Post on 09-Aug-2019
Central Pacific coral evidence for increased ENSOvariance over the past 6,000 years
Helen McGregor, Steven Phipps, Matthew Fischer, Michael Gagan,
Laurent Devriendt, Andrew Wittenberg, Colin Woodroffe, Jian-Xin Zhao,
Jessica J. Gaudry, David Fink, Allan Chivas
Contributions from paleo-ENSO reconstructions
2. Extend the ENSO records to periods when the forcing is different
1. Extend the instrumental record
(e.g. Holocene; orbital forcing changes in seasonal distribution of insolation)
Reduced ENSO variance in the mid-Holocene, but…
J. Lough
<<
Paleo-ENSO variance 50-80% reduction
Model ENSO variance 10-15% reduction
Is Holocene ENSO variance really reduced?
Cobb et al. 2013 Science
Compilation of Holocene coral records suggests otherwise
Test the response of ENSO to changes in orbital forcing over the past 6,000 years
SST anomaly (°C)
• 255 years of new Poritescoral d18O data (non-continuous) combined with published records
• Single site, Kiritimati(Christmas) Island in the heart of the NINO3.4 region
• Well-preserved material
• U-Th dated
Porites coral microatolls at Kiritimati Is.
4 cm
Porites microatoll d18O and instrumental climate data
McGregor et al. Geochim. Cosmochim. Acta 2011
~4,300 yrBP microatoll – reduced ENSO
Positive anomaly years (El Niño)
Negative anomaly years (La Niña) Neutral years
Absence of moderate-strong eventsMcGregor et al. (2013) Nature Geoscience
A slight digression…
% of interannual record explained by the each year type
Inst. SST anomaly
Positive anomaly years (El Niño)Negative anomaly years (La Niña)Neutral years
4,300 yr BP d18O anomaly
What did the average El Niño event look like 4,300 yr BP?
Seasonal timing differences?
ENSO asymmetry?
What do these changes mean?
~4,300 yrBP microatoll – reduced ENSO
Positive anomaly years (El Niño)
Negative anomaly years (La Niña) Neutral years
Absence of moderate-strong eventsMcGregor et al. (2013) Nature Geoscience
New monthly resolved coral d18O data from Kiritimati Island
Pink = new coral data; red = 2-8 year band pass filter; grey = published modern coral data
Compilation of published and new records from KiritimatiIsland, central Pacific
For Kiritimati correlation of s.d. versus time is significant for• all records (pearson r 0.71 99% CI; also significant for all locations)• records < 30 years or < 50 years are excluded (pearson r 0.9, 0.93 99% CI)• the high-variance, late 20th Century coral records are excluded (pearson r 0.65,
95% CI)
0.20
0.15
0.10
0.05
2-8
ye
ar
filter
sta
nd
ard
devia
tion
(‰
)
6000 5000 4000 3000 2000 1000 0
Calendar Year BP
Comparison of coral data to Mk3L 8000 year unforced transient model simulation NINO3.4 SST
Trend coefficient (Ds/ka)
Fre
quency
010
20
30
40
50
−0.008 −0.004 0.000 0.004 0.008
Obs
Coral obs
Trend is not reproduced in the unforced model simulation when sampled with the same autocorrelation characteristics and record lengths as the coral dataset
0.50
0.45
0.40
0.35
0.30
Mk3L 2
-8 b
p filt
er
NIN
O3.4
SS
T s
.d.
(de
g C
)
8000 6000 4000 2000 0
Calendar Year BP
Mk3L 1000-year orbitally forced snapshot simulations: 2-8 bandpass filtered NINO3.4 SST
Holocene Kiritimati Island coral trend in ENSO variance is qualitatively similar to trend simulated by Holocene snapshot simulations by the Mk3L model
Why discrepancies?
Positive anomaly years (El Niño)
Negative anomaly years (La Niña)
Neutral yearsMcGregor et al. (2013) Nature Geoscience
1. Record length (e.g. Wittenberg 2009)
Short records not necessarily representative
ENSO - observations
Jan 1998
Feb 1997
0 4 8 1612 20
CMAP rainfall (mm/day)IGOSS SST (°C)
10 14 18 2622 30
SST Rainfall
Why discrepancies?
