Criteria for Volcano Selection
On Landsat images of the Central Andes there are well in excess of 1,000 topographic features whose distinctive morphologies indicates that they are volcanoes. Clearly, it would be impractical to provide descriptions of all these volcanoes, although we have attempted to catalog them (Appendix II). Furthermore, it is clear from their degraded morphologies that most of these volcanoes can be confidently identified as "extinct"; i.e. unlikely to erupt again. Given the hyperarid environment of the Atacama desert and the slow erosion rates, many of these may be as much as 5-10 million years old. In this compilation, we have attempted to select only those volcanoes that are potentially active. As explained earlier, we eschew the plain term active deliberately, since it can cause confusion.
Identification of a volcano as potentially active is a difficult but important task. We note that several disastrous eruptions of modern times have taken place on volcanoes not thought to be 'active' (e.g. El Chichon, Mexico, 1982), and that the eruption of Mt. Lamington (New Guinea) in 1951 took place on a mountain not even recognised as a volcano. It is relevant to record that the committee for the International Decade for Natural Hazard Reduction (IDNHR) has determined that the "identification and global mapping of all active and potentially active volcanoes" should be an important science objective for the IDNHR (National Research Council, 1987). Identification of potentially active volcanoes from remote sensed images alone is clearly more difficult than when ground studies are possible, but fortunately the enviromental conditions in the Central Andes are optimal.
Morphological criteria
In the absence of current eruptions or documented historic records, a convenient, but not universally accepted, criterion for regarding a volcano as potentially active is that it should exhibit evidence of having erupted during the last 10,000 years (Simkin et al. 1981). In areas like north America, there are some well controlled radiocarbon dated stratigraphies for major volcanoes, but in the Central Andes, radiometrically controlled stratigraphies for volcanoes are absent, so morphological criteria have to be used for those volcanoes for which no more obvious evidence, such as the presence of visible fumaroles is present.
The most useful objective morphological criterion we have used is evidence of eruptive activity which post-dates the last major glaciation in the Central Andes. It has been known for many years that the central Andean volcanoes were extensively glaciated down to altitudes of ~4,300 m (Hollingworth & Guest 1967; Clapperton, 1983). Only one detailed radiocarbon dating study of the timing of central Andean glaciation has been carried out, on the Quelccaya ice cap in southern Peru (Mercer & Palacios, 1977). This ice cap is some 250 km from the nearest volcano that we have studied, but provides a vital frame of reference for our geomorphological observations (Figure 9). Mercer & Palacios' critical observations are firstly that after a major re-advance about 11,000 yr BP (Huancane II stage), the Quelccaya ice cap shrank rapidly until by 10,000 yr BP it was little, if any, larger than at present; and secondly that there was a further minor readvance between 600 and 300 BP corresponding to the world-wide "Little Ice Age" (Grove, 1989, Thompson et al. 1986).
Fortunately, the 30 m pixel size of the TM is sufficiently small that moraines left by ice caps and valley glaciers are unequivocally recognisable. While several sets of moraines are visible on some volcanoes, in the absence of any other evidence except that of Quelccaya, we assume that the most prominent correspond to those left by the last major ice regression (10,000 yr BP; Mercer and Palacios' Huancane II moraines), and use these moraines as our most consistent time marker. These are the moraines which extend down to elevations as low as 4,300m. The "Little Ice Age" moraines of Quelccaya extend only slightly beyond the limits of the present ice cap. On many volcanoes, for example Tacora (VCA no. 9), we observe small moraines sets at high elevations, and we interpret these as corresponding to the Little Ice Age advance, and have used them accordingly to provide a second time marker. Clearly, it would be highly desirable to have independent radiometric confirmation of these correlations, but there seems no prospect of this at present. It is possible that other, older, sets of moraines are present on some volcanoes.
Two "end member" volcano types are therefore easily recognisable:
* Those that have been heavily glaciated and have clearly not experienced any activity since the last deglaciation, and
* Those that show no signs of glaciation, and whose surfaces appear to have been wholly formed in post-glacial times.
Volcanoes which present more difficult problems for interpretation are those which exhibit typical glacial features like valley moraines, but which also exhibit apparently youthful volcanic features. Given that many volcanoes are large mountains over 6,000 m high, with edifices up to 3,000 m above their surroundings, it is not surprising that some exhibit evidence of long and complex glacial and volcanic histories. Many volcanoes are clearly edifices which were largely constructed before the 10,000 yr BP ice regression but have experienced some more recent volcanic activity. Where relationships between volcanic and glacial features are not clear, we have taken a volcano to be potentially active if it possess one or more of the following features:
* A summit crater with pristine morphology
* Flank lava flows with pristine morphology
* Flank lava flows with low albedos.
Lava flows require careful study. On the lower, drier slopes of volcanoes their morphologies tend to be much better preserved than in the summit region, where they are exposed to ice and snow. Thus, fresh lava morphologies at high elevations are much more significant than those at lower elevations. Lava flows also brighten with age as a result of weathering, erosion and accumulation of wind blown dust. Low albedo flows are therefore the best evidence for youthful activity (Rothery & Francis 1984, Rothery et al., 1986).
