Piskozub

Are decadal-scale climate predictions possible?
Large scale air-sea interaction modeling
and process studies
Jacek Piskozub
Institute of Oceanology PAS, Sopot, Poland
COST Conference, Sopot, 7-8 April 2014
Global warming is marching on
Temperature anomalies (global GISS data series) of decades (the last
one ends on February 2014) vs. base period 1961-1990 with uncertainty
(2 σ) shown.
Piskozub 2014 (unpublished)
What causes the multidecadal variability of
global temperature?
On top of the global warming trend, the instrumental temperature records
show significant multidecadal variability. To better learn the actual climate
sensitivity to greenhouse gases, we need to understand how much of it is
caused by changing forcing (natural and anthropogenic) and how much by
natural variability? And what this variability actually is?
The Copenhagen Diagnosis 2009
The trend is easily attributable to CO2. But what
about the “wiggles”?
Using only the
transitional climate
sensitivity (2 K) and the
fact that radiative
forcing depends
logarithmically on
atmospheric CO2
concentration, it is easy
to reconstruct the
general shape of
instrumental global
temperature data
series. The residuum is
dominated by ENSO on
2-7 year time scales
and a 70 year
oscillation around the
CO2 predicted
temperature.
Piskozub 2012 (unpublished)
No 65 year cycle in the solar activity spectrum from
tree ring C14
Solar activity from tree ring 14C shows many cycles but the shorter
important one is 88 year long. It is difficult to explain any multiannual or
multidecadal variability of a shorter period by sun only. We also know
of no other external decadal periodical forcings.
Dean 2000 (USGS)
Modern view of thermohaline circulation
Deep waters return to the ocean surface thanks to turbulent mixing
(especially on ocean ridges) and upwelling driven by Ekman transport
around the Antarctic (pushing surface waters north, that is leftwards from
the wind direction – Southern Hemisphere!)
Rahmstorf 2006
Atlantic Multidecadal Oscillation
Atlantic Multidecadal oscillation is a periodical (60-70 years) warming
and cooling of North Atlantic. The temperature anomaly of North
Atlantic (top) is used as the AMO index.
Sutton, Hodson 2005 (Science)
Attribution of global temperature anomalies: now with AMO
Attribution of 20th
century global
temperature variability
to particulate natural
and anthropogenic
forcings (including
AMO, added as an
additional forcing)
from multiple
regression analysis.
Piskozub & Gutowska 2011 (EGU 2011)
Anthropogenic and AMO related component of
decadal temperature trends
Comparison of anthropogenic and AMO contributions to temperature change
between subsequent decades. It is obvious that AMO may increase or
decrease decadal trends by up to 1/3 of their values and that the next change
will be rather down than up (assuming the 70 year period will continue)
Piskozub & Gutowska 2011 (EGU 2011)
The map show only regions of statistically significant correlation.
Piskozub & Gutowska 2013
The map show only regions of statistically significant correlation.
Piskozub & Gutowska 2013
Atlantic surface
circulation
North Atlantic surface
circulation is controlled by
two large gyres: the
anticyclonical Subtropical
Gyre and cyclonic Subpoolar
Gyre. The wind forced Gulf
Stream and its (supposed)
continuation, the North
Atlantic Drift (or Current)
divide the two gyres of
different water properties
(the Subpolar Gyre is much
colder and less salty.
Tomczak & Godfrey 2002
(a book chapter)
Anticorrelated changes of T and S in the two gyres
1980-2000 minus 1950-1970
Temperature and salinity changes in the Subtropical and Subpolar Gyres are
anticorrelated on multidecadal time scales. In 1950 – 2000 overturning
circulation decreased by -1.5 ± 1 Sv in the former and increased by +0.8 ± 0.5
Lozier et al. 2010 (Nature Geoscience)
Sv in the latter.
AMOC exchanges heat between ocean and the
atmosphere on decadal scales
The temperature of subsurface tropical North Atlantic and surface waters
are anticorrelated on decadal scales showing that AMOC exchanges heat
not only between the hemispheres but also between the ocean and the
atmosphere with an AMO-like timescale
Zhang 2007 (GRL)
Zmienność
Wanner et al. 2001
Wintertime NAO values since 1950
Wintertime has similar values on decadal scales. In fact it look almost like a
cycle of... about 65 years.
http://www.cpc.ncep.noaa.gov/products/precip/CWlink/pna/season.JFM.nao.gif
Linear temperature trends
(a) The linear trend in areaaveraged global land surface
temperature (°C per 5 years)
determined from CRUTEM3
for five different periods:
1979–2010 (dark red), 1987–
2010 (light blue), 1993–2010
(green), 1999–2010
(orange), and 2005–2010
(pink). Filled bars represent
trends that are significant at
the 95% significance level.
(b) As in Figure 1a
but only for the NH (20°N–
90°N). (c) As in Figure 1a but
for the tropics (20°N–20°S).
(d) As in Figure 1a but for
the Southern Hemisphere
(SH; 20°S–90°S).
