GCM sensitivity experiments with locally modified land surface properties over tropical South America

Ensembles of 1-year-long experiments with a relatively high-resolution ECMWF model were conducted in order to investigate the impact of modified land surface properties on local, regional and large-scale atmospheric circulations. The modifications consisted of changes to land cover and increased albedo over the northern part of South America. In many respects the experimental design resembles the setting of classical deforestation experiments. The local model response to imposed modifications, which includes a reduction in precipitation as well as in evaporation and an increase in surface temperature, was found to be stronger in dry (July–September, JAS) than in wet (January–March, JFM) season, and in the ensemble with higher albedo value. Local drying is discussed in terms of locally generated overturning that resembles a direct thermal circulation. The effects of this circulation seem to be dominant over the reduction in large-scale moisture supply from the adjacent ocean. On large scales, changes to the Pacific branch of the Walker circulation lead, through modified divergent flow, to a tropics-wide impact on precipitation. In addition to South America, the largest changes are seen in the south Pacific convergence zone in JFM, while the impact on the Atlantic inter-tropical convergence zone is stronger in JAS. In the extratropics, there is little change in precipitation. In the upper troposphere, a distinctive teleconnection wave-pattern could be seen in the Pacific/North American region during JFM. A notable feature in the upper-air model response in JAS is a wave train extending from South America, over the northern Atlantic into Europe. With regard to the interaction between the land surface response and model systematic errors, our results suggest that the erroneous shift of the downward branch of the Pacific/South American Walker circulation is likely to be a cause, rather than a consequence, of the rainfall deficit over South America in the model climatology.