Method to force surface salinity

1
Method to force surface salinity
2
Arctic fresh water content
3
Force surface salinity
Problem Could be unstable and/or produce
unphysical oscillations Solution Use partially
coupled climate models (simple but costly) Other
suggestions?
Gerdes, Hurlin Griffies, 2006
4
Coupled models much more stable

• Heat advection feedback counters salt advection
  feedback
• Almost closed water balance at each time
• Example
• increase in run-off into the Arctic Ocean
• Reduction in meridional transport of salt
• Accumulation of salt in the subtropics (enhanced
  by the atmospheric transport of water from the
  subtropics into the Arctic)
• Transport of more saline water northward

5
20th century forcing for global ocean-sea ice
models
Rüdiger Gerdes
6
Availabe data and methods

• Reconstruction for the Arctic Ocean (Kauker et
  al., JGR, 2009)
• CALYPSO climate reanalysis
• project (failed to get funded)

• Reanalysis projects for the 20th century (Compo
  et al., BAMS, 2006)

7
CALYPSO - ClimAte variabiLitY assessment through
re-analyses of PaSt Oceanic data
In summary, the specific goals of CALYPSO are the
production of ORE-50 (1958-2008) Ocean
RE-analyses at highest horizontal and vertical
resolutions with different models/assimilation
schemes for the ocean essential variables at
global and regional (Pan- European Seas) scales
forced with atmospheric forcing from 1958-2008
produced specifically for CALYPSO inside the
project and derived from the ECMWF ERA-40 and
other forcing data set. ORE-120 (1888-2008)
Ocean RE-analyses for the ocean essential
variables with different models/assimilation
schemes at global and regional (Pan-European
Seas) scales forced with AMIP forcing from
1888-2008 produced specifically for CALYPSO for
the period 1888-2008. REP-120 and REP-50
REProcessed multi-satellite and in-situ
observational data for the period 1888-2008.
REC-120 and/or REC-50 REConstructed data sets
using statistical algorithms for the longest
possible period. CALYPSO produces both purely
observational analyses or REConstructions (REC)
and the so-called Ocean RE-analyses (ORE) that
meld numerical general circulation model
information with observations. It uses the
longest time series of relevant ocean climate
variables available in historical archives
already available at European level. The
present-day project organizing the archives of
the ocean data is SeaDataNet, in which many of
the CALYPSO partners also participate. CALYPSO
makes use of the metadata infrastructures build
by SeaDataNet and MyOcean and uses both satellite
and in-situ observations from both services.
8
Reconstruction of atmospheric forcing(as in
Kauker et al., 2009)
Can be done in many different ways. Needs to be
validated if possible. Needs expertise and
time. Can not yield patterns that are not
present in the tuning period. (In our case
Tuning period 1948 1978 instead of 1958 1988
did not reproduce the recent warming
anomaly.) Have ot be aware of possible
overfitting. Can give results that are in better
agreement with station data than reanalsis.
9
Reconstruction for the 20th century

• Reconstruct sea ice thickness for the whole 20th
  century
• - have to use ocean-sea ice model
• - needs atmospheric forcing
• - reconstructed forcing has to be validated
• Reconstruction of forcing
• redundancy analysis linking NCEP with
  observational data (station data and gridded
  data)
• validation for periods where comprehensive
  datasets are available
• validation using sea ice simulation and
  observed sea ice extent for the 20th century
• The method yields pairs of patterns of the
  predictor (station data) and the predictand
  (model grid) in which the predictand pattern is
  optimized to represented the highest possible
  variance in the fitting period.

10
AWI-wichtigste Daten
Station data from 1900 to 1997
Kauker, Köberle, Gerdes Karcher, JGR, 2009
Number of missing monthly values (1900 - 1997)
for surface air temperature
11
AWI-wichtigste Daten
2nd redundancy mode for two SAT predictor
datasets
AARI data
AICSEX/IARC data
12
AWI-wichtigste Daten
Explained variance in SLP
Fitting period
Validation period
13
AWI-wichtigste Daten
Simulated and observed ice extent
14
AWI-wichtigste Daten
Simulated and observed ice extent anomalies in
the East Siberian Sea
Observed extent from Polyakov et al., 2003
15
1930s warm anomaly

zonally averaged SAT anomalies (Johannesen et
al., 2004)
16
merge iarc_tryck hadslp1.0-240 mslpg120
fit3
val1
val2
val
17
merge iarc_tryck hadslp1.0-240 mslpg120
iarc_tryck
hadslp1.0-240
mslpg
mslpg-meanropts
Expl. Var
18
merge iarc_tryck hadslp1.0-240 mslpg120
19
Sea ice thickness reconstruction
Trend 110 km3/a
10 km3/a
20
CMIP3 vs. NAOSIM sea ice volume trends
21
AWI-wichtigste Daten
Climate models underestimate the sea ice volume
trend
22
The 20th Century Reanalysis Project

