Utility of Detection and Attribution

1
Utility of Detection and Attribution

• Hans von Storch
• Institute for Coastal ResearchGKSS Research
  Center, Geesthacht, Germanyand
  CLISAP/KlimaCampus, Hamburg University

2

• The issue is
• deconstructing a given record
• with the intention to identify predictable
  components.
• Predictable
• -- either natural processes, which are known of
  having limited life times,
• -- or man-made processes, which are subject to
  decisions (e.g., GHG, urban effect)
• Differently understood in different social and
  scientific quarters.
• The issue is also to help to discriminate between
  culturally supported claims and scientifically
  warranted claims (cf. Myles scepticism)

3
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.

4
Significant trends
Establishing the statistical significance of a
trend may be 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.
5
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

6
Global The utility of global da is to clarify
that the concept of GHG-related anthropogenic
climate change is real. Conclusion from a
successful daThe public is talking about a
real effect.
7
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, doi10.1029/2005GL025591 Zorita, E., T.
Stocker and H. von Storch How unusual is the
recent series of warm years? Geophys. Res. Lett.
8
The Rybski et al- approach

• Global mean air temperature
• Statistics of ?TL,m, which is the difference of
  two m-year temperature means separated by L
  years.
• Temperature variations are modelled as Gaussian
  long-memory process, fitted to various
  reconstructions of historical temperature
  (Moberg, Mann, McIntyre)

Historical Reconstructions their significance
for detection
9
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 for m5 and
L20 (top), andfor m30 and L100 years. The
thresholds R 2, 2.5 and 3s are given as dashed
lines.
Rybski et al., 2006
10
Counting extremely warm years

• Among the last 17 years, 1990-2006, 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?
• Monte-Carlo simulations taking into account
  serial correlation, either AR(1) (with lag-1
  correlation ?) or long-term memory process (with
  Hurst parameter H0.5d).
• Best guesses
• ? 0.85
• d ? 0.45 (very uncertain)

Zorita, et al 2009
11
RegionalIntention Preparation and design of
measures to mitigate expected adverse effects of
climate change. Problems high variability,
little knowledge about natural variability more
human-related drivers (e.g. industrial aerosols,
urban effects)
12
Log-probability of the event E that the m largest
values of 157 values occupy the last17 places in
long-term autocorrelation synthetic series
Zorita, et al., 2009
Derived from Hadley Center/CRU data for Giorgi
bins.
13
For regional mean temperatures we have a signal
and attribute it to GHGs (see also Jonas talk).
What about precip? This information may be
relevant for a few sectors, such as agriculture.
14
Regional DJF precipitation
?0.05
15
Regional JJA temperatures
16
Consistency analysis Baltic Sea catchment

• Consistency of the patterns of model
  predictions and recent trends is found in most
  seasons.
• A major exception is precipitation in JJA and
  SON.
• The observed trends in precipitation are stronger
  than the anthropogenic signal suggested by the
  models.
• Possible causes- scenarios inappropriate
  (false)- drivers other than CO2 at work
  (industrial aerosols?)- natural variability much
  larger than signal (signal-to-noise ratio ?
  0.2-0.5).

17
Local change another major driver urban warming
Gill et al.,2007
18
Local Station data
JonesMoberg until 2000, afterwards NASA-GISS
19
Conclusions Based on my personal experience in
interacting with public, media and policymakers
(German bias all levels) DA is confronted
with requests from different stakeholders, with
stakes at different geographical scales,
woldviews and perceptions.
20
Global clients want to have proof that the
basic concept of man-made global climate change
is real. The best answer for this client is an
answer which is very robust and not critically
dependent on models. Mostly done. Regional
clients want to have best guesses of the
foreseeable future, in order to institute
adaptive measures on the scale of medium-size
catchment basins not many clear results. Local
clients want know how global and local drivers
shape the future of the ocal environment, and
which measures for mitigation are available, and
which levels of adaptation are required. very
little done.
21
Storm surges in Hamburg
22
Sturmfluten in der ElbeVergangenheit
Differenz Scheitelhöhen Hamburg - Cuxhaven
Sturmfluten in der Elbe deutlich erhöht seit 1962
aufgrund wasserbaulicher Maßnahmen, vor allem
wegen der Verkürzung der Deichlinie