1. Introduction
Satellite records of more than 40 years show that the Arctic summer sea ice extent minimum has decreased dramatically and the melt season is lengthening (Petty et al. 2020; Stroeve and Notz 2018). Given the positive ice–albedo feedback in the Arctic (Budyko 1969; Kashiwase et al. 2017; Sellers 1969), not only sea ice retreat but also sea ice surface melting has the potential to influence the surface energy balance and the Arctic summer sea ice minimum. According to Perovich and Polashenski (2012), sea ice surface albedo drops from high values of 0.8–0.9 to near 0.6 and even lower values after liquid water appears on the surface in spring. Previous studies have tried to utilize MO dates as a predictor of the September minimum sea ice extent (Petty et al. 2017; Wang et al. 2011). Research on MO may contribute to sea route safety, which is closely related to summer sea ice cover (Lei et al. 2015).
Sea ice surface melting is usually regarded as a response to the atmospheric state (Bliss and Anderson 2014; Drobot and Anderson 2001), not the underlying ocean, which directly impacts sea ice retreat. Specifically, sea ice MO is associated with higher surface air temperature (SAT), total column water vapor (TWV), and cloud cover, which increases downward longwave radiation (Mortin et al. 2016). The warm, moist atmosphere that favors sea ice surface melting is considered a result of atmospheric circulation (i.e., heat and moisture advection). Advection can be brought on by synoptic storms, surface pressure patterns (Horvath et al. 2021), or other meridional transport from beyond the Arctic (Boisvert et al. 2013; Crawford et al. 2018; Persson 2012; Xu et al. 2020). Existing studies help construct a clear picture describing Arctic sea ice surface melting controlled by atmospheric heat and moisture transport and subsequent local thermodynamic factors.
Previous studies have also analyzed the spatial and temporal characteristics of MO across the Arctic but tend to treat the Arctic as a whole or as separate individual seas. Liang and Su (2021) analyzed the relationship between sea ice MO in the Laptev Sea (LS) and the East Siberian Sea (ESS) in the sense of spatial averaging, which are neighboring seas on the edge of the Arctic Ocean. Also, sea ice cover in these two seas is the heaviest along the northern sea route. In this large continental shelf area, the related sea ice conditions (more or less) would probably have a significant influence on the surface energy exchange, oceanic stratification, and ecosystem activity (Lalande et al. 2009). In this study, we further investigate the spatiotemporal characteristics of MO in this region, identifying several leading modes and the driving mechanisms behind them.
2. Data and methods
The sea ice MO dataset is retrieved from the National Aeronautics and Space Administration (NASA) Cryospheric Sciences Research Portal. This dataset is based on the passive microwave (PMW) algorithm developed by Markus et al. (2009), which diagnoses the daily variation in surface brightness temperature. The physical theory behind the algorithm is that surface brightness temperature increases when liquid water appears on the sea ice surface, as the emissivity of water is larger than that of ice or snow. As in Mortin et al. (2016), Liu and Schweiger (2017), and Huang et al. (2018), this study uses the variable of early MO, which represents the first day in spring when liquid water is detected by a surface microwave signal. In this study, we focus on the yearly MO in the LS and ESS from 1979 to 2018, with a spatial resolution of ∼25 km. Note that this paper focuses on ice surface melt rather than ice basal melt (Lei et al. 2022).
The MO results from the PMW algorithm do not cover all the sea ice grids. For instance, according to the algorithm, if more than four of the surrounding pixels differ by more than one day, values of MO for that pixel are assumed to be invalid (Markus et al. 2009). For the LS and ESS, we retrieve alternative MO values based on daily SAT with a threshold of −1°C (Bliss and Anderson 2018; Rigor et al. 2000) to fill in the missing values and also correct the systematic error of alternative MO based on the original values. The SAT datasets are the International Arctic Buoy Programme/Polar Exchange at the Sea Surface (IABP/POLES) data for 1979–2004 and the Atmospheric InfraRed Sounder (AIRS) data for 2005–18. Details of the two SAT datasets and related processing procedures can be found in Liang and Su (2021).
