The climate signal in the stable isotopes of snow from Summit, Greenland: Results of comparisons with modern climate observations

doi:10.1029/97JC00162

HAL Id: hal-03335034

https://hal.archives-ouvertes.fr/hal-03335034

Submitted on 5 Sep 2021

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of sci-

entic research documents, whether they are pub-

lished or not. The documents may come from

teaching and research institutions in France or

abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est

destinée au dépôt et à la diusion de documents

scientiques de niveau recherche, publiés ou non,

émanant des établissements d’enseignement et de

recherche français ou étrangers, des laboratoires

publics ou privés.

The climate signal in the stable isotopes of snow from

Summit, Greenland: Results of comparisons with

modern climate observations

J. White, L. Barlow, D. Fisher, P. Grootes, J. Jouzel, S. Johnsen, M. Stuiver,

H. Clausen

To cite this version:

J. White, L. Barlow, D. Fisher, P. Grootes, J. Jouzel, et al.. The climate signal in the stable

isotopes of snow from Summit, Greenland: Results of comparisons with modern climate observa-

tions. Journal of Geophysical Research. Oceans, Wiley-Blackwell, 1997, 102 (C12), pp.26425-26439.

�10.1029/97JC00162�. �hal-03335034�

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 102, NO. C12, PAGES 26,425-26,439, NOVEMBER 30, 1997

The climate signal in the stable isotopes of snow

from Summit, Greenland:

Results of comparisons with modern climate observations

J. W. C. White, •,2 L. K. Barlow, • D. Fisher, 3 P. Grootes, 4,s

J. Jouzel, 6 S. J. Johnsen, ?,8 M. Stuiver, 4 and H. Clausen ?

Abstract. Recent efforts to link the isotopic composition of snow in Greenland with

meteorological and climatic parameters have indicated that relatively local information

such as observed annual temperatures from coastal Greenland sites, as well as more

synoptic scale features such as the North Atlantic Oscillation (NAO) and the temperature

seesaw between Jakobshaven, Greenland, and Oslo, Norway, are significantly correlated

with 8180 and 8D values from the past few hundred years measured in ice cores. In this

study we review those efforts and then use a new record of isotope values from the

Greenland Ice Sheet Project 2 and Greenland Ice Core Project sites at Summit,

Greenland, to compare with meteorological and climatic parameters. This new record

consists of six individual annually resolved isotopic records which have been average to

produce a Summit stacked isotope record. The stacked record is significantly correlated

with local Greenland temperatures over the past century (r = 0.471), as well as a

number of other records including temperatures and pressures from specific locations as

well as temperature and pressure patterns such as the temperature seesaw and the North

Atlantic Oscillation. A multiple linear regression of the stacked isotope record with a

number of meteorological and climatic parameters in the North Atlantic region reveals

that five variables contribute significantly to the variance in the isotope record: winter

NAO, solar irradiance (as recorded by sunspot numbers), average Greenland coastal

temperature, sea surface temperature in the moisture source region for Summit

(30ø-20øN), and the annual temperature seesaw between Jakobshaven and Oslo.

Combined, these variables yield a correlation coefficient of r = 0.71, explaining half of

the variance in the stacked isotope record.

1. Introduction

One of the great strengths of ice cores as proxies for past

environmental conditions is that they can provide not only the

long timescale necessary to view the large changes of the past

glacial periods but also the high temporal resolution needed to

look at socially relevant timescales, that is, subannual to dec-

adal changes in climate and environmental conditions. In the

past the focus in ice-core research, particularly for stable iso-

topes, has been mostly on the long timescales, with less em-

1Institute of Arctic and Alpine Research, University of Colorado,

Boulder.

2Also at Department of Geological Sciences, University of Colo-

rado, Boulder.

3Glaciology Section, Terrain Sciences Division, Geological Survey of

Canada, Ottawa, Ontario.

4Department of Geological Sciences and Quaternary Research Cen-

ter, University of Washington, Seattle.

