1. Introduction
Snow cover is widely distributed on the land surface, sea ice, and ice sheets, and is characterized by high surface albedo, absorptivity, and low thermal conductivity (Zhang 2005). In recent decades, changes in snow conditions have been widely observed, including the reductions in Northern Hemisphere snow cover (IPCC 2019), enhanced surface snowmelt on the Greenland Ice Sheet (Flowers 2018), and earlier snowmelt onset and later freeze-up on the Arctic sea ice (Stroeve et al. 2014). As a spatially extensive and seasonally variable component of the cryosphere and an interface between the atmosphere and ground, snowpack exerts an important influence on the global climate system because of its linkages and feedbacks among hydrology, energy fluxes, and atmospheric circulation through seasonal melt (Barry and Gan 2011). The presence of liquid water decreases surface albedo, thereby increasing the solar radiation and temperature, thus enabling further melt through the albedo feedback mechanism. Meltwater runoff contributes to sea level rise directly. Surface snowmelt may endanger the ice shelves through hydrofracture (Lai et al. 2020) and induce the acceleration of ice flow when meltwater penetrates ice crevasses and lubricates the ice–bedrock surface (Poinar et al. 2017).
Continuous in situ observation of snowmelt in polar and mountain regions is challenging due to the formidable weather conditions. Satellite microwave remote sensing is well suited for the monitoring of melt timing and melt extent because of its relative insensitivity to atmospheric contamination and solar illumination effects compared with optical sensors, and strong microwave sensitivity to changes in surface dielectric properties between freeze–thaw transition (Tiuri et al. 1984). Spaceborne radiometers and scatterometers have been widely utilized to examine the melting conditions of terrestrial snow, sea ice, and ice sheet by identifying the sharp changes in brightness temperature (Tb) and radar backscatter. Northern Hemisphere terrestrial snow has been monitored for melt timing from radiometers over a number of years at both high latitudes and high altitudes (Ramage and Isacks 2003; Takala et al. 2009; Smith et al. 2017), showing an overall lengthening of melt season since the late 1970s (Tedesco et al. 2009). Snowmelt on the polar sea ice and ice sheets can be determined when the microwave measurements or their diurnal variation and gradient ratio are above a region-specific constant depending on the local snow properties (Zwally and Fiegles 1994; Abdalati and Steffen 1997; Forster et al. 2001; Nghiem et al. 2001; Tedesco 2007; Markus et al. 2009; Willmes et al. 2009; Trusel et al. 2012; Zheng et al. 2018).
Snowmelt variabilities are jointly affected by the underlying surface conditions (Panday et al. 2011; Ganey et al. 2017; Zhou et al. 2019) and the large-scale atmospheric fields (Mortin et al. 2016; Scott et al. 2019; Zheng et al. 2020a). Relationships between melt conditions in the cryosphere and the large-scale atmospheric conditions were diagnosed in both hemispheres (Tedesco et al. 2009; Tedesco and Monaghan 2009; Wang et al. 2013; Nicolas et al. 2017). Monitoring the large-scale snow melting conditions with a long historical record is essential for understanding the climate dynamics in the cryosphere. However, most of the previous research focuses on either a short period or a specific area. In this regard, Foster et al. (2011) developed a snow product that includes the daily freeze–thaw state of Northern Hemisphere terrestrial snow and the Greenland Ice Sheet. Furthermore, Wang et al. (2011, 2013) investigated the integrated pan-Arctic melt onset over sea ice, ice sheet, and terrestrial snow by combining active and passive microwave measurements. Compared with the Northern Hemisphere, the Southern Hemisphere has received much less attention with regard to the overall melting conditions.
Continued monitoring of the dynamics in integrated global snowmelt will provide useful information on the response of the cryosphere to the current global climate change. To consolidate and analyze snow melting conditions at a global scale, we generate a long time series data record of global melt onset (MO) over sea ice, ice sheets, and terrestrial snow in both hemispheres based on the recent progress in snowmelt monitoring technologies from passive microwave remote sensing measurements. We further explore the dynamics in regional and global MO and their connections to the large-scale atmospheric controls.
2. Data and methodology
a. Datasets
We utilize daily (including both ascending and descending overpasses) Tb retrievals from the Scanning Multichannel Microwave Radiometer (SMMR), Special Sensor Microwave Imager (SSM/I), and the Special Sensor Microwave Imager/Sounder (SSMIS) sensor series to achieve a long MO record at a global scale. With a wide swath and sun-synchronous polar orbits, these sensors allow excellent global coverage and frequent polar coverages. The Nimbus-7 SMMR dataset contains Tb measurements every other day from 1978 to 1987, and the SSM/I and SSMIS sensors can provide daily Tb measurements from 1987 to the present. These sensors obtained Tb retrievals at multiple frequencies with both vertical and horizontal polarization. Vertically polarized measurements at K band [18.7 GHz for SMMR and 19.35 GHz for SSM/I(S)] and Ka band (37.0 GHz for all the sensors) are employed in the monitoring of snowmelt. These datasets with a spatial resolution of 25 km are projected to the EASE-Grid projection and are available from the National Snow and Ice Data Center (NSIDC; https://nsidc.org/). The 25-km digital elevation model (DEM) at EASE-Grid projection used in this study was also obtained from NSIDC (ftp://sidads.colorado.edu/pub/DATASETS/brightness-temperatures/easegrid/tool).
Daily sea ice concentration (SIC) and snow water equivalent records are utilized for pixel-wise identification of the sea ice and snow cover. The Bootstrap SIC product in both hemispheres with a spatial resolution of 25 km obtained from the NSIDC is used to mask sea ice pixels in this study (Comiso 2017). A sea ice pixel was determined when SIC is above 15% (Meier and Stroeve 2008). ERA5 snow water equivalent data provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) is used to mask snow cover (Hersbach and Dee 2016), since the passive microwave snow product does not include the Southern Hemisphere and there is incomplete coverage of the optical remote sensing snow product due to the effect of cloud cover and polar night. With a horizontal resolution of 31 km, ERA5 snow water equivalent shows good performance with a root-mean-square error of about 48 mm (one of the best in the commonly used products) when validated against in situ measurements across the range of snow conditions in the Northern Hemisphere (Mortimer et al. 2020). The ERA5 data are resampled and reprojected to the 25-km EASE-Grid. In situ measurements from buoys on sea ice and from automatic weather stations (AWS) on ice sheet and land surface, including air temperature (Tair) measured below 3 m, snow temperature (Tsnow), albedo, and snow depth observations, are used to evaluate the MO detection algorithms. AWS observations over sea ice, ice sheet, and land surface are obtained from the International Arctic Buoy Programme (IABP) (http://iabp.apl.washington.edu/), the Programme for Monitoring of the Greenland Ice Sheet (PROMICE) (https://www.promice.dk/), and the National Oceanic and Atmospheric Administration (https://gis.ncdc.noaa.gov/maps/ncei), respectively.
