South America regional climate model simulations and seasonal forecast

1
South America regional climate model simulations
and seasonal forecast

• Julio Pablo Reyes Fernandez
• Center for Weather Forecast and Climate Studies
  CPTEC

2
Outline

• Introduction
• Short Review RCM over South America
• RCM History at CPTEC
• Models
• Experiments and results
• Summary
• Future
• Where are going

3
Introduction

• Necessity regional climate information (Added
  value)
• Regional climate models supply adequate
  representation of physical process and high
  spatial resolution
• The RegCM are based on NWP models with suitably
  modified radiative transfer and land surface
  process

4
von Storch, 2004
5

• Review papers Giorgi and Mearns (1991), McGregor
  (1997), Giorgi and Mearns (1999), Giorgi et al.
  (IPCC 2001), Leung et al. (2003), Wang et al.
  (2004), Laprise (2006)

6
Short Review

• RCM studies over South America (15 years)
• Low to high spatial resolution (100-20km)
• Months to years (nowdays)
• Tuning to tropical region (few information)

7
South America is a challenge for climate modeling
community

• Large spatial domain, position, particular
  characteristics

8
Fernandez, 2004
9
(No Transcript)
10
(No Transcript)
11
Applications

• Studies
• Climate change
• Climate system
• Development and validation
• Impact
• Paleoclimate
• Process
• Seasonal prediction

12
Regional models at CPTEC

• History
• 1994 CPTEC
• 1996 operational regional model ETA
• 40km South america
• 1998 ETA extended integrations 1-month long
• GCM-driven Control experiments
• Dry and wet season
• 2000 ETA seasonal integrations
• 4.5 months
• 2000 at present BRAMS quality air

13

• 2003 at present Data Assimilation
• RPSAS and EKF
• 2004 at present Regional climate model
  development
• Perfect LBC experiments (NCEP, ECMWF)
• 2005 at present experiments of climate conditions
  different from present
• Climate change
• HadRCM, ETA, RegCM3

14
Objetives
To use regional climate models to assess regional
consequences of global climate change is
necessary firstly analyzed their present climate
simulations. In this talk show some
characteristic of model performance, namely,
correlation of climatological means and biases to
identify systematic model errors and also others
experiments at CPTEC. Development and
validation, process, climate change and seasonal
prediction
15
Models

• CPTECs ETA
• ICTP RegCM3
• BRAMS

16
ETA model modifications

• ETA NCEP regional model (Mesinger et al. 1988).
  To perform climate integration the original code
  was strongly modified. There were made multiple
  changes and corrections
• developed an interface which allows the ETA
  model to start from different data and to use
  them for the time dependent boundary conditions
• modified the SST update program to use
  climatological and operational SST data

17

• modified the insolation programs to use the 360
  days year HadAM3H data new solar and long wave
  radiation scheme
• restart code
• corrected all shortcomings that dont permit to
  run the model for the period more than 3 years
• The integrations were performed on the
  supercomputer SX6

18
Domain
19
Experiments and results
20
Simulations of the Summer Circulations

• Design
• RegCM3 and ETACLIM
• 10-years integration for Austral summer (January)
• 80 km resolution
• Capabilities

21
Fernandez et al., 2006a TAC
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
Transferability and subgrid

• Design
• 50 kms over different regions
• Reanalysis NCEP/DOE II
• Temperature and Precipitation skill over
  different domains
• Sub-BATS at medium resolution
• (50 km ? 10km)
• INPE, USP, IGP, INM

26
Fernandez, Rocha, Martinez and Silva, 2006
27
(No Transcript)
28
Fernandez, Rocha, Silva, Martinez, Cirisan 2006
29
(No Transcript)
30
Impact the new solar radiation scheme

• Design
• 2003 Austral Summer (January)
• 40km resolution
• Reanalysis NCEP/DOE II
• Reynolds and Smiths SST
• Impact on surface fluxes, temperature and
  precipitation
• Tarasova, Fernandez, Pisnichenko, Marengo,
  Caballos and Bottino, 2006 JAMC

31
(No Transcript)
32
(No Transcript)
33
The INPE ETA model for Climate Change Simulations
(CCS)

