Spatial and temporal variability in snow ... - Semantic Scholar

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, D23102, doi:10.1029/2008JD010235, 2008

for

Full Article

Spatial and temporal variability in snow accumulation at the West Antarctic Ice Sheet Divide over recent centuries J. Ryan Banta,1 Joseph R. McConnell,1 Markus M. Frey,2,3 Roger C. Bales,2 and Kendrick Taylor1 Received 7 April 2008; revised 21 July 2008; accepted 19 August 2008; published 10 December 2008.

[1] Ice cores collected in 2000 (ITASE 00-1) and 2005 (WDC05A, WDC05Q) from the

West Antarctic Ice Sheet Divide (WAIS Divide) project site were used to investigate the spatial and temporal variability in accumulation. The ice cores were dated based on annual layer counting of multiple glaciochemical measurements resulting in bottom depth ages for WDC05A, WDC05Q, and ITASE 00-1 of 1775, 1521, and 1653 A.D., with mean annual accumulation rates of 0.200, 0.204, and 0.221 mweq a 1, respectively. Small-scale spatial variability (SSV) was determined using an analysis of variance of accumulation in the ice core array, thereby quantifying the uncertainty in individual accumulation records. Results indicate that the spatial variability was 0.030 mweq a 1, or approximately 15% of the average annual accumulation. An accumulation record representative of the WAIS Divide local area over recent centuries was developed using a principal component analysis to identify the coherent accumulation signal. The WAIS Divide local record exhibited 14% interannual variability (1 standard deviation of the mean) with the SSV reduced to 0.017 mweq a 1. Correlations of the WAIS Divide local accumulation record with atmospheric indices (e.g., Antarctic Oscillation) exhibited periods when the records oscillate in and out of phase. Thus, reconstructing local and global atmospheric indices from WAIS Divide accumulation records over recent centuries may prove problematic. Citation: 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, doi:10.1029/2008JD010235.

1. Introduction [2] Developing high-resolution regionally representative long-term accumulation records from ice cores which can go back thousands of years are fundamental to better understanding the role of Antarctica in the context of global climate change. Accurate historical records of net accumulation, defined as precipitation minus evaporation, are especially important in the west Antarctic region, which is undergoing significant changes in mass balance and glacier dynamics [Overpeck et al., 2006; Davis et al., 2005; Scambos et al., 2000]. While previous studies have shown that there has not been a significant change in snowfall over the continent for the past 50 years [Monaghan et al., 2006], studies have also indicated that certain regions of the Antarctic ice sheet such as the West Antarctic ice sheet and peninsula, have seen an increase 1 Desert Research Institute, Nevada System of Higher Education, Reno, Nevada, USA. 2 School of Engineering, University of California, Merced, California, USA. 3 Now at British Antarctic Survey, National Environment Research Council, Cambridge, UK.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2008JD010235$09.00

in annual accumulation [Kaspari et al., 2004; MosleyThompson et al., 1999], in some cases doubling since 1850 [Thomas et al., 2008]. Using radar depth-age scales, Neumann et al. [2008] found that accumulation in central West Antarctica decreased approximately linearly across the ice divide, however, accumulation rates in polar regions have been shown to vary significantly over short distances [Spikes et al., 2004; McConnell et al., 2000a]. Hence, more spatially representative historical accumulation records are necessary to further investigate accumulation variability in west Antarctica. [3] Temporal variability in Antarctic accumulation has been linked with past climate conditions. For example, accumulation rates have been correlated with the El Ninõ– Southern Oscillation (ENSO), however, these correlations can be highly variable in strength and sign [e.g., Frey et al., 2006; Kaspari et al., 2004; Bromwich et al., 2000]. Van den Broeke and van Lipzig [2004] used a climate model to demonstrate that the Antarctic Oscillation (AAO) can strongly affect Antarctic accumulation, though the effects vary spatially. [4] In this study, we examine the temporal and spatial trends of accumulation at the West Antarctic Ice Sheet Divide project site area (WAIS Divide) over recent centuries, and explore links between accumulation and atmospheric

D23102

1 of 8

BANTA ET AL.: ACCUMULATION VARIABILITY AT WAIS DIVIDE

D23102

Figure 1. Map of Antarctica showing ice core locations. The West Antarctic Ice Sheet (WAIS) Divide ice cores (WDC05A and WDC05Q) are approximately 100 m apart, and ITASE 00-1 is approximately 20 km to the northwest. Figure modified from Conway et al. [2005]. processes. High-resolution trace element glaciochemical records were used to develop two new accumulation records from WAIS Divide as well as a reanalysis of accumulation from a previously collected nearby ice core (International Trans-Antarctic Scientific Expedition (ITASE) 00-1). To gain a more complete understanding of the climatic processes affecting WAIS Divide, a locally representative accumulation record was developed from the array of three ice cores and compared with atmospheric indices. The uncertainty in the WAIS Divide local accumulation record due to the effects of small-scale spatial variability was quantified using an analysis of variance.