The connections between snowmelt and atmospheric circulations are examined using several important climate indices that have been identified as influential on polar climate, including the Arctic oscillation (AO), North Atlantic Oscillation (NAO), Pacific decadal oscillation (PDO), Niño-3.4, Southern Oscillation index (SOI), and southern annular mode (SAM). AO is obtained by projecting the 1000-mb height anomalies poleward of 20°N and has been reported to can explain the MO variability in the Arctic (Tedesco et al. 2009). The NAO is the leading mode of low-frequency variability in the cool season across the North Atlantic that can modulate the melting condition on the Greenland Ice Sheet (Levitus et al. 1994; Thompson and Wallace 1998; Hahn et al. 2018). Sea ice variability and ice sheet surface melt have been found to be closely linked with the Pacific climate indices, typically including the PDO and Niño-3.4 (Lindsay and Zhang 2005; Tedesco and Monaghan 2009; Meehl et al. 2016; Nicolas et al. 2017; Hirota et al. 2018). PDO is calculated as the leading principal component of the North Pacific sea surface temperature variability poleward of 20°N (Zhang et al. 1997). Niño-3.4 is the area-averaged sea surface temperature anomaly within 5°S–5°N, 170°–120°W (Rayner et al. 2003). The AO, NAO, PDO, and Niño-3.4 indices are available at National Oceanic and Atmospheric Administration (NOAA; https://www.noaa.gov/). A recent study showed that surface melt over the Antarctic Ice Sheet and sea ice is related to the variability in SOI and SAM (Zheng et al. 2020b). The SOI index is defined as the normalized pressure difference between Tahiti and Darwin (Ropelewski and Jones 1987) and is calculated based on the Climatic Research Unit data (https://crudata.uea.ac.uk/cru/data/soi/). SAM is calculated using the zonal pressure difference between 40° and 65°S (Marshall 2003) and is available at https://legacy.bas.ac.uk/met/gjma/sam.html.
b. Melt onset detection algorithms
The SMMR, SSM/I, and SSMIS sensors detect MO by recognizing the sharp changes in Tb in the transition from dry snow to wet regime. A cross-platform calibration between the three sets of Tb measurements was conducted according to Dai et al. (2015). We generate a 40-yr integrated melt onset (MO) record over sea ice, ice sheets, and terrestrial snow from SMMR, SSM/I, and SSMIS based on the corresponding well-established algorithms in recent literature that have been applied at large-scale. MO over all kinds of snow and ice features is determined when melt signals last for at least 4 consecutive days. The Tair values from ERA5 are used to assist with MO detection, which is constrained to the days with compatible thermal regimes (e.g., Tair > −5°C) following Belchansky et al. (2004).
where i is time (day), subscripts “A” and “D” indicate the ascending and descending observations, and Tbow is the received Tb that contributed by the open-water portion. A pixel unmixing process is applied to amplify melt signals on sea ice because open water exhibits a much lower Tb (Markus and Cavalieri 1998), especially in the first-year sea ice zone. If we assume that SIC for the two passes remains unchanged (i.e., SICA = SICD = SIC), then we have DAVice(i) = |Tb37A(i) − Tb37D(i)|/SIC. A melt event is recognized when DAVice of the ice sheet or sea ice exceeds T = 10 K (Zheng et al. 2020b; Willmes et al. 2009). Tedesco et al. (2009) has proposed a dynamic-DAV method for melt detection over terrestrial snow by including the winter mean DAV as an offset because DAV during dry snow conditions could be very high (above the 10-K threshold). Winter DAV over the sea ice and ice sheet are much lower than that over the terrestrial snow, and always keep well below 10 K. We do not include an additional dynamic offset in the DAV algorithm because it may result in the missing of weak melt events.
The DAV algorithm has been validated against extensive field data in the Antarctic including buoys on sea ice and AWS on the ice sheet [see Fig. 1 in Willmes et al. (2009) and Figs. 2 and 3 in Zheng et al. (2020b)]. The overall accuracy (p0, the proportion of observed agreement) and kappa coefficient k = (p0 − pc)/(1 − pc) are used to measure the coincidence based on the confusion matrix, where pc is the proportion in agreement due to chance (Cohen 1960). Zheng et al. (2020b) showed that the overall accuracy and kappa coefficient between the positive maximum Tair and satellite-derived melt signals are 0.82 and 0.60 over sea ice and 0.93 and 0.79 over the ice sheet.
The DAV algorithm is also employed in the MO detection over the Northern Hemisphere. A large quantity of Tair observations in the Arctic Ocean and Greenland Ice Sheet allows a direct evaluation of the satellite-derived MO. Although there are dozens of AWSs and buoys, many stations are located on the land or along the coastline. Moreover, Tair has not been continuously recorded for many buoys. Hourly Tair from 11 AWSs on the Greenland Ice Sheet and 80 buoys in the Arctic were used in the evaluation (Fig. 1). Generally, MO derived from the satellites is well consistent with that from Tair, with a root-mean-square difference of 16 and 22 days over the ice sheet and sea ice, respectively. The accuracy is comparable to that in other studies about MO in both hemispheres (Mortin et al. 2012; Zheng et al. 2018). Differences exist between the two observation systems because ground data are collected at point scale whereas satellite data represent a spatial scale on the order of tens of kilometers. In addition, it takes time to produce meltwater when Tair exceeding the freezing point in the early melt season. Owing to the penetration and absorption of solar radiation within the snowpack, melting can also occur when Tair is below the melting point (Koh and Jordan 1995).
Evaluation of the DAV method over the Northern Hemisphere. (a) Arctic ice cover days; the gray lines and dots show the locations of the buoys and AWSs used in the evaluation, and the colorful lines and red dots denote the buoys and AWSs that presented in Fig. 2. (b) Comparisons between MO derived from satellite observations and that derived from in situ Tair over the Arctic sea ice (blue) and the Greenland Ice Sheet (red).
Citation: Journal of Climate 35, 10; 10.1175/JCLI-D-21-0265.1
We compare the Arctic sea ice MO derived from DAV algorithm and that from Markus et al. (2009) (available at https://earth.gsfc.nasa.gov/cryo/data/arctic-sea-ice-melt). The melt product from Markus et al. (2009) includes both early melt onset (EMO; first occurrence of melt) and continuous melt onset (CMO; first occurrence of free water). MO derived from DAV algorithm is on average 4 days later and 11 days earlier compared with EMO and CMO, with a root-mean-square difference of 13 and 20 days, respectively (Fig. S1 in the online supplemental material). The spatial pattern of MO is very similar to that of EMO, except for the central Arctic where MO occurred slightly later. We observe earlier MO records in the marginal sea ice zone compared with CMO. This is expected as radiometer is capable of detecting melt events before the appearance of free water, even when the liquid water content is 0.1% by volume (Tedesco 2009). CMO is also found to be 8 days later than that from near-surface temperatures (Markus et al. 2009).
