• Behrangi, A., M. Lebsock, S. Wong, and B. Lambrigtsen, 2012: On the quantification of oceanic rainfall using spaceborne sensors. J. Geophys. Res., 117, D20105, https://doi.org/10.1029/2012JD017979.

    • Search Google Scholar
    • Export Citation
  • Bolton, D., 1980: The computation of equivalent potential temperature. Mon. Wea. Rev., 108, 10461053, https://doi.org/10.1175/1520-0493(1980)108<1046:TCOEPT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Boucher, R. J., and J. G. Wieler, 1985: Radar determination of snowfall rate and accumulation. J. Climate Appl. Meteor., 24, 6873, https://doi.org/10.1175/1520-0450(1985)024<0068:RDOSRA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chang, E. K. M., S. Lee, and K. L. Swanson, 2002: Storm track dynamics. J. Climate, 15, 21632183, https://doi.org/10.1175/1520-0442(2002)015<02163:STD>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Davies-Jones, R., 2008: An efficient and accurate method for computing the wet-bulb temperature along pseudoadiabats. Mon. Wea. Rev., 136, 27642785, https://doi.org/10.1175/2007MWR2224.1.

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

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ellis, T. D., T. L’Ecuyer, J. M. Haynes, and G. L. Stephens, 2009: How often does it rain over the global oceans? The perspective from CloudSat. Geophys. Res. Lett., 36, L03815, https://doi.org/10.1029/2008GL036728.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ferraro, R. R., F. Weng, N. C. Grody, and A. Basist, 1996: An eight-year (1987–1994) time series of rainfall, clouds, water vapor, snow cover, and sea ice derived from SSM/I measurements. Bull. Amer. Meteor. Soc., 77, 891905, https://doi.org/10.1175/1520-0477(1996)077<0891:AEYTSO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Field, P. R., and A. J. Heymsfield, 2015: Importance of snow to global precipitation. Geophys. Res. Lett., 42, 95129520, https://doi.org/10.1002/2015GL065497.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Frankoski, N. J., and A. T. Degaetano, 2011: An East Coast winter storm precipitation climatology. Int. J. Climatol., 31, 802814, https://doi.org/10.1002/joc.2121.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Grim, J. A., N. F. Laird, and D. R. Kristovich, 2004: Mesoscale vortices embedded within a lake-effect shoreline band. Mon. Wea. Rev., 132, 22692274, https://doi.org/10.1175/1520-0493(2004)132<2269:MVEWAL>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hamada, A., and Y. N. Takayabu, 2016: Improvements in detection of light precipitation with the Global Precipitation Measurement Dual-Frequency Precipitation Radar (GPM DPR). J. Atmos. Oceanic Technol., 33, 653667, https://doi.org/10.1175/JTECH-D-15-0097.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Han, Y., and E. R. Westwater, 2000: Analysis and improvement of tipping calibration for ground-based microwave radiometers. IEEE Trans. Geosci. Remote Sens., 38, 12601276, https://doi.org/10.1109/36.843018.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hiley, M. J., M. S. Kulie, and R. Bennartz, 2011: Uncertainty analysis for CloudSat snowfall retrievals. J. Appl. Meteor. Climatol., 50, 399418, https://doi.org/10.1175/2010JAMC2505.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hou, A. Y., and Coauthors, 2014: The Global Precipitation Measurement Mission. Bull. Amer. Meteor. Soc., 95, 701722, https://doi.org/10.1175/BAMS-D-13-00164.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Iguchi, T., S. Seto, R. Meneghini, N. Yoshida, J. Awaka, and T. Kubota, 2010: GPM/DPR level-2. Algorithm Theoretical Basis Doc., 72 pp., https://pmm.nasa.gov/sites/default/files/document_files/ATBD_GPM_DPR_n3_dec15.pdf.

  • Iguchi, T., S. Seto, R. Meneghini, N. Yoshida, J. Awaka, M. Le, V. Chandrasekar, and T. Kubota, 2015: GPM/DPR level-2. Algorithm Theoretical Basis Doc., 68 pp., http://www.eorc.jaxa.jp/GPM/doc/algorithm/ATBD_DPR_2015_whole_2a.pdf.

  • Katsumata, M., H. Uyeda, K. Iwanami, and G. Liu, 2000: The response of 36- and 89-GHz microwave channels to convective snow clouds over ocean: Observation and modeling. J. Appl. Meteor., 39, 23222335, https://doi.org/10.1175/1520-0450(2000)039<2322:TROAGM>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kidd, C., A. Becker, G. J. Huffman, C. L. Muller, P. Joe, G. Skofronick-Jackson, and D. B. Kirschbaum, 2017: So, how much of the Earth’s surface is covered by rain gauges? Bull. Amer. Meteor. Soc., 98, 6978, https://doi.org/10.1175/BAMS-D-14-00283.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kneifel, S., U. Löhnert, A. Battaglia, S. Crewell, and D. Siebler, 2010: Snow scattering signals in ground-based passive microwave radiometer measurements. J. Geophys. Res., 115, D16214, https://doi.org/10.1029/2010JD013856.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kongoli, C., 2003: A new snowfall detection algorithm over land using measurements from the Advanced Microwave Sounding Unit (AMSU). Geophys. Res. Lett., 30, 1756, https://doi.org/10.1029/2003GL017177.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kristovich, D. R., and M. L. Spinar, 2005: Diurnal variations in lake-effect precipitation near the western Great Lakes. J. Hydrometeor., 6, 210218, https://doi.org/10.1175/JHM403.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kulie, M. S., and R. Bennartz, 2009: Utilizing spaceborne radars to retrieve dry snowfall. J. Appl. Meteor. Climatol., 48, 25642580, https://doi.org/10.1175/2009JAMC2193.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kulie, M. S., R. Bennartz, T. J. Greenwald, Y. Chen, and F. Weng, 2010: Uncertainties in microwave properties of frozen precipitation: Implications for remote sensing and data assimilation. J. Atmos. Sci., 67, 34713487, https://doi.org/10.1175/2010JAS3520.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kulie, M. S., L. Milani, N. B. Wood, S. A. Tushaus, R. Bennartz, and T. S. L’Ecuyer, 2016: A shallow cumuliform snowfall census using spaceborne radar. J. Hydrometeor., 17, 12611279, https://doi.org/10.1175/JHM-D-15-0123.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kummerow, C., and Coauthors, 2000: The status of the Tropical Rainfall Measuring Mission (TRMM) after two years in orbit. J. Appl. Meteor., 39, 19651982, https://doi.org/10.1175/1520-0450(2001)040<1965:TSOTTR>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Levizzani, V., S. Laviola, and E. Cattani, 2011: Detection and measurement of snowfall from space. Remote Sens., 3, 145166, https://doi.org/10.3390/rs3010145.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, C., 2016: GPM precipitation feature database. Tech. Doc., 15 pp., http://atmos.tamucc.edu/trmm/data/document/GPM_database_description_1.0_201601.pdf.

  • Liu, C., and E. J. Zipser, 2009: “Warm rain” in the tropics: Seasonal and regional distributions based on 9 yr of TRMM data. J. Climate, 22, 767779, https://doi.org/10.1175/2008JCLI2641.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, C., and E. J. Zipser, 2015: The global distribution of largest, deepest, and most intense precipitation systems. Geophys. Res. Lett., 42, 35913595, https://doi.org/10.1002/2015GL063776.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, C., E. J. Zipser, D. J. Cecil, S. W. Nesbitt, and S. Sherwood, 2008: A cloud and precipitation feature database from nine years of TRMM observations. J. Appl. Meteor. Climatol., 47, 27122728, https://doi.org/10.1175/2008JAMC1890.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, G., 2008: Deriving snow cloud characteristics from CloudSat observations. J. Geophys. Res., 113, D00A09, https://doi.org/10.1029/2007JD009766.

    • Search Google Scholar
    • Export Citation
  • Liu, G., and J. A. Curry, 1997: Precipitation characteristics in Greenland-Iceland-Norwegian Seas determined by using satellite microwave data. J. Geophys. Res., 102, 13 98713 997, https://doi.org/10.1029/96JD03090.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, G., and E. K. Seo, 2013: Detecting snowfall over land by satellite high-frequency microwave observations: The lack of scattering signature and a statistical approach. J. Geophys. Res. Atmos., 118, 13761387, https://doi.org/10.1002/jgrd.50172.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, N., and C. Liu, 2016: Global distribution of deep convection reaching tropopause in 1 year GPM observations. J. Geophys. Res. Atmos., 121, 38243842, https://doi.org/10.1002/2015JD024430.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Matrosov, S. Y., 2007: Modeling backscatter properties of snowfall at millimeter wavelengths. J. Atmos. Sci., 64, 17271736, https://doi.org/10.1175/JAS3904.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mülmenstädt, J., O. Sourdeval, J. Delanoë, and J. Quaas, 2015: Frequency of occurrence of rain from liquid-, mixed-, and ice-phase clouds derived from A-Train satellite retrievals. Geophys. Res. Lett., 42, 65026509, https://doi.org/10.1002/2015GL064604.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nesbitt, S. W., and E. J. Zipser, 2003: The diurnal cycle of rainfall and convective intensity according to three years of TRMM measurements. J. Climate, 16, 14561475, https://doi.org/10.1175/1520-0442-16.10.1456.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Noh, Y. J., G. Liu, E. K. Seo, J. R. Wang, and K. Aonashi, 2006: Development of a snowfall retrieval algorithm at high microwave frequencies. J. Geophys. Res., 111, D22216, https://doi.org/10.1029/2005JD006826.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Noh, Y. J., G. Liu, A. S. Jones, and T. H. V. Haar, 2009: Toward snowfall retrieval over land by combining satellite and in situ measurements. J. Geophys. Res., 114, D24205, https://doi.org/10.1029/2009JD012307.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Precipitation Processing System Team, 2014: Global Precipitation Measurement file specification for GPM products, version 1.08. NASA Doc., 1131 pp., https://storm.pps.eosdis.nasa.gov/storm/filespec.GPM.V1.pdf.

  • Seto, S., T. Iguchi, and T. Oki, 2013: The basic performance of a precipitation retrieval algorithm for the global precipitation measurement mission’s single/dual-frequency radar measurements. IEEE Trans. Geosci. Remote Sens., 51, 52395251, https://doi.org/10.1109/TGRS.2012.2231686.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sims, E. M., and G. Liu, 2015: A parameterization of the probability of snow–rain transition. J. Hydrometeor., 16, 14661477, https://doi.org/10.1175/JHM-D-14-0211.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Skofronick-Jackson, G., and B. T. Johnson, 2011: Surface and atmospheric contributions to passive microwave brightness temperatures for falling snow events. J. Geophys. Res., 116, D02213, https://doi.org/10.1029/2010JD014438.

    • Search Google Scholar
    • Export Citation
  • Skofronick-Jackson, G., J. A. Weinman, M. J. Kim, and D. E. Chang, 2004: A physical model to determine snowfall over land by microwave radiometry. IEEE Trans. Geosci. Remote Sens., 42, 10471058, https://doi.org/10.1109/TGRS.2004.825585.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Skofronick-Jackson, G., and Coauthors, 2017: The Global Precipitation Measurement (GPM) mission for science and society. Bull. Amer. Meteor. Soc., 98, 16791695, https://doi.org/10.1175/BAMS-D-15-00306.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Spencer, R., H. Goodman, and R. Hood, 1989: Precipitation retrieval over land and ocean with the SSM/I: Identification and characteristics of the scattering signal. J. Atmos. Oceanic Technol., 6, 254273, https://doi.org/10.1175/1520-0426(1989)006<0254:PROLAO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Staelin, D. H., and F. W. Chen, 2000: Precipitation observations near 54 and 183 GHz using the NOAA-15 satellite. IEEE Trans. Geosci. Remote Sens., 38, 23222332, https://doi.org/10.1109/36.868889.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stephens, G. L., and Coauthors, 2002: The CloudSat mission and the A-Train: A new dimension of space-based observations of clouds and precipitation. Bull. Amer. Meteor. Soc., 83, 17711790, https://doi.org/10.1175/BAMS-83-12-1771.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stephens, G. L., and Coauthors, 2010: Dreary state of precipitation in global models. J. Geophys. Res., 115, D24211, https://doi.org/10.1029/2010JD014532.