2. Location, location, location – different SST and rainfall influence on d18O
Conclusions
Analysis of coral d18O at a single site (Kiritimati Island) suggests
Increase in ENSO variance from the 6000 years ago to present
ENSO variance outside the range expected from internal, unforced variations
An ENSO response to orbital forcing
Long records emerging with the opportunity to calculate ENSO metrics for periods different from today.
0.50
0.45
0.40
0.35
0.30
Mk3L 2
-8 b
p filt
er
NIN
O3.4
SS
T s
.d.
(de
g C
)
8000 6000 4000 2000 0
Calendar Year BP
1. MODERN KIRITIMATI CORAL MICROATOLLS AND THEIR CLIMATE SIGNALS
Microatoll d18O reproducibility
Modern coral X-ray
Within-microatoll variabilityBetween-microatoll variability
McGregor et al. Geochim. Cosmochim. Acta 2011
ENSO variance from ~4,300 yrBP microatoll
79% reduction in ENSO variance ~4,300 years ago(dominent period 6-8 years)
McGregor et al. (2013) Nature Geoscience
79% reduction in ENSO variance ~4.3 kyBP: Outside the range expected for ENSO with no forcing?
Wittenberg (2009) – GFDL CM2.1 unforced model simulation
Woodroffe and Gagan 2000, Woodroffe et al. 2003; McGregor et al. 2011
El Niño year
Kiritimati Porites microatoll d18O and Sr/Ca
Kiritimati microatolls are faithful recorders of ENSO
Compilation of published and new records from the Line Islands, central Pacific
For Kiritimati correlation of s.d. versus time is significant for• all records (pearson r 0.71 99% CI; also significant for all locations)• records < 30 years or < 50 years are excluded (pearson r 0.9, 0.93 99% CI)• the high-variance, late 20th Century coral records are excluded (pearson r 0.65,
95% CI)
Ratio of annual cycle to interannual (ENSO) variance
Annual cycle explains 30% fossil coral variance, compared to 10% for modern coral stack
McGregor et al. (2013) Nature Geoscience
Changes in the annual cycle 4,300 yrBP
Green = modern (1972-2006), orange = 4,300 yrBP
• Strengthened annual cycle
• Present-day annual cycle at KiritimatiIs. is a ‘knock-on’ from eastern equatorial Pacific (EEP)
• Orbital forcing affecting the annual cycle in the EEP and affecting ENSO?
• Today strengthened annual cycle and reduced ENSO (e.g. 1920s-1950s)
• Is this a clue as to the mechanism for reduced mid-Holocene ENSO?
160˚W 140˚W 120˚W 100˚W 80˚W
EQ
10˚N
JAS
b
21 22 23 24 25 26 27 28 29 30
SST (˚C)
a
McGregor et al. (2013) Nature Geoscience
79% reduction in ENSO variance ~4.3 kyBP: Outside the range expected for ENSO with no forcing?
Wittenberg (2009), Phipps et al. (2011)
GFDL CM2.1 Mk3L
~4.3 kyBP ENSO is outside the range of internal, unforced ENSO. Another driver = external forcing (orbital changes)
Interdecadal amplitude variability ~4,300ka
McGregor et al. (In press) Nature Geo
Correlation of Kiritimati microatoll d18O with rainfall and SST
McGregor et al. 2011
Wavelet & phase plots for Kiritimati microatoll d18O, ERSST, & rainfall
Semi-annual
Annual
Inter-annual
McGregor et al. 2011
Compilation of modern d18O coral records from Kiritimati Island
Evans et. al 1998; Woodroffe & Gagan 2000; Woodroffe et al. 2003; Nurhati et. al 2009; McGregor et al. 2011; McGregor, et al. 2013
Porites coral microatolls at Kiritimati Is.
4 cm
Microatoll X-ray
McGregor et al. Geochim. Cosmochim. Acta 2011
Dome-shaped Porites coral
20 cm
Porites microatoll
Porites microatoll vs dome-shaped Porites
20 cm
Porites microatoll
Dome-shaped Porites coral
Fossil microatolls at Kiritimati Is.Kiritimati (Christmas) Island – optimal site
Hundreds of fossil microatolls
~4,300 year-old coral microatoll XM35, 175 year-long monthly resolved d18O record
McGregor et al. Quaternary Geochronology 2011
But what about the longer term picture over the Holocene?
Cobb et al. 2013 Science
3. SEASONAL CYCLE AND ENSO RESULTS FROM FOSSIL CORAL MICROATOLLS
How did El Niño and La Niña events evolve at the seasonal scale?
