Info

Note: The decomposition of the series was carried out using singular spectrum analysis (SSA; M = 31).

Note: The decomposition of the series was carried out using singular spectrum analysis (SSA; M = 31).

Several indexes have been proposed to estimate the strength and persistence of the effects of volcanic eruptions on the atmosphere and climate. We used two indexes: the VEI (Newhall and Self, 1982) and the sulfate loads derived from the ice cores. The parameterization of volcanism in the VEI chronology has been criticized as an index of explosivity and not of stratospheric loading. To produce a more accurate index, the residence times of the aerosol and particles ejected must be taken in account. However, we used the VEI only to assess the relative magnitude of each explosion and to select on this basis the explosions included in further analysis.

Ice cores can provide estimates of atmospheric sulfate contents, but are not always dated precisely and can be biased. It is recognized that high-altitude ice cores may overemphasize mid- and high-latitude volcanic eruptions (Zielinski, 1995). They do, however, provide direct evidence of eruptions, although not the nature of the climatic response. While the historical series for these forcing agents are imperfectly known or measured, they do nonetheless represent our best estimates of the time histories of the corresponding forc ing. We assumed, as did Sato et al. (1993), that the aerosols were uniformly distributed over the globe.

9.3.3. Anomalies in the Temperature Reconstructed Series

It is possible to find several statistical techniques used for evaluating volcanic signals in the literature proposing volcanic impact on climate. Most of them are applied to monthly surface and ocean temperature series and focus on the impact of selected eruptions (Mass and Portman, 1989; Stothers, 1996; Kelly et al., 1996; Robock and Mao, 1995). Other authors worked with annually resolved proxy series and noted the occurrence of minimum values in the series that are coincident with the dates of major explosive eruptions (Briffa et al., 1998; D'Arrigo and Jacoby, 1999).

We used a simple t-test to determine temporal and spatial anomaly patterns in the series, and we related to such episodes the significant low-temperature anomalies occurring immediately or 1 year after known volcanic explosions. The results are shown in Fig. 1.

It seems rather evident that some eruptions are associated with a coherent spatial pattern in both North America temperature reconstructed series and Northern Hemisphere composite reconstructions. By contrast, few eruptions have affected the South America series and the Southern Hemisphere composite reconstruction in a consistent way.

The eruptions of Komagatake (42.07°N) in 1640 and Awu (3.67°N) in 1641 are reported as especially rich in sulfur dioxide (Briffa et al., 1998) and are probably responsible for the cool spikes observed in most parts of the North America and Northern Hemisphere reconstructed series. No effect is evident on South America. Only the Lenca reconstruction shows some effect. Co-incidentally, the eruption of Volcano Llaima, at 38.7°S, occurred in 1640, very close to the site where the Lenca chronology was derived. This result suggests a local, limited effect.

In the decade from 1663-1673, four volcanoes are known to have erupted: Usu, Tarumai, San Salvador, and Gamkonora. The first two were northern midlati-tude volcanoes that erupted in 1663 and 1667, respectively. Both explosions were scaled as VEI 5. The two others are equatorial volcanoes and their eruptions were scaled as VEI 4. Cole-Dai et al. (1997) attributed to these volcanoes the spikes of sulfates that appear in two Antarctic ice cores. Some significant cool departures have been observed only in the northern North America series as of this time.

According to Rampino and Self (1982), high northern latitude eruptions (>40°N) usually disperse aerosols

FIGURE 1 Mean differences. The composite wave modes (2-49.9 year) of the reconstructed series were standardized over the entire period to mean 0 and standard deviation 1. A 7-year running mean was calculated and tested for significance against the general mean (0) using a common t-test (p <0.1). Each dark block indicates points where the 7-year mean is significantly lower than 0 and coincides with the occurrence of one of the volcanic explosions listed in Table 2. The arrows indicate the explosions of E1: Komagatake, Llaima, and Awu; E2: Usu, Tarumai, San Salvador, and Gamkonora; E3: Chikurachki, Hekla, and Long Island; E4: Katla-Qraefajokull; E5: Katla and Jorullo; E6: Raikoke, Asama, and Lakagigar; E7: Soufrière, Mayon, and Tambora; E8: Cosequina; E9: Krakatau and Tarawera; E10: Santa María, Pelée, Soufrière, and Ksudach; and E11: Puye-hue and Cerro Azul.

FIGURE 1 Mean differences. The composite wave modes (2-49.9 year) of the reconstructed series were standardized over the entire period to mean 0 and standard deviation 1. A 7-year running mean was calculated and tested for significance against the general mean (0) using a common t-test (p <0.1). Each dark block indicates points where the 7-year mean is significantly lower than 0 and coincides with the occurrence of one of the volcanic explosions listed in Table 2. The arrows indicate the explosions of E1: Komagatake, Llaima, and Awu; E2: Usu, Tarumai, San Salvador, and Gamkonora; E3: Chikurachki, Hekla, and Long Island; E4: Katla-Qraefajokull; E5: Katla and Jorullo; E6: Raikoke, Asama, and Lakagigar; E7: Soufrière, Mayon, and Tambora; E8: Cosequina; E9: Krakatau and Tarawera; E10: Santa María, Pelée, Soufrière, and Ksudach; and E11: Puye-hue and Cerro Azul.

over only a portion of the Northern Hemisphere and, thus, have much less impact than low-latitude explosive eruptions, which can affect both hemispheres.

A more extensive cool departure was observed in most of the North America and the composite Northern Hemisphere series in 1690-1700. Three explosive eruptions are known from this decade: Chikurachki (60.3°N), Hekla (63.6°N), and Long Island (5.3°S). The Long Island equatorial volcano produced a powerful explosion between 1695 and 1700 that was scaled as VEI 6. The exact date of the explosion is not yet established. A cold anomaly is also observed on the Southern Hemisphere composite for this period. Coinciden-tally, the Antarctic ice cores registered an increase in sulfate content (Cole-Dai et al., 1997).