• Design
• 40km resolution
• Long-term 5-year continuous integration
• 1979-1984 Reanalysis NCEP/DOE II
• Validation and adjust
• Fernandez, Pisnichenko, Tarasova, 2007

34
Forcing and simulations
Salathé, Steed, Mass, and Zahn, 2007 JC
Pisnichenko, I.A., T.A.Tarasova, J.P.R.
Fernandez, and J. Marengo, 2006, 8th CSHMO
35
(No Transcript)
36
(No Transcript)
37
(No Transcript)
38
Fig 4 Simulated rainfall monthly means.
39
Fig 5 Simulated air temperature monthly means.
40
(No Transcript)
41
Comparison
42
Rauscher et al. 2007, MWR
43
ETACLIM, 1979-1983 NNRP2
44
Fig 7 Diurnal cycle (UTC) of rainfall (3 hour
period).
45
(No Transcript)
46
DEEP SOIL TEMPERATURE
Impact in our climate simulations
47
Climate change experiments

• Design
• 40km resolution
• Long-term 2 x 30-year continous integrations
• 1960-1990 and 2070-2100
• HadAM3H data
• Pisnichenko, Tarasova and Fernandez, 2006
• Marengo et al. 2007

48
(No Transcript)
49
(No Transcript)
50
Seasonal simulations hindcasts mode

• Design
• 40 km resolution
• Ensemble mixed physics
• 1979-2002 General Circulation Atmospheric Model
  CPTEC and reanalysis
• Improvement seasonal forecast
• Chou et al. 2005
• In progress

51
Summary

• In order to perform long simulations (climate)
  several refinements are made to the ETA
  configuration used for weather forecasting over
  the South America as the climate model
• The annual cycle of precipitation and temperature
  is well captured
• The simulations shows a tendency to underestimate
  precipitation, mainly in the austral summer and
  transition seasons (DJF) with exception in winter
  (JJA)

52
Summary - cont

• The regional seasonal temperature show a positive
  biases
• Also the values of ? between the simulated and
  observated data are in general high in almost all
  the regions (Table 1)
• Analysis of diurnal cycle of precipitation and
  soil temperatures shows some shortcommings
• The ETA model provides similar skill than others
  regional models

53
Summary - cont

• A new solar radiation scheme has been implemented
  into the Eta Model
• Nevertheless, a systematic difference remains
  (1020), probably because of the incorrect
  description of cloud parameters
• The original version of the model reproduces a
  monthly mean precipitation rate closer to the
  GPCP observational data as compared with the
  modified

54
Summary - cont

• The transferability experiments show that both
  models successful simulate the climate
  variability over different regions
• The coupling of the physics parameterizations
  seems to work better in RegCM3 than in ETACLIM
  for Middle latitude and Caribbean than for the
  Peruvian Andes and Europe
• The sub-BATS experiments show small increment of
  skill over Europe and Caribbean regions, but no
  over the Andes

55
Summary - cont

• Futher analysis of the difference between
  simulations and observations will aid in the
  identification the causes and help to tune the
  RCMs
• Further changes in physical parameterizations of
  the model aimed to obtain better model
  performance
• Identify what regions has better model skill and
  higher climate predictability will be reduce the
  incertain when used in climate change simulations
  

56
Future

• Where are going
• High resolutions
• Longer simulations and ensembles
• Development of Earth regional climate system
  models
• ocean, hydrology, biosphere, chemistry/aerosol

57
Thank you Muito Obrigado Gracias
58
examples

• Cook, KH Vizy, EK South American Climate during
  the Last Glacial Maximum Delayed Onset of the
  South American Monsoon. 2005
• Grimm,A. J. Pal, F. Giorgi Local forcing and
  Intraseasonal modulation of the South American
  Summer Monsoon soil moisture, sea surface
  temperature, and topography. 2007
• Pal, J Giorgi, F Bi, X Elguindi, N et al. The
  ICTP RegCM3 and RegCNET Regional Climate
  Modeling for the Developing World. 2007

59
(No Transcript)
60

• JFM CRU/CMAP

• JFM RegCM3

• Pal et al. 2007, BAMS

61
(No Transcript)
62
(No Transcript)
63
Control run (2)
Precipitation
WindTemp. 850 hap
Grimm et al.,2007 Sensitividade
64
EBU/POM
65
(No Transcript)