Melting poses a challenge in geodynamic numerical modelling: thermodynamic models are
computationally expensive and they present serious restrictions as far as P-T conditions are
concerned; on the other hand, simple parameterizations usually cannot address major
element contents of melts, and thus physical properties. Here, we present a new
polynomic parameterization based on pMELTS [Ghiorso et. al., 2002] to be used
in geodynamic models. In addition, we show a first application to a geodynamic
model.
Our parameterization is adapted for continuous melt fractionation under decompression. The
input parameters are initial pressure of melting, pressure, critical porosity, water content and
temperature. The parameterization can be further calibrated for different rock compositions.
It yields as the amount of melt retained in the rock, total degree of melting plus major
element compositions in the form of wt% of oxides, such as: SiO2, MgO, FeO, CaO, Al2O3
and Na2O.
The parameterization has the same limitations as the thermodynamic model on which it is
based (MELTS), and somewhat bigger errors due to statistical fitting. In turn, it involves
advantages in terms of computational speed, and ease of implementation. Most importantly,
extrapolation of the model along this parameterization can provide statistically meaningful
results. To demonstrate this, we benchmark these results with high pressure melting
experiments.
Finally, we show first applications of our parameterization as it is coupled to simple
thermomechanical plume models. In these models, different melt compositions are obtained
when changing potential temperature, plume buoyancy flux, and plume temperature.
Although the parameterization errors are probably too high for petrological ends (where
MELTS and pMELTS should be used instead), it presents an efficient and suitable option for
geodynamic models. |