Atmospheric Moisture Transports from Ocean to Land and Global Energy Flows in Reanalyses

Abstract An assessment is made of the global energy and hydrological cycles from eight current atmospheric reanalyses and their depiction of changes over time. A brief evaluation of the water and energy cycles in the latest version of the NCAR climate model referred to as CCSM4 is also given. The focus is on the mean ocean, land, and global precipitation P; the corresponding evaporation E; their difference corresponding to the surface freshwater flux E–P; and the vertically integrated atmospheric moisture transports. Using the model-based P and E, the time- and area-average E–P for the oceans, P–E for land, and the moisture transport from ocean to land should all be identical but are not close in most reanalyses, and often differ significantly from observational estimates of the surface return flow based on net river discharge into the oceans. Their differences reveal outstanding issues with atmospheric models and their biases, which are manifested as analysis increments in the reanalyses. The NCAR CCSM4, along with most reanalysis models, the exception being MERRA, has too-intense water cycling (P and E) over the ocean although ocean-to-land transports are very close to observed. Precipitation from reanalyses that assimilate moisture from satellite observations exhibits large changes identified with the changes in the observing system, as new and improved temperature and water vapor channels are assimilated and, while P improves after about 2002, E–P does not. Discrepancies among hydrological cycle components arise from analysis increments that can add or subtract moisture. The large-scale moisture budget divergences are more stable in time and similar across reanalyses than model-based estimates of E–P. Results are consistent with the view that recycling of moisture is too large in most models and the lifetime of moisture is too short. For the energy cycle, most reanalyses have spurious imbalances of ~10 W m−2 within the atmosphere, and ~5–10 W m−2 in net fluxes into the surface and to space. Major improvements are needed in model treatment and assimilation of moisture, and surface fluxes from reanalyses should only be used with great caution.