1. Introduction
Vigorous large-scale oceanic circulations and intense air–sea interactions in the mid- to high-latitude North Atlantic make this region an important component in global climatic studies. Here, tropical warm and saline seawaters, transported northward by the Gulf Stream (GS) and the North Atlantic Current (NAC), meet the cold and fresh Arctic waters that are transported southward by the Greenland and the Labrador Currents. The strong boundary currents constitute the cyclonic North Atlantic subpolar gyre (SPG), whose variability, in both intensity and shape, is critical in the redistribution of heat and salt in the North Atlantic (Hátún et al. 2005; Sarafanov 2009; Häkkinen et al. 2011). The SPG is also known as the formation region of intermediate water due to intense air–sea interaction, especially during severe winters (Dickson et al. 1996; Marshall et al. 1998). Warm and saline waters become dense after releasing heat to the atmosphere and thus sink at certain locations in the SPG, generating an equatorward return flow at depth. This return flow coupled with surface northward western boundary currents together form the lower and the upper limbs of the Atlantic meridional overturning circulation (AMOC), which may have significant impacts on abrupt climate changes (Clark et al. 2002; McManus et al. 2004; Lynch-Stieglitz et al. 2007, 2011).
One of the most challenging aspects in studying the variability of the North Atlantic is the involvement of physical processes acting over a wide range of temporal and spatial scales, from the short time scale local tip jets events (e.g., Våge et al. 2008) to eddy activities with strong seasonality and interannuality (e.g., Lilly et al. 2003; Volkov 2005) to variations in horizontal gyre circulations at intraseasonal to interannual time scales whose spatial scales could have an impact on climate change. The subpolar region is particularly climate-relevant as variability in the SPG intensity in past decades has been linked to variations in the deep-water formation rate in the Labrador Sea and the AMOC (Böning et al. 2006; Bersch et al. 2007; Rhein et al. 2011). The observed spinup (Belkin 2004) and the subsequent spindown of the SPG (Häkkinen and Rhines 2004) and the recent weakening of the AMOC (Latif et al. 2006) have revealed the dramatic variability of the North Atlantic at interannual to decadal time scales. Moreover, the apparent warming of subsurface ocean waters in the subpolar region has been suggested to be a major contributor for accelerated glacier melting (Bindschadler 2006; Holland et al. 2008), implying a potential relationship between the SPG intensity that determines the amount of heat transported along Greenland coast and the melting ice mass along the margin of the SPG. It is, however, still not well understood whether the ongoing warming is part of low-frequency variations or a long-term trend. Therefore, distinguishing the relative importance of different modes of variability and possible related causes has become of great necessity and has increasingly captured the interest and attention of both scientists and policymakers.
Sea surface heights, which have been measured by the repeat coverage of altimeters, provide one of the most important oceanic signals related to the climate variability at both global and regional scales. The sea level responds to high-frequency wind forcing, for example, during storms and hurricanes, and has been used for eddy studies (e.g., Volkov 2005; Chelton et al. 2011). In addition, the sea surface height anomaly (SSHA) can be interpreted as either the variations of upper ocean water properties due to changes of temperature and salinity in the water columns or the changes in the oceanic circulation given that the dynamic sea level variations balance geostrophic velocity anomalies (e.g., Häkkinen and Rhines 2004; Böning et al. 2006). Therefore, the sea level is a good indicator for large-scale oceanic variability particularly at annual and longer time scales. Since the launch of TOPEX/Poseidon and ERS-1 and -2 in the early 1990s, followed by Jason-1 and Jason-2, the precision of an individual sea surface height measurement based on these missions has now reached the 1–2-cm level (e.g., Beckley et al. 2010). This continuous sampling of sea level with increasing accuracy using altimetry measurements provides a powerful tool to study the variability of the sea level in the context of climate variability and climate change.
Sea level variations in the North Atlantic have been investigated by previous studies for certain time scales. Ferry et al. (2000) attributed the seasonality of the SSHA in the North Atlantic to the steric changes of heating by examining in situ data and a numerical simulation. At interannual time scales, the SSHA changes were found to coincide with the shifts in the North Atlantic Oscillation (NAO) index (Häkkinen 2001; Esselborn and Eden 2001; Volkov and van Aken 2003). Based on numerical simulations, Häkkinen (1999, 2001) concluded that the low-frequency SSHA variability along the GS and in the SPG is related to the meridional overturning process. Zhang (2008) further suggested that the SSHA pattern could be used as a “fingerprint” of AMOC variations at middle to high latitudes using a 1000-yr model simulation. Li et al. (2012), based on altimeter observations alone, indicated that the low-frequency SSHA variability in the North Atlantic responds to the cumulative NAO forcing and its distinct reduction of variance between the 1990s and 2000s might relate to the propagation of AMOC variations. However, one should be cautious about using SSHA to represent the AMOC change given the existence of a high-frequency wind-driven response as suggested by Lorbacher et al. (2010), based on a sequence of global ocean-ice model experiments. The authors also pointed out that a careful interpretation of the SSHA patterns is needed since its causes depend on the time scales of interest. Given that the changes associated with eddies and winds are large and the length of altimeter record is relatively short, it is possible that small shifts in the short-term variability will get aliased into the trend. In the present paper, we primarily analyzed 18-yr records of the SSHA (1993–2010) using the merged products of multiple altimetry missions that were or currently are on operation. We explored the characteristic features of the SSHA patterns in the North Atlantic at various time scales and then investigated the possible causes. This analysis helps to advance our understanding of the North Atlantic variability, especially in the SPG, in terms of the SSHA variations. The Hilbert–Huang transform (HHT) was applied to examine different modes of variability as well as the long-term trends in the SSHA. Based on the separation of sea level modes, we intended to find the dominant time scales of variability and factors that may be responsible for each mode of variability.
Contents are organized as follows. Section 2 offers a brief description on the data and methods used. In section 3, the altimeter data and the decomposed components at various time scales are analyzed. Spatial and temporal differences of the SSHA variability are discussed with more emphasis on the SPG. The possible causes of the SSHA variability at corresponding time scales are explored in section 4 by primarily investigating the upper layer heat content anomaly. The conclusions appear in section 5.
2. Data and method
a. Data used
Weekly and monthly altimeter products with ⅓° spatial resolution were produced by SSALTO–Data Unification and Altimeter Combination System (DUACS) and distributed by Archiving, Validation and Interpretation of Satellite Oceanographic data (AVISO), with support from the Centre National d’Etudes Spatiales (CNES). The delayed time (DT) products combine data from different missions, significantly improving the estimation of mesoscale features (Le Traon and Dibarboure 1999).
Quality-controlled subsurface ocean temperature and salinity data were obtained from the Ensemble3 (EN3) dataset (Ingleby and Huddleston 2007). This dataset combines data from the World Ocean Database 2005 (WOD05), the Global Temperature and Salinity Profile Project (GTSPP), Argo, and the Arctic Synoptic Basinwide Oceanography (ASBO) project. The quality control procedure includes the use of altimeter data. Objectively analyzed gridded data at monthly intervals at 1° × 1° resolution were used.