• Jeff Whitaker, Gil Compo, Nobuki Matsui and
  Prashant Sardesmukh
• NOAA/ESRL and Univ. of Colorado/CIRES

23
The 20th Century Reanalysis

• What
• A 6-hourly reanalysis from 1892-present
  (1918-1949 done so far), using only surface
  pressure observations.
• Why
• No daily gridded tropospheric-wide circulation
  dataset before 1948 exists.
• Evaluate models, understand causes for 20th
  century climate variations (e.g. 30s U.S.
  drought, 20-40s polar warming).
• How
• 56 member Ensemble Kalman Filter, T26L28 CFS03
  model.
• Includes analysis error estimate.

24
Feasibility OSE

• Subset of operational ps obs for DJF 2004/2005.
• EnKF with CFS at T62, 3DVar and 4DVar with ECMWF
  IFS at T159 (courtesy Jean-Noel Thepaut).
• OSE with 1998 GFS in Feb 2006 BAMS.

25
EnKF Error34 m
NCEP Operational
EC 4DVar Error31 m
EC 3DVar Error104 m
26
Mean and Spread Dec 1, 1918-1949
27
1918-1948 24-h forecast skill
28
1918-1948 24-h forecast skill
29
1918-1948 24-h forecast skill
30
Uncertainty estimate (blue actual O-F, red
expected O-F)
31
T700 Verification - Independent Obs(courtesy S.
Bronniman and A. Grant, ETH)
MAE 1.58 K Expected Error 1.71 K
MAE 1.59 K Expected Error 1.73 K
Assumes observation error of 1.5 K
32
Z500 Verification - Independent Ob(courtesy S.
Bronniman and A Grant, ETH)
MAE 25.8 m Expected Error 23.7 m
Assumes observation error of 20 m
33
Sep 1938 New England Hurricane
34
1930s U.S. Drought(Precip)
Analyzed Anomalies (1918-1948 base period) with
Z500 field
Observed Anomalies (1895-2000 base period)
35
1930s U.S. Drought(Precip)
Analyzed Anomalies (1918-1948 base period) with
MSLP field
Observed Anomalies (1895-2000 base period)
36
1930s U.S. Drought(Temperature)
Analyzed Anomalies (1918-1948 base period) with
Z500 field
Observed Anomalies (1895-2000 base period)
37
(No Transcript)
38
FIG. 2. Comparison of analyses of 0000 UTC 20 Dec
2001 500-hPa geopotential height from (top left)
full NCEPNCAR reanalysis using all available
observations at all levels (gt 150,000) and
parallel assimilation experiments with a
simulated 1895 network of only 308 surface
pressure observations from (top right) EnsClim
(rms difference with full NCEPNCAR reanalysis is
95.7 m), (bottom left) EnsFilt (rms difference
with full reanalysis is 49.2 m), and (bottom
right) CDAS-SFC (rms differencewith full
reanalysis is 96.0 m). Blue dots indicate the
location of the surface pressure observations
used to make the experimental analyses.The
5500-m line is thickened, and the contour
interval is 50 m.
39
Summary

• Accuracy Mid-tropospheric circulation fields
  about as accurate as a 3-day forecast today.
• Timeline 1918-1949 done, full 1892-present done
  by end of 2008.
• Data Access Will be freely available from NCAR,
  NOAA/ESRL and NOAA/NCDC. 1918-1949 in early
  2008, rest late 2008/early 2009.
• For status updates, email jeffrey.s.whitaker_at_noaa.
  gov or gilbert.p.compo_at_noaa.gov

40
Twentieth Century Reanalysis (V1)

• Objectively-analyzed weather maps with
  uncertainty
• 6-hourly, daily average, monthly values for 1908
  1958
• 2o by 2o, global grid
• The analysis is performed with Ensemble Filter
  (Compo et al., 2006).
• Observations of surface pressure and sea level
  pressure from the International Surface Pressure
  Databank version1.1 and ICOADS version 2.4 were
  assimilated every six hours.

