Arctic Cloud–Radiation–Temperature Associations in Observational Data and Atmospheric Reanalyses

Abstract

Associations between cloudiness, radiative fluxes, and surface air temperature in the central Arctic are evaluated from 1) measurements made at Russian drifting ice stations, and 2) atmospheric reanalyses of the National Centers for Environmental Prediction (NCEP) and the European Centre for Medium-Range Weather Forecasts (ECMWF). In the ice station data, cloudiness is associated with an increase of downward longwave radiation in all months and an increase of net (downward minus upward) total radiation from September through March. The surface air temperatures under overcast skies are 6°?9°C higher than under clear skies during September?March, and the differences are even larger when the observations are stratified by wind as well as cloudiness. The warming by the radiative flux enhancement after a transition from clear skies to overcast has a 1?2-day timescale, while the cooling after the transition to clear skies has a somewhat shorter timescale. The NCEP reanalysis exaggerates slightly the association between cloudiness and surface air temperature, while the ECMWF reanalysis shows a considerably weaker association. The maximum cloud-radiative forcing (MCRF), defined as the difference between the ice station measurements of net surface radiation under cloudy and clear skies, ranges from ?59 W m?2 in June to positive values of 20?30 W m?2 in September?March. The annual mean is small but positive, 3 W m?2, despite the approximately three-month summer period of substantially negative MCRF. These findings are consistent with the conventional cloud-radiative forcing obtained in earlier studies using satellite data and one-dimensional models of the Arctic atmosphere and sea ice. Neither reanalysis captures the seasonality of the observationally deduced effects of clouds on surface radiation. The NCEP reanalysis does not capture the seasonality of the actual cloudiness (as defined by the reported cloud fractions), while the ECMWF reanalysis does not show an impact of clouds on the surface solar flux. Issues needing further attention in the model?data comparison are the effects of surface heterogeneities, the characterization of Arctic clouds, the formulational reasons for the discrepancies between the model-derived reanalyses and the observational data, and the implications for model-derived projections of climate change in the Arctic.