Cohen et al 2012 (GRL)
Seasonal trends and patterns
Surface temperature anomalies and trends (poleward of 20 N) and spatial
patterns for four seasons.
Cohen et al 2012 (GRL)
Can we predict NAO values?
Correlations of summer ocean
SST with NAO index of the
following winter (Rodwell et al.
1999)
British Met Office has a
hindcasting 2/3 successful
prediction of winter NAO
sign from North Atlantic SST
of the previous summer.
Rodwell, Rowell & Folland 1999 (Nature) & www.metoffice.gov.uk
NAO leads AMO by 10 years!
NAO index (shaded) seems to lead North Atlantic temperature (a simple
measure of THC volume) by 10 years (thick line is a 11-year running mean).
The mechanism explaining the phenomenon is supposed to be the NAO
effect on Labrador Sea deep convection (top is Labrador Sea Water
thickness in meters).
Latif et al. 2006 (Journal of Climate)
This is how two sine functions with a phase
difference cross-correlate
The result of cross-correlation of the sine functions of 64 year period and a
15 year phase difference with no white noise added.
Piskozub 2013 (IAPSO)
Same sine functions with some noise added
The result of cross-correlation of the sine functions of 64 year period and a
15 year phase difference with some white noise added.
Piskozub 2013 (IAPASO)
This is NAO cross-correlated with AMO
NAO index seems to lead AMO by about 15 years while AMO leads NAO
9with inverted sign) by another ca. 15 years. It looks almost exactly as if the
two indices were the same approximately 60 year cycle but with different
phases.
Piskozub 2013 (IAPSO)
Do we have a warming “hiatus”?
The graph shows the global mean surface temperature relative to
the 1961–90 mean, based on the HadCRUT4.2.0.0 data set. The inset
shows the 1993–2012 time span, with green denoting La Niña years and
red, El Niño years; the size of the symbol indicates the strength of
La Niña/El Niño (the Niño index for year N is computed by averaging from
October of year N−1 to September of year N).
Held 2013 (Nature)
Two phases
of ENSO
La Niña
El Niño
Global temperature and ENSO
Comparison of global surface temperature (NOAA series) and NINO3.4
index of Central Tropical Pacific SST.
Trenberth & Fasullo 2013 (Earth's Future)
Global temperature: data and modelling
Annual mean time series based on observations, HIST and POGA-H
(model runs with historical forcings and in the latter case forced historical
SST od tropical eastern Pacific. Bars on the right show the ranges of
ensemble spreads of the 2002–2012 averages
Kosaka & Xie 2013 (Nature)
Ocean heat content
OHC integrated from 0 to 300 m (grey), 700 m (blue), and total depth
(violet) from ORAS4 ocean reanalysis, as represented by its 5 ensemble
members
Ballmaseda Trenberth Callen 2013 (GRL)
PDO and NAO
Comparison of North Atlantic Oscillation (NAO) and low pass filtered
Pacific Decadal Oscillation (PDO).
Trenberth & Fasullo 2013 (Earth's Future)
ENSO is in phase with NAO!
The Multivariate ENSO Index (MEI) correlates with NAO with only a 1-2 year
lag (NAO is lagged behind MEI) and anticorrelates for lags > 20 yrs which
suggests both are cycles almost in phase with each other.
Piskozub & Gutowska 2014 (EGO)
If you attend EGU 2014, please do not skip
my talk about the processes driving “teleconnections”
between ENSO, AMO, NAO and PDO.
Conclusions
The Earth climate seems react to cyclical variability with a
prriod of approximately 65 years
●Understanding this variability is crucial for constraining climate
sensitivity to forcings (greenhouse gases) and useful for making
decadal scale climate predictions.
●Variability with this time scale needs to be externally forced (no
evidence for that) or ocean related
●Different ocean basins show variability of similar period but
different phases
●All the cycles of different phases may be symptoms of the
same planetary scale variability involving at minimum Meridional
Overturning Circulation and NH atmospheric circulation
●Models still lack the capability to represents this variability
(though some show its elements)
●
Why sea ice NH & SH trends are so different?
Sea ice in Northern
Hemisphere has a
consistent decreasing
trend.
However the Southern
Hemisphere sea ice has a
weak increasing trend in
the whole satellite era.
IPCC Report 2007 (technical_summary)
The pattern of Antractic sea ice trends
Trends of sea ice cover fraction per decade in 1979 – 2012. Black line
delineates areas of statistically significant trends.
King 2014 (Nature)
Is Atlantic responsible
for sea-ice trends
around the Antarctic?
North Atlantic temperatures (with the satellite era marked by the box) (top)
and (a) SLP (contours indicate), SAT (land-area color) and SIC (ocean-area
color – inverted colors!), individually regressed against the normalized
tropical Atlantic SST, (b) difference of values for 'high” (1996-2012) and “low”
(1979-1995) AMO periods, (c) modeling results.
Li et al 2014 (Nature)
Thanks for attention
Anthony Casay, 1994