    • Search Google Scholar
    • Export Citation
  • Sun, Y., S. Solomon, A. Dai, and R. W. Portmann, 2006: How often does it rain? J. Climate, 19, 916934, https://doi.org/10.1175/JCLI3672.1.

  • Vivekanandan, J., J. Turk, and V. N. Bringi, 1991: Ice water path estimation and characterization using passive microwave radiometry. J. Appl. Meteor., 30, 14071421, https://doi.org/10.1175/1520-0450(1991)030<1407:IWPEAC>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Weng, F., and N. C. Grody, 1994: Retrieval of cloud liquid water using the Special Sensor Microwave Imager (SSM/I). J. Geophys. Res., 99, 25 53525 551, https://doi.org/10.1029/94JD02304.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wood, N. B., T. S. L’Ecuyer, D. G. Vane, G. L. Stephens, and P. Partain, 2013: Level 2C snow profile process description and interface control document. JPL Doc., 21 pp., http://www.cloudsat.cira.colostate.edu/sites/default/files/products/files/2C-SNOW-PROFILE_PDICD.P_R04.20130210.pdf.

  • Xie, X., U. Lhnert, S. Kneifel, and S. Crewell, 2012: Snow particle orientation observed by ground-based microwave radiometry. J. Geophys. Res., 117, D02206, https://doi.org/10.1029/2011JD016369.

    • Search Google Scholar
    • Export Citation
  • You, Y., N. Y. Wang, and R. Ferraro, 2015: A prototype precipitation retrieval algorithm over land using passive microwave observations stratified by surface condition and precipitation vertical structure. J. Geophys. Res. Atmos., 120, 52955315, https://doi.org/10.1002/2014JD022534.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • You, Y., N.-Y. Wang, R. Ferraro, and P. Meyers, 2016: A prototype precipitation retrieval algorithm over land for ATMS. J. Hydrometeor., 17, 16011621, https://doi.org/10.1175/JHM-D-15-0163.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • You, Y., N.-Y. Wang, R. Ferraro, and S. Rudlosky, 2017: Quantifying the snowfall detection performance of the GPM Microwave Imager channels over land. J. Hydrometeor., 18, 729751, https://doi.org/10.1175/JHM-D-16-0190.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zipser, E., C. Liu, D. Cecil, S. W. Nesbitt, and S. Yorty, 2006: Where are the most intense thunderstorms on Earth? Bull. Amer. Meteor. Soc., 87, 10571071, https://doi.org/10.1175/BAMS-87-8-1057.

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

    An example of snow precipitation features over Lake Superior. (a) KuPR near-surface reflectivity, (b) KuPR echo-top height, (c) KuPR near-surface precipitation rate, and (d) minimum 89-GHz PCT from GMI with matched swath to KuPR. (e) The cross section of KuPR reflectivity along the black line in (b) is shown, where the solid line is precipitation rate (in 10 mm h−1), the black dashed line is GMI 89-GHz PCT, the red line is 183 ± 7 GHz, the blue line is 166-GHz V channel, and the color fill is the KuPR near-surface reflectivity. The outline of snow features are indicated by black lines in (c).

  • View in gallery

    (a) Occurrence of near-surface radar reflectivity at 1-dBZ bins from snow pixels over North and South ocean and land in GPM KuPR (April 2014–March 2017) and CloudSat CPR (2007–10) data. The color-shaded regions correspond to the range (minimum and maximum) values of CPR occurrences obtained from any of the four running average of 3, 5, 7, and 9 pixels. The results of 5-pixel average are presented with dashed lines inside color-shaded regions. (b) As in (a), but with near-surface snow precipitation rate. Note that occurrence is calculated by dividing the sum of snow pixels at specific bin size by the total sample pixels (including both snow and not snow), so that the sum of each curve equals the total snow occurrence over each region respectively. (c) Total snowfall contribution from specific snowfall rates in logarithmic bins size.

  • View in gallery

    Geographical distribution of (a) mean unconditional KuPR near-surface snowfall rate (mm yr−1) and (b) percentage of precipitation as snow in each 0.25° × 0.25° box. Note that the mean unconditional snowfall rate is calculated by dividing total snow accumulation by total samples (snow and no snow) in each 0.25° × 0.25° box. (c) Number of KuPR snow features in each 1° × 1° box during April 2014–March 2017.

  • View in gallery

    (a) Percentage contribution of total volumetric snow by SFs of different sizes. (b) The locations of SFs categorized by size and rarity of the events is represented by colored symbols. (c) Contribution of SFs with size >10 000 km2. The fraction is calculated in 2° × 2° boxes.

  • View in gallery

    As in Fig. 4, but by SFs with different maximum KuPR near-surface reflectivity.

  • View in gallery

    As in Fig. 5, but by SFs with different maximum KuPR echo-top height.

  • View in gallery

    (a) Diurnal and (b) seasonal variations of population of SFs over North and South land and ocean. The red dashed line in (a) is diurnal variation of SFs over North land including those with ERA-Interim wet bulb temperature warmer than 1°C.

  • View in gallery

    2D histograms of SFs categorized by maximum KuPR echo-top heights and month over (a) North ocean, (b) North land, and (c) South ocean and (d) 2D histograms of SFs categorized by size and month over (d) North ocean, (e) North land, and (f) South ocean.

  • View in gallery

    CDFs of SFs (a) by sizes, (b) by KuPR maximum near-surface reflectivity, (c) by maximum echo-top height, (d) by minimum 89-GHz PCT, (e) by minimum 166-GHz V TB, and (f) by minimum 183 ± 7 GHz TB.

  • View in gallery

    (a) Cumulative 2D histogram of the SFs over (a) North and (b) South land (contours) and ocean (color filled) as a function of maximum near-surface reflectivity and maximum echo-top height. (c),(d) As in (a),(b), but as a function of minimum 89-GHz PCT and maximum echo-top height.

  • View in gallery

    (a) Cumulative 2D histogram of the SFs over North land (contours) and ocean (color filled) as a function of maximum echo-top heights and minimum 166-GHz (H) TB. (b) As in (a), but as a function of maximum echo-top heights and minimum 166-GHz (V) TB. (c) As in (a), but as a function of maximum near-surface snow rate and minimum 166-GHz (H) TB. (d) As in (a), but as a function of maximum near-surface snow rate and minimum 166-GHz (V) TB.

  • View in gallery

    (a) Cumulative 2D histogram of the SFs over Northern Hemispheric land (contours) and over ocean (color filled) as a function of maximum echo-top heights and min 166 GHz (H) TB. (b) Same as figure (a) but as a function of maximum echo-top heights and min 166 GHz (V) TB. (c) Same as figure (a) but as a function of maximum near-surface snow rate and min 166 GHz (H) TB. (d) Same as figure (a) but as a function of maximum near-surface snow rate and min 166 GHz (V) TB.

  • View in gallery

    (a) Cumulative 2D histogram of the SFs over North land (contours) and ocean (color filled) as a function of maximum echo-top heights and minimum 183 ± 3 GHz TB. (b) As in (a), but as a function of maximum echo-top heights and minimum 183 ± 7 GHz TB. (c) As in (a), but as a function of maximum near-surface snow rate and minimum 183 ± 3 GHz TB. (d) As in (a), but as a function of maximum near-surface snow rate and minimum 183 ± 7 GHz TB.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 32 32 22
PDF Downloads 32 32 24

Global Distribution of Snow Precipitation Features and Their Properties from 3 Years of GPM Observations

View More View Less
  • 1 Department of Physical and Environmental Sciences, Texas A&M University–Corpus Christi, Corpus Christi, Texas
  • 2 Department of Geological and Mining Engineering and Sciences, Michigan Technological University, Houghton, Michigan
© Get Permissions
Full access

Abstract

Using a 3-yr Global Precipitation Mission (GPM) Ku-band Precipitation Radar (KuPR) dataset, snow features (SFs) are defined by grouping the contiguous area of nonzero solid precipitation. The near-surface wet bulb temperatures calculated from ERA-Interim reanalysis data are used to verify that SFs are colder than 1°C to omit snowfall that melts before reaching the surface. The properties of SFs are summarized to understand the global distribution and characteristics of snow systems. The seasonal and diurnal variations of SFs and their properties are analyzed over Northern and Southern Hemispheric land and ocean separately.

To quantify the amount of snow missed by the GPM KuPR and the amount of snow underestimated by the CloudSat Cloud Profiling (CPR), 3-yr KuPR pixel-level data are compared with 4-yr CloudSat CPR observations. The overall underestimation of snowfall during heavy snow events by CPR is less than 3% compared to the combined CPR and KuPR estimates. KuPR underestimates about 52% of weak snow. Only a small percentage of SFs have sizes greater than 10 000 km2 (0.35%), maximum near-surface reflectivity above 30 dBZ (5.1%), or echo top above 5 km (1.6%); however, they contribute 40%, 49.5%, or 30.4% of the global volumetric snow detected by KuPR. Snow in the Northern Hemisphere has stronger diurnal and seasonal variation compared to the Southern Hemisphere. Most of the SFs over the ocean are found with relatively smaller, less intense, and shallower echo tops than over land.

© 2018 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: Abishek Adhikari, aadhikari1@islander.tamucc.edu

Abstract

Using a 3-yr Global Precipitation Mission (GPM) Ku-band Precipitation Radar (KuPR) dataset, snow features (SFs) are defined by grouping the contiguous area of nonzero solid precipitation. The near-surface wet bulb temperatures calculated from ERA-Interim reanalysis data are used to verify that SFs are colder than 1°C to omit snowfall that melts before reaching the surface. The properties of SFs are summarized to understand the global distribution and characteristics of snow systems. The seasonal and diurnal variations of SFs and their properties are analyzed over Northern and Southern Hemispheric land and ocean separately.

To quantify the amount of snow missed by the GPM KuPR and the amount of snow underestimated by the CloudSat Cloud Profiling (CPR), 3-yr KuPR pixel-level data are compared with 4-yr CloudSat CPR observations. The overall underestimation of snowfall during heavy snow events by CPR is less than 3% compared to the combined CPR and KuPR estimates. KuPR underestimates about 52% of weak snow. Only a small percentage of SFs have sizes greater than 10 000 km2 (0.35%), maximum near-surface reflectivity above 30 dBZ (5.1%), or echo top above 5 km (1.6%); however, they contribute 40%, 49.5%, or 30.4% of the global volumetric snow detected by KuPR. Snow in the Northern Hemisphere has stronger diurnal and seasonal variation compared to the Southern Hemisphere. Most of the SFs over the ocean are found with relatively smaller, less intense, and shallower echo tops than over land.

