Content uploaded by Lisa Tauxe
Author content
All content in this area was uploaded by Lisa Tauxe on Dec 28, 2022
Content may be subject to copyright.
Posted on 8 Dec 2022 — CC-BY 4.0 — https://doi.org/10.1002/essoar.10512962.1 — This a preprint and has not been peer reviewed. Data may be preliminary.
Changes in non-dipolar field structure over the Plio-Pleistocene:
New paleointensity results from Hawai‘i compared to global datasets
Brendan Cych1, Lisa Tauxe1, Geo↵rey Cromwell2, John Sinton3, and Anthony Koppers4
1UC San Diego
2U.S. Geological Survey
3University of Hawai’i
4Oregon State University,Oregon State University
December 8, 2022
Abstract
AfoundationalassumptioninpaleomagnetismisthattheEarth’smagneticfieldbehavesasageocentricaxialdipole(GAD)
when averaged over sufficient timescales. Compilations of directional data averaged over the past 5 Ma yield a distribution
largely compatible with GAD, but the distribution of paleointensity data over this timescale is incompatible. Reasons for
the failure of GAD include: 1) Arbitrary “selection criteria” used to eliminate “unreliable” data vary between studies, so the
paleointensity database may include biased results. 2) The age distribution of existing paleointensity data varies from latitude
to latitude so di↵erent latitudinal averages likely represent di↵erent time periods. 3) The time-averaged field could be truly
non-dipolar. Here, we present a consistent methodology for analyzing paleointensity results and comparing time-averaged
paleointensities from di↵erent studies. We apply it to data from Plio/Pleistocene Hawai‘ian igneous rocks, sampled from fine-
grained, quickly cooled material (lava flow tops, dike margins and scoria cones) and subjected to the IZZI-Thellier technique;
the data were analyzed using the BiCEP method of Cych et al (2021, doi:10.1029/2021GC009755), which produces accurate
paleointensity estimates without arbitrarily excluding specimens from the analysis. We constructed a paleointensity curve for
Hawai‘i over the Plio/Pleistocene using the method of Livermore et al (2018, doi:10.1093/gji/ggy383), which accounts for age
distribution and has robust uncertainties. We demonstrate that even with the large uncertainties associated with obtaining a
mean field from temporally sparse data, our average paleointensities obtained from Hawai‘i and Antarctica (from Asefaw et al.,
2021, doi:10.1029/2020JB020834, reanalyzed here) are not GAD-like after about 1.5 Ma.
1
DRAFT
Changes in non-dipolar field structure over the
Plio-Pleistocene: New paleointensity results from
Hawai‘i compared to global datasets
Brendan Cycha,e, Lisa Tauxea, Geoffrey Cromwellb,f, John Sintonc, and Anthony A.P. Koppersd
a
University of California, San Diego, CA, USA;
b
Occidental College, Los Angeles, CA, USA;
c
University of Hawai‘i at Man
¯
oa, HI, USA;
d
Oregon State University, Corvallis, OR,
USA; eNow at University of Liverpool, England, UK; fNow at US Geological Survey, California Water Science Center, Santa Maria, CA, USA
This manuscript was compiled on December 1, 2022
A foundational assumption in paleomagnetism is that the Earth’s
magnetic field behaves as a geocentric axial dipole (GAD) when av-
eraged over sufficient timescales. Compilations of directional data
averaged over the past 5 Ma yield a distribution largely compati-
ble with GAD, but the distribution of paleointensity data over this
timescale is incompatible. Reasons for the failure of GAD include: 1)
Arbitrary “selection criteria” to eliminate “unreliable” data vary be-
tween studies, so the paleointensity database may include biased
results. 2) The age distribution of existing paleointensity data varies
from latitude to latitude so different latitudinal averages likely repre-
sent different time periods. 3) The time-averaged field could be truly
non-dipolar. Here, we present a consistent methodology for analyz-
ing paleointensity results and comparing time-averaged paleointen-
sities from different studies. We apply it to data from Plio/Pleistocene
Hawai‘ian igneous rocks, sampled from fine-grained, quickly cooled
material (lava flow tops, dike margins and scoria cones) and sub-
jected to the IZZI-Thellier technique; the data were analyzed using
the BiCEP method of Cych et al (2021, doi:10.1029/2021GC009755),
which produces accurate paleointensity estimates without arbitrarily
excluding specimens from the analysis. We constructed a paleoin-
tensity curve for Hawai‘i over the Plio/Pleistocene using the method
of Livermore et al (2018, doi:10.1093/gji/ggy383), which accounts
for age distribution and has robust uncertainties. We demonstrate
that even with the large uncertainties associated with obtaining a
mean field from temporally sparse data, our average paleointensi-
ties obtained from Hawai‘i and Antarctica (from Asefaw et al., 2021,
doi:10.1029/2020JB020834, reanalyzed here) are not GAD-like after
about 1.5 Ma.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Paleointensity
|
Time-averaged geomagnetic field
|
Geocentric Axial
Dipole Hypothesis
P
aleomagnetists use the direction of the magnetization
1
acquired in the Earth’s ancient magnetic field to obtain
2
estimates of the ancient latitude at which the rock formed.
3
Calculation of a latitude relies on an assumption that the
4
Earth’s magnetic field is structured like a bar magnet when
5
averaged over sufficiently long timescales, so that the magnetic
6
field is vertical at the poles, and horizontal at the equator,
7
also termed a Geocentric Axial Dipole (GAD). Estimates of
8
the Earth’s magnetic field direction, taken from different lati-
9
tudes over the past 10 Myr conform relatively well to a GAD
10
field, with a small hemispheric asymmetry (
1
). On the other
11
hand, estimates of the Earth’s magnetic field strength (the
12
paleointensity) averaged over the last 5 Myr consistently show
13
a behaviour incompatible with a strongly dipolar field. A seem-
14
ingly persistent feature in paleointensity data is the presence
15
of weak paleofields at high southern latitudes (
2
–
4
), which
16
causes a hemispheric asymmetry in the paleointensity data.
17
This is seen in paleointensities from the MagIC database over
18
the last 5 Ma (plotted in Figure 1a) where the mean paleoin-
19
tensity at 80
¶
S would be produced by a centered magnetic
20
dipole with a moment of around 40 ZAm
2
, whereas the mean
21
paleointensity at 20
¶
N would require a dipole moment with a
22
magnitude closer to 80 ZAm
2
. Attempts to fit Giant Gaussian
23
Process (GGP) models to paleointensity data to determine the
24
structure of the time-averaged field have found that the field
25
consistently requires a strong quadrupole term 15-30% the
26
strength of the dipole field (
5
,
6
), producing this asymmetry.
27
However, such a large quadrupole is completely incompatible
28
with the directional data. 29
Three different hypotheses could explain the non-dipole
30
like behaviour of global time-averaged paleointensity records: 31
bias in paleointensity estimation, comparison of temporally
32
distinct data in a time varying field, and genuine non-dipole
33
field behavior. Regarding the issue of bias, paleointensity
34
estimation involves normalizing the observed natural rema-
35
nent magnetization (NRM) to a magnetization acquired in
36
a known laboratory field. The accurate determination of
37
a paleointensity therefore requires that the acquisition of a
38
magnetization be reproducible. However, it has been shown
39
(e.g., (
7
–
10
)) that some rocks have non reproducible magne-
40
tizations, which can lead to biased paleointensity estimates.
41
Global paleointensity records may be confounded by these
42
biased estimates, leading to an apparent non dipole signature.
43
Alternatively, geomagnetic intensity variations through time
44
Significance Statement
Reconstructions of tectonic plates rely on the assumption that
the time-averaged geomagnetic field behaves like a axial geo-
centric dipole. Global compilations of field directions are close
to a GAD, but field strengths are not. Obtaining estimates of
field strength is difficult because of high experimental failure
rates and inconsistent analysis methods between studies. Here
we present new data collected from igneous rocks in Hawai‘i
and use consistent analytical methods to compare them to
published data from Antarctica and Israel. Our results indicate
a persistent non-dipolar component in the Earth’s magnetic
field over the past 1.5 Ma, but more dipolar behavior prior to
that. This is surprising, given our current understanding of the
processes that give rise to the field.
Please provide details of author contributions here.
Please declare any competing interests here.
2To whom correspondence should be addressed. E-mail: bcychucsd.edu
www.pnas.org/cgi/doi/10.1073/pnas.XXXXXXXXXX PNAS | December 1, 2022 | vol. XXX | no. XX | 1–12
DRAFT
Fig. 1.