The Jorullo (13.06°N) eruption of 1759 is probably reflected in the cool departures observed in six of the

North America reconstructed series and in Lenca. No signal appears in the composite series. The eruptions of Raikoke (48.2°N), Asama (36.3°N), and Lakagigar (12.3°N) may have affected the north and south high-latitude reconstructed series for the years 1778 and 1783.

The decade between 1810 and 1820 has been noted as especially cool by a number of workers. The eruptions of Soufrière (16°N) in 1812, Mayon (13.3°N) in 1814, Tambora (8.25°S) in 1815, and Beerenberg (71°N) in 1818 are probably responsible for much of the cooling observed. The year 1816 often has been referred to as the year without a summer. It was a time of significant weather-related disruptions in New England and in Western Europe, with killing summer frosts in the United States and Canada (1.0°-2.5°C colder than normal). These phenomena were attributed to a major eruption of Tambora (VEI 7) in Indonesia. The volcano threw sulfur dioxide gas into the stratosphere, and the aerosol layer that formed led to striking sunsets seen around the world for several years (Sigurdsson and Carey, 1992).

In our results, this cooling effect is not as widespread as expected. The signal appears in the North Yukon, Columbia, and intermountain basin series and is well defined in the Northern Hemisphere composite reconstructions; it appears in only one of the South American series (Ushuaia). Moreover, the cool anomaly seems to have started sometime earlier, as in 1809. Some evidence (Cole-Dai et al., 1991; Cole-Dai et al., 1997) suggests the occurrence of an unknown volcanic eruption 6 years before Tambora. Rampino et al. (1979) noticed that the average global temperature had already been decreasing since 1810 and then rose again in the 1820s. They also pointed out that the decade 1810-1820 coincided with the most pronounced low in the mean sunspot record of the last 250 years, indicating a lower solar ultraviolet output. This was also a time of low intensity in the solar-terrestrial magnetic wind. These factors plus the volcanic loads may have contributed to the generally cooler climate.

In the South American series, a cool period appears coincidentally with the explosions of Galunggung (7.25°S) in 1822. No signal seems to have been recorded in any of the North American series for that date.

Volcano Cosequina (12.9°N) exploded in 1835 (VEI 5) and probably contributed to the cool pattern observed in three North American reconstructions and two of the Northern Hemisphere composites. No effect was observed in any of the Southern Hemisphere series.

The possible climate effects of the Krakatau (6.12°N) eruption have been the subject of many studies. The consensus seems to be that the depression of Northern Hemisphere temperature was ca. 0.3°-0.4°C for 1 or 2 years following the Krakatau eruption (Jones and Wigley, 1980). The eruptions of Tungurahua (1°S) and Tarawera (38°S) in 1886 may have contributed to the general cooling in the late 1880s (Rampino and Self, 1982). In our data, the signal attributable to these volcanoes is rather weak.

The first two decades of the twentieth century were particularly rich in major volcanic eruptions. Soufrière (16.4°N), Santa María (14.3°N), and Mount Pelée (14.7°n) in 1902; Ksudach (51.2°N) in 1907; Tarumai (42.06°N) in 1909; and Novarupta (58.27°N) in 1912 share a VEI of 5 or higher. The sulfur load has been reported as very high in this period. The reconstructed series show a widespread anomaly, especially in those for North America.

Four explosions have been reported in the Southern

Hemisphere during the period from 1918-1932: Tun-gurahua (1.47°S), Manam (4.1°S), Puyehue (40.57°S), and Cerro Azul (35.6°S). These eruptions could perhaps be linked to the cooling periods observed in some of the South American series and in the Southern Hemisphere composite reconstruction in the same time frame. Hekla (64°N) in 1947, Ambrym (16.2°S) and Lamington (9°S) in 1951, and Bagana (6.14°S) in 1952 are probably connected to the anomalies observed in both North and South American series.

A careful analysis of the results presented here indicates that approximately 38 of the 90 eruptions tested could be linked to significant cool episodes in the reconstructions. Most of the explosions occurred in clusters of years, and due to the window (7 years) used in the analysis, we are not able to separate their effects. The choice of the window has been a compromise between augmenting the noise and reducing the resolution of the estimation. However, several tests using other windows (3, 5, 9, and 11 years) indicate that a 7-year window gives good resolution with more stable means.

The North America and Northern Hemisphere composite reconstructions give more obvious patterns of anomalies linked to the explosions. It is interesting to notice that some of the North America and the Northern Hemisphere composite temperature reconstructions include only tree-ring-density chronologies (Briffa et al., 1998) or a mixture of tree-ring-width and tree-ring-density data (Jones et al., 1998; Mann et al., 1998). No tree-ring-density data have been used in the development of the South America and Southern Hemisphere composite reconstructions. South America and the Southern Hemisphere composite produce a less marked pattern, but apparently more linked to explosions that occurred in the Southern Hemisphere.

According to Mass and Portman (1989), the potential climatic impact of volcanic eruptions is reduced by the inhomogeneous spread of volcanic dust. The largest attenuation of solar radiation does not occur globally, but rather is confined predominantly to the eruption hemisphere.

9.3.4. Sulfate Loads and Climate Relationships

As was expected, the explosions that seem to be more influential in terms of producing cool episodes are those linked to heavy loads of sulfate aerosols in the atmosphere.

In order to investigate this relationship, we ran a superposed event analysis using different combinations of eruption dates (events) and reconstructed temperature series. In a first set of experiments, we tested all the explosion dates against the reconstructed series and

0 0

Post a comment