The atmospheric variable fields are from the ERA5 dataset (Hersbach et al. 2020), which is the updated version of ERA-Interim issued by the European Centre for Medium-Range Weather Forecasts (ECMWF). The factors include monthly SAT, TWV, SLP, surface winds, winds, and geopotential height at 500 hPa in April and May for 1979–2018. Six-hourly geopotential heights at 500 hPa in April and May are also retrieved to represent storm tracks by the standard deviation of daily-mean results after Butterworth bandpass filtering of 2.5–6 days (Blackmon 1976).
Empirical orthogonal function (EOF) analyses are utilized to decompose MO around the LS and ESS, as well as to obtain Arctic Oscillation (AO) indices (Thompson and Wallace 1998) in April and May based on monthly SLP north of 60°N. The choice of this southern boundary is meant to focus on the high Arctic. Even so, the retrieved AO index is statistically correlated with that from the NOAA Climate Prediction Center (https://www.cpc.ncep.noaa.gov/products/precip/CWlink/daily_ao_index/ao.shtml), which is based on 1000-mb height anomalies poleward of 20°N (r = 0.83 and 0.84 with 99% confidence for April and May, respectively). The Barents Oscillation (BO) index (Skeie 2000) in April is the normalized series of the April SLP anomaly averaged over the Barents Sea, which could drive atmospheric moisture and heat transport through the region of the LS and ESS and influence the sea ice MO.
3. Results
a. MO modes in the LS and ESS
MO features around the Arctic have been widely explored in previous studies (Belchansky et al. 2004b; Bliss and Anderson 2018; Drobot and Anderson 2001; Markus et al. 2009; Stroeve et al. 2014). In this study, we aim specifically to reveal the spatiotemporal characteristics of MO in the LS and ESS, two neighboring seas that are part of the northern sea route along the Siberian coast.
EOF decomposition of MO in this region for 1979–2018 exhibits three distinct modes (left column in Fig. 1). The first mode (Fig. 1a1) shows that MO in the LS and ESS varies together, accounting for about 20% of the variance. As the two seas are side by side, it is expected that sea ice in the LS and ESS is usually under the same atmospheric influence. We name this first mode the LE-mode, with the letters L and E indicating LS and ESS. It is worth noticing in this mode the LS has a stronger anomaly while the southeastern part of the ESS is mostly insignificant, which indicates that the LS is under the main influence of the driving factors for this mode. The LS is surrounded by relatively more land, especially to the west and southwest. In the springtime the land warms up faster than the ocean; because the elevation around the LS is relatively low, this may have some influence on the regulation of sea ice MO by atmospheric patterns.
(a1),(b1),(c1) The first three EOF modes of MO in the LS and ESS. (a2),(b2),(c2) Composite difference of original MO between years of the positive and negative phases of the prominent first three EOF modes. Stippling denotes 95% significance by a two-sample t test. Black dots separate the LS and ESS.
Citation: Journal of Climate 36, 18; 10.1175/JCLI-D-22-0791.1
The other two leading modes are somewhat beyond our expectation. The second mode (Fig. 1b1) denotes a seesaw-like pattern in the southwest–northeast direction, while the third mode (Fig. 1c1) is in the southeast–northwest direction. This reminds us of the multiplicity of MO variation in this region. We call the second mode the L-mode, as the anomaly center of earlier MO is located in the LS. Similarly, the third mode is called the E-mode. Both the L-mode and E-mode can explain about 10% of the MO variance.
According to the maxima of the original principal components (PCs) of the first three modes, we can select the prominent mode in each year during 1979–2018. We compare the original MO difference (right column in Fig. 1) between the years of positive and negative phases of the prominent modes (Table 1). The spatial consistency with the LE-mode, L-mode, and E-mode suggests the robustness of the leading modes. Based on the prominent mode and its related PC value, we can reconstruct the spatial distribution of MO anomalies each year (Fig. S1 in the online supplemental material). Comparability between these reconstructions and the original MO anomalies (Fig. S2) also suggests the realistic aspect of the modes obtained from the EOF method. The reconstructed MO anomaly fields grouped into each phase of the prominent modes can be seen in Fig. S3.
List of years categorized as positive and negative phases of the prominent modes.
Note also that the LE-mode is prominent in most years (17 out of 40 years for the whole period), followed by the L-mode (14) and E-mode (9), which corresponds to the variance explained by the three leading modes.
