Transcript Trends - hvonstorch.de

Concepts of Detection and
Attribution
Hans von Storch
Institute for Coastal Research
GKSS Research Center, Geesthacht, Germany
and Meteorological Institute, Hamburg University
Page 1
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Person: Hans von Storch
Hans von Storch
• director of Institute for
Coastal Research @ GKSS
• professor for
Meteorology at the
Meteorological Institute
@ U Hamburg
• author of „Statistical
Analysis in Climate
Research“ @ Cambridge
U Press (with Francis
Zwiers) and other books.
• doctor h.c. at Natural
Science Faculty @ U
Göteborg
Page 2
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Detection and attribution
of ongoing change
Page 3
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
The question if we „see something“ supporting the reality of
a human influence on climate – needs the adoption of a
mathematical language.
Determination of man-made climate change is not a matter
of theory, but of assessing data.
The framework is of statistical nature, and the results are
probability statements condition upon certain assumptions.
The whole process is called „detection and attribution“.
Page 4
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
„Significant“ trends
Often, an anthropogenic influence is assumed to be in operation
when trends are found to be „significant“.
• If the null-hypothesis is correctly rejected, then the
conclusion to be drawn is – if the data collection exercise would
be repeated, then we may expect to see again a similar trend.
• Example: N European warming trend “April to July” as part of
the seasonal cycle.
• It does not imply that the trend will continue into the future
(beyond the time scale of serial correlation).
• Example: Usually September is cooler than July.
Page 5
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
„Significant“ trends
Establishing the statistical significance of a trend is a
necessary condition for claiming that the trend would
represent evidence of anthropogenic influence.
Claims of a continuing trend require that the dynamical
cause for the present trend is identified, and that the
driver causing the trend itself is continuing to operate.
Thus, claims for extension of present trends into the
future require
- empirical evidence for an ongoing trend, and
- theoretical reasoning for driver-response dynamics, and
- forecasts of future driver behavior.
Page 6
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Detection and attribution
of non-natural ongoing change
• Detection of the presence of non-natural signals:
rejection of null hypothesis that recent trends are drawn
from the distribution of trends given by the historical
record. Statistical proof.
•Attribution of cause(s): Non-rejection of the null
hypothesis that the observed change is made up of a sum
of given signals. Plausibility argument.
History:
Hasselmann, K., 1979: On the signal-to-noise problem in atmospheric response studies. Meteorology over the tropical
oceans (B.D.Shaw ed.), pp 251-259, Royal Met. Soc., Bracknell, Berkshire, England.
Hasselmann, K., 1993: Optimal fingerprints for the detection of time dependent climate change. J. Climate 6, 1957 1971
Hasselmann, K., 1998: Conventional and Bayesian approach to climate change detection and attribution. Quart. J. R.
Meteor. Soc. 124: 2541-2565
Page 7
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Where does the stochasticity come from?
Stochasticity is a mathematical construct to allow an
efficient description of the (simulated and observed)
climate variability.
• Simulation data: internally generated by a very large
number of chaotic processes.
• Dynamical “cause” for real world’s natural unforced
variability best explained as in models.
Page 8
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
How do we determine
the „natural climate variability“?
• With the help of the limited empirical evidence from
instrumental observations, possibly after suitable
extraction of the suspected „non-natural“ signal.
• By accessing long „control runs“ done with quasirealistic climate models.
• By projection of the signal on a proxy data space, and
by determining the statistics of the latter from
geoscience indirect evidence (e.g., tree rings).
Page 9
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Global
Page 10
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Cases of Global Climate Change
Detection Studies
In the 1990s … weak, not well documented signals.
Example: Near-globally distributed air temperature
IDAG (2005), Hegerl et al. (1996), Zwiers (1999)
In the 2000s … strong, well documented signals
Examples: Rybski et al. (2006)
Zorita et al. (2009)
IDAG, 2005: Detecting and attributing external influences on the climate system. A review of recent advances. J.
Climate 18, 1291-1314
Hegerl, G.C., H. von Storch, K. Hasselmann, B.D. Santer, U. Cubasch, P.D. Jones, 1996: Detecting anthropogenic climate
change with an optimal fingerprint method. J. Climate 9, 2281-2306
Zwiers, F.W., 1999: The detection of climate change. In: H. von Storch and G. Flöser (Eds.): Anthropogenic Climate
Change. Springer Verlag, 163-209, ISBN 3-540-65033-4
Rybski, D., A. Bunde, S. Havlin,and H. von Storch, 2006: Long-term persistence in climate and the detection problem.
Geophys. Res. Lett. 33, L06718, doi:10.1029/2005GL025591
Zorita, E., T. Stocker and H. von Storch: How unusual is the recent series of warm years? Geophys. Res. Lett.
Page 11
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Trend in air
temperature
1965-1994
Signal or noise?
1916-1945
Hegerl et al., 1996
Page 12
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
“Guess patterns”
The reduction of degrees of
freedom is done by projecting
the full signal S on one or a
few several “guess patterns”
Gk, which are assumed to
describe the effect of a given
driver.
S = k k Gk + n
with n = undescribed part.
