Evaluation of simulated clear-sky longwave radiation using ground-based observations

Surface observations from a tropical ocean and a subarctic land-based site are employed to evaluate the clear-sky surface downwelling longwave irradiance (SDL) simulated using the European Centre for Medium-Range Weather Forecasts reanalysis (ERA). Comparison of simulated clear-sky and observed all-sky SDL highlights coincident periods of irradiance variability on various timescales in both datasets. Measurements during cloudy conditions are subsequently removed using a combination of the measured shortwave and longwave surface irradiances and recorded rainfall. The most reasonable filtering specifications are determined experimentally; clear-sky filtered observations of SDL are compared with corresponding simulated values. The root-mean-square differences between simulated and observed clear-sky SDL are within the observational uncertainty of +/-10 W m(-2). Simulated clear-sky SDL is about 8 W m(-2) more than the measured tropical values. In the subarctic, the simulated clear-sky SDL is less than observed values in the winter and greater than observed values in the summer. The clear-sky SDL differences are explained partially by the differences in ERA moisture profiles and near-surface temperature in comparison with radiosonde ascents. A primary limitation of the radiometric measurements is the lack of information regarding cloud amount; if model-simulated clear-sky fluxes and cloud radiative forcing are to be fully evaluated, it is highly desirable that such information should accompany surface-based radiation data.