http//www.cdc.noaa.gov/data/gridded/data.20thC_Re
an.html
41
Surface temperature April 1910
42
Arctic methods for evaluating simulations
43
Fram Strait 10 years timeseries
44
Changes in Arctic hydrography
Difference fwc 2007 - 2006
Changes ... in fwc ... in Atlantic Water
properties ... in river water distribution ... in
sea ice formation ... in Pacific Water pathways
45
AWI-wichtigste Daten
Decreasing ice thickness in the transpolar drift
Haas et al., 2008
46
AWI-wichtigste Daten
Großräumige Verteilungen aus Computermodellen
North Atlantic-Arctic Ocean-Sea Ice Model
(NAOSIM) Im Gegensatz zu den Klimamodellen
werden hier Beobachtungen der Atmosphäre
berücksichtigt (hindcast-Simulationen).
47
Tracers 18O (meteoric water, run-off) Radioact
ive tracers (Atlantic Water) Silicate (Pacific
Water)
48
Use of cost function of adjoint NAOSIM
49
Difference in 100m salinity (2080-2100)
(1980-2000) CCSM (20C3M, A1B, run1)
50
Difference in surface salinity (2080-2100)
(1980-2000) (20C3M, A1B, run1)
4
0
-4
CCSM
MPI
51
Fresh water content (rel. 35) (2080 2100)
(1980 2000)
CCSM
MPI
52
AOMIP status
53
AOMIP coordinated activities to improve models
and model predictions

• Proshutinsky1, R. Gerdes2, D. Holland3, G.
  Holloway4 and M. Steele5
• Synthesis to identify consistent errors across
  models, propose solutions, and find the most
  suitable and reliable coupled ice-ocean models
  for use in fully coupled regional and global
  climate models
• Process studies to improve models, investigate
  processes using model results and observations.
  In particular, AOMIP focuses on How to better
  model the arctic halocline which creates the
  stratification necessary to insulate perennial
  sea ice from the Atlantic Water layer? How to
  avoid restoring and flux correction? What is the
  role of different mechanisms influencing heat
  fluxes in the ocean - sea-ice - atmosphere
  system?
• AOMIP also focuses on fresh water and heat
  problems to answer the fundamental questions how
  does the fresh water/heat enter the Arctic Ocean
  system? How does it move about which includes
  phase changes, and how does it finally exit the
  system?
• Under the process studies theme, AOMIP
  furthermore investigates the role of tides in
  Arctic climate, parameterization of stress-driven
  and convection-driven mixing and the role of
  small- meso- and large-scale turbulence (eddies
  and gyres).
• The major contributors of the global change in
  Arctic climate are changes associated with
  atmospheric conditions and in the changes brought
  to the Arctic with the Atlantic water. The latter
  was the major topic of recent AOMIP studies and
  AOMIP will continue working with the Atlantic
  water role in the Arctic and global climate
  interaction.

54
AOMIP coordinated experiments
Bering Strait volume, heat and salt
fluxes Canada Basin shelf-basin exchange and
mechanisms   Pacific water circulation (origin,
forcing, pathways)   Canada basin major
mechanisms of halocline formation and
variability Circulation and fate of fresh water
from river runoff (pathways and seasonal
transformation due to mixing and freezing)
55
AOMIP coordinated experiments (cont.)
Beaufort Gyre mechanisms of fresh water
accumulation and release (origin of the BG
freshwater reservoir, sources and sinks, role of
sea ice dynamics and seasonal transformations,
Ekman pumping) Fresh water balance of the Arctic
Ocean seasonal and interannual variability
(sources, sinks, pathways) Fresh water regional
studies diversion of liquid FW north of Fram
Strait and impact of this on sea ice Atlantic
water circulation (circulation patterns,
variability and heat exchange, model validation
based on observations)   Ecosystem
experiments Data assimilation and numerical
methods
56
AOMIP activities

• AOMIP collaborators were invited to organize a
  session in MOCA-09 (IAMAS, IAPSO and IACS Joint
  Assembly, to be held in July 19-29, 2009 in
  Montréal, Québec, Canada). Prior to that
  assembly, AOMIP collaborators are also
  participating in the modeling workshop Arctic
  System Modeling Workshop III (International
  Collaboration in Arctic System Modeling) to be
  held on July 16-17, at the University of Quebec
  at Montreal (UQAM)
• http//www.iarc.uaf.edu/workshops/2009/arctic_syst
  em_model_09/
• 2. The AOMIP participants agreed to meet again
  in fall 2009 (October 21-23, 2009), at Woods Hole
  Oceanographic Institution to report about
  numerical experiments and other project results

57
AOMIP WGOMD collaboration

• Take advantage of
• Regional expertise in key region
• Global exchanges, esp. Bering Strait CAA/Fram
  Strait
• Identical forcing fields (CORE II)
• Data repository/postprocessing

58
Ende
Ende
Foto L. Tadday