of 20
1
Geophysical Research Letters
Supporting Information for
Termination 1 Millennial
-
scale Rainfall Events over the Sunda Shelf
F. Buckingham
1
, S.A. Carolin
1,2
, J.W. Partin
3
, J.F. Adkins
4
, K.M. Cobb
5
, C.C. Day
1
, Q. Ding
6
, C. He
7
,
Z. Liu
7
,
B. Otto
-
Bliesner
8
, W.H.G. Roberts
9
, S. Lejau
10
, J. Malang
10
1
Department of Earth Sciences, University of Oxford, Oxford, UK
2
Department of Earth Sciences, University of Cambridge, Cambridge, UK
3
Institute for Geophysics, Jackson School of
Geosciences, University of Texas at Austin, Austin, TX, USA
4
Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA, USA
5
Department of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, GA, U
SA
6
Department of Geography and Earth Research Institute, University of California Santa Barbara, Santa Barbara, USA
7
Department of Geography, The Ohio State University, Columbus, OH, USA
8
Climate and Global Dynamics Laboratory, National Center for
Atmospheric Research, Boulder, CO, USA
9
Geography and Environmental Sciences, Northumbria University, Newcastle
-
Upon
-
Tyne, UK
10
Gunung Mulu National Park, Sarawak, Malaysia
Contents of this file
Text S1 to S
5
Figures S1 to S
9
Tables S1 to S2
Additional Supporting Information (Files uploaded separately)
Captions for
Datasets
S1 to
S
3
2
Introduction
Here we provide more detailed information regarding: S1) the field site, S2) the measurement
of radiogenic U and Th isotopes and calculation of t
he individual U/Th ages, S3) measurement
of the stable isotopes, S4) construction of the interpolated age model, and S5) calculation of
approximate pseudo
-
stalagmite calcite
δ
18
O from iTRACE modeled rainwater
δ
18
O. Figures
include a map of the field site (Fig. S1), an image of stalagmite SC02 with sample locations (Fig.
S2), the Oxcal interpolated age model (Fig. S3), plot comparison of the timing of millennial
scale events in stalagmite SC02 versus stalagmite SCH
02 given a range of initial detrital
thorium ratios used to correct the stalagmite U
-
Th ages (Fig. S4),
map of the location of the six
regional cave sites and chosen model grid points for model
-
data comparison (Fig. S5),
calculated
δ
18
O
calcite
from iTRACE
δ
18
O
rain
and surface temperature for
the
six cave site locations
around the region
using three different
fractionation factor equations
(Fig.S
6
)
,
iTRACE
modeled change in mean annual rainfall and precipitation
-
weighted rainwater
δ
18
O in
response to HS1 and
YD forcing in the IPWP (Fig. S
7
), maps of iTRACE, TraCE
-
21ka, and
HadCM3 modeled modern mean annual and seasonal rainfall compared with observation
maps (Fig. S
8
a
), modeled LGM mean annual rainfall compared with modeled modern mean
annual rainfall (Fig. S
8
b
), and modeled HS1 mean annual rainfall compared with modeled LGM
mean annual rainfall (Fig. S
8
c
)
, and mapped iTRACE modeled change in precipitation
-
weighted seasonal rainwater
δ
18
O in response to HS1 and YD forcing (Fig. S
9
)
. Two data sets
including U
-
T
h measurements and stable isotope measurements are found in separate excel
files.
Text
S1
Field Site.
Stalagmite SC02 was
collected
during a 2006 field expedition
from
Secret Chamber in
Gunung Mulu National Park (4
°
6
N, 114
°
53’E). Three outcrops of Melinau limestone, covered
with dense tropical rainforest, are located on the western shale and sandstone slopes of
2375m high Gunung Mulu
in northern Borneo
.
The base of the Melinau limestone outcrops
are
~120 meters above sea lev
el (masl). Secret Chamber
is located
within the passages of
Clearwater Cave
, found on
the western slopes of Gunung Api
(Fig. S1).
Text
S
2
U
-
Th dating.
