3.1. Densification model

The 10 msnow temperature at H72 is estimated to be –20.3˚C based on the 10m depth snow-temperature distribution in east Dronning Maud Land (Reference SatowSatow,1978). Bulk density of the entire ice core was measured using an electronic balance, a scale and a caliper. It was measured from the surface to the core bottom (73.285m depth) at about 40 cm intervals. Figure 3 shows the ice-core density data with a ``best-fit’’ profile by the empirical densification model of Reference Herron and LangwayHerron and Langway (1980). An average surface mass balance of 317 mma<sup>–1</sup> w.e. is obtained from the model. for the above calculations, the following values are used: surface snow density (400 kg m<sup>–3</sup>), 10 m snow temperature (–20.3˚C), ice density ρ<sub>i</sub> (917 kg m<sup>–3</sup>), k<sub>1</sub> (0.0218m<sup>–1</sup>) and the slope of the second stage of the densification profile (0.0355m<sup>–1</sup>). the parameter k<sub>1</sub> is defined by equation (6b) in Reference Herron and LangwayHerron and Langway (1980), and ``second stage’’ refers to the density region 550–800 kg m^–3.

Fig. 3 Depth–density data with calculated profile by Reference Herron and LangwayHerron and Langway (1980).

3.2. Time markers by volcanic events

The acid concentration derived from H₂SO₄, HCl and HF is higher in the ice-core signals from volcanic eruptions (e.g. Reference HammerHammer, 1980; Reference Legrand and Delmas.Legrand and Delmas, 1987; Reference Moore, Narita and MaenoMoore and others, 1991). We found several high peaks of SO₄ ^2– and Cl^– in the H72 ice core, but these signals do not indicate significant peaks of elevated concentration. We consider that the peaks of Cl^– are not of volcanic origin but probably originate from sea water since the ratios of Cl^– to Na⁺ at all peaks are similar to that of sea water. the SO₄ ^2– has several origins (e.g. sea water, marine biogenic dimethylsulfide (DMS) and volcanic eruptions). We employed the profile of non-sea-salt SO₄ ^2– (hereafter nssSO₄ ^2–) concentration to detect volcanic signals. the nssSO₄ ^2– in the H72 ice core indicates a periodic seasonal fluctuation caused by marine biogenic DMS, and the maximum value occurs during summer (Reference Ayers, Ivey and GilletAyers and others, 1991; Reference OsadaOsada, 1994; Reference Koga and TanakaKoga and Tanaka, 1996). the highest values are almost the same level as the 2σ level (0.76 μmol L^–1) of nssSO₄ ^2– data from the surface to the bottom (73.16m depth).

We found 11 spike-like signals exceeding 2σ in the nssSO₄ ^2– data and assume that these peaks are candidates for volcanic signals. Using pH and methanesulfonic acid (MSA) concentration at the same depth, we classified the spikes into two categories since methanesulfonic acid (MSA) is formed by atmospheric oxidation of DMS (Reference Savoie, Prospero, Larsen and Saltzman.Savoie and others, 1992). the characteristics of the two categories are as follows. In categoryA, the nssSO₄ ^2– value exceeds 2σ, and both pH and MSA are lower than their average values (5.32 and 0.046 μmolL^–1, respectively); therefore, these signals show the highest reliability in volcanic signals. In category B, the nssSO₄ ^2– value exceeds 2σ, and either the pH or MSA is higher than its average value. These signals show lower reliability than categoryA as indicators of volcanic signals.

We also employed ECM current to detect volcanic signals since ECM current reflects acidity in ice cores (e.g. Reference HammerHammer, 1980). ECM current depends on ice-core density, showing a lower background current at a shallow depth of 24m (density ~650 kg m^–3) than in deeper sections. the spike-like signals in ECM current exceeding 2σ (1.41 μA above 24 m depth, and 2.76 μA below 24 m) are considered to be derived from volcanic acids such as H₂SO₄, HCl, HF (Reference ClausenClausen and others, 1997). Using H⁺ concentration at the same depth, we classified these spikes into two categories as follows. In category A, ECM current is >2σ and H⁺ concentration is high; therefore, these signals show the highest reliability as indicators for volcanic signals. the signals in category B have ECM current >2σ, but the H⁺ concentration is lower than average (5.01 μmol L^–1); they have a lower reliability as volcanic indicators.

