The caldera-forming eruption of Volcan Ceboruco, Mexico

Volcan Ceboruco (Mexico) erupted similar to 1000 years ago, forming a caldera similar to 4 km in diameter. The pyroclastic deposit (Jala pumice) produced was studied at 75 sites, where thicknesses of all tephra layers were measured and, at many sites, the five coarsest lithics in each fall layer were measured. Several fall layers were also sieved for component analysis. White (rhyodacite) and gray (dacite) pumices occur in every tephra layer. The eruption began by dispersing a thin, narrowly distributed fall layer (P0) to the north. Next, a widely dispersed, coarse, Plinian fall deposit (P1), the most voluminous layer in the Jala pumice, erupted. The rest of the eruption alternated between pyroclastic surges and flows (including the Marquesado pyroclastic flow, southwest of the volcano) and Plinian falls (P2-P6). In all, 3-4 km(3) of magma erupted, similar to 95% of which was deposited as fall layers. During most of the deposition of P1, eruptive intensity (mass flux) was almost constant at 4-8x10(7) kg s(-1), producing a Plinian column 25-30 km in height. Size grading at the top of P1 indicates, however, that mass flux waned dramatically, and possibly that there was a brief pause in the eruption. During the post-pi phase of the eruption, a much smaller volume of magma erupted, although mass flux varied by more than an order of magnitude. We suggest that caldera collapse began at the end of the P1 phase of the eruption, because along with the large differences in mass flux behavior between P1 and post-P1 layers, there were also dramatic changes in lithic content: (P1 contains similar to 8% lithics; post-P1 layers contain 30-60%) and magma composition (P1 is 98% rhyodacite; post-P1 layers are 60-90% rhyodacite). However, the total volume of magma erupted during the Jala pumice event is close to that estimated for the caldera. These observations appear to conflict with models which envision that, after an eruption is initiated by overpressure in the magma chamber, caldera collapse begins when the reservoir becomes underpressurized as a result of the removal of magma. The conflict arises because firstly, the P1 layer makes up too large a proportion (similar to 75%) of the total volume erupted to correspond to an overpressurized phase, and secondly, the caldera volume exceeds the post-P1 volume of magma by at least a factor of three. The mismatches between model and observations could be reconciled if collapse began near the beginning of the eruption, but no record of such early collapse is evident in the tephra sequence. The apparent inability to place the Jala pumice eruptive sequence into existing models of caldera collapse, which were constructed to explain the formation of calderas much greater in volume than that at Ceboruco, may indicate that differences in caldera mechanics exist that depend on size or that a more general model for caldera formation is needed.