SNow at Liebniz Laboratory, Christian Albrechts University, Kiel,

Germany.

6Laboratoire de Mod•lisation du Climat et de l'Environnement,

Centre d'Etudes de Saclay, Gif-sur-Yvette, France.

7Department of Geophysics, University of Copenhagen, Copenha-

gen, Denmark.

8Also at Science Institute, University of Reykjavik, Reykjavik, Ice-

land.

Paper number 97JC00162.

0148-0227/97/97JC-00162509.00

phasis on the recent record and even less on the Holocene.

Given our growing understanding of human impact on the

global environment and climate system, one of the primary

goals of the new Summit cores from the Greenland Ice Core

Project (GRIP) and the Greenland Ice Sheet Project 2

(GISP2) was to take fuller advantage of the rich density of

information available in the cores. In particular, there was an

emphasis on determining how the isotopic signal records the

recent climatic conditions of central Greenland and how those

climate conditions can be related to the climate of the North

Atlantic region. Despite the great potential for using isotopic

data from ice cores to look at climate changes of the past few

centuries, these data have been largely ignored and/or dis-

missed in studies of recent climate change. This is due in part

to a lack of high-resolution isotopic data from the upper parts

of cores and in part to a lack of emphasis on understanding the

nature of the isotopic signal and what it records. Recently,

several papers have addressed this issue, using data from the

new GISP2 and GRIP cores in the Summit region as well as

other isotopic data, mostly collected and analyzed by the Dan-

ish and Canadian groups, which have recently become avail-

able for public distribution via the World Data Centers A for

Paleoclimatology and Glaciology (Boulder, Colorado). In this

paper we will review those efforts, which uniformly point to-

ward the conclusion that annually resolved isotopic data from

Greenland ice cores do contain useful climatic information.

We will then take a first look at the climate signals found in the

26,425

26,426 WHITE ET AL.: CLIMATE FROM ISOTOPES IN GREENLAND SNOW

stacked isotopic record from the Summit region. A total of six

cores in the Summit region were drilled, sampled, and mea-

sured with sufficient temporal detail to calculate annual isoto-

pic values. We have for the first time the capability to stack

isotopic records and thus to greatly decrease the isotopic noise

which is present in a single record, focusing attention on the

common signal present in a suite of cores. The stacked record

is described in a companion paper by Johnsen et al. [this issue].

Here we will compare this record to records of climate change

in the region in order to determine what climatic signals are

present in the isotopic composition of Greenland snow from

the Summit region.

2. Stable Isotope Ratios of Greenland Snow

and Regional North Atlantic Climate:

An Overview of Recent Results

Several recent studies have linked the isotopic composition

of snow (•80/•60 and/or D/H, expressed as 8•80 and 8D) in

the Summit area with climatic indicators or climate proxies in

the North Atlantic region. These include the North Atlantic

Oscillation (NAO), which is the atmospheric pressure oscilla-

tion between Iceland and the Azores [Walker, 1924; van Loon

and Rogers, 1978; Moses et al., 1987; Rogers, 1984; Barlow et al.,

1993; White et al., 1996], regional Greenland temperatures

[Fisher et al., 1996], the temperature seesaw, which is an op-

position in atmospheric temperatures between western Green-

land (Jakobshaven) and western Scandinavia (Oslo) [van Loon

and Rogers, 1978; Barlow et al., 1993], and solar activity [Stuiver

et al., 1995]. The results of these studies are reviewed below.

Our focus here will be on these newer studies as they all

examine data from the Summit area. We will later use the

relationships found in these studies between these climatic

parameters and the isotopic record of central Greenland as a

beginning point in examining the climate signals in the stacked

isotopic record from Summit.

