Introduction (Habil)
This is the introduction from my habilitation thesis
Introduction
The transport of melt, from the lower mantle to the upper mantle or from lower/mid-crustal source rocks to (often) upper crustal emplacement levels, is one of the most important mass and heat transfer processes in the earth. Melt is constantly generated in the crust and mantle and may move to different locations (McKenzie, 1984). The whole process of melt generation and migration involves several different processes and takes place at different scales (Ribe, 1987; Bons et al., 2004). Initially, newly generated melt collects in melt pockets (e.g. Rushmer, 1995; Sawyer, 2000; Klepeis et al., 2003; Guernina & Sawyer, 2003) and may or may not form an interconnected network (Bulau et al., 1979; McKenzie, 1985; Wickham, 1987; Petford, 1995; Petford & Koenders, 1998). This initial melt, at one point, leaves the source region and can, if external conditions permit, move for hundreds of meters or kilometers (Weertman, 1971; Nicolas & Jackson, 1982; Miller et al., 1988; Sleep, 1988; Brown, 1994; Rubin, 1998; Rushmer, 1995; Conolly et al., 1997; Weinberg, 1999). Prominent examples for this long range transport of melt are the numerous plutons and batholiths that can be found on each continent and from each geological era (e.g. Becker et al., 2000; Brown, 2001 and references therein).
One of the main problems to understand the active processes during melt segregation and migration are the very different scales involved (Holtzman et al., 2003a; Spiegelman et al., 2001). It is therefore convenient to separate the process of melt generation, early distribution of melt within the source and later migration of melt out of the source into distinct processes (Ribe, 1987). Another reason to separate these processes lies in their dependence on physical parameters. Partial melting is mainly governed by the constraints of conservation of energy and, in natural systems, multicomponent phase equilibrium (Ribe, 1987) while melt migration is mainly governed by the principles of mass conservation (McKenzie, 1984; Riley & Kohlstedt, 1991).
Partial melting occurs when the temperature, pressure and chemical composition of the rock is favorable for it (Harris et al., 2000). As temperatures increase, minerals that constitute the reaction assemblage to form melts become unstable and a melt forms. This preferably occurs at grain triple junctions and at grain boundaries where the reacting minerals are in direct contact (e.g. Sawyer, 1998; Harris et al., 2000). At the onset of partial melting, the rheological parameters of the partially molten rock change dramatically with. The biggest, and most obvious, change is probably that of the bulk viscosity. It changes from that of a solid rock towards that of a melt, e.g. the bulk viscosity decreases dramatically (Barboza & Bergantz, 1998, Walte et al., 2005). The presence of melt also leads to an increase in heat-transfer rates (e.g. Raia & Spera, 1993; Barboza & Bergantz, 1998) and may cause an increase of tectonic strain rates (Dell'Angelo et al., 1987; Hollister & Crawford, 1987).
Below a certain threshold (the critical melt fraction, CMF), melt is distributed in isolated pockets. Currently, this CMF is regarded to be well below 10% (Rushmer, 1995, Vigneresse et al., 1996; Faul, 2001). Above this threshold, the solid matrix slowly disaggregates and melt pockets may become interconnected to form a network (Arzi, 1978, van der Molen & Paterson, 1979). However, melt can only leave the source if there is space where it can flow to (Rabinowitz et al., 2002; Spiegelman & McKenzie, 1987). At the same time, if melt migration is faster than melt generation, the host rock has to deform (compact) to accommodate for the loss of volume (e.g. Hollister & Crawford, 1986; Brown, 1994; Rushmer, 1995, 2001).
Once melt can migrate through the host rock, different ascent mechanisms may be active (e.g. Asimow & Stolper, 1999; Vanderhaeghe, 1999):
Porous flow describes the flow of melt through an aggregate of solid particles. If the melt fraction increases, the solid particles of the host rock may completely loose contact. At this stage, granular flow (suspension flow) occurs (Frank, 1968, Stolper et al., 1981).
Dyking describes the process where a finite volume of melt rises through the earth's crust in fractures and dykes (Petford, 1995) driven mainly by buoyancy (e.g. Petford et al., 1994). Dyking is a relatively fast process and can transport large volumes of melt through rocks with temperatures below the solidus of the melt (e.g., Weertman, 1971; Spence & Turcotte, 1985; Lister & Kerr, 1991; Clemens and Mawer, 1992; Petford et al., 1994; Meriaux & Jaupart, 1998) .
Diapirism describes both, ascent and emplacement of a large portion of melt (e.g. Weinberg & Podladchikov, 1994). Per definition, a diapir is a roughly tear drop-shaped body which has to rise at least one body diameter before its final emplacement (Paterson & Vernon, 1995).
None of these processes is exclusive and a combination of the processes might be applicable to a wide range of tectonic environments (Vanderhaeghe, 1999; Weinberg, 1999).
One of the main problems in studying these processes is that they generally can not directly be observed in nature. Experiments and analogue models can help to better understand these processes. However, it is usually complicated to (up)scale these models to nature (Holtzman et al., 2003b). Another problem with experiments are the often underdefined boundary conditions due to the limited size (spatial as well as temporal) of the experiments. Computer simulations can help to overcome these problems. In computer simulations, the scaling and timing can freely be chosen and boundary conditions can be set as needed. However, the main problem with computer simulations is that they need to be based on a valid physico-chemical models for the process they are supposed to simulate.
In the following, I will briefly introduce some general concepts of numerical simulations. This is followed by a more in depth view of the processes that are active during melt segregation, migration and long range transport of melt.

