Two land surface parameterizations are used to test the impact of land surface processes on the simulations of the South American monsoon. One parameterization is the simple Simplified Biosphere Model (SSiB), which includes explicit vegetation representation and the other is a surface model with two-soil layers (SOIL) with no explicit vegetation parameterizations. Using the NCEP GCM/SOIL and the NCEP GCM/SSiB, we have integrated the models for 12 months from May 1, 1987 to April 30, 1988. In the two simulations, the initial soil moisture and surface albedo are similar. The major differences between these two experiments are the land surface parameterizations and land cover conditions: one with vegetation and the other with only soil layers.
The most substantial differences between the NCEP GCM/SOIL and the NCEP GCM/SSiB are the simulations in the evolution and the spatial distributions of South American monsoon (SAM). The SAM evolves in Central America and then moves towards southeast South America. In the NCEP GCM/SOIL, the development of the SAM is too fast and too strong. In December, instead of moving towards the southeast, the monsoon moves to the northwest. The NCEP/SSiB, on the other hand, correctly simulates the SAM evolution. To understand the mechanisms that contributed to the differences in the simulations, the surface energy and water balances are analyzed.
The NCEP/SSiB produces different spatial distributions of the surface sensible and latent heat fluxes. These differences induce changes in circulation, which produce variations in SAM simulations. In the pre-monsoon stages, land surface processes mainly influence the divergence in Central America. In the monsoon maturation stage, this influence is manifested in its effect on the simulations in the Bolivian Height and in its effect on the South Atlantic Convergence Zone and the equatorial trade winds.
To more accurate simulate the SAM, the Eta/SSiB regional model is also embedded to simulate regional details. Two three-month simulations (one in the dry season and the other in the wet season) with the Eta/SSiB show that the regional model is able to produce better precipitation simulations than the reanalysis data and GCM results.