IRI, Lamont, NY

1
The Pre-Industrial European ClimateWhat were the
dominant mechanisms?

• Lennart Bengtsson
• with
• Kevin Hodges
• ESSC, Reading
• and
• Erich Roeckner
• Renate Brokopf
• MPI, Hamburg

2
What do we know about European climate
variations?In this study we will focus on
surface temperature ( 2m above ground)

• This is probably the best observed part of the
  world.
• Several meteorological stations with daily
  records back to the middle of the 18th century
• A few observational records back to the 17th
  century.
• Documentary proxy evidence of incidental
  character ( e.g. Brazdil et al., 2005 with
  references)
• Sea ice data ( Baltic Sea and Iceland), Greenland
  ice core, tree ring evidence from Scandinavia
  etc.
• We have here used recent data sets compiled by
  Luterbacher (2005)

3
What are the processes leading to climate
variations?

• Internal variations of the climate system
• ( Hasselmann, 1976, Manabe and Stouffer, 1996)
• Volcanic aerosols warming the stratosphere and
  cooling the troposphere
• Solar variations in addition to the 11-year cycle
• Anthropogenic influences (GHG and aerosols, land
  surface changes)

4
What do we know about the causes to European
climate variations?

• Anthropogenic effects have increasingly
  influenced climate since the early 20th century,
  but presumably not so much before that time.
• Volcanic aerosols have influenced climate, but
  only for a few years at most
• Solar variability is still an open issue, no
  variations except the 11-year cycle according to
  recent summary of satellite records (Frölich,
  2005)
• We know that climate varies due to internal
  processes such as ENSO.

5
(No Transcript)
6
Why have we done this study now?

• The latest model at MPI (ECHAM5/MPI-OM) have
  demonstrated features which we believe are
  important. These include
• Realistic portraying of ENSO ( Oldenborgh et al.,
  2005)
• A good simulation of tropical intra-seasonal
  variability(Lin et al., 2005)
• Realistic simulation of extra-tropical and
  tropical storm tracks( Bengtsson et al., 2005)
• Long stable integration without the use of flux
  adjustment
• Relatively high resolution ( T63/L31)
• Recent compilation of satellite records indicate
  that there is no discernible trend in TSI (total
  solar irradiance) 1978-2005 ( Frolich, 2005) and
  that the empirical relation between sun spots
  and TSI is not any longer obvious. Consequently
  previously compiled data sets of long term
  variation in TSI and used by modelers could be
  put into question. Long-term TSI trend has been
  reassessed (2005) and significantly reduced.

7
TSI (1978-2005) After C Frolich (2005)ISSI,
Bern
8
Assessment of long-term variation in TSI

• Judith Lean ( PAGES News, Vol 13, No.3)
• stellar data has been reassessed, instrumental
  drifts are suspected in the aa-index, and it has
  been shown that the long-term trends in the
  aa-index and cosmogenic isotopes do not
  neccessarily imply equivalent long-term trends in
  solar irradiance
• In the latest synthesis by Wang, Lean and
  Sheeley, 2005 (Astrophys. J. 625, 522-538) has
  the amplitude in low frequency irradiation been
  reduced to 0.27x Lean 2000.
• This means a difference between the Maunder
  Minimum and the present TSI of only 0.5 Wper
  m2 ) or 0.09W effectively.

9
Is it at all possible to reconstruct the
evolution of the regional climate?

• Ensemble integrations with GCMs show generally
  large differences between individual members.
• Exceptions can be found in some regions in
  relation to ENSO events. ( but then ENSO must be
  known)
• The European region is under influence of
  Atlantic storm tracks which are only weakly
  constrained by external or remote forcing.
• Reconstruction of climate evolution is closely
  related to climate predictability ( of the 1st
  kind)

10
St. Petersburg prediction 10.1 2006
17th onward ca -30 C
11
IPCC AR4 Arctic Temperature Anomalies by AOGCMs
20th Century (20C3M) 11/20 models have decadal
signal
Courtesy, J Overland
PIcntrl (Control Runs) 10/20 models have decadal
signal
12
The pre-industrial European climate
1500-1900.Some science issues

• What is the typical internal variability?
• How is the variability related to global
  variability?
• Over which periods can trends be observed?
• What are the characteristic features of extreme
  events?
• How are the storm tracks related to climate
  anomalies?
• What is the relation to NAO, PNA and ENSO?

13
Reconstruction of climate from observations

• Upper air
• 1978 until present global 3D-reconstruction
• 1947 until 1978 global reconstruction feasible,
  but significant errors for SH and the tropics
• Surface only
• Late 19th century until present surface
  observations for a major part of the globe
• 19th century climate observations from selected
  regions
• 18th century small number of selected
  observations
• 17th century and earlier (essentially only
  indirect information)

14
Reconstruction of climate from observations and
proxy data

• Global, hemispheric, 1000-2000, annual resolution
• Mann et al, 1999
• Jones and Mann, 2004 and references therein
• Europe (1500-2003, seasonal resolution)
• Luterbacher et al., 2004
• Xoplaki et al., 2005
• Luterbacher et al., 2005 ( this study)

15
Luterbacher, Science 2004
16
Luterbacher Science 2004
17
Warmest and coldest season in Europe 1500-2003
Luterbacher et al(2004), Xoplaki et al (2005)
18
The Luterbacher (2005) data set compared to
Luterbacher et al., (2004)