Monthly surface flux data are provided by the National Centers for Environmental Prediction (NCEP) Reanalysis datasets from Physical Sciences Division (PSD) (Kalnay et al. 1996). Net heat flux is obtained by summing up the sensible and latent heat fluxes and the shortwave and longwave radiations with positive values denoting heat gain by the ocean.
b. Hilbert–Huang transform
The combination of the Hilbert spectral analysis and the empirical mode decomposition (EMD), namely Hilbert–Huang transform, is designed specifically for analyzing nonlinear and nonstationary data (Huang et al. 1998; Huang and Wu 2008). This method, to a large extent, overcomes the difficulties of the more commonly used Morlet wavelet analysis that include the leakage problem due to limited length and the nonadaptive nature of the basic wavelet function. The Hilbert transform is a useful tool to estimate the amplitude and frequency of a signal over time, allowing localization of specific events in time–frequency space. However, the Hilbert transform requires data to be band-limited in order to produce meaningful results, and it is not obvious, a priori, what bands should be selected. To resolve this issue, the EMD method was proposed by Huang et al. (1998), which decomposes any complicated dataset into a finite and, often, small number of intrinsic mode functions (IMFs) that admit well-behaved Hilbert transforms (see the appendix for more details).
The essence of the EMD method is to identify the IMFs by their characteristic time scales in the data empirically, and then decompose the data accordingly. Each IMF must satisfy two conditions: 1) in the whole dataset, the number of extrema and the number of zero crossings in each must either be equal or differ at most by one, and 2) at any data point, the mean value of the envelope defined using the local maxima and the envelope defined using the local minima is zero. In the present study, we applied the improved white noise–assisted decomposition method of ensemble EMD (EEMD). The EEMD method can eliminate, to a large degree, the frequent appearance of the mode mixing associated with traditional EMD (Wu and Huang 2009). This approach defines the true IMF components as the mean of an ensemble of trials, each consisting of the data plus a white noise of finite amplitude. By eliminating the problem of mode mixing, it also produces a set of IMFs that can be physically interpreted and a time–frequency distribution without transitional gaps (Huang and Wu 2008). The IMFs isolate physical processes of various time scales and also give the temporal variation with the processes in their entirety without resorting to the linear assumption. The removal of certain frequency components will therefore not affect other information, either linear or nonlinear, included in the signal. Statistical significance testing of the components is performed based on the distribution of energy as a function of mean period of the IMFs relative to that of pure white noise (Wu and Huang 2005). It is designed to objectively determine the information content of the IMF components from a noisy dataset derived from EEMD relative to the white-noise statistics (see the appendix for description on the statistical significance test of IMFs).
The MSD is a function of frequency, offering a measure of total amplitude (or energy) contribution from each frequency value. It represents the cumulated amplitude over the entire data span in a probabilistic sense, which means a higher likelihood for an oscillation at a frequency have appeared locally in the whole time span of the data (Huang et al. 1998). The IE depends only on time and can be thus used to check the energy fluctuation.
3. SSHA in the North Atlantic
a. Dipole pattern
We investigated the characteristic patterns of the SSHA over the middle to high latitudes of North Atlantic using multisatellite altimeter measurements from 1993 to 2010. The SSHA evolution in the North Atlantic reveals a positive sea level trend centered over the SPG (Fig. 1a), which corresponds to a scenario of weakening SPG surface circulation after mid-1990s that has been seen in previous studies (Verbrugge and Reverdin 2003; Häkkinen and Rhines 2004; Böning et al. 2006; Li et al. 2012). Such sea level changes could also indicate shifts in the SPG shape as suggested by Hátún et al. (2005). The SSHA trends in the other regions, especially in the GS region (weakly negative trend), may not represent a true long-term trend given their small coefficients of determination (Fig. 1b; the coefficient of determination is given by 1-residual sum of squares/total sum of squares). Hence, the SSHA changes in the SPG could be characterized by a linear trend while they may fluctuate without a simple trend in the GS region. The former signal could highlight the persistence of the sea level changes in the SPG since the 1990s.
(a) The linear trend of the sea surface height anomaly in the study domain and (b) the coefficient of determination, which is a measure of the goodness of linear fit (1-residual sum of squares/total sum of squares). The 1000-, 2000-, and 3000-m isobaths are shown in solid gray lines.
Citation: Journal of Climate 29, 13; 10.1175/JCLI-D-12-00670.1
Figure 2 indicates high standard deviation (STD) of the SSHA over the GS region and low STD over the SPG, which is consistent with the low and high determination coefficients, respectively, as shown in Fig. 1. The GS, the NAC, and the Azores Current embrace most of the SSHA variance, in which the highest STD values (>10 cm) are constrained by 1000-m isobath to the west and the east and by the sub-Arctic front (SAF) to the north. The SPG interior has more variability than its periphery, which reflects larger variations of the horizontal circulations. The contrast between the SPG and the GS region in the linear trend as well as in the STD suggests different dominant time scales of variability due to different physical processes. Given that the linear trend might be a part of long time scale variability, a more general conclusion could be made here: the SSHA variability in the SPG has very important low-frequency components compared to that in the GS. This is tested in following sections.
The standard deviation of the sea surface height anomaly.
Citation: Journal of Climate 29, 13; 10.1175/JCLI-D-12-00670.1
b. SSHA variability
Volkov and van Aken (2003) examined the SSHA variability at annual time scales and interannual time scales in the North Atlantic. The authors used a harmonic shape with a frequency of 1 cycle per year (cpy) to determine the annual sea level cycle and applied a running mean with a window width equal to about 1 year to estimate the interannual signal. However, given that even mean sea level fluctuates from year to year, estimation of a signal with a fixed frequency window may remove physically meaningful nonstationary information carried by the signal. A sine function, therefore, may not be suitable to analyze the SSHA variability, especially when the study area has undergone a clear nonstationary warming process. Volkov and van Aken (2003) also noted the difficulties in estimating the characteristic period of the interannual change or in obtaining meaningful long-term trend with only 8 years of measurements. With such a short record, small shifts in the timing of the annual cycle (for example) may be aliased into the trend.
To differentiate the SSHA signals with limited frequency bands and take into account the nonlinear and nonstationary nature of the process, we decomposed the monthly SSHA time series into components with variable periods (the IMFs and residual) by using the EEMD method (see the appendix for description and an example of EEMD decomposition of SSHA time series). The decomposition was applied to every spatial point over the study domain. Figure 3 shows the temporal averages of the IMFs with high to low frequency (C1 to C6) and of the residual (R) resulting from the EEMD method, respectively. Given that the SSHA was obtained referencing to the 7-yr mean sea level from 1993 to 1999 (AVISO procedure), the time-averaged SSHA over the entire period of this study (1993–2010) represents the sea level changes relative to the reference period. Therefore, Fig. 3 reflects the 1993–2010 average relative to the 1993–99 average using monthly mean data. It is worth mentioning that the residual (or the long-term trend) differs from a linear trend because it can still have time-dependent fluctuations yet at a very low frequency. Significance testing was performed and only those values significant at the 95% significance level were kept. C1 represents the intraseasonal sea level variation, which may be associated with mesoscale eddy activity (e.g., Volkov and van Aken 2003). C2 and C3 have periods around 1 year, representing an annual or quasi-annual cycle. C4–C6 correspond to interannual variations. It is notable that the residual and the original SSHA data have the same order of amplitude, which is greater than that of the other components. This again suggests the importance of the long-term SSHA variability in the North Atlantic, whose actual period may not be well resolved because of the limited time span of data used. Moreover, SSHA variations are found to mainly localize in the SPG and the GS regions while the variations in the regions beyond are generally negligible but with increasing trends. The increases and decreases of sea level in the SPG and the GS respectively as shown in Figs. 3a and 3h occurred mainly in the 2000s. For the other components, however, the sea level in the SPG changed slightly (Figs. 3b–d) or even decreased somehow at longer time scales (Figs. 3e,f) in the 2000s. Caution should be taken in the GS region, because in this region only the intraseasonal and annual signals (C1 and C2) are statistically significant. Both of these two modes showed increased sea level in this region, in contrast to the decreasing trend (Fig. 3h).