A reconstructed time series based on the PC value of the prominent mode in each year (bars in Fig. 2) reveals decadal characteristics. During the 1980s, the negative phase of the LE-mode is prominent, which denotes later MO in this region and is consistent with the relatively cold atmospheric condition in the 1980s. In contrast, during the early 1990s, the positive LE-mode is dominant, which indicates earlier MO in both the LS and ESS and is consistent with the Arctic warming and positive AO in this period. After the mid-1990s, the more prominent L-mode and E-mode show up, especially the L-mode. In other words, frequent seesaw-like patterns of MO anomalies exist after the mid-1990s, which echoes the results in Liang and Su (2021). In addition, since 1999, more strongly positive L-modes and LE-modes also suggest earlier MO in the LS than in the ESS.
Reconstructed time series of the PC value of the prominent mode (1979–2018). The markers denote normalized AO and BO indices. Note that the BO index has been multiplied by −1.
Citation: Journal of Climate 36, 18; 10.1175/JCLI-D-22-0791.1
During the 1980s, the prominent LE-mode is always in the negative phase, whereas after the 1990s it is mostly in the positive phase. This means that MO in the LS and ESS has recently occurred earlier as a result of Arctic warming.
b. Driving factors of MO modes in the LS and ESS
In the section above, we introduced three MO modes in the LS and ESS. Here, we explore atmospheric driving factors responsible for these different modes. Previous studies have found that surface air temperature (SAT) and total column water vapor (TWV) are the two main factors for MO through warming effects on the surface (Belchansky et al. 2004a; Huang et al. 2018; Mortin et al. 2016). For the LS and ESS, MO occurs in May and June (Fig. S4), and atmospheric conditions in both April and May can influence MO (Liang and Su 2021). We regress large-scale thermodynamic and dynamic variables in April and May on the PCs of the LE-mode, L-mode, and E-mode, respectively (first two rows in Figs. 3–5).
The regression of (left to right) surface air temperature (SAT), total-column water vapor (TWV), sea level pressure (SLP) with surface wind, and 500-hPa geopotential height (hgt500) with the wind field in (a1)–(a4) April and (b1)–(b4) May against the LE-mode, but (c1)–(c4) simultaneously against AO in April. Magenta lines denote the boundaries of the LS and ESS. Stippling in the left two columns represents 95% confidence. Vectors and regions within yellow contours in the right two columns have 95% confidence.
Citation: Journal of Climate 36, 18; 10.1175/JCLI-D-22-0791.1
As in Fig. 3, but for the L-mode and BO index in April.
Citation: Journal of Climate 36, 18; 10.1175/JCLI-D-22-0791.1
As in Fig. 3, but for the E-mode and AO index in May.
Citation: Journal of Climate 36, 18; 10.1175/JCLI-D-22-0791.1
For the LE-mode, earlier MO in the LS and ESS is significantly related to higher SAT and TWV locally in April and May (Figs. 3a1,a2,b1,b2). In particular, SAT warming in April is even stronger than that in May, which implies more influence in April on the LE-mode. Around the Arctic we can see positive anomalies of SAT and TWV on the Siberian side. Correspondingly, we see stronger and more significant dynamic anomalies in April than in May (Figs. 3a3,a4,b3,b4). Positive AO patterns appear at the surface and in the midtroposphere in April, with a low pressure anomaly near the North Pole and northwest of the LS and ESS. Related cyclonic winds at the surface and 500 hPa bring warm, moist air masses from near the North Atlantic and Siberia toward the Siberian Arctic and cause cold advection around Greenland and the Canadian Arctic Archipelago.
Note that AO has already been supposed to influence sea ice MO in the Arctic (Drobot and Anderson 2001). Atmospheric conditions related to AO in April (Figs. 3c1–c4) resemble those in April related to the LE-mode, and the MO anomaly related to AO in April resembles the LE-mode (Fig. 6a1, spatial correlation r = 0.81 with 99% confidence). Further, the LE-mode PC and AO in April are significantly correlated (Fig. 6a2; r = 0.61 with 99% confidence), even after detrending (r = 0.57 with 99% confidence). Phases of the prominent LE-mode are mostly consistent with AO phases in April (Fig. 2). In addition, both time series have a slight positive tendency, which denotes earlier MO in recent years and is consistent with Arctic warming. Storm tracks are prominent phenomena mainly in the midlatitudes, including the North Atlantic storm track and North Pacific storm track. In the Arctic, storm activity is also significant, and previous studies have suggested a connection between storm tracks and MO in the LS and ESS. Positive AO in April is also related to increased storm activity around the Arctic (left subplot in Fig. 7), which may cause earlier MO in the LS and ESS. The above consistency provides evidence for a causal relationship between AO in April and the LE-mode.