Using the two procedures described above, reliability numbers for each volcanic signal are given in Table 1. Reliability ``1’’ is the highest level, with both the nssSO<sub>4</sub> <sup>2–</sup> and ECM signals in category A. Reliability ``2’’ is the second highest level, corresponding to a single A rating for either the nssSO₄ ^2– or the ECM signal. Reliability ``3’’ is the lowest level, corresponding to remaining events with a B rating for either nssSO₄ ^2– or ECM. Probable source volcanoes for each volcanic signal are investigated using Reference Simkin and SiebertSimkin and Siebert (1994) and Reference Kohno, Fujii, Kusakabe and FukuokaKohno and others (1999) as shown in Table 1.

Table 1. Volcanic signals and their chemical compositions found in the H72 ice core, and their probable source volcanoes

The symbol ``?’’ is added at the end of the names of the source volcanoes for which the reliability of the signals is ``3’’. It is possible that signal b originates from an eruption by Nevado del Ruiz, Colombia, in 1985 (volcanic explosivity index (VEI) = 3), and signal l from Purace, Colombia, in 1869 (VEI = 3). the difference between ages counted by ECM current and volcanic eruption years is also shown in Table 1. the age difference is in the range 0 to +3, which is on the same order as the transportation time from volcanoes to the Antarctic ice sheet.

3.3. Dating by ECM current, visual stratigraphy, tritiumcontent, stable isotopes and major chemical ions

ECM was carried out continuously for the H72 ice core. Reference NishioNishio and others (2001) describe the instrumentation and the analytical procedure used. We can detect the acidity of an ice core with high resolution by ECM current (Reference HammerHammer, 1980), and obtain the total amount of H⁺ ions (Reference Hammer, Oeschger and LangwayHammer, 1989). There are seasonal variations of acidity in snowfall. for instance, high sulfuric acid related to regional oceanic production of sulfur-rich gases, methanesulfonic acid (MSA), is generally detected in January around the Antarctic coastal region (Reference Legrand, Wolff and WagenbachLegrand and others, 1999). These seasonal characteristics are used to detect summer layers in an ice core by ECM.

The characteristics of visual stratigraphy in the H72 ice core are associated with atmospheric conditions. An example is the number of ice-crust layers contained in the H72 ice core. Ice-crust layers are classified into two types according to whether or not the ice crust includes air bubbles, based on observation by eye from digital video images. One is a multilayered ice crust (CR-1) including air bubbles. the other is a single-layered ice crust (CR-2) without air bubbles, which was probably formed by the melting and refreezing process at the snow surface during the warm period (Reference Narita and WatanabeNarita and Watanabe, 1977). Figure 4 shows profiles of visual stratigraphy, ECM current, δ¹⁸O, nssSO₄ ^2–, MSA and NO₃ ^–. Visual stratigraphy is composed of mosaicked pictures, which are continuously taken by digital video camera along the ice core. the high ECM current with ice-crust layer (CR-2) is apparently identified as a summer layer as shown at depths of 19.45–19.50, 20.17–20.21, 20.74–20.79 and 21.09–21.14m. the nssSO₄ ^2– and NO₃ ^– concentrations also indicate high peaks as summer layers at the same depths of ice core; on the other hand, the high δ¹⁸O peak is found at slightly shallower depths, so it seems to appear a few months later than the ECM peak. Thus, we count seasonal variations by the high ECM current with ice-crust layer (CR-2) for the whole H72 core, and we define this dating as the ECM age. Althoughthe δ¹⁸O peak indicates seasonal variations in Figure 4, the seasonal variation of δ¹⁸O becomes unclear below about 40 m depth. When the ECM age and the δ¹⁸O age are compared with the time marker of volcanic events listed in Table 1, the dating by the δ¹⁸O age is younger than the ECM age. the comparison for ECM age with the time marker of volcanic events is discussed in section 3.4.