© 2018 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: Abishek Adhikari, aadhikari1@islander.tamucc.edu

1. Introduction

Snowfall measurement from space is important to understand the global precipitation and hydrologic cycle. Although only about 5% of the total global precipitation falls as a snow (Levizzani et al. 2011), the majority of the precipitation in the mid- and high latitudes falls as snow (Liu and Curry 1997; Ellis et al. 2009; Levizzani et al. 2011; Field and Heymsfield 2015; Mülmenstädt et al. 2015). The early methods to measure snow in the past include ground-based snow gauges, ground-based microwave observations (both active and passive), and spaceborne radiometers. Because gauge deployments are restricted to land and cover less than 1% of the globe (Kidd et al. 2017), and ground-based radar observations (Boucher and Wieler 1985; Ferraro et al. 1996; Han and Westwater 2000; Skofronick-Jackson et al. 2004; Kneifel et al. 2010; Xie et al. 2012) are mainly for validation purposes that cover only certain regions of interest, the global measurement of snowfall was not possible until the recent satellite era. Many past ground-based microwave observations suggested the possibility of snowfall detection from the spaceborne microwave observations (Liu and Curry 1997; Staelin and Chen 2000; Kongoli 2003; Skofronick-Jackson et al. 2004; Noh et al. 2009). In these studies, the in situ measurements, ground-based radar, and spaceborne passive microwave (PMW) radiometers such as Special Sensor Microwave Imager (SSM/I) (e.g., Weng and Grody 1994; Liu and Curry 1997), Advanced Microwave Sounding Unit (AMSU) (e.g., Staelin and Chen 2000; Kongoli 2003; Skofronick-Jackson et al. 2004; Noh et al. 2006), and Airborne Microwave Imager (AMR) (e.g., Katsumata et al. 2000) etc. were used to demonstrate the possibility of detecting snow precipitation by using high frequency microwave channels at 89, 150, and 183 GHz. Recent studies (e.g., You et al. 2015, 2016) have demonstrated the snowfall retrieval performance of Special Sensor Microwave Imager/Sensor (SSMIS) and Advanced Technology Microwave Sounder (ATMS) over land. Also, You et al. (2017) has demonstrated the performance of all Global Precipitation Measurement (GPM) Microwave Imager (GMI) channels over land and showed that the 166-GHz channel is indispensable for snowfall detection. Although the high-frequency spaceborne passive radiometer results compare reasonably well with in situ and surface radar observations for moderate to heavy snow events and are useful to separate snow from liquid precipitation, there are many difficulties associated with the approach. Falling snow detection over land is complex because of diverse background emissivities associated with different land types. Precipitating snow using PMW is particularly difficult to distinguish over a snow-covered or frozen area (Levizzani et al. 2011). The existence of supercooled cloud liquid water may also complicate PMW signatures by causing warm brightness temperatures instead of brightness temperature depressions due to ice particles (Liu and Seo 2013). Also, the physical shape of the snowflakes and their impact on observed brightness temperature is difficult to discern (e.g., Kulie et al. 2010; Skofronick-Jackson and Johnson 2011).

These shortcomings of PMW could be overcome by spaceborne active microwave radars that provide a high-resolution vertical structure of precipitation (Kulie and Bennartz 2009). The Tropical Rainfall Measuring Mission (TRMM), the first satellite that carried a Precipitation Radar to space, was launched in November 1997 (Kummerow et al. 2000). Although TRMM enabled the vertical structure of rain associated with different precipitation systems to be studied, it was not appropriate to detect snowfall because of its low-orbit inclination (35°) and its exclusive tropics and subtropics coverage. Global radar observation of snowfall was possible after the advent of the CloudSat satellite (Stephens et al. 2002), which was launched in 2006 and carried the first millimeter-wavelength (94 GHz) Cloud Profiling Radar (CPR) to space. CPR observations enabled snow-related research globally because of its high radar sensitivity (~−28 dBZ) and high-latitude coverage (82°S–82°N). Liu (2008) derived snow cloud characteristics from CPR observations and investigated global snowfall characteristics. This study defined the snow–rain threshold by using in situ measurements and land station measurements, then converted CPR reflectivity (Ze) to snowfall rate (S) by using modeled ZeS relationships. Kulie and Bennartz (2009) used CPR to study global “dry snow” events, where dry snow is defined as surface snow events occurring at 2-m temperature of 0°C or less to avoid the complications related to partially melted snowflakes. This investigation reported that global snowfall is predominantly composed of light snowfall events with CPR reflectivities below about 10 dBZ. Moreover, the global snowfall associated with different modes such as shallow cumuliform and nimbostratus snowfall was studied by Kulie et al. (2016). However, CPR is a single wavelength radar with nadir-only scanning capabilities observing a strip of ~1.5 km on the earth’s surface for each passing orbit, thus limiting its ability to obtain three-dimensional precipitation structure compared to a scanning radar. Other drawbacks of CPR observations include the difficulties of detecting heavy snowfall due to Mie effects, the upper limit of around 20 dBZ and high attenuation by W-band radar, and fixed diurnal sampling from the sun-synchronous orbit at 0130 and 1330 local time. All of these studies show dual-frequency precipitation radar observations are needed to reduce the uncertainties related to hydrometeor shape and size distribution, since these microphysically induced uncertainties are the major source of error for ZeS relationships (Iguchi et al. 2010; Liu 2008; Kulie and Bennartz 2009).

The Global Precipitation Mission (GPM) core satellite was launched in February 2014 with a primary objective to provide next-generation precipitation observations with high temporal resolution (Hou et al. 2014; Skofronick-Jackson et al. 2017). GPM provides a unique opportunity to detect and measure moderate and high snowfall rates because of its Dual-Frequency Precipitation Radar (DPR) and multichannel GPM Microwave Imager (GMI) and high inclination orbit ~65°. Figure 1 shows a snowfall system case observed from the GPM satellite over Lake Superior (North America, near 47°N, 87°W). This case shows elevated snowfall rates up to ~10 mm h−1 associated with near-surface radar reflectivity values approaching 40 dBZ. This indicates that GPM can be used to detect more intense and heavy snowfall that would be underestimated by the CloudSat satellite because CPR-related reflectivity saturates at ~20 dBZ because of Mie scattering effects (Matrosov 2007; Kulie and Bennartz 2009). Observations from the GPM core satellite with more than 3 yr in orbit provide a unique opportunity to study snow systems as shown in Fig. 1 on a near-global basis. Using 3 years of GPM Precipitation Radar observations, this study aims to address the following scientific questions:

  • How much near-surface light snow is missed by the GPM Ku-band radar when compared to the CloudSat CPR?
  • How much near-surface heavy snow is underestimated by CloudSat CPR when compared to the GPM Ku-band radar?
  • What are the general characteristics of snow precipitation systems observed by GPM Ku-band radar?
  • What is the global geographical distribution of snow systems based on their properties such as sizes, intensities, and storm depths?
  • What is the contribution of the global volumetric snow from large, more intense, and deep snow systems?
  • What are the seasonal and diurnal variations of snow system properties?
  • What is the difference between snow over ocean and land?
  • What is the difference between snow in the Northern Hemisphere and the Southern Hemisphere?
To answer the above questions, general snowfall rates from CloudSat and GPM are first compared in section 2. Section 2 also provides detailed data and methodology descriptions, including definitions and properties of snow precipitation features (SFs) using GPM snow precipitation retrievals. General GPM-derived snowfall rates, snow contributions from SFs of various properties, geographical distributions, seasonal and diurnal variations, and SF land–ocean and hemispheric differences are presented in section 3. A summary is provided in section 4.
Fig. 1.
Fig. 1.

An example of snow precipitation features over Lake Superior. (a) KuPR near-surface reflectivity, (b) KuPR echo-top height, (c) KuPR near-surface precipitation rate, and (d) minimum 89-GHz PCT from GMI with matched swath to KuPR. (e) The cross section of KuPR reflectivity along the black line in (b) is shown, where the solid line is precipitation rate (in 10 mm h−1), the black dashed line is GMI 89-GHz PCT, the red line is 183 ± 7 GHz, the blue line is 166-GHz V channel, and the color fill is the KuPR near-surface reflectivity. The outline of snow features are indicated by black lines in (c).

Citation: Journal of Climate 31, 10; 10.1175/JCLI-D-17-0012.1

2. Data and methodology

The GPM Core Observatory satellite was launched on 27 February 2014. Instruments on board include the DPR, which operates at Ku (13.5 GHz) and Ka (35.5 GHz) bands, and a multifrequency (10–183 GHz) GMI. Increased radar sensitivity (~12 dBZ), dual-frequency capabilities, and a non-sun-synchronous and higher inclination orbit (65°) allow the GPM Core Observatory satellite to detect weaker rainfall and light snowfall in the mid- and higher latitudes compared to the TRMM instrument suite. The high-frequency GMI channels (e.g., 166 and 183 GHz) are also more sensitive to falling snow (You et al. 2017).

Three years (April 2014–March 2017) of Ku-band of precipitation radar (KuPR) version 5 data are used in this study, where the near-surface solid precipitation rate is retrieved from near-surface radar reflectivity observations (Seto et al. 2013). While 3 years does not constitute a true climatological data record, we use the term “satellite climatology” to describe the multiyear GPM data record throughout this study. The solid-only precipitation is referred as “snow or snow precipitation” according to the phase parameter in the KuPR product. The phase state of the hydrometeor is related to radar brightband existence and the temperature of the radar range gate level. For the solid precipitation designation, the phase value has been assigned less than a 100 value as described in Iguchi et al. (2015) and the GPM data product file specification (Precipitation Processing System Team 2014). The KuPR snow retrieval algorithms estimate “near surface” snow (e.g., data bins from about 0.5 to 1 km above ground level) to overcome ground clutter issues, but the near-surface snow might melt before it reaches the surface if ambient atmospheric conditions change appreciably in the layer below the designated near-surface radar bin. To exclude possible melting snow cases, the criteria of surface wet bulb temperature (TW) colder than 1°C are used following Sims and Liu (2015), who suggested that when TW is below 1°C, the probability of precipitation falling as snow is more than 50%. To derive TW, first, 2-m surface temperature, dewpoint temperature, and surface pressure values from the 6-hourly ERA-Interim reanalysis dataset (Dee et al. 2011) are used to estimate the equivalent potential temperature as described in Bolton (1980). Then, these equivalent potential temperatures are used to calculate TW as described in Davies-Jones (2008).

The same criteria of TW colder than 1°C are also deployed to CPR-estimated snow to exclude the possible melting snow cases. Each KuPR and CPR snow pixel is collocated with 6-hourly TW. About 54% (3.4%) of KuPR (CPR) pixels are found with TW greater than 1°C and excluded from this study. Thus, about 39 million snow pixels from 3 years of KuPR observations and 18.5 million “true” surface snow pixels from 4 years of CPR observations are used in this study to compare and estimate snow from KuPR and CPR.

The KuPR is chosen for this study because of its wider swath (245 km) compared to the Ka-band radar (120 km), thus allowing large snow systems to be sampled. This is especially important since most of the snow systems have significant horizontal extension, for example, cold fronts and mid- to high-latitude cyclones. The minimum detectable KuPR reflectivity was designed to be 18 dBZ, similar to the TRMM PR. After 1 year in orbit, however, KuPR observational data analyses have shown that its postlaunch minimum detectability of 12 dBZ has far exceeded prelaunch estimates and is comparable to the KaPR (Hamada and Takayabu 2016).

a. Amount of weak snow missed by GPM KuPR

Although the KuPR can detect snow associated with reflectivities as low as 12 dBZ, a large fraction of snow events are weak and below the KuPR detection threshold (e.g., Kulie and Bennartz 2009; Kulie et al. 2016). Therefore, it is important to quantify the amount of snow missed by KuPR or underestimated by CPR. Here, we use 4-yr CloudSat observations to estimate the amount of weak snow that occurs below the KuPR sensitivity. Note that the temporal resolution of KuPR is ~3.2 days, while CPR is ~16 days. First, the occurrence of near-surface reflectivity and near-surface snowfall rate retrievals from the KuPR are compared with CloudSat CPR. Four years (2007–10) of Cloudsat 2C SNOW-PROFILE (http://www.cloudsat.cira.colostate.edu/data-products/level-2c/2c-snow-profile) and 3 years (April 2014–March 2017) of KuPR pixel-level data (https://pmm.nasa.gov/data-access/downloads/gpm) are analyzed over four different mid- to high-latitude regions.