(a1),(b1),(c1) MO regression on the AO in April, BO in April, and AO in May, respectively. (a2),(b2),(c2) Comparison between PCs of the MO modes and large-scale atmospheric circulation indices. Magenta stippling denotes 95% significance. Correlation coefficients with double asterisks denote 99% confidence, while those with a single asterisk denote 90% confidence.
Citation: Journal of Climate 36, 18; 10.1175/JCLI-D-22-0791.1
Storm-track simultaneous regression against (left) AO in April, (center) BO in April, and (right) AO in May, respectively. Stippling denotes 95% significance.
Citation: Journal of Climate 36, 18; 10.1175/JCLI-D-22-0791.1
For storm tracks, we notice that increasing storms may occur around the LS and ESS, especially in April (Fig. 8), which is consistent with earlier MO there in recent decades. Further, this increased storm activity seems to be linked with Siberia south of the LS and ESS. It is worth noticing that storm track signals are more significant in the midlatitudes than in the high latitudes. Caution should be taken when focusing only on high latitudes.
Storm-track tendency in (left) April and (right) May for 1979–2018. Stippling in limited area denotes 95% significance.
Citation: Journal of Climate 36, 18; 10.1175/JCLI-D-22-0791.1
Related to the positive L-mode (i.e., the seesaw-like pattern in the southwest–northeast direction with earlier MO in the LS), significantly higher SAT and TWV tend to occur around the LS (Figs. 4a1,a2,b1,b2). Anomalies in April around the LS seem to be stronger than those in May. Unlike high pressure anomalies around central and eastern Siberia in May (Figs. 4b3,b4), which may favor southwesterlies in the LS, significant low pressure anomalies occur around the Barents Sea (Figs. 4a3,a4), which may bring warm, moist air masses toward the LS. The latter reminds us of the Barents Oscillation (BO) (Skeie 2000), and previous studies have argued that negative BO in April tends to drive earlier MO in the LS (Liang and Su 2021).
Simultaneous atmospheric responses related to the BO in April (Figs. 4c1–c4) confirm that when the BO in April is in the negative phase, cyclonic anomalies appear around the Barents Sea both at the surface and in the midtroposphere, which tends to bring more warm, moist air masses toward the LS. MO anomalies related to BO in April show a similar spatial pattern to the L-mode, namely a seesaw-like structure in the southwest–northeast direction (Fig. 6b1; spatial correlation r = −0.72 with 99% confidence). The L-mode PC is also closely related to BO in April (Fig. 6b2; r = −0.35 with 90% confidence, still significant after detrending). Phases of the prominent L-mode are mostly consistent with BO phases in April (Fig. 2). Note that for convenience of comparison, the BO index and its related MO anomalies have been multiplied by the opposite sign at the same time. We notice that before 2007, the L-mode is mainly in its negative phase, whereas after 2007 the positive phase is more prominent. This indicates that MO in the LS has recently tended to be earlier. In addition, the BO in April is also related to increased storm activity in the Arctic (middle subplot in Fig. 7), though not as strongly as that related to AO in April. Although we do not see stronger storm activity in the LS than in the ESS, one region southwest of the LS shows significantly stronger storm activity. In all, we can say that BO in April tends to drive the L-mode of MO in the LS and ESS.
For the positive phase of the E-mode, related to earlier MO in the ESS, higher SAT and TWV in the ESS are more prominent in May than in April (Figs. 5a1,a2,b1,b2). We can also see stronger anomalies in the dynamic fields in May than in April (Figs. 5a3,a4,b3,b4). In May, positive-AO-like patterns appear both at the surface and in the midtroposphere. Unlike the case for the LE-mode, the cyclonic anomaly center tends to be located northeast of the LS and ESS, which may cause slight southerlies in the ESS and is consistent with positive anomalies of SAT and TWV.
