreadis and Lettenmaier (2012), Langlois et al. (2012), and Che et al. (2014). In particular, Andreadis and Lettenmaier (2012) discussed the impact of snow stratigraphy representation of the LSM on the performance of the RA scheme. Our recent study (i.e., Kwon et al. 2016) demonstrated that RA can enhance the continental-scale snowpack estimates. However, we also reported that large-scale applications of RA are impeded by the presence of various snow and land-cover types, especially dense forest land cover.

At the continental (or larger) scale, the physical properties of snow vary widely with local climate conditions (e.g., air temperature, precipitation, and wind speed). According to Sturm et al. (1995), seasonal snow cover can be classified into six classes (i.e., tundra, taiga, alpine, maritime, prairie, and ephemeral). Tedesco and Kim (2006) found considerably different TB values can be simulated for such a wide range of snow classes in their four radiative transfer models (RTMs; with three of these based on a single snow layer representation). In an RA scheme, an RTM is an observational operator predicting TB; therefore, the quality of the assimilation results may strongly depend upon the RTM used (Durand et al. 2008) as well as the LSM. Through synthetic experiments, Kwon et al. (2015) showed that two snowpack RTMs, that is, the Dense Media Radiative Transfer–Multi-Layers model (DMRT-ML; Picard et al. 2013) and the Microwave Emission Model of Layered Snowpacks (MEMLS; Wiesmann and Mätzler 1999) coupled with the Community Land Model, version 4 (CLM4; Oleson et al. 2010; Lawrence et al. 2011), have substantially different RA performance.

Meanwhile, vegetation canopy masks the microwave emission from the underlying surface and adds its own emission (Foster et al. 1991; Chang et al. 1996; Pampaloni 2004). In this regard, snow estimates for vegetated areas using microwave radiance observations by retrieval algorithms (e.g., Hall et al. 1982; Chang et al. 1996) or by radiance assimilation methods (e.g., Kwon et al. 2016) involve considerable uncertainties. In particular, the microwave TB exhibits a significantly lower sensitivity to snow in dense forest (Hallikainen and Jolma 1992; Roy et al. 2012). Therefore, precise estimates of the impact of the vegetation (i.e., vegetation emission and transmission) on TB at the top of the atmosphere (TOA) are essential for applications of the RA method in vegetated areas. Without such estimates, the vegetation canopy could result in an inaccurate modeling of the relationship between TB and snow [i.e., snow water equivalent (SWE) or snow depth] and, in turn, snow estimates via RA would be degraded as presented in Kwon et al. (2016).

Our objective in this study is to improve the performance of the snow RA system in estimating continental-scale snow water storage across snow and land-cover types. To this end, we attempt to address the following research questions:

  1. Which microwave frequency channels are most useful in improving snow estimates through RA?
  2. Which of the two snowpack RTMs (i.e., DMRT-ML and MEMLS) performs better for different snow cover types at the continental scale?
  3. Is vegetation single-scattering albedo ω an important parameter for estimating the effect of vegetation on TB at the TOA and improving the RA performance in estimating snow?

Section 2 describes the coupled radiance assimilation system used in this study. Datasets including atmospheric forcing, TB observations, and validation data are briefly described in section 3. The design of the radiance assimilation experiments is explained in section 4. The results are presented and discussed in section 5, and conclusions are drawn in section 6.

2. Coupled radiance assimilation system

In the coupled RA system (Fig. 1) employed in this study, we use CLM4 as an LSM and RTMs as observational operators. Data assimilation is implemented by the Data Assimilation Research Testbed (DART; Anderson et al. 2009), which is a community data assimilation system developed by the National Center for Atmospheric Research (NCAR) for ensemble-based data assimilation. We employ this particular coupled RA system to maintain continuity with our previous RA papers (i.e., Kwon et al. 2015, 2016).

Fig. 1.
Fig. 1.

Schematic diagram of the coupled RA system. Atmospheric ensemble forcing used by the atmospheric RTM includes atmospheric pressure, air temperature, and humidity. The CLM4 surface input data read by the RTMs are sand and clay fractions and vegetation fraction.

Citation: Journal of Hydrometeorology 18, 3; 10.1175/JHM-D-16-0102.1

CLM4 (Oleson et al. 2010; Lawrence et al. 2011) simulates snow dynamics for up to five snow layers, depending on the total snow depth. CLM4 can simulate snowmelt–refreeze cycles for each snow layer. It is also able to simulate snow densification by considering destructive and melting metamorphism and compaction by snow load. These make CLM4 suitable for this RA study. CLM4 estimates soil temperature for 15 soil layers, and it simulates hydrological processes over the top 10. The horizontal spatial heterogeneity of the land surface in CLM4 is represented by a nested subgrid hierarchy in which grid cells consist of land units, snow/soil columns, and plant functional types (PFTs). The land unit that has several snow/soil columns represents the broadest spatial patterns. The snow/soil column that has multiple PFTs is designed to describe potential variability in the soil and snow-related variables within a vegetated land unit. The PFT level includes bare ground and 16 PFTs [see Oleson et al. (2010) for more details on the PFTs] that have different physiology and structure.

The observational operator (i.e., coupled RTMs of snowpack, atmosphere, and vegetation) calculates TB at the TOA based on Durand and Margulis (2007), as follows:
e1
e2
e3
where is the brightness temperature at the TOA (K); tc and ta are the vegetation and atmospheric transmissivity, respectively; Vc is the vegetation fraction within a grid cell; is the brightness temperature from the snowpack (K), which is estimated by DMRT-ML or MEMLS and includes the effect of the underlying soil; is the brightness temperature emitted by the vegetation canopy (K); Tc is the vegetation physical temperature (K); is the atmospheric brightness temperature (K); is the boundary condition TB for snowpack RTMs to model ; and is the space brightness temperature (2.7 K).

Two snowpack RTMs, that is, DMRT-ML and MEMLS, are used in this study while Kwon et al. (2016) used only DMRT-ML to calculate TB from the snowpack. These two RTMs were chosen because of the following reasons. First, they estimate TB using a multilayered snowpack as emphasized by many previous studies (e.g., Mätzler et al. 1984; Rosenfeld and Grody 2000; Brucker et al. 2010; Champollion et al. 2013). Second, they are based on two different theories. DMRT-ML is based on the Dense Media Radiative Transfer (DMRT) theory (Tsang and Kong 2001) while MEMLS is based on the six-flux model [see Wiesmann and Mätzler (1999) for more details]. Last, they have been widely tested and applied in TB estimation or assimilation studies [e.g., Brucker et al. (2011), Roy et al. (2013), and Kwon et al. (2016) for DMRT-ML; Durand et al. (2008, 2009) and Toure et al. (2011) for MEMLS; and Kwon et al. (2015) and Löwe and Picard (2015) for both].

In DMRT-ML, the microwave absorption and scattering coefficients of snow are calculated using the DMRT theory and the microwave emission and propagation are estimated using the Discrete Ordinate Radiative Transfer model (DISORT; Jin 1994), in which multiple scattering between the layers is considered. Meanwhile, the extinction coefficient in MEMLS is calculated based on the improved Born approximation (Mätzler 1998). MEMLS estimates multiple volume scattering, absorption, and propagation of the microwave signal using the six-flux theory. To calculate TB from the snowpack, both RTMs require snow inputs such as snow layer thickness, density, temperature, wetness, and grain radius, all of which are provided by CLM4 in the coupled RA system. The reflectivity of the underlying soil is calculated by the rough bare soil reflectivity model (Wegmüller and Mätzler 1999) using the estimated soil temperature and soil water content from CLM4.

The advantage of using DMRT-ML as an observational operator in our RA system is that DMRT-ML is a physically based model and it represents the snow grain size using an effective snow grain radius as in CLM4. Although the effective snow grain radius used in DMRT-ML is not exactly the same as defined in CLM4 (Brucker et al. 2011; Roy et al. 2013), DMRT-ML includes a tunable stickiness parameter, which is related to the size of the scatterers and can be used to mitigate this discrepancy (e.g., Kwon et al. 2015). However, continental-scale optimization of the stickiness parameter is difficult, and thus it was updated during assimilation as suggested by Kwon et al. (2016) (see Table 1 for a list of the updated states/parameters). Initial values of the stickiness for each grid cell and each ensemble member were randomly created using a range of 0.1–0.5, which was determined based on previous studies (e.g., Mätzler 1998; Tsang et al. 2008; Andreadis and Lettenmaier 2012; Picard et al. 2013; Kwon et al. 2015).

Table 1.

