Comparing surface energy, water and carbon cycle in dry and wet regions simulated by a land-surface model.

In this study, we analyze results from 47-year (1954-2000) offline simulations using an Australian land-surface model CSIRO Atmosphere Biosphere Land Exchange. We focus on exploring its surface mean climatology, interannual and decadal variations in Australia and Amazonia basin in South America which are distinguished by dry and wet climates respectively. Its skill is assessed by using observational datasets and four model products from the Global Land-surface Data Assimilation System. Surface evaporation and runoff climatologies are satisfactorily simulated, including surface energy and water partitions in dry and wet climates. In the Australian continent dominated by dry climate, slowly varying soil moisture processes are simulated in the southeast during austral winter. The model is skilful in reproducing the nonlinear relationship between rainfall and runoff variations in the southwestern part of the Australia. It shows that the significant downward trend of river inflow in the region is associated with enhanced surface evaporation which is caused by increased surface radiation and wind speed. In its carbon-cycle modeling, the model simulates an upward trend of NPP by about 0.69%/year over the Amazonia forest region in the 47-year period. By comparing two sets of the model results with/without CO variations, it shows that 35% of such increases are caused by changes in climatic conditions, while 65% is due to the increase in atmospheric CO concentration. Given the close linkage between climate, water and vegetation (carbon cycle), this work promotes an integrated modeling and evaluation approach for better representation of land-surface processes in Earth system studies. [ABSTRACT FROM AUTHOR]