Atmospheric responses to AO in May (Figs. 5c1–c4) are reminiscent of those related to the E-mode. Note that the dynamic center is northeast of the LS and ESS rather than northwest for AO in April. The thermal responses to AO in May around the LS and ESS are not quite significant. MO anomalies related to AO in May have a similar pattern to the E-mode (Fig. 6c1; spatial correlation r = 0.64 with 99% confidence), while the E-mode PC has a close connection with the AO index in May (Fig. 6c2; r = 0.50 with 99% confidence). Phases of the prominent E-mode are mostly consistent with those of AO in May (Fig. 2). In addition, AO in May is also related to increased storm activity around the Arctic, especially south of the ESS, which may favor earlier MO in the ESS. Nevertheless, the results above suggest that AO in May is the regulator of the E-mode.
4. Discussion and conclusions
In this study, three modes of MO in the LS and ESS are identified by EOF decomposition for 1979–2018, referred to as the LE-mode, L-mode, and E-mode. The LE-mode shows consistent MO anomalies in the LS and ESS. The L-mode denotes a seesaw-like pattern in the southwest–northeast direction, while the E-mode represents a seesaw-like pattern in the southeast–northwest direction. By assigning a prominent mode for each year, we can validate the three modes from the original MO data to some extent. The EOF analysis here reveals a combination of three modes of MO variability in the region of the LS and ESS, while the methodology of Liang and Su (2021) focuses on the early/late relationship (seesaw-like structure) between sea ice MO in the LS and ESS which is effectively a single mode of variability. At the same time, the L-mode and E-mode give a better picture of the seesaw structure.
Based on the maxima of the PCs of the three leading modes, we can define the prominent mode in each year. Then, the composite difference of the original MO between years with positive and negative phases of the same prominent mode exhibits similar patterns to the EOF modes, which suggests the realistic aspect of these MO modes. Another disturbing factor may be polynyas, which often exist in the LS and ESS in the early summer. In this paper, we have not considered polynyas. But given their small area relative to the whole region, their presence should not affect the EOF analyses here. Indeed, the strong air–ocean exchange over polynyas could potentially influence the local sea ice MO due to related downward thermal effects. This is a worthwhile topic for future study.
Decadal features of MO in the LS and ESS are indicated in this study. During the 1980s, the negative phase of the LE-mode is dominant, which represents later MO in the LS and ESS. In contrast, during the early 1990s, the positive phase of the LE-mode is prevalent, which denotes earlier MO and is consistent with Arctic warming during this period. After the mid-1990s, the L-mode and E-mode are more prominent, especially the L-mode, suggesting a more frequent seesaw-like pattern of MO in the LS and ESS. Since 1999, more strongly positive L-modes and the LE-modes imply earlier MO in the LS than in the ESS.
Coupled thermodynamic and dynamic factors indicate that the AO in April, BO in April, and AO in May have a causal linkage with the LE-mode, L-mode, and E-mode, respectively. The driving mechanisms are related mainly to the location of cyclonic anomalies. For the AO in April, the cyclonic anomaly is northwest of the LS and ESS, which brings warm, moist air masses to the LS and ESS. Meanwhile, storm activity increases around the LS and ESS. For the BO in April, the cyclonic anomaly is mainly around the Barents Sea, which causes warm, moist advection toward the LS. In this case, significantly strong storm activity occurs southwest of the LS. For the AO in May, the cyclonic anomaly is northeast of the LS and ESS, which directs cold advection to the LS and warm advection to the ESS. At the same time, storm activity south of the ESS increases (Fig. 9). Although both the AO and BO were implicated by Liang and Su (2021), this analysis shows how the different circulation patterns in April and May related to the AO and BO, respectively, are likely responsible for driving different modes of MO variability found in the EOF.
Schematic mechanism for sea ice MO in the LS and ESS.
