Low (web) Quality - BALTEX
Low (web) Quality - BALTEX
Low (web) Quality - BALTEX
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
8<br />
Improvement of long-term integrations by increasing RCM domain size<br />
Sin Chan Chou, André Lyra, José Pesquero, Lincoln Alves, Gustavo Sueiro, Diego Chagas, José<br />
Marengo and Vladimir Djurdjevic<br />
National Institute for Space Research - INPE, Cachoeira Paulista, SP, Brasil, 12630-000, chou@cptec.inpe.br<br />
1. Introduction<br />
Global models are used to generate climate change<br />
according to different future scenarios. However, these<br />
models have rather coarse resolution. The nesting of a<br />
Regional Climate Model (RCM) over the area of interest<br />
allows the reduction of grid size which is more desirable for<br />
impact studies. In addition, the climate generated from a<br />
regional model is strongly dependent on the lateral boundary<br />
conditions (LBC) and, consequently, on the domain size or<br />
the position of the LBCs (Xue et al. 2007).<br />
The objective of this work is to show the dependence of the<br />
tropical South America climate from a RCM on the choice<br />
of the position of the lateral boundary conditions.<br />
2. The models<br />
The driver model is the INGV-SXG coupled GCM (Gualdi<br />
et al 2003a,b) which uses ECHAM4 (Roeckner, 1996) as<br />
atmospheric component, OPA 8.2 (Madec et al, 1999) as<br />
ocean model and LIM as the sea ice model (Fichefet e<br />
Morales Maqueda, 1999). The atmospheric resolution is<br />
about T106L19.<br />
The RCM is the Eta Model (Black, 1994; Mesinger et al,<br />
1988; Janjic, 1994) configured with 40-km horizontal<br />
resolution and 38 vertical layers. The model has been used<br />
for weather and seasonal forecasts over South America<br />
(Chou et al, 2005) and was adapted to run long-term<br />
integrations (Pesquero et al, 2009).<br />
A continuous integration for the present climate period from<br />
1961-1990 was carried out, with the lateral boundary<br />
conditions provided at every 6 hours. The sea surface<br />
temperature was taken from monthly values produced by the<br />
AOGCM. The small domain was setup with 135x293x38<br />
points, with the boundaries positioned about 15 degrees<br />
outside the continent in the tropics, and the larger domain<br />
with 201x333x38 points, spanned additional 20 degrees in<br />
the east-west direction. In the second domain, the<br />
Intertropical Convergence Zone can produced by the RCM.<br />
2). The small domain nested run did not show this double<br />
structure, however, the ITCZ rain was very weak, and<br />
therefore rains over Amazon region and northern part of<br />
Northeast Brazil were underestimated. The larger domain<br />
run exhibited the ITCZ rain position more correctly and<br />
the rain amounts over Amazonia and northern part of<br />
Northeast of Brazil have increased. The precipitation in<br />
subtropical areas and southern parts of the continent was<br />
not clearly affected by the increase of the domain size.<br />
Similar results were found in other seasons of the year<br />
The large positive temperature bias occurred over<br />
Amazonia and northern part of Northeast Brazil in the run<br />
using small domain area. These errors may have been<br />
caused by the reduced amount of precipitation. These<br />
positive bias have reduced or changed sign in the larger<br />
domain run. Again, negligible temperature effects<br />
occurred in subtropical or higher latitudes.<br />
The annual cycle of precipitation over Amazonia region<br />
showed substantial improvement in the larger domain<br />
setup. The amounts during rainy months were more<br />
accurate and the peak occurred in the correct month.<br />
These results are independent of LBC schemes, as it has<br />
been shown by Veljovic (2009) that using another scheme<br />
(Davies, 1976) large scale structure is retained.<br />
Figure 1.<br />
Domains of the Eta Model.<br />
Figure 2. 1961-1990 DJF Eta precipitation<br />
(mm/d): small (lhs) and large (rhs) domains.<br />
4. Some Conclusions<br />
The Eta model improved the large scale tropical<br />
precipitation and temperatures by positioning the lateral<br />
boundaries away from the region of interest. Negligible<br />
effects could be noticed outside the tropics.<br />
3. Results<br />
During DJF, the austral summer, the major precipitation<br />
areas over South America continent occur in the Amazon,<br />
northern part of Northeast Brazil, central and southeastern<br />
parts of Brazil. These areas are associated with the<br />
Intertropical Convergence Zone (ITCZ), the South Atlantic<br />
Convergence Zone (SACZ) and frontal passages. The driver<br />
model produced the rain band associated to the ITCZ to the<br />
south of the observed position; a secondary band was<br />
positioned to the north suggesting a double structure (Figure<br />
References<br />
Chou, S.C., Bustamante, J., Gomes, J.L, Evaluation of Eta<br />
Model seasonal precipitation forecasts over South<br />
America. Nonlinear Processes in Geophysics, v.12,<br />
pp.537 – 555, 2005.<br />
Pesquero, J.F., Chou, S.C, Nobre, C.A, Marengo, J.A,<br />
Climate downscaling over South America for 1961-<br />
1970 using the Eta Model, Theoretical and Applied<br />
Climatology, 2009 DOI: 10.1007/s00704-009-0123-z.