Transcript Water resources

Water Resources: Hydroclimatic
Forensics Through Reanalysis
Climate Adaptation
Bryson C. Bates
Leader, Pathways to Adaptation Theme
24th June 2008
Acknowledgments
● Colleagues
● Steve Charles (CSIRO)
● Richard Chandler (University College London)
● James Hughes (University of Washington)
● Eddy Campbell (CSIRO)
● Funding
● Australian Climate Change Science Program
● Indian Ocean Climate Initiative
● South Eastern Australian Climate Initiative
ACRE Workshop 2008
Why ACRE & Water Resources?
● The problem
● characteristics of multi-year drought
● short hydrologic record lengths < 100 years
● reliability of water supply systems – over-rated
&/or over-allocation?
● ACRE applications
● deriving the ‘recent’ envelope of natural climate
variability
● documenting atmospheric circulation changes that
caused droughts prior to middle of 20th Century?
● putting anthropogenic climate change in context
● stochastic assessments of system reliability
ACRE Workshop 2008
IWSS Dam Inflow Series
Cost of system expansion: 2 billion AUD spent over last decade
ACRE Workshop 2008
Rules; What Rules?
● Traditional water planning based on
assumption of stationarity (constant mean,
variance & autocorrelation)
● Observed changes in means, variance &
extremes – old rules are breaking down
● Trends or shifts: if, when & why?
● Regimes (periods exhibiting stationarity)?
● Use all of the record; or which part?
ACRE Workshop 2008
Experimental Design
● Dam inflow series
● trend?
● change points & regimes?
● Stochastic downscaling
● changes in atmospheric circulation variables?
● changes in frequencies of synoptic types?
● At-site precipitation
● changes in occurrence?
● changes in amounts?
ACRE Workshop 2008
10
1920
ACRE Workshop 2008
1960
Year
2000
0.2
0.02
1000
0.3
0.03
a
0.1
0.00
0.01
p-value
Probability value
100
Inflow (Gl)
IWSS Inflows
b
6
2
8
6
hh
10
10
12
Site Map
ACRE Workshop 2008
Nonhomogeneous Hidden Markov Model
● Observed process: sequence of regional precipitation
occurrence patterns Rt: t = 1, …, T
● Hidden discrete-valued process: sequence of weather
types (or states) St
● State to state transitions driven by atmospheric
information (predictors) Xt
ACRE Workshop 2008
NHMM Details
● Season: May to October
● Period of interest: 1958 – 2007
● Fitting: sequential estimation – BIC
● Number of weather states: 6
● Atmospheric predictors:
● mean MSLP
● N-S MSLP gradient
● DTd850 = T850 – Td850
● 1st canonical variate
● Testing: split sample & physical scrutiny
ACRE Workshop 2008
Interannual & Split Sample Validation
Validation
Fitting
Simulated daily rainfall
Fitting
Fitting
Validation
Observed daily rainfall
ACRE Workshop 2008
Dry
spell
Wet
spell
Validation
Atmospheric Predictors
(1983:2007)
versus
(1958:1982)
ACRE Workshop 2008
Synoptic Typing & Frequency
Southwest Australia
Type 5: Dry Everywhere
0.4
Probability
0.2
1958
ACRE Workshop 2008
0
0.2
0
Probability
0.4
Type 3: Wet West & Central
1978
1998
1958
1978
1998
At-Site Precipitation
1958 to 2007
● Overall reduction in precipitation occurrence –
changes most notable in west of the region
● Most sites exhibit reductions in mean wet-day
precipitation amounts
● Some indication of increase in precipitation
intensity in SW corner of the region
ACRE Workshop 2008
SE Australia
Total River Murray System Inflows (including Darling River)
Modelled Annual Inflows - current conditions
30 000
Annual Inflow (GL)
25 000
20 000
Long-term Average Inflow (11 200 GL/yr)
15 000
10 000
5 400 GL/yr
5 000
6 300 GL/yr
4 150 GL/yr
Average Inflows during Drought Periods
0
1892 1902 1912 1922 1932 1942 1952 1962 1972 1982 1992 2002
Drivers of Federation, WWII & current droughts?
ACRE Workshop 2008
Concluding Remarks


Water planners confronting

historically-unprecedented drought

non-stationarity in dam inflow series

apparent increase in climatic risk due to
anthropogenic climate change
ACRE can assist water planners by

providing additional information about envelope of
natural climate variability – system reliability

providing explanations for the causes of major
droughts prior to middle of 20th century

putting anthropogenic climate change in context –
if & when will the envelope of natural climate
variability be breached (approximately)?
ACRE Workshop 2008
NHMM vs. Null Model
Occurrence
Amounts
Period
SSM
SSE
1958-77
10306
15567
0.398
919753
2198732
0.295
1978-92
7712
10084
0.433
687208
1640354
0.295
1993-98
3197
4187
0.433
280207
727790
0.278
ACRE Workshop 2008
Er
SSM
SSE
Er
Indian Ocean Climate
Initiative
SW Australia
Informed Adaptation
120
1925-1975
100
1976-2003
mm
80
60
40
20
0
Jan
ACRE Workshop 2008
Feb
Mar
Apr
May
Jun
Jul
Month
Aug
Sep
Oct
Nov
Dec
SEACI
Weather Type Probabilities: SE MDB
"Dry Everywhere"
ACRE Workshop 2008
"Wet Everywhere"