The experimental lines are from
(1) M. Izuha, S. Yamamoto, and S. Saito, 1997, J. Mol. Struct. 413–414, 527.
The present fit employs Watson's S-reduction. With the inclusion of h₃ and d₂^S, the standard deviation drops to 16.5 kHz from 24 kHz reported in (1).
The b-type transitions can not be predicted with confidence, and they are very weak. Therefore, their prediction was omitted. Predictions with calculated uncertainties larger than 0.5 MHz should be viewed with caution. The predictions should be sufficient for astronomical observations toward higher frequencies, whereas some caution is advised at lower frequencies because it is not clear if the hyperfine parameters are known accurately enough.
The dipole moment is derived from an ab initio calculation from
(2) J. Koput, 2002, J. Phys. Chem. A 106, 12067.