• Gridded data from Mitchell and Jones (2005)
• Additional instrumental predictors mainly from
  18th and 19th century
• Additional proxies for the 1500-1650 period

19
Observed winter (JJA) temperature for Europe
1500-2000Luterbacher 2005

-0.1
-5.7
20
Observed summer (JJA) temperature for Europe
1500-2000Luterbacher 2005

18.2
15.6
21
Global annual averaged temperature500-year
integration with ECHAM/MPI-OM Pre-industrial
atmospheric composition. No variation in external
or internal forcing
14.6
Climate drift 0.027/cent.
13.7
22
The pre-industrial climate compared to the
present (90- year mean)

• Atmospheric composition
• CO2 286.2 ppm ( now 382 ppm) ( 1960-1990 350
  ppm)
• CH4 805.6 ppb
• N2O 276.7 ppb
• No CFCs
• Surface temperature effect
• Global - 0.19 K
• European land area winter - 0.54 K
• European land area summer 0

23
Model simulation of el Nino/la Nina (NINO3 index)
during 500 years
24
Model simulation of the North Atlantic
Oscillation (NAO)during 500 years
25
(No Transcript)
26
Storm tracks at high NAO ( gt2 sd, left) and low
NAO ( lt 2 sd, right) Intensity and density
(top)and generation (below)
27
50-year trendsgt0.23 corresponds to 95
significance
T
Sea ice
Z 850
P
28
50-year sea-ice trendgt0.23 corresponds to 95
significance


29
Model winter (DJF) temperature for Europe during
500 years
1.5
- 7.5
30
Pre-industrial temperatures in Europe (DJF)Model
results ( smaller numbers in right column are
observed values prior to 1950 covering ca 200
years
31
Observed winter (JJA) temperature for Europe
1500-2000Luterbacher 2005

1500
1900
32
Model summer (JJA) temperature for Europe during
500 years
19.1
15.6
33
Pre-industrial temperatures in Europe (JJA)Model
results ( smaller numbers in right column are
observed values prior to 1950 covering ca 200
years
34
Observed summer (JJA) temperature for Europe
1500-2000Luterbacher 2005

18.2

15.6
1900
1500
35
Observation and model statisticsLuterbacher et
al., 2005 ( Temp. in C)
36
Coldest winter and warmest summerleft observed,
right model
1941/42
1947
37
(No Transcript)
38
Observation and model statisticsLuterbacher et
al., 2005 ( Temp. in C)
39
Ten coldest European wintersmodel and
observations ( Luterbacher, 2005)
Model
Obs
Model global anomaly
Model height 500 hPa
40
Ten warmest European summersmodel and
observations ( Luterbacher, 2005)
Model
Observ.
41
Largest temperature differences between 30 year
periodswinter ( cold-warm) left, summer
(warm-cold) right
model
observ.
Model global
42
Coldest-warmest winterWarmest-coldest summer
Largest 30 - year averages
43
Summary Pre-industrial European climate over 500
years ( land area)

• There is a good agreement between modeled and
  observed winter and summer temperature
• Modeled variance is slightly higher than observed
  but agreement better in the 19th century
• The coldest winter in the model is colder than
  observations and the warmest summer is warmer
  than observed. There are considerably
  similarities with observed extreme seasons.
• The ten coldest winters are associated with a
  warmer central tropical Pacific and a warmer
  Arctic
• The ten warmest summers are associated with
  colder than normal tropical oceans
• Sustained anomalies over 30 years in the model
  experiment are similar to observations but
  observed summer anomalies are larger

44
Summary Pre-industrial European climate over 500
years ( land area)

• Global temperature anomalies is well correlated
  with ENSO (0.70 in DJF)
• There is virtually no correlation between global
  temperature anomalies and European temperature
  anomalies.
• As in observations modeled European winter
  temperature is correlated with NAO (0.46)
• There are regional trends on the time scale of 50
  years( significant at 95)

45
ConclusionsThe European climate 1500-1900

• It is strongly suggested that that the climate
  of Europe during the period was strongly
  dominated by natural variability and that
  external forcing ( total solar irradiance, TSI
  and volcanic aerosols) only have had a minor
  effect.
• This is supported by the fact that extreme
  temperature for different seasons occur at very
  different times. For example the coldest summer
  and autumn occurred in the beginning of the 20th
  century, while the coldest winter and spring
  occurred during the 17th and 18th century
  respectively.
• We believe it is probably not feasible to
  attribute climate variations in the European
  region to variations in the external forcing as
  these variations are completely dominated by
  internal climate variations
• We believe great care must be applied in
  attribution studies during this period as
  incorrect conclusions may result when model
  variance have deficiencies. It is important to
  recognize that seasons with extreme temperature
  and even longer periods with unusual temperatures
  may just happen by chance

46
Caveats

• External forcing from volcanic aerosols
• ( This is likely to give higher variance during
  the summer)
• Variation in cloudiness due to CCN of cosmic
  origin
• Land-surface changes
• Model artifacts

47
Proposals for further work

• Ensemble integrations ( minimum 3)
• Extend assessment to other parts of the Earth
  with long observational records of good quality
• Including volcanic aerosols
• Including land surface processes
• Including orbital forcing

48
END