List of the CLM4 states and RTM parameters updated during the assimilation. The vegetation single-scattering albedo was updated only in the cases specified in Table 2.

Table 1.

Unlike DMRT-ML and CLM4, MEMLS employs an exponential correlation length for the snow grain size representation. Therefore, for MEMLS, simulated CLM4 snow grain radius is converted to the exponential correlation length using the conversion equation suggested in Kwon et al. (2015) and based on the work by Mätzler (2002) and Debye et al. (1957):
e4
where pex is the exponential correlation length (m), re is the effective grain radius (m), ρ is the snow density (kg m−3), and ρice is the ice density (917 kg m−3). MEMLS is advantageous for large-scale RA studies because the use of MEMLS in the RA system is computationally efficient compared to using DMRT-ML.
The effect of the atmosphere on TB is estimated following Ulaby et al. (1981). The vegetation transmissivity and optical depth τ are estimated using Eq. (5) (Schmugge and Jackson 1992) and Eq. (6) (Jackson and Schmugge 1991), respectively:
e5
e6
where λ is the wavelength (cm); b′ and x are the empirical parameters, which depend upon the vegetation canopy type and are updated in the RA system (see Table 1); wc is the vegetation water content (kg m−2), which is calculated based on Paloscia and Pampaloni (1988) using the leaf area index (LAI); and θ is the incident angle. The vegetation fraction (i.e., Vc) and LAI required by the observational operator to estimate TB were provided by CLM4. As a compromise between the computational efficiency and representation of the spatial heterogeneity, TB was estimated on a column basis and averaged (area weighted) for a CLM4 grid cell. Therefore, LAI (only the exposed LAI, not buried by snow, was considered) and Vc, averaged and summed across all PFTs in a column, respectively, were used in estimating TB.

Among a variety of ensemble-based assimilation algorithms available in DART, we use the ensemble adjustment Kalman filter (EAKF) [see Anderson (2001) for more detailed explanations of the EAKF], which is a deterministic ensemble square root filter (Tippett et al. 2003) and does not need randomly perturbed observations. Anderson (2001) reports that in the cases compared in the paper, the performance of the EAKF was much better than that of the traditional ensemble Kalman filter (EnKF; Evensen 1994; Burgers et al. 1998), especially for a small ensemble size.

3. Datasets

a. DART–CAM4 atmospheric ensemble reanalysis

Raeder et al. (2012) produced an 80-member ensemble atmospheric reanalysis, which includes air temperature, atmospheric pressure, precipitation, humidity, wind speed, and downward shortwave radiation, using the coupled DART–Community Atmosphere Model, version 4 (CAM4; Gent et al. 2011). Among them, we used the 40 randomly chosen ensemble members to drive the model. The advantage of using the DART–CAM4 atmospheric ensemble reanalysis, rather than manually perturbing each atmospheric forcing field, is that the physical consistency (i.e., cross correlation) between forcing fields can be maintained. CLM4 driven by the DART–CAM4 forcing tends to overestimate snow depth (Kwon et al. 2016) because of systematic biases in the atmospheric forcing, in particular the estimated precipitation. However, in this study, we mainly focused on answering our research questions in applying the RA method to the continental-scale snow estimation, given the biased forcing, and thus bias correction of the atmospheric forcing was not conducted.

b. AMSR-E observations

In this study, we assimilated the Advanced Microwave Scanning Radiometer for Earth Observing System (AMSR-E)/Aqua daily global quarter-degree gridded brightness temperatures data (Knowles et al. 2006a,b) into the coupled RA system. AMSR-E observes vertically (V)- and horizontally (H)-polarized TOA microwave radiances at six frequencies (i.e., 6.925, 10.65, 18.7, 23.8, 36.5, and 89.0 GHz). However, only vertically polarized TB observations were assimilated because horizontally polarized TB is significantly influenced by ice layers within the snowpack (Mätzler 1987; Durand et al. 2008; Rees et al. 2010), which are not accurately represented in the five-layer snow model. In addition, only nighttime observations were used, thereby minimizing error due to snow wetness.

It has been reported that the total bias error of the AMSR-E sensor ranges from 0.66 K at 100 K to 0.68 K at 250 K (Lobl 2001). However, AMSR-E TB data products used in this study may have more error resulting from two sources (see http://nsidc.org/data/docs/daac/ae_l2a_tbs.gd.html#errorsource): 1) a mismatch between the constructed (level 2A) antenna pattern and the ideal antenna pattern, which is related to the fit error when the original observations are spatially resampled, and 2) random measurement error. In addition, the original AMSR-E TB data, which have 0.25° spatial resolution, were scaled up to the CLM4 grid (0.9° × 1.25°) for computational efficiency. This spatial averaging over several adjacent grid cells would smooth out random measurement errors from individual grid cells, but it may also introduce additional uncertainties into the TB observations by further simplifying heterogeneous grid cells. Therefore, to consider these additional error sources, the observation error was set to 2 K, as assumed by Durand and Margulis (2007). However, quantification of the AMSR-E TB observational error needs to be further investigated.

c. CMC snow depth data

The assimilation results were evaluated using the Canadian Meteorological Centre (CMC) daily snow depth data (Brasnett 1999; Brown and Brasnett 2010). The CMC analysis data are produced using snow depth data from surface synoptic observations and meteorological and special aviation reports acquired from the World Meteorological Organization (WMO) information system (Brown and Brasnett 2010). Although the CMC product has some deficiencies such as early snowmelt in the spring (Brown et al. 2010; Toure et al. 2016) and fewer in situ observations at higher latitudes (Reichle et al. 2011), it is a spatially complete dataset of daily snow depth for the Northern Hemisphere (Reichle et al. 2011) and is thus considered one of the best available snow references for large-scale model evaluations (e.g., Su et al. 2010; Reichle et al. 2011; Forman et al. 2012; Zhang et al. 2014; Toure et al. 2016; Kwon et al. 2016).

4. Experimental design

The experiments are designed to address our three research questions related to the microwave frequency channels, snowpack RTMs, and vegetation single-scattering albedo. In all experimental cases (Table 2) including the open-loop run (without assimilation), CLM4 was run at 0.9° × 1.25° spatial resolution forced by the DART–CAM4 atmospheric ensemble reanalysis. The experiments were conducted for North America from December 2002 to February 2003.

Table 2.

RA experimental cases with respect to snowpack RTMs, frequency channels, and single-scattering albedo.

Table 2.

As shown in Table 1, the physical states and parameters updated during the assimilation include SWE, snow grain radius, snow temperature, soil temperature, soil water content, snow stickiness (in DMRT-ML), and two empirical parameters [x and b′ in Eq. (6)] of the vegetation RTM. The single-scattering albedo was also updated in some experimental cases. As suggested in Zhang et al. (2014) and Kwon et al. (2016), among the snow-mass-related states in CLM4 [i.e., the mass of snow liquid water and ice (kg m−2), SWE (kg m−2), snow density (kg m−3), and snow depth (m)], only SWE was updated during the assimilation procedure to avoid excessive snow mass updating. Snow depth and the mass of snow liquid water and ice were adjusted according to their physical relationships with SWE, as defined in CLM4. Snow density was not updated in the RA system because it is simply calculated from SWE and snow depth in CLM4.

Kwon et al. (2016) demonstrated two approaches that are effective in improving the continental-scale snow estimates: 1) simultaneous updates of all model physical states and parameters involved in predicting TB and 2) a rule-based approach in which the states and parameters are updated only when the signs of their correlations with the prior TB coincide with those of sensitivity of the estimated TB to the states and parameters. In many cases, the uncertainty of the estimated TB can be primarily determined by the uncertainties of other snow and soil properties, especially snow grain size, instead of SWE (Kwon et al. 2015). This results in undesirable updates of SWE in the RA system. Kwon et al. (2016) hypothesized that only when the sign of the correlation between the forecasted states/parameters and the predicted TB is the same as the sign of the sensitivity index, the TB difference (between simulations and observations) will provide meaningful information for each of the priors. Therefore, for example, SWE was not updated if the correlation between the estimated SWE and TB within the ensemble was positive because TB was negatively sensitive to SWE (Kwon et al. 2016). These two approaches were simultaneously applied in the RA cases. It should be noted that the assimilation was not performed if any ensemble members within a grid cell predicted no snow.