In Fig. 2, we present the cases of the application of these methods in MO detection over first-year sea ice (Fig. 2a), multiyear sea ice (Fig. 2b), the Greenland Ice Sheet (Figs. 2c,d), and terrestrial snow in Siberia (Fig. 2e) and Alaska (Fig. 2f). Generally, the satellite-derived MO occurs in the days when Tair exceeds the freezing point and the surface albedo and snow depth begin to decline.
Cases of the application of MO detection methods adopted from Wang et al. (2013) and Zheng et al. (2020b). (a),(b) Comparisons between DAV, SIC, and the buoy measurements of first-year sea ice (magenta line in Fig. 1a) and multiyear sea ice (cyan line in Fig. 1a), respectively. (c),(d) Comparisons between DAV, surface albedo, and the AWS observations on the eastern (72.39°N, 27.23°W) and western side (67.07°N, 48.84°W) of the Greenland Ice Sheet. (e),(f) Comparisons between ascending TbD, snow depth, and AWS observations of terrestrial snow over Siberia (71.88°N, 82.70°E) and Alaska (64.84°N, 147.61°W), respectively. Red columns are the satellite-derived MO.
Citation: Journal of Climate 35, 10; 10.1175/JCLI-D-21-0265.1
c. Quality control and data analysis
To capture complete melt seasons, a melting year starts on 1 January (1 July) and ends on 31 December (30 June) for the Northern (Southern) Hemisphere. The missing Tb measurements are filled based on timeline interpolation. Only the pixels with sea ice (snow cover) that persist for at least 5 days are considered in snowmelt detection according to Markus et al. (2009). To avoid the effect from very late transient melt events, MO that occurs after 1 October (1 April) in the Northern (Southern) Hemisphere is not included in the analyses. Northern Hemisphere MO in 1987 is excluded in the analyses because of the continuously missing Tb measurements. Analysis of terrestrial snow MO is not conducted at lower latitudes (50°S–50°N) where snow cover is relatively short-lived and the TbD method may fail to work. Linear trends are calculated using a least squares method. Significance levels of the correlations and regressions are determined using Student’s t test. Only the pixels with at least 60% valid records (24 years) are included in the trend analysis. The multivariate regression model is used to quantify the correlation between MO and climate indices. Climate connections were further examined by comparing the time-scale principal components (PCs) of MO and climate modes.
3. Results
a. Climatology and trend
Melt detection is conducted over the polar sea ice and ice sheets, and terrestrial snow based on SMMR, SSM/I, and SSMIS passive microwave observations. An integrated, grid cell-wise record of global MO is generated for each year. Figure 3 shows the mean MO over the past four decades (1979–2018). MO is mapped onto a polar stereographic grid for both the Northern (Fig. 3a) and Southern (Fig. 2b) Hemispheres. MO occurs very late over the sea ice north of 70°N. On the Greenland Ice Sheet, surface snowmelt begins in April in the marginal regions and in late July in the inland (Fig. 3c). Snow cover is sparsely distributed outside the Antarctic Circle in the Southern Hemisphere with the exception of the southern end of South America. MO shows significantly latitudinal zonality, coming later from the marginal sea ice to the inland of the Antarctic Ice Sheet. MO can occur in October on the low-lying ice shelves of the Antarctic Peninsula (Fig. 3d). Generally, MO comes later with increasing latitude and elevation, and appreciably shows both latitudinal and vertical zonality (Fig. 3e). MO can be described as a quadratic function of latitude and elevation in both the Northern (r2 = 0.82, p < 0.01) and Southern (r2 = 0.89, p < 0.01) Hemispheres.
Mean MO over the period 1979–2018: (a) Northern Hemisphere, (b) Southern Hemisphere, (c) Greenland Ice Sheet, (d) Antarctic Peninsula, and (e) MO with varying latitudes and elevations.
Citation: Journal of Climate 35, 10; 10.1175/JCLI-D-21-0265.1
Figure 4 shows the pixel-based trends in global MO. We observe an extensive advance in MO over the Northern Hemisphere during the period 1979–2018, with hotpots distributed in the Barents Sea and the eastern Greenland Sea where MO also showed great temporal variations and the change rate can exceed 2 days yr−1 (Fig. 4a). Most of the statistically significant trends are negative in the Northern Hemisphere. Delay in MO can be found in a limited part of the Bering Sea. MO advance in the high-latitude areas of the Arctic sea ice and Greenland Ice Sheet were both more rapid than that in the lower latitudes (Figs. 4d,e). MO variability in the Southern Hemisphere is relatively discrete (Fig. 4b). Marginal sea ice in the Arctic shows a significant trend toward a later arrival of MO, especially in the Ross Sea.
Trend in MO over the period 1979–2018. (a),(b) Trends in MO in the Northern and Southern Hemisphere, respectively; black points indicate the trends that are significantly above the 95% confidence level. (c)–(f) MO trend with 1° latitude summaries over (left to right) all types, sea ice, ice sheet, and terrestrial snow, respectively. Error bars indicate the uncertainties of the trends that estimated at 95% confidence level; gray points indicate the trends that are significantly above the 95% confidence level.
Citation: Journal of Climate 35, 10; 10.1175/JCLI-D-21-0265.1
Spatially averaged variations in MO at the hemisphere and global scale, as well as over the Arctic sea ice, Greenland Ice Sheet, Arctic terrestrial snow (≥60°N), Eurasian terrestrial snow south of 60°N, North American terrestrial snow south of 60°N, Antarctic sea ice, and Antarctic Ice Sheet are examined to study the global and regional MO variabilities (Fig. 5). MO arrived significantly earlier (−0.44 ± 0.11 days yr−1) in the Northern Hemisphere during the period 1979–2018. Sustained advance in MO was found over the Arctic sea ice and Arctic terrestrial snow, with a rate of −0.23 ± 0.11 days yr−1 and −0.62 ± 0.16 days yr−1, respectively. MO in most of the Northern Hemisphere has occurred earlier from the 1980s to the middle of the 2000s, but afterward presented no significant trend (e.g., Eurasia) or even became later (e.g., Greenland Ice Sheet) in some regions. MO arrived significantly later (0.57 ± 0.17 days yr−1) in the Southern Hemisphere during the period 1979–2018, including both the Antarctic sea ice (0.61 ± 0.18 days yr−1) and Antarctic Ice Sheet (0.20 ± 0.12 days yr−1). The integrated MO at the global scale occurred significantly earlier with a rate of −0.19 ± 0.07 days yr−1. It is worthwhile to note that MO in both hemispheres and the global did not present a significant trend from 1998 to 2012 (i.e., the warming hiatus; IPCC 2013) when the warming rate was reported to become much slower.
