1. Introduction
The Antarctic Ice Sheet (AIS) is the largest ice body on the planet, and stores more than 70% of global freshwater (Fretwell et al. 2013). Changes in its total mass play a crucial role in global sea level, ocean temperature and salinity, atmospheric circulation, and thermohaline circulation. From 1979 to 2017, the total mass loss of the AIS was equivalent to 14.0 ± 2.0 mm global mean sea level rise as a result mostly of basal melting and ice calving into the ocean, and insignificant change in the net snow accumulation, that is, surface mass balance (SMB) (Lenaerts et al. 2019; Rignot et al. 2019), and the loss rate has accelerated in recent two decades (IMBIE team 2018; IPCC 2019, 2021). Despite more important losses due to dynamical factors, net accumulation from snowfall, the sole input of Antarctic mass balance, is vitally important for year-to-year mass variations. For example, the current West Antarctic Ice Sheet (WAIS) would be in a state of mass equilibrium if snow accumulation increased by more than 740 Gt yr−1 in this region (Feldmann et al. 2019). Furthermore, annual snowfall falling on the AIS is estimated to be over 2300 Gt (Mottram et al. 2021), equivalent to ∼6 mm global sea level decline. The short-term evolution of snow accumulation thus directly contributes to the variability of global sea level. When considering the ice sheet dynamics, the increased snow accumulation can enhance the ice flow toward the ocean at the centennial or even longer scales (Winkelmann et al. 2012).
SMB is the result of mass gain through snowfall and drifting-snow deposition subtracted by the losses of sublimation, wind-driven snow erosion, and runoff. In general, SMB is positive almost for the whole AIS, but negative for the blue area covering <7% of Antarctic surface (e.g., Das et al. 2013). The increase in global saturated water vapor pressure is estimated to be 7% if air temperature rises by 1 K. As air warms, the capacity to hold water over the AIS is expected to increase, which gives potential increase in snow accumulation and SMB (Lenaerts et al. 2016; Palerme et al. 2017; Dalaiden et al. 2020). Increased SMB may potentially play a bigger role in the mitigation of future global sea level rise. Thus, besides the concerns of variability in the ice discharge and basal melting, much attention should be also paid to quantification of increased SMB in future. However, this has still not been thoroughly understood because of the large uncertainties in the SMB predicted by global atmospheric climate models (AGCMs), which constrain the accuracy of sea level rise projection. To address this, it is consequently essential to fully understand the past and current variability in Antarctic SMB.
From the 1957–58 International Geophysical Year (IGY) onward, considerable attempts have been made to measure SMB using stakes or stake networks, snow pits/shallow ice cores, ultrasonic sounders, and ground-based and airborne-based ice penetrating radar. At present, more than 3500 sites have several years or even longer measurements (Wang et al. 2021). However, these measurements are still too spatially sparse and temporally discontinuous to directly quantify SMB over the entire AIS. Spatiotemporally complete SMB can be derived from a variety of reanalysis products such as the fifth-generation ECMWF reanalysis (ERA5), the National Aeronautics and Space Administration (NASA) Modern-Era Retrospective Analysis for Research and Applications version 2 (MERRA-2), or from regional atmospheric climate models (RCMs) such as MAR, RACMO, or from atmospheric global climate models (AGCMs). Unfortunately, they are relatively reliable only for the modern satellite era, and notable biases and questionable trends still exist (e.g., Bromwich et al. 2011; Wang et al. 2016, 2020). To make full use of the advantages of in situ observations and climate models and to the most extent overcome their respective disadvantages, recent several studies have attempted to reconstruct SMB changes at the continental or regional scale using ice core records from the compilation of Thomas et al. (2017), combined with atmospheric models (Medley and Thomas 2019; Wang et al. 2019; Dalaiden et al. 2020, 2021). While these reconstructions have substantially improved the understanding of Antarctic SMB variability, limitations still exist. Most of them did not account for the influence of depositional noise in single ice core records. For the reconstruction by data assimilation based on the AGCMs (Dalaiden et al. 2020), the poor skills of the AGCMs in capturing Antarctic precipitation changes and their coarse resolutions are an important limitation. Furthermore, no gridded Antarctic SMB reconstructions based on the observational records cover more than two centuries until now, which is a constraint for the estimation of longer-term Antarctic SMB trends and changes. Additionally, compared with the database of Thomas et al. (2017), one of the main distinguishing features of a new compilation of SMB observations (i.e., the AntSMB database) by Wang et al. (2021) is that it compiled various types of SMB measurements, and included more quality-controlled ice core records at an annual resolution. In particular, the AntSMB dataset improves spatial coverage of SMB from the ice core records, especially from the Filchner-Ronne Ice Shelf to the South Pole, on the Lambert Glacier drain and Dronning Maud Land. The most recently published long-term annually resolved ice core records at the South Pole are also included. Therefore, a new long-term reconstruction by means of these ice core records with higher spatial coverage is still required to improve upon the earlier work. The temporally extended reconstruction is also very useful for understanding Antarctic longer-term SMB trends and changes.
The objective of this study is to generate a new spatiotemporally complete reconstruction of SMB over the full AIS since 1701, based on the newly compiled ice core records from the AntSMB dataset (Wang et al. 2021), integrating with the fields of precipitation and evaporation from multiple reanalysis products, and SMB from two RCMs (RACMO and MAR), to extend the earlier work. Different from the earlier reconstructions, the reconstruction method we used is based on the reconciliation of different spatial weights from different atmospheric climate modes. This new reconstruction is then used to investigate the spatial and temporal changes in SMB on the entire Antarctic continent over the past 300 years, and to quantify the contribution of AIS SMB to the global sea level changes.