A localization distance parameter in DART restricts the effect of assimilated observations on model states of nearby grid cells (Anderson et al. 2009). This mitigates the degradation of assimilation performance resulting from correlations between two uncorrelated variables due to sampling error (Anderson 2007). According to Kwon et al. (2016), the localization distance of 0.01 radians (0.6°) was used in all RA experimental cases.

a. Microwave frequency channels

Previous snow RA studies mostly used passive microwave observations at 18.7 and/or 36.5 GHz (e.g., Durand et al. 2009; Dechant and Moradkhani 2011; Toure et al. 2011; Bateni et al. 2015; Kwon et al. 2016). However, through a synthetic test, Durand and Margulis (2006) suggested that all frequency channels provide valuable information for snowpack. In this paper, we compared the RA performance of the experimental cases (Table 2) using different AMSR-E frequency channels. The 6.925-GHz TB was excluded from the experiments because of the high potential for radio frequency interference over the United States (Njoku et al. 2005).

b. Snowpack radiative transfer models

We employed two snowpack RTMs (DMRT-ML and MEMLS) to estimate the snowpack TB. Because of the different TB sensitivities of DMRT-ML and MEMLS to the physical properties of snow and underlying soil, the correlations between the SWE error (simulation minus observation) and TB error produced by coupled CLM4–DMRT-ML and CLM4–MEMLS also exhibit considerable differences (Kwon et al. 2015). RA using ensemble Kalman–based data assimilation methods assumes that the SWE error is correlated with the TB error (Kwon et al. 2015). Therefore, two coupled models (CLM4–DMRT-ML and CLM4–MEMLS) may show varying RA performance for different snow cover types. Kwon et al. (2016) showed that the RA system using DMRT-ML significantly improves the snow depth estimates for the tundra and maritime snow classes. Here, we compared the performance of the RA system using DMRT-ML and MEMLS for various snow and land-cover types (see Table 2 for the experimental cases).

c. Vegetation single-scattering albedo

The most commonly used vegetation RTM is the τω model (Mo et al. 1982), which is a simplified approach to model the vegetation effect on TB. Two parameters, namely, the vegetation optical depth and the single-scattering albedo, are involved in the model. The single-scattering albedo parameterizes all processes within the vegetation canopy layer, such as the multiple scattering (Kurum et al. 2012). In many cases, ω is neglected (e.g., Kruopis et al. 1999; Langlois et al. 2011) because previous studies have reported that it is generally less than 0.1 (Pampaloni and Paloscia 1986; Grant et al. 2008; Roy et al. 2012).

Our coupled RA system, which uses the approach of Durand and Margulis (2007), initially neglects ω in its estimate of TB at the TOA [Eq. (2)]. However, this assumption of ω = 0 results in an overestimation of the microwave emission of the vegetation canopy (Ferrazzoli et al. 2002). Therefore, the contribution of ω was additionally considered in representing TB as follows:
e7

First, the improvement of the RA performance by considering ω was tested using a constant value of ω = 0.064, as suggested in Roy et al. (2012) [see the D1823(ω), D1836(ω), M1823(ω), and M1836(ω) cases in Table 2]. However, the ω value depends on vegetation type, frequency, and polarization (Langlois et al. 2011). Although Roy et al. (2012) suggested that ω is mostly frequency and polarization independent for coarse-scale observations, it is still influenced by vegetation properties. Yet, no measurements of ω are available at the continental scale and thus an optimization procedure for ω is required. Several studies (e.g., Paloscia 1995; Njoku and Li 1999; Pellarin et al. 2006; Grant et al. 2008; Roy et al. 2012) have been published on the optimization of ω values for different vegetation types, but they are not in agreement. This is attributed to the fact that surface parameterizations (e.g., snow and/or soil), datasets, and assumptions can influence the optimization results (Roy et al. 2012; Pellarin et al. 2006). Fortunately, however, most of the ω values suggested by these studies are within the range of 0.05–0.1. Therefore, in the D1823(ω_up), D1836(ω_up), M1823(ω_up), and M1836(ω_up) cases (Table 2), ω was updated in the RA system using the range of 0.05–0.1. If the updated ω fell outside the specified range, it was set to the closest value within the range. Based on the sensitivity analysis (not shown here), TB at the TOA is negatively sensitive to ω; that is, an increase in ω leads to a decrease in TB at the TOA. Using the rule-based approach suggested by Kwon et al. (2016), the update of ω was performed only if the correlation between the prior TB and ω was negative.

5. Results and discussion

We presented our results separately according to the six snow classes (tundra, taiga, alpine, maritime, prairie, and ephemeral) of Sturm et al. (1995) (Fig. 2a) and the five dominant land-cover types (bare soil, forest, shrub, grass, and crop; Fig. 2b), which were determined by classifying 17 PFTs (including bare soil), defined in the CLM4 surface input dataset, into the five land-cover types and then by comparing the percentage of each land-cover type within a grid cell.

Fig. 2.
Fig. 2.

(a) Snow and (b) land-cover types in North America: snow class are 1) tundra, 2) taiga, 3) maritime, 4) ephemeral, 5) prairie, and 6) alpine; land-cover types are 1) bare soil, 2) forest, 3) shrub, 4) grass, and 5) crop.

Citation: Journal of Hydrometeorology 18, 3; 10.1175/JHM-D-16-0102.1

a. Performance of the RA system using AMSR-E frequencies and snowpack RTMs

The TB observation at each AMSR-E frequency channel (10.65, 18.7, 23.8, 36.5, and 89.0 GHz vertical polarization) was separately assimilated into the RA system, and the resulting snow depth root-mean-square errors (RMSEs) of the RA experimental cases using DMRT-ML and MEMLS for North America are shown in Fig. 3. Among the five frequency channels assimilated, only the 18.7- and 23.8-GHz channels led to an overall improvement in snow depth estimates in both DMRT-ML and MEMLS cases. The performance of the DMRT-ML cases (i.e., 3.1% and 3.6% improvements in D18 and D23, respectively) was slightly better than that of the MEMLS cases (i.e., 1.8% and 2.5% improvements in M18 and M23, respectively). Compared to the open-loop run without assimilation (the horizontal dotted line in Fig. 3), the performance of the RA system was marginally improved in the D36 case, whereas it was degraded in the M36 case. The 23.8-GHz channel is more sensitive to water vapor than the 18.7- and 36.5-GHz channels (Pulliainen et al. 1993; Turner et al. 2007) and thus has not been used in previous snow RA studies, except for the synthetic experiments in Durand and Margulis (2006). Our results imply the feasibility of using the 23.8-GHz channel in the RA system with modeling of the atmospheric effect.

Fig. 3.
Fig. 3.

The snow depth RMSE (m) in the RA experimental cases, using different snowpack RTMs (DMRT-ML and MEMLS) and assimilating different AMSR-E frequency channels, for North America. The horizontal dotted line represents the RMSE of the open-loop run (without assimilation).

Citation: Journal of Hydrometeorology 18, 3; 10.1175/JHM-D-16-0102.1

Meanwhile, the results obtained here contradict the synthetic experiment results of Durand and Margulis (2006) in which the 10.65- and 36.5-GHz channels contributed more than the 18.7- and 23.8-GHz channels in improving the snow estimates in their RA system. There are four potential reasons for our different results. First, RA methods applied in the real world might produce different results than when used in synthetic experiments. Second, different results could be due to differences among RA schemes in terms of LSM, RTM, updated states/parameters during the assimilation, or assimilation methods employed (e.g., EnKF and EAKF). Third, a rule-based approach (Kwon et al. 2016) was applied in our RA experiments but was not used by Durand and Margulis (2006). The use of rule-based approach probably influenced the relative contribution of different frequency channels to the RA performance. Finally, our experiments were conducted for shallow snow (~1 m snow depth), whereas Durand and Margulis (2006) worked with deep snow conditions (~3 m snow depth).

We additionally analyzed the effect of assimilating two frequency channels simultaneously on the RA performance in estimating snow depth (Fig. 3). Because the 23.8- and 18.7-GHz channels were the best- and next best–performing frequency channels, respectively, these two were assimilated in the D1823 and M1823 cases. For comparison purposes, TB observations at 18.7- and 36.5-GHz channels were also simultaneously assimilated in the D1836 and M1836 cases because these two channels have been frequently used in snow RA systems as well as in snow retrieval algorithms.

