SEARCH

SEARCH BY CITATION

Keywords:

  • CloudSat;
  • East Antarctic mass increase;
  • GRACE;
  • high precipitation events

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion and Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[1] An improved understanding of processes dominating the sensitive balance between mass loss primarily due to glacial discharge and mass gain through precipitation is essential for determining the future behavior of the Antarctic ice sheet and its contribution to sea level rise. While satellite observations of Antarctica indicate that West Antarctica experiences dramatic mass loss along the Antarctic Peninsula and Pine Island Glacier, East Antarctica has remained comparably stable. In this study, we describe the causes and magnitude of recent extreme precipitation events along the East Antarctic coast that led to significant regional mass accumulations that partially compensate for some of the recent global ice mass losses that contribute to global sea level rise. The gain of almost 350 Gt from 2009 to 2011 is equivalent to a decrease in global mean sea level at a rate of 0.32 mm/yr over this three-year period.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion and Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[2] Changes in the continental ice mass are the most imminent contribution to future sea level rise in a warming climate. The major ice sheets of Greenland and Antarctica contribute significantly to present-day sea level change [Rignot et al., 2011, 2008; Velicogna, 2009]. In recent years, the total contribution from both ice sheets (excluding surrounding glaciers and ice caps) is estimated to have contributed to 1.3 ± 0.4 mm/yr sea level rise [Rignot et al., 2011].

[3] Over the past decade, observations of the Antarctic ice sheet revealed that West Antarctica and the Antarctic Peninsula are primarily affected by increases in glacial discharge, whereas East Antarctica remains fairly stable [Rignot et al., 2008]. However, strong decadal and interannual variations in the mass balance that are related to highly inhomogeneous regional variations in glacial discharge and opposing solid precipitation potentially mask trends and accelerations in ice sheet melting in East Antarctica [Rignot et al., 2011].

[4] Antarctic temperatures are projected to increase with increasing atmospheric carbon dioxide content. Despite these higher atmospheric temperatures, most of continental Antarctica is projected to remain below freezing temperature, and therefore runoff from surface melting is negligible [Van den Broeke et al., 2011; Kuipers Munneke et al., 2012]. Climate model simulations of global warming also project long-term increases in precipitation over the Antarctic ice sheet, an expected consequence of the higher atmospheric moisture content in a warming climate.

[5] Recent studies are inconclusive about whether projected effects of warming are already detectable in current datasets on snowfall and thus on the Antarctic mass balance. Reanalysis and ice core data show no considerable change in precipitation since the 1950s [Monaghan et al., 2006]. This contradicts studies based on altimetry observations which indicate an increase in surface height over East Antarctica since the 1990s [Davis et al., 2005]. While discrepancies are potentially due to differences in data coverage and measurement technique, the contradicting findings underline the importance of constant monitoring of precipitation changes over the ice sheet and their impact on the total mass balance.

[6] In general, precipitation over Antarctica decreases from the low-altitude coast to the inland plateau [Giovinetto and Bentley, 1985; Vaughan et al., 1999]. Local minima and maxima in precipitation correspond to the leeward and windward sides of topographical ridges, respectively. Interannual variability in Antarctic snowfall is high and attributed to fluctuations in cyclone activity that is associated with the major modes of variability, notably the Southern Annular Mode (SAM) and the El Nino Southern Oscillation (ENSO). Stronger circumpolar Westerlies, related to a positive SAM trend over the past decades, impact cyclogenesis [Sinclair et al., 1997] over the Southern Ocean, and lead to increased precipitation over the circumpolar trough [Noone and Simmonds, 2002]. ENSO has been associated with blocking events induced by Rossby wave trains propagating from the tropics to the Southern Ocean. Increased poleward flow along the western flank of the blocking high pressure system leads to significant changes in synoptic weather along the coast [Hirasawa et al., 2000].