Global and regional variations in MO during the period 1979–2018. Gray and black lines represent the original and 5-yr moving-average time series; red and blue lines are the MO trends during 1979–2018 and the warming hiatus period (1998–2012). Uncertainties of the trends are estimated at 95% confidence level.
Citation: Journal of Climate 35, 10; 10.1175/JCLI-D-21-0265.1
b. Melt onset and near-surface temperature
A trend map shows that annual mean Tair in most of the Northern Hemisphere has increased significantly (Fig. S2). The warming rate at the Northern Hemisphere high latitudes is much higher than that at lower latitudes. A decrease in annual mean Tair (July to June in Southern Hemisphere) was found in the Pacific and Indian Oceans around the Antarctic continent. Generally, trends in MO showed a similar pattern to that in ERA5 Tair (Fig. 4). Regions with a significant trend in MO were characterized by a significant relationship between MO and Tair, especially at the Northern Hemisphere high latitudes (Fig. S2). Overall, the change pattern of MO in the cryosphere is consistent with near-surface temperature. Particularly, a significant anticorrelation (r = −0.77, p < 0.01) is found between the MO and the mean Tair north of 50°N.
c. Melt onset and climatic indices
To better understand the role of atmospheric circulation in regional MO variability, we examine the relationship between MO and the climatic indices based on the multivariate regression model. The correlation analyses are based on detrended time series (Table S1). In the Northern Hemisphere, we find robust correlations (above 95% confidence level) between MO over the Greenland Ice Sheet and the AO or NAO. Particularly, NAO can explain 47% of the MO variability over the Greenland Ice Sheet (Fig. 6). AO+NAO can explain 29% and 28% of the variability of MO over the Arctic and Eurasian terrestrial snow, respectively. The correlations between MO and climate indices are relatively weaker in the Southern Hemisphere. Climate dynamics in the Pacific Ocean play an important role in MO dynamics in the Southern Hemisphere. This is manifested as a significant correlation (90% confidence level) between MO over the Antarctic sea ice and PDO, and between MO over the Antarctic Ice Sheet and Niño-3.4. PDO+Niño-3.4 can explain 13% of the variability of MO over the Antarctic sea ice. MO over the Antarctic Ice Sheet is also found to be related to SAM.
Comparison between detrended MO (blue line) over the Greenland Ice Sheet and NAO (black line). Dashed and solid lines represent the original and 5-yr moving-average time series.
Citation: Journal of Climate 35, 10; 10.1175/JCLI-D-21-0265.1
We further extract the space–time modes of MO variability based on empirical orthogonal function (EOF) analysis. EOF1 and EOF2 of MO each explain less than 10% of the variability in both hemispheres (Fig. 7). High variabilities are found in Europe and Siberia. EOF1 and EOF2 have an opposite sign in Europe. Both EOF1 and EOF2 show high variabilities over the Antarctic marginal sea ice zone. In Fig. 8, we present the correlation between the first five time-scale principal components of MO and the climate indices. The highest correlations were found between the first and second PCs and the climate modes. PC1 for the Arctic sea ice and Eurasian terrestrial snow is negatively related to AO (95% confidence level). PC2 for the Greenland Ice Sheet is significantly linked with both AO and NAO (99% confidence level). In the Southern Hemisphere, we find significant correlations between PC1 (PC2) and Niño-3.4 and SAM over the Antarctic Ice Sheet (sea ice).
Normalized EOF patterns for MO during the period 1979–2018: (a),(b) EOF1 and EOF2 in the Northern Hemisphere and (c),(d) EOF1 and EOF2 in the Southern Hemisphere.
Citation: Journal of Climate 35, 10; 10.1175/JCLI-D-21-0265.1
Correlation between the PCs of regional MO and climatic indices. Superscript letters a, b, and c indicate the correlation is significant above 99%, 95%, and 90% confidence level, respectively.
Citation: Journal of Climate 35, 10; 10.1175/JCLI-D-21-0265.1
The above correlation analyses suggest AO and NAO can considerably explain the MO variability over the Northern Hemisphere, while Niño-3.4 and SAM play important roles in the surface melt condition in the Southern Hemisphere. In Fig. 9, pixel-wise correlation analysis is conducted to show the relationships between these climate indices and MO. Tair shows a similar response to AO and NAO in the Northern Hemisphere (Fig. S3). A significant correlation between AO and MO is found in Europe where Tair is positively correlated with AO. A significant correlation between NAO and MO is found in western Canada and the Greenland Ice Sheet where Tair is negatively correlated with NAO. In the Southern Hemisphere, MO is strongly linked to Niño-3.4 and SAM in the Ross Sea where Tair is significantly related to these indices.
Map showing the pixel-wise correlations between climate indices and MO: (a),(b) correlations between AO/NAO, and MO and (c),(d) correlations between Niño-3.4/SAM, and MO. Black points indicate the correlations that are significantly above the 95% confidence level.
Citation: Journal of Climate 35, 10; 10.1175/JCLI-D-21-0265.1
d. Melt onset and extreme weather conditions
Besides the year-to-year connections between MO and large-scale climate variability, MO dynamics can also be driven by short-term climate extremes for weeks. Figure 10 presents an example of the very early arrival of Arctic sea ice MO as a result of the anomalous warm and southerly flow from the North Atlantic Ocean owning to the polar vortex breakdown. MO was observed in early spring in 2018 over the eastern Greenland and Barents Seas, mainly in the period of 22–26 February (Fig. 10a). Actually, extensive surface melt events occurred in the Arctic Ocean during this period (Fig. 10b). ERA5 shows an abnormal warming event during this period when maximum Tair increased to be close to or above the melting point (Fig. 11b). Figures 11a and 11c show the early stage (17–21 February) of the warming event, and the quick decrease in Tair with the recovery of the polar vortex in five days (27 February–2 March). The extreme warming event was characterized by anomalously high downward longwave radiation and surface sensible heat flux.
Early melt events in late February 2018. (a) MO in 2018 over the Arctic sea ice. (b) Number of melt events over the Arctic Ocean during 22–26 Feb 2018.
Citation: Journal of Climate 35, 10; 10.1175/JCLI-D-21-0265.1
Atmospheric conditions (left) before, (center) during, and (right) after the extreme warming event in early spring 2018, showing ERA5 (top) maximum Tair and mean wind field, (middle) downward longwave radiation, and (bottom) sensible heat flux.