As shown in Fig. 3, except for the D1836 case, the assimilation of two frequency channels improved the overall performance of the RA system as compared to the separate use of each frequency channel, especially in MEMLS cases. The D1836 case performed better than D36 but worse than D18. One plausible reason is the effect of the stickiness parameter in DMRT-ML. Because the stickiness was updated during the assimilation, a different combination of assimilated microwave frequency channels may lead to slightly varying stickiness values. Therefore, the stickiness determined by assimilating both the 18.7- and 36.5-GHz channels could possibly result in the degradation of the RA performance compared to the case assimilating only the 18.7-GHz channel. It should be noted that compared to the 1836 cases, slightly improved estimates of snow depth were obtained in the 1823 cases for both DMRT-ML and MEMLS (i.e., 3% and 2% decrease in the RMSE, respectively). The period of our experiments only covered the snow accumulation season. Therefore, note that including the snow-melting period may influence the relative performance of each frequency channel in the RA system due to wet snow conditions.

Assimilating the 18.7- and 23.8-GHz channels without considering the vegetation single-scattering albedo, DMRT-ML cases produced about 5.2% better estimates of snow than MEMLS cases for the alpine snow class while MEMLS cases were more effective for tundra (about 4.7% better performance; Fig. 4). DMRT-ML and MEMLS cases showed comparable performance in estimating snow depth for the taiga, maritime, ephemeral, and prairie snow classes, that is, about 2% increase, 3% decrease, 46% increase, and 4% increase in the RMSE, respectively. Compared to the open-loop run, however, DMRT-ML and MEMLS cases only improved snow depth estimates for the tundra (14.1% and 18.2% decrease in the RMSE, respectively) and maritime snow classes (3% decrease in the RMSE for both). While the snow depth RMSE of the open-loop run for the ephemeral and prairie snow classes was already very small and RA marginally increased the RMSE, the degradation of the snow depth estimates by RA was noticeable for the taiga and alpine snow classes. The poor performance of the RA system for the taiga and alpine snow classes could be due to the effect of vegetation (especially forest) on TB estimations. The difficulty of characterizing snowpack under the vegetation canopy using microwave radiance observations has been reported by many previous studies (e.g., Hallikainen and Jolma 1992; Chang et al. 1996; Foster et al. 2005). Figure 5 shows that regardless of the snowpack RTMs and microwave frequency channels, the RA system could not improve the snow depth estimates for forest land cover, the dominant land-cover type for the taiga and alpine snow classes (Fig. 2). This implies that the effect of the vegetation canopy on TB at the TOA was not accurately represented in our current RA system. Further improvement of the performance of the RA system for vegetated areas by introducing ω is discussed in the next section.

Fig. 4.
Fig. 4.

The snow depth RMSE (m) in the RA experimental cases, using different snowpack RTMs (DMRT-ML and MEMLS) and assimilating different AMSR-E frequency channels, for snow classes as defined in Sturm et al. (1995): (a) tundra, (b) taiga, (c) maritime, (d) ephemeral, (e) prairie, and (f) alpine. The horizontal dotted line represents the RMSE of the open-loop run (without assimilation).

Citation: Journal of Hydrometeorology 18, 3; 10.1175/JHM-D-16-0102.1

Fig. 5.
Fig. 5.

The snow depth RMSE (m) in the RA experimental cases, using different snowpack RTMs (DMRT-ML and MEMLS) and assimilating different AMSR-E frequency channels, for land-cover types: (a) bare soil, (b) forest, (c) shrub, (d) grass, and (e) crop. The horizontal dotted line represents the RMSE of the open-loop run (without assimilation).

Citation: Journal of Hydrometeorology 18, 3; 10.1175/JHM-D-16-0102.1

b. The impact of the vegetation single-scattering albedo on the RA performance

To analyze the effect of ω on the estimation of TB emitted by vegetation and on the performance of the RA system, two approaches were compared: 1) the use of the constant ω value (0.064) obtained in Roy et al. (2012) and 2) the use of an ω updated during the assimilation. The experimental results for North America, for six snow classes, and for five land-cover types are presented in Figs. 6, 7, and 8, respectively.

Fig. 6.
Fig. 6.

The snow depth RMSE (m) in the RA experimental cases, using different snowpack RTMs (DMRT-ML and MEMLS) and assimilating two frequency channels simultaneously (i.e., 18.7 and 23.8 GHz or 18.7 and 36.5 GHz) for North America. The vegetation single-scattering albedo was neglected (in the D1823, D1836, M1823, and M1836 cases), was set to 0.064 [in the D1823(ω), D1836(ω), M1823(ω), and M1836(ω) cases], or was updated during the assimilation [in the D1823(ω_up), D1836(ω_up), M1823(ω_up), and M1836(ω_up) cases]. The horizontal dotted line represents the RMSE of the open-loop run (without assimilation).

Citation: Journal of Hydrometeorology 18, 3; 10.1175/JHM-D-16-0102.1

Fig. 7.
Fig. 7.

The snow depth RMSE (m) in the RA experimental cases, using different snowpack RTMs (DMRT-ML and MEMLS) and assimilating two frequency channels simultaneously (i.e., 18.7 and 23.8 GHz or 18.7 and 36.5 GHz) for six snow classes as defined in Sturm et al. (1995): (a) tundra, (b) taiga, (c) maritime, (d) ephemeral, (e) prairie, and (f) alpine. The vegetation single-scattering albedo was neglected (in the D1823, D1836, M1823, and M1836 cases), was set to 0.064 [in the D1823(ω), D1836(ω), M1823(ω), and M1836(ω) cases], or was updated during the assimilation [in the D1823(ω_up), D1836(ω_up), M1823(ω_up), and M1836(ω_up) cases]. The horizontal dotted line represents the RMSE of the open-loop run (without assimilation).

Citation: Journal of Hydrometeorology 18, 3; 10.1175/JHM-D-16-0102.1

Fig. 8.
Fig. 8.

The snow depth RMSE (m) in the RA experimental cases, using different snowpack RTMs (DMRT-ML and MEMLS) and assimilating two frequency channels simultaneously (i.e., 18.7 and 23.8 GHz or 18.7 and 36.5 GHz) for five land-cover types: (a) bare soil, (b) forest, (c) shrub, (d) grass, and (e) crop. The vegetation single-scattering albedo was neglected (in the D1823, D1836, M1823, and M1836 cases), was set to 0.064 [in the D1823(ω), D1836(ω), M1823(ω), and M1836(ω) cases], or was updated during the assimilation [in the D1823(ω_up), D1836(ω_up), M1823(ω_up), and M1836(ω_up) cases]. The horizontal dotted line represents the RMSE of the open-loop run (without assimilation).

Citation: Journal of Hydrometeorology 18, 3; 10.1175/JHM-D-16-0102.1

By considering the constant ω, the overall snow depth RMSE for North America was reduced (Fig. 6). The most considerable improvement was achieved for forest land cover (about 5%–11% decrease in the RMSE; Fig. 8), and accordingly, the snow estimates for the taiga and alpine snow classes were also enhanced (about 5%–11% and 4%–8% decrease in the RMSE, respectively) compared to the RA cases neglecting ω (Fig. 7). For other vegetated areas except crop land cover, only negligible improvement (less than 2%) was observed (Fig. 8). Improvements in performance resulting from the use of ω were more noticeable in MEMLS cases [M1823(ω) and M1836(ω)] than in DMRT-ML cases [D1823(ω) and D1836(ω)], especially for the forest land-cover type and the taiga snow class (Figs. 7, 8). The overall performance of the DMRT-ML case [D1823(ω)] was worse than the MEMLS case [M1823(ω)] (Fig. 6). The performance of the cases assimilating the 18.7- and 36.5-GHz channels was slightly better than that of the cases assimilating the 18.7- and 23.8-GHz channels for crop land cover (Fig. 8e). This is likely because the 36.5-GHz channel exhibited better performance in estimating snow depth than the 23.8-GHz channel for this land-cover type (Fig. 5e). Our RA experimental results show that substantial improvements to snow estimates made through RA are achieved by taking ω into account even when the ω value is less than 0.1. This suggests that ω should be considered in representing the effect of vegetation on TB at the TOA from the RA perspective.

Updating ω during the assimilation in the RA system did not improve the RA performance beyond what was already achieved using a constant ω (Figs. 68). Rather, the snow depth RMSE slightly increased in the D1823(ω_up) and M1823(ω_up) cases. Nonetheless, the performance of the RA cases using an updated ω was still superior to that of the cases neglecting ω.