Lake terraces
Related glacial features that we have found useful in providing time markers are glacial pluvial lake terraces. These are well developed around the rim of the ancient lake basin now occupied by the Salars de Uyuni and Coipasa, and which in glacial times was occupied by Lake Tauca (Figure 10). This lake was similar in extent to Lake Bonneville, the ancestral Great Salt Lake, Utah. Like Bonneville, Lake Tauca exhibits a series of terraces which are clearly visible on TM images. The youngest of these has been dated at ~10,000 years (Servant & Fontes, 1978; Clapperton, 1983). These radiocarbon dates are the only dates of lake terraces in the Central Andes, and it is not yet clear whether the pattern of high and low lake levels can be matched with those in the northern hemisphere. The terraces, however, provide a means of dating several major volcanoes. For example, the major debris flow of Ollague volcano, Chile is older than the youngest terrace, while that from Tata Sabaya, Bolivia, is younger.
Volcan Ollague (VCA no. 18) is an interesting example of a volcano whose history can be well constrained; its debris avalanche clearly pre-dates the last high stand of the lake, but it summit region also shows a partial girdle of moraine at approximately 5,000 m. Ollague has significant fumarolic activity at the present day. We infer that the high level moraine represents the Little Ice Age advance; thus we infer that Ollague has not erupted in the last several hundred years.
Regional variations
An important difficulty in applying glacial geomorphological criteria to dating volcanoes in the central Andes is that there are major climatic differences between the northernmost part of the study area (14ºS) and the southernmost (28ºS). Volcanoes in the the north such as Coropuna in Peru (VCA no. 1) are heavily covered by snow and ice at the present day, while those at similar or higher elevations in the south such as Ojos del Salado (VCA no. 43) are virtually ice free. Similar variations appear to have existed during glacial times, since there is little evidence for obvious moraine deposits south of about 24ºS. While the contemporary ice and snow cover creates some problems for interpreting images of the northern volcanoes, the chief problems actually arise with those in the south: the area appears to have been hyper-arid throughout the Late Tertiary (Galli-Olivier, 1967, Mortimer, 1973). Erosion rates have therefore been exceedingly slow and it is consequently difficult to distinguish between recently active volcanoes, and those that may have been inactive for tens of thousands of years. Examples:
Volcan Palpana (21º32'S, 68º31'W, altitude 6,023 m) is an example of a volcano which we have determined to be not potentially active (Figure 11) . Although presenting a youthful, conical shape this major volcano does not exhibit any well preserved lava flows on its flanks and has parasol-type radial drainages incised on the upper slopes, in which valley and terminal moraines are recognisable. Additionally, its southern flank has a large amphitheatre shaped hollow, which has clearly been glaciated.
Volcan Licancabur (VCA no. 23) is an example of a closely similar type of volcano, which, although not listed in the Catalog of Active volcanoes of the World by Casertano (1963), should be regarded as potentially active. The cone is steep; has a pristine summit crater containing a small warm crater lake, and the flanks display numerous pristine lava flows. Although of similar altitude to Palpana, the cone shows no evidence of glacial deposits, and appears to have been active since the 10,000 yr BP deglaciation.
Volcan Paniri (22º03'S; 68º14'W; altitude 5,946 m) is an example of a more problematical volcano. A similar composite cone to Palpana and Licancabur, Paniri has what appears to be a relatively well shaped summit crater and fairly well preserved lava flows on its flanks(Figure 12). However, it also exhibits unambiguous evidence of valley glacier deposits on its south east flanks, and aerial photographs reveal that most of the summit region is extensively degraded. This coupled with the lack of any unambiguously pristine lava flows on the upper parts of the edifice have led us to exclude Paniri fromVolcanoes of the Central Andes. A few other volcanoes have posed similar problems (i.e. Pular-Pajonales, VCA no. 32). Doubtless, more detailed future studies will prove some of our interpretrations to be erroneous, and will lead to re-assignation of individual volcanoes from one category to another.
Monogenetic centres and large silicic caldera systems
The discussion so far has dealt only with large composite conical volcanoes that have obvious individual identities. In attempting to identify the most likely types of future volcanic activity (as opposed to identifying volcanoes that have erupted in the recent past), eruptions taking place in two other volcanological contexts must be considered: small monogenetic events, and those related to the evolution of large silicic calderas.
In some areas (such as central Mexico) basaltic scoria cones are extremely abundant, and there are historical records of the eruption of new cones such as Paricutin, Mexico, which was active between 20 February 1943 and 1952 (Foshag & Gonzalez, 1956). There are no such records for the Central Andes where scoria cones are also common, but a new cone could potentially erupt at any time anywhere within the province. In fact, many of the morphologically most youthful volcanos in the Central Andes are small monogenetic scoria cones and their associated lava fields. These are important because they are exemplars of likely future volcanic constructs. An unrelated type of likely future activity is the eruption of small monogenetic silicic lava domes, of which there are many in the region. We have described several of the best examples of these youthful monogenetic centers in section 3.
Large silicic caldera systems are long lived and may erupt many times over periods of millions of years (Smith & Bailey, 1968). While no major caldera system appears to have erupted in the Central Andes within the last million years, there are many young lava domes, dome complexes and geothermal manifestations associated with calderas between 1 and 8 million years old. Such features are extremely important as they indicate active magmatic systems beneath these calderas, which we believe should be considered as likely to erupt in the future; perhaps catastrophically, or more probably in minor eruptions of silicic lavas and pyroclastic rocks. We have described large silicic systems in section 4.