Season 1 — S0k

The top 3 m of the S0k firn core, collected during the first season (1999/2000), preserves material from summer 1999/2000 back to the installation of the AWS during summer 1997/98. This core section preserves chemical and isotopic signals consistent with two seasonal cycles identified from the AWS dating scale. The summer species (MSA, Δ¹⁸O (not shown), nitrate (not shown) and non-sea-salt sulphate (not shown)) are in phase with the summer periods of the AWS dating scale, while sea-salt concentrations (sodium, chloride (not shown) and magnesium (not shown)) are preserved during winter deposition layers (Fig. 3). This is consistent with previous high-resolution studies that resolve the seasonal cycle of ice-core signals both at the Law Dome site (Reference Curran, van Ommen and MorganCurran and others, 1998) and at other Antarctic sites (Reference Legrand and DelmasLegrand and Delmas, 1984; Reference Minikin, Wagenbach, Graf and KipfstuhlMinikin and others, 1994; Reference KreutzKreutz and others, 1999). These results suggest that densification over the top 3m of the snowpack does not significantly affect the dating of events at the seasonal scale at Law Dome.

Fig. 3.

MSA and sodium records preserved in S0k for the 0–3m depth range. Seasonal (dotted line) boundaries determined from the AWS are also shown.

Season 2 — DSS0102

The DSS0102 firn core was collected 2 years later than S0k, and the top 6 m of DSS0102 covers the period from summer 2001/02 back to the installation of the AWS during summer 1997/98 (Fig. 2). The core section 3 –6 m in DSS0102 overlaps with the time period preserved in the core section 0–3m in S0k. Figure 4 illustrates that for the top 3 m of the snowpack the seasonal cycles of the chemical and isotopic signals are in phase with the AWS dating scale. This is consistent with the results presented in Figure 3 and confirms that the top 3 m of the snowpack is not significantly affected by densification at the seasonal time-scale. However, for depths >3m the impact of densification becomes increasingly more pronounced and the firn-core record compacts significantly compared to the AWS dating scale (Fig. 4). This is indicated by the seasonal signals in the firn core becoming out of phase with the seasonal boundaries identified in the AWS dating scale. For example, the clear summer peak in MSA at 5.25 m depth is identified from the AWS dating scale as occurring during autumn 1998 (Fig. 4). This same summer peak is identified in the S0k firn core at around 2.5m depth and identified from the AWS dating scale as occurring during summer 1997/98 (Fig. 3).

Fig. 4.

MSA and sodium records preserved in DSS0102 for the 0–6m depth range. Seasonal (dotted line) boundaries determined from the AWS are also shown.

Correcting for densification— DSS0102

The AWS dating scale may be ‘corrected’ for depths >3m by considering the amount of densification for events preserved in both the S0k and DSS0102 firn cores. Figure 5 illustrates the effect of compacting the AWS dating scale for DSS0102 to bring the seasonal signals in the firn core in phase with seasons identified from the AWS net accumulation record. The AWS dating scale ‘correction’ was calculated using an interpolative method considering significant chemical and Δ¹⁸O events preserved in both firn cores. Nineteen significant events were identified in both the S0k and DSS0102 firn cores. The depth of each event in the firn cores was used as a known input range, and a simple linear interpolation was conducted on the AWS dating scale. This resulted in ‘compacting’ the AWS dating scale with respect to the DSS0102 firn core to bring the significant events in line with the dates identified from the S0k firn core.

Fig. 5.

Trace chemical and Δ¹⁸O signals associated with the events preserved in DSS0102. (a) MSA, sodium; (b) nitrate, non-sea-salt sulphate; and (c) Δ¹⁸O. Seasonal (dotted) boundaries determined from the AWS but corrected for densification are also shown.