Citation: Journal of Climate 35, 10; 10.1175/JCLI-D-21-0265.1
4. Discussion
Snow melting conditions over sea ice, ice sheets, and terrestrial snow have been previously investigated at a point or regional scale (e.g., Yan et al. 2009; Kuipers Munneke et al. 2018; Markus et al. 2009; Zheng et al. 2020b; Tedesco 2007). In this study, a global-scale MO dataset is generated based on spaceborne radiometer observations. Previous studies have investigated the ice retreat and snow end date from satellite-based sea ice and snow cover products (e.g., Stroeve et al. 2016; Chen et al. 2015; Stammerjohn et al. 2008). MO generally occurs earlier than ice retreat and snow-free dates, allowing a more timely prediction of the subsequent melting condition (Zheng et al. 2021). Although they have proved to perform well when validated against meteorological observations, the methods used in this study may still fail to detect MO because the optimal threshold may differ temporally and regionally with varying surface snow properties (Zheng et al. 2020b), especially in regions with complex topography where ERA5 always show great uncertainties and it is difficult for radiometers to work (Mortimer et al. 2020). MO for shallow snowpack is not identified in some years when snow cover disappears quickly. The DAV algorithm may fail to work due to ice disintegration and flooding effects in the marginal sea ice zone. It is worth noting that MO in the marginal sea ice zone is sometimes determined as the advance of a melting sea ice pixel rather than the timing with energy available to melt snow or ice, which thus may obscure the trend analysis. Although an intersensor calibration has been applied to minimize the difference in Tb, the varying crossing times due to orbit degradation and cross-platform may still introduce some uncertainties in the long-term trends of MO (Picard and Fily 2006; Tedesco et al. 2009).
A significant advance in MO was observed at the global scale during the period 1979–2018, especially in the Northern Hemisphere high latitudes. This is in line with global warming and the Arctic amplification given that the spatial distribution of trend in MO agrees well with that in the near-surface temperature (Figs. 4 and 6). The advance in MO in the Northern Hemisphere and globally has been slowing down since the 1990s (Fig. 4). A significant delay in MO was observed in the Southern Hemisphere. Although some studies have doubted the authenticity of the warming hiatus due to the incomplete observational coverage (Karl et al. 2015; Wang et al. 2017), no significant trend in MO was observed in the warming hiatus period in both hemispheres and globally from a continuous satellite-based dataset.
Arctic amplification has contributed to the dramatic melting of Arctic sea ice and snow cover, partly due to the surface albedo feedback (Winton 2006; Cohen et al. 2014). The sustained advance in MO over the Arctic sea ice indicates a longer melt season, which further enhances the heat exchange between the atmosphere and ocean (Serreze et al. 2009; Stroeve et al. 2014). About 47% of MO variability over the Greenland Ice Sheet can be explained by the NAO (Fig. 7), which is characterized by a dipole in the sea level pressure field with one anomaly center near Greenland. Greenland temperatures have been reported to show a close relationship with the NAO in recent decades (Hanna and Cappelen 2003; Hanna et al. 2013). Statistically significant correlations were found between MO of the terrestrial snow north of 50°N (except for North America) and the AO. This is consistent with the findings in Schaefer et al. (2004) and Tedesco et al. (2009). The earlier arrival of MO was correlated with increased temperatures in northern Eurasia (Fig. S2), which is associated with a positive phase of AO (Kryzhov and Gorelits 2015).
Snowmelt in the pan-Antarctic region was found to be strongly associated with the atmospheric component of El Niño–Southern Oscillation (ENSO) and SAM (Turner 2004; Tedesco and Monaghan 2009; Clem and Fogt 2013). The ENSO-/SAM-related sea ice MO changes were found in the Ross Sea where MO has occurred significantly later. This connection has been previously documented in Zheng et al. (2020b). As a consequence of a poleward propagation of Rossby wave train induced by the ENSO events, height anomalies were found over the Pacific sector of the Antarctic where the sea ice melting conditions were found to be strongly correlated with the summer SOI or Niño-3.4 (Turner 2004; Zheng et al. 2020b). MO dynamics over the Antarctic Ice Sheet were found to be linked with the SAM, which is characterized by zonally symmetric atmospheric pressure anomalies of opposite sign between Antarctica and midlatitudes and can affect the poleward heat transport (Marshall 2007; Tedesco and Monaghan 2009; Zheng et al. 2020b).
Extreme melt events over sea ice and ice sheet were always found to be attributed to the advection of warm air as a result of abnormal circulation conditions (Ballinger et al. 2018; Kuipers Munneke et al. 2018; Zheng et al. 2020a). In February 2018, the early melt events in the Arctic Ocean occurred in conjunction with a period of sustained anomalously warm southerly flow due to sudden stratospheric warming (Moore et al. 2018). Reanalysis data and in situ observations suggest a warm air intrusion from the midlatitudes elevated Tair up to 20°C above average (even above the freezing point) (Ludwig et al. 2019). The sudden stratospheric warming event has also contributed to a rapid decline of sea ice and the opening of the North Greenland Polynya in this period (Moore et al. 2018). These findings suggest that MO is a good indicator of long-term climate, as well as an indicator of extreme weather conditions.
5. Conclusions
Based on the recent progress in snowmelt detection technologies, a 40-yr integrated global MO dataset (1979–2018) is produced by combining the passive microwave estimates over sea ice, ice sheets, and terrestrial snow. Generally, the spatial distribution patterns of MO show latitudinal zonality and vertical zonality in both hemispheres. The temporal variabilities of MO agree well with near-surface temperature. An extensive advance in MO was found over the Northern Hemisphere (−4 days decade−1), particularly at the high latitudes where the warming rate was much higher than the global mean. MO occurred later in the Southern Hemisphere (6 days decade−1) with hotspots distributed in the marginal sea ice zone. Overall, the global cryosphere presented a trend toward an earlier arrival of MO with a rate of about −2 days decade−1. This trend, however, has slowed down since the 1990s. No significant trend in MO was observed in the so-called warming hiatus period (1998–2012) when the warming rate slowed down.
The globally integrated MO dataset brings new insights into the climate dynamics in the cryosphere and the associated atmospheric controls. Correlation analyses suggest MO dynamics over the high latitudes of the Northern Hemisphere were related with the Arctic Oscillation and North Atlantic Oscillation. Particularly, the North Atlantic Oscillation can explain 47% of the MO variability over the Greenland Ice Sheet. Connections between MO and climate indices are relatively weaker in the Southern Hemisphere. However, empirical orthogonal function analysis suggests that the first two principal components of MO were strongly correlated with Niño-3.4 and the southern annular mode over the Antarctic Ice Sheet and sea ice. Further studies are required to investigate the mechanisms responsible for these linkages. We observe a very early occurrence of MO in the Arctic Ocean in late February 2018, which was attributed to an anomalously warm southerly flow that elevates the near-surface temperature to be close to or above the melting point as a result of sudden stratospheric warming. These findings suggest the variabilities in MO are closely linked with both long-term climate dynamics and short-term extreme weather conditions.