What can this tell us about the ENSO processes during the mid-Holocene?
ENSO: interannual variance yet seasonal changes
Jan 1998 - El Niño pattern
Feb 1997 - normal pattern
0 4 8 1612 20
CMAP rainfall (mm/day)IGOSS SST (°C)
10 14 18 2622 30
Inst. SST anomaly
1
0
-1
SS
Tan
om
(°C
)
J A S O N D J F M A M J
Month
2525
1515
55
Po
wer
S(f
)
1086420
Period (years)
ENSO power spectra
~4,300 yrBP microatoll – seasonal-scale ENSO processes
Positive anomaly years (El Niño)
Negative anomaly years (La Niña) Neutral years
McGregor et al. (2013) Nature Geoscience
East-central Pacific SST annual cycle
Apr-Jun
Jul-Sep
WOA09 SST data
Oct-Dec
Jan-Mar
4.3 kyBP insolation, annual, trade winds, changes counter ENSO development…
• Increased insolation north of equator• Strengthened ITCZ• Strengthened east Pacific
meridional winds and SST gradient • Cooler EEP SSTs• Strengthened E-W trade winds• Enhanced annual cycle
• Amping up the trade winds during the season when ENSOs develop• Reduced ENSO variability
Jul-Sep
East-central Pacific SST annual cycle
Apr-Jun
Jul-Sep
WOA09 SST data
Oct-Dec
Jan-Mar
Mid-Holocene: Stronger EP annual cycle, weaker ENSO
Jul 1996
120˚E 150˚E 180˚ 150˚W 120˚W 90˚W 60˚W
Longitude
20˚S
20˚N
Latitu
de
Jul 2012
120˚E 150˚E 180˚ 150˚W 120˚W 90˚W 60˚W
Longitude
20˚S
20˚N
Latitu
de
Nov 1996
120˚E 150˚E 180˚ 150˚W 120˚W 90˚W 60˚W
Longitude
20˚S
20˚N
Latitu
de
Nov 1997
120˚E 150˚E 180˚ 150˚W 120˚W 90˚W 60˚W
Longitude
20˚S
20˚N
Latitu
de
Nov 2012
120˚E 150˚E 180˚ 150˚W 120˚W 90˚W 60˚W
Longitude
20˚S
20˚N
Latitu
de
Jul 1997
120˚E 150˚E 180˚ 150˚W 120˚W 90˚W 60˚W
Longitude
20˚S
20˚N
Latitu
de
Reference neutral: July & November 1996
Reference El Niño: July & November 1997
Partial ~4,300 yrBP analogue: July & November 2012
18˚C 20˚C 22˚C 24˚C 26˚C 28˚C 30˚C
Sea Surface Temperature Reynolds et al. 2002 (ERSST)
Mid-Holocene: Stronger EP annual cycle, weaker ENSO
•More northerly position of the ITCZ (Haug et al. Science 2001) increased cross-equatorial winds, stronger meridional SST gradient, stronger annual cycle•Paleo evidence
•Cooler SSTs eastern equatorial Pacific (e.g. Koutavas et al. Geology 2006)•Stronger upwelling (e.g. Carre et al. Quat. Int. 2012)•Modelling study suggests seasonality may be an important factor (Braconnot et al. Cli. Dynam. 2011) and our hypothesis is consistent with idealised model experiments (Timmermann et al. J. Clim. 2007)
• Implication: ENSO is affected by external forcing via changes in month-to-month annual cycle processes
Jul-Sep
2. MID-HOLOCENE CHANGES IN ENSO VARIANCE
Origins of interest in mid-Holocene ENSO
Holocene: 0-10,000 years ago (mid-Holocene: 4,000-6,000 years ago)Proxy records: Indirect measures of climate from biological or inorganic archives
Early work: Pollen records from lakes, lake sedimentation records, mollusc species changes, model studies, coral oxygen isotope records (d18O) Indicated ENSO had strengthened since the early to mid-Holocene
Could the mid-Holocene be a test-bed for improving our understanding of ENSO?Similar but different (orbital forcing)Longer recordsAlternate ENSO behaviours
Tudhope et al. Science 2001; Koutavas et al. Geology 2006 after Moy et al. Nature 2002
Laguna Pallcacocha (Ecuador) sediment color intensity
Papua New Guinea coral d18O ENSO standard deviation
Challenges to understanding ENSO from proxy records
1. Seasonally-resolved records capture the full range of ENSO variance but are not long enough (e.g. Wittenberg 2009)
2. Forced models do not fully capture the mid-Holocene changes
3. If long enough then do not have the seasonal resolution and/or only sample part of ENSO variance (e.g. Moy et al. 2002)
Laguna Pallcacocha (Ecuador) sediment color intensity
4. Many records are from teleconnected regions (e.g. Donders et al. 2008)
Interdecadal amplitude variability past 1ka
Fowler et al. Nature Climate Change (2012)
Insolation at the equator and 10N for 4.3 kyBP and present
El Niño-Southern
Oscillation
http://svs.gsfc.nasa.gov/vis/a000000/a000200/a000287/index.html
Southern Oscillation Index –pressure difference between Tahiti and Darwin
NINO3.4 Index – SST anomaly in the NINO3.4 region in the central/eastern equatorial Pacific (5°N-5°S, 120-170°W)
Contribution from paleo-ENSO research:
• Extend ENSO records from present back in time a few hundred centuries
• Extend the ENSO records to periods when the forcing is different enough to detect a change
– Improve models and predictability (Models do not fully capture ENSO processes)
Can ENSO respond to external forcing or is it a random oscillation of the climate system
(internal variability)?
