Journal of Geophysical Research: Atmospheres

Evidence for climate modulation of the 10Be solar activity proxy

Authors


Abstract

[1] We used a snow pit record in conjunction with detailed snow accumulation data and oxygen isotope records to examine atmospheric transport and deposition effects on 10Be at Law Dome, Antarctica. Data from an adjacent automatic weather station was used to date the record at snowfall event-scale resolution. In contrast to prior ice core studies in Antarctica, the snow pit record is of a sufficiently short duration (∼1 year) that 10Be fluctuations reflect mainly atmospheric transport processes rather than solar modulation of production. Elevated concentrations of 10Be were found in the late austral summer and early autumn snow, synchronous with the seasonal increase in stratospheric aerosols at Antarctic stations. A significant (P < 0.01) anticorrelation of 10Be with δ18O occurs at the snowfall event scale. Fractionation of water isotopes at Law Dome is controlled by local and regional processes, specifically transport and local temperature. The anticorrelation seen here implies that 10Be concentration was reduced in snow from warmer air masses (characterized by less negative δ18O). There is potential for confounding solar modulation with climatic modulation if at sites such as this one, warmer meteorological influences may be associated with reduced 10Be concentrations. Quantification of the significance of this effect for the longer-term 10Be record will require analysis of longer 10Be records from different sites.

1. Introduction

[2] Although Antarctic snow pit studies of the cosmogenic (cosmic ray produced) radionuclide 10Be have been undertaken previously [e.g. Steig et al., 1996], none approach the snowfall event-scale resolution of this study. Prior high resolution studies at the high accumulation Law Dome site have yielded valuable results for palaeoclimate reconstructions from ice core records [Morgan and van Ommen, 1997; Curran et al., 1998; McMorrow et al., 2004]. We show that high temporal resolution analyses of 10Be can contribute to understanding of meteorological influences on 10Be.

[3] Ice core records of 10Be have been used to reconstruct the history of factors which control the flux of cosmic rays to Earth: namely, variations in the Earth's geomagnetic field strength [e.g. Wagner et al., 2000; McCracken, 2001] and variations in solar activity [e.g. Raisbeck and Yiou, 1980; Beer et al., 1990; Beer, 2000]. In addition to variations in production rate in the atmosphere, the 10Be ice core record also is influenced by atmospheric mixing, transport pathways, depositional processes and changes in precipitation rate [Lal, 1987; Yiou et al., 1997; McCracken and McDonald, 2001]. Although production in the atmosphere is captured in models [Masarik and Beer, 1999; Webber and Higbie, 2003], the influences of meteorology and deposition are less well understood [Beer, 2000].

[4] Variations of 10Be concentration in polar ice are considered primarily a function of changes in precipitation rate and secondarily of changes in production rate due to solar variability [Yiou et al., 1997]. Under the relatively stable precipitation rates of the late Holocene solar variability has been reconstructed by assuming a simple linear relationship between 10Be concentration in ice cores and the rate of 10Be production in the atmosphere [e.g. McCracken and McDonald, 2001; Bard et al., 2000; Usoskin et al., 2004]. Techniques employed to reduce the meteorological noise in 10Be records include smoothing over multiple years [e.g. Bard et al., 2000] and using a synthesis of 10Be data from Antarctica and Greenland [e.g. McCracken and McDonald, 2001]. Comparisons of 10Be with 14C from tree rings have also been valuable in separation of system effects from production effects. Both 14C and 10Be are produced by cosmic ray induced interactions, but they experience quite different geochemical behaviour. Using models of the carbon cycle it has been shown that there is a close agreement between 14C and 10Be variations in the recent past [e.g., Beer et al., 1984; Bard et al., 1997]. This agreement is evidence that the dominant source of variability in 10Be during the Holocene is the production rate. However, climatic effects continue to complicate interpretation of 10Be records. For example, the Greenland, Dye 3, 10Be record shows a strong decrease in the later part of the 20th century record which is not observed in the South Pole record [see Raisbeck and Yiou, 2004]. This discrepancy has been an important reason behind the contrasting reconstructions of recent changes in solar activity [Bard et al., 2000; Usoskin et al., 2003; Raisbeck and Yiou, 2004; Muscheler et al., 2005; Solanki et al., 2004].

[5] A number of reconstructions of cosmogenic production rates and solar activity indices incorporating 10Be ice core records have demonstrated significant correlations with reconstructions of certain climate parameters. For example, Northern Hemisphere temperature over the last millennium [Usoskin et al., 2005], increases in North Atlantic drift ice cycles [Bond et al., 2001] and most famously, the “little ice age” and “medieval warm period” [e.g. Bard et al., 2000]. However, the magnitude of solar forcing appears too small to explain the observed climate changes and mechanisms proposed, as yet, fail in important respects to generate the expected climatic responses [Rind, 2002].

