[1] We present a model of melt segregation in a mush submitted to both compaction and shear. It applies to a granitic melt imbedded within a partially molten continental crust, able to sustain large stress values. The mathematical derivation starts with the equations for melt and plastic flow in a mush. They are manipulated to obtain equations for the mean flow field and for the separation velocity. Assuming that the mean flow field is simple shear, a specific set of equations for the melt flow in a shear field is obtained. After simplifying the equations, they finally reduce to two systems of coupled equations. One is the wellknown equation for compaction. The other is new and describes melt channelling during shear in a mush with a constant viscosity plastic matrix. Three free parameters are observed. One is the usual compaction length, and the other two are functions of the stress and strain amplitude during shear. Compaction instabilities lead to the development of spherical pockets rich in melt while shear channelling instability segregates melt in parallel veins. The size of the pockets and the distance between veins remain close to the compaction length. Actually, the viscosity ratio between the matrix and its melt controls the compaction length L, which is found metric or submetric. The two types of instability segregate melt. However, the compaction process is generally so sluggish that it cannot compete with the channelling one. The channelling time is controlled by the amount of intergranular melt present in the system and of the amplitude of the shear stresses. During each channelling cycle, lasting for about 30 to 300 kyr, the intergranular melt is completely squeezed out from the volume in between veins. As melting progresses, the successive batches of melt, as well as the residual solid matrix, are increasingly more dehydrated. As a result, both phases progressively stiffen without changing their viscosity contrast and the associated compaction length. The segregation process stops when the dehydration process clamps the deformation of the solid matrix.

No abstract

A mantle diapir of 8‐km radius has been recognized by systematic structural mapping of the Oman ophiolite in the Maqsad district. This diapir chilled while still active under the ridge crest. Streamlines rotated in the diverging part of the diapir a few hundred meters under the Moho. This implies a decrease by several orders of magnitude in the effective viscosity of the peridotites to a depth of 1 km beneath the Moho. We present dynamic models for diapiric flow across such an interface. A step of 0.5 MPa in dynamical pressure is found along the high/low viscosity interface; the highest pressure occurs within the uppermost low‐viscosity layer. This implies that magma percolating through the porous peridotite network is retained below this interface. This promotes a rapid increase of the melt/rock ratio, hence also a dramatic decrease in the effective viscosity of the peridotites. The feedback between magma percolation and decreasing viscosity produces the steep rotation of the mantle flow under the Moho.

Abstract. The heat flow map derived from 550 measurements collected in a the southern portion of the sedimented rift in Middle Valley, northern Juan de Fuca Ridge, displays kilometersized quasi-circular regions of very high heat flow. Some of these domains, explored during Ocean Drilling Program (ODP) leg 139, are thought to be discharge zones of large-scale hydrothermal plumes. To understand this unique data set, we modeled the kilometer-scale hydrothermal circulation within both the sedimentary and the igneous crust, using a set of two-and threedimensional models that use a numerical technique based on horizontal spectral decomposition of the flow equations. These models include variations in the viscosity and density of the hydrothermal fluids with temperature. We examine the variations in flow patterns due to different permeability-versus-depth distribution within sediment and pillow layers. Models with the same permeability in both layers do not match the seafloor heat flow field in Middle Valley. When the permeability decreases from the bottom to the top of the simulation domain by a factor greater than 20, convection assumes a plume pattern to produce surface heat flow comparable to that observed in Middle Valley. Within the models the ratio of the heat flux above the recharge and discharge domains is directly related to the vertical harmonic mean of the permeability field. A value of 7 X 10 -16 m 2 provides a good match to the heat flow observations. The Darcy velocities of the hydrothermal fluids in the discharge areas approach 16 crn/yr while in the recharge areas they are lower than 3 crn/yr. These rates and the temperature inside the plumes are sufficiently high to produce the observed massive sulfide deposits and mineral alterations in 1-2 X 105 years.The dynamic pressure produced by the hydrothermal flow matches the pressure measured in drill sites. This process may play a major role in compaction, fracturing, and uplift of the sediment cover. For example, the dynamic pressure in the ascending plume equals the lithostatic pressure at a depth of 50 m. Resulting hydrofracturing could explain the genesis of the vent fields associated with the hydrothermal discharge

International audienceWe report on the detection of air convection with infrared thermal images for two quasi-circular craters, 20 m and 40 m wide, forming the volcanically inactive cone of Formica Leo (Reunion Island). The thermal images have been acquired from an infrared camera at regular time intervals during a complete diurnal cycle. During the night and at dawn, we observe that the rims are warmer than the centers of the craters. The conductivity contrast of the highly porous soils filling the craters and their30° slopes are unable to explain the systematic temperature drop from rim to centers. We suggest that this signal could be attributed to air convection with gas entering the highly permeable soil at the center of each crater, then flowing upslope along the bottom of the soil layer, before exiting it along the crater rims. To quantify this process, we present a two-dimensional numerical modelling of air convection in a sloped volcanic soil with a surface temperature evolving between day and night Thisconvection depends on a unique dimensionless equivalent Rayleigh number Racq which is the product of the standard Rayleigh number with the volumetric heat capacity ratio of the air and the soil. The convective flow is unsteady: during some periods, the convective flow is entirely confined within the soil, and at other times air enters the crater at its center and exits it at the rim crests. When Racq =6000, a value likely compatible with the soil permeability and the geothermal heat flux. avery strong transient cold air plume occasionally develops along the center of the crater. The interval of time between two plumes only depends on the thermal fluctuations within the top boundary layer of the convective cell, and thus is not contrasted by the diurnal cycle. The detachment of a cold plume can occur at any time, after few days of quiescence, and lasts several hours. During the whole convective cycle, the rim to center temperature drop persists and has an amplitude and a shape having an excellent agreement with that found in the IR-images. This work constitutes a preliminary step to explore the deep thermal structure of the active caldera of Bory-Dolomieu and could help to improve the understanding of volcanic hazards of the Reunion volcano