A comparison of surface observations and ECHAM4-GCM experiments and its relevance to the indirect aerosol effect

Abstract

The observations of solar irradiance at the surface, total cloud cover, and precipitation rates have been used to evaluate aerosol-cloud interactions in a GCM. Records from Germany and the United States were available for the time period from 1985 to 1990 and 1960 to 1990. The model used here is the European Centre for Medium-Range Weather Forecasts-Deutsches Klimarechenzentrum: Hamburg (ECHAM4) GCM as run for a 5-yr period with a fully coupled sulfur chemistry-cloud scheme by Lohmann and Feichter. Two experiments- one with an annual mean sulfate load of 0.36 Tg S for the preindustrial simulation and one with 1.05 Tg S for the present day simulation were studied. The goal was to confirm indirectly the existence of the indirect aerosol effect by finding indices for a better agreement of observations with the present-day experiment as compared with the preindustrial experiment. The authors were able to draw such a conclusion only for the German data but not for the United States. The model correctly predicts the annual mean total cloud cover in Germany and the United States, whereas global solar radiation is underestimated by 13 W m-2. This deficiency stems from cloudy conditions. Clouds are either optically too thick or the vertical distribution of clouds is erroneous. This is confirmed by the modeled overcast solar irradiance, which is 27 W m-2 lower than observed, whereas, for the clear sky, model and observations agree. Precipitation rates are underestimated by 42% in the United States. The seasonal cycle of the precipitation rate is incorrect in all U.S. regions. The modeled cloud cover is too low over the central United States in July and August, and consequently the solar irradiance exceeds the observations during these months. The opposite occurs in winter, when the model overestimates the cloud cover and thus underestimates solar irradiance. The nonseasonality of vegetation and soil parameters is suggested as a possible cause for these deficiencies. The convective precipitation formation might also contribute to these discrepancies. On the other hand, this drying out effect of the inner continent is not as pronounced in coastal regions, and, in particular, the comparisons for the German grid box provide indications for the validity of the indirect aerosol effect. The modeled annual cloud cover and solar radiation cycles for the present-day aerosol load are in better agreement with observations. Furthermore, the model shows an interesting shift from low-cloud reduction to cirrus formation in spring as a consequence of the indirect aerosol effect, a result that is confirmed by observational data.