Estimating climate variability over the next 1-25 years

1
Estimating climate variability over the next
1-25 years

• Dr Scott Power
• IOCI, August 2005

2
Using history as a guide (for 2006-2024)
1911-1974 1975-2001
Data courtesy WA Water Corp
3
Can we use climate models to provide better PDFs?
4
Australian rainfall v. NINO4 SST in BMRC Climate
Model
5
Models data provide climate predictions for
6-12 months ahead. They exhibit some skill in
predicting some things.
6
Using initial data can change PDFs (Probability
Density Functions) if there is predictability
A prediction as a change in a PDF
Data Courtesy Samoa Meteorology Division
7
Can we predict beyond 2006 years?

• BMRC CGCM (Power et al. 1998)
• MOM OGCM - Pacanowski et al. 1991
• L25, 2 deg by (0.5, 6 deg)
• hybrid mixing (ml, Ri) see Power et al. 1995
• thermodynamic sea-ice
• R21 L17 unified AGCM - Colman (2000)
• Colman 2000
• spectral, Rotstayn (1999) prognostic clouds
  Tiedtke (1989) convection GW drag (Palmer et al.
  1986) McAvaney Hess (1996) BL scheme
• Q, Sf flux adjusted

8
Climate models suggest that ENSO predictability
is very limited beyond 1-2 years
Sensitivity of NINO4 index to small initial
nudges
NINO4
Chaos limits predictability
Time (Years 1 to 4))
BMRC CGCM (Power et al. 1998)
9
Predictability beyond 2 years is present, e.g.
off-equatorial, deep (310m) Pacific Ocean
Deep Ocean Temperature

• lt. 100 years ..gt

10
Off-Equatorial, Deep Pacific Ocean - highly
predictable
Exhibits predictability

• lt.. 13 years ....gt

11
Thermohaline Circulation
Power et al. (2005, in press)
12
Kick-starting forecasts with data
Subsurface Ocean Temperature
Sea-level from satellite
Winds from satellite
XBTs moored instruments
Courtesy Neville Smith, BMRC
13
A big step forward, but approach neglects
information about initial state of climate system
IPCC model output courtesy Pandora Hope, BMRC
14
Estimating future PDFs

• Approach will borrow from
• seasonal prediction e.g. initialisation,
  ensembles
• climate change projections e.g. scenarios for
  future CO2 emissions
• strategic research on decadal predictability
• Challenging, strategic, resource intensive
• Improve models, secure obs networks
• Requires closer collaboration between CSIRO,
  Bureau
• ACCESS timely ( exciting possibility)

15
(No Transcript)
16
Seamless prediction

• Increasingly, decade- and century-long climate
  projection will become an initial-value problem
  requiring knowledge of the current observed state
  of the atmosphere, the oceans, cryosphere, and
  land surface to produce the best climate
  projections as well as state-of-the-art decadal
  and interannual predictions (WCRP, 2005)

17
ACCESS

• Australian Climate Community Earth System
  Simulator
• New initiative in planning stages
• Bureau, CSIRO, AGO
• Universities, other agencies (federal and state)

18
Thermohaline Circulation
19
Variability in models conveyor belt
Variability in models Southern Ocean Temperature
20
Using initial data can change PDFs (Probability
Density Functions) if there is predictability
A prediction as a change in the PDF
Data Courtesy Samoa Meteorology Division
21
Decadal changes in southern Indian Ocean linked
with Africa
22
Decadal changes in Southern Indian Ocean linked
with Australia (in Model)

• Research Only!

23
Courtesy J Arblaster (NCAR/BMRC)
24
Argo floats supply temperature, salinity,
pressure, velocity information - a revolution in
data acquisition
Courtesy Howard Freeland, Institute of Ocean
Sciences, CANADA
25
CaveatDecadal predictability arising from
Initial Conditions might be substantial in some
things (e.g. deep ocean) but low in variables of
more significance to humans (e.g. rainfall over
land)Strategic research in this area continues
26
Statistical Downscaling Techniques
Provide realistic local information for Impact
Studies using coarse information from Global
Climate Models
From BoM booklet The greenhouse effect and
climate change, 2004.
Courtesy Bertrand Timbal, BMRC
27
Coordinated Observation and Prediction of the
Earth System, COPES

• Aim
• To facilitate analysis and prediction of Earth
  system variability and change for use in an
  increasing range of practical applications of
  direct relevance, benefit and value to society

28
Conveyor belt variability appears to precede (by
4 years) SST possibly some Africa/Australia
variability in BMRC CGCM
29
Climate Change Projections can help
Courtesy CSIRO
30
Estimating future

• Approach will borrow from
• seasonal prediction (e.g. using data,
  ensembles)
• climate change projections (e.g. scenarios for
  future CO2 emissions)
• strategic research on decadal predictability
• Challenging, strategic, resource intensive
• Requires closer collaboration between CSIRO,
  Bureau beyond ACCESS
• Intermediate steps will be used, e.g.
• selective/nudged climatologies
• use existing climate change projections
• strategic research on decadal prediction