The temporal averages over the period of 1993–2010 of (a) original SSHA, (b)–(g) the IMFs at increasing time scales, and (h) the residual (long-term trend).
Citation: Journal of Climate 29, 13; 10.1175/JCLI-D-12-00670.1
After the decomposition of the sea level variability, we examined the contribution from each component to the total variance, showing the percentage of the variance explained by sea level variability over certain time scales (Fig. 4). Sea level variations in the SPG are mostly explained by the long-term trend (R, about 50%; Fig. 4g), confirming a persistent change in the SPG sea level from the 1990s to the 2000s. Besides the long-term trend, a significant part of the apparent sea level variance in the SPG is due to shifts in the seasonal cycle (C2, about 30%; Fig. 4b). In contrast, the total variance of the sea level in the GS region is dominated by the intraseasonal (C1, about 40%) and the annual signals (C2, about 30%), while less than 20% is explained by the residual. The intraseasonal SSHA signals may relate to direct wind forcing and current meandering and resultant persisting eddy generation. The annual signals could be related to shifts in the seasonal cycle. It was also noted that in the regions beyond the SPG and the GS, small SSHA variations with amplitude of ±1 mm are dominated by the annual signal.
The variance explained by the signals over different time scales corresponding to components in Figs. 3b–h.
Citation: Journal of Climate 29, 13; 10.1175/JCLI-D-12-00670.1
c. Characteristic SSHA variations in the subpolar gyre
We focused on the SPG given its crucial role in linking the upper and the lower limbs of the AMOC (Bersch et al. 2007; Rhein et al. 2011) and its undergoing changes in terms of both its intensity and its shape (Hátún et al. 2005; Sarafanov et al. 2008; Sarafanov 2009). Weekly SSHA was spatially averaged over the SPG domain (52°–66°N, 65°–20°W) for analysis. The EEMD results are presented in Fig. 5, which includes the full decomposition to illustrate the high- to low-frequency progression. In total eight IMFs and a residual were obtained due to the improved temporal resolution of the dataset. All IMF components are statistically significant at 99% confidence level. C1, C2, and C3 are the intraseasonal signals with periods ranging from weeks to months. They correspond to fast sea level variations that might relate to regional events such as storm surge occurrences. Given that the contributions from the intraseasonal signals are generally small in comparison to the total SSHA variance (Fig. 4), we did not perform further interpretation on this time scale. C4 has the largest amplitude of variations with clear seasonality, which is associated with the solar radiation in this region. At interannual to decadal time scales (C5, C6, C7, and C8), signals have irregular cycles in which a depressed sea level was observed in the 2000s. The long-term trend (R) shows an overall sea level rise at the rate of approximately 4 mm yr−1, suggesting a regional exaggeration of the sea level rise in the SPG (a global mean rate is 3.2 ± 0.6 mm yr−1 reported by AVISO; http://www.aviso.oceanobs.com/en/news/ocean-indicators/mean-sea-level/). However, there is variable rate of the sea level rise in the long-term trend and a reduction of the rates was observed after around 2002.
The IMFs and the residual decomposed from the spatially averaged SSHA over the subpolar gyre. Units are cm.
Citation: Journal of Climate 29, 13; 10.1175/JCLI-D-12-00670.1
Hilbert transforms were then performed on all IMFs in order to examine not only the contribution from each mode to the total SSHA variations but also how this contribution changes with time. The resultant Hilbert amplitude spectrum provides a unique amplitude–frequency–time plot in which the amplitude is a function of both frequency and time. The MSD and IE were computed and displayed along for illustration. Figure 6 shows the results after applying Hilbert transform to all IMFs (C1–C8). The amplitudes with frequency higher than 2 cpy are not displayed. A maximum in the Hilbert spectrum was found at the frequency of 1 cpy, corresponding to the annual signal dominating over the time span of this study. This annual signal is also reflected in a rather broad peak in the MSD given that small fluctuations still exist in the annual signal. Another persistent maximum appeared at very low frequencies around approximately 0.1 cpy. This maximum yields a relatively sharp peak in the MSD, indicating its periodic nature. Besides the dominant annual signal, the appearance of signals within a small band of low frequencies may be of more interest given their potential connection to long time scale processes in the North Atlantic. There are also some local maxima at the interannual time scales, revealing more complex and nonperiodic phenomena associated with the sea level variations in the SPG. Furthermore, the IE plot shows a decreasing spectrum density for the period from the 1990s to the early 2000s but this is followed by a period of increasing energy. While the decrease may be related to reduction in kinetic energy due to weakening of the SPG surface circulation since the 1990s (Häkkinen and Rhines 2004), the following regaining of energy might be a result of a major “shift in the system” that occurred in the beginning of the 2000s (Sarafanov et al. 2012). The latter may relate to the changes in the production of the intermediate layer water mass in the Labrador Sea (van Aken et al. 2011) as well as to the changes in the SPG intensity indicated by the first principal component of the sea surface heights (Häkkinen and Rhines 2009).
The Hilbert transforms of the IMFs (C1–C8). (a) The instantaneous energy density level (IE) in cm2, (b) the Hilbert spectrum, and (c) marginal spectral density (MSD) of each frequency (in cm).
Citation: Journal of Climate 29, 13; 10.1175/JCLI-D-12-00670.1
Given the importance of low-frequency SSHA variations, we then isolated the signals at longer than annual time scales (C5, C6, C7, and C8) and applied the Hilbert transform to them. Figure 7 shows the Hilbert spectrum of the selected low-frequency components. After removing the intraseasonal and annual signals, the progression of IE becomes smoother with a reversal around the early 2000s (Fig. 7a). This suggests that such a reversal is mainly influenced by the low-frequency signals, and the high-frequency signals only produce fluctuations along this trend. Therefore, the importance of the low-frequency signals and their contributions to the total SSHA variability should be especially considered in an analysis of ongoing processes in the North Atlantic from the 1990s to the 2000s. Maximum amplitudes appear in the frequency range of 0.05–0.1 cpy, corresponding to signals with periods between 10 and 20 years (Figs. 7b,c). Furthermore, there are several local maxima of amplitude existing in 0.1–1 cpy frequency range, suggesting more complex interannual components of the sea level variations in the SPG. However, these maxima were generally shifted toward smaller frequency bands between 0.1 and 0.2 cpy (periods of 5–10 yr) in the late 2000s. The very low-frequency component decomposed from the SSHA may be spurious given the limited length of data record and therefore should be interpreted with caution. Moreover, sudden drops of the IE in 1995/96 and 2009/10 (Fig. 7a) were found to be concomitant with drops in the wintertime NAO index, which indicates the close relationship between the SSHA and the sign changes of the NAO index that has been previously observed (Esselborn and Eden 2001; Volkov and van Aken 2003).
Similar to Fig. 6, but only applying the Hilbert transforms to the selected IMFs (C5–C8). (a) The instantaneous energy density level (IE; in cm2), (b) the Hilbert spectrum, and (c) marginal spectral density (MSD) of each frequency (in cm).
Citation: Journal of Climate 29, 13; 10.1175/JCLI-D-12-00670.1
4. Possible causes of the variability in the SPG
The total SSHA can be split into two components including contributions from steric effects and variations in bottom pressure, respectively (Gill and Niiler 1973). We first tested the steric effects in the subpolar region. To make the steric height anomalies 
The altimeter SSHA and steric height anomaly calculated from temperature and salinity observations in the SPG.
