Transcript Smith_2009_v2

Current issues with climate
change projections
Ian Smith
May 21 2009
The Murray-Darling Basin
14% of Australia
1million sq. km
Over 2 million people
O’Reilly’s
Charleville
Major river systems
Warrego
Murray River
2530 km
Darling River
2740 km
Balonne
QUEENSLAND
Brisbane
Border
Barwon
Moree
Bourke
SOUTH
AUSTRALIA
Darling
Meninde
Lakes
Morgan
Macquarie
Dubbo
Menindee
NEW SOUTH WALES
Forbes
Lake
Vi ctoria
Lachlan
Mildura
Adelaide
Murray
Sydney
Murrumbidgie
Murray Bridge
Canberra
Swan Hill
Albury
VICTORIA
Melbourne
200 km
Basin characteristics
Length
3,370km
Basin size
1,050,116 km²
Population
2 million
Population density
2 people/km²
Key economic activity
agriculture, tourism, mining,
manufacturing
Key issues
risks to shared water resources,
overallocation
Average yearly rainfall in the MDB
River Murray Inflows
“…an appalling start to 2009…lowest 3 months of water flows on record…rain in the northwest were not enough to
offset climate extremes in the southern regions…extreme temperatures .. The correlation between inflows and
rainfall is now broken…algal blooms…critical human needs not guaranteed…”
Understanding observed
changes in runoff
Percent difference (1997-2006 relative to 1895-2006)
Rainfall
Rainfall
Runoff
January to March 2009
rainfall deciles
April 2006 to March 2009
36-month rainfall deciles
Matching observed trends
Role of:
GH gases ?
Aerosols ?
Ozone ?
Land cover
change ?
Natural variability
?
Other ?
SUMMER
AUTUMN
Regional climate drivers
Expansion of tropics
Weaker trade winds
Shift in the Sub-Tropical Ridge
Weaker westerlies
High-Resolution Climate Projections and Impacts
Ranking GCMs
Climate projections
•Rankings based on simulations of present
day climate
•There are models which consistently
perform relatively well, and also models
which consistently underperform
•Provides a basis for better weighting, if not
excluding, some model results when
forming projections
•There is (but not always) evidence of
clustering in the projected changes from
better performing models
GCM ID
Weighted failure rate (%)
(Table 2)
UKMO-HadCM3
0
MIROC3.2(hires)
8
GFDL-CM2.1
13
GFDL-CM2.0
20
MIROC3.2(medres)
25
ECHO-G
33
UKMO-HadGEM1
33
ECHAM5/MPI
38
MRI-CGCM2.3.2
40
CCSM3
44
CGCM3.1(T63)
50
GISS-AOM
58
INM-CM3.0
59
CGCM3.1(T47)
63
FGOALS-G1.0
63
CSIRO-Mk3.0
73
CNRM-CM3
75
IPSL-CM4
75
BCCR-BCM2.0
88
GISS-ER
88
PCM
89
GISS-EH
100
Assessment of
GCMs
Ranking of (AR4) GCM performance to improve
of regional climate change projections and
impacts.
•There are models which consistently perform
relatively well, and also models which
consistently underperform
•Provides a basis for better weighting, if not
excluding, some model results when forming
projections
•There is (but not always) evidence of clustering
in the projected changes from better performing
models
Relating local-scale weather
& climate to large-scale
atmospheric variables
(modelled or observed)
Getting from GCM coarse scale results
(100 to 200 km**2)
to catchment scales.
GCMs cannot represent regional scale
features that drive local climate
Solution 1: Statistical downscaling.
WINDS (U,V)
GEOPOTENTIAL HEIGHT (Z)
RELATIVE HUMIDITY (RH)
TEMPERATURE (T)
Rainfall = f(T,RH,Z,U,….)
Solution 2: Dynamical downscaling.
Downscaling can be complicated…
Integration of historical climate data
with projection information
Natural variability
PDF for natural variability and
model greenhouse signal
uncertainty
1900
rainfall
rainfall
2050 climate
projection
2008
2100
1900
2008
Sample obs
Currently
We need
2100