Figure 9 shows that the vegetation transmissivity estimated in the RA experimental cases [D1823, M1823, D1823(ω), and M1823(ω)] ranges from 0.4 to 0.6 for forested areas. The saturation levels of the vegetation transmissivity obtained by Langlois et al. (2011) for the Canadian boreal forest are 0.51 and 0.55 for the 19-GHz-V and 19-GHz-H channels, respectively, and 0.53 for both the 37-GHz-V and 37-GHz-H channels. Roy et al. (2012) suggested that the vegetation transmissivities for dense boreal forests are 0.69, 0.62, 0.497, and 0.423 for 6.9-, 10.7-, 18.7-, and 36.5-GHz channels, respectively. Based on the values suggested in previous studies, the range of vegetation transmissivity we obtained for forest land cover is reasonable.

Fig. 9.
Fig. 9.

The estimated (ensemble mean) vegetation transmissivity in the (a) D1823, (b) M1823, (c) D1823(ω), and (d) M1823(ω) cases. The values were averaged over two frequency channels (18.7- and 23.8-GHz vertical polarizations) during the assimilation period.

Citation: Journal of Hydrometeorology 18, 3; 10.1175/JHM-D-16-0102.1

However, in the D1823 and M1823 cases, about 60%–70% of TB at the TOA was contributed by TB emission from the vegetation canopy in areas dominated by forest land cover (Figs. 10a,b). Compared to the results by Roy et al. (2012) in which constituted about 46% and 50% of at 18.7 and 36.5 GHz-V, respectively, the contribution of was overestimated in our experiments by neglecting ω. By considering ω in the D1823(ω) and M1823(ω) cases, more reasonable estimates of the vegetation contribution were obtained (Figs. 10c,d). The changes [the 1823(ω) cases minus the 1823 cases] in the estimated ratio for forested areas by considering ω ranged from −5% to −15% (Figs. 10e,f). A better estimate of the contribution was observed in the MEMLS case than in the DMRT-ML case (Fig. 10). This could explain why the reduction in the snow depth RMSE associated with ω was greater in MEMLS cases than in DMRT-ML cases for forest land cover (Fig. 8b).

Fig. 10.
Fig. 10.

The ratio of the estimated (ensemble mean) to the estimated (ensemble mean) in the cases neglecting ω [(a) D1823 and (b) M1823] and in the cases considering ω [(c) D1823(ω) and (d) M1823(ω)]. The values were averaged over two frequency channels (18.7- and 23.8-GHz vertical polarizations) during the assimilation period. (e),(f) Changes in the estimated ratio by considering ω.

Citation: Journal of Hydrometeorology 18, 3; 10.1175/JHM-D-16-0102.1

While the RA performance was enhanced for vegetated areas by considering ω (Figs. 8, 11a,b), the snow depth estimates by the RA system were still worse than the open-loop run for the ephemeral, prairie, and alpine snow classes, although the degradation was marginal and insignificant (p > 0.05; Figs. 7, 11c,d). This may be partly attributed to the following three reasons (the first and second reasons may only be applicable to the ephemeral and prairie snow classes). First, the effect of the underlying soil conditions on TB for shallow snowpack would not be accurately estimated and thus RA could not further improve the snow estimates compared to the open-loop run for the ephemeral and prairie snow classes where the open-loop run already showed good performance (the snow depth RMSE was 0.027 and 0.045 m, respectively). Second, our approach, in which the priors were not updated when one or more of the ensemble members predicted no snow, might affect the results, especially for the ephemeral snow class. Third, the effect of topography on snow and TB simulations was not explicitly considered in this study because of the use of the coarse CLM4 spatial resolution (0.9° × 1.25°), in particular for the alpine snow class. All these possible sources of degradation need to be further explored.

Fig. 11.
Fig. 11.

Spatial distributions of the snow depth RMSE difference: (a) D1823(ω) − D1823, (b) M1823(ω) − M1823, (c) D1823(ω) − open loop, and (d) M1823(ω) − open loop. Negative and positive values denote the improvement and degradation of the RA performance, respectively.

Citation: Journal of Hydrometeorology 18, 3; 10.1175/JHM-D-16-0102.1

c. Snow water volume and snow cover area in North America

Snow water volume was calculated by multiplying the estimated (ensemble mean) SWE by grid area and summing the result over North America (Fig. 12a). Compared to the open-loop run, the snow water volume was reduced in the RA cases throughout the assimilation period (i.e., the snow accumulation season). The difference in the snow water volume between the RA cases and open-loop run increased with time and approached 110 and 132 km3 for the D1823(ω) and M1823(ω) cases, respectively, by the end of the assimilation period. This represented 7.8% and 9.5% of the snow water volume from the open-loop run, respectively. As shown in Fig. 12a, the snow water volume estimated by the RA system was closer to the CMC data than that estimated by the open-loop run. However, the magnitude of the improvement was less than half of the difference between our estimates and the CMC data.

Fig. 12.
Fig. 12.

The estimated (ensemble mean) (a) snow water volume (103 km3) and (b) snow cover area (106 km2) in North America (asterisk indicates the difference was divided by the value of the open-loop run). The CMC snow water volume was calculated using the SWE estimates obtained by Reichle et al. (2011) from the CMC snow depth and climatological snow densities suggested by Sturm et al. (2010).

Citation: Journal of Hydrometeorology 18, 3; 10.1175/JHM-D-16-0102.1

In CLM4, snow cover fraction (SCF) is a diagnostic variable and is estimated from snow density and depth using the snow cover parameterization suggested by Niu and Yang (2007). The ensemble mean snow cover area in North America was calculated from SCF estimated by CLM4 (Fig. 12b). For comparison purposes, the snow cover area calculated from the Moderate Resolution Imaging Spectroradiometer (MODIS) SCF (Hall et al. 2006) was also presented in Fig. 12b. Compared to the open-loop run, the snow cover area also decreased in the RA cases but the magnitude of the decrease was very small. The error (simulation minus MODIS observation) was much greater than the change (improvement). The maximum difference in the snow cover area between the RA cases and open-loop run was 68 969 km2 for the D1823(ω) case and 122 998 km2 for the M1823(ω) case, corresponding to 0.5% and 0.8% of the snow cover area from the open-loop run, respectively. The insignificant changes in the snow cover area through RA may be attributed to the fact that RA was primarily effective in improving snow depth (and SWE) estimates for relatively deep snowpack regions where SCF was already saturated (i.e., 100% SCF). As previously mentioned, our current RA system leaves the priors unchanged when one or more of the ensemble members have no snow, which would also affect the results for areas near the snow/no-snow boundary where SCF values are small.

6. Conclusions

In this study, three research questions related to microwave frequency channels, snowpack RTMs, and vegetation single-scattering albedo ω were addressed for the purpose of improving the snow RA performance across snow classes and land-cover types in North America. Our coupled RA system employed CLM4 (for snow energy and mass balance modeling), RTMs (for TB estimates), and DART (for ensemble-based data assimilation). Two different snowpack RTMs (DMRT-ML and MEMLS) were compared in terms of their relative performance in improving snow depth estimates through RA. RA experiments were conducted during the snow accumulation season (from December 2002 to February 2003) by assimilating AMSR-E TB observations using the EAKF. The experimental results were assessed for six snow classes (tundra, taiga, alpine, maritime, prairie, and ephemeral) and five land-cover types (bare soil, forest, shrub, grass, and crop) using the CMC snow depth data.

The results showed that the 23.8- and 18.7-GHz channels are the best- and next best–performing frequency channels, respectively, in our RA system. We obtained a significant improvement in snow depth estimates, especially in MEMLS cases, by assimilating the two best-performing frequency channels (18.7 and 23.8 GHz). Simultaneous assimilation of the 18.7- and 36.5-GHz channels also improved the overall performance of the RA system compared to the open-loop run, in particular, when ω was taken into account, although it was still slightly less effective than the use of the 18.7- and 23.8-GHz channels. It should also be noted that the 1836 cases using MEMLS were superior to the 1823 cases using DMRT-ML when ω was considered.

By introducing ω, the contribution of vegetation TB emission to TB at the TOA was more reasonably represented in the RA system. Consequently, substantial improvements in the RA performance were achieved for vegetated areas, in particular for the forest land-cover type and the taiga and alpine snow classes. However, note that the performance of the RA system was still degraded for the alpine snow class compared to the open-loop run. Although we could not further improve the RA performance by updating ω during the assimilation, the results suggested that from the RA perspective, ω is an essential factor in the RA system for characterizing snow under the vegetation canopy. By establishing the ω values for various land-cover types (or vegetation types), the performance of snow RA will be further enhanced.

When ω was neglected, the DMRT-ML cases were superior to the MEMLS cases for the alpine snow class while the MEMLS cases produced better snow estimates for the tundra snow class. When ω was considered, improvement of the RA performance was more noticeable in the MEMLS cases than in the DMRT-ML cases, and the MEMLS cases outperformed the DMRT-ML cases for the taiga snow class as well as for the tundra snow class.