[6] Improved understanding of meteorological and climatic influences on 10Be delivery to polar sites would be valuable for refining reconstructions of past solar variability. The high temporal resolution of the record considered here has an advantage in this respect: the atmospheric residence time of 10Be is approximately one to two years in the stratosphere and several weeks in the troposphere, therefore ice measurements of less than one year duration should reflect mainly atmospheric transport processes [Beer, 2000].

2. Experimental Results

2.1. Sample Site and Characteristics

[7] Samples analysed in this study came from a snow pit located at 66°46′09″S, 112°48′26″E, Dome Summit South (DSS), Law Dome. This site is well suited to study of chemical and other signals in fine detail [Curran et al., 1998]. It experiences high snow accumulation (0.68 m per year ice equivalent), and low annual average wind speed (8.3 ms−1), which minimises surface disturbance, and perennially low temperature (summer mean, −12.6 °C), which precludes alteration by summer melt [Morgan et al., 1997; Allison et al., 1993].

2.2. Sample Collection

[8] A snow pit was excavated to a depth of 2.25 m adjacent to an automatic weather station (AWS 1181) near Dome Summit South (DSS), Law Dome, East Antarctica (66°44′S, 112°45′E) during December 2001. Twenty 10 cm thick contiguous samples of approximately 1.5 kg mass were taken down the face of the snow pit by insertion of a 10 cm high × 20 cm × 20 cm stainless steel scoop. Samples were immediately sealed in zip lock bags. Oxygen isotope samples were also collected from the pit [see McMorrow et al., 2002]. To avoid contamination of the 10Be samples, all equipment was pre-rinsed with >18 MΩcm−1 Milli-Q water and personnel were equipped with dust masks and gloves.

2.3. Sample Preparation and Measurement

[9] Sample preparation [see Child et al., 2000] and accelerator mass spectrometer (AMS) measurements [see Fink et al., 2000] were carried out at the Australian National Tandem Accelerator for Applied Research (ANTARES) AMS. 10Be concentrations were normalised to the National Institute of Standards (NIST) SRM 4325 10Be standard with an adjusted ratio of 3.02 × 10−11 [Fink et al., 2000]. Measurements of the NIST standard were reproducible within ±3%. Aldrich BeO blanks indicated that the machine background 10Be/9Be was negligible (<5 × 10−15) compared to the ratios being measured. Chemistry procedural blanks were in the range 5–15 (×10−15). After normalisation to the NIST standard, error weighted mean 10Be/9Be ratios ranged from 180–760 (×10−15) with overall errors ranging from (7–22) × 10−15.

[10] Melt water was acidified prior to filtration during sample preparation, in exception to the procedure of Child et al. [2000]. Acidification immediately after filtration had been advised by Smith et al. [2000] in order to separate terrestrial dust and extraterrestrial material (micrometeorites and cosmic spherules) from the melted sample with minimal exchange of 10Be. However, we have shown that even under worst case assumptions of terrestrial dust levels and extraterrestrial material flux, the potential quantity of 10Be exchanged from these sources is insignificant to this study [Pedro, 2002]. Acidification prior to filtration is consistent with techniques used in prior ice core studies and potentially reduces adsorption of 10Be to the filter and bottle walls [Finkel and Nishiizumi, 1997; Yiou et al., 1997].

2.4. Sample Dating

[11] Event-scale dating of the snow pit was carried out using snow accumulation data from an adjacent Automatic Weather Station (AWS 1181, Australian Antarctic Division, Glaciology Program) according to the technique described in McMorrow et al. [2002]. Ultrasonic soundings from the AWS measured snow height at approximately hourly intervals, providing a record of the difference between snow accumulation and snow removal events over time. Dating of the 2.00 m 10Be record indicates the record spans less than 1 year of accumulation, from early January 2001 to 25 December 2001. The 10-cm resolution 10Be samples were determined to span time periods ranging from days to several months.

[12] Prior studies at Law Dome have found that the snow pit dating technique provides records which are regionally reproducible over a 12 km transact, approximately following an accumulation isopleth at Law Dome [McMorrow et al., 2002].

