Magma Residence and Evolution Within the Crust

In principle, U-series isotope data can also be used to assess the residence times of lavas in shallow magma chambers beneath active volcanoes, either from variations in lavas shown to be derived from a common parental magma [e.g. George et al. (2004)] or from mineral isochrons whose ages exceed eruption ages [e.g. Volpe & Hammond (1991); Heath et al. (1998)]. Perhaps the simplest way forward is to look for systematic variations in the disequilibria observed in whole rocks (see below), so long as the halflife is appropriate to the magma residence times being investigated. One of the more robust observations is that the crustal residence time for magmas containing significant 226Ra-excesses cannot have been greater than 8000 yrs, so long as those 226Ra-excesses were produced in the mantle wedge (see Sec. 3 above). For example, in the Tonga-Kermadec and Aleutian arcs, Turner et al. [2000] and George et al. [2004] have found that Ra-Th dis-equilibria decrease with increasing SiO2 suggesting that the time scale of differentiation was on the order of a few 1000 years. An encouraging point is that these time scales are similar to those obtained in recent numerical thermal modelling based around constraints on energy loss available from the thermal output at volcanoes [Hawkesworth et al. (2000)]. The main results are summarised in Fig. 9 which shows that 10 km3 of basaltic magma losing heat at 100 MW will undergo 20% crystallisation to reach a basaltic

Fig. 8 Plot of (226Ra/230Th)o versus SiO2 showing that (226Ra/230Th)o decreases with increasing SiO2 which places constraints on the time scale of differentiation (time elapsed indicated in years along the light grey Tonga-Kermadec array). A model, instantaneous gabbroic fractionation vector shows that Ra/Th remains essentially constant during fractionation from basalt to dacite. Thus, if the observed decrease in (226Ra/230Th)o is due to the time taken for differentiation, then this must have taken less than the time for 226Ra-230Th to return to equilibrium (8000 years).

Fig. 8 Plot of (226Ra/230Th)o versus SiO2 showing that (226Ra/230Th)o decreases with increasing SiO2 which places constraints on the time scale of differentiation (time elapsed indicated in years along the light grey Tonga-Kermadec array). A model, instantaneous gabbroic fractionation vector shows that Ra/Th remains essentially constant during fractionation from basalt to dacite. Thus, if the observed decrease in (226Ra/230Th)o is due to the time taken for differentiation, then this must have taken less than the time for 226Ra-230Th to return to equilibrium (8000 years).

andesitic composition in about 1000 years, 60% crystallisation to reach a dacitic composition in about 3000 years and closer to 5000-8000 years to reach a rhyolitic composition. These estimates can be directly compared with those derived from the Ra-Th disequilibria data on Fig. 8.

Studies of mineral ages have met with many complications [see Hawkesworth et al. (2004) for a recent review]. Detailed studies of mineral separates have revealed evidence that phenocryst populations may often have mixed ages and/or consist of old cores with young rims such that the U-Th and Ra-Th systems give differing ages [Cooper et al. (2003); Turner et al. (2003)]. Thus, there is growing evidence that phenocrysts within these lavas could be older than the estimated ages for the liquids and may reflect incorporation of older cumulate or wall rock materials into young magma batches [e.g. Pyle et al. (1988); Sparks et al. 1990; Heath et al. (1998)]. The corollary is that the observed phenocrysts were not the ones responsible for differentiation of their enclosing liquid and Sr isotope profiles in plagioclase phenocrysts provide independent evidence for complex crystal histories [Davidson & Tepley (1997)].

AT - decrease in temperature below the liquidus

Fig. 9 Results of a numerical power-output model for basaltic systems showing the time taken for crystallisation as a function of temperature decrease below the solidus [after Hawkesworth et al. (2000, 2001)].

AT - decrease in temperature below the liquidus

Fig. 9 Results of a numerical power-output model for basaltic systems showing the time taken for crystallisation as a function of temperature decrease below the solidus [after Hawkesworth et al. (2000, 2001)].

These differentiation time scales imply that there is no direct link between magma residence time and eruptive periodicity, since most island volcano eruptions re-occur on the order of 10's to 100's years. Rather, eruptive periodicity may be linked to degassing (Jaupart 1996) and, for example, Tait et al. [1989] developed a model which predicted eruptive periodicity on the scale of years to 10-100's years due to crystallisation induced increases in volatile over-pressure. Recently, Gauthier & Condomines [1999] have exploited the fact that 210Pb has a gaseous parent, 222 Rn (see Fig. 2), to constrain the time scales of magma degassing and recharge at Stromboli and Merapi volcanoes. Applying their approach to a global survey of 210Pb systematics in arc lavas, Turner et al. [2004] estimated that most island arc magmas undergo degassing for several decades prior to eruption. Moreover, in at least one example, from Sangeang Api volcano in the Sunda arc, there is evidence that this degassing occurred much more recently than the bulk compositional differentiation of the magmas. Therefore, the crystals formed by degassing did not separate from their liquid and cause bulk compositional changes. Finally, Berlo et al. [2004] found that the 210Pb systematics of lavas from Mount St. Helens varied with eruption style and through time following the cataclysmic May 1980 eruption. Importantly, these studies have both recognised that 210Pb-exceses may indicate the presence of fresh degassing magma at depth. Since the build up of gas pressure and the injection of hot mafic inputs into existing magma

LITHOSPHERE

Fig. 10 Schematic cross section of the plumbing and reservoir system for an island arc volcano modified from Gill [1981] to include element transfer (+U and +Ra are intended to schematically illustrate the spatial and temporal separation of addition of U and Ra due to dehydration reactions in the subducting plate), magma transport and residence time scales discussed in text.

Fig. 10 Schematic cross section of the plumbing and reservoir system for an island arc volcano modified from Gill [1981] to include element transfer (+U and +Ra are intended to schematically illustrate the spatial and temporal separation of addition of U and Ra due to dehydration reactions in the subducting plate), magma transport and residence time scales discussed in text.

chambers [e.g. Sparks et al. (1977)] are likely triggers for eruption, such data may have important future application in eruptive hazard prediction.

6. Conclusions

The use of U-series disequilibria in unravelling the physical processes of fluid transfer, partial melting, melt migration and modification at convergent margins is still a new and rapidly expanding field of research. The presently available constraints on the time scales of magma formation, storage, ascent and degassing at island arcs are illustrated on Fig. 10. Advances in analytical techniques are allowing for more rapid and precise analysis and new data sets, particularly on fully characterised and well dated lavas, can only improve our understanding of convergent margin processes. However, these data will need to be combined with numerical models if their full significance is to be realised.

Was this article helpful?

0 0

Post a comment