2. Data and methods
a. Data
1) Global atmospheric reanalysis products and regional climate models
Annual precipitation minus evaporation/sublimation (P − E) is calculated by five global reanalysis products, namely ERA5 (Hersbach et al. 2020), MERRA-2 (Gelaro et al. 2017), the European Centre for Medium-Range Weather Forecasts (ECMWF) interim reanalysis (ERA-Interim, herein ERAI) (Dee et al. 2011), the Japanese 55-year Reanalysis (JRA-55) (Kobayashi et al. 2015), and versions 1 and 2 of the Climate Forecast System Reanalysis (CFSR) (Saha et al. 2014). In particular, we use monthly total precipitation and evaporation of MERRA-2, ERA5, and JRA-55, monthly forecast accumulations of total precipitation and evaporation from ERAI, and CFSR monthly total of 6 h forecast accumulations of precipitation. It is notable that evaporation is not available in the CFSR dataset, which is calculated from monthly mean of the 6-h mean latent heat flux at the surface by a transformation coefficient of latent heat to sublimation (2.838 × 106 J kg−1). Annual SMB products are calculated by monthly SMB outputs of the regional climate model RACMO2 (1979–2016) (van Wessem et al. 2018) and MAR (1979–2017) (Agosta et al. 2019), which are both forced by ERAI.
2) Ice core records
Among the 175 records (Fig. 1a), we only utilize the records spanning the periods overlapping with the reanalyses/RCMs. Spatial representativeness of the remaining individual data is still impacted by depositional noises from drifting snow processes (e.g., Ding et al. 2011; Frezzotti et al. 2007, 2013). To improve the signal-to-noise ratio, if two or more individual ice core records are located at the same location including South Pole and WDC (Table S1 in the online supplemental material), we use their average to form a reconciled time series. If the annual net snow accumulation at a single core over Antarctica has a value above 700 kg m−2 yr−1, the ice core record at this core is representative for the estimation of year-to-year SMB changes, with the accuracy comparable to stake network measurements at this site (Frezzotti et al. 2007, 2013). Thus, we directly use the single time series of ice core records from the five sites (numbers 1–5 in Table S1) over the Antarctic Peninsula (AP), where have annual snow accumulation of >700 kg m−2 yr−1. Dense networks of ice core data are observed on the western Dronning Maud Land (DML), Ronne Ice Shelf, and Berkner Island (Fig. 1a). To decrease possible bias toward these high data density areas, records from the same subregions on western DML defined by Altnau et al. (2015) are averaged to form the corresponding stacked time series. We average all records on the Berkner Island. The Ronne Ice Shelf is divided into three regions (western, eastern, and interior areas), and then records in respective regions are averaged. Following Frezzotti et al. (2013) and data providers in the AntSMB compilation, the other records from a single ice core over West Antarctica and East Antarctica (numbers 7–10, 12–33, 35, and 41–58 in Table S1 in the online supplemental material) are considered to be less or even not influenced by small-scale perturbation from topography/wind driven effects, and thus they are directly used for the reconstruction. By filtering and reconciling, we finally obtain SMB time series at 62 locations (Fig. 1b and Table S1).
(a) Spatial distribution of 175 ice core locations, and multiyear averaged SMB at each core site is also shown; (b) a map showing 62 locations of the stacked ice core. The map is generated using the QAntarctica software (Matsuoka et al. 2021). The datasets are used as the background, including Antarctic Digital Database (ADD) low-resolution coastlines and ADD Rock_outcrop low-resolution polygons, from the Scientific Committee on Antarctic Research (SCAR) ADD version 7.0 (http://www.add.scar.org/). The black lines on inland Antarctica are the boundaries of the three major grounded Antarctic areas, i.e., the East Antarctic Ice Sheet (EAIS), West Antarctic Ice Sheet (WAIS), and Antarctic Peninsula (AP).
Citation: Journal of Climate 36, 23; 10.1175/JCLI-D-22-0747.1
b. Methods
1) Reconstruction methodology
Here we use the same ice core observations, but different spatial weights calculated by five global reanalysis products including ERA5, ERAI, MERRA-2, JRA-55, and CFSR, and two RCMs (i.e., RACMO2 and MAR) to generate seven reconstructions, respectively. Discrepancies of spatial weights from different models exist due to the different assimilated observations and data assimilation systems in the different global reanalyses, and different parameterization schemes in the climate models. However, all seven spatial weight fields are considered to be equally valid. The final reconstruction is assembled as the mean of the seven independent products.
To estimate the contribution of snow accumulation changes to global sea level, we convert the reconstructed normalized SMB values to SMB using the standard deviation of the reanalysis/regional climate model, and multiyear averaged fields of P − E or simulated snow accumulation. The procedure is performed separately for each dataset of reanalyses and RCMs. Before conversion, we evaluate the biases in the simulated snow accumulation fields using the ice core records above, and the quality-controlled AntSMB datasets including radar-based snow accumulation data by Wang et al. (2021) (Fig. S1). Only measurements falling within the period 1979–2018 are used. We calculate the difference between the observed values and the simulated snow accumulation at the grid cell where they have contemporaneous time spans, and then the difference is divided by the simulated values [i.e., bias = (simulation − observation)/simulation]. If two or more observations occur at a single grid cell, we first average all observations and then compare to model simulations in this grid cell. The true kriging interpolation method (i.e., distance-based interpolation) is used to interpolate the relative errors in the simulated SMB over the whole AIS. By bias correction, the resulting SMB fields are closer to observed values than those without correction.
2) Reconstruction validation
To estimate the robustness of our reconstructions, we use full reconstruction and verification statistics. The full reconstruction statistics are to estimate the reconstructions in which the weights are calculated using the full period (from 1979/80 onward). The verification reconstruction statistics estimate the addition reconstructions by means of a split calibration/verification method, previously used in the validation of the reconstructions of Antarctic temperature (Nicolas and Bromwich 2014) and atmospheric pressure (Fogt et al. 2017, 2018). The full period is divided into two nonoverlapping periods (1979/80–1997 and from 1998 onward). If the kriging weights are determined for one subperiod to generate a reconstruction, another subperiod is used to independently validate the reconstruction. This yields two sets of validation statistics, and the final verification statistics are the average of the two. The skill statistics are estimated using explained variance, the squared Pearson correlation coefficient, root-mean-square error (RMSE), reduction of error (RE), and coefficient of efficiency (CE).