3. Results

3.1. 10Be Concentrations and Variability

[13] The mean 10Be concentration (and standard error in the mean) of the twenty snow pit samples was (5.6 ± 0.5) × 103 atoms g−1. This is comparable to the (7.6 ± 0.3) × 103 atoms g−1 reported by Smith et al. [2000] in ice core samples from DSS, Law Dome, spanning 1832 to 1913 AD. Figure 1 shows the variability in concentration with snow pit depth. Concentration ranges from (3.2 ± 0.1) × 103 atoms g−1 to (12.7 ± 0.3) × 103 atoms g−1. In Figure 2 the snow pit depth profile of Figure 1 has been referenced temporally using snow accumulation data from the AWS. A prominent peak in concentrations is noted from 1.7 to 1.9 m below the snow pit surface. This corresponds to late summer and early autumn of 2001. Concentrations are lowest from 0.7 to 1.6 m, during mid to late autumn 2001. Generalisation of this pattern to other years is restricted by the short data set. Notably, an assessment of archived ice core measurements from the DE08 drilling site (located 8 km east of DSS) also revealed a statistically significant summer maximum in 10Be concentrations [Pedro, 2002].

Figure 1.

10Be concentrations in samples from the snow pit against sample depth. 0 m represents snow pit surface at time of sampling (25 December 2001). Vertical bars represent standard measurement error in 10Be concentrations.

Figure 2.

10Be concentrations from Figure 1 referenced temporally using snow accumulation data from the AWS. Horizontal bars indicate the range of days over which two-thirds of the snow contributing to each sample was deposited (this allows consideration of the time interval over which the bulk of the sample accumulated and minimizes potential error associated with classification of small snowfall events to samples). Five days have been added to the upper and lower bounds of these bars as a best estimate of the error potentially associated with the date scale, on the basis of earlier work [McMorrow et al., 2002]. Data points are positioned within the horizontal bars by weighting the timing of accumulation events contributing to each sample according to their size. Horizontal bars which overlap indicate that a common precipitation event may have contributed to the samples. Vertical bars represent standard measurement error in 10Be concentrations.

[14] A late summer peak in 10Be concentration is concordant with the observed timing of enhanced arrival of stratospheric aerosol markers to Antarctic research stations, for example, elevated 10Be/7Be ratios [Raisbeck et al., 1981] at Nuemayer [Wagenbach et al., 1988; Wagenbach, 1996] and enhanced 7Be levels at South Pole and coastal Terre Adèlie [Maenhaut et al., 1979; Raisbeck et al., 1981; Sanak et al., 1985].

3.2. 10Be Flux

[15] It is simple to estimate the annual flux rate (atoms cm−2 yr−1) of 10Be during the study period. We used the following quantities: length of time sampled by the snow pit, 349 days; surface area of samples, 400 cm2; and the measured value (±standard measurement error) of 10Be atoms in the record, (182 ± 6.4) × 106 atoms. From these, flux was estimated to be (4.8 ± 1.6) × 105 atoms cm−2 yr−1, calculated with best estimates of error in the accuracy of the time scale and sample size, and scaling for a full year. It will be interesting to compare this result with new longer term 10Be records from Law Dome. This flux is comparable to values reported for Greenland e.g., Milcent (70°18′N), 5.0 × 105 atoms cm−2 yr−1, Dye 3 (65°11′N), 4.7 × 105 atoms cm−2 yr−1, and higher than those reported from central Antarctica e.g., Vostok (78°28′S), 1.8 × 105 atoms cm−2 yr−1, Dome C (74°39′S) 1.9 × 105 atoms cm−2 yr−1 (from Raisbeck and Yiou, 1985).

3.3. Deposition Mechanism

[16] Dry deposition of 10Be is dominant on the high Antarctic plateau. This is evident from the inverse relationship between 10Be concentration and accumulation rate at the ice core drilling sites, Vostok, Dome C and South Pole [Raisbeck and Yiou, 1985]. At Law Dome, 10Be concentrations at three sites were independent of their eightfold difference in accumulation rate, indicating dominance of wet deposition [Smith et al., 2000].

[17] A temporal approach is used here to test the findings of Smith et al. [2000]. Each successive snow sample of 10-cm depth resolution was deposited over a known time interval, as determined from AWS data. This time interval, or “exposure time”, represents the number of days over which a sample was exposed to deposition processes (Figure 3). Some dependence of 10Be concentrations on sample exposure time would be expected if dry deposition dominated 10Be accumulation. There is no statistically significant relationship between exposure time and 10Be concentration in our data (for both error weighted and unweighted data). For example, some samples which had exposure times of ∼30 days exhibit similar 10Be concentrations to samples with exposure times of several days. This result is concordant with the work of Smith et al. [2000] and demonstrates that dry deposition of 10Be was not important during the study period.

Figure 3.