The North Atlantic Oscillation is a unifying atmospheric

feature with associated variability in sea surface temperature

(SST) and land surface temperatures in the North Atlantic

region. The NAO denotes a sea level pressure contrast be-

tween the Icelandic Low and Azores High, specifically the

tendency of these two centers of action to strengthen or

weaken together [Walker, 1924; van Loon and Rogers, 1978;

Moses et al., 1987]. The NAO index is defined as the normal-

ized monthly mean sea level pressure difference between

Akureyri, Iceland, and Ponta Delgada, Azores [van Loon and

Rogers, 1978; Rogers, 1984]. Sea level pressure difference in the

North Atlantic is strongest in the winter, and generally the

NAO is discussed in terms of its winter character.

Investigations of the NAO by van Loon and Rogers [1978],

Rogers and van Loon [1979], Lamb [1977], Kelly et al. [1987],

and Moses et al. [1987] indicate two distinctly different atmo-

spheric and oceanic temperature regimes for the two extreme

modes of the NAO. The temperature seesaw, or tendency for

air temperatures in western Greenland to be warmer when air

temperatures in western Scandinavia are cooler (and vice ver-

sa), describes this opposition of temperatures in the region.

Intensification of the "normal" mode of the NAO (deep Ice-

landic Low, strengthened Azores High) has associations with

the Greenland below (GB; Greenland colder) mode of the

temperature seesaw as well as with stronger westerly winds and

higher than normal SSTs for large parts of the North Atlantic

Drift (Figure la). Stronger zonal westerlies are also associated

with warmer temperatures in England and northern Europe.

Intensification of the "reverse" mode of the NAO has associ-

ations with the Greenland above (GA; Greenland warmer)

mode of the temperature seesaw, as well as with weaker west-

erlies, stronger meridional winds, and lower SSTs in parts of

the North Atlantic Drift (Figure lb). Stronger meridional

winds are also associated with warmer temperatures along the

west Greenland coast.

The link between stable isotope ratios from snow in central

Greenland and the NAO was originally established in the

GISP2 core [Barlow et al., 1993] through association with the

seesaw in winter temperatures between Jakobshaven, Green-

land, and Oslo, Norway. Using the high-resolution isotopic

record, which contains at least eight samples per annual layer,

Barlow et al. compared the isotopic values of snow averaged

annually as well as the lowest isotopic values (isotope "win-

ters") and the highest values (isotopic "summers") with the

temperature seesaw. The focus was on the strong GA and GB

years taken from the winter seesaw values of van Loon and

Rogers [1978] for the years 1840-1970. Within this time period,

there were 23 GA winters and 26 GB winters. A one-tailed

t-test showed high statistical significance for annual and winter

as well as summer signals. The isotopic signal from the GISP2

site responds in agreement with the west Greenland (Jakob-

shavn) side of the temperature seesaw. It should be noted that

the temperature seesaw is not well defined in coastal temper-

ature stations on east Greenland. This suggests that while

there is some regional coherence in the coastal Greenland

temperatures, part of the variability in atmospheric tempera-

tures in Greenland is different from the western to the eastern

sides of the ice sheet. An investigation of storm tracks, cyclo-

genesis, atmospheric circulation patterns, and automatic

weather station (AWS) records suggests that precipitation ar-

rives in central Greenland primarily from both the southeast

and southwest/west [Barlow, 1994; Barlow et al., this issue].

This observation suggests that the isotopic composition of

snow should respond to both the average coastal temperatures

and the temperature pattern described by the temperature

seesaw, a suggestion that will be supported by the results of the

multiple linear regression analysis of the stacked isotope

record described in the second part of this paper.

In a different approach to relating the stable isotopes in the

GISP2 core to climate change, White et al. [1996] examined the

power spectrum of the annually averaged 8D values and com-

pared the periodicities found in the isotopic record with oscil-

lations found in records of climate change. The isotopic record

at GISP2 exhibits many statistically significant oscillations.