[7] While observational coverage of high-latitude regions was limited in the past, recently launched satellites provide new perspectives on the present-day ice sheet changes. In addition to surface height measurements from altimetry, which require an accurate estimate of firn density to infer mass and have accuracy limitations near coastal regions, gravity measurements over the past decade provide direct measurements of the net ice sheet mass changes. Observations from the Gravity Recovery And Climate Experiment (GRACE) yield unprecedented insights into spatial and temporal evolution of the ice mass [Velicogna, 2009; Rignot et al., 2011; Sasgen et al., 2010]. The CloudSat mission [Stephens et al., 2008] was launched in 2006 and the CloudSat radar has a unique capability for detecting snowfall [Liu, 2008]. The combination of both GRACE ice mass change and CloudSat snowfall thus offers an opportunity to gain new insights on changes to Antarctic ice mass described below.

[8] This study reports on observations of a substantial mass increase in Eastern Antarctica that began in 2009. CloudSat observations confirm re-analysis estimates of snowfall and these reveal that the regional ice sheet growth was related to high precipitation events along the coast of Dronning Maud Land. Further analysis indicates that atmospheric blocking over the Atlantic sector of Antarctica led to increased poleward moisture transport from the ocean to the region from 10 W to 70 E. The mass gain was produced primarily from several discrete precipitation events that occurred over two months in the years 2009 and 2011.

2. Data and Methods

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion and Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[9] We estimate Antarctic ice sheet mass changes using GRACE data derived from the JPL RL05 time variable gravity field solutions. The data have been corrected for geocenter motion using estimates by Swenson et al. [2008]; glacial isostatic adjustment is subtracted from the GRACE solutions using the model of Paulson et al. [2007]. The C2,0 spherical harmonic coefficients, describing the Earth's oblateness, derived from satellite laser ranging measurements using the estimates by Cheng and Tapley [2004] are substituted for the C2,0 coefficients in the GRACE product. The spherical harmonics are filtered using the method by Swenson and Wahr [2006] and additionally smoothed with a 300 km Gaussian. The gravity changes, expressed in spherical harmonic coefficients. were converted to equivalent changes in ice mass following Wahr et al. [1998]. Error estimates for the monthly GRACE fields are based on the method of Wahr et al. [2006]. For comparison, an estimate of the local ice sheet mass change from CSR RL05 is shown in Figure S2 in the auxiliary material.

[10] To obtain estimates of net precipitation, we combine monthly means of precipitation and evaporation from the ERA Interim re-analysis [Dee et al., 2011]. We use the same re-analysis products for daily fields of geopotential height at 500 mbar at a horizontal resolution of 1.5 × 1.5 degrees.

[11] In addition to the re-analysis output, we compare our findings to precipitation estimates based on CloudSat data. The mean snowfall rate is estimated from the CloudSat [Stephens et al., 2008] cloud profiling radar, which is extremely sensitive to the occurrence of precipitation of all phases due to the high sensitivity of the radar. Snowing pixels are identified using the ‘snow certain’ determination in the 2C-Precip-Column (Release 04) product, which flags the occurrence and determines the phase of precipitation. The algorithm first identifies the phase of the precipitation based on the 2-meter air temperature derived from the European Center for Medium range Weather Forecasting (ECMWF) weather analysis. FollowingLiu [2008], pixels with temperatures less than 2C at 2-meters altitude are considered snowing. Subsequent to phase identification the likelihood of precipitation is determined using a temperature dependent reflectivity threshold applied to the sixth radar range gate above the surface and vertical continuity tests to mitigate the effects of surface clutter.

[12] After snow certain pixels are identified the more challenging task of quantifying the snowfall intensity is undertaken using a simple approach. Snowfall intensity (S) is estimated using the reflectivity (Z) in the sixth range bin above the surface (∼1300 m) using a straightforward Z-S relationship of the form Z = aSb. Some complicating factors for quantifying snowfall rate include (a) uncertainties in ice crystal habit, (b) uncertainties in the ice Particle Size Distribution (PSD), (c) multiple scattering of the radar beam, and (d) attenuation of the radar beam. Matrosov and Battaglia [2009]show that multiple scattering (c) and attenuation (d) effects largely cancel each other in snowfall so these effects are neglected here. Uncertainty resulting from crystal habit and PSD are the dominant error sources because of the strong sensitivity of the radar reflectivity to particle size and shape. To account for these primary uncertainties sources, snowfall intensity is calculated using three Z-S relationships followingHiley et al. [2011]who use a wide range of theoretically possible ice crystal habits to calculate a best-fit Z-S with uncertainty bounds. Note that this provides an upper bound on uncertainty because it is derived from a diverse collection of ice crystal habits some of which are unlikely to occur in nature. The regional monthly mean snowfall intensities are calculated within 1° latitude/longitude bins, weighting by their respective areas, and integrated over the study area resulting in a mean snowfall rate.

3. Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion and Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[13] Between 2003 and 2011, GRACE observed mass loss of the West Antarctic Ice Sheet and along the Antarctic Peninsula Trend maps reveal that the mass loss in West Antarctica is primarily located around Pine Island Glacier (PIG) (Figure 1). While several studies [e.g., Rignot et al., 2008; Velicogna, 2009] indicate no significant changes in the East Antarctic ice sheet between 2002 and 2009, GRACE recently observes mass gain along the coast (Figure S1). A spatial average over the region where this mass gain is observed (30 W–60 E, 65 S–80 S) shows a relatively stable mass budget from 2003–2008, followed by a strong increase that began in 2009 (Figure 2). The cumulative mass gain observed by GRACE between early 2009 and 2011 is approximately 350 Gt, which is equivalent to about 0.32 mm/yr global sea level decrease.

image

Figure 1. GRACE mass trends from 2004 to 2011 in cm/yr equivalent water column height change. Only values within the land mask and trends at a 95% significance level are shown.

Download figure to PowerPoint

image

Figure 2. (top) GRACE mass average over 30W–60E, 65S–80S (green) compared to integrated net precipitation from ERA Interim (red) and CloudSat accumulated snowfall (black). (bottom) ERA Interim net precipitation (black) compared to CloudSat snowfall accumulation (red) and ERA interim precipitation (blue dashed). Gray shading indicates CloudSat error bars.

Download figure to PowerPoint

[14] These regional mass increases may be caused either by a) surface mass balance variations due to precipitation and snow drift or b) ice dynamic changes. The rapidity of the observed increase suggests that an increase in precipitation is the most likely explanation for the mass gain since ice dynamical processes are usually slower to unfold [e.g., Rignot, 2006] and snowdrift has only a small contribution to interannual changes in the surface mass balance [Lenaerts et al., 2012]. We evaluate this hypothesis by first comparing the mass gain that results from the accumulated snowfall observed by CloudSat. We then use this snowfall to verify the mass accumulation deduced from the precipitation contained in ERA Interim re-analysis. Once verified, we use reanalysis data to show how precipitation changes induced by atmospheric circulation changes explain the observed sudden increases in ice mass.

[15] The increase in mass due to accumulated precipitation is equivalent to the integral over time of net precipitation in a region. The snowfall observations of CloudSat confirm the anomalously large accumulation of snow over Dronning Maud Land starting in 2009 (Figure 2, top). The monthly CloudSat and reanalysis precipitation time series are well correlated (r = 0.63) over the CloudSat period. This suggests that the anomalous mass gain observed by GRACE is primarily a result of excess precipitation during the period between 2009 and 2011, whereas ice dynamical processes in this region have a rather small contribution. This is further confirmed in reanalysis data. CloudSat precipitation estimates and the re-analysis model output (Figure 2) are very similar when integrated over the multi-year period, a process that naturally reduces the sampling noise inherent in the observations. The accumulation of net precipitation anomaly over the region of interest derived from the ERA Interim re-analysis also resembles the mass time series from GRACE (Figure 2) for the entire GRACE period. All three estimates of mass accumulation agree within the uncertainty of the respective datasets. The mass increase from ERA Interim's forecasted net precipitation fields agree to within 10% with the mass increase based on the atmospheric moisture convergence fields from the Japanese JRA-25 re-analysis (not shown [Onogi et al., 2007; Landerer et al., 2010]) which suggests that sublimation has little effect on the accumulated mass.