10Be concentration is plotted against the “exposure time” (the time that a sample was exposed to potential dry deposition processes). Regression analysis shows a weak positive linear correlation (r2 = 0.15) that is not statistically significant, indicating that dry deposition was not important during the study period.

3.4. Comparison With the Snow Pit Oxygen Isotope Ratio Records

[18] The snow pit record of oxygen isotope ratios (δ18O) from McMorrow et al. [2004], averaged to 10-cm resolution, is shown in Figure 4. A typical seasonal pattern can be seen: lower (more negative) ratios in winter and higher (less negative) ratios in summer. The record of 10Be, extracted at 10-cm resolution, is overlain for comparison. The 10Be record is inversely related to δ18O at the snowfall event-scale. To assess the significance of this relationship we plotted the change in 10Be concentration between adjacent samples against the change in δ18O (Figure 5). This method largely removes the lower frequency component of the δ18O and 10Be data and focuses on the event-scale. A negative correlation was obtained with r2 = 0.32. We note that the technique applied, by consideration of differences, reduces the number of statistically independent data points and potentially the significance of the correlation. However, in this instance the lag-one autocorrelations for δ18O and 10Be in Figure 4 are small and from a sample size (n) of 19 we obtain an effective sample size (neff), after taking differences in Figure 5, of 18. With n = 18 the statistical significance of the 10Be - δ18O negative correlation is high (P < 0.01).

Figure 4.

Comparison between the 10Be record (solid curve) and a contemporaneous δ18O record (dashed curve) from the snow pit considered at 10-cm resolution. Vertical bars indicate standard measurement errors.

Figure 5.

Change in δ18O between adjacent samples plotted against change in 10Be concentration between adjacent samples. Analysis shows that the relationship is significant (r2 = 0.32, and P < 0.01), suggesting a common meteorological influence on delivery of these species to the study site. In calculation of P, the values n = 19 and neff = 18 were used; refer to text.

4. Discussion

[19] Although polar meteorology strongly affects the 10Be ice core record [Lal, 1987] there has been little progress in defining and quantifying these effects. While the data set considered here is small (due to unforeseen circumstances at the time of extraction of the record in the field), the high resolution and access to detailed AWS accumulation data provides an insight into meteorological influences that could not be considered previously. The most notable features of the 10Be record are the peak in concentration during late summer to early autumn and the relationship with δ18O. These are discussed in turn.

[20] The peak in 10Be concentration may correspond to a period of enhanced stratospheric exchange. Elevated concentrations of stratospheric aerosol markers are also reported to arrive to Antarctic stations during late summer and early autumn [Raisbeck et al., 1981; Wagenbach et al., 1988; Wagenbach, 1996; Maenhaut et al., 1979; Sanak et al., 1985]. Wagenbach [1996] suggested that this trend may be related to a weakened tropospheric temperature inversion layer over Antarctica and enhanced convective activity and vertical exchange within the troposphere at this time [also Koch and Mann, 1996; Dibb et al., 1994]. However, there is no clear evidence that enhanced convective activity should increase stratospheric exchange. Notably, the peak in 10Be concentration does not coincide with the break down of the polar vortex in spring to early summer. Nor does it coincide with the increased stratospheric subsidence that is expected to occur during winter [see König-Langlo et al., 1998].

[21] The significant negative correlation observed between changes in δ18O and 10Be indicates a common meteorological influence on δ18O and 10Be at Law Dome. Previous high resolution ice and firn studies at Law Dome have developed an understanding of the meteorological influences affecting δ18O, which facilitates interpretation of the 10Be - δ18O relationship.

[22] Variations in δ18O at Law Dome are associated with changes in the source and transport properties of incoming moisture bearing air masses and local air temperature [van Ommen and Morgan, 1997; McMorrow et al., 2001; McMorrow et al., 2002; McMorrow et al., 2004]. Less negative (“warmer”) oxygen isotope signals are generally characterized by the rapid meridional transport of marine air to Law Dome and warmer temperatures, whereas more negative (“colder”) signals are often characterised by weak cyclonic development in the vicinity, enhanced influence of continental air and cooler temperatures [McMorrow et al., 2001; McMorrow et al., 2002]. On the basis of this understanding we propose the following explanation for the 10Be, δ18O relationship: (1) depletion of 10Be in marine air masses (characterized by warmer δ18O,) due to higher frequency of precipitation events (“washout”) with respect to continental air [see Aldahan et al., 2001] and (2) air masses with colder δ18O are expected to have traveled greater distances from their oceanic moisture sources [Morgan, 1982], thus increasing their range for aerosol scavenging, and experienced greater altitudes of trajectory [Morgan, 1982], thus increasing exchange with stratospheric and upper-tropospheric aerosols with elevated 10Be [Koch and Mann, 1996; Dibb et al., 1994].