Citation: Journal of Climate 36, 18; 10.1175/JCLI-D-22-0791.1
Some research has explored the independence of AO and BO. Tremblay (2001) indicated that the BO represents a shift in the action center associated with the AO, as the nonstationary AO causes an artifact of the BO in EOF analyses. However, Chen et al. (2013) suggested that a robust pattern resembling BO appears during different time periods, even when AO is relatively stationary. In this paper, although the time series of AO and BO in April are correlated, they have different influences on the modes of sea ice MO around the LS and ESS mainly due to their different location of pressure anomaly. For example, in 1991 (1996), only AO in April is substantially positive (negative), which leads to a positive (negative) LE-mode; in 1989, only BO in April is substantially negative, which leads to a negative L-mode (Fig. S5). In a later study, it would be worth exploring the specific mechanisms behind the AO and BO. Another atmospheric pattern, the Arctic dipole (Wu et al. 2006), seems to be not critical here for the MO variability, which is reasonable if the Arctic dipole configuration results in main meridional heat transport through the Chukchi Sea and the central Arctic Ocean. But, as the position of high and low pressure centers changes, the channel of meridional heat transport as well as the Arctic transpolar drift may influence the region of the LS and ESS (Watanabe et al. 2006). In this sense, the Arctic dipole could have potential effects on the MO variability in the LS and ESS.
There are other pathways regulating sea ice MO around the LS and ESS. Specifically for sea ice MO in the southern LS, Crawford et al. (2018) proposed that earlier retreat of snow cover in spring over the West Siberian Plain tends to form greater ridging over the ESS and result in more frequent southerly flow of warm and moist air masses over the LS. Figure 4b shows that a high pressure anomaly in May around eastern Siberia tends to drive air masses from the lower latitudes toward the LS, causing higher SAT and TWV there. This mechanism in May related to the L-mode, in addition to BO in April, to some extent explains the relatively weak correlation between BO in April and the L-mode (Fig. 6). At the same time, it is echoing the mechanism in Crawford et al. (2018) to some extent.
Employing the self-organizing mapping method on daily surface pressure (April–July), Horvath et al. (2021) tried to establish simultaneous relevance between synoptic surface atmospheric circulation and Arctic sea ice MO. Despite the different methodology, time scale, and time span, their result shows that a leading surface atmospheric pattern (node 13 in their paper) contributes to sea ice MO in the LS and ESS, reminiscent of the April AO pattern in this paper. The secondary pattern contributing to sea ice MO in the LS (ESS) is cyclonic (anticyclonic) circulation in the Arctic Ocean, which blows westerlies (easterlies) into the region. These results together show that from synoptic to subseasonal scales, various mechanisms could influence sea ice MO around the LS and ESS, suggesting the diversity of the atmospheric circulation in spring and summer.
Acknowledgments.
This work was supported by the National Natural Science Foundation of China (Grants 42288101 and 42120104001). We also give thanks to three anonymous reviewers who have shared constructive comments for this work.
Data availability statement.
The sea ice MO dataset from NASA Cryospheric Sciences Research Portal was downloaded at https://earth.gsfc.nasa.gov/cryo/data/arctic-sea-ice-melt. SAT from IABP/POLES was downloaded at https://arcticdata.io/catalog/view/doi:10.18739/A2J598, while SAT from AIRS was downloaded at https://disc.gsfc.nasa.gov/datasets/AIRS3STD_006/summary. The ERA5 dataset can be retrieved at https://cds.climate.copernicus.eu/cdsapp#!/search?type=dataset&keywords=((%20%22Product%20type:%20Reanalysis%22%20). In this study, we used ERA5 six-hourly and monthly averaged data for 1979–2018, at single levels and pressure levels.
REFERENCES
Belchansky, G. I., D. C. Douglas, I. N. Mordvintsev, and N. G. Platonov, 2004a: Estimating the time of melt onset and freeze onset over Arctic sea-ice area using active and passive microwave data. Remote Sens. Environ., 92, 21–39, https://doi.org/10.1016/j.rse.2004.05.001.
Belchansky, G. I., D. C. Douglas, and N. G. Platonov, 2004b: Duration of the Arctic sea ice melt season: Regional and interannual variability, 1979–2001. J. Climate, 17, 67–80, https://doi.org/10.1175/1520-0442(2004)017<0067:DOTASI>2.0.CO;2.
Blackmon, M. L., 1976: A climatological spectral study of the 500 mb geopotential height of the Northern Hemisphere. J. Atmos. Sci., 33, 1607–1623, https://doi.org/10.1175/1520-0469(1976)033<1607:ACSSOT>2.0.CO;2.