2. Methods [5] Two ice cores were collected in 2005 using a 10 cm electromechanical drill from the West Antarctic Ice Sheet Divide project site (WAIS Divide), located between the Pine Island– Thwaites and Ross drainage basins: WDC05A and WDC05Q (hereafter also referred to as WAIS cores; Figure 1 and Table 1). The ice cores, approximately 1 km apart, were analyzed to 70 m and 130 in depth, respectively, corresponding to the past 230 and 484 years. Following standard procedures, the ice cores were measured, weighed and processed in the field in order to calculate depthdensity profiles. The WAIS ice cores were shipped to the National Ice Core laboratory (NICL), Denver, CO where they were cut and then shipped to the Desert Research

D23102

Institute (DRI), Reno, NV for laboratory analysis. Both the NICL and DRI facilities maintained a constant ambient temperature of 20°C or below while storing the ice cores prior to laboratory analysis. An additional 105 m deep ice core located 20 km to the northwest (ITASE 00-1) was collected in 2000 [Kaspari et al., 2004]. We reanalyzed the ITASE 00-1 core at the Desert Research Institute using the same laboratory methodology as the WAIS cores, which differs from the previous studies. [6] In order to obtain high-fidelity, high-resolution glaciochemical records, the ice cores were analyzed using the Continuous Flow Analysis – Trace Elements Dual (CFATED) methodology (adapted from McConnell et al. [2002]). The CFA-TED analytical system allows for (1) very high depth resolution (approximately 0.01 m water equivalent per sample corresponding to an average of 21 samples per year for these ice cores); (2) measurements of a broad spectrum of elemental and chemical species (>20); (3) exact depth coregistration of all measurements; and (4) very low detection limits (detection limits for some elements are in the low parts per quadrillion range). [7] Developing accurate accumulation records from ice cores requires accurate identification of annual layers. Previous studies have demonstrated that several glaciochemical parameters exhibit seasonal cycles [e.g., Banta and McConnell, 2007; Frey et al., 2006; Dixon et al., 2005; McConnell et al., 1998]. However, accumulation rates of individual years may differ by as much as 50% depending on the seasonal characteristics of the parameter used to develop the depth-age scale [Anklin et al., 1998]. Consistent layer identification is best accomplished when robust seasonal indicators, such as the midwinter H2O2 concentration minima or the summer non– sea salt sulfur (nssS) concentration maxima, are combined with multiple annually varying parameters to resolve any ambiguities [e.g., Banta and McConnell, 2007; Frey et al., 2006; McConnell and Bales, 2004]. The use of multiple parameters is especially important if the total accumulation for a given year is low or if the chemical seasonal signal is dampened for one or more years. [8] The WAIS and ITASE 00-1 ice cores were dated using annual layer identification based on multiple seasonal markers, with nssS to sodium ratio (nssS/Na) as the primary indicator. Seasonal cycles of hydrogen peroxide were not well preserved below the top 12 m. The nssS/Na annual cycle covaries in time with well known seasonal tracers such as hydrogen peroxide and dust tracers, with maxima occurring during the summer and minima occurring midwinter [Dixon et al., 2005; McConnell and Bales, 2004]. Where the nssS/Na annual layers were ambiguous (e.g., a low accumulation year, a doublet peak in one year, or due to

Table 1. Summary of Ice Core Accumulation Recordsa

Site

Latitude (°S)

Longitude (°W)

Elevation (m)

Mean Annual Accumulation (mweq a 1)

WDC05A WDC05Q ITASE 00-1

79.463 79.468 79.383

112.125 112.086 111.239

1759 1759 1791

0.200 ± 0.034 0.204 ± 0.035 0.221 ± 0.041

a

Symbol ± represents the range (1 s) of the annual values.

2 of 8

Time Span (years)

Coefficient of Variation

1775 – 2005 1521 – 2005 1653 – 2000

0.18 0.19 0.21

D23102


D23102

Figure 2. Accumulation records of WDC05A, WDC05Q, and ITASE 00-1 (mweq a 1). The bold line represents a three-point running average filter.

volcanic fallout), annual layers of sea salt and dust tracers were used to aid in dating the ice cores (e.g., Mg, Sr, Ca, Al, Ce). [9] Additional cores from across the Antarctic ice sheet were previously analyzed and dated using this methodology. Owing to very high accumulation rates at these other sites, there was minimal ambiguity in the dating (e.g., Gomez accumulation rate = 0.72 mweq a 1 [Thomas et al., 2008], James Ross Island accumulation rate = 0.57 mweq a 1 [McConnell et al., 2007]). Accumulation records from other high accumulation sites are also known to a high level of confidence (e.g., Law Dome accumulation rate = 0.7 mweq a 1, updated from Palmer et al. [2001]). The glaciochemical records from these sites were used for comparisons to reduce uncertainties in the WAIS and ITASE 00-1 depth-age scales. The depth-age scale was verified in each ice core using nssS and conductivity peaks that are associated with known volcanic eruptions. The final dating uncertainty for the three ice cores was estimated to be less than one year. [10] A seventh-order polynomial depth-density model was fit to the field generated density measurements to convert snow depth to annual accumulation in water equivalent (reported in mweq a 1). An additional 30 m ice core (WDC05E; T. Sowers, personal communication, 2008) and 3 m snow pit samples (Science Coordination Office, personal communication, 2008) were collected in the immediate area of the WDC05Q core. These additional density profiles were used to help constrain the WAIS depth-density curve in the shallower depths where the density measurement uncertainties are greatest. The ITASE 00-1 depthdensity profile was also generated from field measurements of that ice core as well as from a nearby 1 m snow pit using an eighth-order polynomial model. Because the WAIS and ITASE 00-1 ice cores were less than 4% of the ice sheet thickness, they were not corrected for vertical strain rates or ice flow. While not incorporated here, thinning is estimated to be no more than 3.0% of the annual layer thickness at a