Violin plots showing latitudinal binned distributions of a) paleointensity and b) age for reported paleointensity results from the MagIC database aged between 50 ka and
5 Ma. In a violin plot, the width of the violin represents the frequency of intensities in that latitude bin, with the widest point in the violin representing the modal value. The
number of data points in each bin are noted above the violins. The yellow stars in a) are the mean paleointensity value at each latitude bin and the solid blue, dashed black and
solid red lines represent the expected mean values for a dipole field with a strength of 40, 60 and 80 ZAm2respectively.
may not be well averaged. The majority of paleointensity
45
determinations are made with volcanic rocks, which record an
46
instantaneous snapshot of the magnetic field at the time they
47
cool. Archeomagnetic data indicate that the Earth’s magnetic
48
field strength can vary strongly over decades to centuries (e.g.,
49
(
11
)), so numerous paleointensity estimates are necessary for a
50
good average. If the field strength varies over long timescales
51
(e.g. millions of years), then comparing the “average” of two
52
studies may not be meaningful if the units sampled are of
53
different ages. And finally, it is also possible that the geomag-
54
netic field is not in fact GAD-like but has long-term non-axial
55
dipole contributions (as suggested by (4,12,13)).56
Paleomagnetists have identified behaviors in a paleointen-
57
sity experiment that deviate from theoretical expectations
58
and may lead to bias and recent studies have made a greater
59
effort to eliminate such biased results. In most paleointensity
60
studies, results from paleomagnetic specimens are excluded
61
from the analysis if they fail a set of “selection criteria” which
62
are phenomenological descriptions of these behaviors. Alterna-
63
tively, the BiCEP method (
14
) attempts to find a relationship
64
between the apparent paleointensity and one of these com-
65
monly used selection criteria (curvature (
15
)), and attempts
66
to correct for the bias induced by the non-ideal behavior,
67
obtaining accurate results without excluding data from the
68
analysis based on arbitrary criteria. Recently, a study (
4
)
69
which used the strict CCRIT criteria (
16
) and the BiCEP
70
method on paleointensity studies from several latitudes found
71
that there is still a discrepancy between these time-averaged
72
paleointensities and those expected for a GAD field, making
73
our first hypothesis (apparent non-dipole behavior is caused
74
by bias in paleointensity estimation) unlikely to be the cause
75
of inaccurate paleointensities.76
Figure 1b shows the age distribution of latitudinally binned
77
absolute paleointensity data in the MagIC database (without
78
selection). It is apparent that different latitude bins have
79
different age distributions. Because of this, the average pale-
80
ointensity from each bin is representative of a different time
81
period, and is not an average paleointensity for the whole of
82
the last 5 Ma. High quality paleointensity data, analyzed in a
83
consistent manner, are needed to determine whether temporal
84
sampling is the cause of apparent non-dipolar behavior, or if
85
the time-averaged field is truly non-dipolar, as outlined in our
86
third hypothesis. 87
In this paper, we present paleointensity estimates from
88
rapidly cooled volcanic material from lava flows, dikes and
89
vent deposits (scoria and spatter cones) aged 0-4 Myr from the
90
Hawai‘ian islands. In Section 1, we describe how we collect
91
samples in the field (1.A), how we conduct paleointensity
92
experiments (1.B) on specimens therefrom, how we analyze
93
our results using the BiCEP method which produces accurate
94
estimates for specimens magnetized in known fields (1.C), and
95
how we obtain ages for our samples using
40
Ar/
39
Ar dating
96
(1.D). In Section 2, we show the results of our paleointensity
97
study in Hawai‘i. Section 2.Bdiscusses how our results suggest
98
that scoria may be a useful lithology for obtaining high quality
99
paleointensity estimates, and are in agreement with estimates
100
from other lithologies. In Section 2.Cwe fit a model to our
101
paleointensity data in an attempt to derive a time average
102
that accounts for uneven temporal sampling. We then apply
103
the same methodology to studies from Northern Israel and
104
Antarctica. This allows us to test whether poor temporal
105
sampling or non dipole behavior is responsible for the weaker
106
paleointensity at high latitudes. Our results indicate that there
107
is a persistent non-dipole component in the Earth’s magnetic
108
field over at least the past 1.5 Myr with older data being much
109
more consistent with a GAD field. 110
1. Methods 111
A. Field Methods.
Our results come from samples collected
112
over three field seasons from outcrops on the Hawai‘ian islands.
113
Samples were collected from the islands of Hawai‘i, Maui,
114
Moloka‘i, and O‘ahu in an attempt to get a representative av-
115
erage paleointensity over the past 4 Myr. This study targeted
116
predominantly glassy and fine grained igneous material from
117
lava flow tops and bottoms, scoria cones and dike margins.
118
Néel theory (
18
) predicts the physics of “uniaxial single do-
119
main” grains which should behave ideally in a paleointensity
120
experiment. Only very small magnetic particle sizes exhibit
121
single domain behavior, and so we sampled rapidly cooled
122
materials most likely to contain these fine grains. 123
2| www.pnas.org/cgi/doi/10.1073/pnas.XXXXXXXXXX Cych et al.
DRAFT
Fig. 2.
Maps showing sampling localities for successful sites used in this study (blue stars). Insets are labeled with the name of each island in capital letters and the name of
the volcano (if applicable) in lowercase. Each map shows samples from a different Volcano/Island. Colors represent ages of units (
17
), with darker colors indicating younger
flows (see colorbar), and dike locations indicated by red lines. Topographic data: U.S. Geological Survey (USGS). 2015. USGS 10-m Digital Elevation Model (DEM): Hawai‘i.
Coastline data: Hawai‘i Statewide GIS Program.
In the field, we collected small unoriented hand samples
124
using a hammer and chisel; this allowed us to obtain smaller
125
pieces of material and was less destructive than obtaining
126
oriented specimens with a drill. Maps of our sampling localities
127
are shown in Figure 2and details regarding location, age and
128
material are given in Table 1.129
B. Laboratory Work.
Each sample was crushed with a mortar
130
and pestle to produce multiple paleomagnetic specimens with
131
masses on the order of 0.1 g. Specimens were weighed and
132
glued into 1 cm wide borosilicate glass tubes using a high
133
temperature, low magnetic moment glue (KaSil). We subjected
134
each specimen to the IZZI-Thellier method (
21
,
22
). This is
135
a step-wise double heating experiment in which the NRM
136
is replaced by a Thermal Remanent Magnetization (TRM)
137
acquired in a known lab field. Under the IZZI protocol, the
138
order of the in-field and zero-field steps alternates at each
139
temperature step. Under ideal conditions, the ratio of the
140
magnetization lost in a zero-field step to the magnetization
141
gained in an in-field step is the ratio of the ancient field (
Banc
)
142
to the laboratory field (
Blab
). For this study, multiple lab
143
fields were used for different specimens, as we observed that
144
the choice of
Blab
affected whether our specimens passed or
145
failed some of our criteria (see Section 1.C). 146
C. Analysis of Data.
To make sure that we have unbiased re-
147
sults, we used two different analysis methods on our data to
148
obtain an estimate of the ancient field. Primarily, we used the
149
BiCEP method (
14
) of estimating paleointensities, but we also
150
looked at results using the CCRIT criteria of (
16
). BiCEP
151
assumes that the magnetization records a single field, and ther-
152
mochemical alteration of the specimen has not occurred. To
153
make certain of this, we used the minimal selection criteria (see
154
(
23
) for definitions and references), DANG<10, DRAT<10. In
155
addition, we use a new parameter, MAD
Coe
<5 which just uses
156
the zero-field first steps. The set of temperature steps on the
157
Cych et al. PNAS | December 1, 2022 | vol. XXX | no. XX | 3
DRAFT
Table 1. Ages and locations for sites from this study that passed CCRIT or BiCEP. Locations for all sites, including those that did not pass
CCRIT or BiCEP are listed in the supporting information. Latitudes and Longitudes are referenced to the WGS84 standard.
Site Island Lithology Lat. (¶N) Lon. (¶E) Age (Ma) ±2‡
HW306 Hawai‘i Vent Deposit 20.04470 -155.73437 0.1900 0.0700
ML001 Moloka‘i Dike 21.13719 -157.15547 2.0700 0.0200
ML012 Moloka‘i Vent Deposit 21.08955 -157.01053 1.6100 0.0300
ML015 Moloka‘i Vent Deposit 21.19876 -157.24734 1.7700 0.0200
MU004 Maui Vent Deposit 20.77605 -156.53433 1.4300 0.0200
MU009 Maui Vent Deposit 20.81885 -156.61782 0.6100 0.0120
MU011 Maui Vent Deposit 20.83016 -156.63110 1.2300 0.0690
MU012 Maui Vent Deposit 20.88931 -156.67484 0.3000 0.0216
MU013 Maui Vent Deposit 20.92685 -156.69633 0.5840 0.0100
MU023 Maui Vent Deposit 20.61085 -156.31100 0.0765 0.0635
MU025 Maui Vent Deposit 20.70692 -156.25424 0.0950 0.0450
MU027 Maui Vent Deposit 20.70551 -156.25857 0.0950 0.0450
MU031 Maui Vent Deposit 20.69669 -156.28040 0.0670 0.0404
MU036 Maui Vent Deposit 20.63397 -156.45102 0.0106 0.0085
MU106 Maui Dike 20.83446 -156.59879 1.4900 0.0500
MU109 Maui Dike 20.83440 -156.59798 1.5500 0.0500
MU111 Maui Dike 20.83471 -156.59808 1.4500 0.0600
MU113 Maui Lava Flow 20.78467 -156.54893 1.1000 0.0600
OA003 O‘ahu Flow 21.29434 -157.81123 2.5500 0.0800
OA008 O‘ahu Flow 21.40440 -158.17461 3.7100 0.0600
OA014 O‘ahu Dike 21.51972 -158.22772 3.4900 0.1700
OA015 O‘ahu Flow 21.46033 -158.21154 3.1000 0.0300
OA019 O‘ahu Flow 21.30938 -157.65783 2.8400 0.0600
OA026 O‘ahu Flow 21.29836 -157.65380 2.7700 0.1300
OA028 O‘ahu Flow 21.29907 -157.65273 2.7200 0.0800
OA030 O‘ahu Vent Deposit 21.27831 -157.79929 0.3800 0.1100
OA100 O‘ahu Vent Deposit 21.28628 -157.79791 0.4800 0.0400
OA101 O‘ahu Vent Deposit 21.28521 -157.79900 0.4800 0.0400
OA104 O‘ahu Flow 21.30080 -157.65320 2.1800 0.3500
OA108 O‘ahu Dike 21.30527 -157.65027 2.2500 0.1700
OA114 O‘ahu Dike 21.41002 -157.76354 2.8700 0.0600
OA116 O‘ahu Dike 21.40308 -158.17264 3.7200 0.0500
OA117 O‘ahu Dike 21.40308 -158.17264 3.7200 0.0500
OA123 O‘ahu Sill Margin? 21.40149 -158.17141 2.5900 0.0900
OA124 O‘ahu Dike 21.40168 -158.16927 3.2500 0.0100
4| www.pnas.org/cgi/doi/10.1073/pnas.XXXXXXXXXX Cych et al.