Example: guess pattern
supposedly representative of
increased CO2 levels
Hegerl et al., 1996
Page 13
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Trends in a scenario
calculation until 2100
Trends in temperature until 1995
Page 14
Hegerl et al., 1996
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
163-209, ISBN 3-540-65033-4
Zwiers, F.W., 1999: The detection of climate change. In: H. von Storch
and G. Flöser (Eds.): Anthropogenic Climate Change. Springer Verlag,
Page 15
Attribution
diagram for
observed 50year trends in
JJA mean
temperature.
The ellipsoids enclose non-rejection regions for testing the null hypothesis
that the 2-dimensional vector of signal amplitudes estimated from
observations has the same distribution as the corresponding signal
amplitudes estimated from the simulated 1946-95 trends in the greenhouse
gas, greenhouse gas plus aerosol and solar forcing experiments.
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
attribution
From:
Hadley Center,
IPCC TAR, 2001
Page 16
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
The Rybski et al- approach
• Global mean air temperature
• Statistics of ΔTL,m, which is
the difference of two myear temperature means
separated by L years.
• Temperature variations are
modelled as Gaussian longmemory process, fitted to
various reconstructions of
historical temperature
(Moberg, Mann, McIntyre)
Historical Reconstructions – their significance for “detection”
Page 17
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Historical Reconstructions – their significance for “detection”
Temporal development of
Ti(m,L) = Ti(m) – Ti-L(m)
divided by the standard
deviation of the m-year
mean reconstructed temp
record
The thresholds R = 2, 2.5
and 3σ are given as dashed
lines.
Page 18
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Rybski et al., 2006
for m=5 and L=20 (top), and
for m=30 and L=100 years.
Counting extremely warm years
Monte-Carlo simulations taking
into account serial correlation,
either AR(1) (with lag-1
correlation ) or long-term
memory process (with Hurst
parameter H=0.5+d).
Best guesses
  0.8
d  0.3 (very uncertain)
Page 19
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Zorita, et al 2009
Among the last 17 years, 19902006, there were the 13
warmest years since 1880 (i.e.,
in 127 samples) – how probable
is such an event if the time
series were stationary?
Zorita, et al., 2009
Page 20
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Regional:
the Baltic Sea
catchment
Bhend, J., and H. von Storch, 2007: Consistency of observed winter precipitation trends in northern
Europe with regional climate change projections, Climate Dynamics, DOI 10.1007/s00382-007-03359
Bhend, J., and H. von Storch, 2009: Consistency of observed temperature trends in the Baltic Sea
catchment area with anthropogenic climate change scenarios, Boreal Environment Research,
accepted
Page 21
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Options
• Detection:
“Is the observed change different from what we
expect due to internal variability alone?” – not doable at
this time, since natural variability not known well
enough.
• Trends – are there significant trends? – no useful
results.
• Consistency:
“Are the observed changes similar to what we expect
from anthropogenic forcing?”
Doable: Plausibility argument using an a priori known
forcing.
Page 22
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Consistency analysis:
attribution without detection
The check of consistency of recent and ongoing trends
with predictions from dynamical (or other) models
represents a kind of „attribution without detection“.
The idea is to estimate the driver-related change from a
(series of) model scenarios (or predictions), and to
compare this “expected change” with the recent trend.
If recent change  expectation, then we may conclude
that the recent change is not due to the suspected driver,
at least not completely.
Page 23
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
DJF mean precipitation
in the Baltic Sea catchment
Example:
Recent 30-year
trend
Trend of DJF precip
according to
different data
sources.
Page 24
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Consistency analysis
Expected signals
Six simulations with regional coupled atmosphere-Baltic
Sea regional climate model RCAO (Rossby-Center,
Sweden)
• three simulations forced with HadCM3 global scenarios,
three with ECHAM4 global scenarios; 2071-2100
• two simulation exposed to A2 emission scenario, two
simulations exposed to B2 scenario; 2071-2100
• two simulations with present day GHG-levels; 1961-90
• Regional climate change in the four scenarios relatively
similar.
Page 25
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Δ=0.05%
Regional DJF precipitation
Page 26
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Regional JJA temperatures
Page 27
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Consistency analysis: Baltic Sea catchment
1. Consistency of the patterns of model “predictions” and
recent trends is found in most seasons.
2. A major exception is precipitation in JJA and SON.
3. The observed trends in precipitation are stronger than
the anthropogenic signal suggested by the models.
4. Possible causes:
- scenarios inappropriate (false)
- drivers other than CO2 at work (industrial aerosols?)
- natural variability much larger than signal (signal-tonoise ratio  0.2-0.5).
Page 28
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Overall summary
How do we know that human influence is changing
(regional) climate?
- Statistical analysis of ongoing change with
distribution of “naturally” occurring changes –
detection, statistical proof.
- ok for global and continental scale temperature.
- In the 1990s, advanced statistical analysis needed,
today also done with simpler methodology.
- Consistency of continental temperature change with
projected (expected) change in regions such as Baltic
Sea catchment (temperature and related variables);
problem with precipitation.
Page 29
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Thanks for your attention
When you want more to know:
http://coast.gkss.de/staff/storch
Contact: [email protected]
Page 30
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007
Page 31
Koninklijk Nederlands Geologisch en Mijnbouwkundig Genootschap: climate symposium, 20. November 2007