depth of 130 m using a Dansgaard Johnsen model (E. Waddington, personal communication, 2008).

3. Results [11] On the basis of the multiparameter dating, the bottom depths of the WDC05A, WDC05Q, and ITASE 00-1 cores were determined to be 1775, 1521, and 1653 A.D., respectively, with mean annual accumulation rates (±1 standard deviation) of 0.200 ± 0.036, 0.204 ± 0.038, and 0.221 ± 0.046 mweq a 1, respectively (Table 1 and Figure 2). The difference in accumulation between the two WAIS records is largely due to the difference in length of the records. Over the common 229 year period, the accumulation rates for both cores were 0.200 mweq a 1. The resulting accumulation records of all three ice cores were approximately normally distributed (Chi squared analysis). [12] An increasing trend of 0.005 m per 100 years was identified over the length of the ITASE 00-1 accumulation record (accounting for autocorrelation [Santer et al., 2000]; Mann-Kendall analysis, statistically significant at the 95% confidence level [CL]). A lesser decreasing trend was identified in the WDC05Q record ( 0.004 m per 100 years or 10% over the length of the record, 95% CL incorporating autocorrelation), though much of the decrease occurred between 1650 and 1725 (average accumulation was 0.215 and 0.198 mweq a 1 for pre 1650 and post 1725, respectively). Prior to 1650 and after 1725, no significant trends were identified in WDC05Q. This matches the finding that the WDC05A accumulation record (which only goes back to 1775) did not exhibit a trend. While not included here, the estimated 3% thinning over 130 m snow depth accentuates the decreasing trend in the WDC05Q record to approximately 13% and reduces the increasing trend in the ITASE 00-1 record to below statistical significance.

3 of 8


D23102

Table 2a. Correlation Matrix (Pearson’s r) for 1775 – 1999 of Normalized Average Annual Accumulationa WDC05Q WDC05A ITASE 00-1

WDC05Q

WDC05A

ITASE 00-1

1.00 0.44 0.39

1.00 0.35

1.00

a

All correlations are significant at the 99% confidence level.

[13] The temporal variability, defined here as the coefficient of variation (CV = standard deviation/mean), was 0.18, 0.19 and 0.21 for WDC05A, WDC05Q, and ITASE 00-1, respectively (Table 1). While higher CVs were identified for short periods, long-term trends in temporal variability were not present over the length of the records. For example, the coefficient of variation at ITASE 00-1 for the 1725 – 1750 period was 0.17, 0.25 for the 1750 – 1775 period, and 0.17 from 1775 to 1800. The WAIS cores exhibited lesser fluctuations in CVs, with the maxima/ minima occurring during different time periods. Thus, short-term fluctuations in temporal variability are likely caused from local small-scale spatial variability rather than regional-scale atmospheric processes.

4. Discussion 4.1. Temporal Variability [14] The measured accumulation rates for the WAIS and ITASE 00-1 cores agree well with previous west Antarctica accumulation maps [e.g., Arthern et al., 2006; Monaghan et al., 2006], as well as with previously reported accumulation rates for the ITASE 00-1 core [Frey et al., 2006; Kaspari et al., 2004]. The difference in accumulation between the WAIS and ITASE 00-1 cores is also in agreement with radar accumulation studies [e.g., Neumann et al., 2008] who found an approximately linear decreasing trend in accumulation across the West Antarctic ice divide. The increasing trend in accumulation over the length of the core at ITASE 00-1 matches closely with the previously reported accumulation record (7% versus 8%, respectively for the 1653 – 1999 period) and is consistent with other ITASE accumulation rates in the Pine Island – Thwaites drainage basin [Kaspari et al., 2004]. Kaspari et al. [2004] also reported a 7% increase in accumulation rate for 1922 – 1991, however, we did not find a statistically significant trend over that time period (Man-Kendall analysis, 95% CL). It is important to note that the previous accumulation study using the ITASE 00-1 ice core [Kaspari et al., 2004] was based on a completely independent glaciochemical analysis (ion chromatography), different seasonal markers (e.g., Na+, SO24 , NO3 , etc.), and a different depth-density model. Comparisons between our results and the previous study indicate the same number of years were identified over the full length of the record (bottom of the core was estimated to be 1653 A.D.). The correlation between the two records (Pearson’s r) was statistically significant (99% CL), and difference in the mean annual accumulation was < 0.001 mweq a 1 over the length of the record. [15] The coefficient of variation of the WAIS and ITASE 00-1 cores were in close agreement, suggesting that similar atmospheric circulation processes are affecting the local area. The common temporal precipitation signal of two