Fig. 4 Profiles of visual stratigraphy, ECM current, δ¹⁸O, nssSO₄ ^2–, MSA and NO₃ ^– from 19.40 to 21.60 m. Visual stratigraphy is composed of mosaicked pictures continuously taken by digital video camera along the ice core. Ice crust CR-1 is multi-layered and includes air bubbles. Ice crust CR-2 is identified as probably summer layers at depths of 19.45–19.50, 20.17–20.21m, 20.74–20.79 and 21.09–21.14 m. C.B. signifies a break in the core.

Figure 5 shows tritium content, snow-stake record at H72 (as shown in Fig. 2) and ECM current from surface to 25m depth. Solid lines below the ECM age indicate the annual-layer boundary detected by ECM current with CR-2 type ice-crust layer. the ECM age agrees with the snow-stake record back to AD 1972 from the surface to 15 m depth. As shown on the left side in Figure 5, we analyzed tritium content from 2.17 to 24.85 m and detected seven tritium peaks at depths of 14.00–14.25, 15.24–15.49, 16.22–16.46, 17.69–17.94, 19.39–19.57, 20.47–20.68 and 22.60–22.85m. According to AD 1978 pit results at the South Pole (Reference Jouzel, L., Pourchet and LoriusJouzel and others, 1979), we can identify these peaks as AD1973, 1971,1969,1966,1963,1962 and 1958, respectively. ECM ages are closely coincident with tritium ages of AD1963,1962 and 1958, however; the other four tritium ages of AD 1966,1969, 1971 and 1973 are about 1 year older than the ECM ages. the reason for the age difference is not clear, since substances from the Northern Hemisphere (NH) will transfer to the interior of the Antarctic ice sheet through the stratosphere, and transportation time from NH to the South Pole is considered to be shorter than that from NH to the coastal region at the H72 site (Reference Kamiyama, Ageta and FujiiKamiyama and others, 1989).

Fig. 5 Comparison of dating with tritium, snow-stake record and ECMfrom snow surface to 25 m depth. the solid line indicates annual boundary detected by ECM current with visual stratigraphy (CR-2). Arrows show the ages measured by snow stakes. the tritium profile is compared with ECM ages from the surface to 25 m depth. Seven tritium peaks are inferred from the South Pole (Reference Jouzel, L., Pourchet and LoriusJouzel and others, 1979) except for the unknown peak of 1991.

3.4. Annual-layer boundary by ECM current profile

As described in section 3.3, ECM current detects the acidity variations of ice due to seasonal signals such as oceanic sea-salt ions, marine biogenic origin material and background nitric acid, and non-seasonal signals caused by volcanic eruptions. Therefore, we must distinguish between seasonal and volcanic components in the ECM-current data.

To discriminate between the non-seasonal signal and the ECM current, the ECM current is compared with visual stratigraphy in Figure 4. Generally, a winter layer has a low ECM current and includes an ice-crust layer (CR-1) in the visual stratigraphy; on the other hand, high ECM currents often appear not merely in summer layers but also in spring or autumn layers due to oceanic acidity, so that high ECM current is sometimes detected twice or more in an annual layer. An annual boundary, determined by ECM current with visual stratigraphy, is identified at the transition boundary between low ECM current and high current as indicated in Figure 4.

Figure 6 shows the dating of the entire H72 ice core obtained by ECMcurrents. the H72 core is dated as AD 1831 by the ECM age at 73.16 m depth. the ECM age agrees with the snow-stake record back to AD 1972 (from the surface to 15 m depth). Furthermore, we compared the ECM age with some time-marker volcanic events to evaluate the ECM age error below 23 m depth. As described in section 3.2, 16 volcanic signals are inferred by ECM current and chemical composition. the ECM age error was estimated by comparing with nine time-marker volcanic events labeled ``e, f, g, i, j, l, m, n, p’’, of which the reliability is 1or 2 and the probable source volcanoes of which are listed in Table1. We found that the ECM age indicates the time difference by 0–3 years with the volcanic eruptions in the probable volcanic signals. the acid fallout travel time from the NH to the Antarctic is 2– 3 years (Reference Jouzel, L., Pourchet and LoriusJouzel and others, 1979), and 0–2 years from the Equator to the Antarctic, so the maximum ECM age error is probably 1–3 years at 73.285 m depth.