[16] Given the good agreement between the re-analysis and CloudSat precipitation and the overall consistency between the snowfall information and GRACE mass anomalies, we use the re-analysis data to place the 2009–2011 anomalies in a longer-term context. The longer re-analysis time series demonstrates that the mass accumulation in 2009–2011 is exceptional over this particular coastal region compared to the three preceding decades (Figure 2). While the snow accumulation shows interannual fluctuations of ±50 Gt before 2009, over the past 3 years the mass increases by about 350 Gt. Both time series of precipitation rates from the ERA Interim re-analysis and CloudSat suggest that the high snowfall events leading to the mass accumulation primarily occurred in May 2009 and June 2011. The precipitation in these two months is 5–6 times higher than the standard deviation of the ERA-Interim time series up to 2008 (Figure 2, bottom). Because the evaporation anomaly is small and ice dynamical process are presumed to act at longer time-scales, we attribute the GRACE mass anomaly in East-Antarctica to these two distinct months with anomalously high precipitation.

[17] To determine the origin of the snowfall anomalies occurring in 2009 and 2011, we analyze the synoptic-scale snowfall variability in May, 2009 and June, 2011. A statistical analysis indicates that the majority of snowfall in these two months can be attributed to 5 periods of several days each, 77% of the precipitation over Dronning Maud Land in May 2009 occurred during the periods of May 6–7 (∼15%), May 17–20 (∼28%) and May 24–27 (∼34%). In June 2011 the highest amounts of snowfall are observed during June 19–21 (∼20%) and June 23–28 (∼43%). During these 9 days snowfall amounted to 63% of total June 2011 precipitation.

[18] Figure 3 shows the spatial patterns of maximum snowfall during these periods. Regions of high precipitation are clearly restricted to the coast along Dronning Maud Land. This spatial distribution is consistent with findings by Schlosser et al. [2008]who showed that while the intensity in snowfall is highly variable, precipitation is mostly limited to the low-altitude coastal areas decreasing toward the higher altitude inland plateau. In conjunction with these high snowfall events, a significant change in the atmospheric pressure fields also occurred over Antarctica and the ocean north of Dronning Maud Land (Figure 3). A seesaw pattern of high and low pressure systems encircles the continent during the periods of the precipitation events in May 2009 and June 2011. A dipole pattern of low and high pressure intersects the continent and induces a strong pressure gradient over Dronning Maud Land. High-pressure systems are associated with an anticyclonic wind circulation that induces a poleward flow along their western flanks. These anomalous pressure patterns suggest that the northerly winds had driven warm and moist air to the continent inducing cloud formation and subsequent precipitation.

image

Figure 3. Daily precipitation (color shading) and geopotential height at 500 mbar (contours) in (left) May, 2009 and (right) June, 2011. Periods with high snowfall are May, 6–7, 17–20, 24–27, 2009 and June, 19–21, 23–27, 2011.

Download figure to PowerPoint

[19] In summary, the analysis of synoptic scale precipitation and sea level pressure indicates that the stable and strong pressure patterns over periods of several days in May 2009 and June 2011, have led to increased moisture flux toward the Antarctic coast that resulted in anomalously high precipitation events. Most of the increase in precipitation in 2009 and 2011 was related to these events, which suggests that the primary driver for the observed mass increase was the prolonged changes in pressure patterns and induced poleward wind in the two years.

4. Discussion and Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion and Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[20] We observed an abrupt mass increase in East Antarctica along the coast of Dronning Maud Land in the GRACE satellite data in 2009–2011. The analysis of precipitation from re-analysis and CloudSat satellite data indicates that the mass gain was caused by anomalously high precipitation events during the Southern Hemisphere winter. Snowfall accumulation derived from ERA Interim precipitation and CloudSat observations is nearly equivalent to the mass increase. Evaporation has only a small impact on the accumulation. Approximately 350 Gt of snow accumulated during the period from 2009 to 2011. This increase in ice sheet mass is equivalent to about 0.32 mm/yr of decrease in global sea level. Recent studies have estimated Antarctic mass loss to amount to 0.4 mm/yr equivalent sea level between 2003 and 2009 [Velicogna, 2009]. The mass loss from the entire ice sheet between 2009 and 2011 amounts to about 325 Gt (not shown) or 0.3 mm/yr equivalent sea level. Hence, despite the regional mass gain the total amount of mass loss remained comparable to the rate between 2003 and 2009.