We perform a cross-validation of our reconstruction using a leave-one-out procedure to further validate our reconstructions. Each site’s ice core records are excluded one by one, and the records of the remaining 61 sites are used to generate a prediction at the excluded site. This way, cross-validation can be made using the 62 alternative reconstructions.
3) Uncertainty estimation
The uncertainties in the individual reconstructions largely result from ice core record errors caused by depositional processes because of wind effects (i.e., small-scale noises) and the errors of the interpolation scheme. Following Monaghan et al. (2006), small-scale noises in the ice core records are quantified by the standard deviation of the average of reconstructed normalized SMB. To calculate uncertainty of the interpolation technique, we make the reconstructions of reanalysis P − E and RCM SMB records over the entire AIS using a subset of time series of SMB from reanalyses and RCMs corresponding to the 62 ice core locations. These reconstruction uncertainties can be determined by comparing them with the actual reanalysis or RCMs. Then the interpolation errors at the individual SMB reconstructions were approximated as the standard deviation of the difference between the reconstructed and original reanalysis and RCM records. The uncertainty of each reconstructed product based on ice core records is the root-mean-square of the sum of squares of the two types of uncertainty. The uncertainties in the final SMB reconstructions are estimated as the square root of the sum of squares of uncertainties of each individual gridded annual SMB reconstructions based on seven climate models by
The uncertainties of the trends in time series of snow accumulation are estimated here, accounting for the reconstruction errors using a Monte Carlo approach. We produce 10 000 Monte Carlo simulations of reconciled time series of snow accumulation for the 1701–2010 period, by adding the perturbations using white Gaussian noises with a variance equal to the squared reconstruction uncertainty. This variance integrated with the standard error of the regression coefficients is used to generate the final trend errors. We apply a two-tailed Student’s t test to test the statistical significance of the trends.
3. Results and discussion
a. Skill of Antarctic snow accumulation reconstruction
Figure 2 shows the validation statistics of seven full and verification SMB reconstructions against normalized ice core records. Regardless of the statistical parameter or the period for the reconstruction, we find that the respective skills of the seven reconstructions are not visibly differentiable. Each full reconstruction and independent verification significantly correlate with the ice core records at the vast majority of the sites, with median squared correlation coefficients of about 0.65 and 0.60, respectively (Figs. 2a,b). In particular, there are high similarities of squared correlations between the respective full reconstruction and verification (Figs. 2a,b), which means that reconstructions using the weights generated from different periods are nearly identical. This also demonstrates the insensitivity of the seven reconstructions to the timespan used for the generation of the kriging weights. The mean and median RMSE values of the seven different reconstructions are smaller than 0.20 (Fig. 2c), suggesting the annual reconstructions are robust. Furthermore, the positive RE and CE of the reconstructions at most ice core locations (Figs. 2d,e) demonstrate that the performance of our reconstructions is higher than application of the calibration and verification climatology. These suggest that a large portion of observed changes are replicated by the reconstructions, which can be further confirmed by the leave-one-out cross validation (Table 1).
Boxplots showing the statistics of the validation for the reconstructions based on five reanalyses and two regional climate models against the 62 reconciled ice core records. All boxes show the first quartile and third quartile, median of validations, and the whiskers are the limits of the distribution. (a) The squared Pearson correlation coefficients for the full reconstruction (
Citation: Journal of Climate 36, 23; 10.1175/JCLI-D-22-0747.1
Leave-one-out cross-validation statistics of the reconstructions in relation to independent observations, including squared correlation coefficients (R2), root-mean-square error (RMSE), and the coefficient of efficiency (CE).
To test the sensitivity of our reconstruction to the availability of the observations, we compare SMB time series from seven full reconstructions and the corresponding leave-one-out reconstructions at the 62 ice-core sites during 1701–2010. The linear regression between the independent dropped-site reconstructed time series and the full reconstruction yields slope values close to 1, with the intercept not significantly different from zero (Fig. S2). This suggests the minor impact of the removal of individual observations on our reconstruction.
b. A comparison with the previous Antarctic SMB reconstructions
The pioneering reconstruction based on ice-core records in combination with reanalysis shows an insignificantly negative trend of Antarctic continent-scale SMB since 1957 (Monaghan et al. 2006). In contrast to this, a positive but statistically insignificant trend (8.0 ± 5.0 Gt yr−1 decade−1) is observed in our reconstruction for the 1957–2000 period. This discrepancy may be caused by the lack of recent ice core records especially from Antarctic coastal regions in the reconstruction by Monaghan et al. (2006), but they are used for our reconstruction. Moreover, our result agrees well with that from the data-assimilation-based Antarctic SMB reconstruction during the same period (7.0 Gt yr−1 decade−1) (Dalaiden et al. 2020).
We also compare with spatial and temporal reconstructions of AIS SMB over the past 200 years using similar methods to our reconstruction (Medley and Thomas 2019) and data assimilation (Dalaiden et al. 2020, 2021) (Fig. S3). Time series of SMB averaged over the entire AIS from our reconstruction correlate robustly with the Medley and Thomas (2019) reconstruction (r = 0.80, p < 0.01) and two data-assimilation-based reconstructions (r = 0.64 and 0.45, p < 0.01) for the overlapping period (1801–2000). All four reconstructions exhibit a significantly upward trend from 1801 to 2000. The trend value of our reconstruction (4.5 Gt yr−1 decade−1) is very close to that from the Medley and Thomas (2019) reconstruction (4.0 Gt yr−1 decade−1), but slightly higher than the two data assimilation reconstructions (2.6 and 3.2 Gt yr−1 decade−1, respectively). A comparison with SMB reconstruction by Wang et al. (2019) shows high and significant correlations between the two reconstructions at the AP (r = 0.87, p < 0.01) and WAIS (r = 0.86, p < 0.01), respectively (Fig. S4).