Carolin et al. (2013) reported U
-
Th ages on the lower portion of stalagmite SC02, 639
-
274
mm distance from stalagmite top, 105
-
31 ka, analysed in the Division of Geological and
Planetary Sciences at
the California Institute of Technology
. For this study, 1
4
additional U
-
Th
ages were measured in the upper portion of SC02 in the Department of Earth Sciences at the
University of Oxford. Three
additional
previously unpublished U
-
Th ages
(sample IDs cc09,
cd06, cd17)
measured at Caltech in 2011 following the met
hods of Carolin et al. (2013)
, as well
as one
U
-
Th age
published in
Carolin et al. (2013
) (sample ID cd19), are used to construct the
upper portion SC02 age model (Figure S2)
.
For the
U
-
Th sample
analyses at the University of Oxford, each
U
-
Th sample
trench was
drilled using a handheld drill with a 0.8mm diameter spherical drill
-
tip, followed by a 0.8mm
diameter flat
-
headed drill
-
tip.
The drill
-
tips were cleaned after each sample in separate
solutions of 2% nitric acid and de
-
ionised water
. The eighteen sampling sites were chosen
based upon the appearance of the lamina, whereby the surface of each stalagmite was
illuminated and dampened with de
-
ionised water until distinct, light
-
coloured laminas were
visible. Contamination was avoided by d
rilling the surface layer of each sampling site and
3
removing the surface powder with compressed air. Each trench followed the curvature of the
growth lamina and ranged from 10
-
20mm long and 2
-
4mm deep.
In a metal free laboratory, samples were weighed to th
e nearest 0.1 mg. Samples were
dissolved in concentrated HNO
3
and a known mass of
236
U/
229
Th spike was added. The samples
were refluxed on a hot plate overnight, dried down and dissolved in nitric acid. U and Th were
extracted from the sample solution usin
g anion exchange columns. 875
μ
L of 2% HNO3 was
added to the samples before U and Th was measured on a Nu
-
Instrument Multicollector
-
Inductively
-
Coupled
-
Plasma Mass Spectrometer (MC
-
ICP
-
MS) at the University of Oxford.
Abundance sensitivity for U measurement
s was corrected based on the half
-
mass
dynamic measurements at 236.5, 235.5, 234.5 and 233.5 on the ion
-
counter during the initial
part of the U measurement. Abundance sensitivity for Th measurements was corrected based
on the half
-
mass dynamic measurement
s at 230.5, 229.5 and 228.5 on the ion
-
counter during
the final part of the Th measurement. Machine biases (ion counter gains and mass
fractionation) were corrected using sample
-
standard bracketing with CRM
-
145 for U, and two
in house 229Th
-
230Th
-
232Th sta
ndards for Th.
The initial detrital
230
Th/
232
Th value of 111 ± 41 ppm was previously calculated for SC02
based on two isochrons measured in separate stalagmites from Secret Cave in Mulu (Carolin et
al., 2013).
We use this value (111 ppm) for calculating t
he initial
-
Th corrected U
-
Th ages. For
this study, we also calculate a suite of initial
-
Th corrected U
-
Th ages using the full range of
initial
230
Th/
232
Th value
s at 10ppm intervals (70ppm, 80ppm, 90ppm, 100ppm, 110ppm,
120ppm, 130ppm, 140ppm, and 150ppm) f
or comparison (e.g. Figure S4).
A Matlab Monte Carlo script was used to calculated the absolute age and age errors
associated with each sample using the half lives presented in Cheng et al. (2013), 245,620kyrs
(234U) and 75,584kyrs (230Th), following the
procedure of Carolin et al. (2013).
Text
S
3
Stable isotope measurements.
Stable isotope samples were milled along the growth axis of SC02 in the Department of
Earth Sciences at the University of Oxford using a New Wave Micromill. Samples were milled at
1 mm intervals in the upper portion of the SC02
-
B slab piece, 182.4
-
147.4 mm
distance from
top of SC02. Sampling resolution was increased to 0.2 mm interval in the lower portion of the
SC02
-
A slab piece, 148.3
-
103.3 mm distance from top. Two trenches were milled in the SC02
-
A
slab due to the shift in growth axis at 117.1 mm distan
ce from top.
In the 1 mm increment section, calcite was excavated by drilling 1 mm diameter circular
holes to a depth of 0.3mm. In the 0.2 mm increment section, a 0.8mm flat
-
headed drill
-
bit was
used to drill a 3mm wide trough perpendicular to the growth
axis, to a depth of 1.5mm. Before
sampling, the top 0.037
-
0.050mm slab surface was drilled and removed with compressed air
to remove surface contamination.