Citation: Journal of Climate 29, 13; 10.1175/JCLI-D-12-00670.1
Direct measurements of the bottom pressure anomalies are very scarce and the data are usually from numerical modeling. The bottom pressure anomalies have amplitudes of less than 1 cm over most parts of the deep ocean and of several centimeters over shallow boundary regions at high latitudes and short time scales (Ponte 1999; Ferry et al. 2000; Vinogradova et al. 2007; Bingham and Hughes 2008). Quinn and Ponte (2012) demonstrated coherence between ocean bottom pressure and sea level anomalies at middle and high latitudes at periods less than 100 days based on observations from the Gravity Recovery and Climate Experiment (GRACE) and altimetry. On longer time scales, the sea level and bottom pressure variability are essentially different with the former prevailing (Vinogradova et al. 2007; Ivchenko et al. 2008; Stammer et al. 2013). Jayne et al. (2003) found that combining altimetry observations with satellite measurements of the time-varying bottom pressure improves the ocean heat storage estimates well on time scales of the annual cycle and shorter. However, in the SPG, inclusion of bottom pressure anomalies contributed to a reduction in RMS error of about 0.05 GJ m−2 at semiannual and annual cycles and in the linear trends [see Figs. 4–6 in Jayne et al. (2003)]. Piecuch et al. (2013) demonstrated that considerable sea level variance could be explained by bottom pressure variance at interannual scales in several regions by using the newly updated GRACE data along with altimetry. Nevertheless, their results identified the northeastern corner of the North Atlantic as the region that bottom pressure is important at interannual scales [see Fig. 5 in Piecuch et al. (2013)].
We are then concerned with the sea level variations induced by changes in heat exchange between ocean and atmosphere and by advective processes (i.e., changes at annual and longer time scales). Recall that the SSHA variability in the SPG is dominated by annual cycles and the long-term increasing trend, which can explain 50% and 30% of the total SSHA variance, respectively. At these dominant time scales, the SSHA and heat content anomaly estimated from the temperature are strongly correlated with correlation coefficients of 0.89 for the annual cycles and 0.97 for the long-term trends.
We obtained the total heat content anomaly, 
Figure 9 illustrates the progression of signals with increasing periods that are decomposed from the spatially averaged 
The IMFs and the residuals decomposed from the heat content anomalies estimated from surface heat flux (gray dotted line; right y axis) and the SSHA (solid black line; left y axis). The depth-integrated heat content anomaly (dashed black line; left y axis) is also shown for comparison. Units are GJ m−2.
Citation: Journal of Climate 29, 13; 10.1175/JCLI-D-12-00670.1
Figure 9 also presents heat content anomalies estimated from the temperature for comparison (only C6 in this time series is not significant). Heat content anomalies in the SPG estimated from the temperature and the SSHA agree well with each other in terms of both amplitude and phase. The strongest correlations for these two time series exist in the long-term trends (correlation coefficient r = 0.96), the annual cycles (r = 0.84), and the decadal cycles (r = 0.76). Therefore, great care should be taken in the estimation of heat content anomalies by altimetry alone when considering interannual scales. Moreover, Fig. 9 shows that the discrepancies between the two heat content anomalies have reduced from the 1990s to the 2000s at the decadal time scale (C5) and in the residuals (R), which may relate to the concomitant warming observed in the northern North Atlantic (e.g., Häkkinen and Rhines 2009; Häkkinen et al. 2011).
We then examined the relationship between the SSHA variability in the SPG and possible underlying processes at longer than annual time scales. These low-frequency changes are typically attributed to the variability in the deep ocean convection (DOC) processes in the Labrador Sea (Häkkinen and Rhines 2004; Bersch et al. 2007; Rhein et al. 2011) and large-scale AMOC variations (Zhang 2008; Lohmann et al. 2009a,b; Robson et al. 2012; Li et al. 2012). The DOC is a direct form of heat exchange between atmosphere and the intermediate to deep ocean. A large amount of heat is released to the atmosphere during strong DOC years from the ocean, resulting in a reduction of heat content in the upper to intermediate water column. The associated interannual trends reflect the atmospheric winter conditions persisting for more than a year (Lazier et al. 2002). Therefore, we chose the depth-integrated heat content of the central Labrador as a DOC index for the analysis of the interannual and longer time scale variability. The heat content has increased almost steadily since the mid-1990s. This is closely associated with the interannual changes of the DOC: after a period of enhanced deep convection with production of large volume of the classic Labrador Seawater (CLSW) in the early 1990s, there were some years of ventilation of the CLSW and formation of the upper Labrador Seawater (ULSW) due to reduced DOC intensity and activity in the 2000s (Kieke et al. 2007; Rhein et al. 2011). The SPG index is defined as the SSHA of the central Labrador Sea (see Häkkinen and Rhines 2004; Böning et al. 2006). The AMOC transport index used here is the meridional transport at 26.5°N (Cunningham et al. 2007) from the RAPID-WATCH MOC monitoring project (http://www.rapid.ac.uk/rapidmoc/). We considered the NAO forcing as a possible driving force in the low-frequency changes in the North Atlantic (e.g., Hurrell 1995; Marshall et al. 2001). Since the ocean signal reflects a time integration of the atmospheric forcing, for example, through mixed layer “memory” and Rossby wave propagation (Curry and McCartney 2001), the cumulative NAO index (CumNAO) was adopted instead, following Li et al. (2012), by integrating monthly NAO index over time (https://climatedataguide.ucar.edu/climate-data/hurrell-north-atlantic-oscillation-nao-index-station-based), which may more accurately represent the cumulative effects of oceanic signal variations. Since the NAO index includes positive or negative phases over time, this time integral could also provide a clear view of any accumulations as well as any shifts in time. A period of persistent positive NAO index becomes an increasing slope in the CumNAO index, and vice versa. The CumNAO can therefore highlight the shifts in the atmospheric conditions compared to the normal NAO index. For instance, a sign change of the NAO index that happened between 1995 and 1996 corresponds to a peak in the CumNAO. In addition, using different time origin of integration does not change the shape of the cumulative NAO index, but instead introduces a shift in magnitude for every point.
Figure 10 shows the close relationship between DOC, the SPG, the AMOC variability, and the cumulative NAO forcing especially at interannual to decadal time scales. The CumNAO index matches with the DOC and the SPG indices well and leads the AMOC variability with some time lags. A persistent high NAO index in the early 1990s lead to consecutive winters of strong DOC events accompanied by a reduction of heat content and an enhancement of the SPG circulation. After mid-1990s, a steady warming of the central Labrador Sea was observed with the exception of the periods of 1999 to 2003 and 2008 to 2010—two periods of observed large production of CLSW (Yashayaev 2007; Våge et al. 2009; Yashayaev and Loder 2009). The AMOC variability follows the changes in DOC and the SPG with a time lag of approximately 1 to 2 years, which agrees with previous findings based on numerical simulations (Zhang 2008, 2010). Therefore, at interannual to decadal time scales, the SSHA variations in the North Atlantic closely represent the AMOC variability driven by the NAO forcing.
