Transcript Predictability of weather and climate

Predictability of weather and climate
What have we learned from comprehensive
modeling studies?
Lennart Bengtsson
MPI for Met. Hamburg
ESSC, Uni. Reading
Many thanks to
J. Shukla
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Laplace
Essai philosophique sur les probabilités
We may regard the present state of the universe as the
effect of its past and the cause of its future. An intellect
which at a certain moment would know all forces that set
nature in motion, and all positions of all items of which
nature is composed, if this intellect were also vast enough
to submit these data to analysis, it would embrace in a
single formula the movements of the greatest bodies of the
universe and those of the tiniest atom; for such an intellect
nothing would be uncertain and the future just like the
past would be present before its eyes.
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
• CLIMATE PREDICTION AND CHAOS
• “ For want of a nail, the shoe was lost;
• For want of a shoe, the horse was lost;
• For want of a horse, the rider was lost;
• For want of a rider, the battle was lost;
• For want of a battle, the kingdom was lost “
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
• Predictability of weather
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
St. Petersburg prediction 10.1 2006
17th onward
ca -30 C
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Example of ultra-high predictability
Observed and simulated QBO
Note the marked changes in wind direction at 10-30 hPa
every 26-28 month
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Predictability of weather and climate.
What have we learned from comprehensive modeling
studies?
• Predictability of atmospheric flow
(from mid latitude weather prediction to the quasi-biennal oscillation)
Predictability of weather
(how close are we to the limit?)
Coupled ocean atmosphere modes
(El Nino-Southern Oscillation)
What do we mean by climate predictability?
(what is predictable?)
Climate and climate change predictability
Concluding remarks
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
A Fundamental Question in Weather
Predictability
A Dynamical System is :
TYPE 1 – characterized by an infinite range of predictability
TYPE 2 – the range of predictability is finite, but can be increased
indefinitely by decreasing the size of the initial error
TYPE 3 – the range of predictability is finite and intrinsically limited
Does the Weather Constitute a Type 2 or a Type 3 System?
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
The Growth of Very Small Errors
• Basic Idea – Reduce the Size of the Initial Error by putting it on
smaller and smaller scales
• Ultimate Predictability controlled by the predictability time T = time
necessary for the error to propagate “upscale” from very, very small
initial scale to a finite, pre-chosen scale
• How does T behave as the initial error gets infinitely small? This
tells us if we have TYPE 2 or TYPE 3 behavior!
• For a Spectrum E(k) ~
k -3 or steeper :
T becomes infinite (thus TYPE 2)
• For a Spectrum E(k) less steep than k
T is finite (thus TYPE 3)
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
-3 :
Nastrom and Gage, 1985:
A climatology of
atmospheric
wavenumber spectra of
wind and temperature
observed by commercial
aircraft. J. Atmos. Sci., 42,
950–960.
Does the Observed -5/3 Spectrum Imply that the Range of
Predictability Cannot be Lengthened by Reducing Initial
Error?
• Perhaps the eddies associated with the observed -5/3 spectrum do not
interact with the large scale
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Predictability of weather and climate.
What have we learned from comprehensive modeling
studies?
• Predictability of atmospheric flow
(from mid latitude weather prediction to the quasi-biennal oscillation)
Predictability of weather
(how close are we to the limit?)
Coupled ocean atmosphere modes
(El Nino-Southern Oscillation)
What do we mean by climate predictability?
(what is predictable?)
Climate and climate change predictability
Concluding remarks
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Improvements in NWP
from Miyakoda (1972) to 2002. Courtesy ECMWF
How long
to get
to D+10
in winter?
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
The principle
of error
reduction in
data
assimilation
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Improvement in
medium-range
forecast skill
12-month running
mean of anomaly
correlation (%) of
500hPa height
forecasts
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Lorenz, E. N., 1982: Atmospheric Predictability Experiments with a Large
Numerical Model. Tellus, 34, 505-513
• Estimates of the lower and upper
bounds of predictability of
instantaneous weather patterns
for ECMWF forecast system
• Lower bound: skill of “current”
operational forecasting procedures
• Upper bound: Growth of initial error,
defined as the difference between
two forecasts valid at the same time
(Lorenz curves)
“Additional improvements at extended range may be realized if the
one-day forecast is capable of being improved significantly.”