[23] Precipitation rate changes cannot explain the relationship between δ18O and 10Be because wet deposition is dominant at the study site (Figure 3). Solar variability cannot be implicated since the event-scale resolution of the snow pit data is shorter than the residence time of 10Be in the atmosphere. Examination of a longer 10Be record in conjunction with meteorological data will be required to test and improve this mechanism.

[24] If meteorological variability results in covariance of 10Be and δ18O over longer time intervals than those identified in this study then there are implications for interpretation of 10Be records. Comparisons of the Law Dome δ18O record with instrumental records (NCEP, 1958–1999) of surface pressure anomalies show significant correlations in the Antarctic region (T. van Ommen, manuscript in preparation, 2006), suggesting that fluctuations in δ18O from the mean are also driven by variability in atmospheric transport pathways and processes [see also Noone and Simmonds, 2002]. This identifies the possibility that annual and longer term variations in δ18O at Law Dome could be associated with 10Be variations due to changes in atmospheric transport and moisture sources, irrespective of changes in solar activity or precipitation rate.

[25] The relationship identified here explains ∼30% of the variance in the 10Be record; this is considerable given that models predict that the fluctuation in 10Be production from sunspot minimum to sunspot maximum is 15 to 30% [Masarik and Beer, 1999]. From the small data set the significance (if any) of this climate modulation on the longer term record cannot be established. The crucial test will be to determine if the relationship holds after a record is smoothed over several years. If it does not, then we can conclude that current methods of reconstructing solar parameters from 10Be records are adequate in their treatment of meteorological effects (e.g. smoothing of records over multiple years [e.g. Bard et al., 2000] and consideration of 10Be flux rather than concentration [e.g. Bond et al., 2001]). If the 10Be, δ18O relationship does hold, then important questions will be raised as to the presence of a climate signal in reconstructions of solar parameters from 10Be at affected ice core sites.

[26] Atmospheric transport influences on 10Be are of interest in the context of recent debate concerning evidence for increased solar activity in recent decades and the disparity in Greenland and South Pole 10Be records [Usoskin et al., 2003; Raisbeck and Yiou, 2004]. The 10Be records from South Pole and Greenland disagree in the late 20th century; the Greenland record shows a strong decrease, which Usoskin et al. [2003] argue is consistent with a level of solar activity that is unprecedented for the past 1150 years, however the strong decrease is not observed in the South Pole record [Raisbeck and Yiou, 2004]. As at Law Dome, 10Be is predominantly wet deposited in Greenland [Yiou et al., 1997] and may exhibit greater transport sensitivity than at South Pole, where deposition is dry and therefore independent of atmospheric moisture sources and less dependent on transport pathways. Furthermore, the result here raises the possibility that 10Be records extending to the present day could have been influenced by the climate change of recent decades at some sites.

5. Conclusion

[27] We have demonstrated a significant anti-correlation between changes in 10Be and oxygen isotope ratio (which we interpret as a climate variable) at Law Dome. The anti-correlation occurs over a period of time short enough to exclude variation in 10Be production. Climate modulation of 10Be transport and deposition must therefore be implicated. Of particular relevance for interpretation of 10Be records is the potential this relationship implies for the covariance of 10Be and climate parameters independently of solar variability. Further research is required to quantify this climate influence on 10Be over longer term records and at different sites. Consideration of additional high temporal resolution 10Be records and longer 10Be records will be useful to achieve this objective. Until the extent of climatic modulation of the 10Be proxy can be more adequately quantified the technique of reconstructing cosmogenic isotope production rates directly from 10Be concentrations in ice should be exercised with caution.

6. Future Work

[28] Previously, access to radionuclide records of sufficiently high resolution for comparison with annual aerosol data has been restricted by the large sample sizes required for measurement combined with the low accumulation rates and high wind speeds characteristic of polar ice coring sites [Wagenbach et al., 1988]. The snowfall event-scale resolution of the record presented here identifies the potential of the Law Dome site for consideration of 10Be concentration in ice and contemporaneous aerosol measurements. Such work could improve understanding of atmosphere ice relationships for 10Be. Extraction of an event-scale record spanning several years of 10Be data will be valuable to test findings from the snow pit record.

Acknowledgments

[29] This work was supported by the Australian Government's Cooperative Research Centres Programme, through the Antarctic Ecosystems and Climate Cooperative Research Centre (ACE CRC).

Ancillary