In order to understand the mechanism of melt generation and the origin
of high heat flow in subduction zones; a series of numerical models o
f the thermal and flow structures in the mantle have been tested in tw
o-dimensional boxes with an inclined subducting slab of constant veloc
ity. In contrast to previous models, (1) large convecting cells are us
ed without imposing a high-temperature profile on the backarc boundary
, which enables us to discuss the global heat balance and to seek for
the heat sources responsible for the melting and high heat flow observ
ed in subduction zones: and (2) various mechanical conditions (e.g., c
oupling between the slab and overlying mantle wedge: buoyancy associat
ed with melting) and heat sources (e.g., heat flux from below, interna
l heat generation, viscous heating) are tested in varying proportions.
In all the calculations, steady state or near-steady state with a sma
ll instability periodically occurring at the upper thermal boundary la
yer is achieved. Under these conditions, the global heat balance can b
e described by a simple boundary layer argument. The results show that
, in order to attain a high enough temperature for melting and a high
heat flux, a large amount of internal heating (i.e., more than 2.5 x 1
0(-7) W/m(3)) is required if the convection is limited within the uppe
r mantle. The high internal heating required may be explained if the r
adioactive nuclides in fluids expelled from the subducting slab are ad
ded to the wedge and circulate in the convection cell for a sufficient
time. Another possible explanation for melting and the high heat flux
is that the hot material is supplied from the lower mantle. Based on
the thermal and flow structures obtained, melting regimes in subductio
n zones are discussed, in which the following key processes take place
: (1) melting associated with pressure change of a rock packet with it
s movement, especially compression melting in the downward flow of the
mantle wedge along the slab, and (2) melting due to compositional cha
nge of the rock packet associated with migration of H2O and melt. To s
olve these problems in detail, further studies on distribution and mig
ration of the fluids will be required.