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Predictive skill ( Z 500 hPa) for the NH
and predictability estimates ( for 6 ( red) and 24 hr (blue)
increments)
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Evolution of 1-Day Forecast Error, Lorenz Error
Growth, and Forecast Skill for ECMWF Model
(500 hPa NH Winter)
1982 1987 1992 1997 2002
ÒInitial errorÓ(1-day
forecast error) (m)
20
15
14
14
8
Doubling time (days)
1.9
1.6
1.5
1.5
1.2
Forecast skill (day 5 AC C)
L Bengtsson 14.6.07
Bosön, Stockholm
0.65 0.72 0.75 0.78 0.84
Multiscale modeling and
simulation in Science
2007
8
1.2*
0.91
From ECMW *est
ECMWF EPS: Forecast Started 8th
January 2005 00UTC (GUDRUN)
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
From L Froude, ESSC
ECMWF EPS: Forecast Started 6th
January 2005 00UTC
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
From L Froude, ESSC
Predictive skill and predictability of storm tracks
for different observing systems
NH
SH
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
From L Froude, ESSC
Conclusions
Weather predictability
• Predictability of actual weather is limited to a few days.
• Predictability of synoptic weather systems is likely limited by ca
two weeks. Improvements has come through smaller initial
errors due to better observations and more advanced dataassimilation.
• Predictability of the general weather type varies between
different regions, for different seasons and for different
situations and can be from weeks to several months.
• There are atmospheric patterns such as QBO that have almost
unlimited predictability.
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Predictability of weather and climate.
What have we learned from comprehensive modeling
studies?
• Predictability of atmospheric flow
(from mid latitude weather prediction to the quasi-biennal oscillation)
Predictability of weather
(how close are we to the limit?)
Coupled ocean atmosphere modes
(El Nino-Southern Oscillation)
What do we mean by climate predictability?
(what is predictable?)
Climate and climate change predictability
Concluding remarks
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
El Nino/Southern Oscillation
1998 JFM SST [oC]
JFM SST Climatology [oC]
1998 JFM SST Anomaly [oC]
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
El Nino changes precipitation patterns
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Evolution of
Climate Models
1980-2000
Model-simulated and observed
500 hPa height anomaly (m)
1983 minus 1989
Vintage 2000
AGCM
Current Limit of Predictability of ENSO (Nino3.4)
Potential Limit of Predictability of ENSO
20 Years: 1980-1999
4 Times per Year: Jan., Apr., Jul., Oct.
6 Member Ensembles
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Kirtman, 2003
Prediction of Atlantic hurricanes
with a general circulation model
integrated over 30 years
ECHAM5/OM
L Bengtsson 14.6.07
Bosön, Stockholm
ERA-40
Multiscale modeling and
simulation in Science
Predictability of large-scale climate
anomaly patterns
• The predictability of ENSO is likely to be from several months to a
few years. There are large variations in predictability. Present
predictability assessment suffers from rather poor coupled models and
later work is expected to change this.
• The same is true for well developed land surface patterns.
• Long term anomalies over Europe (NAO) have limited predictability
• The fact that very long semi-persistent pattern occur both in reality
and in models suggest that predictability in some regions (such as in
the Sahel region) are longer.
• Here we need more active basic research.
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Predictability of weather and climate.
What have we learned from comprehensive modeling
studies?
• Predictability of atmospheric flow
(from mid latitude weather prediction to the quasi-biennal oscillation)
Predictability of weather
(how close are we to the limit?)
Coupled ocean atmosphere modes
(El Nino-Southern Oscillation)
What do we mean by climate predictability?
(what is predictable?)
Climate and climate change predictability
Concluding remarks
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Köppen climate zones
•
•
•
•
•
•
Main groups
A: Tropical rainy climate, all months > +18 C
B: Dry climate, Evaporation > Precipitation
C: Mild humid climate, coldest month +18 C - -3 C
D: Snowy - forest climate, coldest month < -3C but warmest > +10
E: Polar climate , warmest month < +10 C
ET: Tundra climate, warmest month > 0 C
•
•
•
•
Subgroups
f : Moist, no dry seasons
w: Dry season in winter
s : Dry season in summer
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Köppen climate zones
ECHAM5
simulated
ERA40
determined
from analyses.
