Regional climate modeling over South America: challenges and perspectives

1
Regional climate modeling over South America
challenges andperspectives

• Silvina A. Solman
• CIMA (CONICET-UBA)
• DCAO (FCEN-UBA)

UMI- IFAECI 2nd Meeting, Buenos Aires.
Argentina April 25-27- 2011
2
Outline

• Why do we need Regional Climate models?
• How well do models represent regional climate
  over South America?
• Main shortcomings and strengths of RCMs over
  South America the CLARIS-LPB contribution.
• Sources of uncertainty in regional climate
  simulations
• Possible research topics

3
Why do we need Regional Climate models?
La información climática a escala regional es
crítica para los estudios de impacto
AOGCM
Regional Climate Model (RCM)
4
Why do we need Regional Climate models?
5
How well do models represent regional climate
over South America?

• CORDEX
• Initiative promoted by the TFRCD /WCRP
• Main goal To Provide a quality-controlled data
  set of RCD-based information for the recent
  historical past and 21st century projections,
  covering the majority of populated land regions
  on the globe.
• To Evaluate the ensemble of RCD simulations.
• to provide a more solid scientific basis for
  impact assessments and other uses of downscaled
  climate information

• CLARIS-LPB
• The EU FP7 CLARIS LPB project
• Main goal To predict the regional climate change
  impacts on La Plata Basin (LPB) in South America,
  and at designing adaptation strategies
• To provide an ensemble of regional hydroclimate
  scenarios and their uncertainties for climate
  impact studies.

6
CORDEX Domains
7
CORDEX South America/CLARIS-LPB
Model Evaluation Framework
Climate Projection Framework
ERA-Interim LBC 1989-2008
A1B Continuous runs Timeslices (2010-2040 and
2070-2100)
Regional Analysis Regional Databanks
Multiple AOGCMs HadCM3-Q0, ECHAM5OM-R3, IPSL
8
CLARIS-LPB coordinated experiments over South
America ERA-Interim boundary forcing
RCM/Institution Country Contact person
RCA/SHMI Sweden Patrick Samuelsson
MM5/CIMA Argentina Silvina Solman, Natalia Pessacg
RegCM3/USP Brazil Rosmeri Porfirio da Rocha
REMO/MPI Germany Armelle Reca Remedio, Daniela Jacob
PROMES/UCLM Spain Enrique Sánchez , R. Ochoa
LMDZ/IPSL France Laurent Li
ETA/INPE Brazil Sin Chou, José Marengo
WRF/CIMA Argentina Mario Nuñez
9
Mean Temperature (DJF) 1990-2006
BIAS
RCMs Ensemble
Warm/cold bias
10
Ensemble spread
DJF
JJA
How large is the ensemble spread?
RATIOspread/IV
11
Temperature Annual cycle
12
Precipitation (DJF) 1990-2006
BIAS
RCMs Ensemble
Wet/dry bias
13
Ensemble spread
DJF
JJA
RATIOspread/IV
14
Precipitation Annual cycle
15

• Up to date most RCMs evaluations have been
  focused on the mean climate, but what about
  higher order climate variability?

Mesoscale variability
Diurnal cylce
Intraseasonal variability
Examples of precipitation variability over
different time-scales
Interannual to interdecadal variability
16
What do we know?

• Overall model performance of the mean climate
• Systematic biases of the simulated mean climate
• Largest biases mainly over tropical South America
• Warm and dry biases over tropical regions Land
  surface?
• Dry and bias over LPB resolution?
• Uncertainty on simulating mean climate
  (inter-model spread)
• Largest biases mainly over tropical regions

But we dont know much about

• Model performance on higher order variability
  patterns
• Systematic biases on higher order variability
  patterns
• Uncertainty in simulating higher order
  variability patterns

17
Internal variability of a RCM over South America

• MM5 model
• OND 1986
• 4 members
• (Solman and Pessacg, 2010)

• How large is the internal variability for
  long-term climate simulations?
• Annual cycle of the internal variability?

18
CLARIS-LPB CORDEX
Model Evaluation Framework
Climate Projection Framework
ERA-Interim LBC 1989-2008
A1B Continuous runs Timeslices 2010-2040
2070-2100
RCP4.5, RCP8.5 1951-2100 or timeslices
Regional Analysis Regional Databanks
Need for a collaborative framework to provide
CORDEX projections over South America
19
RCM perspectives

• Need for evaluating RCMs in terms of variability
  patterns.
• Understanding the causes for the systematic
  biases of the simulated mean climate
• Need for evaluating the internal variability of
  RCMs to put the climate response patterns in the
  context of the noise level.
• Need for a collaborative framework to provide
  CORDEX projections over South America

20
Conclusions

• South American climate is characterized by
  variability patterns on a broad range of
  timescales and different spatial distributions.
• Regional climate models are able to simulate the
  mean climatic conditions, though large
  uncertainties and systematic biases can be
  identified over some regions /variables.
• Studies using Regional Climate models focused on
  the response of the regional climate to external
  forcings (increasing CO2 land use changes or
  soil moisture conditions) show that the climate
  response is very heterogeneous both spatially and
  temporally.
• Some particular regions of South America exhibit
  large responses, mainly in terms of changes in
  precipitation, temperature and moisture flux to
  these external forcings.