Uncertainties due to different resolutions and interannual variability

Since the CPR footprint resolution (1.4 km cross track and 1.7 km along track) is much smaller than KuPR resolution [~(5 × 5 km2)] at nadir, the primary step should be making them comparable in a spatial scale. One possible approach is to upscale the CPR pixels to make them comparable with the KuPR pixels. The upscaling approach has been used by Sun et al. (2006) and Stephens et al. (2010) by averaging the CPR pixels to the required resolutions. The same approach is adapted by Behrangi et al. (2012) to compare oceanic CPR and TRMM KuPR rain pixels. Because of the different sensitivities of the TRMM KuPR and CPR, the CPR footprint resolution was reduced by averaging 3, 5, and 11 contiguous pixels. This allowed for a better comparison between the two radars with comparable footprints. Here, we follow the method of Behrangi et al. (2012) and calculate the running mean of snowfall rate and reflectivity factor at 3, 5, 7, and 9 contiguous CPR pixels. While performing the running mean of snowfall rate and reflectivity factors with different CPR pixels, the sensitivity is found to be very low when compared to the estimated snowfall rates. The spread in the snowfall rates is ~1 mm yr−1 or less, which translates to <1% of uncertainty. The snowfall rates slightly decrease when averaging over more pixels. Because of the low sensitivity in uncertainties and slightly changing snowfall rate values, the 5-pixel running mean methods are used in the actual analysis. The uncertainties in snowfall statistics caused by footprint differences are shown with the value ranges among four different running mean results in the following discussion.

The amount of snowfall from the two datasets in the two time periods considered might be different due to the interannual variability of snowfall. The average snowfall rate of each year has been calculated and compared with the 3-yr average separately over global and hemispheric scales. Note that the global data (here, in this study, global means the region between 65°S and 65°N) are categorized into four regions consisting of 40°–65°S (hereafter referred to as “South or Southern Hemisphere” for brevity) and 40°–65°N (hereafter referred to as “North or Northern Hemisphere”) over land and ocean separately. Each year’s snowfall rate, entire dataset duration averages, and standard deviation method are used to calculate the uncertainties and interannual variability. The KuPR snowfall rate uncertainties on a global (hemispheric) scale are less than 4% (10%). Similarly, snowfall rate uncertainties are also calculated for CPR. The 4-yr CPR estimates have an uncertainty of only 5.6% on a global scale and less than 10% on a hemispheric scale (Table 1). Thus, snowfall estimation statistics presented in this study from KuPR and CPR have an uncertainty of less than 6% for global scales (65°S–65°N) and less than 10% for hemispheric scales (see Table 1 for details). Also, different retrieval algorithms between KuPR and CPR themselves could lead to differences in estimated snowfall rates. With the 10% interannual changes for each region in mind, two datasets of different time periods have been compared assuming that the spectrum of snow systems would not change significantly in these two periods.

Table 1.

Occurrence of snow pixels and mean unconditional snowfall rate over North and South ocean and land from GPM KuPR and CloudSat CPR. The KuPR and CPR snowfall rates are compared to the combined KuPR–CPR snowfall rates to estimate the missing–underestimated snow. The uncertainty percentages are also calculated by using the entire dataset duration average and each year’s average for each global region to address snowfall rate interannual variability.

Table 1.

Figure 2a compares the occurrence of the near-surface snow reflectivity in each 1-dBZ bin from CloudSat and KuPR over both hemispheric land and ocean regions. The orbital granule data from CPR and KuPR have been analyzed over the entire periods considered for each dataset. The pixels with TW colder than or equal to 1°C are binned by either near-surface reflectivity values (in 1 dBZ) or snow rates (in logarithmic snow rate bins). Then, the occurrence of each snow rate or snow reflectivity was found by dividing the sum of snow pixels in each specific bin by the total number of pixels (including both snow and not snow). Since KuPR can detect to only ~12 dBZ, it misses all light snow with reflectivity less than 12 dBZ. CPR can detect reflectivity to its −28-dBZ detection threshold, although the CPR snow algorithm uses −15 dBZ as a minimum cutoff to consider a CPR observation as a snow-producing cloud structure. The CPR reflectivity occurrence starts to decline above 10 dBZ and saturates near 20 dBZ over ocean. Over land, the CPR reflectivity distribution shows values above 20 dBZ, including a secondary peak near 23 dBZ. These elevated CPR reflectivity values greatly exceeding 20 dBZ are most likely ground clutter contamination over high terrain mountainous regions (Kulie and Bennartz 2009). On the other hand, KuPR reflectivity has a maximum occurrence at ~21 dBZ for both land and ocean and can detect snow above 40 dBZ. Some of these high values of reflectivity might be from wet snow or possibly from ground clutter contamination in complex terrain, although the KuPR dataset does not contain obvious secondary reflectivity peaks at high reflectivity values like the possible clutter contaminated CPR land dataset displays. CloudSat snow occurrence estimates over the South ocean are significantly higher than over the North ocean, but KuPR estimates snow occurrence over both hemispheric oceans are almost identical, with the North ocean displaying a slightly higher frequency of occurrence for more intense snowfall events with near-surface reflectivity >22 dBZ. It is worth noting that the respective KuPR and CPR occurrences intersect when reflectivity is ~(14–15) dBZ for both land and ocean.

Fig. 2.
Fig. 2.

(a) Occurrence of near-surface radar reflectivity at 1-dBZ bins from snow pixels over North and South ocean and land in GPM KuPR (April 2014–March 2017) and CloudSat CPR (2007–10) data. The color-shaded regions correspond to the range (minimum and maximum) values of CPR occurrences obtained from any of the four running average of 3, 5, 7, and 9 pixels. The results of 5-pixel average are presented with dashed lines inside color-shaded regions. (b) As in (a), but with near-surface snow precipitation rate. Note that occurrence is calculated by dividing the sum of snow pixels at specific bin size by the total sample pixels (including both snow and not snow), so that the sum of each curve equals the total snow occurrence over each region respectively. (c) Total snowfall contribution from specific snowfall rates in logarithmic bins size.

Citation: Journal of Climate 31, 10; 10.1175/JCLI-D-17-0012.1

Figure 2b compares the occurrence of snow pixels at different snowfall rates from KuPR and CPR. Note that KuPR estimates snowfall rate by using near-surface radar reflectivity (Seto et al. 2013), whereas 2C SNOW-PROFILE products estimate the vertically resolved probability density function of snowfall rate considering the size, attenuation, and multiple scattering of snow particles. These vertically retrieved profiles are used to estimate near-surface snowfall rate (Wood et al. 2013). Therefore, the KuPR and CPR retrieval algorithms could derive different snowfall rates from the exact same near-surface reflectivity value. Here, snow occurrence is estimated by dividing total snow pixels in each snowfall rate bin by total number of samples. Figure 2b shows that KuPR detects snowfall rates as low as 0.15 mm h−1, whereas CPR can detect down to 0.01 mm h−1. KuPR therefore misses all the lighter snowfall less than 0.15 mm h−1 that comprise a significant number of occurrences in the CPR snowfall rate distribution. However, CPR underestimates frequency of snowfall rates greater than 2.5 mm h−1 over both hemispheric oceans compared to KuPR. However, over land, most CPR snowfall rate occurrences are significantly higher than KuPR up to ~10.5 mm h−1, and CPR snow occurrences are lower than KuPR for higher snowfall rate values. Some of the very high snowfall rate values from the CPR observations correspond to elevated reflectivity values (Fig. 2a), which are likely due to ground clutter.

To demonstrate the amount of total snow volume missed by KuPR, Fig. 2c shows the mean annual snowfall contribution from each snowfall rate bin for both KuPR and CPR. In Fig. 2c, the average annual snowfall contribution (mm yr−1) from each incremental snowfall rate (mm h−1) bin has been calculated for different regions. The summation of the contributions from all the snowfall rate bins represents the global mean unconditional snowfall rate. Note that this is the average annual value of snowfall rate over the entire region during the time period. Although the heavy snowfall events shown in Fig. 2b are less frequent, they contribute significantly to the global snow volume. A peak contribution is found from ~0.8 mm h−1snowfall rate for all the regions except South land (~1.5 mm h−1). The mean occurrence of snow sample and mean unconditional snowfall rate is summarized in Table 1 for both CPR and KuPR radars. Globally (65°N–65°S), KuPR demonstrates 0.6% snow occurrence, whereas CPR observes 3.1% snow occurrence. So, on both hemispheric and global scales, the CPR snow occurrences and mean unconditional snow estimates are much higher than KuPR estimates (Table 1, Fig. 2c). These CloudSat snowfall rate estimates over different regions are close to previous CloudSat estimates. For example, Hiley et al. (2011) estimated about 0.8–1 mm day−1 peak snow [~(300–360) mm yr−1] in the South ocean with the first year of CloudSat data and with a very close magnitude as in Table 1. Kulie et al. (2016) did not present zonal averages, but the magnitudes of the average global snowfall rate are similar to the results reported in this study. Neglecting the diurnal sample bias of CloudSat, the total amount of snow precipitation can be estimated by considering the higher values of snow precipitation from both CPR and KuPR in Fig. 2c. This synergistic CPR–KuPR approach accumulates the higher value of annual snowfall contribution from either CPR or KuPR in each snowfall rate bin [i.e., CPR (GPM) snowfall rates are utilized in the lower (higher) part of the snowfall rate spectrum]. Therefore, estimated total snowfall rate will always be higher than the mean unconditional snowfall rates from GPM or CloudSat, providing that a higher snowfall rate of the two datasets is taken into account.

By combining the lighter snow from CPR and heavier snow from KuPR, the unconditional global (65°S–65°N) total snow rate is about 72.8 mm yr−1, of which KuPR underestimates about 51.8% in total. Over the four defined regions, KuPR systematically underestimates snow over North land (70%) and South ocean (61%). This trend, however, is reduced over North ocean (48%) and South land (44%), indicating that snow systems over the four regions contain different properties; this could be due to different atmospheric/surface conditions. Note that the above comparisons between KuPR and CPR are based on two different retrieval algorithms, which themselves could lead to differences in the snowfall rates, all else being equal as well as the snow detection criteria employed by both retrievals.

The global mean unconditional snowfall rate estimated by CloudSat is 71.1 mm yr−1, which means CloudSat CPR underestimates about 2.3% of the estimated total snowfall rate (72.8 mm yr−1; Table 2). South Ocean has relatively shallow light snow, with a significant amount of snow occurring with CPR reflectivities less than 12 dBZ. CPR underestimates only about 2.1% of the total estimated snow, and KuPR misses about 61.7% of snow over South ocean (Table 2). CPR underestimates a significant amount of snow over North ocean (6.3%) and South land (13.4%) because of the high occurrence of intense reflectivities (KuPR dBZ > 20 dBZ) and heavy snowfall rates (>1 mm h−1) over these regions (Figs. 2a,b). Considering ocean and land, CPR underestimates 1.5% over the Northern Hemisphere and 2.1% over the Southern Hemisphere. This indicates that although the CPR reflectivity saturates at ~20 dBZ, the CPR snow algorithm still estimates a reasonable amount of heavy snowfall despite complicating factors like increased W-band attenuation. CPR underestimates only a small fraction (2.3%) of heavy snow.

Table 2.

Global and regional population of SFs. Note that these populations include only those SFs that have ERA-Interim wet bulb temperature (TW) below 1°C.

Table 2.

b. Definition of snow precipitation features

Precipitation features (PFs) from the GPM KuPR have already been used to demonstrate properties of global precipitation systems and extreme convection (Liu and Zipser 2015; Liu and Liu 2016). To describe the properties of snow precipitation systems, an approach similar to that used to define PFs from the TRMM and GPM observations (Liu et al. 2008; Liu 2016) has been used to define SFs, more specifically by grouping the contiguous area of nonzero solid precipitation observed by the GPM Ku-band radar. Note that PFs or some precipitation systems consist of both liquid and solid precipitation areas. SFs are only defined over the solid precipitation area under such circumstances for all available times. This definition may produce unrealistic snowfall results because initially frozen particles could melt on the way down to the surface. To ensure “near surface” solid precipitation is still solid at the ground, the surface wet bulb temperatures derived from ERA-Interim reanalysis (Dee et al. 2011) data have been used. Each SF has been collocated with the ERA-Interim reanalysis data and validated with TW colder than or equal to 1°C following Sims and Liu (2015). A large proportion of SFs with phase value indicating snow at the near surface have TW values exceeding 1°C and are excluded. The characteristics of SFs are then summarized, including geocenter location, total volumetric snow (sum of the instantaneous snow rate times snow area), maximum near-surface KuPR reflectivity, maximum echo-top heights, maximum echo-top heights of 20 dBZ, and minimum 89-GHz polarization corrected temperature (PCT; Spencer et al. 1989). During 3 years of GPM observations, 3.53 million SFs are defined (Table 2). In this study, to remove noisy signals and ground clutter contamination in the reflectivities associated with high terrain and mountainous regions, SFs with a center location at elevations higher than 2 km are excluded. Also, to exclude some of the artifacts in GPM datasets, SFs with maximum echo tops greater than 8 km are excluded from this study (details to be revisited in section 3f). After excluding these SFs, ~3.47 million SFs are used in this study.

