Understanding caldera structure and development: An overview of analogue models compared to natural calderas Valerio Acocella ⁎ Dipartimento Scienze Geologiche Roma Tre, Roma, Italy Received 1 August 2006; accepted 15 August 2007 Available online 28 August 2007 Abstract Understanding the structure and development of calderas is crucial for predicting their behaviour during periods of unrest and to plan geothermal and ore exploitation. Geological data, including that from analysis of deeply eroded examples, allow the overall surface setting of calderas to be defined, whereas deep drillings and geophysical investigations provide insights on their subsurface structure. Collation of this information from calderas worldwide has resulted in the recent literature in five main caldera types (downsag, piston, funnel, piecemeal, trapdoor), being viewed as end-members. Despite its importance, such a classification does not adequately examine: (a) the structure of calderas (particularly the nature of the caldera's bounding faults); and (b) how this is achieved (including the genetic relationships among the five caldera types). Various sets of analogue models, specifically devoted to study caldera architecture and development, have been recently performed, under different conditions (apparatus, materials, scaling parameters, stress conditions). The first part of this study reviews these experiments, which induce collapse as a result of underpressure or overpressure within the chamber analogue. The experiments simulating overpressure display consistent results, but the experimental depressions require an exceptional amount of doming, seldom observed in nature, to form; therefore, these experiments are not appropriate to understand the structure and formation of most natural calderas. The experiments simulating underpressure reveal a consistent scenario for caldera structure and development, regardless of their different boundary conditions. These show that complete collapse proceeds through four main stages, proportional to the amount of subsidence, progressively characterized by: (1) downsag; (2) reverse ring fault; (3) peripheral downsag; (4) peripheral normal ring fault. The second part of this study verifies the possibility that these latter calderas constitute a suitable analogue to nature and consists of a comprehensive comparison of the underpressure experiments to natural calderas. This shows that all the experimental structures, as well as their progressive development, are commonly observed at natural calderas, highlighting a consistency between models and nature. As the shallow structure of experimental calderas corresponds to a precise architecture at depth, it provides a unique key to infer the deeper structure of natural calderas: recognizing diagnostic surface features within a caldera will thus allow it to be categorized within a precise structural and evolutionary context. The general relationship between the evolutionary stage of a caldera and its d/s (diameter/subsidence) ratio allows such a quantification, with stage 1 calderas characterized by d/s N 40, stage 2 by 18 b d/s b 40, stage 3 by 14 b d/s b 18 and stage 4 by d/s b 14. The consistency between experiments and nature suggests that, in principle, the d/s ratio may permit to evaluate the overall structure and evolutionary stage of a caldera even when its surface structure is poorly known. The volume of erupted magma associated with caldera collapse is poorly dependent on the d/s ratio or evolutionary stage; however, the location of sin- and post-collapse volcanism may depend not only upon the amount of collapse, but also on the roof aspect ratio. As the regional tectonic control is concerned, the experiments Available online at www.sciencedirect.com Earth-Science Reviews 85 (2007) 125 – 160 www.elsevier.com/locate/earscirev ⁎ Fax: +39 6 54888201. E-mail address: acocella@uniroma3.it. 0012-8252/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.earscirev.2007.08.004