After a long period of subaerial fissure-fed extrusions of basaltic magmas (∼ 12 to > 3 Ma) volcanic activity was then concentrated in the central part of Tenerife. Phonolitic magma chambers formed and a central volcanic complex was constructed (the Las Canadas edifice). The formation of a large depression (the Las Canadas caldera) truncated the top of the edifice. The active twin strato-cones Teide—Pico Viejo are sited in this depression. The history of the Las Canadas caldera and edifice are established from stratigraphy, geochronology (K—Ar dates) and volcanological studies. Two different groups are recognized, separated by a major unconformity. The Lower Group is dated at 2 to 3 Ma and includes the products of several volcanic centres, which together represent several cycles. The Upper Group ranges from 1.56 to 0.17 Ma and includes three different formations representing three long-term (∼ 100 to 300 Ka) volcanic cycles. The periods of dormancy between each formation were of ∼ 120 to 250 Ka duration. The Las Canadas caldera is a multicyclic caldera which formed over the period 1.18–0.17 Ma. Each cycle of activity represented by a formation culminated in caldera collapse which affected different sectors of the Las Canadas edifice. Geological observation and geochronology support an origin by collapse into a magma chamber. The minimum volume of pyroclastic ejecta is substantially greater than the present caldera depression volume (45 km³), but approaches the inferred volume of the original caldera depression (> 140 km³). After the formation of the caldera, sector collapses could also occur at the northern flank of the volcano causing the disappearance of the northern side of the caldera wall.

We present experiments to investigate the formation of calderas using inflatable balloons in a medium of fused alumina powder. This system provides a model of a magma chamber in an homogeneous elastic media which scales to volcanic systems in terms of geometry and strength of the media. Three kinds of experiment were performed: (i) the balloon was inflated and then deflated simulating pre-emption tumescence and caldera collapse: (ii) the balloon was deflated without prior inflation, simulating caldera formation without pre-eruption tumescence: (iii) three balloons were inflated and deflated in a line, simulating multiple magma chambers. The influence of magma chamber shape was investigated using different shaped balloons. Balloon inflation formed a dome with a surface pattern of shallow polygonal fractures centrally, and radial fractures peripherally. Uplift occurs along deep inward-dipping concentric reverse faults. Deflation generated depressions with an outer set of concentric ring fractures surrounding a central funnel formed by sagging along numerous minor faults, some of which formed during doming. During collapse the concentric reverse faults were converted to normal faults and the radial normally-faulted fractures were converted to reverse faults. Without prior tumescence, an outer set of near vertical ring faults surround a flat central depression due to collapse of a coherent block. An inner set of outward-dipping reverse ring faults also develop. The experimental caldera area increased with balloon size and increased as the balloon depth decreased. For a given balloon volume and depth, collapse area is significantly smaller if doming had occurred prior to collapse. Arcuate fractures can provide the pathways for magma ascent during pre-emption tumescence (Smith & Bailey 1968), but the shallow radial and polygonal fracture systems may also influence vent location. Changes in fault geometry and sense of movement (extension to compression and vice versa) occur when subsidence initiates, and can account for changes in vent location that occur in many eruptions. Multiple overlapping collapse calderas are associated with magma chamber migration, whereas nested calderas can result from the activity of a single magma chamber.