3. Results

a. General climatology of snow from KuPR

Using 3 years of GPM Ku-band products, the geographical distribution of mean unconditional near-surface snowfall rate (excluding the undetectable light snow) and their fractions to the total precipitation are shown on 0.25° × 0.25° grids in Figs. 3a and 3b, respectively. Note that near-surface snowfall rate is derived from the reflectivity at the lowest reliable altitudes above the surface without ground clutter contamination. The altitude could vary from about 500 m above ocean surface up to 2 km above mountain regions. The highest snowfall rates (>270 mm yr−1) are mainly found over oceanic regions such as the North Atlantic Ocean, North Pacific Ocean near the Sea of Japan, the Gulf of Alaska, and Southern Hemispheric oceans between 50° and 60°S. Over land, mainly high mountain regions such as the Himalayas, Rockies, and Andes experience the highest snowfall during the GPM observations. The global mean unconditional near-surface snowfall rate over the 65°S–65°N domain is estimated as 35.1 mm yr−1 whereas the mean unconditional near-surface precipitation rate over the same domain is estimated as about 940 mm yr−1. Note that the mean unconditional snow rate is calculated by dividing total snow accumulation by the total number of samples. The fraction of global volumetric snow in each region is listed in Table 3. It is found that oceanic regions have the highest volume of snow contribution with ~81% (South ocean ~65% and North ocean ~16%), and continental regions contribute only ~16% of the global snow volume. With an extremely small areal coverage compared to other defined regions, South land contributes only less than 0.1% global snow volume.

Fig. 3.
Fig. 3.

Geographical distribution of (a) mean unconditional KuPR near-surface snowfall rate (mm yr−1) and (b) percentage of precipitation as snow in each 0.25° × 0.25° box. Note that the mean unconditional snowfall rate is calculated by dividing total snow accumulation by total samples (snow and no snow) in each 0.25° × 0.25° box. (c) Number of KuPR snow features in each 1° × 1° box during April 2014–March 2017.

Citation: Journal of Climate 31, 10; 10.1175/JCLI-D-17-0012.1

Table 3.

Fraction of features with sizes >10 000 km2, maximum echo-top heights >5 km, and maximum near-surface reflectivity >30 dBZ and their contribution to the total snow over 65°S–65°N and four subregions.

Table 3.

Figure 3b shows the fraction of the near-surface precipitation that falls as a snow. Here, the fraction is determined by dividing the snowfall rate by precipitation rate in each 0.25° × 0.25° grid box. Consistent with Liu (2008), in the Northern Hemisphere, the higher fractions (>50%) of precipitation as snow are found over specific geographical locations, such as the Gulf of Alaska, North Atlantic Ocean, Himalaya region, and the Sea of Japan. In the Southern Hemisphere, the snow fraction increases toward high latitudes, and most of the precipitation (~75%) falls as snow at the latitudes south of 50°. Over the high mountain regions, near-surface precipitation tends to be in the form of snow, as shown in Fig. 3b. This is likely due to the defined near-surface height being located near or above the freezing level; it is possible the actual precipitation at the surface is in liquid form, especially over mountains in summer. Also, the areas of large snowfall rate correspond to the areas of high snow fractions, which is consistent with what was reported by Liu and Curry (1997). Figure 3c shows the geographical distribution of the KuPR SFs at 1° × 1° grid boxes. All the features above mountain regions are not considered, and most of the features are associated with snow precipitation at the ground.

b. Snow features of different sizes

To analyze the snow contributions from different SF sizes, histograms of regional volumetric contribution to global snow over each region are categorized by SF size and presented in Fig. 4a. The South ocean has a large proportion of total snow from larger SF sizes (40 000–80 000 km2), which accounts for ~9% of the total global snow. Another peak is also found for relatively smaller feature sizes (~300 km2), which accounts for ~5.5% of global snow. These results are consistent with small size shallow cumuliform and large size nimbostratus snow systems found by Kulie et al. (2016). Other regions also have the maximum contribution from larger features greater than 10 000 km2 in areal extent. The locations of the snow features categorized by their sizes are shown in Fig. 4b with different colors. Note that the primary focus is to highlight the extreme cases, so the majority of the SFs that are smaller in size might be located underneath the larger ones. The majority of large SFs are found over oceanic regions such as the Gulf Stream, North Atlantic Ocean, Sea of Japan, Bay of Alaska, and over the higher latitude Southern Ocean. The North Atlantic coastal winter snow storms mainly form near 35°N in the Atlantic Ocean and track northward by following the coastal path (Frankoski and Degaetano 2011). The storm tracks over the North Atlantic reach a maximum during January–February, while over the North Pacific, storm tracks reach a maximum during fall and spring (Chang et al. 2002). Over the high-latitude South ocean, large snow systems are evenly distributed from east to west. These extensive snow systems over mid- and high-latitude ocean are consistent with locations of extensive precipitation systems that were reported by Liu and Zipser (2015). Over land, the sizes of features are relatively larger compared to their oceanic counterparts. Additionally, large snow features are observed over specific regions, such as the east-central United States, eastern Russia, the western coast of Canada, and the eastern part of China. The common feature about both hemispheres is that snow systems with large sizes tend to occur at higher latitudes. Over South (North) Ocean, the fraction of SFs with area >10 000 km2 is only 0.34% (0.26%), but they contribute 44% (35%) of the volume snow, respectively (Table 3). The small fraction of large systems contributing a large proportion of total snow is also found over North and South land.

Fig. 4.
Fig. 4.

(a) Percentage contribution of total volumetric snow by SFs of different sizes. (b) The locations of SFs categorized by size and rarity of the events is represented by colored symbols. (c) Contribution of SFs with size >10 000 km2. The fraction is calculated in 2° × 2° boxes.

Citation: Journal of Climate 31, 10; 10.1175/JCLI-D-17-0012.1

The geographical distributions of the fraction of total snow from SFs greater than 10 000 km2 are shown in Fig. 4c in each 1° × 1° box. These large fractions are geographically locked in the Northern Hemisphere and are observed mainly over the central and East Coast portions of the United States, central Europe, eastern Russia, and the northeast coast of China. Although, the population of larger SFs is relatively less over the United States (Fig. 4b), the fraction of SFs with size greater than 10 000 km2 is high (Fig. 4c). Over the Southern Hemisphere, this fraction increases toward higher latitudes. It is interesting to note that the low fraction of snow from large systems over the Southern Ocean southwest of Chile coincide with sea surface temperatures that are relatively warmer at that latitude compared to surrounding regions.

c. Snow features of different intensity

The global volumetric snow contribution based on KuPR maximum near-surface reflectivity in each 2-dBZ bin size (Fig. 5a) shows that contributions of total snow peak from features with maximum reflectivity ~(28–30) dBZ. Globally, about 5.1% of SFs have maximum reflectivity greater than 30 dBZ and they contribute about ~49% total volumetric snow (Table 3). Over the Northern Hemisphere, land contributes more snow volume for relatively less intense (<20 dBZ) and extremely intense (>40 dBZ) features whereas oceanic regions dominate for the moderately intense features (28–38 dBZ).

Fig. 5.
Fig. 5.

As in Fig. 4, but by SFs with different maximum KuPR near-surface reflectivity.

Citation: Journal of Climate 31, 10; 10.1175/JCLI-D-17-0012.1

The geographical distribution of features based on their maximum near-surface reflectivity is presented in Fig. 5b. The most intense SFs are found more frequently over the Northern Hemisphere versus the Southern Hemisphere. It is worth noting that the SFs with maximum reflectivity greater than 40 dBZ also have significant global snow contribution (Fig. 5a). These extremely intense snow features (>40 dBZ) are found mainly in Northern Hemisphere regions such as the northeastern United States, Russia, the North Atlantic Ocean, and the Sea of Japan. The Southern Hemisphere experiences relatively intense features at latitudes higher than 55°S.

The fraction of the features with high reflectivity value (>30 dBZ) in each 1° × 1° grid box is shown in Fig. 5c. In the Southern Hemisphere, a distinct band of high fraction (~20%) between 45° and 50°S is present all the way from east to west across this latitudinal belt. In the Northern Hemisphere, the high fraction of most intense features is observed mainly over the central North Atlantic Ocean, the eastern part of the Sea of Japan, the east-central regions of China, and the southern and eastern parts of the United States. It is interesting to note that snowfall fraction near the 45°S latitudinal belt is relatively lower (Fig. 3b) but fractions of intense features are higher (Fig. 5c). A similar trend of heavy snow was reported by CPR observations (Liu 2008). Note that these regions are the transition between the warm and cold ocean surfaces. Many of the snow systems are likely wet snow or melting snow. For example, it rarely snows over Texas, but when snow occurs it is very likely to be wet snow because of high near-surface temperatures. The maxima in Fig. 5c over Texas is caused by a very small denominator (total snow amount). The radar precipitation retrieval algorithm could have difficulties determining the precipitation phase, as well as dealing with possible brightband signals. Therefore, at the moment we are skeptical about some of the heavy snow precipitation rates over these regions. Further validation of these snow systems is needed.

d. Snow features of different depth

Three years of KuPR observations show that about 98% of the SFs have echo-top heights less than 5 km, which is consistent with CloudSat shallow cumuliform snowfall observations (Kulie et al. 2016). The geographical distribution of SFs based on their maximum 20-dBZ echo-top height shows that very few of the snow clouds reach deeper than 8-km echo-top heights (Fig. 6b), which was also reported by CloudSat observations (Liu 2008). We speculate that some of these very deep SFs are related to very heavy snowfall (>2 mm h−1) events. CPR observations demonstrated that some of the deeper cloud layers could produce heavy snowfall events. These deep SFs are mainly found over land regions, especially over the western coast of Canada, the central region and East Coast of the United States, central Europe, the eastern part of China, and high mountainous areas (not shown in Fig. 6). A significant number of deep features are also observed over the South ocean and higher latitudes of the North Atlantic and Pacific Ocean. The majority of the SFs over ocean are shallower when compared to land. Although approximately less than 2% of the SFs have a 20-dBZ echo located above 5 km, they contribute a significant amount of the total snow volume (~30%). Table 3 lists the fraction of deep SFs (echo top >5 km) to the four respective regions and globally. Out of the four regions, the deep SFs over North land contribute the most volumetric snow (~42%). Over the South ocean, maximum snow contributions originate from 2- to 5-km echo tops with distinct echo-top peaks at ~2.5–3 km, which account for ~7% of total snow volume (Fig. 6a). Over the Northern Hemisphere, ocean regions have maximum contribution from ~(2.5 to 3.5) km, whereas land has maximum snow contribution from ~4- to 5.5-km echo-top height—each account for ~2% of the total snow volume (Fig. 6a). The fraction of deep SFs (with 20-dBZ echo top >5 km) is high for both North land (2.3%) and South land (2.2%). The highest fractions of total snow from these deep SFs are found in the central and East Coast regions of the United States (more than 70%; Fig. 6c).

Fig. 6.
Fig. 6.

As in Fig. 5, but by SFs with different maximum KuPR echo-top height.

