Geology and geochemistry of the mansion pyroclast fall succession, St. Kitts

Abstract

Some 4000 years ago Mt. Misery volcano was in a particularly active state, emitting a sequence of pyroclastic deposit that are widely distributed over the island and show a compositional range from basalt (SiO₂ 48%) to andesite (SiO₂ 62%). The type section at Mansion, on the east coast, has been the subject of a study byBaker andHolland (1973).

Of special interest in this succession is the intimate association of basic and relatively salic products. It constitutes a detailed record of a short period (a few centuries?) in the volcano’s history, which properly interpreted may tell us something of the processes of magmatic differentiation and replenishment. The compositional patterns may also provide some guide to the course likely to be followed in future eruptions.

Some degree of caution is necessary in considering the chemistry of pyroclastic rocks, which between eruption and deposition may have been influenced by aerial fractionation or winnowing processes. There may also be problems of partial redistribution, weathering and also the inclusion of accessory or accidental lithic fragments.

Isopach maps show that deposition of the Mansion succession was partly governed by the prevailing ENE wind. The total thickness varied from about 45 m at a distance of 4 km west of the crater compared with about half that thickness an equal distance east of the crater. The more complete sections occur on the eastern, windward side, whereas on the west much of the upper part has been removed, presumably by mudflows and floods. The marked unconformity over the basic cinder zone in western sections makes correlation with eastern sections more difficult.

Sharp changes in composition and eruptive pattern tend to be heralded by particularly coarse horizons which usually contain fragments of coarse grained cumulates: these were presumably dislodged by the influx of new magma from depth. The coarsest horizon of all precedes the most basic phase in the middle of the sequence, when basalt flows were also discharged. It is not possible to identify individual basaltic horizons over any distance in the field but correlation does become possible when chemical profiles are integrated with stratigraphic data. It is generally sufficient to correlate by means of one elemente.g. Mg. The basaltic units are both thicker and coarser to the west of Mt. Misery. The most mafic beds have been found at Mansion in the east, probably because of the westward winnowing of the least dense fractions.

Units of coarse greenish angular andesitic lapilli which occur near the top of the succession are the most amenable to wider correlation. There are three major units of these lapilli though often only one is exposed. Though indistinguishable in the field chemical analysis reveals that the middle unit is decidedly more basic and this becomes a useful criterion in correlation. It has been demonstrated that there are slight but significant variations in the chemistry of the upper unit over St. Kitts. Samples from the west are relatively enriched in SiO₂ (61.5%) compared with those to the east of the crater (59.0%). The pattern of variation can be matched closely to the isopachs, pointing to an influence of the wind on the ultimate composition of the deposits: presumably the less dense fractions were enhanced downwind. As a test of consistency, four samples were analysed from different heights in a single unit of andesitic lapilli but there was no significant difference.

The results suggest that when conventional stratigraphic methods fail, chemical profiles may play a useful role. The pattern of variation in these profiles also suggests that basaltic andesites and perhaps some andesites are derived by fractional crystallization of basaltic magma. However, other andesites which break the pattern and appear suddenly in large volume may have a quite independent origin.