Line parameters for the A 2 & ` ! X 2 % bands of OH

Updated line parameters, including line positions and intensities, have been generated for the six strongest A2&`(v@)!X2% i (vA) bands of OH, i.e., v@!vA"0,#1, vA"0, 1, 2, applicable to atmospheric and high temperatures. Results from recent laboratory measurements and theoretical studies have been incorporated into the calculations. Tables of line parameters are available for 296 and 4600 K. ( 2000 Elsevier Science Ltd. All rights reserved. The OH line parameters for the A2&`(v@"0)!X2% i (vA"0) band generated almost 20 years ago by Goldman and Gillis [1] have been used routinely in atmospheric and astrophysical studies (e.g., [2}12] and references therein). Other atmospheric UV studies of OH have used very similar values, e.g., Burnett and Burnett [13], Armerding et al. [14], Notholt et al. [15] and references therein. While UV}VIS line parameters are now a formal part of the HITRAN database [16], these OH lines have not been included yet in the database. The IR OH line parameters have been updated extensively in the recent work of Goldman et al. [17]. The calculations by Goldman and Gillis [1] were based on the spectroscopic constants of Destombes et al. [18] for energy levels and line positions, the relative Einstein coe$cients A(v@J@!vAJA) of Chidsey and Crosley [19], and on the intensity normalization according to the rotationless (N@"0) v@"0 lifetime as measured by German [20], q 0,1@2 "(0.688$0.007)]10~6 s. The line positions were estimated to be accurate to better than 0.1 cm~1. In subsequent, unpublished, calculations, Gillis and Goldman [21] revised the A2&`(v@"0)!X2%(vA"0) line parameters calculations, using the spectroscopic constants of Coxon [22], with the previous [1] intensity normalization. The accuracy of the new line positions 0022-4073/00/$ see front matter ( 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 0 2 2 4 0 7 3 ( 0 0 ) 0 0 0 1 1 X was estimated to be better than 0.01 cm~1. The line parameters for the A2&`(v@)!X2%(vA), (v@, vA)"(1,1), (2,2), (1,0), (2,1), (3,2), were also calculated in that work. According to the recent work of Cageao et al. [8], the Goldman and Gillis (0,0) line parameters remain the adopted database. It was recognized, however, by Cageao et al. [8] that a signi"cant improvement in the (0,0), (1,1), (2,2) line positions was provided by the high-resolution FTS work of Stark et al. [23], but that no signi"cant change in the intensity normalization can be concluded from the available publications. Comparing (0,0), (1,1) and (2,2) line positions of Gillis and Goldman [21] with those lines from Stark et al. [23] which are reported with small experimental error shows that in most cases the agreement is better than 0.004 cm~1 (it was &0.08 cm~1 for the (0,0) line positions of Ref. [1]). With no recent measurements, the estimated accuracy for the (1,0), (2,1) and (3,2) lines remains &0.05 cm~1. The intensity normalizations used for the (0,0) and the additional (v@, vA) lines in Refs. [1,21] is very close to those reported by Copeland et al. [24], who measured relative vibrational band intensities but applied slightly di!erent absolute normalization, and to the update by Luque and Crosley [25]. We thus "nd that the line parameters of Gillis and Goldman [21] should be adequate for atmospheric and astrophysical quantitative modeling. The additional studies of OH and its isotopomers in the A2&`!X2%(v@, vA) bands that became available in the last 15 years (e.g., [24}31]) enable further extension of the database. In particular, the studies by Stark et al. [23] and by Luque and Crosley [25] provide improvements for the line positions and transition probabilities, respectively. The work reported recently by Levin et al. [32] incorporated the same 6 bands and transition probabilities used in Gillis and Goldman [21]. Levin et al. [32] applied the results of the spectroscopic constants analysis by Stark et al. [23], but not the updated transition probabilities of Luque and Crosley [25]. We have been able to duplicate the (0,0), (1,1), (2,2) calculations by Stark et al. [23] and include them in our new line parameters calculations. This required the correction of some typographical errors in the Hamiltonian matrix elements listed in Table 3 of Stark et al. [23]. Most notably, the q term in S2% 1@2 D2% 1@2 T needs to have G2x, not $x (where x"J#1 2 ), and the c@ D for A(2)!X(2) in Table 4 should have been !1.8]10~4, not !1.8]10~5. Furthermore, it was necessary to increase the number of digits in the spectroscopic constants, and these were supplied for the current work by Stark from his original work for the published paper [23]. The corrections needed for the matrix elements were independently recognized by Levin et al. [32], but the published spectroscopic constants were not changed in their work, which was aimed at low resolution spectral modeling. The updated transition probabilities p(v@J@!vAJA) and Einstein-A coe$cients A(v@J@!vAJA) of Luque and Chidsey [25] were obtained via the LIFBASE database provided by Luque and Chidsey [33] and incorporated into our program. Comparison with our previous results [1,21] show similar A-values for most of the transitions. For J[25, the di!erences are in the range of 0.5}5% for main transitions, and 5}20% for satellite transitions. The later increase up to factor of &2 for J&35. The normalization to experiment used in the LIFBASE is essentially unchanged (686 ns instead of the 688 ns quoted above; the average of recent values presented by Cageao et al. [8] is 690$70 ns). The updated intensity parameters (Einstein A and HITRAN S) re#ect the improvement achieved in the calculated transition probabilities, as described by Luque and Crosley [25]. 226 J.R. Gillis et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 68 (2001) 225}230