Investigation of moisture field assimilation in global reanalysis

The impact of humidity data on a global reanalysis system is investigated. It is found that the inclusion of moisture in the reanalysis system causes the forecast model to produce a systematically wetter atmosphere. As such, the data assimilation tends to dry out the atmosphere by a negative analysis increment. Meanwhile, an analysis cycle without moisture beneficially affects the other first guess variables, especially the mass field. This indicates that the present reanalysis system seems to be inconsistent with the humidity field and that reanalysis without the use of humidity data is, for the most part, able to reproduce dynamic fields that are similar to reanalysis with all of the observation data. Our results also suggest that continuous refinement of the internal physics in the forecast model can be an efficient way to provide an enhanced analysis.