Table 1. Definitions of benchmark cases

--------------------------------------------------------------------------------
Case     Ra_0.5    Viscosity        l and m
                   variation    for initial perturbation      Resolution       
--------------------------------------------------------------------------------
BM1A      7e3        1e0             (3,2)                     32x(12x32x32)
BM1B      7e3        1e1             (3,2)                     32x(12x32x32)
BM1C      7e3        2e1             (3,2)                     32x(12x32x32)
BM1D      7e3        1e2             (3,2)                     32x(12x32x32)
BM1E      7e3        1e3             (3,2)                     32x(12x32x32)
BM1F      7e3        1e4             (3,2)                     32x(12x32x32)
BM1G      7e3        1e5             (3,2)                     32x(12x32x32)
BM1H      7e3        1e6             (3,2)                     32x(12x32x32)
BM2A      7e3        1e0             (4,0)+(4,4)               32x(12x32x32)
BM2B      7e3        2e1             (4,0)+(4,4)               32x(12x32x32)
BM2C      7e3        3e1             (4,0)+(4,4)               32x(12x32x32)
BM2D      7e3        1e2             (4,0)+(4,4)               32x(12x32x32)
BM2E      7e3        1e3             (4,0)+(4,4)               32x(12x32x32)
BM2F      7e3        1e4             (4,0)+(4,4)               32x(12x32x32)
BM2G      7e3        1e5             (4,0)+(4,4)               32x(12x32x32)
BM2H      7e3        1e6             (4,0)+(4,4)               32x(12x32x32)
BM3A      1e5        1e0             (4,0)+(4,4)               48x(12x48x48)
BM3B      1e5        1e1             from BM3A                 48x(12x48x48)
BM3C      1e5        3e1             from BM3A                 48x(12x48x48)
BM3D      1e5        1e2             from BM3C                 48x(12x48x48)
--------------------------------------------------------------------------------
Note: 1) Resolution in radial direction (the number outside of the parentheses)
         for our calculations is refined near the top and bottom boundaries. 
      2) Resolution for comparison cases is 48x(6x48x48) for Stemmer et al. {2006},
         102x(102x204) for Yoshida and Kageyama {2004}, where the number outside
         parentheses represents the resolution in radial direction. 
      3) Ratcliff et al {1996} used different resolution for cubic and tetrahedral 
         symmetry cases, as they considered the symmetry of these cases and did not 
         compute them in a full sphere. For cubic symmetry cases, the equivalent 
         resolution is 32, 64, and 128 cells in radial, co-latitude and longitude 
         directions, respectively, and these numbers are 32, 32, and 64 for the
         tetrahedral cases.