Citation: Journal of Climate 31, 10; 10.1175/JCLI-D-17-0012.1

e. Diurnal and seasonal variations

To demonstrate the diurnal variation of the SFs, fractions of SFs are calculated by dividing the number of SFs in each 2-h bin by the total number of SFs in each region respectively. Among all the regions, both Northern Hemispheric land and ocean show a relatively stronger diurnal variation than in the Southern Hemisphere, and a distinct early morning peak is observed over the North ocean. We speculate that the early morning peak over the Northern Hemisphere is possibly due to the surface roughness differences between continent and ocean that leads to frictional convergence and/or the land breeze circulation that typically develops early in the morning. In the afternoon, the dewpoint depression increases with increasing of air temperature due to sensible heating, which leads to the minimum snowfall feature during the afternoon. The early morning peak of lake effect snow was also reported by past regional research (Grim et al. 2004; Kristovich and Spinar 2005). Southern Hemispheric land (not shown in Fig. 7 because of insufficient samples) and ocean have weak diurnal variation.

Fig. 7.
Fig. 7.

(a) Diurnal and (b) seasonal variations of population of SFs over North and South land and ocean. The red dashed line in (a) is diurnal variation of SFs over North land including those with ERA-Interim wet bulb temperature warmer than 1°C.

Citation: Journal of Climate 31, 10; 10.1175/JCLI-D-17-0012.1

The red dashed curve in the Fig. 7a shows the diurnal variation of the snow features without applying the surface wet bulb temperature restriction. A distinct early afternoon peak is observed over North land, which is close to the typical diurnal cycle of PFs over the tropics that has been shown by many previous studies (Nesbitt and Zipser 2003; Liu and Zipser 2009). However, the peak over land occurs slightly earlier in the day compared to the tropics. With the additional temperature constraint, this afternoon peak of snow precipitation is removed. This surface wet bulb temperature sensitivity test indicates that the KuPR snow retrieval algorithm needs further validation and improvement to distinguish between solid and liquid precipitation at the surface.

The seasonal variations of SF populations over the respective Southern and Northern Hemispheric ocean and land categories are shown in Fig. 7b. Here, the fractions of SFs are calculated by dividing the number of SFs in each 1-month bin by the total number of SFs for the entire time period and then normalizing the result. The Northern Hemisphere shows a stronger seasonal variation both over land and ocean than the Southern Hemisphere. Over the North ocean during the summer months (JJA), the population of SFs reaches a minimum and is less than 1%. However, the population of SFs increases rapidly and reaches a maximum of ~21% in December and then decreases until midsummer. Over North land, the variation is not strong as over ocean, but two distinct peaks are found in April and October. The population of SFs during these peak months is ~13%. In the Southern Hemisphere, the seasonal variation over ocean is weak and populations of SFs do not vary much except for a slight decrease in summer months (DJF). Over land, the variation is relatively stronger than over ocean and a peak is observed during July. These strong SF seasonal variations over the Northern Hemisphere versus the Southern Hemisphere are probably due to strong sensible heating over a greater land fraction (Field and Heymsfield 2015).

Further analyses of SF seasonal variation over Northern Hemispheric ocean and land are conducted based on KuPR maximum echo-top heights and SF area, which are shown in Fig. 8 as two-dimensional histograms for every 1-month bin size. Over the North ocean during the fall season, SFs start to develop and display relatively deeper vertical structure in December and January (Fig. 8a). During the winter months, most of the SFs have echo tops between 2 and 4 km. During summer, few SFs are observed with relatively shallow features (Fig. 8a). Over the North land, SFs are relatively deeper than over ocean and may reach up to 8 km in all the seasons except in summer (Fig. 8b). The majority of the SFs have shallow (2–4 km) echo tops in late fall and winter seasons and another peak in the spring season. In the summer, although the population of SFs is lower, land experiences more features than ocean that are relatively shallow. Over the South ocean, seasonal SF variations are weaker than over the North ocean (Fig. 8c). Furthermore, SF size over Northern Hemispheric land is relatively smaller than over ocean in all seasons and tends to reach a peak during April (Figs. 8d,e). Over the South ocean, SF sizes are consistent in all months except in austral summer (DJF), where SF sizes are relatively smaller (Fig. 8f). Here, it is worth noting that over the Northern Hemisphere, land SFs are deeper and relatively smaller in size, but land SFs have both shallower echo tops and are smaller in size over the Southern Hemisphere (figure not shown).

Fig. 8.
Fig. 8.

2D histograms of SFs categorized by maximum KuPR echo-top heights and month over (a) North ocean, (b) North land, and (c) South ocean and (d) 2D histograms of SFs categorized by size and month over (d) North ocean, (e) North land, and (f) South ocean.

Citation: Journal of Climate 31, 10; 10.1175/JCLI-D-17-0012.1

f. Properties of snow features

The above results demonstrate that SF properties are different over each hemispheric land and ocean region. Further analysis of SFs based on their properties such as the maximum near-surface reflectivity; maximum echo-top height; size; minimum 89-GHz PCT; and minimum 166 GHz [both horizontal (H) and vertical (V)], 183 ± 3, and 183 ± 7 GHz brightness temperatures (TB) over land and ocean for both hemispheres are conducted separately in this section. Figure 9 shows cumulative distribution frequency (CDF) diagrams of SF properties for all the regions. Figure 9a shows that land SFs are relatively larger than oceanic SFs. More than 98% of the SFs have a size less than 10 000 km2. South ocean and North ocean have no significant difference in size, but South land has relatively larger SF sizes compared to North land, especially for areas smaller than 1000 km2. Also, a majority of the SFs (~93% over ocean and ~85% over land) have a size less than 1000 km2 (Fig. 9a). Figure 9b shows the CDF comparison of maximum reflectivity in SFs. North ocean tends to have more intense features than South ocean. South land has significantly more intense features than North land, likely due to the ground clutter (see discussion below). North land has only 5% of SFs with reflectivity greater than 30 dBZ, while South land has ~15%. Although the fraction of 30-dBZ features is low, they contribute a significant amount of volumetric snow globally (Table 3). SFs over ocean are slightly more intense than over North land and less intense than South land when the maximum reflectivity is below 30 dBZ. This result indicates that both hemispheric oceans tend to have more moderate snowfall than North land and lighter snowfall than South land. However, North land has a larger fraction of SFs with more than 30 dBZ compared to over ocean, which is consistent with the more extreme intense SFs over land in Fig. 5.

Fig. 9.
Fig. 9.

CDFs of SFs (a) by sizes, (b) by KuPR maximum near-surface reflectivity, (c) by maximum echo-top height, (d) by minimum 89-GHz PCT, (e) by minimum 166-GHz V TB, and (f) by minimum 183 ± 7 GHz TB.

Citation: Journal of Climate 31, 10; 10.1175/JCLI-D-17-0012.1

Figure 9c shows that land SFs are deeper than ocean SFs over both hemispheres and only about 1% of the SFs are deeper than 5 km both over land and ocean. Land–ocean differences are also larger over the Southern Hemisphere than over the Northern Hemisphere. There is no significant difference for very shallow features (<3 km) over North and South land. North ocean tends to have deeper SFs than over South ocean.

SFs over the four regions are also analyzed based on minimum 89-GHz PCT and minimum 166, 183 ± 3, and 183 ± 7 GHz GMI TB. Depression of 89-GHz PCT is related to the amount of ice in the column (Vivekanandan et al. 1991) and is used as a convective intensity indicator (e.g., Zipser et al. 2006). The ice scattering signatures clearly show that ocean features are warmer than land features in both hemispheres (Fig. 9d). For example, over the North (South) land, ~65% (80%) of features are colder than 260 K, whereas only ~18% (22%) of North (South) ocean features are below this threshold. Also, the North ocean tends to have warmer PCT than the South ocean (Fig. 9d). At 166 GHz, there are more land SFs with colder minimum TB than over ocean for values below 250 K, which is consistent with stronger ice scattering in land SFs with higher echo tops. However, there are more SFs with minimum 166-GHz TB warmer than 250 K over ocean (Fig. 9e). The reason for this is not clear. One possibility is that land systems have more supercooled liquid water due to more aerosols for cloud condensation nuclei. Relatively more supercooled liquid water present above 250 K increases the TB, thus counteracting ice scattering TB depressions. This supercooled liquid water relative TB warming effect was shown in recent research (e.g., Kulie et al. 2010; Liu and Seo 2013). However, the TB warming cause is currently speculative, and further studies are needed to fully understand it. The land versus ocean contrast in 166-GHz TB is much smaller than land–ocean 89-GHz PCT differences. There is even less land versus ocean differences in 183 ± 7 GHz TB (Fig. 9f) since these higher-frequency channels are almost surface blind and are influenced less by surface emissivity due to the water vapor masking effect. Thus, the land versus ocean 89-GHz PCT and 166-GHz TB contrasts are likely influenced by surface emissivity (You et al. 2017).

Further comparisons of SF properties over land and ocean are shown in two-dimensional cumulative histograms in Figs. 10 and 11. In both hemispheres, ocean SFs tend to be shallower and less intense, indicated by lower maximum reflectivity (Figs. 10a,b) and warmer 89-GHz PCT (Figs. 10c,d) than over land. Most of the oceanic features have echo-top heights below 4 km, and most intense features (>35 dBZ) have echo tops greater than 3 km (Fig. 10a). Southern Hemispheric land has both deeper and more intense features than over ocean. It is worth noting that KuPR reflectivity never exceeds 35 dBZ for very shallow (<2 km) SFs for all the regions except for South land, where many intense features are associated with very shallow echo-top heights. More than 20% of the SFs over South land have reflectivities greater than 30 dBZ and are mainly concentrated in the 2–4-km echo-top height category (Fig. 10b). Those SFs are found in the southern part of the Andes (figure not shown), and we speculate that these features are from ground clutter. Most of the shallow features have relatively colder PCT over land than over ocean (Figs. 10c,d). The Southern Hemisphere (both land and ocean) tends to have warmer PCT and shallower echo tops compared to the Northern Hemisphere. This trend could be from ice scattering effects, but more likely results from the surface emissivity differences between land and ocean.

Fig. 10.
Fig. 10.

(a) Cumulative 2D histogram of the SFs over (a) North and (b) South land (contours) and ocean (color filled) as a function of maximum near-surface reflectivity and maximum echo-top height. (c),(d) As in (a),(b), but as a function of minimum 89-GHz PCT and maximum echo-top height.

Citation: Journal of Climate 31, 10; 10.1175/JCLI-D-17-0012.1

Fig. 11.
Fig. 11.

(a) Cumulative 2D histogram of the SFs over North land (contours) and ocean (color filled) as a function of maximum echo-top heights and minimum 166-GHz (H) TB. (b) As in (a), but as a function of maximum echo-top heights and minimum 166-GHz (V) TB. (c) As in (a), but as a function of maximum near-surface snow rate and minimum 166-GHz (H) TB. (d) As in (a), but as a function of maximum near-surface snow rate and minimum 166-GHz (V) TB.

Citation: Journal of Climate 31, 10; 10.1175/JCLI-D-17-0012.1

Further land and ocean differences can be illustrated by using GMI high-frequency channels and various SF properties. Figures 11 and 12 show Northern Hemisphere 166- and 183-GHz TB results combined with SF properties. At 166 GHz, land SFs have some cases with horizontal polarization TB colder than 200 K, which corresponds to the stronger ice scattering in systems with high radar echo tops (Fig. 11a). However, for shallow SFs with radar echo top around 2–3 km, many land SFs have TB around 250–260 K, which is warmer than most of the oceanic SFs (around 230–240 K; Fig. 11a). Vertical 166-GHz polarization TB land and ocean SF distributions are relatively similar, with the exception of the colder SFs when echo tops are higher (Fig. 11b). The 166-GHz horizontal and vertical polarization TB differences may imply a different shape of snow particles between land and ocean. We speculate that land SFs could have more flat ice particles than ocean SFs, but this land versus ocean microphysical composition difference needs further validation in future work. For the same near-surface precipitation rate, land SFs have a broader range of minimum 166-GHz TB than over ocean SFs (Figs. 11c,d). It is interesting to note that for extreme snow rates (>5mm h−1), land SFs tends to have warmer 166-GHz TB than ocean SFs. The reason for this trend is unknown, but it could be related to supercooled liquid water as previously discussed. However, this is just a speculation, and further study and observations are warranted. Near 183 GHz, land and ocean differences are relatively smaller than at 166 GHz. Some features are the same, including the colder TBs for land systems with higher radar echo tops (Figs. 12a,b). Ocean SFs tend to have colder TBs for those with extreme snow rates (>5 mm h−1) compared to over land. Because the 183-GHz microwave channels are less sensitive to the surface type, these land and ocean contrasts should mostly result from the ice scattering signature.