White et al. identified several which can also be found in other

annually resolved isotopic records from Greenland, including

sites in southern Greenland (Dye 3) and northern Greenland

(Camp Century). These periods, in years, are 4.5, 6.3, 7.5, 9.8,

12.2, and 16. These oscillations were filtered from the isotopic

time series and then plotted against similarly filtered oscilla-

tions from (1) the NAO, which has its strongest oscillation at

7.6 years, (2) sunspots, with a strong oscillation at 11-12 years,

and (3) the 8180 record from corals, which has a strong oscil-

lation at 4.6 years and which is a proxy of the E1 Nifio-

Southern Oscillation (ENSO). The results of this study found

no statistical coherence, or similarity in the temporal pattern of

the amplitudes, for the isotopic signal versus ENSO, either for

the pressure indices or isotopic values from corals.

There was also no coherence between the l 1-year cycle of

sunspot numbers and the l 1-year cycle found in the GISP2

WHITE ET AL.: CLIMATE FROM ISOTOPES IN GREENLAND SNOW 26,427

a

0 ø,

30'

85'

70'

Higher

$ST

55' W

80;

L

Zonal

winds

40' W

b

H

25' W

10'W

5'E

5O:

winds

0 ø,

100'

Lower L

SST

85'

70'

55' W 40' W 25' W

10'W

5'E

Figure 1. (a) Diagram of Greenland below (GB) mode of the temperature seesaw. In this end-member,

Jakobshavn, Greenland, is relatively cold (marked by a minus sign) and Oslo, Norway, is relatively warm

(marked by a plus sign). Accompanying oceanic and atmospheric phenomena, all associated with extreme

normal mode of the NAO, are also noted: zonal winds with lower pressure north and higher pressure south

and higher sea surface temperatures in the western North Atlantic. The ice-core sites for GRIP and GISP2

are marked (Summit). The diagram is a compilation of observations of van Loon and Rogers [1978], Rogers and

van Loon [1979],Lamb [1977],Kelly et al. [1987], and Moses et al. [1987]. (b) As in Figure la, but for Greenland

above (GA) mode of the temperature seesaw (extreme reverse mode of the NAO).

isotopic record. Comparison of these two time series filtered to

isolate the 11-year component showed a slight correspondence,

as the strongest sunspot minima tended to occur in phase with

the strongest isotope minima. This correspondence was not

compelling evidence, however, that isotopes and sunspots are

linked. A different conclusion was reached by Stuiver et al.

[1995], who compared the GISP2 isotopic record with sunspots

for much of the Holocene. Their conclusion was that sunspots

and isotopes did indeed appear to show common variance over

this longer time period. Stuiver et al. also noted that volcanic

26,428 WHITE ET AL.: CLIMATE FROM ISOTOPES IN GREENLAND SNOW

Band-pass Filtered Components of the GISP2 •SD and the Winter NAO Indices

0

•-2

-4

-8

1880

-4

-2

-1•c •

0

- 3

1900 1920 1940 1960 1980

Year

Band-pass Filtered Components of GISP2 •SD and the Winter NAO Reconstruction

-4

m 0 •'

-4 2

3

-8 4

1700 1750 1800 1850 1900 1950 2000

Figure 2. (a) Comparison of time series of/SD values from

the GISP2 ice core (thin line) and winter NAO indices (thicker

line) which have been band-pass filtered at 7.5 years to isolate

this oscillation in both records [from White et al. 1996]. In the

isotope data this is one of many significant oscillations. In the

NAO data this oscillation is significant and is the strongest in

the power spectrum. Note that both the phase and amplitude

of the two time series are very similar, implying that they are

responding to the same forcing or that this oscillation in the

isotope record is forced by the NAO oscillation. (b) As in

Figure 2a, except the winter NAO indices are inferred from

tree ring indices [Cook et al., 1997]. The match for these two

series is again excellent in the 1900s, but both the phasing and

particularly the amplitude deviate in the previous 2 centuries.

The implications of this are discussed in the text.

events appear to have an impact on the isotopic record, with

evidence of lower/5•80 (cooling) corresponding with volcanic

eruptions. They separated the volcanic events by size using

indices of explosivity and showed that larger eruptions tended

to be associated with larger/5•80 drops.