D23102

cores can be identified by examining the correlation coefficient between the annual accumulation rates. Though statistically significant, on an annual basis these intercore correlations often resulted in low correlations (Table 2a). This was likely due to variability in the timing of the nssS/ Na maxima from year to year (e.g., effects of depositional noise). Spatial variability can also contribute to low intercore correlations, as a few anomalously large and/or small accumulation years in one of these records can noticeably decrease the correlation, especially in the few meters near the surface where the firn density measurement uncertainties are greatest. For example, the annual Pearson’s r correlation between WDC05A and WDC05Q was 0.66 (99% CL) for 1850 – 1875, but only 0.44 (99% CL) over the common 225 year period. 4.2. Spatial Variability [16] Small-scale spatial variability (SSV), defined here as the incoherent signal due to blowing snow or sastrugi, has significant implications for ice core record interpretations, especially at sites where the SSV may constitute a significant percentage of the annual accumulation such as WAIS Divide. For instance, correlations between accumulation records and high-frequency events, such as the El Ninõ – Southern Oscillation with a 3- to 5-year periodicity, may be substantially reduced. Quantifying the SSV allows error bars to be placed on annual accumulation records, resulting in more meaningful comparisons with atmospheric processes. Using an approach described by McConnell et al. [2000b], we calculated the SSV for the WAIS Divide local area using the WAIS and ITASE 00-1 ice cores. For two adjacent ice cores, X and Y, the annual accumulation time series was defined as X(t) = P(t) + ex(t) and Y(t) = P(t) + ey(t), respectively; where P(t) was the coherent accumulation signal common to both ice cores and e(t) was the incoherent signal composed of spatial variability and methodological noise. The variability in the accumulation signal of the two adjacent cores was separated into a coherent and incoherent component, defined here as Var(P) = rxyVar(X) and Var(ex) = (1 rxy)Var(X); where rxy was the correlation coefficient between X and Y, Var(X) = variance of X(t) and Var(ex) = Var(ey) [McConnell et al., 2000b; Fisher et al., 1985]. Var(P) was calculated using the average of all possible rxy values between sites, and Var(e) was calculated using the mean squared variance from an analyses of variance. SSV was reported as Var(ex)0.5 (mweq a 1). For the 1775 – 2000 period common to the three ice cores, small-scale spatial variability was 0.030 mweq a 1, accounting for 61% of the annual variance which is equivalent to 15% of the annual accumulation rate. Previous studies have reported similar SSVs in polar regions, ranging between 0.029 and 0.040 mweq a 1 [Banta and McConnell, 2007; McConnell et al., 1997, 2000b]. 4.3. Local Accumulation [17] An accumulation record generated from an array of ice cores is more representative of the accumulation patterns for the region than a record from a single ice core because of the reduced impact of SSV. An accumulation record of the WAIS Divide project area was created from an array including WDC05A, WDC05Q, and ITASE 00-1 (hereafter referred to as the WAIS Divide local record). We did not

4 of 8

D23102


D23102

Figure 3. WAIS Divide local accumulation record (red) compared with ECMWF ERA40 estimated accumulation (1980 –1999, blue). Error bars represent the standard error of the mean from small-scale spatial variability, or SSV * n 0.5, where n is the number of cores in the array. Three-point running average filter applied to all data sets. include other ice cores further away as they were likely influenced by different hydrologic processes and/or were analyzed using different methodologies. For example, the Gomez core, located 1200 km from WDC05Q, experienced a doubling in accumulation over recent centuries [Thomas et al., 2008]. The WAIS Divide local record was generated from a principal component analysis (first eigenvector) using annual data for those years common to the ice cores. While the mean accumulation rates are very similar in the three ice cores, the annual data were normalized (Z = observation/mean) to ensure records with higher accumulation did not overly influence the composite record. The WAIS Divide local mean accumulation (1775 –1999) was 0.207 mweq a 1 and varied by approximately 14% of the mean (1 standard deviation). It did not exhibit a statistically significant linear trend (95% CL, accounting for autocorrelation). [18] The WAIS Divide local accumulation record generally agreed with the European Centre for Medium-Range Weather Forecasts reanalysis (ECMWF ERA-40) in the post satellite era (Spearman rank correlation, rs = 0.36, 90% CL; 1985 – 1999; Figure 3). Note that Spearman rank correlation coefficients (rs) were used here in place of Pearson’s r correlations because a Pearson’s r correlation is neither robust nor resistant, making it highly sensitive to data points with large leverage. ECMWF ERA-40 precipitation data from the modern satellite era (post-1979) have higher degrees of confidence than presatellite data [Bromwich et al., 2007; Bromwich and Fogt, 2004; Marshall, 2003]. ECMWF has also issued a statement that ‘‘1979 – 2001 [is] the period with the best and most time-consistent product quality for the globe as a whole’’ (http:// www.ecmwf.int/research/era). Hence, presatellite comparisons with ECMWF ERA-40 were not included in this study. The comparison also highlighted the uncertainty in the ECMWF ERA-40 data prior to 1985 where rs was not significant (90% CL; 1980 – 1985; Figure 3). While this is in the postsatellite era, Bromwich et al. [2007] point out that the bias correction scheme adjustment between and pre- and postsatellite data did not fully adjust until 1985.