Acknowledgments.
We thank the NASA DAAC at the NSIDC for providing the passive microwave datasets and sea ice concentration product (SMMR, https://n5eil01u.ecs.nsidc.org/PM/NSIDC-0071.001; SSM/I and SSMIS, https://n5eil01u.ecs.nsidc.org/DP1/PM/NSIDC-0032.002; Bootstrap sea ice concentration, https://n5eil01u.ecs.nsidc.org/PM/NSIDC-0079.003). The European Centre for Medium-Range Weather Forecasts (ECMWF) is thanked for providing the ERA5 data. Sea ice AWS observations are obtained from the buoys deployed by the International Arctic Buoy Programme (IABP) (http://iabp.apl.washington.edu/). Measurements from the Greenland Ice Sheet AWS are provided by the Programme for Monitoring of the Greenland Ice Sheet (PROMICE) (https://www.promice.dk/). Land surface AWS observations are obtained from the National Oceanic and Atmospheric Administration (https://gis.ncdc.noaa.gov/maps/ncei). This study was supported by the National Natural Science Foundation of China (Grant 42006192), the Innovation Group Project of Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai) (311021008), the China Postdoctoral Science Foundation (2020M683054 and 2021T140756), and the National Key Research and Development Program of China (Grant 2019YFC1509104). The authors declare that they have no conflict of interest.
Data availability statement.
The melt onset dataset over the period 1979-2018 in both hemispheres is available online at https://doi.org/10.6084/m9.figshare.14273576.v1.
REFERENCES
Abdalati, W., and K. Steffen, 1997: Snowmelt on the Greenland Ice Sheet as derived from passive microwave satellite data. J. Climate, 10, 165–175, https://doi.org/10.1175/1520-0442(1997)010<0165:SOTGIS>2.0.CO;2.
Ballinger, T. J., E. Hanna, R. J. Hall, T. E. Cropper, J. Miller, M. H. Ribergaard, J. E. Overland, and J. L. Høyer, 2018: Anomalous blocking over Greenland preceded the 2013 extreme early melt of local sea ice. Ann. Glaciol., 59, 181–190, https://doi.org/10.1017/aog.2017.30.
Barry, R. G., and T. Y. Gan, 2011: The Global Cryosphere: Past, Present, and Future. Cambridge University Press, 472 pp.
Belchansky, G. I., D. C. Douglas, and N. G. Platonov, 2004: 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.
Chen, X., S. Liang, Y. Cao, T. He, and D. Wang, 2015: Observed contrast changes in snow cover phenology in northern middle and high latitudes from 2001–2014. Sci. Rep., 5, 16820, https://doi.org/10.1038/srep16820.
Clem, K. R., and R. L. Fogt, 2013: Varying roles of ENSO and SAM on the Antarctic Peninsula climate in austral spring. J. Geophys. Res. Atmos., 118, 11 481–11 492, https://doi.org/10.1002/jgrd.50860.
Cohen, J., 1960: A coefficient of agreement for nominal scales. Educ. Psychol. Meas., 20, 37–46, https://doi.org/10.1177/001316446002000104.
Cohen, J., and Coauthors, 2014: Recent Arctic amplification and extreme mid-latitude weather. Nat. Geosci., 7, 627–637, https://doi.org/10.1038/ngeo2234.
Comiso, J., 2017: Bootstrap Sea Ice Concentrations from Nimbus-7 SMMR and DMSP SSM/I-SSMIS, version 3. NASA National Snow and Ice Data Center Distributed Active Archive Center, accessed 11 June 2020, https://doi.org/10.5067/7Q8HCCWS4I0R.
Dai, L., T. Che, and Y. Ding, 2015: Inter-calibrating SMMR, SSM/I and SSMI/S data to improve the consistency of snow-depth products in China. Remote Sensing, 7, 7212–7230, https://doi.org/10.3390/rs70607212.
Flowers, G. E., 2018: Hydrology and the future of the Greenland Ice Sheet. Nat. Commun., 9, 2729, https://doi.org/10.1038/s41467-018-05002-0.
Forster, R. R., D. G. Long, K. C. Jezek, S. D. Drobot, and M. R. Anderson, 2001: The onset of Arctic sea-ice snowmelt as detected with passive- and active-microwave remote sensing. Ann. Glaciol., 33, 85–93, https://doi.org/10.3189/172756401781818428.
Foster, J. L., and Coauthors, 2011: A blended global snow product using visible, passive microwave and scatterometer satellite data. Int. J. Remote Sensing, 32, 1371–1395, https://doi.org/10.1080/01431160903548013.
Ganey, G. Q., M. G. Loso, A. B. Burgess, and R. J. Dial, 2017: The role of microbes in snowmelt and radiative forcing on an Alaskan icefield. Nat. Geosci., 10, 754–759, https://doi.org/10.1038/ngeo3027.
Hahn, L., C. C. Ummenhofer, and Y. O. Kwon, 2018: North Atlantic natural variability modulates emergence of widespread Greenland melt in a warming climate. Geophys. Res. Lett., 45, 9171–9178, https://doi.org/10.1029/2018GL079682.
Hanna, E., and J. Cappelen, 2003: Recent cooling in coastal southern Greenland and relation with the North Atlantic Oscillation. Geophys. Res. Lett., 30, 1132, https://doi.org/10.1029/2002GL015797.
Hanna, E., J. M. Jones, J. Cappelen, S. H. Mernild, L. Wood, K. Steffen, and P. Huybrechts, 2013: The influence of North Atlantic atmospheric and oceanic forcing effects on 1900–2010 Greenland summer climate and ice melt/runoff. Int. J. Climatol., 33, 862–880, https://doi.org/10.1002/joc.3475.
Hersbach, H., and D. Dee, 2016: ERA5 reanalysis is in production. ECMWF Newsletter, No. 147, ECMWF, Reading, United Kingdom, 7, http://www.ecmwf.int/sites/default/files/elibrary/2016/16299-newsletter-no147-spring-2016.pdf
Hirota, N., H. Shiogama, H. Akiyoshi, T. Ogura, M. Takahashi, Y. Kawatani, M. Kimoto, and M. Mori, 2018: The influences of El Niño and Arctic sea-ice on the QBO disruption in February 2016. npj Climate Atmos. Sci., 1, 10, https://doi.org/10.1038/s41612-018-0020-1.