~40
% of the milled samples were analysed to produce a
stable isotope record with an average temporal
resolution of ~
50
years
after 15 ka, and ~200
years before 15 ka
.
30
-
60μg of each sample was analysed on a Delta V Advantage isotope ratio mass
spectrometer coupled to either a Kiel IV carbonate device or Gas Bench II introductory system
at the Universit
y of Oxford.
A selection of repeat
samples were measured on both
Kiel and
Gasbench
introductory systems to confirm consistency.
The calcite powders were reacted with
100% phosphoric acid at
71
°
C.
The
relative
13
C/
12
C
values
are reported in the conventiona
l
δ
13
C (‰) notation, relative to
the Vienna Pee Dee Belemnite (
V
-
PDB
), by assigning a value of
+1.95
exactly to NBS
-
19
.
The relative
18
O/
16
O values are reported in the conventional
d
18
O
(‰)
notation, relative to V
-
PDB, on a normalised scale such that the
d
18
O of NBS
-
19 is
-
2.2‰.
4
Primary and secondary standards were used for calibration and for determining precision and
accuracy.
Average m
easurement error
is
less than 0.0
5
‰ (1
σ
) for
δ
13
C
and
0.
10
‰ (1
σ
) for
δ
18
O.
Text
S4 Oxcal
Poisson
-
process deposition
age model and interpolated ages.
The Poisson
-
process deposition model feature in OxCal(v4.4) was used to interpolate
between the eighteen U/Th ages in the younger portion (31 ka to present) of S
C02 (Bronk
Ramsey, 2008, 2009; Bronk Ramsey and Lee, 2013). OxCal uses Markov Chain Monte Carlo
(MCMC) sampling to produce a distribution of potential solutions using the Metropolis
-
Hastings algorithm (Bronk Ramsey, 2008). The P_sequence() function allows
for fluctuations in
the deposition rate, of which the model finds the most appropriate value through
computation. We chose the default initial conditions for the Poisson
-
process model:
k
0
= 0.1
mm
-
1
, log
10
(k/k
0
) = U(
-
2,2). Boundaries were included in the model input at 143.9 mm, 140.9
mm, 96.6 mm, and 94.8 mm distance from
the
top of SC02, where detrital layers were visible.
The final modeled ages are in stratigraphic order and a probability distribution is provided f
or
each interpolated age at all sampling depths, with a statistical mean and 68% and 95%
confidence ranges.
The OxCal Poisson
-
process deposition model was also used to recalculate the age model
of SCH02
for Figure S4
, using the same initial conditions as u
sed for SC02:
k
0
= 0.1 mm
-
1
,
log
10
(k/k
0
) = U(
-
2,2).
SCH02
U
-
Th ages
input into the Oxcal model
were recalculated using the
Cheng et al. (2013) half lives and Carolin et al. (2013) Monte Carlo
age error analysis.
Boundaries were included in the OxCal model i
nput at the hiatus locations identified in Partin
et al. (2007): 236mm, 235mm, 198.5mm, 197.5mm, 102.5mm, 101mm, 76mm, and 75mm
distance from top of stalagmite.
Text S
5
Forward modeled
δ
18
O
calcite
from iTRACE
δ
18
O
rain
To forward model
δ
18
O
calcite
from
iTRACE
δ
18
O
rain
, we use a best
-
estimate approximation of
the equilibrium fractionation factor of oxygen isotopes between calcite and water at low
temperatures.
Equation S.1
relates the water
-
calcite fractionation factor (
α
calcite
-
H2O
) to
temperature (T = K
evin), based on a collection of modern cave measurements covering a
range of temperatures (Tremaine, Froelich, & Wang, 2011). We feel this equation is the best
available to represent the
18
α
cc/w
versus temperature relationship in
the
tropical cave
environm
ent
s
.
10
3
In
α
calcite
-
H2O
(
18
O) = 16.1(10
3
/T)
24.6
(S.1)
iTRACE
δ
18
O
rain
is reported in per
mil (‰) notation, relative to the water standard VSMOW,
which has oxygen
-
18 / oxygen
-
16 ratio =2.0052x10
-
3
.
Eqns. S.
2
-
S.
