The generation of magmas in Earth's mantle and their segregation deep inside the Earth result in lavas that erupt from volcanoes on Earth's surface. How these processes take place and how they impart geochemical signatures in the resulting lavas is fundamental to our knowledge of how Earth behaves as a system. However, the processes that take place and complex interplay of geochemical and geophysicsl parameters that allow them to happen are presently inadequately understood. This research uses state-of-the-art numerical methods and computer algorithms and combines the expertise of both geosciences and applied mathematics to tackle the issues surrounding the physical and chemical proceses involved in partial melting of the mantle, melt migration, and melt-rock reaction in the generation of magma. From geochemical evidence gleaned from basaltic lavas erupted at mid-ocean ridge spreading centers, it is clear that the primary mechanism of melt transport requires the focusing of melt through channels in the mantle. However, the formation, distribution,and longieivy of these high-porosity channels is still a matter of speculation. To shed light on this problem, high order, adaptve, and parallel numerical models for simulating reactive magma migration in 2D and 3D settings directly relevant to melt migration below mid ocean ridges is being developed. The resulting simulation software will incorporate algorithms, approaches, and parameters that allow reactive dissolution of minerals, shear deformation, and decomperssional melting to be taken into account in models that calculate the distribution and segregation of melt. Additional considerations that link the geochemcial consequences and geophysical implications of melt migration will also be developed. The results will provide critical imput to general time-dependent melting models that allow geochemists to relate chemical heterogenaities observed in erupted basaltic lavas to their mantle sources. Broader impacts of the work include the funding of researchers at an institution in an EPSCoR state (Rhode island), building fundamental infrastructure for science via developing software that will be part of the Computational Infrastructure for Geodynamics software library which is publicly accessible, and training of undergraduates and graduate students in computational methods.