Close relationships among the heat content of the central Labrador Sea (dashed black line), the sea surface height anomaly of the central Labrador Sea (solid black line), the meridional overturning circulation transport at 
Citation: Journal of Climate 29, 13; 10.1175/JCLI-D-12-00670.1
5. Concluding remarks
This study has presented an analysis of 18-yr SSHA records in the mid- to high-latitude North Atlantic that examined the characteristic features of sea level variations. A dipole pattern, centered between the SPG and the GS path, appears in both the linear trends and the STD of the SSHA, indicating different modes of variability in the two regions. A simple increase in the SSHA could represent most changes in the SPG throughout the entire time span of the study. This increase corresponds to a continuous weakening of surface circulation. In contrast, the linear trend along the GS path is not representative of the overall sea level changes that are instead associated with greater variance. The separation of the SSHA time series into components with various time scales and the long-term trend showed more details about the contrast between the SPG and the GS regions. It is worth mentioning, however, that the long-term trend may have a contribution from aliasing due to the limited length of data record. The SSHA variability along the GS is dominated by intraseasonal to annual fluctuations, which together could account for 70% of the total sea level variance. These intense, high-frequency variations in the SSHA time series would effectively mask the low-frequency signals of greater interest (e.g., Lorbacher et al. 2010). In the SPG, the long-term trend and the annual signal accounts for 50% and 30% of the total variance, respectively. The long-term trend might also be a part of longer time scale variability that cannot be resolved in the limited altimeter record length of 18 years. Therefore, the dominance of the long-term trend in the SPG sea level changes may indicate the critical role of low-frequency processes in the North Atlantic. Furthermore, by applying the Hilbert transforms to the IMF components of the SPG sea level, signals at decadal and longer time scales were observed in addition to the well-known strong annual cycle, confirming the important role of the low-frequency components in the SPG variability. In the regions beyond the SPG and the GS, the SSHA variations are smaller and are dominated by annual variations with an overall increasing trend.
Although the long-term trends have similar pattern of the linear trends, the EEMD-derived trends contain more time-dependent information. This observation was not possible using most traditional time series analysis methods that assume stationary changes. In the SPG, a sea level rise rate reduction was found around the early 2000s, which is concomitant with the reduced amplitudes in the low-frequency signals. Spectral analysis indicated coinciding changes in the energy with the amplitude changes, which increases after the early 2000s following a period of energy loss from the system since the early 1990s. Such reversal of energy loss was influenced by the low-frequency SSHA signals (removal of the high-frequency signals did not change the result). The timing of such changes may correspond to a “shift in the system” around 2002, which was also found in the observations of the LSW formation and the first principal component of the altimetry-derived sea level in the northern North Atlantic (Sarafanov et al. 2012, and references therein). These changes in the SSHA from the 1990s to the 2000s might suggest a new cycle of SPG intensification as observed in part of its boundary currents from repeat hydrographic sections in the early 2000s (Han et al. 2010; Daniault et al. 2011). Actually, deep-water transport strengthened in opposition to its surface layer counterpart in the 1990s (Dengler et al. 2006; Sarafanov et al. 2010). Moreover, after years of weak or absent DOC activities, the Labrador Sea experienced a strong DOC event in the winter of 2007/08 even without a clear phase of preconditioning (Våge et al. 2009). Rapid development of convection under intensified atmospheric forcing greatly deepened the mixed layer to more than 1600 m, disrupting steady warming of the intermediate depth waters since 1994 (Yashayaev and Loder 2009). All of these recent observations cast questions about the future of the SPG under a warming scenario and its relationship to both external atmospheric forcing as well as internal oceanic dynamics. This requires further investigation on understanding the low-frequency variability in the subpolar North Atlantic.
This study also showed that, in the factors responsible for the SPG sea level variability, advection is important at interannual to decadal time scales while the air–sea heat flux is not negligible at annual time scale. The result is in line with previous studies on the interannual sea level anomaly in the North Atlantic (Reverdin et al. 1999; Häkkinen 2001; Esselborn and Eden 2001; Verbrugge and Reverdin 2003; Volkov and van Aken 2003). Moreover, it has been suggested that air–sea heat flux explains most of the sea level variance at annual scale in the northeastern North Atlantic (Ferry et al. 2000; Volkov and van Aken 2003). This, therefore, suggests different underlying mechanisms accounted for the SSHA in the western and eastern subpolar North Atlantic (Herbaut and Houssais 2009; Li et al. 2012). Such importance has been studied in the GS given that advection plays a more determinant role in regulating the heat content than air–sea fluxes on interannual-to-decadal time scales (Dong and Kelly 2004; Kelly and Dong 2004). Based on results from numerical simulations, Häkkinen (1999, 2001) attributed the low-frequency sea level changes predominately to overturning changes. The close relationship among DOC, the SPG, and the AMOC especially at interannual time scales was also presented here by comparing adequate indices of each process. Matches between these crucial physical processes, as well as their relationship to the cumulative NAO forcing, may explain changes in the advection and thus account for low-frequency variability in the SSHA in the North Atlantic. Since the AMOC index used here is available for a very limited time period, we tested the role of the AMOC in another way to see how low-frequency SSHA variability relates to AMOC variations. Li et al. (2012) suggested the potential of using low-frequency SSHA to represent changes in AMOC variations between different latitudes based on analysis of two regional SSHA covering the SPG and the GS respectively. Here, we extended their works by dividing the study domain into three adjacent regions in order to examine the meridional propagation of SSHA variations (Fig. 11). For consistency, we considered only significant low-frequency components (period >1 yr) in these three regions (regions 1, 2, and 3). Unless otherwise noted, the SSHA variations discussed hereinafter represent the low-frequency changes. The propagation of SSHA variations from high- to midlatitude regions was observed with certain time lags between two adjacent regions. The SSHA variations of region 2 lag region 1 by about 2 years with a significant correlation of 0.52. Toward the south, there is a 5-yr lead between SSHA variations in regions 2 and 3 with a significant correlation of −0.36. The propagation velocity is estimated by V = L/T, where L is the meridional distance between 50° and 30°N (about 2297 km) and T is 5 yr, which gives a propagation velocity of approximately 1.46 cm s−1. This finding is consistent with modeled climatological mean meridional velocity of 1.5 cm s−1 speed at which AMOC variations at high latitudes propagate to the midlatitudes (Curry et al. 1998; Zhang 2010) following interior pathways observed recently (Bower et al. 2009). Therefore, it confirms that the low-frequency SSHA variations in the SPG and along the GS are mostly related to AMOC variations, which might be a part of multidecadal variability as suggested in numerical models (Zhang 2008; Mahajan et al. 2011). Moreover, Chen and Tung (2014) indicate that increasing heat content in the Atlantic Ocean effectively contributes to ongoing surface warming slowdown (“climate hiatus”). The authors further propose the underlying mechanism of corresponding climate shifts in both salinity and ocean heat content in the subpolar North Atlantic with involvement of the AMOC variations. Therefore, characteristic modes of sea level variability especially in the subpolar North Atlantic might have further implications for the hiatus and its relationship to the AMOC. Observational verification is of great importance when a longer altimeter record becomes available.
Domains of areas 1, 2, and 3 used to spatially average the SSHA and examine the lagged correlation between each other. Arrows indicate possible propagation pathway of AMOC variations. The 1000-, 2000-, and 3000-m isobaths are shown in solid gray lines.
Citation: Journal of Climate 29, 13; 10.1175/JCLI-D-12-00670.1
It is worth mentioning that in most areas of the SPG, surface heat flux is positively correlated with the heat content, which indicates that surface flux introduces variations to the heat content (and thus the sea level anomaly) in the upper oceans. In contrast, the situation in the GS is different and has opposite values of 
Acknowledgments
The authors thank the editor and two anonymous reviewers for their helpful comments that greatly improved this manuscript. The authors also thank Autumn Kidwell for editorial assistance in English. This research was partially supported by NASA Physical Oceanography Program, NASA EPSCoR Program, NASA Space Grant, and NOAA Sea Grant.