DRAFT
Fig. 3.
Example of BiCEP being used to obtain a paleointensity for site MU111. a) Arai plot (
19
) for specimen MU111A05, red dots represent steps where the zero-field
measurement was made first, and blue dots represent in-field first steps. Open circles represent temperature steps not used for this analysis. Triangles are pTRM checks and
green curves are BiCEP’s circular fits to the data. b) Zijderveld plot (
20
) showing magnetic direction data. Open symbols are steps where the temperature steps were not used.
Green line is a principal component analysis fit to the directional data. c) Histogram of possible site mean intensities from BiCEP. d) BiCEP fit showing the predicted relationship
(blue lines) between intensity (y axis) and the curvature criterion (˛
k, x axis).
Arai plot which maximize the FRAC criterion while passing
158
the MAD
Coe
, DANG and DRAT criteria. The vast majority of
159
our specimens pass these criteria with ease, and the ones that
160
do not would unambiguously be rejected by almost any other
161
set of criteria. Site results from BiCEP have a 95% credible
162
interval which is equivalent to the full width of the 2
‡
interval
163
from traditional selection criteria methods (e.g., CCRIT). We
164
considered a site level result from BiCEP acceptable if it has a
165
credible interval with a full width less than 40% of the median
166
value, or 16
µ
T, whichever is greater (the original BiCEP A or
167
B criteria of Cych et al. (2021) (
14
) only include the former
168
criterion). This is equivalent to criteria of
±
10%or4
µ
T used
169
for the CCRIT at a site level. An example of BiCEP being
170
used to estimate
Banc
and its uncertainty for a site is shown
171
in Fig. 3.172
D. Age Constraints.
We obtained a range of radiometric ages
173
for our samples that span the past 4 Ma. Rocks from 23 of
174
our successful sites were analyzed at the Argon Geochronology
175
lab at Oregon State University (OSU) for age determination.
176
200-300
µ
m pieces from each sample were prepared by acid
177
leaching in an ultrasonic bath according to the procedure of
178
(
24
). This was followed by irradiation of the samples in the
179
OSU TRIGA CLICIT nuclear reactor. Samples were then
180
incrementally heated using a defocused CO
2
laser, and the
181
isotopic composition of the released argon was measured using
182
an ARGUS-VI multi-collector mass spectrometer. Seventeen
183
of our ages were calculated using argon-argon (Ar-Ar) plateaus.
184
Three ages from sites OA019, OA116 and OA124 were calcu-
185
lated using Mini-Plateau ages. Sites MU011 and MU036 were
186
calculated using inverse isochron ages and site ML001 was cal-
187
culated using a total fusion age. For sites OA030, OA100 and
188
OA101, we used existing potassium-argon (K-Ar) ages, (
25
)
189
and on West Maui, existing K-Ar ages (
26
) were similarly used
190
for sites MU009 and MU013. Mapped scoria cones at sites
191
MU023, MU025 and MU027 have good age constraints over
192
the timescale we are interested in from K-Ar dating and strati-
193
graphic relationships outlined in (
27
). Finally, site OA026 has
194
its age constrained by stratigraphic relationship with our other
195
Ar-Ar dated flows. A full table of ages is given in Table 1.196
2. Results 197
Results are listed in Table 2. We obtain passing results from
198
35 sites (Table 2): 31 passed BiCEP and 21 passed CCRIT.
199
Some of the results that pass CCRIT do not pass BiCEP, but
200
those sites that pass both methods exhibit good agreement
201
between one another. Because BiCEP gives a more objective
202
analysis, and because we obtain more passing results with this
203
Cych et al. PNAS | December 1, 2022 | vol. XXX | no. XX | 5
DRAFT
Fig. 4.
Paleointensity and age estimates from this study using the BiCEP method
from lava flows (purple squares), scoria cones (red circles) and dike margins (pink
diamonds). Error bars represent the 95% credible interval for intensity estimates, and
the 2
‡
interval for age estimates. Grey triangles are other Hawai‘ian results from the
HSDP2 core (
28
), which have a similar distribution over this time period to our results.
Blue envelope represents the 95% credible interval for the AH-RJMCMC model (
29
)
fit to the data (see section C).
method, we use only the results that pass BiCEP for the rest
204
of our analyses.205
We plot our results versus age in Fig. 4. It is apparent
206
that our results support the hypothesis that the more recent
207
field (over the past
≥
1.5 Ma) is considerably higher than that
208
from 1.5-4 Ma (e.g., (
30
)), supporting the hypothesis of a
209
potential long period variation in the field strength (
30
–
32
).
210
It is also worth noting that in Fig. 1, latitudes which have age
211
distributions which skew towards ages older than 1 Ma (e.g.
212
80
¶
S, 60
¶
N, 0
¶
) tend to have averages that agree with a
≥
40
213
ZAm
2
dipole, whereas the majority of latitudes with mostly
214
younger results tend to agree with a 60-70 ZAm
2
dipole
215
moment, so qualitatively our hypothesis that the missing
216
dipole may be caused by temporal sampling seems plausible.
217
However, the data from Antarctica (
3
) span the entire last 4
218
Ma but also have an average field consistent with a 40 ZAm
2
219
axial dipole strength, so temporal sampling alone does not
220
explain all of the deviation from a GAD field.221
The high paleointensity results over the past 1.5 Ma come
222
predominantly from vent deposits (scoria and spatter cones),
223
whereas older results come predominantly from dikes and lava
224
flows. The dikes and lava flows are associated with the early
225
shield building stages of Hawai‘ian volcanoes, whereas the vent
226
deposits are predominantly from the later stages of volcanic
227
construction. The difference in lithology being coupled with
228
a difference in field strength may be concerning, however our
229
young, high field strength results agree well with the average230
paleointensity from lava flows in the HSDP2 core ((
28
,
33
), re-
231
analyzed in (
4
)), shown as grey triangles in Figure 4, although
232
the variance of the HSDP2 data is larger. Additionally, results
233
from several scoria cones yielded much weaker fields, including
234
for two cones on Moloka‘i older than 1.5 Ma. This leads us to
235
believe that our results from scoria are accurate.236
3. Discussion237
A. Pitfalls of selection criteria.
We used the BiCEP method
238
to obtain site level paleointensity estimates, and prefer this
239
over the CCRIT method (and all other sets of selection criteria
240
Site npass/ntot Bmin Banc Bmax Method
HW306 8/8 30.8 36.8 42.9 BiCEP
ML001 7/7 23.2 31.2 39.2 BiCEP
ML012 6/6 28.1 29.0 30.2 BiCEP
ML015 5/5 5.5 12.0 16.7 BiCEP
MU004 11/11 39.3 42.3 45.5 BiCEP
MU009 6/6 31.1 36.6 42.4 BiCEP
MU011 5/9 19.2 26.5 33.8 CCRIT
MU012 6/6 31.8 34.6 37.6 BiCEP
MU013 8/8 14.8 19.2 23.8 BiCEP
MU023 8/8 26.1 31.0 35.6 BiCEP
MU025 7/7 33.9 42.1 50.2 BiCEP
MU027 6/6 19.7 24.7 30.7 CCRIT
MU031 10/10 34.6 40.4 46.0 BiCEP
MU036 9/9 10.4 10.9 11.4 BiCEP
MU106 10/12 22.1 28.8 35.0 BiCEP
MU109 7/7 15.9 18.8 21.9 BiCEP
MU111 6/6 12.1 14.3 16.2 BiCEP
MU113 8/8 38.1 43.7 49.7 BiCEP
OA003 11/11 26.9 29.2 31.3 BiCEP
OA008 4/4 14.9 20.2 26.2 BiCEP
OA014 10/12 10.3 13.0 15.6 BiCEP
OA015 8/8 35.3 39.7 44.5 BiCEP
OA019 15/15 20.5 22.9 25.3 BiCEP
OA026 8/8 12.5 15.0 17.4 BiCEP
OA028 8/8 29.4 33.1 36.8 BiCEP
OA030 16/16 45.6 48.9 52.2 BiCEP
OA100 6/12 50.0 51.0 52.0 CCRIT
OA101 9/9 37.3 43.0 48.3 BiCEP
OA104 3/8 15.8 17.6 19.3 CCRIT
OA108 8/8 13.2 19.5 25.5 BiCEP
OA114 6/6 21.8 25.3 30.2 BiCEP
OA116 8/8 21.7 24.9 28.2 BiCEP
OA117 5/5 19.2 23.7 28.1 BiCEP
OA123 6/8 10.3 13.8 19.0 BiCEP
OA124 7/7 33.8 36.8 40.2 BiCEP
Table 2. Paleointensity results from specimens in this study which
passed BiCEP and CCRIT. npass: Number of passing specimens.
ntot: Total number of specimens. For CCRIT results Bmin and Bmax
represent the bounds of the 2‡interval, and so a full width of 40% or
16 µT is considered to have passed. The method column represents
the preferred paleointensity result (BiCEP) when a site passed both
BiCEP and CCRIT
6| www.pnas.org/cgi/doi/10.1073/pnas.XXXXXXXXXX Cych et al.
DRAFT
in use by various authors) as BiCEP produces many more
241
site level results than CCRIT. Often, BiCEP passed sites
242
where specimens failed the FRAC criterion of CCRIT, which
243
specifies that a large proportion of the total magnetization
244
of the specimen is needed to make a paleointensity estimate.