Bliss, A. C., and M. R. Anderson, 2014: Snowmelt onset over Arctic sea ice from passive microwave satellite data: 1979–2012. Cryosphere, 8, 2089–2100, https://doi.org/10.5194/tc-8-2089-2014.
Bliss, A. C., and M. R. Anderson, 2018: Arctic sea ice melt onset timing from passive microwave-based and surface air temperature-based methods. J. Geophys. Res. Atmos., 123, 9063–9080, https://doi.org/10.1029/2018JD028676.
Boisvert, L. N., T. Markus, and T. Vihma, 2013: Moisture flux changes and trends for the entire Arctic in 2003–2011 derived from EOS Aqua data. J. Geophys. Res. Oceans, 118, 5829–5843, https://doi.org/10.1002/jgrc.20414.
Budyko, M. I., 1969: The effect of solar radiation variations on the climate of the Earth. Tellus, 21A, 611–619, https://doi.org/10.3402/tellusa.v21i5.10109.
Chen, H. W., Q. Zhang, H. Körnich, and D. Chen, 2013: A robust mode of climate variability in the Arctic: The Barents Oscillation. Geophys. Res. Lett., 40, 2856–2861, https://doi.org/10.1002/grl.50551.
Crawford, A. D., S. Horvath, J. Stroeve, R. Balaji, and M. C. Serreze, 2018: Modulation of sea ice melt onset and retreat in the Laptev Sea by the timing of snow retreat in the West Siberian Plain. J. Geophys. Res. Atmos., 123, 8691–8707, https://doi.org/10.1029/2018JD028697.
Drobot, S. D., and M. R. Anderson, 2001: An improved method for determining snowmelt onset dates over Arctic sea ice using scanning multichannel microwave radiometer and special sensor microwave/imager data. J. Geophys. Res., 106, 24 033–24 049, https://doi.org/10.1029/2000JD000171.
Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803.
Horvath, S., J. Stroeve, B. Rajagopalan, and A. Jahn, 2021: Arctic sea ice melt onset favored by an atmospheric pressure pattern reminiscent of the North American–Eurasian Arctic pattern. Climate Dyn., 57, 1771–1787, https://doi.org/10.1007/s00382-021-05776-y.
Huang, Y., X. Dong, B. Xi, and Y. Deng, 2018: A survey of the atmospheric physical processes key to the onset of Arctic sea ice melt in spring. Climate Dyn., 52, 4907–4922, https://doi.org/10.1007/s00382-018-4422-x.
Kashiwase, H., K. I. Ohshima, S. Nihashi, and H. Eicken, 2017: Evidence for ice–ocean albedo feedback in the Arctic Ocean shifting to a seasonal ice zone. Sci. Rep., 7, 8170, https://doi.org/10.1038/s41598-017-08467-z.
Lalande, C., S. Bélanger, and L. Fortier, 2009: Impact of a decreasing sea ice cover on the vertical export of particulate organic carbon in the northern Laptev Sea, Siberian Arctic Ocean. Geophys. Res. Lett., 36, L21604, https://doi.org/10.1029/2009GL040570.
Lei, R., H. Xie, J. Wang, M. Leppäranta, I. Jónsdóttir, and Z. Zhang, 2015: Changes in sea ice conditions along the Arctic Northeast Passage from 1979 to 2012. Cold Reg. Sci. Technol., 119, 132–144, https://doi.org/10.1016/j.coldregions.2015.08.004.
Lei, R., and Coauthors, 2022: Seasonality and timing of sea ice mass balance and heat fluxes in the Arctic transpolar drift during 2019–2020. Elementa, 10, 000089, https://doi.org/10.1525/elementa.2021.000089.
Liang, H., and J. Su, 2021: Variability in sea ice melt onset in the Arctic northeast passage: Seesaw of the Laptev Sea and the East Siberian Sea. J. Geophys. Res. Oceans, 126, e2020JC016985, https://doi.org/10.1029/2020JC016985.
Liu, Z., and A. Schweiger, 2017: Synoptic conditions, clouds, and sea ice melt onset in the Beaufort and Chukchi seasonal ice zone. J. Climate, 30, 6999–7016, https://doi.org/10.1175/JCLI-D-16-0887.1.
Markus, T., J. C. Stroeve, and J. Miller, 2009: Recent changes in Arctic sea ice melt onset, freezeup, and melt season length. J. Geophys. Res., 114, C12024, https://doi.org/10.1029/2009JC005436.