In addition to our previous study (Kwon et al. 2016), which introduced and demonstrated the rule-based approach, this study showed that the performance of the RA system in estimating snow depth over North America can be enhanced by using the best-performing frequency channels (i.e., 18.7 and 23.8 GHz) [3% (in D1823) and 2% (in M1823) improvements compared to the cases using the 18.7- and 36.5-GHz channels] and by considering ω {2.5% [in D1823(ω)] and 4.8% [in M1823(ω)] improvements compared to the 1823 cases neglecting ω} (Fig. 13a). In particular, the RA system performed much better than the open-loop run for areas without vegetation cover (Fig. 13b). The snow depth estimates by RA were also enhanced for forested areas (Fig. 13c) by better representing the vegetation contribution to TB at the TOA. However, the overall performance (150–160-mm snow depth RMSE) of our RA system does not yet meet the objective requirement of satellite measurement accuracy (i.e., 20-mm SWE and 60-mm snow depth for shallow snow conditions) suggested by the Integrated Global Observing Strategy (see WMO 2007). In addition, we focused on limited spatial (North America) and temporal scales (single snow accumulation season), which may not be enough to fully consider all snow conditions (especially wet snow). Our RA system needs to be further improved and evaluated at the global scale over multiyear snow seasons. This will be our aim in future studies.

Fig. 13.
Fig. 13.

The improvement of the RA performance in estimating snow depth for (a) North America, (b) bare soil land cover, and (c) forest land cover by simultaneously updating all model physical states and parameters determining TB based on a rule (RArule); by assimilating the best-performing frequency channels, that is, 18.7 and 23.8 GHz (RA1823); and by considering the vegetation single-scattering albedo [RA1823(ω)].

Citation: Journal of Hydrometeorology 18, 3; 10.1175/JHM-D-16-0102.1

Quantifying and correcting systematic biases in LSMs and RTMs are important to improve the RA performance. Although we used a rule-based approach to mitigate the effect of the systematic biases on the RA performance by simultaneously updating all model physical states and RTM parameters involved in predicting TB, it may not be enough. Further research is required to minimize these biases for high-quality continental-scale snow products using the RA method.

Acknowledgments

This work is supported in part by the National Natural Science Foundation of China under Grant 91337217, in part by the National Aeronautics and Space Administration under Grant NNX11AJ43G, and in part by the National Science Foundation under Grant M0856145. The authors thank Michael Durand, Ghislain Picard, and Christian Mätzler for providing their computer codes and for discussions on the use of them. The Data Assimilation Research Testbed (DART)–Community Atmospheric Model, version 4 (CAM4), atmospheric ensemble reanalysis data are prepared by Kevin Raeder (raeder@ucar.edu). The authors would also like to thank Adam S. Papendieck for language assistance and Long Zhao for processing the time-shifted atmospheric forcing. Computational resources were provided by the UT Texas Advanced Computing Center (TACC).

REFERENCES

  • Anderson, J. L., 2001: An ensemble adjustment Kalman filter for data assimilation. Mon. Wea. Rev., 129, 28842903, doi:10.1175/1520-0493(2001)129<2884:AEAKFF>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Anderson, J. L., 2007: Exploring the need for localization in ensemble data assimilation using a hierarchical ensemble filter. Physica D, 230, 99111, doi:10.1016/j.physd.2006.02.011.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Anderson, J. L., T. Hoar, K. Raeder, H. Liu, N. Collins, R. Torn, and A. Arellano, 2009: The data assimilation research testbed: A community facility. Bull. Amer. Meteor. Soc., 90, 12831296, doi:10.1175/2009BAMS2618.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Andreadis, K. M., and D. P. Lettenmaier, 2012: Implications of representing snowpack stratigraphy for the assimilation of passive microwave satellite observations. J. Hydrometeor., 13, 14931506, doi:10.1175/JHM-D-11-056.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bateni, S. M., S. A. Margulis, E. Podest, and K. C. McDonald, 2015: Characterizing snowpack and the freeze–thaw state of underlying soil via assimilation of multifrequency passive/active microwave data: A case study (NASA CLPX 2003). IEEE Trans. Geosci. Remote Sens., 53, 173189, doi:10.1109/TGRS.2014.2320264.

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

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brown, R. D., and B. Brasnett, 2010: Canadian Meteorological Centre (CMC) Daily Snow Depth Analysis Data (updated annually). National Snow and Ice Data Center, Boulder, CO, digital media, accessed 26 April 2016. [Available online at https://nsidc.org/data/docs/daac/nsidc0447_CMC_snow_depth/.]

  • Brown, R. D., C. Derksen, and L. Wang, 2010: A multi-dataset analysis of variability and change in Arctic spring snow cover extent, 1967–2008. J. Geophys. Res., 115, D16111, doi:10.1029/2010JD013975.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brucker, L., G. Picard, and M. Fily, 2010: Snow grain size profiles deduced from microwave snow emissivities in Antarctica. J. Glaciol., 56, 514526, doi:10.3189/002214310792447806.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brucker, L., A. Royer, G. Picard, A. Langlois, and M. Fily, 2011: Hourly simulations of the microwave brightness temperature of seasonal snow in Quebec, Canada, using a coupled snow evolution–emission model. Remote Sens. Environ., 115, 19661977, doi:10.1016/j.rse.2011.03.019.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Burgers, G., P. J. van Leeuwen, and G. Evensen, 1998: Analysis scheme in the ensemble Kalman filter. Mon. Wea. Rev., 126, 17191724, doi:10.1175/1520-0493(1998)126<1719:ASITEK>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Champollion, N., G. Picard, L. Arnaud, E. Lefebvre, and M. Fily, 2013: Hoar crystal development and disappearance at Dome C, Antarctica: Observation by near-infrared photography and passive microwave satellite. Cryosphere Discuss., 7, 175217, doi:10.5194/tcd-7-175-2013.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chang, A. T. C., J. L. Foster, and D. K. Hall, 1996: Effects of forest on the snow parameters derived from microwave measurements during the BOREAS winter field experiment. Hydrol. Processes, 10, 15651574, doi:10.1002/(SICI)1099-1085(199612)10:12<1565::AID-HYP501>3.0.CO;2-5.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chapin, F. S., and et al. , 2005: Role of land-surface changes in Arctic summer warming. Science, 310, 657660, doi:10.1126/science.1117368.

  • Che, T., X. Li, R. Jin, and C. Huang, 2014: Assimilating passive microwave remote sensing data into a land surface model to improve the estimation of snow depth. Remote Sens. Environ., 143, 5463, doi:10.1016/j.rse.2013.12.009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cohen, J., and D. Entekhabi, 1999: Eurasian snow cover variability and Northern Hemisphere climate predictability. Geophys. Res. Lett., 26, 345348, doi:10.1029/1998GL900321.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Debye, P., H. R. Anderson, and H. Brumberger, 1957: Scattering by an inhomogeneous solid. II. The correlation function and its application. J. Appl. Phys., 28, 679683, doi:10.1063/1.1722830.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dechant, C., and M. Moradkhani, 2011: Radiance data assimilation for operational snow and streamflow forecasting. Adv. Water Resour., 34, 351364, doi:10.1016/j.advwatres.2010.12.009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Durand, M., and S. A. Margulis, 2006: Feasibility test of multifrequency radiometric data assimilation to estimate snow water equivalent. J. Hydrometeor., 7, 443457, doi:10.1175/JHM502.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Durand, M., and S. A. Margulis, 2007: Correcting first-order errors in snow water equivalent estimates using a multifrequency, multiscale radiometric data assimilation scheme. J. Geophys. Res., 112, D13121, doi:10.1029/2006JD008067.

