Modelling South American Climate and Climate Change

Simulations of anthropogenically induced change in the coupled climate-carbon cycle system suggest that the tropical forest of South America could become unsustainable through the 21st century in response to reduced precipitation and increased air temperature. This thesis examines the ability of the Hadley Centre land-surface model (MOSES2) to simulate observed mass and energy fluxes from Amazon forest and the role of sea surface temperature (SST) change in South American climate change. MOSES2 simulated reasonable long-term mean heat and water fluxes, but did not simulate the observed rate of daytime carbon uptake at two central Amazon sites. This was attributed to a persistent midday reduction in stomatal conductance due to the saturation of the solar radiation response at low light levels. Calibration of the photosynthesis sub-model remedied this error but at the expense of degrading modelled evaporation, and suggested that “big-leaf” leaf- to canopy-level scaling assumptions are inappropriate for these study sites. The revised MOSES2 stomatal parameters had only a weak effect on climate simulated by an atmosphere-only model (HadAM3). HadAM3 was used to examine the contributions of global mean and regional gradients of SST change to simulated future climate change over South America. It was found that regional gradient changes contributed to approximately 72% and 30% of the precipitation reduction and temperature increase, compared to 5% and 50% contributions from SST mean warming. Precipitation reductions corresponded to a deeper, longer dry season, through Hadley and Walker cell anomalies associated with an early withdrawal and delayed return of the Atlantic ITCZ and enhanced Central American monsoon. The strength of precipitation and forest carbon uptake reductions were dependent on the coexistence of El Nino-like tropical Pacific and meridional gradient tropical Atlantic SST changes. The atmospheric response was only partly analogous to that through an observed El Nino event