245
BiCEP accounts for the uncertainty in curvature (and therefore
246
bias), introduced by using only part of a specimen’s Arai plot
247
for a paleointensity estimate. This can be seen in Fig. 3a,
248
where specimen MU111A05 fails CCRIT due to low FRAC,
249
but using a smaller part of the Arai plot translates to only a
250
small increase in the uncertainty in curvature, shown by the
251
green curves fit to the data.252
In addition to the FRAC criterion in CCRIT, we identify
253
cases in which criteria may reject a specimen if it has an ancient
254
field much lower than the lab field. The MAD criterion may
255
be exceeded if the laboratory magnetization acquired in an
256
in-field step is not fully removed during a zero-field step, a
257
consequence of a “high temperature pTRM tail” (
34
). This
258
behavior is very noticeable in IZZI experiments (Fig. 5), as
259
the in-field first steps are more strongly affected by this effect.
260
This leads to a zig-zag appearance in the Zijderveld plot. The
261
sizes of these tails are dependent on both the magnitude of the
262
lab field, and the effect the tails have on MAD is dependent on
263
the angle between lab and ancient field. If we call this angle
264
◊
, then the perpendicular part of the tails will be controlled
265
by
Blab sin ◊
. If we assume no other sources of deflection to
266
the MAD angle, the equation for the effect is:267
tan(MAD)ÃBlab
Banc
sin ◊.[1]268
This equation demonstrates that in the same lab field, sites
269
with low ancient fields will be preferentially rejected with
270
higher MAD, and sites with high ancient fields will be prefer-
271
entially accepted.272
To counteract the lab field-dependent effects, we used 10,
273
30 and 70
µ
T fields in our studies, which captures the range
274
of the ancient field. At some sites with low estimated
Banc
,
275
there was an observably higher pass rate in lower fields. An
276
example of this for site OA014 is illustrated in Fig. 5. To treat
277
specimens magnetized in different fields fairly, it is tempting
278
to come up with a criterion for MAD which is dependent on
279
Equation 1. However, effects that we may be using MAD
280
to look for (e.g. two component magnetizations) will not
281
be dependent on the lab field, and so we suggest calculating
282
MAD for exclusively the zero-field first or “Coe” type steps
283
(
35
). Although pTRM tails may still be present in these steps,
284
they will be significantly reduced in in-field first steps. We
285
call a MAD calculated using these steps MAD
Coe
and how
286
it compares to MAD for site OA014 is shown in Fig. 5d.
287
This significantly reduces the lab field-dependent effects, but
288
does not eliminate them entirely. Because pTRMs scale with
289
the lab field used, there may be other unrecognized pTRM
290
dependent effects. We recommend using a range of lab fields in
291
paleointensity studies as the most robust way of compensating
292
for these effects.293
B. Sample Characterization.
We have demonstrated our abil-
294
ity to obtain high quality paleointensity results from our sam-
295
ples using the BiCEP method. However, it is not clear what
296
the primary carriers of the magnetization are for these samples,
297
particularly for samples from vent deposits, which are rela-
298
tively unstudied in the paleointensity literature. To attempt
299
Fig. 5.
a)-c) Zijderveld plots of specimens from site OA014, showing zig-zagging
behavior that progressively increases with lab field and d) Scatter plot showing the
relationship between the MAD criterion, and the magnitude and angle of the lab field
for all ten fully demagnetized specimens from this site. Paleointensity experiments
were performed laboratory fields of a) a 10
µ
T, b) 30
µ
T and c) 70
µ
T. d) MAD (green
circles) angle against the strength of the component of the lab field perpendicular to
the ancient field direction (calculated by the PCA of the zero-field first steps). Orange
triangles are the MAD of the zero-field first steps only (MAD
Coe
). Horizontal dashed
line represents the selection criterion (5) used in this study. Using MAD
Coe
improves,
though does not completely eliminate, the lab-field dependence of MAD. All MADs
were calculated using temperature steps above 400-600
¶
C to avoid any potential
viscous remanent magnetization (VRM).
Cych et al. PNAS | December 1, 2022 | vol. XXX | no. XX | 7
DRAFT
Ha (T)
1.750.00 0.25 0.50 0.75 1.00 1.25 1.50
0.0
0.2
0.4
0.6
0.8
1.0
NRM / NRM0
0
100
200
300
350
400
425
450
475
500
510
520
530
540
5
5
0
-0.3
-0.2
-0.1
-0.0
0.1
0.2
0.3
-0.3 -0.2 -0.1 0
0.1 0.2 0.3
-0.3 -0.2 -0.1 0
0.1 0.2 0.3
-0.3 -0.2 -0.1 0
0.1 0.2 0.3
a) b) c) d)
Ha (T)
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
-0.1 -0.05 0
0.05 0.1
-0.1 -0.05 0
0.05 0.1
-0.1 -0.05 0
0.05 0.1
300
350
0.0 0.1 0.2 0.3 0.4 0.5
0.0
0.2
0.4
0.6
0.8
1.0
NRM / NRM0
540
510
500
475
450
425
400
e) f) g) h)
-0.2 -0.1 0
0.1 0.2
Ha (T)
-0.1
-0.2
0
0.1
0.2
-0.2 -0.1 0
0.1 0.2
Hb (T) Hb (T)
-0.2 -0.1 0
0.1 0.2
Hb (T)
i) j) k)
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75
0.0
0.2
0.4
0.6
0.8
1.0 0100
200
300
350
400
425450
475 500
510 520
530 540
550
560
570
58
0
59
0
600
610
620
T
R
M
/
N
R
M
0
NRM / NRM0
l)
Fig. 6.
First Order Reversal Curves (FORCs) a),e),i), iFORCs b),f),j), and tForcs c),g),j) calculated using the xFORC protocol (
36
). All FORCs calculated using a smoothing
factor of 2 and a non-linear color scale of 1, except for iFORCs which were calculated using a smoothing factor of 3 and a non-linear color scale of 10. Arai plots are plotted in
d),h),k). FORCs use sister specimens from two sites that yielded passing results: OA030 (top row), OA014 (center row) and a site which did not pass CCRIT or BiCEP, HW305
(bottom row). Sites which yielded specimens with linear Arai plots tend to have an elongated central ridge and have 3 lobes in the iFORC (top and center rows), whereas sites
with curved Arai plots tend to have more spread along the Ha=Hbdirection and have extremely noisy iFORCs with little information.
to characterize the domain state of our samples, we obtained
300
First Order Reversal Curves (FORCs, (
37
)) using the xFORC
301
protocol of (
36
) on selected material from sites which passed
302
BiCEP (and from some which failed). For this analysis we
303
used sister specimens from the same samples for which the
304
paleointensity results were acquired. FORCs are a qualitative
305
way of assessing the domain state of a specimen using its hys-
306
teresis properties. Specimens which contain “Single-Domain”
307
(SD) grains which are ideal for the paleointensity experiment
308
will have FORCs with a central ridge of positive values along
309
the
Ha
=-
Hb
axis (see e.g. Fig. 6a). Specimens with higher
310
numbers of non SD grains will have FORCs which have a
311
spread along the
Ha
=
Hb
axis. The iFORC which represents
312
the induced part of the magnetization displays a pattern of
313
three distinct “lobes” (e.g. Fig 6b,e) for a sample containing
314
SD grains, whereas it may display four “lobes” or be extremely
315
noisy for samples containing non-SD grains. The tFORC rep-
316
resents “transient hysteresis” which occurs in non-SD grains;
317
specimens with just noise on the tFORC (e.g. Fig. 6c) are
318
most likely to be single domain.319
Examples of FORCs and Arai plots for different samples
320
are displayed in Fig. 6. The FORC interpretations generally
321
agree with the paleointensity experimental results. FORCs
322
obtained from dike samples have pronounced central ridges and
323
three lobes in the iFORC if visible, and effectively no tFORC
324
(Fig. 6a-d). These samples generally had Arai plots which
325
were straight lines, but sometimes underwent thermochemical
326
alteration at high temperatures. Samples from lava flows
327
and vent deposits had central ridges, with small amounts of
328
transient hysteresis and spreading along the
Ha
=
Hb
axis.
329
These samples still have linear Arai plots, and often have three
330
lobes present in the iFORC, which suggests that the majority
331
of carriers in these specimens are single domain (see Fig. 6e-
332
h). An example from a relatively coarse grained lava flow is
333
given in Fig. 6i-l. Samples like these had highly curved or
334
zig-zagging Arai plots (Fig. 6l) generally had no central ridge
335
and lots of spreading along the
Ha
=
Hb
axis (Fig. 6i). These
336
samples had pronounced tFORCs(Fig. 6k) , and only noise
337
in the iFORCs away from the Haaxis (Fig. 6j), observations 338
which are consistent with the curved and zig-zagging Arai
339
plots. 340
We also obtained Back Scattered Electron (BSE) images
341
using an Scanning Electron Microscope (SEM), and Electron
342
Dispersive X-Ray Spectroscopy (EDS) element maps to iden-
343
8| www.pnas.org/cgi/doi/10.1073/pnas.XXXXXXXXXX Cych et al.
DRAFT
c)
e)
MChr
Ol
Cpx
Plg
f)
a)
b)
6.5 μm
62.4 μm
h
)
Fe-Mg-Ti Oxides
Plg
Cpx
759 μm
134 μm
d)
g)
Fig. 7.
Back Scattered Electron (BSE) images and Electron Dispersive X-Ray Spec-
troscopy (EDS) maps of sister specimens from selected samples used in this study.