Fig. 12.
Fig. 12.

(a) Cumulative 2D histogram of the SFs over Northern Hemispheric land (contours) and over ocean (color filled) as a function of maximum echo-top heights and min 166 GHz (H) TB. (b) Same as figure (a) but as a function of maximum echo-top heights and min 166 GHz (V) TB. (c) Same as figure (a) but as a function of maximum near-surface snow rate and min 166 GHz (H) TB. (d) Same as figure (a) but as a function of maximum near-surface snow rate and min 166 GHz (V) TB.

Citation: Journal of Climate 31, 10; 10.1175/JCLI-D-17-0012.1

Finally, the sizes of SFs over North land are relatively larger, slightly less intense (Fig. 13a), and deeper than over ocean (Fig. 13c). The largest snow systems (10 000 km2) tend to have deeper echo tops (Figs. 13c,d) and larger maximum reflectivity (Figs. 13a,b) over both hemispheric land and ocean. For all the large SFs (>2000 km2), values of maximum reflectivity and echo top are never lower than 20 dBZ and 2 km, respectively.

Fig. 13.
Fig. 13.

(a) Cumulative 2D histogram of the SFs over North land (contours) and ocean (color filled) as a function of maximum echo-top heights and minimum 183 ± 3 GHz TB. (b) As in (a), but as a function of maximum echo-top heights and minimum 183 ± 7 GHz TB. (c) As in (a), but as a function of maximum near-surface snow rate and minimum 183 ± 3 GHz TB. (d) As in (a), but as a function of maximum near-surface snow rate and minimum 183 ± 7 GHz TB.

Citation: Journal of Climate 31, 10; 10.1175/JCLI-D-17-0012.1

4. Summary

The GPM Core Observatory satellite with its KuPR provides a unique opportunity to observe snow precipitation globally. Compared to its predecessor (TRMM), the GPM highly inclined orbit (65°) and improved sensitivity of the KuPR help to detect and observe global snowfall. Using 3 yr of GPM Ku-band radar observations, properties of snow systems are summarized by snow features of different sizes, intensities, depths, and volumetric snow. The main findings are listed as follows:

  • The GPM KuPR has a minimum detectable signal of 12 dBZ and misses a significant amount of weak snow compared to CloudSat. However, it provides a relatively better estimate of heavy snow information, which CloudSat likely underestimates. Because of attenuation and Mie scattering effects, CloudSat underestimates heavier snow associated with reflectivities greater than ~14 dBZ when compared to KuPR reflectivity distributions. The occurrence of KuPR snow pixels and the mean unconditional annual snowfall rate estimated by KuPR are low compared to CPR over all the hemispheric and global regions. In general, KuPR underestimates about 52% of total snow between 65°S and 65°N compared to a global snowfall estimate created from a combined CPR and KuPR snowfall dataset.
  • Although CloudSat W-band radar underestimates a significant fraction of intense reflectivities (>14 dBZ) and saturates at ~20 dBZ due to Mie scattering effects for large snow particles, the snowfall retrievals underestimate only 2.3% of snowfall when compared to the combined KuPR–CloudSat estimate. Specifically, CPR underestimates about 6.3% (13.4%) of snow over North ocean (South land) because of frequent heavy snowfall rates observed over those regions.
  • Geographical distributions of snow show that large annual snowfall rates (>270 mm yr−1) occur over some oceanic regions such as the North Atlantic Ocean, the North Pacific Ocean near the Sea of Japan, the Gulf of Alaska, and many oceanic regions located between 50° and 60°S in the Southern Hemisphere. The fraction of snow precipitation over the Southern Hemisphere increases consistently toward the pole, whereas over the Northern Hemisphere, a high fraction of snow precipitation is geographically oriented. Both general hemispheric snowfall patterns are consistent with CloudSat observations.
  • The important findings from KuPR snow features are, although only a small percentage of the snow features have larger sizes greater than 10 000 km2 (0.35%), maximum near-surface reflectivity above 30 dBZ (5.1%) and echo top above 5 km (1.66%); they contribute a large proportion (40%, 49.5%, and 30.4%) of the global volumetric snow, respectively.
  • The extensive SFs (with size greater than 10 000 km2) are mainly observed in the U.S. East Coast, central Europe, the northeast coast of China, and high-latitude ocean regions in the Southern Hemisphere. In the Northern Hemisphere, the locations of the most intense SFs (with maximum near-surface reflectivity greater than 30 dBZ) are almost similar with extensive SF locations. However, in the Southern Hemisphere, a distinct band of high total snow fractions from extensive SFs are observed between the 45° and 50°S latitudinal belt. The SFs with maximum echo top deeper than 5 km are found in central and East Coast regions of the United States, the west coast of Canada, the eastern part of China, and high latitudes of the North Atlantic, Pacific, and Southern Hemispheric oceans.
  • Northern Hemispheric snow systems have relatively stronger diurnal variation than the Southern Hemisphere, with a distinct peak in the early morning. Also, Northern Hemispheric snow systems show strong seasonal variations in comparison to the Southern Hemisphere. Most of the snow precipitation occurs during late fall and winter seasons.
  • Snow features with larger sizes are observed mainly over land in both hemispheres. Over land, relatively cold, deep, and intense features are observed. Over the Southern Hemisphere, although the snow systems are larger than the Northern Hemisphere, they are relatively shallow and less intense. Most of the intense features from GPM KuPR are found between 2- and 5-km cloud-top heights, and these features are relatively small in size. Interestingly, these more intense features occur more often over midlatitudes (~45°) than the higher latitudes.
  • At high-frequency PMW channels, it is clear that the ice scattering signal associated with TB depressions is related to the depth of the snow system. Shallow snow systems over land have relatively warmer TB at horizontal polarized 166 GHz than over ocean, but land and ocean 166-GHz vertical polarization TB distributions are similar. This result implies important snow microphysical property differences over land versus ocean.

While the above findings are compelling and illustrate the scientific worth of the GPM dataset for global snowfall feature characterization, a few KuPR-related issues should be highlighted. First, the KuPR can detect precipitation events associated with a radar reflectivity down to about ~12 dBZ. Therefore, a large portion of the lighter snow systems and their associated SFs are not detected by KuPR and are therefore not discussed in this study. This study also uses radar reflectivity and retrievals from single-frequency (Ku band) radar observations to estimate snowfall rates. Dual-frequency observations potentially offer more robust quantitative snowfall rate retrievals than single-frequency retrievals, but the larger-swath Ku band observations and retrievals were deemed more useful for the current study to most effectively create the snow features database. Ground clutter contamination poses another daunting challenge for spaceborne radar. Although the KuPR radar algorithm uses a reliable near-surface bin to overcome most ground clutter issues, there is also large uncertainty whether this near-surface snow reaches the ground or melts before it hits the ground under a wide array of possible meteorological conditions. Additionally, possible biases and errors present in ERA-Interim wet bulb surface temperature datasets and interpolation from 6-hourly products to SFs could introduce uncertainties in the snowfall results. The wet bulb temperature constraint alone might remove some of the actual SFs events. In fact, if wet bulb temperature less than 0°C is used, the total global snow volume from KuPR would be reduced by 35%. Further uncertainty and possible unrealistic results are related to very deep echo tops of up to 12 km and very high radar reflectivity values up to 50 dBZ. Phase detection also complicates the KuPR retrievals, as it is always difficult to separate rain and snow from a single-frequency channel since snow and light rain produce a similar range of reflectivities. The use of the brightband signature as a phase indicator is not promising for convective clouds. These complicating factors and uncertainties can be mitigated by improved future algorithm versions containing further focused validation efforts for the standard GPM radar precipitation retrieval algorithm to separate solid and liquid precipitation, including validating snow at ground using the surface temperature. Last, the 3-yr GPM statistics in this study do not take into account the interannual variations in the snowfall. Despite these inherent complications associated with the GPM snowfall data products, GPM provides a unique global snowfall perspective for the 3-yr dataset presented in this study. The extended GPM dataset collected in future years will provide extremely useful information regarding global snowfall as the community builds a true long-term, observational high-latitude precipitation record.

Acknowledgments

This research was supported by NASA Precipitation Measurement Mission Grants NNX16AD76G and NNX16AE21G under the direction of Dr. Ramesh Kakar and NNX16AH74G under the direction of Dr. Erich Stocker. Thanks to the Precipitation Processing System (PPS) team at NASA Goddard Space Flight Center, Greenbelt, MD, for data processing assistance.

REFERENCES

  • Behrangi, A., M. Lebsock, S. Wong, and B. Lambrigtsen, 2012: On the quantification of oceanic rainfall using spaceborne sensors. J. Geophys. Res., 117, D20105, https://doi.org/10.1029/2012JD017979.

    • Search Google Scholar
    • Export Citation
  • Bolton, D., 1980: The computation of equivalent potential temperature. Mon. Wea. Rev., 108, 10461053, https://doi.org/10.1175/1520-0493(1980)108<1046:TCOEPT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Boucher, R. J., and J. G. Wieler, 1985: Radar determination of snowfall rate and accumulation. J. Climate Appl. Meteor., 24, 6873, https://doi.org/10.1175/1520-0450(1985)024<0068:RDOSRA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chang, E. K. M., S. Lee, and K. L. Swanson, 2002: Storm track dynamics. J. Climate, 15, 21632183, https://doi.org/10.1175/1520-0442(2002)015<02163:STD>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Davies-Jones, R., 2008: An efficient and accurate method for computing the wet-bulb temperature along pseudoadiabats. Mon. Wea. Rev., 136, 27642785, https://doi.org/10.1175/2007MWR2224.1.

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

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ellis, T. D., T. L’Ecuyer, J. M. Haynes, and G. L. Stephens, 2009: How often does it rain over the global oceans? The perspective from CloudSat. Geophys. Res. Lett., 36, L03815, https://doi.org/10.1029/2008GL036728.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ferraro, R. R., F. Weng, N. C. Grody, and A. Basist, 1996: An eight-year (1987–1994) time series of rainfall, clouds, water vapor, snow cover, and sea ice derived from SSM/I measurements. Bull. Amer. Meteor. Soc., 77, 891905, https://doi.org/10.1175/1520-0477(1996)077<0891:AEYTSO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Field, P. R., and A. J. Heymsfield, 2015: Importance of snow to global precipitation. Geophys. Res. Lett., 42, 95129520, https://doi.org/10.1002/2015GL065497.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Frankoski, N. J., and A. T. Degaetano, 2011: An East Coast winter storm precipitation climatology. Int. J. Climatol., 31, 802814, https://doi.org/10.1002/joc.2121.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Grim, J. A., N. F. Laird, and D. R. Kristovich, 2004: Mesoscale vortices embedded within a lake-effect shoreline band. Mon. Wea. Rev., 132, 22692274, https://doi.org/10.1175/1520-0493(2004)132<2269:MVEWAL>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hamada, A., and Y. N. Takayabu, 2016: Improvements in detection of light precipitation with the Global Precipitation Measurement Dual-Frequency Precipitation Radar (GPM DPR). J. Atmos. Oceanic Technol., 33, 653667, https://doi.org/10.1175/JTECH-D-15-0097.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Han, Y., and E. R. Westwater, 2000: Analysis and improvement of tipping calibration for ground-based microwave radiometers. IEEE Trans. Geosci. Remote Sens., 38, 12601276, https://doi.org/10.1109/36.843018.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hiley, M. J., M. S. Kulie, and R. Bennartz, 2011: Uncertainty analysis for CloudSat snowfall retrievals. J. Appl. Meteor. Climatol., 50, 399418, https://doi.org/10.1175/2010JAMC2505.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hou, A. Y., and Coauthors, 2014: The Global Precipitation Measurement Mission. Bull. Amer. Meteor. Soc., 95, 701722, https://doi.org/10.1175/BAMS-D-13-00164.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Iguchi, T., S. Seto, R. Meneghini, N. Yoshida, J. Awaka, and T. Kubota, 2010: GPM/DPR level-2. Algorithm Theoretical Basis Doc., 72 pp., https://pmm.nasa.gov/sites/default/files/document_files/ATBD_GPM_DPR_n3_dec15.pdf.

