Understanding how internal climate variability influences arctic regions is
required to better forecast future global climate variations. This paper
investigates an annually-laminated (varved) record from the western Canadian
Arctic and finds that the varves are negatively correlated with both the
instrumental Pacific Decadal Oscillation (PDO) during the past century and
also with reconstructed PDO over the past 700 years, suggesting drier Arctic
conditions during high-PDO phases, and vice versa. These results are in
agreement with known regional teleconnections, whereby the PDO is negatively
and positively correlated with summer precipitation and mean sea level
pressure respectively. This pattern is also evident during the positive
phase of the North Pacific Index (NPI) in autumn. Reduced sea-ice cover
during summer–autumn is observed in the region during PDO
In the North Pacific region, the Pacific Decadal Oscillation (PDO) is the
major mode of multi-decadal climate variability (Mantua et al., 1997). The
PDO can be described as a long-lived pattern of Pacific sea surface temperature (SST) variability similar to the El Niño–Southern Oscillation
(ENSO)
(Allan et al., 1996; Zhang et al., 1997) or as a low-frequency residual of
ENSO variability on multi-decadal timescales (Newman et al., 2003). During
the warm (positive) PDO phase (PDO
PDO modulation of western Canadian Arctic climate.
In recent years, several varved records have been established in the Arctic (at Cape Bounty: Cuven et al., 2011; Lapointe et al., 2012; at South Sawtooth Lake: Francus et al., 2008; at Lake C2: Douglas et al., 1996; at Murray Lake: Besonen et al., 2008; and Lower Murray Lake: Cook et al., 2009) in order to investigate past climate variations. Amongst them, the Cape Bounty record is most probably the best documented because it has been supported by climate, hydrological and limnological research at the Cape Bounty Arctic Watershed Observatory since 2003. The annual nature of this sedimentary record, its duration (1750 years) and the above-average quality of its chronology opens the opportunity to investigate (1) correlations with instrumental records, (2) cyclicity of this record by time-series analysis, (3) teleconnections with major climate indices and (4) the long-term influence of the climate mode of variability on the western Canadian Arctic.
Cape Bounty East Lake (hereafter CBEL, 5 m a.s.l.; Fig. 1 black asterisk)
is located on southern Melville Island in the western Canadian High Arctic
(74
To understand the recent relationship between the western Canadian Arctic
climate and the PDO, a one-point correlation map was calculated using the
Pearson's correlation. These were prepared using the Climate Explorer tool
that is managed by the Royal Netherlands Meteorological Institute (Trouet and
Van Oldenborgh, 2013; Van Oldenborgh and Burgers, 2005). Precipitation,
sea level pressure, temperature and sea-ice anomalies were obtained from the
ERA-Interim reanalysis (Dee et al., 2011), a dataset that provides robust
observations of mean temperature and precipitation in the Canadian Arctic
(Lindsay et al., 2014; Rapaić et al., 2015). For zonal and meridional
wind, NCEP/NCAR (Kalnay et al., 1996), which covers the period 1950–2016
was used. The PDO as defined in Mantua et al. (1997) is derived as the
leading principal component of monthly SST anomalies in the North Pacific
Ocean, poleward of 20
The methods used to count varves rely on both visual examination of thin
sections and the use of
Varve thickness (VT) and grain-size data (Lapointe et al., 2012), available from the NOAA palaeoclimate database, were linearly detrended. A Box-Cox transformation was then used to stabilize variance in the time series (note that the use of both no transformation or a log transformation of the time series yielded similar results). The data were normalized to allow for a comparison with other time series. Three PDO reconstructions (D'Arrigo et al., 2001; Gedalof and Smith, 2001; MacDonald and Case, 2005) were used for comparison with the CBEL record. Spectral analyses were carried out using REDFIT (Schulz and Mudelsee, 2002), and wavelet analyses were performed with the software R (R Development Core Team, 2008) using the package bi-wavelet (Gouhier and Grinsted, 2012). For wavelet analysis the interval 244–2000 CE was analysed as the lake was fully isolated by glacio-isostatic uplift from the ocean after 244 CE (Cuven et al., 2011; Lapointe et al., 2012).
North Pacific Index (NPI) and precipitation during
September–November.
Several key climate indices demonstrate the present-day influence of the PDO
on the western Canadian Arctic. The correlation between the PDO index (Mantua
et al., 1997) based on ERSSTv4 (Huang et al., 2015) and sea-ice cover (Dee et
al., 2011) is positive during summer and autumn over the region (Figs. 1a,
S1). An anomalous surface high-pressure system develops in the vicinity of
southern Melville Island from July to September (JAS) (Fig. 1b) during
positive PDO phases (PDO
Another important teleconnection is revealed in the spatial correlation
between PDO and mean sea level pressure (MSLP) during winter (Fig. 1d). The
mid- to high-latitude manifestation of the PDO includes a wave train that is
characterized by a deepening of the Aleutian Low and a high-pressure system
to the northeast over the Canadian Arctic during PDO
Instrumental PDO (NOAA) compared with grain size at Cape Bounty East Lake from 1900 to 2000 (best correlation is achieved when CBEL lags PDO by 1 year). Bold lines are 10-year low-pass filtered. A turbidite dated to 1971 CE eroded 7 years (Lapointe et al., 2012).
