Slide sem t

1
Performance of the HadRM3P model for downscaling
of present climate in South American  Lincoln
Muniz Alves, José A. Marengo Centro de
Previsão de Tempo e Estudos Climáticos
(CPTEC/INPE) 12630-000 Cachoeira Paulista, São
Paulo, Brazil (Contact lincoln_at_cptec.inpe.br)
SKILL OF THE MODEL SIMULATION
SCIENCE GOAL A regional program led by CPTEC/INPE
is CREAS (Regional Climate Change Scenarios for
South America). CREAS represents a collaboration
between the UK-Met Hadley Centre Regional and
various programs from the Brazilian government
funded by GEF. In CREAS, the HadAM3P global model
is used together with the HadRM3P regional model
to downscale climate variability and change in
South America at the resolution of 50 km. In
this poster, we analyze simulations of climate
variability for South America during the present
(1961-1990), at the annual and seasonal levels.
CONTROL CLIMATE SIMULATED BY THE RCM
How far does the RCM diverge from its driving
AGCM (HadAM3P)?
How HadRM3P add value to the AGCM?
OBSERVATION
HadRM3P
HadAM3P
Precipitation (mm/dia)
1983
1985

• SUMMARY AND CONCLUSIONS
• In general the HadRM3P is capalbe of simulating
  the mean climatological features over South
  America
• The HadRM3P resolves features on finer scales
  than the GCM. This is particularly clear for
  precipitation.
• The model is found to represent quite accurately
  the primary features of observed circulation,
  temperature and precipitation patterns, including
  their seasonal cycle and the main modes of
  interannual variability. But, there are
  significant biases.
• The model must be adequalety tuned in order to
  give reliabe for climate change, but there are a
  number of uncertainties and caveats associated
  with the RCMs predictions of climate change over
  South America.

Simulated precipitation, temperature and
atmospheric circulation at 850 and 200 hPa for
DFJ 1983 and 1985. HadAM3P (first column),
HadRM3P (center), Observed (third column).
REGIONAL CHARACTERISTICS AND TIME VARIABILITY
Interannual variability of observed and modeled
normalized rainfall departures during the peak of
the rainy season. Tick black line represents the
mean rainfall from the model ensemble. Thin blue
lines represent each member of the ensemble. Tick
red line shows the observed rainfall
Table - Bias, standard deviation (STD), root mean
square error (rmse) and correlation coefficient
(?) of annual cycle.
Annual cycle of observed (CRU) and modeled
rainfall and temperature in several regions of
SA. Tick orange line shows observations. Thick
black line represents the mean from the model
ensemble. Others colors represent each member of
the ensemble.
Acknowledgements This poster is part of the a
Master Degree Thesis of the first autor. We thank
WMO and CPTEC for partially grants this
conference. Thanks also to the UK-Met Offices
staffs for the valuable assistance. CREAS is
funded by the UK FCO-GOF Program and the
PROBIO-MMA-GEF project (Brazil)