APPENDIX
Hilbert–Huang Transform
Empirical mode decomposition
Figure A1 gives an example of the progression of the IMFs with high to low frequency (C1–C6) and the residual (R) decomposed from a SSHA time series at a location within the subpolar gyre (60°N, 42°W).
The IMFs (C1–C6) and the residual (R) decomposed from a SSHA time series (input) at 60°N, 42°W. Units are cm.
Citation: Journal of Climate 29, 13; 10.1175/JCLI-D-12-00670.1
Statistical significance test of IMFs
Wu and Huang (2005) studied the characteristics of IMFs from uniformly distributed white noise, and gave the empirical relationship between the energy density and the mean period. This leads to the establishment of the energy distribution function for each IMF and the spread function of the energy distribution for various percentiles. Based on the characteristics of the white noise, the information content of IMF components from a noisy dataset derived from EMD can be objectively determined relative to the white noise statistics. Therefore, statistical significance of the IMFs is based on the distribution of energy as a function of mean period of the IMF relative to that of pure white noise. This method allows one to differentiate true signals from components of noise with any selected statistical significance level (95%, 99%) (Wu and Huang 2004, 2005; Huang and Wu 2008). Insignificant IMFs are interpreted as noise inherent to the original signal. Significance testing shows that all IMF components shown in Fig. A1 are significant at the 95% significance level.
An alternative way of testing the significance of an IMF of data is implemented based on the Monte Carlo method (Coughlin and Tung 2004, 2005; Flandrin et al. 2004, 2005). The special characteristics of data, such as lagged autocorrelation, are used to make a null hypothesis for the type of noise process for data. One can generate a number of samples of the noise series with the same length as that of data and decompose them into a large number of sets of IMFs. By comparing the energy or variance of each IMF from data with that from random samples, one can tell whether an IMF contains any statistically significant information. A minor drawback of this approach is the demand of large computational resources to obtain an accurate distribution of energy of an IMF when the size of data to be analyzed is very large (Wu and Huang 2005).
Hilbert transforms of IMFs
This enables us to represent the amplitude and the instantaneous frequency as functions of time in a three-dimensional plot, in which each point of amplitude corresponds to a pair of [t, 
REFERENCES
Antonov, J. I., S. Levitus, and T. P. Boyer, 2002: Steric sea level variations during 1957–1994: Importance of salinity. J. Geophys. Res., 107 (C12), 8013, doi:10.1029/2001JC000964.
Beckley, B. D., N. P. Zelensky, S. A. Holmes, F. G. Lemoine, R. D. Ray, G. T. Mitchum, S. D. Desai, and S. T. Brown, 2010: Assessment of the Jason-2 extension to the TOPEX/Poseidon, Jason-1 sea-surface height time series for global mean sea level monitoring. Mar. Geod., 33, 447–471, doi:10.1080/01490419.2010.491029.
Belkin, I. M., 2004: Propagation of the “Great Salinity Anomaly” of the 1990s around the northern North Atlantic. Geophys. Res. Lett., 31, L08306, doi:10.1029/2003GL019334.
Bersch, M., I. Yashayaev, and K. Koltermann, 2007: Recent changes of the thermohaline circulation in the subpolar North Atlantic. Ocean Dyn., 57, 223–235, doi:10.1007/s10236-007-0104-7.
Bindschadler, R., 2006: Hitting the ice sheets where it hurts. Science, 311, 1720–1721, doi:10.1126/science.1125226.
Bingham, R. J., and C. W. Hughes, 2008: The relationship between sea-level and bottom pressure variability in an eddy permitting ocean model. Geophys. Res. Lett., 35, L03602, doi:10.1029/2007GL032662.
Böning, C. W., M. Scheinert, J. Dengg, A. Biastoch, and A. Funk, 2006: Decadal variability of subpolar gyre transport and its reverberation in the North Atlantic overturning. Geophys. Res. Lett., 33, L21S01, doi:10.1029/2006GL026906.
Bower, A. S., M. S. Lozier, S. F. Gary, and C. W. Böning, 2009: Interior pathways of the North Atlantic meridional overturning circulation. Nature, 459, 243–248, doi:10.1038/nature07979.
Cazenave, A., and W. Llovel, 2010: Contemporary sea level rise. Annu. Rev. Mar. Sci., 2, 145–173, doi:10.1146/annurev-marine-120308-081105.
Chambers, D. P., B. D. Tapley, and R. H. Stewart, 1997: Long-period ocean heat storage rates and basin-scale heat fluxes from TOPEX. J. Geophys. Res., 102, 10 525–10 533, doi:10.1029/96JC03644.
Chelton, D. B., M. G. Schlax, and R. M. Samelson, 2011: Global observations of nonlinear mesoscale eddies. Prog. Oceanogr., 91, 167–216, doi:10.1016/j.pocean.2011.01.002.
Chen, X., and K. K. Tung, 2014: Varying planetary heat sink led to global-warming slowdown and acceleration. Science, 345, 897–903, doi:10.1126/science.1254937.
Clark, P. U., N. G. Pisias, T. F. Stocker, and A. J. Weaver, 2002: The role of the thermohaline circulation in abrupt climate change. Nature, 415, 863–869, doi:10.1038/415863a.
Coughlin, K. T., and K. K. Tung, 2004: 11-year solar cycle in the stratosphere extracted by the empirical mode decomposition method. Adv. Space Res., 34, 323–329, doi:10.1016/j.asr.2003.02.045.
Coughlin, K. T., and K. K. Tung, 2005: Empirical mode decomposition and climate variability. Hilbert–Huang Transform: Introduction and Applications, N. E. Huang and S. S. P. Shen, Eds., World Scientific, 149–165.
Cunningham, S. A., and Coauthors, 2007: Temporal variability of the Atlantic meridional overturning circulation at 26.5°N. Science, 317, 935–938, doi:10.1126/science.1141304.
Curry, R. G., and M. S. McCartney, 2001: Ocean gyre circulation changes associated with the North Atlantic Oscillation. J. Phys. Oceanogr., 31, 3374–3400, doi:10.1175/1520-0485(2001)031<3374:OGCCAW>2.0.CO;2.
Curry, R. G., M. S. McCartney, and T. M. Joyce, 1998: Oceanic transport of subpolar climate signals to mid-depth subtropical waters. Nature, 391, 575–577, doi:10.1038/35356.
Daniault, N., H. Mercier, and P. Lherminier, 2011: The 1992–2009 transport variability of the East Greenland–Irminger Current at 60°N. Geophys. Res. Lett., 38, L07601, doi:10.1029/2011GL046863.
Dengler, M., J. Fischer, F. A. Schott, and R. Zantopp, 2006: Deep Labrador Current and its variability in 1996–2005. Geophys. Res. Lett., 33, L21S06, doi:10.1029/2006GL026702.
Dickson, R., J. Lazier, J. Meincke, P. Rhines, and J. Swift, 1996: Long-term coordinated changes in the convective activity of the North Atlantic. Prog. Oceanogr., 38, 241–295, doi:10.1016/S0079-6611(97)00002-5.
Dong, S., and K. A. Kelly, 2004: Heat budget in the Gulf Stream region: The importance of heat storage and advection. J. Phys. Oceanogr., 34, 1214–1231, doi:10.1175/1520-0485(2004)034<1214:HBITGS>2.0.CO;2.