Different from the Medley and Thomas (2019) reconstruction, the linearly detrended time series of reanalyses and RCMs are utilized to calculate the weighting coefficients, which greatly reduces the influence of suspicious trends of the reanalysis datasets. Compared with the reconstructions by Medley and Thomas (2019), and Wang et al. (2019), the reconstruction method used here is an improvement by considering spatial weights from seven different reanalysis products and RCMs to be equally valid. In such a manner, the final reconstruction is generated as the mean of seven independent reconstructions. We include ice core records with higher spatial coverage in our reconstruction than the four previous reconstructions. Furthermore, rather than single ice core records, we use the reconciled records when multiple ice cores are located in the same geographic area to minimize the effect of depositional noise and hence enhance the signal-to-noise ratio.
c. Antarctic SMB changes over the past three centuries
1) Temporal variability in snow accumulation
We use the final reconstruction (i.e., the average of the seven reconstructions described above) to investigate the historical variability in Antarctic SMB over the past 300 years (Fig. 3). Trends in snow accumulation are clearly spatially heterogeneous over all centuries considered. Statistically significant (p < 0.01) but opposing trends exist between the western and eastern WAIS for the 1801–1900 and 1901–2010 periods. Over the western WAIS, SMB trends are significantly positive over 1800–1900, but negative over 1900–2010, while the opposite occurs in eastern WAIS and the AP for the two 100‐yr time slices. The contrasting trends are also present over coastal regions of eastern DML for the 1701–1800 and 1801–1900 periods. However, for the same two time slices, the trends on George V Land and Adéle Land are significant (p < 0.01) and positive. In the 0°–120°E sector of the EAIS, directions of trends also vary over the three different centennial time slices.
Spatial distribution of SMB trends (kg m−2 yr−1 century−1) in the final reconstruction, i.e., the average of the seven reconstructions based on ERA5, ERAI, CFSR, JRA-55, MERRA-2, MAR, and RACMO for the (left) 1701–1800, (center) 1801–1900, and (right) 1901–2010 periods, respectively. Each 100-yr timespan is shown at the top of each panel. Areas significant at the confidence level of 99% are enclosed by dashed lines.
Citation: Journal of Climate 36, 23; 10.1175/JCLI-D-22-0747.1
Figure 4 shows the time series of reconstructed SMB averaged over the AP, WAIS, EAIS, and AIS including ice shelves. Over the AP, SMB shows an upward trend over 1701–1800, and then decreases for about 100 years (Fig. 4a). However, a rapid and potentially accelerated increase is observed over the twentieth and early twenty-first centuries (3.2 ± 0.9 Gt yr−1 decade−1 for 1901–2010, p < 0.01). There is no significant trend over the WAIS during the past 300 years (Fig. 4b), likely as a result of the counteracting seesaw pattern of SMB changes on the western and eastern WAIS. The EAIS SMB exhibits an obvious increase since the mid-1800s but shifts to decrease over the late twentieth century (Fig. 4c). When integrated over the whole ice sheet and ice shelves, a significant upward trend in SMB (3.6 ± 0.8 Gt yr−1 decade−1) is found for the 1701–2010 period, and the trend is larger during 1801–2010 (4.8 ± 1.5 Gt yr−1 decade−1) (Fig. 4d).
Time series of annual surface mass balance (SMB) from 1701 to 2010 averaged over the (a) AP, (b) WAIS, (c) EAIS, and (d) AIS, relative to the 1801–1900 mean, as presented by the thick black lines. The gray-shaded bounds are their corresponding uncertainties (±1σ).
Citation: Journal of Climate 36, 23; 10.1175/JCLI-D-22-0747.1
To further investigate the sensitivity of long-term SMB changes to the starting and ending years of the calculations, the trends for different periods of at least 50 years are estimated for the reconstructions averaged over the AP, WAIS, EAIS, and AIS, respectively (Fig. 5). At the AP, the negative trends (p < 0.05) occur between the early 1800s to mid-1840s and mid-1850s to early 1970s, whereas since 1900 all 50-yr or longer trends are positive and statistically significant. In particular, the most recent 50-yr period (1961–2010) has experienced an unpreceded 50-yr increasing trend over the past three centuries. It is clear that the significance of SMB trends over the WAIS is most sensitive to the calculation period. Before 1800, negative trends prevail over the EAIS. However, significantly positive trends are present between the 1740s to mid-1870s and 1850s to early 2000s. For the entire ice sheet, from the 1740s onward, the marked positive trends are overwhelmingly dominant.
The trends in annual SMB over the (a) AP, (b) WAIS, (c) EAIS, and (d) AIS for various timespans with the starting year on the y axis and ending year on the x axis in each panel. Each panel presents the trends calculated only by the data with at least 50 years long, with the longest period at the bottom right. The stippling in each panel shows where trends are significant at the 95% confidence interval.
Citation: Journal of Climate 36, 23; 10.1175/JCLI-D-22-0747.1
2) Spatial variability in snow accumulation
Empirical orthogonal functions (EOFs) are used to examine the leading patterns of reconstructed Antarctic SMB, shown in Fig. 6. In total, about 72.5% of the total spatial variability is explained by the first, second, and third EOFs (Fig. 6). EOF1 accounts for 38.0% of spatial variability in the Antarctic SMB reconstruction and shows a dipole pattern with opposite sign anomalies from the AP to the western WAIS. The dipole pattern is related to the precipitation spatial changes resulting from the depth of the Amundsen Sea low (ASL), which is tied to the phase of the Southern Annular Mode (SAM) and El Niño–Southern Oscillation (ENSO). A deepening of ASL drives a clockwise rotation of moisture advection, with a significant amount of northerly warm moisture to the AP and eastern WAIS, and an opposite decreased availability of moisture to the western WAIS. EOF2 explains 24.6% of spatial variability and consists of a strong signal over the entire AP and toward the coastal region of West Antarctica, where precipitation is largely driven by the ASL changes (e.g., Hosking et al. 2013; Turner et al. 2013; Wang et al. 2017). Due to the little precipitation over the EAIS by comparison, the AP and WAIS coastal regions dominate the overall AIS precipitation signal. About 9.9% of the total variability is explained by EOF3, which consists of two opposite strong signals in coastal Wilkes Land, and edges of the eastern DML and Adéle Land.
