A. HECTOR VALDES CARRASQUILLA Alcalde Municipal

SiO2 (wt. %)

Fig. 3.16. (a-h) Selected radiogenic isotopes plotted against SiO2 (wt.%). Data fields as for figure 3.9. Note elevations in 87Sr86Sr for Sumbing pyroclastic deposits (Merapi and Merbabu) relative to the lavas (and Ungaran).

 The Pyroclastic Deposits contain major and trace element and isotopic ratios very similar to Merapi. This includes notably higher Sr (ppm), 87Sr/86Sr and 206Pb/204Pb and lower HFSE and 143Nd/144Nd than the lavas. 87Sr/86Sr ratios become progressively higher in samples with lower SiO2 (wt.%)

 The lavas are more heterogeneous in their chemistry with variable MgO/FeO*, Ni, Cr and La/Yb, lower 87Sr/86Sr and 206Pb/204Pb, higher 143Nd/144Nd, and similar

176_Hf/177_{Hf compared to the pyroclastic rocks. Elevated HFSE contents in the lavas}

resemble those at Ungaran. There is little variation between isotope ratios and differentiation, other than lava Sumb78, which contain ratios similar to the pyroclastic deposits.

The aim of the next section is to examine the processes which are capable of causing the magmatic variation between the groups described above.

3.5. Discussion

3.5.1. Discriminating between Pyroclastic Deposits and Lavas through upper crustal processes

The absence of primitive rock compositions at Sumbing suggest that the magmas were modified from an original mantle source as they rose through the crust. It is therefore important to establish the nature and extent of the processes that occur via differentiation in crustal reservoirs to see if they are capable of creating the decoupling of elements between magmatic groups. Island arc magmas are rarely emplaced at the surface without having experienced some degree of fractional crystallisation, mixing between magmas of different compositions or growth conditions, and/or assimilation of the arc crust (Davidson, 1987; Gasparon et al., 1994; Handley, 2006; Deegan et al., 2010).

Evidence for upper crustal processes typically includes textural signs of disequilibrium in minerals, incorporation of crustal and cumulate xenoliths, and systematic element variations against indices of differentiation with low concentrations of compatible elements (e.g. Davidson et al., 2005). It is also suggested that dominance of intermediate compositions in volcanic arcs may reflect mixed hybrids, where reservoirs with slowly evolving melts are

replenished by an influx of primitive mafic magma prior to eruption (Sudradjat, 1991; Reubi et al., 2002). Petrographic observations suggest that mixing between compositionally distinct magmas are common beneath Sumbing, and minerals which have experienced different growth conditions have also been mixed together. The question is; to what extent does this modify the chemistry of the lava?, does it affect all of the rocks?, and are any other open system processes in operation?

Simple binary mixing between two magmas would be expected to produce near linear relationships on element-element plots, or hyperbolic arrays on ratio-element plots (Langmuir et al., 1978; Vogel, 1982; Flood et al., 1989). This type of geochemical trend is not obvious in the element-element plots shown in figures 3.9 to 3.13. However, the addition of plutonic xenoliths, and mingling between magmas and distinct crystal residues, would create disturbances in magmatic trends by the addition (or removal) of mineral phases (Luhr & Haldar, 2006). This provides a more realistic process for some groups at Sumbing.

The pyroclastic samples at Sumbing show a highly porphyritic appearance with abundant clinopyroxene and plagioclase phenocrysts. Plagioclase core-to-rim compositions (An36-90)

are particularly variable in these rocks compared to the lavas. A wide range in anorthite content is not an unusual feature in Sunda arc volcanoes (e.g. Handley, 2006; Chadwick et al., 2007). However, the wide range is limited to the pyroclastic rocks.

The lava deposits show a restricted range of anorthite contents which do not exceed An59. It is

evident that mineral chemistry and petrography in addition to chemistry can distinguish two groups of rocks at Sumbing which have potentially had different pre-eruptive histories. It is also evident that both groups have experienced differentiation prior to emplacement. To address these issues in greater detail, the two groups are evaluated on an individual basis below.

3.5.1.1. Differentiation of magmas 1: Fractional Crystallisation

The most notable variation in chemistry between the pyroclastic rocks and lavas are the concentrations in Sr and HFSE. Strontium is characteristically compatible in plagioclase, where it readily substitutes for calcium. Therefore, one way of producing the distinction between groups could be to either accumulate plagioclase in the pyroclastic rocks, or remove plagioclase from the lavas.

Higher concentrations of CaO, Al2O3, Ba and Eu for a given SiO2 and MgO contents in the

pyroclastic samples add support this idea. To quantify this hypothesis, modelling has been applied to both groups using the XLFRAC Least Squares modelling technique (after Stormer & Nicholls, 1978).

Least Squares Modelling

The model requires an initial (parent) and final (daughter) magma (whole-rock oxide values), together with a number of phases representative of mineral compositions. It works by calculating the probability of adding or subtracting the phases in order to derive the daughter from the parent, in which a good solution minimises the sum of all residuals (∑r2_{). For this}

model, 8 oxides are used and the five phases present in most rocks; plagioclase, amphibole, clinopyroxene, orthopyroxene and titanomagnetite. The objectives are firstly, to determine whether pyroclastic deposits and lavas can be derived from the addition and/or removal of plagioclase; and secondly, the plausibility of fractional crystallisation as a driving force between group compositions.

Results of the modelling are shown in table 3A (1-14). For each sample, SiO2 and Mg#

values are given, together with the mineral phases required to be removed or added during differentiation. Models 1-7 use lava Sumb81 as a parent magma composition because it is the least evolved rock on the basis of SiO2 and Mg#. Because there are no obvious liquid lines of

decent (based upon geochemistry), daughter derivatives from a number of the lavas have been calculated (see figure 3.17a).

Models 1-7 produce acceptable results with ∑r2 values of between 0.06 and 0.26 in accordance with a crystallisation assemblage of plagioclase + amphibole + opx + cpx + Fe-Ti oxide. Results show that between 21.80% and 55.68% crystallisation of the minerals are required to produce the daughter compositions from Sumb81. Models 1 to 3 use CBL‟s as daughter compositions with progressive elevations in SiO2 and decreases in Mg#. This series

of rocks do not form a liquid line of decent because of inconsistencies in the removal of amphibole and clinopyroxene. Similarly, the same is true for models between Sumb81 and the lavas (4-7). For this case the only inconsistency is in model 6, where Sumb78 requires less clinopyroxene removal than adjacent lavas.