The western Canadian Arctic is also strongly influenced by the NPI during September–November (SON) (Fig. 2). The NPI is a more
direct measure of the strength of the Aleutian Low (Trenberth and Hurrell,
1994) and has been shown to be part of the PDO North Pacific teleconnection
(Schneider and Cornuelle, 2005). A weakened Aleutian Low (increased MSLP) is
seen in the Pacific during times of positive NPI (NPI
The sedimentary varve record gives support to these instrumental climate
observations. Annual coarse grain size (98th percentile) (Lapointe et al.,
2012) is negatively correlated with the instrumental PDO (Mantua et al.,
1997) during the last 100 years (no lag:
Correlation analysis for the varve thickness at Cape Bounty East
Lake and different proxy records of PDO.
For the time interval 1300–1900 CE, a single 1.34 cm thin erosive bed is
evident in the sedimentary record (Supplement Text 1, Supplement Fig. S4),
making the comparison of the CBEL VT with other palaeo-PDOs
acceptable. The three reconstructed PDOs (MacDonald and Case, 2005; Gedalof
and Smith, 2001; D'Arrigo et al., 2001) show periods of high coherency, but
there are periods of low consistency between them (Fig. 4a–c), as reported
in the literature (Kipfmueller et al., 2012; Wise, 2015). These reconstructed
PDOs are probably best interpreted as reflections of the PDO at their given
study sites, explaining the lack of co-variability during certain periods. To
better explain the variance in the palaeo-PDO time series, a principal
component analysis (PCA) was performed on the three reconstructed PDOs. The
PC1 (Fig. 4d) explains 51 % of the variability (loadings factors: 0.58,
D'Arrigo et al., 2001; 0.68, Gedalof and Smith, 2001; and 0.65, MacDonald and
Case, 2005) and its highest correlation with VT is achieved with an 18-year
lag (Fig. S3:
To obtain accurate confidence intervals for the linear correlation between the PDO records and CBEL, a non-parametric stationary bootstrap, using 1000 iterations, is used (Mudelsee, 2010). The optimal average block length is determined using the method described in Patton et al. (2009), which is well suited for autocorrelated time series. The correlation analysis performed on both raw and 5-year filtered data shows a large improvement of the significance levels with filtered data (Fig. S8) as well as stronger correlation coefficients (Table 1). Also, the use of filtered data provides narrower confidence intervals, that is, less uncertainty. The visible and statistically significant negative correlation between three independent PDO records and CBEL strongly support our assumption that the VT at our site is influenced by the PDO.
To further support the link between the Cape Bounty sequence and the PDO
(NPI), spectral analysis of the entire VT record for the 244–2000 CE period
found significant (
When the western Canadian Arctic is characterized by lower-pressure system
anomalies when the Aleutian Low is in a weakened state (increased sea level pressure,
NPI
These meridional wind anomalies appear to persist during the cold season
(Fig. S9), although they are not as pronounced over the western Canadian
Arctic as in SON (Fig. 6a). This is consistent with annual
surface wind stress differences between PDO phases over the North Pacific
(Zhang and Delworth, 2015) during the 20th century (Fig. 7). Indeed,
sustained southerly wind anomalies are observed in the northernmost part of
the Pacific during PDO
PDO modulation of winds and sea surface temperature in the Pacific (Zhang and Delworth, 2015). Regression of SST
(
Warmer summer temperatures during PDO
This study suggests a significant influence of the PDO (NPI) on the climate
of the western Canadian Arctic, a region where instrumental data coverage is
very sparse and the duration of available records is short. Spatial
correlations using both instrumental and reanalysis data indicate a strong
atmospheric teleconnection, likely responsible for the increase of
precipitation during PDO
All of the palaeo data can be found on the NOAA server
The authors declare that they have no conflict of interest.
We wish to thank the Polar Continental Shelf Program for their field logistic
support and NSERC grants to Pierre Francus and Scott F. Lamoureux.
François Lapointe is grateful to grants provided by the FRQNT and the
W. Garfield Weston Foundation. We thank Geert Jan van Oldenborgh for advice
with the use of the KNMI database. We also thank James Screen for
constructive advice and Byron Steinman and Ze'ev Gedalof, who provided
information on PDO datasets. François Lapointe would also like to thank David Fortin and the Ouranos Consortium
for constructive conversations. Palaeo-data used in this study can be found on
the NOAA server