[19] The SSV of the WAIS Divide local record was reduced by n 0.5, or to 0.017 mweq a 1, where n was the number of ice cores in the array. Arrays with more ice cores would further reduce the SSV (e.g., four ice core records are necessary to reduce the impact of SSV by 50%; nine ice cores are required to reduce it to 0.010 mweq a 1 or less than 5% of the average annual accumulation). Alternatively, SSV can also be reduced by temporal averaging (versus spatial). Multiyear filters dampen the incoherent noise, however, they also reduce the high-frequency coherent signal [van der Veen and Mosley-Thompson, 1999]. For example, averaging 4 years of annual accumulation from a single ice core will also reduce the SSV by 50%, however, this temporal average will also smooth out any 3- to 4-year periodicities. Longer decadal- and millennial-scale averages further reduce the noise (e.g., important for examining glacial versus interglacial periods), but such long averages also reduce the 5- to 7-year coherent signal, thereby preventing meaningful comparisons with many atmospheric processes (e.g., El Ninõ – Southern Oscillation). Hence, a balance is necessary between reducing incoherent noise and preserving higher-frequency information. Here, a threepoint running average filter was applied to the annual WAIS Divide local accumulation record (hereafter referred to as the filtered WAIS Divide local record; Figure 3). The improved signal-to-noise ratio in the filtered record was reflected by higher correlation coefficients of accumulation rates between sites (Table 2b). For this study, a three-point running average filter was the shortest filter width which (1) sufficiently preserved high-frequency events and (2) allowed identification of the coherent signal. Table 2b. Correlation Matrix (Pearson’s r) for 1775 – 1999 of Normalized Average Annual Accumulation, Three-Point Filter Applieda WDC05Q WDC05A ITASE 00-1

5 of 8

a

WDC05Q

WDC05A

ITASE 00-1

1.00 0.73 0.61

1.00 0.54

1.00

All correlations are significant at the 99% confidence level.

D23102


D23102

Figure 4. Comparison of the filtered WAIS Divide local accumulation record with the three dominant modes of atmospheric variability (AAO, SA1, SA2) and the Southern Oscillation Index (SOI) using decadal-scale Spearman rank correlations (three-point running average filter applied to all data sets). Dashed lines indicate 95% confidence level. [20] Following Kuhns et al. [1997], a signal-to-noise rxy). A variance ratio F was calculated as F = rxy/(1 high F ratio denotes a high degree of coherent signal between the ice cores, while an F ratio below 1 indicates that the variance in the ice core records are largely composed of incoherent signals. The average F ratio for the filtered WAIS accumulation records was 1.67, confirming that the common coherent precipitation signal can be readily identified. In addition, the SSV of the filtered WAIS Divide local record was reduced to 0.007 mweq a 1, or about 3% of the mean accumulation. 4.4. Comparison to Atmospheric Circulation Processes [21] The Antarctic Oscillation (AAO), also known as the Southern Hemisphere annular mode (SAM), is commonly defined as the first principal component of 850 hPa geopotential height anomalies from 20S to the South Pole [Thompson and Wallace, 2000; Thompson and Solomon, 2002]. The AAO accounts for 33% of the variance, while the second (SA1) and third (SA2) principle components explain 11% and 9%, respectively [Thompson and Wallace, 2000]. The AAO, SA1, and SA2 indices used in this study (as reported by Thompson and Wallace [2000]) were generated from the National Center for Atmospheric Research/National Center for Atmospheric Research (NCEP/ NCAR) reanalysis data. Similar to the ECMWF ERA-40 data, NCEP/NCAR reanalysis atmospheric pressure data have a much higher level of confidence in the post satellite era [Bromwich et al., 2007; Bromwich and Fogt, 2004; Marshall, 2003]. Hence, for this study, we examined the role of the AAO and subsequent modes over the entire length of the record, but primarily focused on the 1970s to present by calculating running decadal-scale Spearman rank correlations (three-point running average filter applied to all data sets, hereafter referred to as filtered). [22] The AAO is approximately centered over South Pole, the SA1 is centered off west Antarctica with a weaker opposite sign anomaly centered over the South Pole, and