White et al. [1996] did find strong statistical coherence be-

tween the 7.6-year oscillations in the isotopic and NAO

records. In addition, the temporal patterns of these oscilla-

tions, when filtered from the individual records, match well for

the past 100 years. This is shown in Figure 2a. One drawback

of comparing the isotope values with NAO indices is that the

length of the NAO record is relatively short. In order to extend

the NAO record, a record of winter NAO indices derived from

tree rings [Cook et al., 1997] was also filtered for the 7.6-year

oscillation and then compared with the 7.6-year oscillation

filtered from the isotope time series. While all three filtered

records were very similar for the period 1880 to the present,

the correspondence between the tree ring NAO proxy and the

isotope series was not good between 1700 and 1880. This is

shown in Figure 2b. White and coworkers speculated that this

finding could mean that the NAO had a different behavior

prior to 1880, at least in the nature of the oscillations in the

indices, and that this difference could be driven by a different

climate mode during the Little Ice Age or by anthropogenic

influences in climate which began to manifest themselves in the

late 1800s. Of course, the result could also mean that one or

both of the climate proxies did not track the 7.6-year oscillation

in the NAO as well prior to 1880 for reasons having to do with

the proxy record and not the climate system. Nonetheless, the

apparent difference in behavior between 1700-1880 and 1880

to the present in the isotopic record and the tree ring NAO

record does raise some important issues for this study. First, it

reinforces the importance of using long records of climate

change to establish a baseline behavior in the climate system,

even if they are proxy records with the accompanying caveats

in interpretation, and second, it raises the point that calibrating

proxy records with historical observations of climate carries

some risk that uniformitarianism may not apply and thus the

proxy may not be correctly interpreted beyond the period of

historical observations.

One of the most exhaustive studies to date of the climate

information in stable isotope ratios in Greenland ice cores is

that of Fisher et al. [1996]. In this study, isotope records from

about 20 ice cores were used, each with isotopes measured at

subannual (>8 samples per year) resolution. Regional chro-

nologies of isotope ratios covering the last few centuries were

constructed for southern Greenland, central west Greenland,

central east Greenland, and northern Greenland, as well as for

the Canadian island sites Aggasiz, Ellesmere, and Devon.

For each region, isotope records from several individual

cores were stacked in order to reduce noise. Fisher et al. [1996]

discuss in detail the issue of noise in the isotopic times series

from individual cores. Noise in this sense is defined as isotopic

variability due to postdepositional processes which can alter

the original isotopic signal in the snow. These processes in-

clude unevenness in drifting or sastrugi, wind scouring, sum-

mer melting and penetration of the melt into the lower layers,

and vapor diffusion, which redistributes the isotopic gradients

occurring naturally due to seasonal changes in the isotopic

composition of snow.

The signal to noise ratio (S/N = (signal variance)/(noise

variance)) for a single isotopic profile in Greenland varies with

accumulation and latitude [Fisher et al., 1985]. For Greenland,

S/N appears to be smallest in the Summit area. The frequency

distribution of both the noise and signal variances in the ice

cores depends on the variable. For example, noise in snow

accumulation time series is concentrated in the higher frequen-

cies. Such noise can be reduced in a single time series by

averaging over relatively short time steps. For isotopic time

series, however, vapor diffusion tends to smooth out the high-

er-frequency variability and transfer it to the lower frequen-

cies, reddening the spectrum. This means that longer averaging

times, of the order of a decade or more, are needed to smooth

out the noise in individual isotopic time series. The necessity

for long averaging times has been observed when comparing

isotopic records from Greenland, although such averaging is

also required when cores are separated by large distances over

which climate is not correlated on the subdecade timescale

[e.g., Dansgaard et al., 1975].

The climate signal in the stable isotopes of snow from Summit, Greenland: Results of comparisons with modern climate observations

Summary (1 min read)

Introduction

3.2.1. Comparison with coastal temperatures. On the basis of both theoretical considerations and observations, stable

Citations

Cites background from "The climate signal in the stable is..."

References

Related Papers (5)