Resumen
The sensitivity of the surface hydrology to the parameterization scheme incorporated in three different AGCMs is investigated. It is found that the simulations of the surface hydrology by the NCAR CCMO and GFDL models (which incorporate the bucket type hydrological parameterization) differ markedly from the CCM1B model (which incorporates the Biosphere Atmosphere Transfer Scheme, BATS). The AGCMs which incorporate the simpler parameterizations of the land surface simulate a number of physical phenomena unrealistically. Many of these gross problems in the land surface climatology are eliminated in the CCM1B model which incorporates BATS. When present-day and doubled CO2 scenarios are compared, it is shown that the differences between model results incorporating different surface schemes are greater than the differences between 1 x and 2 x CO2. The implications of these results are that the surface hydrological fields from AGCMs, and their predictions of climatic change at the land surface, are likely to be unreliable. Investigations of the impact of climatic change consequent upon changes in the land surface climatology should employ one of the advanced parameterizations of the land surface coupled into the general circulation model, and develop and validate those components of the atmospheric model, such as the surface radiative forcing, regional precipitation patterns and the diurnal cycle within the planetary boundary layer, that are most important for deriving land surface processes.