Data and methods

4.2.1

DSS site characteristics and

_{Be composite record}

Dome Summit South (DSS, 66◦_{46.18’S 112}◦_{48.69’E, Fig. 4.1) is the main ice core drilling}

site on Law Dome, coastal East Antarctica (see Morgan et al. [1997] for site charac- teristics). High temporal resolution (∼monthly) 10_{Be measurements have been carried} out at DSS over a number of years, with good agreement observed between multiple records sampled up to 500 m apart [Pedro et al. 2011a/Chapter 2]. This indicates that variations in 10_{Be concentrations at the site are primarily driven by production and} atmospheric transport and deposition related factors and are not strongly affected by post-depositional processes. The 10-year10_{Be record used here is a composite derived}

from three discrete, but identically processed ice cores. Full details concerning the sampling, chemical preparation, Accelerator Mass Spectrometer (AMS) measurement and construction of the composite 10_{Be record are available elsewhere [Pedro et al.} 2011a/Chapter 2]. Measurement uncertainties for the 10_{Be record are small (∼ 3%)} and will be neglected in the analysis to follow.

The cores used in the composite record were originally sampled at an average tempo- ral resolution of 10 samples per year. To better facilitate intercomparison with other data sets the10_{Be data was interpolated onto a monthly grid. The timescale has been} improved (over Pedro et al. [2011a]/Chapter 2) through the use of reanalysis precipi- tation data from the European Centre for Medium-Range Weather Forecasts Interim Re-Analysis (ERA-Interim, [Simmons et al., 2006; Uppala et al., 2008]) to provide information on the seasonal distribution of precipitation variations within years (see Appendix A.). This technique is justified by the excellent agreement between ERA- Interim precipitation data and measured snow accumulation data from a downward pointing radar on a automatic weather station located near the site for the period 1998 to 2002 (Fig. A1). In transferring from the depth-scale to the new time-scale the information from ERA-Interim regarding the time distribution of snow fall leads to changes in the amplitude of some 10_{Be concentration peaks relative to Pedro et al.} [2011a]/Chapter 2 (see Fig. A2). These changes do not affect any conclusions of our previous work including the strength of the correlation between neutron counting rates and 10_{Be concentrations.}

In measuring the shared variance between time series in this paper we use the bivariate Pearson’s correlation coefficient (rxy), calculated using the method detailed inMudelsee

[2003]. This method employs the nonparametric stationary bootstrap with an average block length proportional to the maximum estimated persistence time of the data. The method is preferred over standard regression techniques since it yields robust results for

rxy and associated 95% confidence intervals (CIs) even for data that is autocorrelated,

unevenly spaced and non-normally distributed such as the DSS 10_{Be data.}

4.2.2

ECHAM5-HAM model description and setup

For the model runs we use the ECHAM5 GCM [Roeckner, 2003] coupled to the HAM aerosol module [Stier et al., 2005], which simulates the production, transport and depo-

110˚E 110˚E 112˚E 112˚E 114˚E 114˚E 116˚E 116˚E 68˚S _68˚S 67˚S _67˚S 66˚S _66˚S 65˚S _65˚S 0 100 200

km

Figure 4.1: Location of the Dome Summit South (DSS) sample site, Law

Dome with surface elevation contours (m, solid lines) and accumulation isopleths (mm ice equivalent, dashed lines). The centre point of the ECHAM5-HAM grid cell used in the analysis is marked by the star.

sition processes of aerosols. Solar modulation of the 10_{Be production rate is simulated} using the 10_{Be production functions ofMasarik and Beer [2009] and monthly-resolved} values for the solar activity parameter (Φ) from [Usoskin et al., 2005, available at

http://cosmicrays.oulu.fi/phi/phi.html]. Since the period under investigation

is brief, geomagnetic modulation of production is negligible and the geomagnetic field strength is held constant at the modern level. A detailed evaluation of ECHAM5-HAM’s ability to simulate the general patterns of atmospheric circulation and deposition of beryllium radionuclides is provided in Heikkil¨a et al.[2008b].

For this study, the model was run on a T42L39 resolution (∼ 300 km horizontal res- olution, 39 vertical levels reaching up to ∼80 km), which is the state of the art for global models incorporating chemical modules. Including the whole stratosphere into the model domain is important for a tracer with stratospheric origin, such as10_{Be. The}

model was run for the years 1993 to 2004. The first 5 years were used to allow 10_Be reach equilibrium in the atmosphere and the final 6 years (1998 to 2004) are used here in the analysis. For logistical reasons it was not possible to run the model over the full 10 years of observed data.

Model results were taken from the grid cell centered at 66◦_{35’S 112}◦_{30’E (marked in}

Fig. 4.1) that covers the Law Dome site. Model resolution is a limitation for modeling coastal Antarctica as precipitation and temperature are very sensitive to the steep (sub- grid scale) gradients in elevation between the ocean and the ice sheet [Bromwich et al., 2011]. The accumulation isopleths shown in Fig. 4.1 illustrate the importance of the orographic effect on snow accumulation rates at Law Dome, e.g. the accumulation rate 10 km to the east of the summit is ∼7 times higher than that observed 20 km to the west of the summit. Clearly, such sub-grid scale gradients in accumulation cannot be resolved by this model. However, the transport paths and the life time of 10_Be from source to sink are long and controlled mostly by large scale atmospheric processes [Heikkil¨a et al., 2009]. To be helpful in understanding the factors driving seasonality in10_{Be concentrations it is more important for the model to capture these large scale} processes than to describe orography in high detail.