IPCC, 2013: Climate Change 2013: The Physical Science Basis. Cambridge University Press, 1535 pp., https://doi.org/10.1017/CBO9781107415324.
IPCC, 2019: Special Report on the Ocean and Cryosphere in a Changing Climate. IPCC, 765 pp., https://www.ipcc.ch/srocc/.
Karl, T. R., and Coauthors, 2015: Possible artifacts of data biases in the recent global surface warming hiatus. Science, 348, 1469–1472, https://doi.org/10.1126/science.aaa5632.
Koh, G. and R. Jordan, 1995: Sub-surface melting in a seasonal snow cover. J. Glaciol., 41, 474–482, https://doi.org/10.3189/S002214300003481X.
Kryzhov, V. N., and O. V. Gorelits, 2015: The Arctic Oscillation and its impact on temperature and precipitation in northern Eurasia in the 20th century. Russ. Meteor. Hydrol., 40, 711–721, https://doi.org/10.3103/S1068373915110011.
Kuipers Munneke, P., and Coauthors, 2018: Intense winter surface melt on an Antarctic ice shelf. Geophys. Res. Lett., 45, 7615–7623, https://doi.org/10.1029/2018GL077899.
Lai, C., J. Kingslake, M. G. Wearing, P. C. Chen, P. Gentine, H. Li, J. J. Spergel, and J. M. Van Wessem, 2020: Vulnerability of Antarctica’s ice shelves to meltwater-driven fracture. Nature, 584, 574–578, https://doi.org/10.1038/s41586-020-2627-8.
Levitus, S., J. I. Antonov, and T. P. Boyer, 1994: Interannual variability of temperature at a depth of 125 meters in the North Atlantic Ocean. Science, 266, 96–99, https://doi.org/10.1126/science.266.5182.96.
Lindsay, R. W., and J. Zhang, 2005: The thinning of Arctic sea ice, 1988–2003: Have we passed a tipping point? J. Climate, 18, 4879–4894, https://doi.org/10.1175/JCLI3587.1.
Ludwig, V., G. Spreen, C. Haas, L. Istomina, F. Kauker, and D. Murashkin, 2019: The 2018 North Greenland polynya observed by a newly introduced merged optical and passive microwave sea-ice concentration dataset. Cryosphere, 13, 2051–2073, https://doi.org/10.5194/tc-13-2051-2019.
Markus, T., and D. J. Cavalieri, 1998: Snow depth distribution over sea ice in the Southern Ocean from satellite passive microwave data. Antarctic Sea Ice: Physical Processes, Interactions and Variability, Vol. 74, Amer. Geophys. Union, 19–39.
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.
Marshall, G. J., 2003: Trends in the southern annular mode from observations and reanalyses. J. Climate, 16, 4134–4143, https://doi.org/10.1175/1520-0442(2003)016<4134:TITSAM>2.0.CO;2.
Marshall, G. J., 2007: Half-century seasonal relationships between the southern annular mode and Antarctic temperatures. Int. J. Climatol., 27, 373–383, https://doi.org/10.1002/joc.1407.
Meehl, G. A., J. M. Arblaster, C. M. Bitz, C. T. Y. Chung, and H. Teng, 2016: Antarctic sea-ice expansion between 2000 and 2014 driven by tropical Pacific decadal climate variability. Nat. Geosci., 9, 590–595, https://doi.org/10.1038/ngeo2751.
Meier, W. N., and J. Stroeve, 2008: Comparison of sea-ice extent and ice-edge location estimates from passive microwave and enhanced-resolution scatterometer data. Ann. Glaciol., 48, 65–70, https://doi.org/10.3189/172756408784700743.
Moore, G. W. K., A. Schweiger, J. Zhang, and M. Steele, 2018: What caused the remarkable February 2018 North Greenland Polynya? Geophys. Res. Lett., 45, 13 342–13 350, https://doi.org/10.1029/2018GL080902.
Mortimer, C., L. Mudryk, C. Derksen, K. Luojus, R. Brown, R. Kelly, and M. Tedesco, 2020: Evaluation of long-term Northern Hemisphere snow water equivalent products. Cryosphere, 14, 1579–1594, https://doi.org/10.5194/tc-14-1579-2020.
Mortin, J., T. M. Schrøder, A. Walløe Hansen, B. Holt, and K. C. McDonald, 2012: Mapping of seasonal freeze-thaw transitions across the pan-Arctic land and sea ice domains with satellite radar. J. Geophys. Res. Oceans, 117, C08004, https://doi.org/10.1029/2012JC008001.
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.
Nghiem, S. V., K. Steffen, R. Kwok, and W. Y. Tsai, 2001: Detection of snowmelt regions on the Greenland ice sheet using diurnal backscatter change. J. Glaciol., 47, 539–547, https://doi.org/10.3189/172756501781831738.
Nicolas, J. P., and Coauthors, 2017: January 2016 extensive summer melt in West Antarctica favoured by strong El Niño. Nat. Commun., 8, 15799, https://doi.org/10.1038/ncomms15799.
Panday, P. K., K. E. Frey, and B. Ghimire, 2011: Detection of the timing and duration of snowmelt in the Hindu Kush–Himalaya using QuikSCAT, 2000–2008. Environ. Res. Lett., 6, 024007, https://doi.org/10.1088/1748-9326/6/2/024007.
Picard, G., and M. Fily, 2006: Surface melting observations in Antarctica by microwave radiometers: Correcting 26-year time series from changes in acquisition hours. Remote Sens. Environ., 104, 325–336, https://doi.org/10.1016/j.rse.2006.05.010.
Poinar, K., I. Joughin, D. Lilien, L. Brucker, L. Kehrl, and S. Nowicki, 2017: Drainage of southeast Greenland firn aquifer water through crevasses to the bed. Front. Earth Sci., 5, 1–15, https://doi.org/10.3389/feart.2017.00005.
Ramage, J. M., and B. L. Isacks, 2003: Interannual variations of snowmelt and refreeze timing on southeast-Alaskan icefields, U.S.A. J. Glaciol., 49, 102–116, https://doi.org/10.3189/172756503781830908.
Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, https://doi.org/10.1029/2002JD002670.
Ropelewski, C. F., and P. D. Jones, 1987: An extension of the Tahiti–Darwin Southern Oscillation index. Mon. Wea. Rev., 115, 2161–2165, https://doi.org/10.1175/1520-0493(1987)115<2161:AEOTTS>2.0.CO;2.
Schaefer, K., A. S. Denning, and O. Leonard, 2004: The winter Arctic Oscillation and the timing of snowmelt in Europe. Geophys. Res. Lett., 31, L22205, https://doi.org/10.1029/2004GL021035.