[21] Further analysis of the precipitation data indicated that the mass gain primarily occurred during May-2009, and June-2010, two months with exceptionally high precipitation. Most of the accumulation in these months resulted from snowfall from only a few main events (three in 2011 and two in 2011). Analysis of synoptic scale precipitation and sea level pressure indicates that these events were associated with anomalous atmospheric blocking events led to an increased poleward flow from the ocean to the coast of Dronning Maud Land.

[22] The events are similar to the abrupt changes in weather conditions observed at Dome Fuji in 1997 where a persistent Rossby wave train originating in the subtropics led to wintertime blocking and sudden change in atmospheric conditions [Hirasawa et al., 2000]. The temperature increased by 40 C and the nearly cloud-free condition in the pre-blocking period was suddenly turned into complete overcast skies with relatively thick clouds in the two-day period of the blocking formation.Hirasawa et al. [2000]suggest that the event in June-1997 was related to the formation of the exceptionally strong El Nino in 1997/98, inducing a persistent Rossby wave train to high latitudes.

[23] Our data suggests that a similar mechanism led to the high precipitation events in 2009 and 2011. The seesaw pattern in geopotential height around Antarctica with a dipole over the Ross and Weddell Sea is indicative of the typical impact of ENSO on atmospheric conditions over the Southern Ocean and Antarctica [Karoly, 1989]. The strong Central Pacific El Nino in 2009/10 and subsequent La Nina 2010/11 potentially induced a similar wintertime blocking over Dronning Maud Land as observed during the 1997/98 El Nino at Dome Fuji.

[24] While the analysis of the atmospheric conditions during the snowfall events suggests the possibility of ENSO induced Rossby wave trains having played a role in drastically changing synoptic weather patterns over coastal Antarctica and the adjacent Southern Ocean, it remains unclear why these particular events would have caused such a dramatic change. The ERA Interim re-analysis precipitation indicates that no significant change in snowfall frequency or strength has occurred in the period between 1979–2008 even though several ENSO events occurred in the same period, including the strong Eastern Pacific El Nino in 1997/98 influencing synoptic weather in Dome Fuji. Whether the 2009 and 2011 events are related to recent changes in the flavor of El Nino [e.g.,Lee and McPhaden, 2010] causing regionally different responses will be subject of follow-up studies.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion and Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[25] We would like to thank two anonymous reviewers for their helpful comments and suggestions. The work was performed at the Jet Propulsion Laboratory, California Institute of Technology, under contract with NASA and is partially supported through NASA contract NMO710771 to JPL. We thank the GRACE analysis centers at University of Texas, JPL, and Geoforschungszentrum Potsdam, and the German Space Operations Center (GSOC) of the German Aerospace Center (DLR).

[26] The Editor thanks two anonymous reviewers for their assistance in evaluating this paper.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion and Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion and Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

Auxiliary material for this article contains two figures.

Auxiliary material files may require downloading to a local drive depending on platform, browser, configuration, and size. To open auxiliary materials in a browser, click on the label. To download, Right-click and select “Save Target As…” (PC) or CTRL-click and select “Download Link to Disk” (Mac).

Additional file information is provided in the readme.txt.

FilenameFormatSizeDescription
grl29727-sup-0001-readme.txtplain text document1Kreadme.txt
grl29727-sup-0002-fs01.epsPS document928KFigure S1. Difference of annual mean ice mass from GRACE expressed in equivalent water column height.
grl29727-sup-0003-fs02.epsPS document68KFigure S2. GRACE mass average over 30W–60E, 65S–80S derived from JPL RL05 and CSR RL05 compared to integrated net precipitation from ERA Interim.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.