2.1 Numerical constraints on phase separation timescales
Numerous compaction models have been developed for application to geologic problems and include descriptions of phase separation driven by gravity (Bercovici et al. , 2001, Huber & Parmigiani, 2018, McKenzie, 1984, McKenzie, 1985, McKenzie, 2011, Ricard et al. , 2001) or by external stresses (Jackson et al. , 2018, Koenders & Petford, 2000, Petford et al. , 2020, Solano et al. , 2012). One of the ultimate challenges for these models is to calculate melt extraction timescales sufficiently low to coincide with petrologic observations. For instance, velocity and melt fraction averaged calculations performed by Bachmann and Bergantz (2004) as well as calculations from McKenzie (1985) and more recently Jackson et al. (2018) have been used to investigate the timescales required to separate large amounts of melt by compaction and found that in some cases such timescales are comparable to the longevity of the intrusive bodies hosting them (Bachmann and Bergantz, 2004; Lee and Morton, 2015). In all cases, these models use rheologic descriptions of the crystal matrix informed by viscous creep.
Other recent works have made strides towards developing compaction models equipped to capture the change in rheologic properties across a range of melt fractions. Wong and Keller (2022), for instance, parameterize the effective matrix viscosity – which relates to the ability of the matrix to resist reduction in melt fraction – such that at low melt fractions an endmember controlled by the creep of individual crystals is approached, while at large melt fractions phase separation is primarily governed by crystal settling. However, constraints on the effective matrix viscosity at melt fractions between these endmembers are necessary, as silicic, crustal magma systems are believed to spend the majority of their lifetime at intermediate melt fractions (Huberet al. , 2009, Szymanowski et al. , 2017). The timescales of melt segregation calculated by numerical compaction models are extremely sensitive to the effective matrix viscosity law used and its evolution as a function of melt fraction.