    • Search Google Scholar
    • Export Citation
  • Durand, M., E. J. Kim, and S. A. Margulis, 2008: Quantifying uncertainty in modeling snow microwave radiance for a mountain snowpack at the point-scale, including stratigraphic effects. IEEE Trans. Geosci. Remote Sens., 46, 17531767, doi:10.1109/TGRS.2008.916221.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Durand, M., E. J. Kim, and S. A. Margulis, 2009: Radiance assimilation shows promise for snowpack characterization. Geophys. Res. Lett., 36, L02503, doi:10.1029/2008GL035214.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Evensen, G., 1994: Sequential data assimilation with a nonlinear quasi-geostrophic model using Monte Carlo methods to forecast error statistics. J. Geophys. Res., 99, 10 14310 162, doi:10.1029/94JC00572.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ferrazzoli, P., L. Guerriero, and J.-P. Wigneron, 2002: Simulating L-band emission of forests in view of future satellite applications. IEEE Trans. Geosci. Remote Sens., 40, 27002708, doi:10.1109/TGRS.2002.807577.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Forman, B. A., R. H. Reichle, and M. Rodell, 2012: Assimilation of terrestrial water storage from GRACE in a snow-dominated basin. Water Resour. Res., 48, W01507, doi:10.1029/2011WR011239.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Foster, J. L., A. T. C. Chang, D. K. Hall, and A. Rango, 1991: Derivation of snow water equivalent in boreal forests using microwave radiometry. Arctic, 44, 147152, doi:10.14430/arctic1581.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Foster, J. L., C. Sun, J. P. Walker, R. Kelly, A. Chang, J. Dong, and H. Powell, 2005: Quantifying the uncertainty in passive microwave snow water equivalent observations. Remote Sens. Environ., 94, 187203, doi:10.1016/j.rse.2004.09.012.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gent, P. R., and et al. , 2011: The Community Climate System Model version 4. J. Climate, 24, 49734991, doi:10.1175/2011JCLI4083.1.

  • Gong, G., D. Entekhabi, and J. Cohen, 2003: Modeled Northern Hemisphere winter climate response to realistic Siberian snow anomalies. J. Climate, 16, 39173931, doi:10.1175/1520-0442(2003)016<3917:MNHWCR>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Grant, J. P., K. Saleh, J.-P. Wigneron, M. Guglielmetti, Y. H. Kerr, M. Schwank, N. Skou, and A. Van de Griend, 2008: Calibration of the L-MEB model over a coniferous and a deciduous forest. IEEE Trans. Geosci. Remote Sens., 46, 808818, doi:10.1109/TGRS.2007.914801.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hall, D. K., J. L. Foster, and A. T. C. Chang, 1982: Measurement and modeling of microwave emission from forested snowfields in Michigan. Nord. Hydrol., 13, 129138.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hall, D. K., V. V. Salomonson, and G. A. Riggs, 2006: MODIS/Terra Snow Cover Daily L3 Global 0.05deg CMG, version 5 (updated daily). National Snow and Ice Data Center, Boulder, CO, digital media, accessed 26 April 2016. [Available online at http://nsidc.org/data/docs/daac/modis_v5/mod10c1_modis_terra_snow_daily_global_0.05deg_cmg.gd.html.]

  • Hallikainen, M. T., and P. A. Jolma, 1992: Comparison of algorithms for retrieval of snow water equivalent from Nimbus-7 SMMR data in Finland. IEEE Trans. Geosci. Remote Sens., 30, 124131, doi:10.1109/36.124222.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jackson, T. J., and T. J. Schmugge, 1991: Vegetation effects on the microwave emission of soils. Remote Sens. Environ., 36, 203212, doi:10.1016/0034-4257(91)90057-D.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jin, Y. Q., 1994: Electromagnetic Scattering Modelling for Quantitative Remote Sensing. World Scientific, 348 pp.

    • Crossref
    • Export Citation
  • Knowles, K., M. Savoie, R. Armstrong, and M. Brodzik. 2006a: AMSR-E/Aqua daily global quarter-degree gridded brightness temperatures, version 1. National Snow and Ice Data Center, Boulder, CO, digital media, accessed 26 April 2016, doi:10.5067/RRR4WWORG070.

    • Crossref
    • Export Citation
  • Knowles, K., M. Savoie, R. Armstrong, and M. Brodzik, 2006b: AMSR-E/Aqua daily EASE-grid brightness temperatures. National Snow and Ice Data Center, Boulder, CO, digital media, accessed 26 April 2016. [Available online at https://nsidc.org/data/docs/daac/nsidc0301_amsre_gridded_tb.gd.html.]

  • Kruopis, N., J. Praks, A. N. Arslan, H. M. Alasalmi, J. T. Koskinen, and M. T. Hallikainen, 1999: Passive microwave measurements of snow-covered forest area in EMAC’95. IEEE Trans. Geosci. Remote Sens., 37, 26992705, doi:10.1109/36.803417.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kurum, M., P. E. O’Neill, R. H. Lang, A. T. Joseph, M. H. Cosh, and T. J. Jackson, 2012: Effective tree scattering and opacity at L-band. Remote Sens. Environ., 118, 19, doi:10.1016/j.rse.2011.10.024.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kwon, Y., A. M. Toure, Z.-L. Yang, M. Rodell, and G. Picard, 2015: Error characterization of coupled land surface–radiative transfer models for snow microwave radiance assimilation. IEEE Trans. Geosci. Remote Sens., 53, 52475268, doi:10.1109/TGRS.2015.2419977.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kwon, Y., Z.-L. Yang, L. Zhao, T. J. Hoar, A. M. Toure, and M. Rodell, 2016: Estimating snow water storage in North America using CLM4, DART, and snow radiance data assimilation. J. Hydrometeor., 17, 28532874, doi:10.1175/JHM-D-16-0028.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Langlois, A., A. Royer, F. Dupont, A. Roy, K. Goïta, and G. Picard, 2011: Improved corrections of forests effects on passive microwave satellite remote sensing of snow over boreal and subarctic regions. IEEE Trans. Geosci. Remote Sens., 49, 38243837, doi:10.1109/TGRS.2011.2138145.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Langlois, A., A. Royer, C. Derksen, B. Montpetit, F. Dupont, and K. Goïta, 2012: Coupling the snow thermodynamic model SNOWPACK with the microwave emission model of layered snowpacks for subarctic and arctic snow water equivalent retrievals. Water Resour. Res., 48, W12524, doi:10.1029/2012WR012133.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lawrence, D., and et al. , 2011: Parameterization improvements and functional and structural advances in version 4 of the Community Land Model. J. Adv. Model. Earth Syst., 3, M03001, doi:10.1029/2011MS00045.

    • Search Google Scholar
    • Export Citation
  • Lobl, E., 2001: Joint Advanced Microwave Scanning Radiometer (AMSR) Science Team meeting. Earth Observer, Vol. 13, Issue 3, NASA Goddard Space Flight Center, Greenbelt, Maryland, 3–9. [Available online at https://eospso.nasa.gov/sites/default/files/eo_pdfs/may_jun01.pdf.]

  • Löwe, H., and G. Picard, 2015: Microwave scattering coefficient of snow in MEMLS and DMRT-ML revisited: The relevance of sticky hard spheres and tomography-based estimates of stickiness. Cryosphere, 9, 21012117, doi:10.5194/tc-9-2101-2015.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mätzler, C., 1987: Applications of the interaction of microwaves with the natural snow cover. Remote Sens. Rev., 2, 259387, doi:10.1080/02757258709532086.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mätzler, C., 1998: Improved Born approximation for scattering of radiation in a granular medium. J. Appl. Phys., 83, 61116117, doi:10.1063/1.367496.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mätzler, C., 2002: Relation between grain-size and correlation length of snow. J. Glaciol., 48, 461466, doi:10.3189/172756502781831287.

  • Mätzler, C., H. Aebischer, and E. Schanda, 1984: Microwave dielectric properties of surface snow. IEEE J. Oceanic Eng., 9, 366371, doi:10.1109/JOE.1984.1145644.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mo, T., B. J. Choudhury, T. J. Schmugge, J. R. Wang, and T. J. Jackson, 1982: A model for microwave emission from vegetation-covered fields. J. Geophys. Res., 87, 11 22911 237, doi:10.1029/JC087iC13p11229.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Namias, J., 1985: Some empirical evidence for the influence of snow cover on temperature and precipitation. Mon. Wea. Rev., 113, 15421553, doi:10.1175/1520-0493(1985)113<1542:SEEFTI>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Niu, G.-Y., and Z.-L. Yang, 2007: An observation-based formulation of snow cover fraction and its evaluation over large North American river basins. J. Geophys. Res., 112, D21101, doi:10.1029/2007JD008674.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Njoku, E. G., and L. Li, 1999: Retrieval of land surface parameters using passive microwave measurements at 6–18 GHz. IEEE Trans. Geosci. Remote Sens., 37, 7993, doi:10.1109/36.739125.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Njoku, E. G., P. Ashcroft, T. K. Chan, and L. Li, 2005: Global survey and statistics of radio-frequency interference in AMSR-E land observations. IEEE Trans. Geosci. Remote Sens., 43, 938947, doi:10.1109/TGRS.2004.837507.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Oleson, K. W., and et al. , 2010: Technical description of version 4.0 of the Community Land Model (CLM). NCAR Tech. Note NCAR/TN-478+STR, 257 pp., doi:10.5065/D6FB50WZ.

