THE DEPTHS OF PARTIAL CRYSTALLIZATION IN MAFIC SYSTEMS: IMPORTANCE OF FILTERING INPUT DATA AND RESULTS FOR MAJOR ELEMENT GEOBAROMETRY

Liquid (liq), olivine (ol), clinopyroxene (cpx), plagioclase (plag) equilibrium geobarometry is one of the petrological methods for determining pressures of partial crystallization, and hence the depths of storage of mafic magmas. This approach relies on the pressure-dependent location of the liq-ol-cpx-plag (LOCP) cotectic in multidimensional dimensional space. LOCP equilibrium geobarometers require that melts be multiply saturated with respect to ol, plag, and cpx, and therefore the interpretation of pressure results calculated for off-cotectic samples is complex and often unclear. Various LOCP equilibrium geobarometers have been applied to mafic magmas, including those by Yang et al. (1996), Herzberg (2004), Villiger et al. (2007), Kelley and Barton (2008), and Hartley et al. (2018). The latter two methods have addressed the need to exclude pressure results for samples that do not lie on the LOCP cotectic, but disagreement between results obtained with these methods highlights the need to develop a comprehensive and systematic approach for the removal of samples that are outside of the assumptions made for cotectic crystallization. We describe a procedure for filtration of input data before pressure calculation using the Kelley and Barton (2008) method, and a procedure for the correction of pressure results. In the filtration process, samples that are not consistent with LOCP cotectic crystallization, as defined by tight arrays on variation diagrams, are removed. Using this filtration method, we reevaluate the depths of partial crystallization for the Juan de Fuca Ridge, eruptive centers in Iceland, the Reykjanes Ridge, and Hawaii. For these locations, we find that filtration generally decreases the average calculated pressure and reduces the range of pressures of partial crystallization for each locality. Filtered results show better agreement with the results of seismic studies than do unfiltered results. This filtration procedure significantly changes the interpretation of the results for the depths of LOCP cotectic crystallization at these localities, and allows for a more accurate estimate of depths of partial crystallization in mafic systems. Such filtering is therefore necessary to reconstruct realistic models for magma plumbing systems.