Red text gives vertical field of view (FOV) for each image. a) BSE image of sample
ML015A, a scoria vent deposit. b) Zoomed in image of large oxide in a), showing
Fe-Ti exsolution textures. c) Zoomed in image of small oxide in a), showing elongate
skeletal/cruciform structure. d) EDS element map of a typical oxide from another sco-
ria vent deposit, ML012A, showing heterogeneous composition in the Iron-Titanium
oxides. The atomic content of Si is shown in yellow, Fe in red, and Ti in pink. e)
BSE image of sample OA030A, an agglutinated basanite vent deposit. f) Close up
of high temperature alteration texture in olivine phenocryst. g) The same texture
present in sample MU012A, a breccia from the bottom of a basanite lava flow. h)
Close up of this texture with EDS element map. Colors are the same as d), with
purple representing Mg. Note that the light colors in the BSE image represent an
iron rich phase (interpreted as magnetite), which is surrounded by a phase richer in
silicon than the surrounding olivine, interpreted as enstatite. Dominant mineral phases
written on a) and e): Plg: plagioclase feldspar, Cpx: clinopyroxene, Ol: olivine, MChr:
chrome spinel. Horizontal banding present in b),c),d),f),h) is an artifact of charging
the sample that occurs in the SEM’s EDS element mapping mode.
tify iron oxides in several thin sections taken from our samples.
344
Several pictures from these analyses are displayed Fig. 7. Dike
345
samples we analyzed contained no visible iron oxides in the
346
glass, and almost no iron oxides in the groundmass. This is
347
consistent with our FORCs and Arai plots (Fig. 6a-d), which
348
are indicative of this specimen containing a predominance of
349
single domain grains, which are 10s of nm in scale and not
350
resolvable by the SEM used in this analysis. By contrast, sam-
351
ples from vent deposits contained numerous micron-scale iron
352
bearing oxides in the groundmass, and in some cases, larger
353
iron oxides on the scales of 100s of microns (Fig. 7a-d), size
354
ranges where we would expect the grains to yield curved Arai
355
plots. Many of these grains have elongated “cruciform” tex-
356
tures (Fig. 7c) or have heterogeneous compositions (Fig. 7a,d).
357
One possibility is that these textures may persist to smaller
358
scales, causing the larger grains to behave like assemblages
359
of smaller, single domain, grains, due to their elongation or
360
having smaller magnetic subregions separated by nonmagnetic
361
lamellae. Another possibility is that these large grains do not
362
contribute to the remanence. However, the lava flows and
363
vent deposits have much higher NRM moments than the dikes,
364
with mass normalized NRMs on the order of 10
≠2
to 10
≠3365
Am
2
/kg, as opposed to the dikes which have moments on the
366
order of 10≠4to 10≠5Am2/kg. 367
Two thin sections from sites MU012 and OA030 have nu-
368
merous olivine grains which exhibit an unusual texture, as
369
displayed in Fig. 7e-h. This texture has been observed previ-
370
ously (
38
,
39
) and is interpreted as being caused by oxidation
371
of olivine at temperatures above 800
¶
C, which causes break-
372
down into an iron oxide (magnetite or hematite depending on
373
formation conditions) and enstatite (see Fig. 7h and figure
374
caption). The temperature of the oxidation means that the
375
samples were oxidized prior to gaining a magnetization, which
376
means the NRM is a primary TRM acquired during cooling.
377
Oxidation of this kind seems to typically occur in fire foun-
378
taining strombolian type eruptions e.g. (
40
) where the lavas
379
remain at high temperatures in an oxidizing environment for
380
a while (e.g. 950
¶
C for 24-48 hours as per (
41
)). OA030 is an
381
agglutinated basanitic vent deposit, agreeing with this oxida-
382
tive environment, whereas the MU012 sample was taken from
383
breccia/clinkers in an a‘¯a lava flow, which may also undergo
384
high temperature oxidation although the source is less clear. 385
Both sites with evidence for high temperature oxidation of
386
olivines had highly linear Arai plots (see Figure 6h), with 16/16
387
specimens passing the strict CCRIT criteria for OA030, and
388
6/6 passing for MU012. Additionally a sample from OA030 has
389
a FORC indicative of single-domain to single-vortex domain
390
state, with a central ridge and three lobes in the iFORC (see
391
Fig. 6, middle row). This indicates that the oxides formed
392
by this breakdown may have extremely desirable properties
393
for paleointensity experiments. Similar to the smaller oxides
394
found in our other vent deposits (Fig. 7c), the elongation
395
and finger-like structures present in these oxides could also
396
explain their ideal behavior in the paleointensity experiment.
397
These thin sections also contained numerous micron scale iron-
398
titanium-magnesium oxides (interpreted as magnesioferrite) in
399
the groundmass and around the outside of the olivine grains
400
(Fig. 7e), but because the majority of the remanence unblocks
401
between 400 and 600
¶
C(seeFig.6d), we believe that magnetite
402
is the dominant remanence carrier in these specimens. 403
Despite the large iron oxides observed in vent deposits and
404
Cych et al. PNAS | December 1, 2022 | vol. XXX | no. XX | 9
DRAFT
Fig. 8.
a) - c) Plots of VADM against age (symbols), and 95% credible envelopes for AH-RJMCMC models (
29
) (shaded areas) for studies from a) Antarctica (purple plus
symbols), b) Hawai‘i (green dots), and c) Israel (orange triangles). Horizontal dashed lines are the average VADM of all paleointensity estimates (symbols) for each plot. In b),
all unfiltered data in the MagIC database from Hawai‘i aged between 50 ka and 3.8 Ma are plotted as grey diamonds, and the average VADM from these data are plotted as a
grey horizontal line. d) Violin plots showing the distribution of averaged VADMs over different time periods, numbers refer to the number of paleointensity within these temporal
ranges, although data outside these ranges may also contribute to these averages. Data from Hawai‘i have a significantly higher average VADM than in Israel and Antarctica
over the past 1.5 Ma, which is reflected in the averages from 0-2.5 Ma. Average VADMs for data older than 1.5 Ma appears to agree for all three locations.
lava flows from this study, we conclude that these lithologies
405
provide a good source for paleointensity estimates, as they
406
have a high success rate relative to our other lithologies owing
407
to their strikingly linear Arai plots (see Fig. 6, top row). Site
408
MU113 provides further evidence for this, as material sampled
409
from the inside of a lava tube gave an identical result to mate-
410
rial sampled from a scoriaceous bomb entrained in the same
411
flow. There are other reasons to favour these types of litholo-
412
gies: The formation of these samples in an oxic environment
413
at high temperature may help prevent thermochemical alter-
414
ation during the paleointensity experiment, and fresh scoria
415
is also easy to come by in Hawai‘i, as many scoria cones are
416
quarried. However, most preserved vent deposits are typically
417
formed during the later stages of Hawai‘ian volcanism, and
418
consequently we have no results from scoria older than 2 Ma.
419
C. Temporal Distributions of Intensity.
Mismatch between the
420
observed distribution of paleointensities with latitude and the
421
expected distribution for a GAD field Fig 1a) could poten-
422
tially be caused by inconsistencies in treatment of data among
423
different paleointensity studies. To compare the time-averaged
424
field from our model to data from different latitudes, we rean-
425
alyzed results from recent paleomagnetic studies in Northern
426
Israel (
4
) and Antarctica (
3
) using the BiCEP method and
427
the same criteria used for the Hawai‘i samples. Tables of
428
results from these re-analyses can be found in the Supporting
429
Information. Each of these studies yielded passing sites with
430
results spanning the past 2.5 Myr. For direct comparisons
431
between locations, we convert each paleointensity result to a
432
Virtual Axial Dipole Moment (VADM) which is the moment
433
of the geocentric dipole (measured in ZAm
2
) that would yield
434
the observed paleointensity at a given latitude. Our average
435
VADM for Hawai‘i is 62.4 ZAm
2
, which is similar to the 64.2
436
ZAm
2
value from Israel, but is significantly higher than the
437
average in Antarctica (39.6 ZAm
2
). Plots of VADMs with age
438
for each location are shown in Figures 8a)-c), with average
439
VADMs plotted as horizontal dashed lines. In Figure 8bwe
440
also plot all the data from Hawai‘i in the MagIC database
441
from this time interval in grey. The unfiltered data have a
442
significantly higher variance than our data, and the weaker
443
field seen prior to 1.5 Ma in our data is not apparent in the
444
unfiltered Hawai‘ian data, which have an average VADM of
445
77.2 ZAm
2
. These differences could occur because more field
446
variation is being captured by the larger dataset, or because
447
the unfiltered data have more variance due to inconsistency
448
in their analysis (for example, preferentially taking the low
449
10 | www.pnas.org/cgi/doi/10.1073/pnas.XXXXXXXXXX Cych et al.
DRAFT
temperature steps in a potentially sagging Arai plot). Despite
450
the consistency in analysis of our data, the average VADM
451
in Hawai‘i and Israel is still very different to that found in
452
Antarctica, indicating that inconsistency in analyses and bi-
453
ased paleointensities caused by Arai plot curvature are not the
454
source of this mismatch.455
Taking an average VADM of the entire age range of our data
456
may not be representative of the time-averaged field, because
457
our data have different temporal distributions, with no data
458
in Israel older than 2.75 Ma. In Hawai‘i, this average does not
459
capture the change in average field strength seen at 1.5 Ma,
460
and in Israel, we have many paleointensity data which record
461
a strong field and come from a small range of time around 850
462
ka B.P. Because this time interval is oversampled, it will bias
463
our average VADM towards these higher values. To account
464
for these problems, it would make sense to fit a curve to our
465
VADMs and take an average of the curve over an interval of
466
interest. We do this using the “Age Hyperparameter Reversible
467
Jump Markov Chain Monte Carlo” (AH-RJMCMC) method
468
(
29
). This model fits piecewise linear curves to paleointensity
469
data in a probabilistic fashion, with curves with less linear
470
pieces being preferable for the model. At times when there
471
are few data, the model uncertainties become very large and
472
revert to a uniform prior distribution, which we set as 0-220
473
ZAm
2
. At times where we have no data, the uncertainty in
474
the average VADM will increase, and so any differences in the
475
average VADM using this method are driven by the data.476
We computed the AH-RJMCMC models, which output a
477
series of possible piecewise linear curves at each locality. We
478
took the average value of each curve over the past 2.5 Ma, and
479
converted these averages to VADMs. The models produced
480
by this analysis are shown in Fig. 8a-c, and the distributions
481
of the time-averaged VADMs for each locality are plotted on
482
the violin plots in Figure 8d. Using this methodology, it is
483
apparent that the time-averaged VADMs over the last 1.5 Ma
484
from Hawai‘i and Antarctica are indeed not consistent with
485
each other, but the time-averaged VADM in Israel could be
486
compatible with either of the other latitudes. However, there
487
is not enough evidence to confirm a difference in the tempo-
488
ral average between Hawai‘i and Antarctica from 1.5-2.5 Ma,
489
with the average VADMs appearing consistent. This implies
490
that poor temporal sampling is not the reason for inconsistent
491
paleointensities at different latitudes, but that some form of
492
genuine non-dipolar field behavior that causes higher fields
493
in Hawai‘i than Antarctica at least since 1.5 Ma. More pa-
494
leointensity studies with high quality paleointensity data at
495
different latitudes (especially from the southern hemisphere)
496
are needed to better understand the sources of this non-dipolar
497
behavior.498
Conclusions499
In this paper, we obtained 31 high quality paleointensity results
500
from dikes, lava flow tops and vent deposits collected in the
501
Hawai‘ian islands, with ages ranging from 0-4 Ma. We demon-
502
strate a methodology for obtaining accurate time-averaged
503
paleointensities, with uncertainties which allow direct com-
504
parison between paleointensity studies at different latitudes.