Fig. 6 Profiles of ECM current of the entireH72 ice core.The solid line indicates the annual boundary detected by ECM current with visual stratigraphy.The ECM age was dated as AD1831at 73.16 m depth.

4.1. Time series of surface mass balance

Using the annual-layer boundaries described in section 3, we can determine the surface mass-balance variation over the past 167 years. Figure 7 shows the time series of annual and 10 year averaged surface mass balance. the time series of the yearly surface mass balance at H72 shows a slight increase, with 19 mm (100 years)^–1 w.e. over the last 167 years, but on the basis of t-distribution analysis this trend is not significant. Average surface mass balance is 311 mma^–1 w.e. for AD 1831– 1998, which is in good agreement with the annual accumulation of 317 mma^–1 w.e. obtained by the empirical densification method described in section 3.1, and the value of 307 mma^–1w.e. obtained by snow-stake data for AD1973–98.

Fig. 7 Surface mass balance in water equivalent at H72, AD 1831–1998. Average surface mass balance is 311 mm a^–1w.e. Although the surface mass balance is scattered during the 167 years, the average surface mass balance is almost constant within the 95% confidence limit.

4.2. Characteristics of isotopes and chemical condition for the last 167 years

Figure 8 shows the oxygen isotope ratio (δ¹⁸O), nssSO₄ ^2–, pH, MSA and NO₃ ^–. These values were continuously measured from surface to bottom at an average interval of 16.3 cm; a total of 450 samples were measured. Analytical methods are described in detail in Reference NishioNishio and others (2001). In Figure 8, the ages are ECM ages described in section 3; the thin line in the δ¹⁸O profile shows raw data, and the thick line indicates the nine-point running mean of raw data. the vertical solid line indicates the average value as –25.5‰. for the oxygen isotope record, a significant trend to lower values, with a negative gradient of 1.7‰ (100 years)^–1, can be seen in the H72 ice-core record. More detailed oxygen isotope records show a gradual increase from AD 1840 to 1880, and afterward a gradual decrease to about AD 1900; subsequently, during the 20th century, the isotope record seems to oscillate at an interval of about 20 years. According to previous studies, δ¹⁸O in an ice core is related to the temperature (Reference Johnsen, Dansgaard and WhiteJohnsen and others, 1989), sea-ice extent (Reference KatoKato, 1978; Reference Bromwich and WeaverBromwich and Weaver, 1983) and the seasonal variation of snow accumulation (Reference Steig, Grootes and StuiverSteig and others, 1994).

Fig. 8 Profiles of δ¹⁸O, chemical constituents (nssSO₄ ^2–,MSA and NO₃ ^–) and pH. Age is due to ECM age. Thick lines in the δ¹⁸O profile show nine-point running means. Possible volcanic eruptions that could give high nssSO₄ ^2– peaks (>0.76 μmol L^–1) are given to the right of the nssSO₄ ^2–profile.

The concentration of nssSO₄ ^2– shows similar variations to δ¹⁸O for AD 1840–1900 and gradually increases from AD 1900 to the present. Some spike-like peaks higher than 2σ (>0.76 μmolL^–1) are identified as volcanic eruptions, and the probable source volcanoes are shown in Figure 8. MSA is an atmospheric oxidation product of dimethylsulfide (DMS), which is produced by biological sources in the ocean (Reference Savoie, Prospero, Larsen and Saltzman.Savoie and others, 1992). Fluctuation of MSA is small for all depths, except for some spikes. It appears that the chemical constituents of snow are influenced by the ocean when these anomalous peaks are observed. In previous studies (Fisher and others, 1998), NO₃ ^– has been found to derive from the oxidation of NO and related species that are produced by anthropogenic combustion sources and natural sources, principally soil emissions, lightning and stratospheric processes (Reference Savoie, Prospero, Larsen and Saltzman.Savoie and others, 1992). In the Arctic region, NO₃ ^– concentration increases from about AD 1900 with increased consumption of fossil fuel. Although a number of spike peaks of NO₃ ^– are observed in the H72 core, the basic concentration of NO₃ ^– is almost constant. the pH value has increased over the last 30 years. Therefore, we consider that the NO₃ ^– in the spike-like peaks originates from natural sources, and not from anthropogenic combustion sources.