Scott, R. C., J. P. Nicolas, D. H. Bromwich, J. R. Norris, and D. Lubin, 2019: Meteorological drivers and large-scale climate forcing of West Antarctic surface melt. J. Climate, 32, 665–684, https://doi.org/10.1175/JCLI-D-18-0233.1.
Serreze, M. C., A. P. Barrett, J. C. Stroeve, D. N. Kindig, and M. M. Holland, 2009: The emergence of surface-based Arctic amplification. Cryosphere, 3, 11–19, https://doi.org/10.5194/tc-3-11-2009.
Smith, T., B. Bookhagen, and A. Rheinwalt, 2017: Spatiotemporal patterns of High Mountain Asia’s snowmelt season identified with an automated snowmelt detection algorithm, 1987–2016. Cryosphere, 11, 2329–2343, https://doi.org/10.5194/tc-11-2329-2017.
Stammerjohn, S. E., D. G. Martinson, R. C. Smith, X. Yuan, and D. Rind, 2008: Trends in Antarctic annual sea ice retreat and advance and their relation to El Niño–Southern Oscillation and Southern Annular Mode variability. J. Geophys. Res., 113, C03S90, https://doi.org/10.1029/2007JC004269.
Stroeve, J. C., 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.
Stroeve, J. C., A. D. Crawford, and S. Stammerjohn, 2016: Using timing of ice retreat to predict timing of fall freeze-up in the Arctic. Geophys. Res. Lett., 43, 6332–6340, https://doi.org/10.1002/2016GL069314.
Takala, M., J. Pulliainen, S. J. Metsamaki, and J. T. Koskinen, 2009: Detection of snowmelt using spaceborne microwave radiometer data in Eurasia from 1979 to 2007. IEEE Trans. Geosci. Remote Sensing, 47, 2996–3007, https://doi.org/10.1109/TGRS.2009.2018442.
Tedesco, M., 2007: Snowmelt detection over the Greenland ice sheet from SSM/I brightness temperature daily variations. Geophys. Res. Lett., 34, L02504, https://doi.org/10.1029/2006GL028466.
Tedesco, M., 2009: Assessment and development of snowmelt retrieval algorithms over Antarctica from K-band spaceborne brightness temperature (1979–2008). Remote Sensing Environ., 113, 979–997, https://doi.org/10.1016/j.rse.2009.01.009.
Tedesco, M., and A. J. Monaghan, 2009: An updated Antarctic melt record through 2009 and its linkages to high-latitude and tropical climate variability. Geophys. Res. Lett., 36, L18502, https://doi.org/10.1029/2009GL039186.
Tedesco, M., M. Brodzik, R. Armstrong, M. Savoie, and J. Ramage, 2009: Pan Arctic terrestrial snowmelt trends (1979–2008) from spaceborne passive microwave data and correlation with the Arctic Oscillation. Geophys. Res. Lett., 36, L21402, https://doi.org/10.1029/2009GL039672.
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.
Tiuri, M., A. Sihvola, E. Nyfors, and M. Hallikaiken, 1984: The complex dielectric constant of snow at microwave frequencies. IEEE J. Oceanic Eng., 9, 377–382, https://doi.org/10.1109/JOE.1984.1145645.
Trusel, L. D., K. E. Frey, and S. B. Das, 2012: Antarctic surface melting dynamics: Enhanced perspectives from radar scatterometer data. J. Geophys. Res., 117, F02023, https://doi.org/10.1029/2011JF002126.
Turner, J., 2004: The El Niño–Southern Oscillation and Antarctica. Int. J. Climatol., 24 (1), 1–31, https://doi.org/10.1002/joc.965.
Wang, K., and Coauthors, 2017: Continuously amplified warming in the Alaskan Arctic: Implications for estimating global warming hiatus. Geophys. Res. Lett., 44, 9029–9038, https://doi.org/10.1002/2017GL074232.
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.
Wang, L., C. Derksen, R. Brown, and T. Markus, 2013: Recent changes in pan-Arctic melt onset from satellite passive microwave measurements. Geophys. Res. Lett., 40, 522–528, https://doi.org/10.1002/grl.50098.
Willmes, S., C. Haas, M. Nicolaus, and J. Bareiss, 2009: Satellite microwave observations of the interannual variability of snowmelt on sea ice in the Southern Ocean. J. Geophys. Res., 114, C03006, https://doi.org/10.1029/2008JC004919.
Winton, M., 2006: Amplified Arctic climate change: What does surface albedo feedback have to do with it? Geophys. Res. Lett., 33, L03701, https://doi.org/10.1029/2005GL025244.
Yan, F., J. Ramage, and R. McKenney, 2009: Modeling of high-latitude spring freshet from AMSR-E passive microwave observations. Water Resour. Res., 45, 11408, https://doi.org/10.1029/2008WR007370.
Zhang, T., 2005: Influence of the seasonal snow cover on the ground thermal regime: An overview. Rev. Geophys., 43, RG4002, https://doi.org/10.1029/2004RG000157.
Zhang, Y., J. M. Wallace, and D. S. Battisti, 1997: ENSO-like interdecadal variability: 1900–93. J. Climate, 10, 1004–1020, https://doi.org/10.1175/1520-0442(1997)010<1004:ELIV>2.0.CO;2.
Zheng, L., C. Zhou, R. Liu, and Q. Sun, 2018: Antarctic snowmelt detected by diurnal variations of AMSR-E brightness temperature. Remote Sensing, 10, 1391, https://doi.org/10.3390/rs10091391.
Zheng, L., C. Zhou, and K. Wang, 2020a: Enhanced winter snowmelt in the Antarctic Peninsula: Automatic snowmelt identification from radar scatterometer. Remote Sensing Environ., 246, 111835, https://doi.org/10.1016/j.rse.2020.111835.
Zheng, L., C. Zhou, T. Zhang, Q. Liang, and K. Wang, 2020b: Recent changes in pan-Antarctic region surface snowmelt detected by AMSR-E and AMSR2. Cryosph., 14, 3811–3827, https://doi.org/10.5194/tc-14-3811-2020.
Zheng, L., X. Cheng, Z. Chen, and Q. Liang, 2021: Delay in Arctic sea ice freeze-up linked to early summer sea ice loss: Evidence from satellite observations. Remote Sensing, 13, 2162, https://doi.org/10.3390/rs13112162.
Zhou, C., L. Zheng, Q. Sun, and R. Liu, 2019: Amery Ice Shelf surface snowmelt detected by ASCAT and Sentinel-1. Remote Sensing Lett., 10, 430–438, https://doi.org/10.1080/2150704X.2018.1553317.
Zwally, H. J., and S. Fiegles, 1994: Extent and duration of Antarctic surface melting. J. Glaciol., 40, 463–475, https://doi.org/10.3189/S0022143000012338.
