The Global Moisture Balance and its Variability in the MERRA-2 Reanalysis

Several advancements addressing water cycle issues have been included in the second Modern-Era Retrospective analysis for Research and Applications (MERRA-2). The analysis of water vapor significantly affects the total mass of the atmosphere. In MERRA-2, a constraint has been placed on the water vapor analysis so that it does not create or destroy mass in the global average, so that in the long term, the global average water cycle reduces to a balance of evaporation and precipitation. In any given region, however, the water cycle will also include an analysis increment and the flux divergence of water. In addition, modeled land precipitation biases can have far reaching effects on the subsequent forecasts and climate of the land. In MERRA-2, a bias corrected precipitation data set is used to force the land surface preferentially over the modeled precipitation, when observations are available. This bias correction should aid in the quality of land surface meteorology and fluxes in MERRA-2.

Here, we evaluate several aspects of the MERRA-2 global water cycle in comparison with observations (when available) and other reanalyses, considering the impact of these advancements on the reanalyzed water cycle. Spurious effects of the observing system on the interannual variability of total column water have been significantly ameliorated; likewise, the interannual variability of the transport of water from ocean to land does not exhibit changes depending on the observing system. Precipitation variability is more dependent on ocean evaporation, which is a function of the prescribed SSTs and assimilated surface winds, and so those observations can affect the global water cycle noticeably. In general the water cycle is generally too intense, with too much Ocean evaporation, precipitation and transport, but still within the range of the most recent reanalyzes estimates. Improvements in the boundary layer parameterization are also allowing for the development of stronger gradients in weather systems, which in turn is allowing for the production of more intense precipitation events. An experiment has been run without the AIRS instrument, showing that its influence in the global sense is small, but more significant when considering land and ocean areas separately. The MERRA-2 water cycle shows some distinct advantages over MERRA, but areas for further improvement will also be discussed.