    • Crossref
    • Export Citation
  • Paloscia, S., 1995: Microwave emission from vegetation. Passive Microwave Remote Sensing of Land–Atmosphere Interactions, B. J. Choudhury et al., Eds., VSP Press, 357–374.

    • Crossref
    • Export Citation
  • Paloscia, S., and P. Pampaloni, 1988: Microwave polarization index for monitoring vegetation growth. IEEE Trans. Geosci. Remote Sens., 26, 617621, doi:10.1109/36.7687.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pampaloni, P., 2004: Microwave radiometry of forests. Waves Random Media, 14, S275S298, doi:10.1088/0959-7174/14/2/009.

  • Pampaloni, P., and S. Paloscia, 1986: Microwave emission and plant water content: A comparison between field measurements and theory. IEEE Trans. Geosci. Remote Sens., 24, 900905, doi:10.1109/TGRS.1986.289705.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pellarin, T., Y. H. Kerr, and J.-P. Wigneron, 2006: Global simulations of brightness temperature at 6.6 and 10.7 GHz over land based on SMMR data set analysis. IEEE Trans. Geosci. Remote Sens., 44, 24922505, doi:10.1109/TGRS.2006.874139.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Picard, G., L. Brucker, A. Roy, F. Dupont, M. Fily, and A. Royer, 2013: Simulation of the microwave emission of multi-layered snowpacks using the dense media radiative transfer theory: The DMRT-ML model. Geosci. Model Dev., 6, 10611078, doi:10.5194/gmd-6-1061-2013.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pulliainen, J., J.-P. Kärnä, and M. Hallikainen, 1993: Development of geophysical retrieval algorithms for the MIMR. IEEE Trans. Geosci. Remote Sens., 31, 268277, doi:10.1109/36.210466.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Raeder, K., J. L. Anderson, N. Collins, T. J. Hoar, J. E. Kay, P. H. Lauritzen, and R. Pincus, 2012: DART/CAM: An ensemble data assimilation system for CESM atmospheric models. J. Climate, 25, 63046317, doi:10.1175/JCLI-D-11-00395.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rees, A., J. Lemmetyinen, C. Derksen, J. Pulliainen, and M. English, 2010: Observed and modelled effects of ice lens formation on passive microwave brightness temperatures over snow covered tundra. Remote Sens. Environ., 114, 116126, doi:10.1016/j.rse.2009.08.013.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reichle, R. H., R. D. Koster, G. J. M. De Lannoy, B. A. Forman, Q. Liu, S. P. P. Mahanama, and A. Toure, 2011: Assessment and enhancement of MERRA land surface hydrology estimates. J. Climate, 24, 63226338, doi:10.1175/JCLI-D-10-05033.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rosenfeld, S., and N. Grody, 2000: Anomalous microwave spectra of snow cover observed from Special Sensor Microwave/Imager measurements. J. Geophys. Res., 105, 14 91314 925, doi:10.1029/1999JD900486.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Roy, A., A. Royer, J.-P. Wigneron, A. Langlois, J. Bergeron, and P. Cliche, 2012: A simple parameterization for a boreal forest radiative transfer model at microwave frequencies. Remote Sens. Environ., 124, 371383, doi:10.1016/j.rse.2012.05.020.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Roy, A., G. Picard, A. Royer, B. Montpetit, F. Dupont, A. Langlois, C. Derksen, and N. Champollion, 2013: Brightness temperature simulations of the Canadian seasonal snowpack driven by measurements of the snow specific surface area. IEEE Trans. Geosci. Remote Sens., 51, 46924704, doi:10.1109/TGRS.2012.2235842.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schmugge, T. J., and T. J. Jackson, 1992: A dielectric model of the vegetation effects on the microwave emission from soils. IEEE Trans. Geosci. Remote Sens., 30, 757760, doi:10.1109/36.158870.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stewart, I. T., D. R. Cayan, and M. D. Dettinger, 2004: Changes in snowmelt runoff timing in western North America under a ‘business as usual’ climate change scenario. Climatic Change, 62, 217232, doi:10.1023/B:CLIM.0000013702.22656.e8.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sturm, M., J. Holmgren, and G. E. Liston, 1995: A seasonal snow cover classification system for local to regional applications. J. Climate, 8, 12611283, doi:10.1175/1520-0442(1995)008<1261:ASSCCS>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sturm, M., B. Taras, G. E. Liston, C. Derksen, T. Jonas, and J. Lea, 2010: Estimating snow water equivalent using snow depth data and climate classes. J. Hydrometeor., 11, 13801394, doi:10.1175/2010JHM1202.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Su, H., Z.-L. Yang, R. E. Dickinson, C. R. Wilson, and G.-Y. Niu, 2010: Multisensor snow data assimilation at the continental scale: The value of Gravity Recovery and Climate Experiment terrestrial water storage information. J. Geophys. Res., 115, D10104, doi:10.1029/2009JD013035.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tedesco, M., and E. J. Kim, 2006: Intercomparison of electromagnetic models for passive microwave remote sensing of snow. IEEE Trans. Geosci. Remote Sens., 44, 26542666, doi:10.1109/TGRS.2006.873182.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tippett, M. K., J. L. Anderson, C. H. Bishop, T. M. Hamill, and J. S. Whitaker, 2003: Ensemble square root filters. Mon. Wea. Rev., 131, 14851490, doi:10.1175/1520-0493(2003)131<1485:ESRF>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Toure, A. M., K. Goïta, A. Royer, E. J. Kim, M. Durand, S. A. Margulis, and H. Lu, 2011: A case study of using a multilayered thermodynamical snow model for radiance assimilation. IEEE Trans. Geosci. Remote Sens., 49, 28282837, doi:10.1109/TGRS.2011.2118761.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Toure, A. M., M. Rodell, Z.-L. Yang, H. Beaudoing, E. Kim, Y. Zhang, and Y. Kwon, 2016: Evaluation of the snow simulations from the Community Land Model, version 4 (CLM4). J. Hydrometeor., 17, 153170, doi:10.1175/JHM-D-14-0165.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tsang, L., and J. A. Kong, 2001: Scattering of Electromagnetic Waves: Advanced Topics. Vol. 3. Wiley-Interscience, 413 pp.

    • Crossref
    • Export Citation
  • Tsang, L., P. Xu, and K. S. Chen, 2008: Third and fourth Stokes parameters in polarimetric passive microwave remote sensing of rough surfaces over layered media. Microwave Opt. Technol. Lett., 50, 30633069, doi:10.1002/mop.23892.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Turner, D. D., S. A. Clough, J. C. Liljegren, E. E. Clothiaux, K. E. Cady-Pereira, and K. L. Gaustad, 2007: Retrieving liquid water path and precipitable water vapor from the Atmospheric Radiation Measurement (ARM) microwave radiometers. IEEE Trans. Geosci. Remote Sens., 45, 36803690, doi:10.1109/TGRS.2007.903703.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ulaby, F. T., R. K. Moore, and A. K. Fung, 1981: Microwave Remote Sensing: Active and Passive. Vol. 1. Addison-Wesley, 456 pp.

  • Vernekar, A. D., J. Zhou, and J. Shukla, 1995: The effect of Eurasian snow cover on the Indian monsoon. J. Climate, 8, 248266, doi:10.1175/1520-0442(1995)008<0248:TEOESC>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wegmüller, U., and C. Mätzler, 1999: Rough bare soil reflectivity model. IEEE Trans. Geosci. Remote Sens., 37, 13911395, doi:10.1109/36.763303.

  • Wiesmann, A., and C. Mätzler, 1999: Microwave emission model of layered snowpacks. Remote Sens. Environ., 70, 307316, doi:10.1016/S0034-4257(99)00046-2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • WMO, 2007: Integrated Global Observing Strategy for the monitoring of our environment from space and from Earth. WMO/TD-1405, 100 pp. [Available online at http://cryos.ssec.wisc.edu/docs/cryos_theme_report.pdf.]

  • Zhang, Y.-F., T. J. Hoar, Z.-L. Yang, J. L. Anderson, A. M. Toure, and M. Rodell, 2014: Assimilation of MODIS snow cover through the Data Assimilation Research Testbed and the Community Land Model version 4. J. Geophys. Res. Atmos., 119, 70917103, doi:10.1002/2013JD021329.

    • Crossref
    • Search Google Scholar
    • Export Citation
Save