ALBERT

Notes: Abstract The 14.1 Ma composite welded ignimbrite P1 (45 km3 DRE) on Gran Canaria is compositionally zoned from a felsic lower part to a basaltic top. It is composed of four component magmas mixed in vertically varying proportions: (1) Na-rhyolite (10 km3) zoned from crystal-poor to highly phyric; (2) a continuously zoned, evolved trachyte to sodic trachyandesite magma group (6 km3); (3) a minor fraction of Na-poor trachyandesite (〈1 km3); and (4) nearly aphyric basalt (26 km3) zoned from 4.3 to 5.2 wt% MgO. We distinguish three sites and phases of mixing: (a) Mutual mineral inclusions show that mixing between trachytic and rhyolitic magmas occurred during early stages of their intratelluric crystallization, providing evidence for long-term residence in a common reservoir prior to eruption. This first phase of mixing was retarded by increasing viscosity of the rhyolite magma upon massive anorthoclase precipitation and accumulation. (b) All component magmas probably erupted through a ring-fissure from a common upper-crustal reservoir into which the basalt intruded during eruption. The second phase of mixing occurred during simultaneous withdrawal of magmas from the chamber and ascent through the conduit. The overall withdrawal and mixing pattern evolved in response to pre-eruptive chamber zonation and density and viscosity relationships among the magmas. Minor sectorial variations around the caldera reflect both varying configurations at the conduit entrance and unsteady discharge. (c) During each eruptive pulse, fragmentation and particulate transport in the vent and as pyroclastic flows caused additional mixing by reducing the length scale of heterogeneities. Based on considerations of magma density changes during crystallization, magma temperature constraints, and the pattern of withdrawal during eruption, we propose that eruption tapped the P1 magma chamber during a transient state of concentric zonation, which had resulted from destruction of a formerly layered zonation in order to maintain gravitational equilibrium. Our model of magma chamber zonation at the time of eruption envisages a basal high-density Na-poor trachyandesite layer that was overlain by a central mass of highly phyric rhyolite magma mantled by a sheath of vertically zoned trachyte-trachyandesite magma along the chamber walls. A conventional model of vertically stacked horizontal layers cannot account for the deduced density relationships nor for the withdrawal pattern.