505
The use of BiCEP allows for consistent comparison of results
506
between different studies, and using the methodology of Liver-
507
more et al. (2018) (
29
) allows us to obtain a time-averaged
508
intensity, with uncertainty, which accounts for the tempo-
509
ral distribution of our paleointensity. Because these robust
510
statistical approaches are used for calculating time-averaged
511
paleointensities, we are able to exclude the hypotheses that
512
inconsistency of our time-averaged VADMs is due to either
513
biased paleointensity data, or inconsistent temporal sampling
514
of paleointensities. 515
Applying the new methodology to data from the Hawai‘ian
516
islands, we find that the time-averaged paleointensity in
517
Hawai‘i over the past 1.5 Ma was higher than during the
518
period from 1.5-4 Ma. Comparing results from paleointensity
519
studies at three latitudes, we find that this period of high
520
paleointensity is not recorded in rocks from Antarctica or Is-
521
rael. We reiterate the conclusion of other recent papers (e.g.
522
(
4
)) that the Earth’s magnetic field averaged over the past 1.5
523
Ma does not conform to a Geocentric Axial Dipole. Further
524
time averages at a greater range of latitudes and times will
525
be needed to obtain better estimates of the structure of this
526
time-averaged field. 527
Our results also indicate that vent deposits containing scoria
528
and olivine bearing rocks which are oxidized at high tempera-
529
tures are potentially good lithologies for obtaining high quality
530
paleointensity estimates, with higher success rates in the pale-
531
ointensity experiment. Specimens from these lithologies have
532
strong magnetizations and tend to alter less in paleointensity
533
experiments. Additionally, these deposits are frequently quar-
534
ried, allowing for easy access to fresh material in the field.
535
Despite their useful properties in paleointensity experiments,
536
and their single-domain like FORCs, the size of iron oxides in
537
these samples when viewed under a microscope is orders of
538
magnitude larger than would be expected for single domain
539
grains. Further study of the magnetic carriers in these samples
540
should be undertaken to understand why they have such ideal
541
rock magnetic properties. 542
References. 543
1. G Cromwell, CL Johnson, L Tauxe, CG Constable, NA Jarboe, PSV10: A Global Data Set 544
for 0–10 Ma Time-Averaged Field and Paleosecular Variation Studies. Geochem. Geophys. 545
Geosyst. 19, 1533–1558 (2018). 546
2. KP Lawrence, et al., Paleomagnetic field properties near the southern hemisphere tangent 547
cylinder. Geochem. Geophys. Geosyst. 10, Q01005 (2009). 548
3. H Asefaw, L Tauxe, AAP Koppers, H Staudigel, Four-Dimensional Paleomagnetic Dataset: 549
Plio-Pleistocene Paleodirection and Paleointensity Results From the Erebus Volcanic 550
Province, Antarctica. J. Geophys. Res. Solid Earth 126, e2020JB020834 (2021). 551
4. L Tauxe, H Asefaw, N Behar, AAP Koppers, R Shaar, Paleointensity Estimates from the 552
Pleistocene of Northern Israel: Implications for hemispheric asymmetr y in the time-averaged 553
field. Geochem. Geophys. Geosyst. n/a, e2022GC010473 (2022). 554
5. AR Muxworthy, Considerations for Latitudinal Time-Averaged-Field Palaeointensity Analysis 555
of the Last Five Million Years. Front. Earth Sci. 0(2017). 556
6. VP Shcherbakov, AV Khokhlov, NK Sycheva, Analysis of the Hypothesis of a Giant Gaussian 557
Process as a Means for Describing Secular Variations of the Geomagnetic Field Vector. Izv. 558
Phys. Solid Earth 55, 182–194 (2019). 559
7. S Levi, The effect of magnetite particle size on paleointensity determinations of the geomag- 560
netic field. Phys. Earth Planet. Inter. 13, 245–259 (1977). 561
8. D Dunlop, O Özdemir, Beyond Néel’s theories: thermal demagnetization of narrow-band 562
partial thermoremanent magnetization. Phys. Earth Planet. Int. 126, 43–57 (2001). 563
9. D Krása, C Heunemann, R Leonhardt, N Petersen, Experimental procedure to detect mul- 564
tidomain remanence during thellier–thellier experiments. Phys. Chem Earth (A/B/C) 28, 681 565
– 687 (2003) Paleo, Rock and Environmental Magnetism 2002. 566
10. L Tauxe, et al., Understanding Nonideal Paleointensity Recording in Igneous Rocks: Insights 567
From Aging Experiments on Lava Samples and the Causes and Consequences of “Fragile” 568
Curvature in Arai Plots. Geochem. Geophys. Geosyst. 22, e2020GC009423 (2021). 569
11. R Shaar, et al., Synchronizing archaeomagnetic field intensity records in the levant be- 570
tween the 23rd and 15th cennturies bce: chronological and methodological implications. 571
Geochem. Geophys. Geosystems 21, e2020GC009251 (2020). 572
12. RL Wilson, Permanent Aspects of the Earth’s Non-dipole Magnetic Field over Upper Tertiary 573
Times. Geophys. J. Int. 19, 417–437 (1970). 574
13. G Cromwell, et al., In search of long-term hemispheric asymmetry in the geomagnetic field: 575
Results from high northern latitudes. Geochem. Geophys. Geosyst. 14, 3234–3249 (2013). 576
14. B Cych, M Morzfeld, L Tauxe, Bias Corrected Estimation of Paleointensity (BiCEP): An Im- 577
proved Methodology for Obtaining Paleointensity Estimates. Geochem. Geophys. Geosyst. 578
22, e2021GC009755 (2021). 579
Cych et al. PNAS | December 1, 2022 | vol. XXX | no. XX | 11
DRAFT
15. GA Paterson, A simple test for the presence of multidomain behavior during paleointensity580
experiments. J. Geophys. Res. Solid Earth 116 (2011).581
16. G Cromwell, L Tauxe, H Staudigel, H Ron, Paleointensity estimates from historic and modern582
Hawaiian lava flows using glassy basalt as a primary source material. Phys. Earth Planet.583
Inter. 241, 44–56 (2015).584
17. DR Sherrod, JM Sinton, SE Watkins, KM Brunt, Geologic map of the state of hawaii. US585
geological survey open-file report 1089, 83 (2007).586
18. L Néel, Théorie du traînage magnétique des ferromagnétiques en grains fins avec applica-587
tions aux terres cuites. Ann. géophys. 5, 99–136 (1949).588
19. T Nagata, Y Arai, K Momose, Secular variation of the geomagnetic total force during the last589
5000 years. J. Geophys. Res. 68, 5277–5282 (1963).590
20. JDA Zijderveld, A.C. demagnetization of rocks: Analysis of results, Methods in Paleomag-591
netism. (Chapman and Hall), (1967).592
21. L Tauxe, H Staudigel, Strength of the geomagnetic field in the Cretaceous Normal Super-593
chron: New data from submarine basaltic glass of the Troodos Ophiolite. Geochem. Geophys.594
Geosyst. 5(2004).595
22. Y Yu, L Tauxe, A Genevey, Toward an optimal geomagnetic field intensity determination tech-596
nique. Geochem. Geophys. Geosyst. 5(2004).597
23. G Paterson, L Tauxe, A Biggin, R Shaar, L Jonestrask, On improving the selection of thellier-598
type paleointensity data. Geochem. Geophys. Geosystems 15 (2014).599
24. AAP Koppers, H Staudigel, JR Wijbrans, Dating crystalline groundmass separates of altered600
Cretaceous seamount basalts by the 40Ar/39Ar incremental heating technique. Chem. Geol.601
166, 139–158 (2000).602
25. A Ozawa, T Tagami, MO Garcia, Unspiked K–Ar dating of the Honolulu rejuvenated and603
Ko‘olau shield volcanism on O‘ahu, Hawai‘i. Earth Planet. Sci. Lett. 232, 1–11 (2005).604
26. T Tagami, Y Nishimitsu, DR Sherrod, Rejuvenated-stage volcanism after 0.6-m.y. quiescence605
at West Maui volcano, Hawaii: new evidence from K–Ar ages and chemistry of Lahaina606
Volcanics. J. Volcanol. Geotherm. Res. 120, 207–214 (2003).607
27. DR Sherrod, Y Nishimitsu, T Tagami, New K-Ar ages and the geologic evidence against608
rejuvenated-stage volcanism at Haleakala, East Maui, a postshield-stage volcano of the609
Hawaiian island chain. GSA Bull.115, 683–694 (2003).610
28. S Cai, L Tauxe, G Cromwell, Paleointensity From Subaerial Basaltic Glasses From the Sec-611
ond Hawaii Scientific Drilling Project (HSDP2) Core and Implications for PossibleBias in Data612
