Phreatic and hydrothermal eruptions often occur with little or no warning representing a significant hazard within geothermal regions. These violent eruptions occur at a range of temperatures and pressures within varying rock types. A range of mechanisms including heating or decompression, allows hydrothermal/supercritical fluid to rapidly flash to steam, expanding and shattering the surrounding rock to produce an eruption, with no direct magmatic influence.
These eruptions are highly variable resulting in the current wide ranging classification schemes, many of which are based on characteristics that are hard to observe and define. This has resulted in confusing nomenclature with many different terms used to describe the same eruptive phenomena. Here a new classification scheme is presented, based on the easily definable features of eruption size, trigger type (natural or anthropogenic) and geological setting (volcanic or hydrothermal). This ultimately produces a classification dividing the eruptions into either phreatic, where magma interacts with cold water but no juvenile material is erupted; or hydrothermal where eruption occurs from an already heated hydrothermal system. Examples are then provided for each classification type. Previous studies have focused exclusively on either physical characteristics of eruptions, small scale experimental modelling of trigger processes or mathematical modelling of various eruption characteristics. Here, a new experimental procedure has been developed to model phreatic fragmentation, based on shock tube experiments for magmatic fragmentation by Alidibirov and Dingwell (1996). Water saturated samples are fragmented from a combination of argon gas overpressure and steam flashing within vesicles. In this thesis, these experimental results have been integrated with the physical characteristics of porosity, permeability and mineralogy to create two new models of phreatic fragmentation. Firstly a generalised model to explain fragmentation processes and secondly a specific model describing the eruption forming Lake Okaro, within the Taupo Volcanic Zone of New Zealand. These models were developed with the overall aim to improve understanding of these eruption types, ultimately improving future hazard modelling.

Experiments were performed on Rangitaiki ignimbrite, through which the Okaro eruption occurred. In order to evaluate alteration effects, both unaltered ignimbrite and hydrothermally altered ignimbrite samples were analysed. Experiments were performed at room temperature and 300°C with pressures from 4 to 15 MPa, to reflect likely geothermal conditions while also assessing the effect of liquid water on fragmentation. Results indicate that within these samples 5 to 8 MPa of decompression is required to trigger an eruption, fitting well with the previously identified trend between decompression and porosity for magmatic samples. The fragmentation front propagates through the sample at speeds ranging between 14 m/s to 42 m/s, increasing with higher applied pressures and higher sample porosity. Most importantly, grain size analysis from these experiments show a clear shift to smaller grain sizes when liquid water flashes to steam (independent of pressure or sample type), reflecting the greater energy involved with steam flashing. Previous grain size analysis of the Okaro breccia deposits have indicated that the highest weight percentage of fragments fall between -3.5 and 1.5 phi, with our experimentally produced fragments fitting right within this range at -0.5 to 1.0 phi. Here the first parameterisation of conditions for phreatic and hydrothermal eruptions is presented creating a general fragmentation model along with a case study on Lake Okaro. These models describe how eruptions occur, with stages from initial system priming and overpressure development through to the last stages of eruption and crater formation.