Assessment of a Regional Climate for South America: A Dynamical Downscaling Approach

MENENDEZ C.G; SAULO CELESTE; SOLMAN S A; NUÑEZ M N.

Detecting and Modelling Regional Climate Change

Springer-Verlag

Lugar: Berlin; Año: 2001; p. 515 - 523

The current coarse resolution of general circulation models (GCMs) does not provide reliable estimates of precipitation and other variables at the appropriate scales required for regional climate studies over the South American region. To overcome this problem, one possibility is to perform regional climate simulations using lirnited-area models nested in global models. A research effort aimed at the development of this downscaling technique for the region extending from the South Pacific across South America to the western South Atlantic is now under way at the Centro de Investigaciones dei Mar y la Atm6sfera (CIMA), Buenos Aires. The technique consists of using the output of GCM simulations to provide initial driving conditions and time-dependent lateral boundary conditions for regional climate model (RCM) simulations over South America and the adjacent oceans (one way nesting). The singularity of this nesting system is that the global model itself has a stretched, variable horizontal resolution, with the grid irregularly spaced in the meridional direction. This stretched grid is introduced in order to improve resolution in a latitudinal band over the region of interest. Hence, we use a "hybrid" strategy in which a regional model is nested in aglobai variable-resolution model, combining traditional nesting with GCM zoorning, at a relatively low computational cost. This pilot study introduces an initial diagnosis of the capabilities of the RCM for simulating climate in the South American region. Our prelirninary results suggest that the nesting technique is a computationally low-cost alternative suitable for simulating regional climate features. However, before applying this nesting system to problems involving the local response to climate change, additional simulations, tuning of parameters and further diagnosis are clearly needed to represent regional patterns more precisely.