BULLETIN OF VOLCANOLOGY, cilt.79, sa.2, 2017 (SCI-Expanded)
Agri Dagi (Ararat), whilst being the tallest volcano in Turkey, is largely understudied. Two predominant peaks, Greater and Lesser Agri, make up the main edifice, which has been built during four main phases. The most recent phase consisted of two volcanic eruptions. The respective surface area and volume of the first volcanic eruption were estimated at 96 km(2) and 3.2 km(3), whereas those of the second eruption were much smaller with the surface area and volume estimated at 25 km(2) and 0.6 km(3). It is unusual for stratovolcanoes to produce basaltic eruptions of more than 3 km(3), although these and larger volumes are not uncommon in flood basalt-type eruptions. Large basaltic eruptions from stratovolcanoes normally require volcano-tectonic forcing (e.g. subsidence of collapse caldera and graben). However, there is no evidence for such volcano-tectonic forcing, during the most recent eruptions at Agri Dagi (Ararat), and therefore, their comparatively large volume basaltic lavas need to be explained in a different way. Here, we present an analytical method for calculating the source volume needed to supply magma to the eruptions at Agri Dagi. We found that the lava flow of 3.2 km(3) was likely fed by a very large magma reservoir (similar to 13,000 km(3)), while the second flow of 0.6 km(3) was fed by a reservoir of a much smaller effective size (or similar to 2000 km(3)). 'Effective size' depends on what fraction of the reservoir participates in the eruption. We propose that the entire reservoir supplied magma to the larger eruption, but only one of its compartments (about one fifth of the total volume of the reservoir) supplied magma to the smaller eruption. Although seismic tomography indicates a magma reservoir at great depths (>20-30 km) below the Agri Dagi volcano, geochemical constraints on some of the later-formed rocks suggest an interaction between a shallow chamber (at 8-10-km depth) and the deep reservoir approximately 0.5 Ma. We provide numerical models whose results indicate that dykes injected from the lateral margins of the deep-seated reservoir are more likely to reach the surface directly rather than replenish the shallow magma chamber, suggesting also that the compartment for the second eruption was at the margin of the reservoir.