El Niño-Southern Oscillation – driver of global interannual climate variability
-4 -3 -2 0-1 4321
IGOSS SST anomaly (°C)
El Niño
La Niña
Global impacts (e.g. Australia)
http://esminfo.prenhall.com/science/geoanimations/animations/26_NinoNina.html
ENSO predictability
Forecast skill 3-6 months in advance
“Tropical Pacific Ocean moves closer to El Niño” BOM, 18 November 2014
Australian Bureau of Meteorology
ENSO predictability
Forecast skill 3-6 months in advance
Results from IPCC-class models
IncreasedENSO variability
DecreasedENSO variabilityC
han
ges
in s
td. d
ev. o
f a
mea
n s
ea le
vel p
ress
ure
in
dex
(re
late
d t
o t
he
SOI)
✖ Models give conflicting forecasts for longer timescales
Collins et al. Nature Geoscience 2010
Barrier to ENSO predictability – intrinsic variability
CSIRO & BOM Climate change in Australia 2007
La
Niña
El
NiñoAnnu
al S
outh
ern
Oscill
ation
Ind
ex
Can ENSO respond to external forcing or is it a random oscillaton of the climate system (internal variability)?
ENSO - observations
Jan 1998 - El Niño pattern (high rainfall and warm SST central & eastern Pacific)
Feb 1997 - normal pattern (high rainfall and warm SST western Pacific)
-4 -3 -2 0-1 43210 4 8 1612 20
CMAP rainfall (mm/day) IGOSS SST anomaly (°C)IGOSS SST (°C)
10 14 18 2622 30
Jan 1999 - La Niña pattern (low rainfall and very cool SST central & eastern Pacific)
Middle Holocene ENSO
Early work: ENSO had strengthened since the early to mid-Holocene Inferred changes due to orbital
forcing
New coral records suggest ENSO displays large internal variability irrespective of forcing
Laguna Pallcacocha (Ecuador) sediment color intensity
Coral d18O ENSO variance
Koutavas et al. Geology 2006 after Moy et al. Nature 2002; Cobb et al. 2013 Science
Controversies & future directions
Teasing out the forced vsinternal variability signals
Seasonally-resolved records are still temporally and spatially sparse
Working with models to ENSO processes Understand the drivers of
past ENSO
Improve models and data-model comparison (e.g. seasonality and ENSO)
Document and understand the nature of ENSO teleconnections
Kiritimati (Christmas) Island
Kiritimati fossil microatoll d18O - details
Kiritimati fossil microatoll d18O
Woodroffe et al. (2003), McGregor et al. (in prep)
ENSO strength & relationship to annual cycle strength
a) Ratio of Annual:Interannual variance versus timeb) Interannual variance versus time c) A bivariate plot of the two indices in a, b
J
j j
jn
ns
sW
C
tjx
05.0
0
5.0 )(
)0(
dd
d
Wavelet Reconstruction:
Contributions from paleo-ENSO reconstructions
2. Extend the ENSO records to periods when the forcing is different
1. Extend the instrumental record (e.g. ICOADS Gilbert Islands, 3S-4N, 172E-177E)
(e.g. Orbital forcing changes in seasonal distribution of insolation)