Partition function data and impact on retrieval quality for an mm/sub-mm limb sounder

The partition function plays a major role as it mainly governs the dependence of the spectral line strength (or line intensity) on the temperature. A wrong partition function will lead to an incorrect line strength, which will lead to a systematic error in the retrieval. Therefore, a careful investigation of the available partition function data and the sensitivity of the retrieval to this is required. Two partition function data sets, the HITRAN and the JPL, are considered in this paper. In a first step, we investigate the spread of the partition function ratios from the simple approximation values given by theory. We find that the HITRAN partition functions are usually higher than the values quoted by the simple approximation. The JPL partition functions are much closer to the values given by the approximation. Comparing directly the two datasets, with some exceptions, a good agreement is found. Only for 16 molecules (from 66 considered molecules), the deviations in the two data sets are larger than 2%. The retrieval error analysis shows that, for the case of the molecular species with strong signatures, an uncertainty in the partition function is directly translated into a retrieval error of the species in question. However, the uncertainty in the partition function of this species can have a high impact on the quality of weak species retrieval.