Esselborn, S., and C. Eden, 2001: Sea surface height changes in the North Atlantic Ocean related to the North Atlantic Oscillation. Geophys. Res. Lett., 28, 3473–3476, doi:10.1029/2001GL012863.
Ferry, N., G. Reverdin, and A. Oschlies, 2000: Seasonal sea surface height variability in the North Atlantic Ocean. J. Geophys. Res., 105, 6307–6326, doi:10.1029/1999JC900296.
Flandrin, P., G. Rilling, and P. Gonçalves, 2004: Empirical mode decomposition as a filter bank. IEEE Signal Process. Lett., 11, 112–114, doi:10.1109/LSP.2003.821662.
Flandrin, P., P. Gonçalves, and G. Rilling, 2005: EMD equivalent filter bank, from interpretation to applications. Hilbert–Huang Transform and Its Applications, N. E. Huang and S. Shen, Eds., World Scientific, 57–74.
Gill, A. E., and P. Niiler, 1973: The theory of the seasonal variability in the ocean. Deep-Sea Res., 20, 141–177, doi:10.1016/0011-7471(73)90049-1.
Häkkinen, S., 1999: Variability of the simulated meridional heat transport in the North Atlantic for the period 1951–1993. J. Geophys. Res., 104, 10 991–11 007, doi:10.1029/1999JC900034.
Häkkinen, S., 2001: Variability in sea surface height: A qualitative measure for the meridional overturning in the North Atlantic. J. Geophys. Res., 106, 13 837–13 848, doi:10.1029/1999JC000155.
Häkkinen, S., and P. B. Rhines, 2004: Decline of subpolar North Atlantic circulation during the 1990s. Science, 304, 555–559, doi:10.1126/science.1094917.
Häkkinen, S., and P. B. Rhines, 2009: Shifting surface currents in the northern North Atlantic Ocean. J. Geophys. Res., 114, C04005, doi:10.1029/2008JC004883.
Häkkinen, S., P. B. Rhines, and D. L. Worthen, 2011: Warm and saline events embedded in the meridional circulation of the northern North Atlantic. J. Geophys. Res., 116, C03006, doi:10.1029/2010JC006275.
Han, G., K. Ohashi, N. Chen, P. G. Myers, N. Nunes, and J. Fischer, 2010: Decline and partial rebound of the Labrador Current 1993–2004: Monitoring ocean currents from altimetric and conductivity–temperature–depth data. J. Geophys. Res., 115, C12012, doi:10.1029/2009JC006091.
Hátún, H., A. B. Sandø, H. Drange, B. Hansen, and H. Valdimarsson, 2005: Influence of the Atlantic subpolar gyre on the thermohaline circulation. Science, 309, 1841–1844, doi:10.1126/science.1114777.
Herbaut, C., and M. Houssais, 2009: Response of the eastern North Atlantic subpolar gyre to the North Atlantic Oscillation. Geophys. Res. Lett., 36, L17607, doi:10.1029/2009GL039090.
Holland, D. M., R. H. Thomas, B. de Young, M. H. Ribergaard, and B. Lyberth, 2008: Acceleration of Jakobshavn Isbrae triggered by warm subsurface ocean waters. Nat. Geosci., 1, 659–664, doi:10.1038/ngeo316.
Huang, N. E., and Z. Wu, 2008: A review on Hilbert–Huang transform: Method and its applications to geophysical studies. Rev. Geophys., 46, RG2006, doi:10.1029/2007RG000228.
Huang, N. E., and Coauthors, 1998: The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis. Proc. Roy. Soc. London, 454A, 903–995, doi:10.1098/rspa.1998.0193.
Hurrell, J. W., 1995: Decadal trends in the North Atlantic Oscillation: Regional temperatures and precipitation. Science, 269, 676–679, doi:10.1126/science.269.5224.676.
Ingleby, B., and M. Huddleston, 2007: Quality control of ocean temperature and salinity profiles—Historical and real-time data. J. Mar. Syst., 65, 158–175, doi:10.1016/j.jmarsys.2005.11.019.
Ivchenko, V. O., S. Danilov, D. Sidorenko, J. Schröter, M. Wenzel, and D. L. Aleynik, 2008: Steric height variability in the northern Atlantic on seasonal and interannual scales. J. Geophys. Res., 113, C11007, doi:10.1029/2008JC004836.
Jayne, S. R., J. M. Wahr, and F. O. Bryan, 2003: Observing ocean heat content using satellite gravity and altimetry. J. Geophys. Res., 108, 3031, doi:10.1029/2002JC001619.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471, doi:10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.
Kelly, K. A., 2004: The relationship between oceanic heat transport and surface fluxes in the western North Pacific: 1970–2000. J. Climate, 17, 573–588, doi:10.1175/1520-0442(2004)017<0573:TRBOHT>2.0.CO;2.
Kelly, K. A., and S. Dong, 2004: The relationship of western boundary current heat transport and storage to midlatitude ocean–atmosphere interaction. Earth’s Climate: The Ocean–Atmosphere Interaction, Geophys. Monogr., Vol. 147, Amer. Geophys. Union, 347–363.
Kieke, D., M. Rhein, L. Stramma, W. M. Smethie, J. L. Bullister, and D. A. LeBel, 2007: Changes in the pool of Labrador Sea Water in the subpolar North Atlantic. Geophys. Res. Lett., 34, L06605, doi:10.1029/2006GL028959.
Latif, M., C. Böning, J. Willebrand, A. Biastoch, J. Dengg, N. Keenlyside, U. Schweckendiek, and G. Madec, 2006: Is the thermohaline circulation changing? J. Climate, 19, 4631–4637, doi:10.1175/JCLI3876.1.
Lazier, J., R. Hendry, A. Clarke, I. Yashayaev, and P. Rhines, 2002: Convection and restratification in the Labrador Sea, 1990–2000. Deep-Sea Res. I, 49, 1819–1835, doi:10.1016/S0967-0637(02)00064-X.
Le Traon, P. Y., and G. Dibarboure, 1999: Mesoscale mapping capabilities of multiple-satellite altimeter missions. J. Atmos. Oceanic Technol., 16, 1208–1223, doi:10.1175/1520-0426(1999)016<1208:MMCOMS>2.0.CO;2.
Li, F., Y.-H. Jo, W. T. Liu, and X.-H. Yan, 2012: A dipole pattern of the sea surface height anomaly in the North Atlantic: 1990s–2000s. Geophys. Res. Lett., 39, L15604, doi:10.1029/2012GL052556.
Lilly, J. M., P. B. Rhines, F. Schott, K. Lavender, J. Lazier, U. Send, and E. D’Asaro, 2003: Observations of the Labrador Sea eddy field. Prog. Oceanogr., 59, 75–176, doi:10.1016/j.pocean.2003.08.013.
Lohmann, K., H. Drange, and M. Bentsen, 2009a: A possible mechanism for the strong weakening of the North Atlantic subpolar gyre in the mid-1990s. Geophys. Res. Lett., 36, L15602, doi:10.1029/2009GL039166.
Lohmann, K., H. Drange, and M. Bentsen, 2009b: Response of the North Atlantic subpolar gyre to persistent North Atlantic oscillation like forcing. Climate Dyn., 32, 273–285, doi:10.1007/s00382-008-0467-6.
Lorbacher, K., J. Dengg, C. W. Boning, and A. Biastoch, 2010: Regional patterns of sea level change related to interannual variability and multidecadal trends in the Atlantic meridional overturning circulation. J. Climate, 23, 4243–4254, doi:10.1175/2010JCLI3341.1.