Spatial distribution of the top three EOF modes (EOF1, EOF2, and EOF3) of 310-yr Antarctic SMB anomalies (1701–2010). The percentage of variance explained by each EOF is given above each panel.
Citation: Journal of Climate 36, 23; 10.1175/JCLI-D-22-0747.1
d. Influence of large-scale atmospheric circulation conditions on Antarctic SMB changes
The possible effects of large-scale mode changes on interannual variability in Antarctic SMB are investigated using the SAM index of Marshall (2003) (SAMM), the Southern Oscillation index (SOI; Trenberth and Stepaniak 2001), the interdecadal Pacific oscillation (IPO) index (Henley et al. 2015), Niño-3.4 index (Rayner et al. 2003), and Niño-4 index (Trenberth and Stepaniak 2001), as well as the time series of EOF1 and EOF2, which explain >60% of variability in Antarctic SMB. In addition, in order to examine the influence of the tropical Pacific anomalies and the phase of SAM on Antarctic SMB at decadal or even longer scales, we use paleo-reconstruction of the Niño-3.4 index over the past 800 years (Emile-Geay et al. 2013), which represents the ENSO signal in the central-eastern tropical Pacific well, and the SAM index developed by Abram et al. (2014) (SAMA), which exists now at its most positive state during the past millennium (Abram et al. 2014).
A significantly positive correlation between annual SAMM and EOF1 principal component time series (r = 0.59, p < 0.05) is observed for the modern period 1979–2010 (Table 2). The correlation remains robust and statistically significant (r = 0.52, p < 0.05) based on the detrended time series. Furthermore, EOF1 shows linear statistically significant relationships with ENSO (SOI, Niño-3.4, and Niño-4) and ENSO-like variability (IPO) (r = −0.32, p < 0.05) (Table 2). ENSO also robustly correlates with the residuals of EOF1 regression with SAMM (r = −0.37, p < 0.05 for Niño 3.4, and r = 0.45, p < 0.05 for SOI) (Fig. S5). The sign of correlations suggests that positive phases of EOF1 (i.e., widespread positive SMB anomalies over the AP) are associated with the SAM positive polarity and also with the negative sea surface temperature (SST) anomalies over the central and eastern tropical Pacific (i.e., La Niña conditions), which agrees with the findings of Wang et al. (2017) and Thomas et al. (2008, 2015). In contrast, there are not any significant correlations of EOF2 time series with SAM, ENSO, and IPO (Table 2).
Correlations between annual EOF1 and EOF2 time series and SAMM, ENSO (Niño-3.4, Niño-4, and SOI indices) and IPO, respectively, spanning 1979–2010. Correlations of 10-yr running-average time series of EOF1 and EOF2 with reconstructed SAMA and Niño-3.4, respectively, spanning from 1701 onward. Boldface italic and boldface fonts denote the correlations statistically significant at the 95% and 99% confidence level, respectively.
The 10-yr moving-average time series of the normalized SAMA and Niño-3.4, EOF1, and EOF2 (Fig. S6) are used to investigate their relationships extending over the past 300 years. Robust correlation (r = 0.62, p < 0.01) occurs between EOF1 and the normalized SAMA, explaining ∼39% of EOF1 variability. The SAM dramatically transited toward a positive phase from 1701 to early nineteenth century, and then this trend reversed over the nineteenth century, and recommenced during the twentieth century. EOF1 variability is consistent with the SAM, with a positive trend during the eighteenth century, and a trend reversal in the nineteenth century, recommencing over the past 100 years. EOF1 running-averaged time series are weakly correlated with Niño-3.4 (r = −0.22, p < 0.01), seeming to suggest a secondary role of tropical forcing in the EOF1 (i.e., AP snowfall changes). Before the late 1800s, the Niño-3.4 SST negative trend exerted a positive forcing on the mean state of AP snowfall, but after this it acted in a way that mitigated the positive influence of increasing SAM on SMB of the AP. No significant relationship is observed between SAMA and EOF2, but EOF2 is positively and significantly correlated with the Niño-3.4 (r = 0.49, p < 0.01). This means that the positive phase of EOF2 (i.e., increase in SMB along the coastal AP and west Antarctica) is associated with positive Niño-3.4 SST anomalies (El Niño period), and the negative EOF2 phase with La Niña periods. Kaspari et al. (2004) also reported that increased SMB is related to El Niño events at most ice core sites over the Pine Island–Thwaites drainage system. While these suggest a widespread ENSO teleconnection with precipitation on the WAIS for prolonged periods, some studies point out that teleconnection is temporally unstable (e.g., Bromwich et al. 2000; Cullather et al. 1996; Banta et al. 2008; Wang et al. 2019).
e. Contribution of Antarctic SMB variability to global sea level rise
To convert AIS SMB changes to global sea level rise, it is an important prerequisite to define a long-term (at least one century) average of the ice sheet in balance. According to Medley and Thomas (2019), mass input through SMB over the entire nineteenth-century can be a tenable representation of the long-term mean for the AIS as a reference state. We calculate cumulative mass variability by summing up a SMB anomaly, relative to the nineteenth century with time, and then translate it into global sea level changes through dividing by the ocean area (362 × 106 km2; Parker 1980). This means that contribution of AIS SMB to global sea level is zero over the entire nineteenth century (Fig. 7). For the twentieth century, the positive anomalies of SMB over the whole AIS relative to the baseline cumulatively dampened global sea level rise by about 14.1 ± 4.8 mm, with a rate of 1.41 ± 0.48 mm decade−1(Fig. 7). The mitigation of global sea level rise from EAIS and WAIS is estimated to be 9.0 ± 3.2 and 2.7 ± 1.2 mm from 1901 to 2010 (Fig. 7). The cumulative contribution of AP SMB to global sea level is positive over most of the nineteenth century, but negative from 1901 to the mid-1980s. Due to the recent increase in SMB, AP becomes an important resource of global sea level mitigation since the mid-1980s. Conversely, a potential slowdown mitigation from EAIS is observed, caused by decreasing SMB from the early 1980s onward. Furthermore, the global sea level mitigation from the entire AIS decreased after entering the twenty-first century.
