Abstract
Energy fluxes important for determining the Arctic surface temperatures during winter in present-day simulations from the Coupled Model Intercomparison Project phase 3 (CMIP3) multimodel dataset are investigated. The model results are evaluated over different surfaces using satellite retrievals and ECMWF interim reanalysis (ERA-Interim). The wintertime turbulent heat fluxes vary substantially between models and different surfaces. The monthly median net turbulent heat flux (upward) is in the range 100–200 W m−2 and −15 to 15 W m−2 over open ocean and sea ice, respectively. The simulated net longwave radiative flux at the surface is biased high over both surfaces compared to observations but for different reasons. Over open ocean, most models overestimate the outgoing longwave flux while over sea ice it is rather the downwelling flux that is underestimated. Based on the downwelling longwave flux over sea ice, two categories of models are found. One group of models that shows reasonable downwelling longwave fluxes, compared with observations and ERA-Interim, is also associated with relatively high amounts of precipitable water as well as surface skin temperatures. This group also shows more uniform airmass properties over the Arctic region possibly as a result of more frequent events of warm-air intrusion from lower latitudes. The second group of models underestimates the downwelling longwave radiation and is associated with relatively low surface skin temperatures as well as low amounts of precipitable water. These models also exhibit a larger decrease in the moisture and temperature profiles northward in the Arctic region, which might be indicative of too stagnant conditions in these models.
