Low (web) Quality - BALTEX

205
CLARIS project: Towards climate downscaling in South America using
RCA3
Claudio G. Menéndez (1, 2), Anna Sörensson (1), Patrick Samuelsson (3), Ulrika Willén (3), Ulf
Hansson (3), Manuel de Castro (4), Jean-Philippe Boulanger (5)
1. Centro de Investigaciones del Mar y la Atmósfera, CONICET-UBA, Buenos Aires, Argentina,
menendez@cima.fcen.uba.ar
2. Departamento de Ciencias de la Atmósfera y los Océanos, FCEN, Universidad de Buenos Aires, Argentina
3. Rossby Centre, SMHI, Norrköping, Sweden
4. Facultad de Ciencias del Medio Ambiente, Universidad de Castilla-La Mancha, Toledo, Spain
5. Laboratoire d'Océanographie et du Climat, UMR CNRS/IRD/UPMC, Paris, France
1. Introduction
This paper documents coordinated work carried out with the
Rossby Centre regional atmospheric climate model (RCA3,
Kjellström et al., 2005) within the work package on
downscaling in the sub-tropical and mid-latitude South
America of the European Union project ‘A Europe-South
America Network for Climate Change Assessment and
Impact Studies’ (CLARIS, http://www.claris-eu.org).
The goal of this 3-year interdisciplinary project was to build
an integrated European-South American network dedicated
to promote common research strategies to observe and
predict climate changes and their consequent socioeconomic
impacts taking into account the climate and
societal peculiarities of South America. CLARIS was built
on experience obtained through other European Projects
such as e.g. PRUDENCE, MICE and ENSEMBLES and
was, in a more modest way, a counterpart of these projects
in South America. Details on the project and its main results
are summarized in Boulanger et al. (2009).
The work package on downscaling has promoted the coordinated
participation of European and South American
research teams in the use and development of regional
dynamical models and statistical downscaling techniques. In
particular, a collaboration between Rossby Centre (Sweden)
and the Centro de Investigaciones del Mar y la Atmósfera
(Argentina) was established with the purpose of pursuing
climate research in South America using RCA3.
Concerning dynamical downscaling activities within
CLARIS, two multi-RCM ensemble downscaling were
carried-out: (i) Case studies of months with observed
extreme precipitation in south-eastern South America
(hereafter EXP1), and (ii) Multiyear simulations of the
recent present climate (EXP2). Models were run at
horizontal spatial scales of ~50 km and were driven by
reanalysis data ERA-40. In addition, regional climate change
scenarios were generated during the last part of the project
(e.g. Nuñez et al., 2008; Sörensson et al., 2009a).
For details on models participating in CLARIS downscaling
activities the reader is referred to Boulanger et al. (2009) and
Menéndez et al. (2009). More detailed information on RCA3
setup for South America can be found in Sörensson et al.
(2009b).
The first part of this study (section 2) briefly evaluates the
ability of models forced by analyzed boundary conditions
(i.e. quasi-observed) to simulate case studies of intense
precipitation in the monthly time-scale near the Rio de la
Plata (EXP1) and a 10 year period (EXP2). The analysis is
focused on the comparison between RCA3, CLARIS
ensembles and observations. Simulations were evaluated
against high-resolution data compiled by the Climatic
Research Unit (CRU) (New et al., 2000).
In the second part (section 3) a regional climate change
scenario developed for the CLARIS project is presented.
RCA3 is nested into a coupled global model to study the
effects of twenty-first-century climate change over South
America. Three 20-year time-slice simulations were
performed for the periods 1980-1999 (using reanalysis and
global model driving fields) and 2080-2099. We assess the
response of temperature and precipitation both in terms of
seasonal means and changes in distributions for daily and
monthly-to-annual time-scales.
2. Ensemble RCM experiments within CLARIS
The CLARIS ensemble for EXP1 consists of simulations
performed with five RCMs (MM5, PROMES, RCA3,
REMO and WRF) and one stretched-grid global model
(LMDZ). The domain of analysis is restricted to southern
South America where CLARIS is exploring models'
behaviour and sources of uncertainty in more detail. For a
description of the main results of the ensemble of model
simulations over South Eastern South America (southern
Brazil, Uruguay, north-eastern Argentina), including a
comparison with station data and with results from a
statistical downscaling method, we refer the reader to
Menéndez et al. (2009).
The CLARIS ensemble for EXP2 consists of regional
simulations performed with four models (LMDZ,
PROMES, RCA3 and REMO) for the period 1991-2000.
Models' domains are somewhat different from model to
model but include most of South America (the domain of
analysis covers from 50S to the equator and 85W to 35W).
The CLARIS ensemble is able to reproduce the basic
large-scale characteristics of the observed precipitation,
such as the migration of rainfall associated with the South
American monsoon system and the precipitation
maximum associated with the South Atlantic Convergence
Zone. The relatively good performance of the multi-model
ensemble average compared to any one individual model,
already found in previous studies with global models and
with RCMs for other regions, is true in this regional study
over South America as well. Nevertheless, this results
from the cancellation of offsetting errors in the individual
models.
Models have difficulties in capturing accurate mean
precipitation amounts over the Amazon and La Plata
basins (the two main hydrological basins of the region).
RCA3 exhibits a reasonably good agreement with
observations, but underestimates the annual mean rainfall
of south-eastern South America (wintertime is too dry in
this model) and overestimates precipitation in parts of
western and southern Amazonia and along the Andes.
These biases are often also found in other global and