Cumulative global sea level mitigation caused by SMB variations over the AP, WAIS, EAIS, and AIS from 1801 onward, relative to the nineteenth-century average, respectively. A negative value stands for the positive contribution of SMB in Antarctica to global sea level rise, and in contrast, a positive one is the mitigation of global sea level rise. All error bounds in each Antarctic section indicate the ±1σ range. They are generated by a Monte Carlo method, and a detailed description is provided in section 2.
Citation: Journal of Climate 36, 23; 10.1175/JCLI-D-22-0747.1
4. Conclusions
In this study, to improve the understanding of long-term Antarctic SMB variations and their impact on the global sea level, we present a new reconstruction of ice-sheet scale SMB spanning the past 310 years, which merges the superiority of ice core records in terms of time coverage with the superiority of recent reanalysis products and RCMs in terms of spatial coverage. Cross-validation shows the robust skill of our reconstruction to represent the spatial and temporal variability in observed SMB during the past 300 years.
Spatial heterogeneity in magnitude and sign of the reconstructed SMB variability is found across the AIS. However, SMB averaged over the entire ice sheet experienced a significantly increasing trend from 1701 to 2010 (3.6 ± 0.8 Gt yr−1 decade−1), and a larger positive trend over 1801–2010 (4.8 ± 1.5 Gt yr−1 decade−1). The increased SMB results in a mitigation of global sea level rise by 14.1 ± 4.8 mm from 1901 to 2010. There is no significant trend in AIS SMB since the 1980s, suggesting a weakening of the long-term buffering effect. It is notable that our estimation of sea level mitigation only accounts for the mass input of SMB to the AIS, not mass loss due to dynamical imbalance.
About 38.0% and 24.6% of the total variability in the reconstruction can be explained by EOF1 and EOF2, respectively. EOF1 is characterized by an SMB dipole between AP (including eastern WAIS) and western WAIS. The dipole is closely related to SAM, ENSO, and IPO. Annual EOF2 changes are not driven by the three modes, but decadal variations are predominantly tied to ENSO.
Our reconstruction can be used to constrain SMB simulations over Antarctica by climate models such as the Coupled Model Intercomparison Project phase 6 (CMIP6) multimodels to estimate Antarctic mass balance, and to drive ice sheet and hydrological models.
Acknowledgments.
This work was funding by the National Natural Science Foundation of China (41971081), the Strategic Priority Research Program of the Chinese Academy of Sciences (XAD19070103), the National Key Research and Development Program of China (2020YFA0608202), and the Project for Outstanding Youth Innovation Team in the Universities of Shandong Province (2019KJH011). The authors thank Shawn Marshall (the editor), Marie Cavitte (the reviewer), and two anonymous reviewers for their constructive comments and suggestions to improve the paper.
Data availability statement.
Monthly MAR SMB data are available at https://doi.org/10.5281/zenodo.2547637. RACMO outputs were provided by Michiel R. van den Broeke, and also can be freely downloaded at https://zenodo.org/record/6367850#.YzY8HMhBxPY. The reanalysis data are available as follows: ERA5 (https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era5), ERA-Interim (https://www.ecmwf.int/en/forecasts/datasets/archive-datasets/reanalysis-datasets/era-interim), CFSR (https://rda.ucar.edu/datasets/ds093.2/), JRA-55 (http://rda.ucar.edu/datasets/ds628.1/), and MERRA-2 (https://disc.gsfc.nasa.gov/). The ice core records are downloaded from the AntSMB dataset (https://doi.org/10.11888/Glacio.tpdc.271148).
REFERENCES
Abram, N. J., R. Mulvaney, F. Vimeux, S. J. Phipps, J. Turner, and M. H. England, 2014: Evolution of the Southern Annular Mode during the past millennium. Nat. Climate Change, 4, 564–569, https://doi.org/10.1038/nclimate2235.
Agosta, C., and Coauthors, 2019: Estimation of the Antarctic surface mass balance using the regional climate model MAR (1979–2015) and identification of dominant processes. Cryosphere, 13, 281–296, https://doi.org/10.5194/tc-13-281-2019.
Altnau, S., E. Schlosser, E. Isaksson, and D. Divine, 2015: Climatic signals from 76 shallow firn cores in Dronning Maud Land, East Antarctica. Cryosphere., 9, 925–944, https://doi.org/10.5194/tc-9-925-2015.
Banta, J. R., J. R. McConnell, M. M. Frey, R. C. Bales, and K. Taylor, 2008: Spatial and temporal variability in snow accumulation at the West Antarctic ice sheet divide over recent centuries. J. Geophys. Res., 113, D23102, https://doi.org/10.1029/2008JD010235.
Bromwich, D. H., A. N. Rogers, P. Kallberg, R. I. Cullather, J. W. C. White, and K. J. Kreutz, 2000: ECMWF analyses and reanalyses depiction of ENSO signal in Antarctic precipitation. J. Climate, 13, 1406–1420, https://doi.org/10.1175/1520-0442(2000)013<1406:EAARDO>2.0.CO;2.