Lynch-Stieglitz, J., and Coauthors, 2007: Atlantic meridional overturning circulation during the Last Glacial Maximum. Science, 316, 66–69, doi:10.1126/science.1137127.
Lynch-Stieglitz, J., M. W. Schmidt, and W. B. Curry, 2011: Evidence from the Florida Straits for Younger Dryas ocean circulation changes. Paleoceanography, 26, PA1205, doi:10.1029/2010PA002032.
Mahajan, S., R. Zhang, T. Delworth, S. Zhang, A. J. Rosati, and Y.-S. Chang, 2011: Predicting Atlantic meridional overturning (AMOC) variations using subsurface and surface fingerprints. Deep-Sea Res. II, 58, 1895–1903, doi:10.1016/j.dsr2.2010.10.067.
Marshall, J., and Coauthors, 1998: The Labrador Sea Deep Convection Experiment. Bull. Amer. Meteor. Soc., 79, 2033–2058, doi:10.1175/1520-0477(1998)079<2033:TLSDCE>2.0.CO;2.
Marshall, J., and Coauthors, 2001: North Atlantic climate variability: Phenomena, impacts and mechanisms. Int. J. Climatol., 21, 1863–1898, doi:10.1002/joc.693.
McManus, J. F., R. Francois, J.-M. Gherardi, L. D. Keigwin, and S. Brown-Leger, 2004: Collapse and rapid resumption of Atlantic meridional circulation linked to deglacial climate changes. Nature, 428, 834–837, doi:10.1038/nature02494.
Piecuch, C. G., K. J. Quinn, and R. M. Ponte, 2013: Satellite-derived interannual ocean bottom pressure variability and its relation to sea level. Geophys. Res. Lett., 40, 3106–3110, doi:10.1002/grl.50549.
Ponte, R. M., 1999: A preliminary model study of the large-scale seasonal cycle in bottom pressure over the global ocean. J. Geophys. Res., 104 (C1), 1289–1300, doi:10.1029/1998JC900028.
Quinn, K. J., and R. M. Ponte, 2012: High frequency barotropic ocean variability observed by GRACE and satellite altimetry. Geophys. Res. Lett., 39, L07603, doi:10.1029/2012GL051301.
Reverdin, G., N. Verbrugge, and H. Valdimarsson, 1999: Upper ocean variability between Iceland and Newfoundland, 1993–1998. J. Geophys. Res., 104, 29 599–29 611, doi:10.1029/1999JC900182.
Rhein, M., and Coauthors, 2011: Deep water formation, the subpolar gyre, and the meridional overturning circulation in the subpolar North Atlantic. Deep-Sea Res. II, 58, 1819–1832, doi:10.1016/j.dsr2.2010.10.061.
Robson, J., R. Sutton, K. Lohmann, D. Smith, and M. D. Palmer, 2012: Causes of the rapid warming of the North Atlantic Ocean in the mid-1990s. J. Climate, 25, 4116–4134, doi:10.1175/JCLI-D-11-00443.1.
Sarafanov, A., 2009: On the effect of the North Atlantic Oscillation on temperature and salinity of the subpolar North Atlantic intermediate and deep waters. ICES J. Mar. Sci., 66, 1448–1454, doi:10.1093/icesjms/fsp094.
Sarafanov, A., A. Falina, A. Sokov, and A. Demidov, 2008: Intense warming and salinification of intermediate waters of southern origin in the eastern subpolar North Atlantic in the 1990s to mid-2000s. J. Geophys. Res., 113, C12022, doi:10.1029/2008JC004975.
Sarafanov, A., A. Falina, P. Lherminier, H. Mercier, A. Sokov, and C. Gourcuff, 2010: Assessing decadal changes in the Deep Western Boundary Current absolute transport southeast of Cape Farewell, Greenland, from hydrography and altimetry. J. Geophys. Res., 115, C11003, doi:10.1029/2009JC005811.
Sarafanov, A., and Coauthors, 2012: Mean full-depth summer circulation and transports at the northern periphery of the Atlantic Ocean in the 2000s. J. Geophys. Res., 117, C01014, doi:10.1029/2011JC007572.
Stammer, D., A. Cazenave, R. M. Ponte, and M. E. Tamisiea, 2013: Causes for contemporary regional sea level changes. Annu. Rev. Mar. Sci., 5, 21–46, doi:10.1146/annurev-marine-121211-172406.
Våge, K., R. S. Pickart, G. W. K. Moore, and M. H. Ribergaard, 2008: Winter mixed layer development in the central Irminger Sea: The effect of strong, intermittent wind events. J. Phys. Oceanogr., 38, 541–565, doi:10.1175/2007JPO3678.1.
Våge, K., and Coauthors, 2009: Surprising return of deep convection to the subpolar North Atlantic Ocean in winter 2007–2008. Nat. Geosci., 2, 67–72, doi:10.1038/ngeo382.
van Aken, H. M., M. Femke de Jong, and I. Yashayaev, 2011: Decadal and multi-decadal variability of Labrador Sea Water in the north-western North Atlantic Ocean derived from tracer distributions: Heat budget, ventilation, and advection. Deep-Sea Res. I, 58, 505–523, doi:10.1016/j.dsr.2011.02.008.
Verbrugge, N., and G. Reverdin, 2003: Contribution of horizontal advection to the interannual variability of sea surface temperature in the North Atlantic. J. Phys. Oceanogr., 33, 964–978, doi:10.1175/1520-0485(2003)033<0964:COHATT>2.0.CO;2.
Vinogradova, N. T., R. M. Ponte, and D. Stammer, 2007: Relation between sea level and bottom pressure and the vertical dependence of oceanic variability. Geophys. Res. Lett., 34, L03608, doi:10.1029/2006GL028588.
Volkov, D. L., 2005: Interannual variability of the altimetry-derived eddy field and surface circulation in the extratropical North Atlantic Ocean in 1993–2001. J. Phys. Oceanogr., 35, 405–426, doi:10.1175/JPO2683.1.
Volkov, D. L., and H. M. van Aken, 2003: Annual and interannual variability of sea level in the northern North Atlantic Ocean. J. Geophys. Res., 108, 3204, doi:10.1029/2002JC001459.
Wu, Z., and N. E. Huang, 2004: A study of the characteristics of white noise using the empirical mode decomposition method. Proc. Roy. Soc. London, 460A, 1597–1611, doi:10.1098/rspa.2003.1221.
Wu, Z., and N. E. Huang, 2005: Statistical significance test of intrinsic mode functions. Hilbert–Huang Transform and Its Applications, N. E. Huang and S. S. Shen, Eds., World Scientific, 107–127.
Wu, Z., and N. E. Huang, 2009: Ensemble empirical mode decomposition: A noise-assisted data analysis method. Adv. Adapt. Data Anal., 1, 1–41, doi:10.1142/S1793536909000047.
Yashayaev, I., 2007: Hydrographic changes in the Labrador Sea, 1960–2005. Prog. Oceanogr., 73, 242–276, doi:10.1016/j.pocean.2007.04.015.
Yashayaev, I., and J. W. Loder, 2009: Enhanced production of Labrador Sea Water in 2008. Geophys. Res. Lett., 36, L01606, doi:10.1029/2008GL036162.
Zhang, R., 2008: Coherent surface-subsurface fingerprint of the Atlantic meridional overturning circulation. Geophys. Res. Lett., 35, L20705, doi:10.1029/2008GL035463.
Zhang, R., 2010: Latitudinal dependence of Atlantic meridional overturning circulation (AMOC) variations. Geophys. Res. Lett., 37, L16703, doi:10.1029/2010GL044474.