SA2 exhibits several locations of activity over the oceans at 50°S [Thompson and Wallace, 2000]. It is expected that each mode should exhibit a relationship with the filtered WAIS Divide local accumulation record, given (1) the AAO accounts for a relatively large proportion of the atmospheric variability, (2) the center of action of the SA1 is located over west Antarctica, and (3) the SA2 center of actions are located over moisture bearing regions. The expected correlations are also supported by van den Broeke and van Lipzig [2004] who used a high spatial resolution climate model to identify significant correlations between West Antarctic precipitation and AAO related parameters (i.e., circumpolar vortex). Additionally, the Southern Oscillation Index, defined as the normalized surface pressure difference between Tahiti and Darwin, is likely to also correlate with Antarctic precipitation data. Using a principal component analysis of 500 hPa atmospheric pressure, Genthon et al. [2003] found that AAO and SOI are both present in the first component. However, they also found it took the first six components to reconstruct 60% of the precipitation. Hence, it is expected that both the AAO and SOI should correlate with precipitation, though the correlation may have low significance and/or only exhibit short periods of significance. [23] Using the 1958 – 1999 period, only the SA1 was found to be significantly correlated with the filtered WAIS accumulation record (rs = 0.37, 95% CL, three-point running average filter applied to all data sets). Correlations on decadal scales demonstrated that the filtered AAO was highly anticorrelated with the filtered WAIS Divide local accumulation record in the 1960s and early 1970s, followed by periods of positive correlations in the late 1970s to early 1980s and then switched back to anticorrelated in the mid 1980s to present (Figure 4). The correlation switched signs four times over the length of the record. Hence, while the overall correlation for 1958 – 2000 was negative, the relationship goes in and out of phase several times, even during the postsatellite period, where there is a higher degree of confidence in the atmospheric reanalysis data. The phase

6 of 8

D23102


reversal was even more pronounced in the comparison between the filtered SA1 and the filtered WAIS accumulation record, switching from rs < 0.60 (1976 –1986, 95% CL) to rs > 0.80 (1986 to present, 99% CL). The filtered SA2 exhibited a generally positive correlation with accumulation, though there are periods where the correlation drops to zero (e.g., 1980 – 1990). [24] Previous studies have seen similar phase reversals between accumulation rates in west Antarctica and atmospheric processes [e.g., Mayewski et al., 2004]. Bromwich et al. [2000] and Cullather et al. [1996] found that net precipitation just outside of the WAIS region was positively correlated with the Southern Oscillation Index (SOI) from 1980 to 1990 and anticorrelated with the SOI in the 1990s. Kaspari et al. [2004] did not find this relationship within the ITASE 00-1 ice core, but reported an overall anticorrelation between SOI and accumulation. Comparing the filtered WAIS Divide local accumulation record to the filtered SOI, we found a negative correlation (rs = 0.31, 99% CL for 1867– 1998; rs = 0.43 for 1958 – 1998, 99% CL; Figure 4). However, similar to the filtered AAO, the decadal-scale correlations switched signs multiple times throughout the record. For example, the correlation to the filtered SOI was highly negative (rs = 0.75, 95% CL) from 1990 to present with no statistically significant correlation in the 1980s (95% CL). Consistent with Kaspari et al. [2004], the filtered SOI was anticorrelated during periods of prolonged El Nino events (1940 –1942 and 1991 –1995). This is also consistent with Frey et al. [2006], who reported that the first principal component of accumulation from a 20 ice core array in west Antarctica showed correlations of both signs with SOI during the 1911 – 1993 period.

5. Conclusion [25] Three multicentury high-resolution accumulation records from the West Antarctic Ice Sheet (WAIS) Divide area presented here were developed using multiparameter annual layer counting. A WAIS Divide local accumulation record was generated from the array of ice cores, with an average accumulation rate of 0.207 mweq a 1, varying approximately 14% of the mean (1 standard deviation). The small-scale variability (SSV), which inhibits correlations to geophysical processes, was determined to be 0.030 mweq a 1 or 15% of the annual accumulation for an individual ice core, but was reduced to 0.017 mweq a 1 (8% of the accumulation rate) in the WAIS Divide local accumulation record. The SSV quantified here is consistent with previous studies, suggesting that it is likely driven by local-scale physical processes common to polar regions (e.g., sastrugi). Hence, caution is warranted when interpreting high temporal resolution records from a single ice core in the WAIS Divide area. [26] The reduced SSV in the local WAIS Divide accumulation record allowed for more meaningful comparisons between large climate indices and accumulation, though additional ice cores may be necessary to sufficiently reduce the SSV impacts depending on the application of the accumulation records. The relationship between atmospheric variability and accumulation in the WAIS Divide area is neither simple nor consistent in time. Temporal correlations between the WAIS Divide local accumulation record and

D23102

atmospheric indices, such as the Antarctic Oscillation and subsequent modes (SA1 and SA2), varied in strength and exhibited periods where the relationship switched signs several times over recent decades. Consequently, reconstructing long-term records of high-frequency geophysical processes from accumulation records at WAIS Divide may prove problematic. [27] Acknowledgments. This research was supported through funding from the National Science Foundation, grants OPP 9904294, OPP 0538427, OPP 0636929, and OPP 9814810. We would like to thank R. Edwards, T. Cox, A. Ellis, and C. Fox for their assistance in the DRI ultra-trace ice core laboratory.