MSA

The marine biogenic signal represented by MSA in DSS0102 shows a clear seasonal signal for the four annual cycles, with maximum concentrations occurring during summer, and minima during autumn, winter and spring (Fig. 5). The record is characterized by sharp boundaries either side of the summer peak, with few additional peaks preserved through the remaining annual cycle. The main summer MSA peak tends to occur simultaneously with the main midsummer Δ¹⁸O peak. These results are consistent with MSA sourced from DMS production during summer marine biogenic activity. Non-summer MSA peaks are much smaller and confined to either late-summer/early-autumn or spring events. These events are also generally characterized by elevated levels in Δ¹⁸O, sea salts and nitrate (Fig. 5). Previous studies have attributed similar events to the intrusion of low-latitude air through strong cyclonic systems (Reference McMorrow, Curran, van Ommen, Morgan, Pook and AllisonMcMorrow and others, 2001, Reference McMorrow, Curran, van Ommen, Morgan and Allison2002).

Sea salts

The major sea-salt species (sodium, chloride, magnesium) preserved in DSS0102 reflect a noisy record, although maximum concentrations tend to be preserved during spring, winter and autumn, and minimum levels during summer (Fig. 5). This is in general agreement with previous studies from coastal Antarctic sites where higher levels of sea salts are preserved during the austral winter (Reference Minikin, Wagenbach, Graf and KipfstuhlMinikin and others, 1994; Reference Curran, van Ommen and MorganCurran and others, 1998; Reference Wagenbach, Legrand, Fischer, Pichlermayer and WolffWagenbach and others, 1998a). The noisy nature of the sea-salt species indicates the record is driven by specific events rather than the broad seasonality in climate conditions. High winter sea-salt concentrations are consistent with increased frequency of cyclonic events in the Antarctic region during winter (Reference King and TurnerKing and Turner, 1997). The fine-resolution sampling of DSS0102, coupled with the accurate, independent dating scale, allows the extension of the sea-salt seasonal cycle to include maximum concentrations preserved during spring and autumn. This result is consistent with the synoptic-scale meteorology for these seasons. Meteorological studies for the East Antarctic region report a semi-annual variation in atmospheric pressure with minimum pressures observed in the austral spring and autumn (Reference SchwerdtfegerSchwerdtfeger, 1984; Reference Allison, Wendler and RadokAllison and others, 1993; Reference King and TurnerKing and Turner, 1997). Atmospheric pressure minima in the Southern Ocean are associated with more intense cyclonic systems characterized by stronger surface winds. These strong winds provide the mechanism for producing enhanced sea-salt concentrations in the ice-core record.

Nitrate

Nitrate shows a more erratic seasonal cycle characterized by narrow peaks during spring and early summer, and lower levels during late summer, autumn and winter (Fig. 5). Previous studies from Law Dome nitrate records have suggested that nitrate concentrations increase from June until a relative maximum is reached in November, with further sharp rises in concentrations until an absolute maximum is attained in January (Reference Curran, van Ommen and MorganCurran and others, 1998). However, the results presented here indicate nitrate maxima occur earlier in the year, and precede the main mid-summer Δ¹⁸O and MSA peaks (Fig. 5). Similarly, high-resolution records from Siple Dome (Reference KreutzKreutz and others, 1999) and the coastal Antarctica stations of Mawson, Neumayer and Dumont d’Urville (Reference WagenbachWagenbach and others, 1998b) show late-spring/early-summer enhancement of nitrate. Spring enhancement of nitrate is consistent with the transport of nitrogen species from the stratosphere to the troposphere through the weakening of the polar vortex at this time of year. These results suggest the nitrate signal preserved in Law Dome ice cores is sourced from the stratosphere rather than local regions, although long-range tropospheric transport of nitrate cannot be ruled out.

Non-sea-salt sulphate

The non-sea-salt sulphate record preserved in DSS0102 shows a clear seasonal cycle with a pronounced midsummer maximum and lower concentrations during winter (Fig. 5). The summer maximum coincides with the main mid-summer peaks in MSA and Δ¹⁸O, and is consistent with marine biogenic sulphur compounds dominating the non-sea-salt sulphate signal at this time of year. However, outside the main mid-summer peak the non-sea-salt sulphate signal is more comparable to the nitrate record (Fig. 5). This indicates a common source for non-sea-salt sulphate and nitrate during spring, autumn and winter, with potential sources including the downward mixing of stratospheric air or long-range tropospheric transport.