Bromwich, D. H., J. P. Nicolas, and A. J. Monaghan, 2011: An assessment of precipitation changes over Antarctica and the Southern Ocean since 1989 in contemporary global reanalyses. J. Climate, 24, 4189–4209, https://doi.org/10.1175/2011JCLI4074.1.
Cullather, R. I., D. H. Bromwich, and M. L. van Woert, 1996: Interannual variations in Antarctic precipitation related to El Niño-Southern Oscillation. J. Geophys. Res., 101, 19 109–19 118, https://doi.org/10.1029/96JD01769.
Dalaiden, Q., H. Goosse, F. Klein, J. T. M. Lenaerts, M. Holloway, L. Sime, and E. R. Thomas, 2020: How useful is snow accumulation in reconstructing surface air temperature in Antarctica? A study combining ice core records and climate models. Cryosphere, 14, 1187–1207, https://doi.org/10.5194/tc-14-1187-2020.
Dalaiden, Q., H. Goosse, J. Rezsöhazy, and E. R. Thomas, 2021: Reconstructing atmospheric circulation and sea-ice extent in the West Antarctic over the past 200 years using data assimilation. Climate Dyn., 57, 3479–3503, https://doi.org/10.1007/s00382-021-05879-6.
Das, I., and Coauthors, 2013: Influence of persistent wind scour on the surface mass balance of Antarctica. Nat. Geosci., 6, 367–371, https://doi.org/10.1038/ngeo1766.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597, https://doi.org/10.1002/qj.828.
Ding, M., C. Xiao, Y. Li, J. Ren, S. Hou, B. Jin, and B. Sun, 2011: Spatial variability of surface mass balance along a traverse route from Zhongshan station to Dome A, Antarctica. J. Glaciol., 57, 658–666, https://doi.org/10.3189/002214311797409820.
Emile-Geay, J., K. M. Cobb, M. E. Mann, and A. T. Wittenberg, 2013: Estimating central equatorial Pacific SST variability over the past millennium. Part II: Reconstructions and implications. J. Climate, 26, 2329–2352, https://doi.org/10.1175/JCLI-D-11-00511.1.
Feldmann, J., A. Levermann, and M. Mengel, 2019: Stabilizing the West Antarctic ice sheet by surface mass deposition. Sci. Adv., 5, eaaw4132, https://doi.org/10.1126/sciadv.aaw4132.
Fogt, R. L., C. A. Goergens, J. M. Jones, D. P. Schneider, J. P. Nicolas, D. H. Bromwich, and H. E. Dusselier, 2017: A twentieth century perspective on summer Antarctic pressure change and variability and contributions from tropical SSTs and ozone depletion. Geophys. Res. Lett., 44, 9918–9927, https://doi.org/10.1002/2017GL075079.
Fogt, R. L., L. N. Clark, and D. H. Bromwich, 2018: A new monthly pressure dataset poleward of 60°S since 1957. J. Climate, 31, 3865–3874, https://doi.org/10.1175/JCLI-D-17-0879.1.
Fretwell, P., and Coauthors, 2013: Bedmap2: Improved ice bed, surface and thickness datasets for Antarctica. Cryosphere, 7, 375–393, https://doi.org/10.5194/tc-7-375-2013.
Frezzotti, M., S. Urbini, M. Proposito, C. Scarchilli, and S. Gandolfi, 2007: Spatial and temporal variability of surface mass balance near Talos Dome, East Antarctica. J. Geophys. Res., 112, F02032, https://doi.org/10.1029/2006JF000638.
Frezzotti, M., C. Scarchilli, S. Becagli, M. Proposito, and S. Urbini, 2013: A synthesis of the Antarctic surface mass balance during the last 800 yr. Cryosphere, 7, 303–319, https://doi.org/10.5194/tc-7-303-2013.
Gelaro, R., and Coauthors, 2017: The Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2). J. Climate, 30, 5419–5454, https://doi.org/10.1175/JCLI-D-16-0758.1.
Henley, B. J., J. Gergis, D. J. Karoly, S. Power, J. Kennedy, and C. K. Folland, 2015: A tripole index for the Interdecadal Pacific Oscillation. Climate Dyn., 45, 3077–3090, https://doi.org/10.1007/s00382-015-2525-1.
Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803.
Hosking, J. S., O. Andrew, J. Gareth, J. Turner, and T. Phillips, 2013: The influence of the Amundsen–Bellingshausen Seas low on the climate of West Antarctica and its representation in coupled climate model simulations. J. Climate, 26, 6633–6648, https://doi.org/10.1175/JCLI-D-12-00813.1.
IMBIE team, 2018: Mass balance of the Antarctic ice sheet from 1992 to 2017. Nature, 558, 219–222, https://doi.org/10.1038/s41586-018-0179-y.
IPCC, 2019: The Ocean and Cryosphere in a Changing Climate. Cambridge University Press, 755 pp., https://doi.org/10.1017/9781009157964.
IPCC, 2021: Climate Change 2021: The Physical Science Basis. Cambridge University Press, 2391 pp., https://doi.org/10.1017/9781009157896.
Kaspari, S., P. A. Mayewski, D. A. Dixon, V. B. Spikes, S. B. Sneed, M. J. Handley, and G. S. Hamilton, 2004: Climate variability in West Antarctica derived from annual accumulation-rate records from ITASE firn/ice cores. Ann. Glaciol., 39, 585–594, https://doi.org/10.3189/172756404781814447.
Kobayashi, S., and Coauthors, 2015: The JRA-55 reanalysis: General specifications and basic characteristics. J. Meteor. Soc. Japan, 93, 5–48, https://doi.org/10.2151/jmsj.2015-001.
Lenaerts, J. T. M., M. Vizcaino, J. Fyke, L. van Kampenhout, and M. R. van den Broeke, 2016: Present-day and future Antarctic ice sheet climate and surface mass balance in the Community Earth System Model. Climate Dyn., 47, 1367–1381, https://doi.org/10.1007/s00382-015-2907-4.