From Lava Flow Interiors. J. Geophys. Res. Solid Earth 122, 8664–8674 (2017).613
29. PW Livermore, A Fournier, Y Gallet, T Bodin, Transdimensional inference of archeomagnetic614
intensity change. Geophys. J. Int. 215, 2008–2034 (2018).615
30. L Tauxe, Long-term trends in paleointensity: The contribution of dsdp/odp submarine basaltic616
glass collections. Phys. Earth Planet. Interiors 156, 223–241 (2006).617
31. P Selkin, L Tauxe, Long-term variations in paleointensity. Phil. Trans. Roy. Soc. Lond. 358,618
1065–1088 (2000).619
32. L Ziegler, C Constable, CL Johnson, L Tauxe, Padm2m: a penalized maximum likelihood620
model of the 0-2 ma paeomagneitc axial dipole moment. Geophys. J. Int. 184, 1069–1089621
(2011).622
33. L Tauxe, JJ Love, Paleointensity in Hawaiian Scientific Drilling Project Hole (HSDP2): Results623
from submarine basaltic glass. Geochem. Geophys. Geosyst. 4(2003).624
34. D Dunlop, O Özdemir, Effect of grain size and domain state on thermal demagnetization tails.625
Geophys. Res. Lett. 27, 1311–1314 (2000).626
35. RS Coe, The determination of paleo-intensities of the earth’s magnetic field with emphasis627
on mechanisms which could cause non-ideal behavior in thellier’s method. J. Geomag. Geo-628
electr. 19, 157–178 (1967).629
36. X Zhao, et al., Magnetic domain state diagnosis using hysteresis reversal curves. J. Geophys.630
Res. Solid Earth 122, 4767–4789 (2017).631
37. C Pike, A Roberts, K Verosub, Characterizing interactions in fine magnetic particle systems632
using first order reversal curves. J. Appl. Phys. 85, 6660–6667 (1999).633
38. T Ejima, et al., Precipitates within olivine phenocrysts in oxidized andesitic scoria from634
Kasayama volcano, Hagi, Japan. J. Miner. Petrol. Sci. 112, 116–126 (2017).635
39. MS Blondes, MT Brandon, PW Reiners, FZ Page, NT Kita, Generation of Forsteritic Olivine636
(Fo99·8) by Subsolidus Oxidation in Basaltic Flows. J. Petrol. 53, 971–984 (2012).637
40. S Del Moro, A Renzulli, P Landi, S La Felice, M Rosi, Unusual lapilli tuff ejecta erupted at638
Stromboli during the 15 March 2007 explosion shed light on the nature and thermal state639
of rocks forming the crater system of the volcano. J. Volcanol. Geotherm. Res. 254, 37–52640
(2013).641
41. SE Haggerty, I Baker, The alteration of olivine in basaltic and associated lavas. Contrib. Miner.642
Petrol. 16, 233–257 (1967).643
Data Archival.
All data and interpretations are avail-
644
able at https://earthref.org/MagIC/19614/9208acad-0f62-
645
4d9e-b265-4c8907d40eb7 and will be made available in the
646
MagIC database at earthref.org/MagIC/19614 on acceptance
647
of this paper. Python notebooks for producing figures can be
648
found at https://github.com/bcych/hawaiian_paleointensity .
649
ACKNOWLEDGMENTS.
This work was partially supported by
650
EAR1827263 to LT and EAR1520788 to GC. We would like to thank
651
the Hawai‘i Department of Land and Natural Resources’ Forestry
652
and Wildlife Program for issuing sampling permits in West Maui,
653
and Moloka‘i Ranch Ltd. and Moloka‘i Land Trust for allowing us
654
to sample on their land. We are grateful for comments from Cathy
655
Constable and Jeffery Gee which improved the manuscript. Finally
656
we would like to thank the late Jasper Konter for his help in the
657
field and hospitality during our field work on Oahu. He will be
658
missed. 659
12 | www.pnas.org/cgi/doi/10.1073/pnas.XXXXXXXXXX Cych et al.
1
Supporting Information for2
Changes in non-dipolar field structure over the Plio-Pleistocene: New paleointensity results3
from Hawai‘i compared to global datasets4
Brendan Cycha, Lisa Tauxea, Geoff Cromwellb, John Sintonc, and Anthony A.P. Koppersd1
5
aUniversity of California, San Diego; bOccidental College; cUniversity of Hawai‘i at Man¯
oa; dOregon State University6
Brendan Cych. E-mail: bcych@ucsd.edu7
This PDF file includes:8
Fig. S19
Table S110
SI References11
Brendan Cycha, Lisa Tauxea, Geoff Cromwellb, John Sintonc, and Anthony A.P. Koppersd1 of 4
Fig. S1.
Plots of Age against cumulative Argon released for age plateau and mini-plateau (OA019, OA116, OA124) age experiments. The mini-plateau ages for OA124
are concordant with total fusion ages from other samples from same site, and the mini-plateau age for OA019 is close to the age for OA028, a nearby lava flow which it is
stratgraphically below.
2 of 4Brendan Cycha, Lisa Tauxea, Geoff Cromwellb, John Sintonc, and Anthony A.P. Koppersd
Table S1. Results analyzed using BiCEP from Antarctica (1) (sites with the prefix ’mc’) and Northern Israel (2) (sites with the prefix ’GHI’)
Site Latitude Longitude Age (Ma) Age 2‡Bmin Bmedian Bmax
mc1001 -77.850000 166.640000 1.1800 0.0100 12.7 18.6 24.9
mc1002 -77.850000 166.690000 0.3300 0.0200 22.3 29.1 36.3
mc1009 -77.550000 166.200000 0.0740 0.0150 23.9 28.0 35.1
mc1015 -77.470000 169.230000 1.3300 0.0200 22.7 26.7 30.6
mc1020 -77.880000 165.020000 0.7700 0.0320 55.6 62.3 69.3
mc1029 -78.310000 164.800000 0.1800 0.0800 41.7 44.7 48.0
mc1031 -78.350000 164.300000 0.1330 0.0117 23.5 31.2 39.0
mc1032 -78.360000 164.300000 0.0078 0.0120 27.9 31.2 34.9
mc1036 -78.390000 164.270000 0.1200 0.0400 22.2 28.8 35.3
mc1103 -78.240000 163.360000 1.4210 0.0300 14.5 20.1 27.7
mc1109 -78.280000 163.540000 1.2610 0.0400 29.5 33.2 37.0
mc1111 -78.220000 162.790000 1.9900 0.0400 18.5 21.4 24.6
mc1115 -78.240000 162.960000 2.4600 0.3100 26.7 31.2 35.8
mc1119 -78.240000 162.960000 1.0800 0.2200 34.6 37.7 40.9
mc1120 -78.240000 163.090000 1.7560 0.0500 22.4 24.7 27.0
mc1121 -78.240000 162.950000 2.5050 0.0600 28.8 32.8 35.7
mc1127 -78.250000 163.730000 1.9420 0.0680 33.7 37.0 40.4
mc1135 -78.230000 166.560000 3.6000 0.0100 29.4 31.7 33.9
mc1139 -78.260000 163.080000 0.8820 0.0800 24.6 27.8 31.0
mc1140 -78.280000 163.000000 2.0430 0.0900 28.4 34.2 39.1
mc1145 -78.240000 162.893000 1.9000 0.1200 3.3 7.0 10.2
mc1147 -78.200000 162.960000 1.6300 0.3200 16.5 22.4 27.2
mc1155 -77.700000 162.250000 1.5000 0.0500 23.3 30.8 38.3
mc1157 -77.700000 162.260000 1.7100 0.0100 31.4 37.9 44.9
mc1160 -77.690000 162.350000 3.4700 0.0500 18.1 24.9 31.4
mc1165 -77.510000 169.330000 1.4510 0.0600 20.6 27.8 35.1
mc1167 -77.490000 169.290000 1.3800 0.1000 38.9 43.5 48.4
mc1200 -77.550000 166.160000 0.0730 0.0100 21.2 26.6 31.8
mc1302 -78.190000 165.320000 0.0400 0.0200 23.7 28.9 34.3
mc1304 -78.240000 163.360000 0.2900 0.0400 18.8 24.7 29.5
mc1305 -78.240000 163.230000 0.9000 0.2000 30.6 34.4 38.2
mc1306 -77.700000 162.690000 2.5600 0.2600 4.5 6.9 9.5
mc1307 -77.850000 166.670000 1.3300 0.2400 39.7 46.1 53.5
GHI01 33.126350 35.782270 0.1177 0.0358 20.1 25.2 30.2
GHI02 33.158050 35.776730 0.1296 0.0012 20.5 24.5 27.9
GHI03B 33.122790 35.724160 0.8420 0.0233 66.7 69.3 72.2
GHI03C 33.122790 35.724160 0.8420 0.0233 36.8 45.2 52.5
GHI03D 33.122790 35.724160 0.8420 0.0233 47.0 59.2 70.1
GHI05 32.960510 35.862240 0.1679 0.0255 19.7 22.6 25.0
GHI06 33.069580 35.771430 0.1145 0.0085 26.4 27.4 28.5
GHI07 33.085810 35.755890 0.6805 0.0183 33.6 40.9 47.7
GHI07C 33.085810 35.755890 0.6805 0.0183 21.0 23.2 25.2
GHI10 33.051680 35.849680 0.6149 0.0349 18.1 19.8 21.5
GHI18 33.025833 35.494912 1.6700 0.0400 30.9 37.3 43.6
GHI19 32.995278 35.525986 2.4500 0.0226 27.1 32.8 39.4
GHI20 32.926290 35.849940 1.6500 0.0200 29.9 31.8 33.5
GHI21 32.926290 35.849940 1.6765 0.0302 21.7 23.6 25.6
GHI25 33.218726 35.777062 0.8723 0.0053 44.7 52.9 60.8
GHI26 33.220000 35.776833 0.8704 0.0169 46.2 50.0 53.7
GHI27 33.212500 35.786157 1.1498 0.0348 33.5 36.9 40.6
GHI28 33.212500 35.786157 1.1912 0.0152 21.5 28.0 33.7
GHI29 33.179444 35.793218 0.7496 0.0945 28.7 31.1 33.2
GHI39 33.141000 35.682000 0.8476 0.1165 5.9 14.8 21.7
GHI40 33.141000 35.682000 0.7736 0.1949 4.7 7.1 9.8
GHI41 33.141000 35.683000 0.7902 0.0058 4.7 7.1 10.0
GHI44 33.042000 35.836000 1.4369 0.0195 45.6 48.9 52.6
GHI46 32.868290 35.829050 2.7442 0.0475 51.2 61.2 75.4
Brendan Cycha, Lisa Tauxea, Geoff Cromwellb, John Sintonc, and Anthony A.P. Koppersd3 of 4
References12
1.