4
are used to calculate
δ
18
O
calcite
(
in
per
mil notation, relative to calcite standard VPDB
)
, given
δ
18
O
rain
and
the
fractionation factor
α
calcite
-
H2O
,
R
calcite
=
α
calcite
-
H2O
* (
δ
18
O
rain,VSMOW
/ 1000 + 1) * 2.0052x10
-
3
(S.
2
)
δ
18
O
calcite,VSMOW
= (R
calcite
/
2.0052x10
-
3
1) * 1000
(S.
3
)
δ
18
O
calcite,VPDB
= (
δ
18
O
calcite,VSMOW
30.92) / 1.03092
(S.
4
)
where R
calcite
is the oxygen
-
18 / oxygen
-
16 ratio of the precipitated calcite from rainwater,
δ
18
O
calcite,VSMOW
is the oxygen isotope ratio of the precipitated calcite in per mil notation relative
5
to the VSMOW standard, and
δ
18
O
calcite,VPDB
is the oxygen isotope ratio of the precipitated calcite
in per mil notation relative to the VPDB standard. Eqn. S.
4
is given i
n Kim et al., 2015.
Two other
prominent equations relating the water
-
calcite fractionation factor to
temperature may be considered
when forward
-
modeling
δ
18
O
calcite
from GCM
-
simulated
δ
18
O
rain
: (S.
5
) the experimental calibration of Kim and O’Neil (1997);
and (S.
6
) an empirical
relationship found using
extremely slow
-
growing calcite samples from Devils Hole and
Corchia Cave (Daëron et al., 2019)
,
10
3
In
α
calcite
-
H2O
(
18
O) = 18.03(10
3
/T)
32.42
(
S.
5
)
10
3
In
α
calcite
-
H2O
(
18
O) = 17.57(10
3
/T)
29.13
(S.
6
)
Figure S6
compare
s
forward
-
modeled
δ
18
O
calcite
records, calculated from each of the
fractionation factor
equation
s
S.1, S.
5
, and S.
6
, for the six model grid sites.
All
forward
-
modeled
δ
18
O
calcite
records show similar structure, but with the mean
δ
18
O
calcite
value of the
record calculated using the Kim and O’Neil (1997) fractionation factors offset by
~
-
1.5‰
from the mean
δ
18
O
calcite
value of the records calculated using the other fractionation factors.
Note,
Figure 3 right panel (dark blue reco
rds) shows the
δ
18
O
calcite
records
calculated using
equation S.
1.
6
Figure S1
.
From Ellis et al., 202
0
. Detailed map of Gunung Mulu National Park.
A
dapted from
Mulu
Caves Project, with permission from Jerry Wooldridge.
7
Figure S
2
.
Scan of working slab top
of SC02. U
-
Th sample locations indicated with sample IDs
(see Data Set 1).
The
Oxcal
Poisson
-
process
modeled
mean
U
-
Th age
at each U
-
Th sample
location is list
ed on the right
(kiloyears before 1950). Stable isotope sample trench indicated
with thick yellow line. Sample cd
19
is youngest U
-
Th sample published in Carolin et al. (2013).
SC02 exhibits a dense calcite structure which lacks porosity, suggesting closed
-
system
behaviour.
8
9
Figure S
3
.
SC02
original and
Ox
c
al
-
modeled U
-
Th ages
plotted against
distance
-
from
-
top of
stalagmite
with Ox
c
al
-
modeled interpolated age model.
(a) Full plot, (b) zoomed in on
Termination 1 section (20
10 ka). The location of five detrital
-
rich lamina visible on the
stalagmite face and used as Ox
c
al deposition model
“boundary”
input parameters are noted
with text labels along the y
-
axis, and ind
icated with
a
change in color of the interpolated age
model.
Original U
-
Th ages are plotted as light grey probability density functions, with Oxcal
-
model adjusted ages
plotted as dark grey probability density functions. 68% confidence
ranges of Oxcal
-
model
ed interpolated ages are plotted in dark color, with 95% confidence
ranges of Oxcal
-
modeled interpolated ages plotted in light color. Individual U
-
Th sample IDs
are listed at their corresponding distance
-
from
-
top of stalagmite values along the y
-
axis.
Mode
led boundary ages
are shown as dark grey probability density functions.