Mortin, J., G. Svensson, R. G. Graversen, M.-L. Kapsch, J. C. Stroeve, and L. N. Boisvert, 2016: Melt onset over Arctic sea ice controlled by atmospheric moisture transport. Geophys. Res. Lett., 43, 6636–6642, https://doi.org/10.1002/2016GL069330.
Perovich, D. K., and C. Polashenski, 2012: Albedo evolution of seasonal Arctic sea ice. Geophys. Res. Lett., 39, L08501, https://doi.org/10.1029/2012GL051432.
Persson, P. O. G., 2012: Onset and end of the summer melt season over sea ice: Thermal structure and surface energy perspective from SHEBA. Climate Dyn., 39, 1349–1371, https://doi.org/10.1007/s00382-011-1196-9.
Petty, A. A., D. Schröder, J. C. Stroeve, T. Markus, J. Miller, N. T. Kurtz, D. L. Feltham, and D. Flocco, 2017: Skillful spring forecasts of September Arctic sea ice extent using passive microwave sea ice observations. Earth’s Future, 5, 254–263, https://doi.org/10.1002/2016EF000495.
Petty, A. A., N. T. Kurtz, R. Kwok, T. Markus, and T. A. Neumann, 2020: Winter Arctic sea ice thickness from ICESat‐2 freeboards. J. Geophys. Res. Oceans, 125, e2019JC015764, https://doi.org/10.1029/2019JC015764.
Rigor, I. G., R. L. Colony, and S. Martin, 2000: Variations in surface air temperature observations in the Arctic, 1979–97. J. Climate, 13, 896–914, https://doi.org/10.1175/1520-0442(2000)013<0896:VISATO>2.0.CO;2.
Sellers, W. D., 1969: A global climatic model based on the energy balance of the earth–atmosphere system. J. Appl. Meteor., 8, 392–400, https://doi.org/10.1175/1520-0450(1969)008<0392:AGCMBO>2.0.CO;2.
Skeie, P., 2000: Meridional flow variability over the Nordic Seas in the Arctic oscillation framework. Geophys. Res. Lett., 27, 2569–2572, https://doi.org/10.1029/2000GL011529.
Stroeve, J., and D. Notz, 2018: Changing state of Arctic sea ice across all seasons. Environ. Res. Lett., 13, 103001, https://doi.org/10.1088/1748-9326/aade56.
Stroeve, J., T. Markus, L. Boisvert, J. Miller, and A. Barrett, 2014: Changes in Arctic melt season and implications for sea ice loss. Geophys. Res. Lett., 41, 1216–1225, https://doi.org/10.1002/2013GL058951.
Thompson, D. W. J., and J. M. Wallace, 1998: The Arctic Oscillation signature in the wintertime geopotential height and temperature fields. Geophys. Res. Lett., 25, 1297–1300, https://doi.org/10.1029/98GL00950.
Tremblay, L.-B., 2001: Can we consider the Arctic Oscillation independently from the Barents Oscillation? Geophys. Res. Lett., 28, 4227–4230, https://doi.org/10.1029/2001gl013740.
Wang, L., G. J. Wolken, M. J. Sharp, S. E. L. Howell, C. Derksen, R. D. Brown, T. Markus, and J. Cole, 2011: Integrated pan-Arctic melt onset detection from satellite active and passive microwave measurements, 2000–2009. J. Geophys. Res., 116, D22103, https://doi.org/10.1029/2011jd016256.
Watanabe, E., J. Wang, A. Sumi, and H. Hasumi, 2006: Arctic dipole anomaly and its contribution to sea ice export from the Arctic Ocean in the 20th century. Geophys. Res. Lett., 33, L23703, https://doi.org/10.1029/2006GL028112.
Wu, B., J. Wang, and J. E. Walsh, 2006: Dipole anomaly in the winter Arctic atmosphere and its association with sea ice motion. J. Climate, 19, 210–225, https://doi.org/10.1175/JCLI3619.1.
Xu, D., L. Du, J. Ma, and H. Shi, 2020: Pathways of meridional atmospheric moisture transport in the central Arctic. Acta Oceanol. Sin., 39, 55–64, https://doi.org/10.1007/s13131-020-1598-9.