Δ18O

The seasonal signal in Δ¹⁸O preserved in DSS0102 is generally characterized by a narrow summer peak and broad winter trough (Fig. 5). All four years presented in Figure 6 show a maximum in Δ¹⁸O occurring during midsummer, with smaller peaks occurring periodically throughout the seasonal cycle. These smaller Δ¹⁸O peaks will diffuse over time and result in a smoothed record characterized by a broad seasonal signal in Δ¹⁸O for deeper ice-core records. The seasonal cycle in Δ¹⁸O presented in Figure 5 is in agreement with longer-term ice-core studies (Reference Schotterer, Frhlich, Gäggeler, Sandjordj and StichlerSchotterer and others, 1997; Reference Van Ommen and Morganvan Ommen and Morgan, 1997) and compares well with the annual cycle in temperature for the East Antarctic region (Reference Allison, Wendler and RadokAllison and others, 1993; Reference Van Ommen and Morganvan Ommen and Morgan, 1997).

Fig. 6.

Δ¹⁸O and air temperatures associated with events preserved in DSS0102. Air temperatures correspond to daily averages for net accumulation days determined from the AWS accumulation record and thus represent a discontinuous record in time. Seasonal (dotted) boundaries determined from the AWS (corrected for densification) are also shown.

Non-summer peaks in Δ¹⁸O often occur during periods of enhanced sea-salt concentrations (Fig. 5), and indicate a common source of Δ¹⁸O and sea salt unrelated to the seasonal cycle in temperature. Potential sources for the link between non-summer Δ¹⁸O and sea salts include the intrusion of strong cyclonic events onto the Antarctic continent (Reference McMorrow, Curran, van Ommen, Morgan, Pook and AllisonMcMorrow and others, 2001). The results presented in Reference McMorrow, Curran, van Ommen, Morgan, Pook and AllisonMcMorrow and others (2001) suggest that cyclonic events facilitate the rapid advection of marine air from lower latitudes onto Law Dome. This rapid advection of low-latitude air provides an explanation for enhanced non-summer Δ¹⁸O signals, while strong winds associated with the cyclone direct sea-salt aerosols over the sampling site.

Temperature and Δ¹⁸O

Figure 6compares the Δ¹⁸O record from DSS0102 with associated air temperatures recorded by the co-located AWS. The temperature records correspond to daily averages for net accumulation days as determined from the AWS accumulation record, and thus represent a discontinuous record in time. Variations in Δ¹⁸O preserved over the four annual cycles are in excellent agreement with observed variations in temperature at the coring site (Fig. 6). The narrow summer peak and broad winter trough in Δ¹⁸O for all four years is reflected in the AWS air-temperature observations. In addition, many non-summer peaks in Δ¹⁸O are also accompanied by peaks in air temperatures. This suggests a strong relationship between Δ¹⁸O and air temperatures at Law Dome. The results are in agreement with Reference Van Ommen and Morganvan Ommen and Morgan (1997), who compare Δ¹⁸O values from a composite seasonal signal calculated from 700 years of Law Dome data with a modern 5 year composite in air temperatures from an AWS 5km north of the drilling site. However, this longer-term study only considers the broad seasonal cycle in Δ¹⁸O as fine-resolution features such as non-summer peaks are smoothed by Δ¹⁸O diffusion in the deep ice record. Figure 6 indicates a strong relationship between Δ¹⁸O and air temperatures at Law Dome at a resolution finer than the annual cycle. In addition, Figure 6 provides a direct comparison between contemporaneous Δ¹⁸O and site temperature observations on an event-by-event basis. Thus, the results presented here provide further evidence for the use of Δ¹⁸O from deep ice cores as a proxy for palaeotemperatures.

The relationship between Δ¹⁸O measurements and observed site temperatures can be used to calibrate the ice-core palaeothermometer. The dependence of observed Δ¹⁸O measurements on temperature is generally well approximated by a linear relationship:

The slope coefficient, α, can be derived either from spatial variability over a region or from temporal variations at a site. A spatially derived slope of 0.6–0.7%˚C^–1 has been determined for the Law Dome region (Reference MorganMorgan, 1979), and a temporally derived slope of 0.44%˚C^–1 has been calculated from 700 years of annual cycles in Δ¹⁸O at Law Dome (Reference Van Ommen and Morganvan Ommen and Morgan, 1997). More recent studies from a short length of Law Dome firn core examined the correlation between Δ¹⁸O and temperature values on an event-by-event basis (Reference McMorrow, Curran, van Ommen, Morgan, Pook and AllisonMcMorrow and others, 2001). In this study, a first attempt was made to account for the episodic nature of net accumulation and potential temperature biases preserved in the ice core. The temporal calibration determined from Reference McMorrow, Curran, van Ommen, Morgan, Pook and AllisonMcMorrow and others (2001) yields a Δ¹⁸O temperature slope of 0.21%˚C^–1, and agrees with the finding from Reference Van Ommen and Morganvan Ommen and Morgan (1997) that temporally derived calibrations give lower slopes than spatially derived calibrations.

The similarity between the values calculated in this work and Reference McMorrow, Curran, van Ommen, Morgan, Pook and AllisonMcMorrow and others (2001) indicates consistency in event-scale calibration between Δ¹⁸O and temperature. The similar calibration coefficients calculated in each study provide evidence for the accuracy in determining temporal Δ¹⁸O temperature calibrations using this event-by-event technique.

The observed slope of temporal calibrations will be influenced by isotopic diffusion in the firn column, which reduces the amplitude of the annual cycle in Δ¹⁸O. The calibration coefficients developed in Reference McMorrow, Curran, van Ommen, Morgan, Pook and AllisonMcMorrow and others (2001) and this work are not corrected for isotopic diffusion in the firn. The degree to which diffusion affects Δ¹⁸O values at shallow depths is not well understood, but may have decreased the amplitude of the Δ¹⁸O annual cycle by up to 30%, which is the integrated effect over the full firn column at DSS (Reference Van Ommen and Morganvan Ommen and Morgan, 1997). Therefore, the 0.24%˚C^–1 calibration determined here may be corrected upwards at most by a corresponding factor. Correcting the 0.24%˚C^–1 calibration upwards by 30% yields a maximum corrected coefficient of 0.31%˚C^–1. This value is still considerably lower than the temporally derived value of 0.44%˚C^–1 (corrected for diffusion) determined by Reference Van Ommen and Morganvan Ommen and Morgan (1997). Thus, correlating Δ¹⁸O and temperature at the event level appears to reflect a genuine difference in the Δ¹⁸O temperature calibration compared to longer-term, seasonally based studies.

The difference between the two temporal calibration techniques invites speculation as to the appropriate coefficient to use for interpretation of the longer-term Δ¹⁸O record. Assuming that diffusion effects in the shallow firn-core material used in this work can correct the event-scale coefficient of 0.24%˚C^–1 upwards by a maximum of 30%, then another explanation for the difference between this coefficient and the longer-term temporal calibration calculated by Reference Van Ommen and Morganvan Ommen and Morgan (1997) is required. A number of factors have been identified as likely contributors to lower temporal than spatial values for the Δ¹⁸O temperature calibration, and these factors may also explain differences in temporal values. First, snowfall typically occurs when the temperature is warmer than average, and this has been shown to affect the temporal calibration (Reference Peel, Bradley and JonesPeel, 1992). It is possible that the event-scale temporal calibration determined in this work accentuates the effects of such a temperature bias in the firn core. The calibration calculated by Reference Van Ommen and Morganvan Ommen and Morgan (1997) will be influenced by this warm-temperature bias during accumulation events. In contrast, the event-by-event calibration of 0.24%˚C^–1 inherently accounts for the potential warm bias, as only temperatures associated with net accumulation events preserved in the firn core are used in the calculation. This will have the effect of increasing temperatures compared to Δ¹⁸O values in the calibration, thus reducing the overall calibration coefficient.

Another important factor that will influence temporal Δ¹⁸O temperature calibrations involves temperature variations at the moisture source. A change in temperature at both the source and deposition sites will reduce the relative temperature variation and lead to reduced fractionation and smaller fluctuations in the Δ¹⁸O signal. The short-term study conducted here (four annual cycles) samples the modern climate regime for the Law Dome region. Temporal calibrations from long-term ice-core records sample a wide range of different climate regimes with greater variations in moisture source and deposition temperatures. Thus, these longer-term coefficients will include more fluctuations in climate cycles and may be more appropriate for calibrating the long-term isotope palaeothermometer.