Lenaerts, J. T. M., B. Medley, M. R. van den Broeke, and B. Wouters, 2019: Observing and modeling ice sheet surface mass balance. Rev. Geophys., 57, 376–420, https://doi.org/10.1029/2018RG000622.
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.
Matsuoka, K., and Coauthors, 2021: Quantarctica, an integrated mapping environment for Antarctica, the Southern Ocean, and sub-Antarctic islands. Environ. Modell. Software, 140, 105015, https://doi.org/10.1016/j.envsoft.2021.105015.
Medley, B., and E. R. Thomas, 2019: Increased snowfall over the Antarctic ice sheet mitigated twentieth-century sea-level rise. Nat. Climate Change, 9, 34–39, https://doi.org/10.1038/s41558-018-0356-x.
Monaghan, A. J., and Coauthors, 2006: Insignificant change in Antarctic snowfall since the International Geophysical Year. Science, 313, 827–831, https://doi.org/10.1126/science.1128243.
Monaghan, A. J., D. H. Bromwich, W. Chapman, and J. C. Comiso, 2008: Recent variability and trends of Antarctic near-surface temperature. J. Geophys. Res., 113, D04105, https://doi.org/10.1029/2007JD009094.
Mottram, R., and Coauthors, 2021: What is the surface mass balance of Antarctica? An intercomparison of regional climate model estimates. Cryosphere, 15, 3751–3784, https://doi.org/10.5194/tc-15-3751-2021.
Nicolas, J. P., and D. H. Bromwich, 2014: New reconstruction of Antarctic near-surface temperatures: Multidecadal trends and reliability of global reanalyses. J. Climate, 27, 8070–8093, https://doi.org/10.1175/JCLI-D-13-00733.1.
Palerme, C., C. Genthon, C. Claud, J. E. Kay, N. B. Wood, and T. L’Ecuyer, 2017: Evaluation of current and projected Antarctic precipitation in CMIP5 models. Climate Dyn., 48, 225–239, https://doi.org/10.1007/s00382-016-3071-1.
Parker, S. P., 1980: McGraw-Hill Encyclopedia of Ocean and Atmospheric Sciences. McGraw-Hill, 580 pp.
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.
Rignot, E., J. Mouginot, B. Scheuchl, M. van den Broeke, M. J. van Wessem, and M. Morlighem, 2019: Four decades of Antarctic ice sheet mass balance from 1979–2017. Proc. Natl. Acad. Sci. USA, 116, 1095–1103, https://doi.org/10.1073/pnas.1812883116.
Saha, S., and Coauthors, 2014: The NCEP Climate Forecast System version 2. J. Climate, 27, 2185–2208, https://doi.org/10.1175/JCLI-D-12-00823.1.
Thomas, E. R., G. J. Marshall, and J. R. McConnell, 2008: A doubling in snow accumulation in the western Antarctic Peninsula since 1850. Geophys. Res. Lett., 35, L01706, https://doi.org/10.1029/2007GL032529.
Thomas, E. R., and Coauthors, 2017: Regional Antarctic snow accumulation over the past 1000 years. Climate Past, 13, 1491–1513, https://doi.org/10.5194/cp-13-1491-2017.
Thomas, E. R., J. S. Hosking, R. R. Tuckwell, R. A. Warren, and E. C. Ludlow, 2015: Twentieth century increase in snowfall in coastal West Antarctica. Geophys. Res. Lett., 42, 9387–9393, https://doi.org/10.1002/2015GL065750.
Trenberth, K. E., and D. P. Stepaniak, 2001: Indices of El Niño evolution. J. Climate, 14, 1697–1701, https://doi.org/10.1175/1520-0442(2001)014<1697:LIOENO>2.0.CO;2.
Turner, J., T. Phillips, J. S. Hosking, G. J. Marshall, and A. Orr, 2013: The Amundsen Sea low. Int. J. Climatol., 33, 1818–1829, https://doi.org/10.1002/joc.3558.
van Wessem, J. M., and Coauthors, 2018: Modelling the climate and surface mass balance of polar ice sheets using RACMO2—Part 2: Antarctica (1979–2016). Cryosphere, 12, 1479–1498, https://doi.org/10.5194/tc-12-1479-2018.
Wang, Y., and Coauthors, 2016: A comparison of Antarctic ice sheet surface mass balance from atmospheric climate models and in situ observations. J. Climate, 29, 5317–5337, https://doi.org/10.1175/JCLI-D-15-0642.1.
Wang, Y., and Coauthors, 2017: Snow accumulation variability over the West Antarctic ice sheet since 1900: A comparison of ice core records with ERA-20C reanalysis. Geophys. Res. Lett., 44, 11 482–11 490, https://doi.org/10.1002/2017GL075135.
Wang, Y., B. Huai, E. R. Thomas, M. R. van den Broeke, J. M. van Wessem, and E. Schlosser, 2019: A new 200‐year spatial reconstruction of West Antarctic surface mass balance. J. Geophys. Res. Atmos., 124, 5282–5295, https://doi.org/10.1029/2018JD029601.
Wang, Y., S. Hou, M. Ding, and W. Sun, 2020: On the performance of twentieth century reanalysis products for Antarctic snow accumulation. Climate Dyn., 54, 435–455, https://doi.org/10.1007/s00382-019-05008-4.
Wang, Y., M. Ding, C. H. Reijmer, P. C. J. P. Smeets, S. Hou, and C. Xiao, 2021: The AntSMB dataset: A comprehensive compilation of surface mass balance field observations over the Antarctic Ice Sheet. Earth Syst. Sci. Data, 13, 3057–3074, https://doi.org/10.5194/essd-13-3057-2021.
Winkelmann, R., A. Levermann, M. A. Martin, and K. Frieler, 2012: Increased future ice discharge from Antarctica owing to higher snowfall. Nature, 492, 239–242, https://doi.org/10.1038/nature11616.