References Anklin, M., R. C. Bales, E. Mosley-Thompson, and K. Steffen (1998), Annual accumulation at two sites in northwest Greenland during recent centuries, J. Geophys. Res., 103, 28,775 – 28,783, doi:10.1029/ 98JD02718. Arthern, R. J., D. P. Winebrenner, and D. G. Vaughan (2006), Antarctic snow accumulation mapped using polarization of 4.3-cm wavelength microwave emission, J. Geophys. Res., 111, D06107, doi:10.1029/ 2004JD005667. Banta, J. R., and J. R. McConnell (2007), Annual accumulation over recent centuries at four sites in central Greenland, J. Geophys. Res., 112, D10114, doi:10.1029/2006JD007887. Bromwich, D. H., and R. L. Fogt (2004), Strong trends in the skill of the ERA-40 and NCEP-NCAR Reanalyses in the high and midlatitudes of the Southern Hemisphere, 1958 – 2001, J. Clim., 17, 4603 – 4619, doi:10.1175/3241.1. 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. Clim., 13, 1406 – 1420, doi:10.1175/1520-0442(2000)0132.0.CO;2. Bromwich, D. H., R. L. Fogt, K. I. Hodges, and J. E. Walsh (2007), A tropospheric assessment of the ERA-40, NCEP, and JRA-25 global reanalysis in the polar regions, J. Geophys. Res., 112, D10111, doi:10.1029/ 2006JD007859. Conway, H., T. A. Neumann, S. Price, E. D. Waddington, D. L. Morse, K. Taylor, P. A. Mayewski, D. A. Dixon, E. Pettit, and E. J. Steig (2005), Candidate drill site near the Ross-Amundsen ice divide, West Antarctica, paper presented at the U. S. Ice Core Working Group meeting, U. S. Ice Core Working Group, Madison, Wis. Cullather, R. I., D. H. Bromwich, and M. L. van Woert (1996), Interannual variations in Antarctic precipitation related to El Nino – Southern Oscillation, J. Geophys. Res., 101, 19,109 – 19,118, doi:10.1029/96JD01769. Davis, C. H., Y. H. Li, J. R. McConnell, M. M. Frey, and E. Hanna (2005), Snowfall-driven growth in East Antarctic ice sheet mitigates recent sealevel rise, Science, 308, 1898 – 1901, doi:10.1126/science.1110662. Dixon, D. A., P. A. Mayewski, S. Kaspari, K. Kreutz, G. Hamilton, K. Maasch, S. B. Sneed, and M. J. Handley (2005), A 200 year sulfate record from 16 Antarctic ice cores and associations with Southern Ocean sea-ice extent, Ann. Glaciol., 41, 155 – 166. Fisher, D. A., N. Reeh, and H. B. Clausen (1985), Stratigraphic noise in time series derived from ice cores, Ann. Glaciol., 7, 76 – 83. Frey, M. M., R. C. Bales, and J. R. McConnell (2006), Climate sensitivity of the century-scale hydrogen peroxide (H2O2) record preserved in 23 ice cores from west Antarctica, J. Geophys. Res., 111, D21301, doi:10.1029/ 2005JD006816. Genthon, C., G. Krinner, and M. Saccnettini (2003), Interannual Antarctic tropospheric circulation and precipitation variability, Clim. Dyn., 21, 289 – 307, doi:10.1007/s00382-003-0329-1. 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, doi:10.3189/ 172756404781814447. Kuhns, H., C. Davidson, J. Dibb, C. Stearns, M. Bergin, and J. L. Jaffrezo (1997), Temporal and spatial variability of snow accumulation in central Greenland, J. Geophys. Res., 102, 30,059 – 30,068, doi:10.1029/ 97JD02760. Marshall, G. (2003), Trends in the Southern Annular Mode from observations and reanalyses, J. Clim., 16, 4134 – 4143, doi:10.1175/15200442(2003)0162.0.CO;2. Mayewski, P. A., et al. (2004), A 700 year record of Southern Hemisphere extratropical climate variability, Ann. Glaciol., 39, 127 – 132, doi:10.3189/172756404781814249.