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Warmest and coldest season in Europe
1500-2003
Luterbacher et al(2004), Xoplaki et al (2005)
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
50-year trends
>0.23 corresponds to 95% significance
T
Sea
ice
Z
850
P
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Observation and model statistics
Luterbacher et al., 2005 ( Temp. in C)
Mean
Stand. dev.
Min.
Max.
Max.-Min.
L Bengtsson 14.6.07
Bosön, Stockholm
(a) Luterbacher (2005)
Annual
DJF
JJA
8.13
-2.31
16.83
0.41*
1.15
0.47
6.61
-5.69
15.62
(1695) (1695/96) (1821)
9.40
-0.09 18.18
(1822) (1842/43) (1757)
2.79
5.60
2.56
Annual
7.39
0.55
5.41
(16)
9.41
(78)
4.00
Multiscale modeling and
simulation in Science
(b) Model
DJF
-2.30
1.23
-7.45
(15)
1.47
(52)
8.92
JJA
16.96
0.55
15.55
(170)
19.08
(459)
3.53
Climate predictability
• Internal climate modes lasting up to several decades are
likely to exist in the climate system. Whether these are
predictable is still an open question
• Model studies suggest that such internal modes have
dominated climate variations during the last several
centuries
• It is hardly feasible to infer any changes in external
forcing from meteorological records for the period 1500 to
1900.
• Models are capable to reproduce the observed
climatology with considerably accuracy ( e.g.Koeppen)
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Predictability of weather and climate.
What have we learned from comprehensive modeling
studies?
• Predictability of atmospheric flow
(from mid latitude weather prediction to the quasi-biennal oscillation)
Predictability of weather
(how close are we to the limit?)
Coupled ocean atmosphere modes
(El Nino-Southern Oscillation)
What do we mean by climate predictability?
(what is predictable?)
Climate and climate change predictability
Concluding remarks
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Delworth and Knutson, 2000
Monte-Carlo simulations with a coupled AO GCM: one out five simulations almost
perfectly reproduced the observed global temperature variability.
obs
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
exp 3
Resultat från den senaste
klimatutvärderingen
Observerad och beräknad
temperaturändring
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Present
climate
Coupled
Model
T63L31
Future
climate
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
• Predictability of snow in Germany
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Ensemble climate trends averaged for
different time-periods
(T/decade)
1-30 years
L Bengtsson 14.6.07
Bosön, Stockholm
1-80 years
Multiscale modeling and
simulation in Science
Climate change predictability
•
•
•
•
•
Changes in climate patterns during the last 50 years are now with high
probability due to changes in atmospheric composition (greenhouse gases and
aerosols)
The trade off between the two cannot easily be done as they both have very
similar climate pattern signatures. It can well be that we are underestimating
the effect of greenhouse gases and overestimating reflecting aerosols or vice
versa
However, a marked positive feedback from anthropogenic greenhouse gases
via water vapor and surface albedo is most likely.
Future climate scenarios include considerable internal regional variations
which can mask or enhance climate warming for several decades. Typical
examples from the last century are the warm 1930s and 40s and cold 1960s
and 70s.
Robust climate change trends for specific regions require probably 50-100
years
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Predictability of weather and climate.
What have we learned from comprehensive modeling
studies?
• Predictability of atmospheric flow
(from mid latitude weather prediction to the quasi-biennal oscillation)
Predictability of weather
(how close are we to the limit?)
Coupled ocean atmosphere modes
(El Nino-Southern Oscillation)
What do we mean by climate predictability?
(what is predictable?)
Climate and climate change predictability
Concluding remarks
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
Concluding Remarks
• The largest obstacles in realizing the potential predictability of weather and
climate are inaccurate models and insufficient observations, rather than an
intrinsic limit of predictability.
– In the last 30 years, most improvements in weather forecast skill have arisen
due to improvements in models and assimilation techniques
• The next big challenge is to build a hypothetical “perfect” model which can
replicate the statistical properties of past observed climate (means, variances,
covariances and patterns of covariability), and use this model to estimate the
limits of weather and climate predictability
– The model must represent ALL relevant phenomena, including ocean,
atmosphere, and land surface processes and the interactions among them
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science
END
Any questions?
L Bengtsson 14.6.07
Bosön, Stockholm
Multiscale modeling and
simulation in Science