H Asefaw, L Tauxe, AAP Koppers, H Staudigel, Four-Dimensional Paleomagnetic Dataset: Plio-Pleistocene Paleodirection
13
and Paleointensity Results From the Erebus Volcanic Province, Antarctica. J. Geophys. Res. Solid Earth
126
,e2020JB020834
14
(2021).15
2.
L Tauxe, H Asefaw, N Behar, AAP Koppers, R Shaar, Paleointensity Estimates from the Pleistocene of Northern Israel:
16
Implications for hemispheric asymmetry in the time-averaged field. Geochem. Geophys. Geosyst.
n/a
,e2022GC010473
17
(2022).18
4 of 4Brendan Cycha, Lisa Tauxea, Geoff Cromwellb, John Sintonc, and Anthony A.P. Koppersd
1
Supporting Information for2
Changes in non-dipolar field structure over the Plio-Pleistocene: New paleointensity results3
from Hawai‘i compared to global datasets4
Brendan Cycha, Lisa Tauxea, Geoff Cromwellb, John Sintonc, and Anthony A.P. Koppersd1
5
aUniversity of California, San Diego; bOccidental College; cUniversity of Hawai‘i at Man¯
oa; dOregon State University6
Brendan Cych. E-mail: bcych@ucsd.edu7
This PDF file includes:8
Fig. S19
Table S110
SI References11
Brendan Cycha, Lisa Tauxea, Geoff Cromwellb, John Sintonc, and Anthony A.P. Koppersd1 of 4
Fig. S1.
Plots of Age against cumulative Argon released for age plateau and mini-plateau (OA019, OA116, OA124) age experiments. The mini-plateau ages for OA124
are concordant with total fusion ages from other samples from same site, and the mini-plateau age for OA019 is close to the age for OA028, a nearby lava flow which it is
stratgraphically below.
2 of 4Brendan Cycha, Lisa Tauxea, Geoff Cromwellb, John Sintonc, and Anthony A.P. Koppersd
Table S1. Results analyzed using BiCEP from Antarctica (1) (sites with the prefix ’mc’) and Northern Israel (2) (sites with the prefix ’GHI’)
Site Latitude Longitude Age (Ma) Age 2‡Bmin Bmedian Bmax
mc1001 -77.850000 166.640000 1.1800 0.0100 12.7 18.6 24.9
mc1002 -77.850000 166.690000 0.3300 0.0200 22.3 29.1 36.3
mc1009 -77.550000 166.200000 0.0740 0.0150 23.9 28.0 35.1
mc1015 -77.470000 169.230000 1.3300 0.0200 22.7 26.7 30.6
mc1020 -77.880000 165.020000 0.7700 0.0320 55.6 62.3 69.3
mc1029 -78.310000 164.800000 0.1800 0.0800 41.7 44.7 48.0
mc1031 -78.350000 164.300000 0.1330 0.0117 23.5 31.2 39.0
mc1032 -78.360000 164.300000 0.0078 0.0120 27.9 31.2 34.9
mc1036 -78.390000 164.270000 0.1200 0.0400 22.2 28.8 35.3
mc1103 -78.240000 163.360000 1.4210 0.0300 14.5 20.1 27.7
mc1109 -78.280000 163.540000 1.2610 0.0400 29.5 33.2 37.0
mc1111 -78.220000 162.790000 1.9900 0.0400 18.5 21.4 24.6
mc1115 -78.240000 162.960000 2.4600 0.3100 26.7 31.2 35.8
mc1119 -78.240000 162.960000 1.0800 0.2200 34.6 37.7 40.9
mc1120 -78.240000 163.090000 1.7560 0.0500 22.4 24.7 27.0
mc1121 -78.240000 162.950000 2.5050 0.0600 28.8 32.8 35.7
mc1127 -78.250000 163.730000 1.9420 0.0680 33.7 37.0 40.4
mc1135 -78.230000 166.560000 3.6000 0.0100 29.4 31.7 33.9
mc1139 -78.260000 163.080000 0.8820 0.0800 24.6 27.8 31.0
mc1140 -78.280000 163.000000 2.0430 0.0900 28.4 34.2 39.1
mc1145 -78.240000 162.893000 1.9000 0.1200 3.3 7.0 10.2
mc1147 -78.200000 162.960000 1.6300 0.3200 16.5 22.4 27.2
mc1155 -77.700000 162.250000 1.5000 0.0500 23.3 30.8 38.3
mc1157 -77.700000 162.260000 1.7100 0.0100 31.4 37.9 44.9
mc1160 -77.690000 162.350000 3.4700 0.0500 18.1 24.9 31.4
mc1165 -77.510000 169.330000 1.4510 0.0600 20.6 27.8 35.1
mc1167 -77.490000 169.290000 1.3800 0.1000 38.9 43.5 48.4
mc1200 -77.550000 166.160000 0.0730 0.0100 21.2 26.6 31.8
mc1302 -78.190000 165.320000 0.0400 0.0200 23.7 28.9 34.3
mc1304 -78.240000 163.360000 0.2900 0.0400 18.8 24.7 29.5
mc1305 -78.240000 163.230000 0.9000 0.2000 30.6 34.4 38.2
mc1306 -77.700000 162.690000 2.5600 0.2600 4.5 6.9 9.5
mc1307 -77.850000 166.670000 1.3300 0.2400 39.7 46.1 53.5
GHI01 33.126350 35.782270 0.1177 0.0358 20.1 25.2 30.2
GHI02 33.158050 35.776730 0.1296 0.0012 20.5 24.5 27.9
GHI03B 33.122790 35.724160 0.8420 0.0233 66.7 69.3 72.2
GHI03C 33.122790 35.724160 0.8420 0.0233 36.8 45.2 52.5
GHI03D 33.122790 35.724160 0.8420 0.0233 47.0 59.2 70.1
GHI05 32.960510 35.862240 0.1679 0.0255 19.7 22.6 25.0
GHI06 33.069580 35.771430 0.1145 0.0085 26.4 27.4 28.5
GHI07 33.085810 35.755890 0.6805 0.0183 33.6 40.9 47.7
GHI07C 33.085810 35.755890 0.6805 0.0183 21.0 23.2 25.2
GHI10 33.051680 35.849680 0.6149 0.0349 18.1 19.8 21.5
GHI18 33.025833 35.494912 1.6700 0.0400 30.9 37.3 43.6
GHI19 32.995278 35.525986 2.4500 0.0226 27.1 32.8 39.4
GHI20 32.926290 35.849940 1.6500 0.0200 29.9 31.8 33.5
GHI21 32.926290 35.849940 1.6765 0.0302 21.7 23.6 25.6
GHI25 33.218726 35.777062 0.8723 0.0053 44.7 52.9 60.8
GHI26 33.220000 35.776833 0.8704 0.0169 46.2 50.0 53.7
GHI27 33.212500 35.786157 1.1498 0.0348 33.5 36.9 40.6
GHI28 33.212500 35.786157 1.1912 0.0152 21.5 28.0 33.7
GHI29 33.179444 35.793218 0.7496 0.0945 28.7 31.1 33.2
GHI39 33.141000 35.682000 0.8476 0.1165 5.9 14.8 21.7
GHI40 33.141000 35.682000 0.7736 0.1949 4.7 7.1 9.8
GHI41 33.141000 35.683000 0.7902 0.0058 4.7 7.1 10.0
GHI44 33.042000 35.836000 1.4369 0.0195 45.6 48.9 52.6
GHI46 32.868290 35.829050 2.7442 0.0475 51.2 61.2 75.4
Brendan Cycha, Lisa Tauxea, Geoff Cromwellb, John Sintonc, and Anthony A.P. Koppersd3 of 4
References12
1.
H Asefaw, L Tauxe, AAP Koppers, H Staudigel, Four-Dimensional Paleomagnetic Dataset: Plio-Pleistocene Paleodirection
13
and Paleointensity Results From the Erebus Volcanic Province, Antarctica. J. Geophys. Res. Solid Earth
126
,e2020JB020834
14
(2021).15
2.
L Tauxe, H Asefaw, N Behar, AAP Koppers, R Shaar, Paleointensity Estimates from the Pleistocene of Northern Israel:
16
Implications for hemispheric asymmetry in the time-averaged field. Geochem. Geophys. Geosyst.
n/a
,e2022GC010473
17
(2022).18
4 of 4Brendan Cycha, Lisa Tauxea, Geoff Cromwellb, John Sintonc, and Anthony A.P. Koppersd