  • Iguchi, T., S. Seto, R. Meneghini, N. Yoshida, J. Awaka, M. Le, V. Chandrasekar, and T. Kubota, 2015: GPM/DPR level-2. Algorithm Theoretical Basis Doc., 68 pp., http://www.eorc.jaxa.jp/GPM/doc/algorithm/ATBD_DPR_2015_whole_2a.pdf.

  • Katsumata, M., H. Uyeda, K. Iwanami, and G. Liu, 2000: The response of 36- and 89-GHz microwave channels to convective snow clouds over ocean: Observation and modeling. J. Appl. Meteor., 39, 23222335, https://doi.org/10.1175/1520-0450(2000)039<2322:TROAGM>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kidd, C., A. Becker, G. J. Huffman, C. L. Muller, P. Joe, G. Skofronick-Jackson, and D. B. Kirschbaum, 2017: So, how much of the Earth’s surface is covered by rain gauges? Bull. Amer. Meteor. Soc., 98, 6978, https://doi.org/10.1175/BAMS-D-14-00283.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kneifel, S., U. Löhnert, A. Battaglia, S. Crewell, and D. Siebler, 2010: Snow scattering signals in ground-based passive microwave radiometer measurements. J. Geophys. Res., 115, D16214, https://doi.org/10.1029/2010JD013856.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kongoli, C., 2003: A new snowfall detection algorithm over land using measurements from the Advanced Microwave Sounding Unit (AMSU). Geophys. Res. Lett., 30, 1756, https://doi.org/10.1029/2003GL017177.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kristovich, D. R., and M. L. Spinar, 2005: Diurnal variations in lake-effect precipitation near the western Great Lakes. J. Hydrometeor., 6, 210218, https://doi.org/10.1175/JHM403.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kulie, M. S., and R. Bennartz, 2009: Utilizing spaceborne radars to retrieve dry snowfall. J. Appl. Meteor. Climatol., 48, 25642580, https://doi.org/10.1175/2009JAMC2193.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kulie, M. S., R. Bennartz, T. J. Greenwald, Y. Chen, and F. Weng, 2010: Uncertainties in microwave properties of frozen precipitation: Implications for remote sensing and data assimilation. J. Atmos. Sci., 67, 34713487, https://doi.org/10.1175/2010JAS3520.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kulie, M. S., L. Milani, N. B. Wood, S. A. Tushaus, R. Bennartz, and T. S. L’Ecuyer, 2016: A shallow cumuliform snowfall census using spaceborne radar. J. Hydrometeor., 17, 12611279, https://doi.org/10.1175/JHM-D-15-0123.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kummerow, C., and Coauthors, 2000: The status of the Tropical Rainfall Measuring Mission (TRMM) after two years in orbit. J. Appl. Meteor., 39, 19651982, https://doi.org/10.1175/1520-0450(2001)040<1965:TSOTTR>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Levizzani, V., S. Laviola, and E. Cattani, 2011: Detection and measurement of snowfall from space. Remote Sens., 3, 145166, https://doi.org/10.3390/rs3010145.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, C., 2016: GPM precipitation feature database. Tech. Doc., 15 pp., http://atmos.tamucc.edu/trmm/data/document/GPM_database_description_1.0_201601.pdf.

  • Liu, C., and E. J. Zipser, 2009: “Warm rain” in the tropics: Seasonal and regional distributions based on 9 yr of TRMM data. J. Climate, 22, 767779, https://doi.org/10.1175/2008JCLI2641.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, C., and E. J. Zipser, 2015: The global distribution of largest, deepest, and most intense precipitation systems. Geophys. Res. Lett., 42, 35913595, https://doi.org/10.1002/2015GL063776.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, C., E. J. Zipser, D. J. Cecil, S. W. Nesbitt, and S. Sherwood, 2008: A cloud and precipitation feature database from nine years of TRMM observations. J. Appl. Meteor. Climatol., 47, 27122728, https://doi.org/10.1175/2008JAMC1890.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, G., 2008: Deriving snow cloud characteristics from CloudSat observations. J. Geophys. Res., 113, D00A09, https://doi.org/10.1029/2007JD009766.

    • Search Google Scholar
    • Export Citation
  • Liu, G., and J. A. Curry, 1997: Precipitation characteristics in Greenland-Iceland-Norwegian Seas determined by using satellite microwave data. J. Geophys. Res., 102, 13 98713 997, https://doi.org/10.1029/96JD03090.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, G., and E. K. Seo, 2013: Detecting snowfall over land by satellite high-frequency microwave observations: The lack of scattering signature and a statistical approach. J. Geophys. Res. Atmos., 118, 13761387, https://doi.org/10.1002/jgrd.50172.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, N., and C. Liu, 2016: Global distribution of deep convection reaching tropopause in 1 year GPM observations. J. Geophys. Res. Atmos., 121, 38243842, https://doi.org/10.1002/2015JD024430.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Matrosov, S. Y., 2007: Modeling backscatter properties of snowfall at millimeter wavelengths. J. Atmos. Sci., 64, 17271736, https://doi.org/10.1175/JAS3904.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mülmenstädt, J., O. Sourdeval, J. Delanoë, and J. Quaas, 2015: Frequency of occurrence of rain from liquid-, mixed-, and ice-phase clouds derived from A-Train satellite retrievals. Geophys. Res. Lett., 42, 65026509, https://doi.org/10.1002/2015GL064604.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nesbitt, S. W., and E. J. Zipser, 2003: The diurnal cycle of rainfall and convective intensity according to three years of TRMM measurements. J. Climate, 16, 14561475, https://doi.org/10.1175/1520-0442-16.10.1456.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Noh, Y. J., G. Liu, E. K. Seo, J. R. Wang, and K. Aonashi, 2006: Development of a snowfall retrieval algorithm at high microwave frequencies. J. Geophys. Res., 111, D22216, https://doi.org/10.1029/2005JD006826.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Noh, Y. J., G. Liu, A. S. Jones, and T. H. V. Haar, 2009: Toward snowfall retrieval over land by combining satellite and in situ measurements. J. Geophys. Res., 114, D24205, https://doi.org/10.1029/2009JD012307.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Precipitation Processing System Team, 2014: Global Precipitation Measurement file specification for GPM products, version 1.08. NASA Doc., 1131 pp., https://storm.pps.eosdis.nasa.gov/storm/filespec.GPM.V1.pdf.

  • Seto, S., T. Iguchi, and T. Oki, 2013: The basic performance of a precipitation retrieval algorithm for the global precipitation measurement mission’s single/dual-frequency radar measurements. IEEE Trans. Geosci. Remote Sens., 51, 52395251, https://doi.org/10.1109/TGRS.2012.2231686.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sims, E. M., and G. Liu, 2015: A parameterization of the probability of snow–rain transition. J. Hydrometeor., 16, 14661477, https://doi.org/10.1175/JHM-D-14-0211.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Skofronick-Jackson, G., and B. T. Johnson, 2011: Surface and atmospheric contributions to passive microwave brightness temperatures for falling snow events. J. Geophys. Res., 116, D02213, https://doi.org/10.1029/2010JD014438.

    • Search Google Scholar
    • Export Citation
  • Skofronick-Jackson, G., J. A. Weinman, M. J. Kim, and D. E. Chang, 2004: A physical model to determine snowfall over land by microwave radiometry. IEEE Trans. Geosci. Remote Sens., 42, 10471058, https://doi.org/10.1109/TGRS.2004.825585.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Skofronick-Jackson, G., and Coauthors, 2017: The Global Precipitation Measurement (GPM) mission for science and society. Bull. Amer. Meteor. Soc., 98, 16791695, https://doi.org/10.1175/BAMS-D-15-00306.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Spencer, R., H. Goodman, and R. Hood, 1989: Precipitation retrieval over land and ocean with the SSM/I: Identification and characteristics of the scattering signal. J. Atmos. Oceanic Technol., 6, 254273, https://doi.org/10.1175/1520-0426(1989)006<0254:PROLAO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Staelin, D. H., and F. W. Chen, 2000: Precipitation observations near 54 and 183 GHz using the NOAA-15 satellite. IEEE Trans. Geosci. Remote Sens., 38, 23222332, https://doi.org/10.1109/36.868889.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stephens, G. L., and Coauthors, 2002: The CloudSat mission and the A-Train: A new dimension of space-based observations of clouds and precipitation. Bull. Amer. Meteor. Soc., 83, 17711790, https://doi.org/10.1175/BAMS-83-12-1771.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stephens, G. L., and Coauthors, 2010: Dreary state of precipitation in global models. J. Geophys. Res., 115, D24211, https://doi.org/10.1029/2010JD014532.

    • Search Google Scholar
    • Export Citation
  • Sun, Y., S. Solomon, A. Dai, and R. W. Portmann, 2006: How often does it rain? J. Climate, 19, 916934, https://doi.org/10.1175/JCLI3672.1.

  • Vivekanandan, J., J. Turk, and V. N. Bringi, 1991: Ice water path estimation and characterization using passive microwave radiometry. J. Appl. Meteor., 30, 14071421, https://doi.org/10.1175/1520-0450(1991)030<1407:IWPEAC>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Weng, F., and N. C. Grody, 1994: Retrieval of cloud liquid water using the Special Sensor Microwave Imager (SSM/I). J. Geophys. Res., 99, 25 53525 551, https://doi.org/10.1029/94JD02304.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wood, N. B., T. S. L’Ecuyer, D. G. Vane, G. L. Stephens, and P. Partain, 2013: Level 2C snow profile process description and interface control document. JPL Doc., 21 pp., http://www.cloudsat.cira.colostate.edu/sites/default/files/products/files/2C-SNOW-PROFILE_PDICD.P_R04.20130210.pdf.

  • Xie, X., U. Lhnert, S. Kneifel, and S. Crewell, 2012: Snow particle orientation observed by ground-based microwave radiometry. J. Geophys. Res., 117, D02206, https://doi.org/10.1029/2011JD016369.

    • Search Google Scholar
    • Export Citation
  • You, Y., N. Y. Wang, and R. Ferraro, 2015: A prototype precipitation retrieval algorithm over land using passive microwave observations stratified by surface condition and precipitation vertical structure. J. Geophys. Res. Atmos., 120, 52955315, https://doi.org/10.1002/2014JD022534.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • You, Y., N.-Y. Wang, R. Ferraro, and P. Meyers, 2016: A prototype precipitation retrieval algorithm over land for ATMS. J. Hydrometeor., 17, 16011621, https://doi.org/10.1175/JHM-D-15-0163.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • You, Y., N.-Y. Wang, R. Ferraro, and S. Rudlosky, 2017: Quantifying the snowfall detection performance of the GPM Microwave Imager channels over land. J. Hydrometeor., 18, 729751, https://doi.org/10.1175/JHM-D-16-0190.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zipser, E., C. Liu, D. Cecil, S. W. Nesbitt, and S. Yorty, 2006: Where are the most intense thunderstorms on Earth? Bull. Amer. Meteor. Soc., 87, 10571071, https://doi.org/10.1175/BAMS-87-8-1057.

    • Crossref
    • Search Google Scholar
    • Export Citation
Save