7 of 8

D23102


McConnell, J. R., and R. C. Bales (2004), Investigation of chemical transfer processes between atmosphere and snow at South Pole, Summary Rep. 2002 – 2003 27, Clim. Monitor. and Diag. Lab. Natl. Oceanic and Atmos. Admin., Boulder, Colo. McConnell, J. R., R. C. Bales, and D. R. Davis (1997), Recent intra-annual snow accumulation at South Pole: Implications for ice core interpretation, J. Geophys. Res., 102, 21,947 – 21,954, doi:10.1029/97JD00848. McConnell, J. R., R. C. Bales, R. W. Stewart, A. M. Thompson, M. R. Albert, and R. Ramos (1998), Physically based modeling of atmosphereto-snow-to-firn transfer of H2O2 at South Pole, J. Geophys. Res., 103, 10,561 – 10,571, doi:10.1029/98JD00460. McConnell, J. R., R. J. Arthern, E. Mosley-Thompson, C. H. Davis, R. C. Bales, R. Thomas, J. F. Burkhart, and J. D. Kyne (2000a), Changes in Greenland ice sheet elevation attributed primarily to snow accumulation variability, Nature, 406, 877 – 879, doi:10.1038/35022555. McConnell, J. R., E. Mosley-Thompson, D. H. Bromwich, R. C. Bales, and J. D. Kyne (2000b), Interannual variations of snow accumulation on the Greenland ice sheet (1985 – 1996): New observations versus model predictions, J. Geophys. Res., 105, 4039 – 4046, doi:10.1029/ 1999JD901049. McConnell, J. R., G. W. Lamorey, S. W. Lambert, and K. C. Taylor (2002), Continuous ice-core chemical analyses using inductively coupled plasma mass spectrometry, Environ. Sci. Technol., 36, 7 – 11, doi:10.1021/ es011088z. McConnell, J. R., A. J. Aristarain, J. R. Banta, P. R. Edwards, and J. C. Simoes (2007), 20th century doubling in atmospheric dust archived in an Antarctic Peninsula ice core parallels climate change and desertification in South America, Proc. Natl. Acad. Sci. U.S.A., 104, 5743 – 5748, doi:10.1073/pnas.0607657104. Monaghan, A. J., et al. (2006), Insignificant change in Antarctic snowfall since the International Geophysical Year, Science, 313, 827 – 831, doi:10.1126/science.1128243. Mosley-Thompson, E., J. F. Paskievitch, A. J. Gow, and L. G. Thompson (1999), Late 20th century increase in South Pole snow accumulation, J. Geophys. Res., 104, 3877 – 3886, doi:10.1029/1998JD200092. Neumann, T. A., H. Conway, S. F. Price, E. D. Waddington, G. A. Catania, and D. L. Morse (2008), Holocene accumulation and ice sheet dynamics in central West Antarctica, J. Geophys. Res., 113, F02018, doi:10.1029/ 2007JF000764. Overpeck, J. T., B. L. Otto-Bliesner, G. H. Miller, D. R. Muhs, R. B. Alley, and J. R. Kiehl (2006), Paleoclimatic evidence for future ice-sheet instability and rapid sea-level rise, Science, 311, 1747, doi:10.1126/ science.1115159.

D23102

Palmer, A., T. van Ommen, M. Curran, V. Morgan, J. Souney, and P. Mayewski (2001), High-precision dating of volcanic events (A.D. 1301 – 1995) using ice cores from Law Dome, Antarctica, J. Geophys. Res., 106, 28,089 – 28,095, doi:10.1029/2001JD000330. Santer, B. D., T. M. L. Wigley, J. S. Boyle, D. J. Gaffen, J. J. Hnilo, D. Nychka, D. E. Parker, and K. E. Taylor (2000), Statistical significance of trends and trend differences in layer-averaging atmospheric temperature time series, J. Geophys. Res., 105, 7337 – 7356, doi:10.1029/ 1999JD901105. Scambos, T. A., C. Hulbe, M. Fahnestock, and J. Bohlander (2000), The link between climate warming and break-up of ice shelves in the Antarctic Peninsula, Ann. Glaciol., 46, 516 – 530, doi:10.3189/ 172756500781833043. Spikes, V. B., G. S. Hamilton, S. A. Arcone, S. Kaspari, and P. A. Mayewski (2004), Variability in accumulation rates from GPR profiling on the west Antarctic plateau, Ann. Glaciol., 39, 238 – 244, doi:10.3189/ 172756404781814393. 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, doi:10.1029/2007GL032529. Thompson, D. W. J., and S. Solomon (2002), Interpretation of recent Southern Hemisphere climate change, Science, 296, 895 – 899, doi:10.1126/ science.1069270. Thompson, D. W. J., and J. M. Wallace (2000), Annual modes in the extratropical circulation. part 1: Month-to-month variability, J. Clim., 13, 1000 – 1016, doi:10.1175/1520-0442(2000)0132.0.CO;2. van den Broeke, M. R., and N. P. M. van Lipzig (2004), Changes in Antarctic temperature, wind and precipitation in response to the Antarctic Oscillation, Ann. Glaciol., 39, 119 – 126. van der Veen, C. J., and E. Mosley-Thompson (1999), Accumulation at South Pole: Comparison of two 900-year records, J. Geophys. Res., 104, 31,067 – 31,076, doi:10.1029/1999JD900501. R. C. Bales, School of Engineering, University of California, Merced, CA 95344, USA. J. R. Banta, J. R. McConnell, and K. Taylor, Division of Hydrologic Science, Desert Research Institute, 2215 Raggio Parkway, Reno, NV 89509, USA. ([email protected]) M. M. Frey, British Antarctic Survey, National Environment Research Council, Cambridge CB3 0ET, UK.

8 of 8