Fluid-flow and fluid-rock interaction models of thermal aureoles around shallow intrusives can be grouped according to the approach: a) inverse models based on observed geochemical alteration (e.g. Gerdes *et al*., 1995) and mineral reaction progress patterns (e.g. Ferry, 1995) ; and b) heuristic forward modeling studies (e.g. Hanson, 1995). More often than not, these two approaches yield results for fluid source, amounts, and fluid flow direction that contradict each other. Similarly, inverse models based on stable isotopes often contradict results obtained from those based on mineral reaction progress. While stable isotope signatures in the metamorphic host rocks typically require significant fluid influx from the intrusive - e.g. fluid fluxes away from the intrusion, down a temperature gradient-, reaction progress of metamorphic reactions documents significant up-temperature fluid flow towards the intrusion. Detailed field and modeling studies (Gerdes *et al*., 1995) on the influence of spatially heterogeneous permeability have led us to believe that some of the contradictions are due to spatially and temporally evolving heterogeneous pore space, its connectivity, and the resulting heterogeneous rock permeability.

Using geologically reasonable distributions of permeability values based on a review of sedimentary basin literature, 2-D stochastic permeability fields were created by the turning bands algorithm (Zimmerman and Wilson, 1990). Finite element calculations describing heat and fluid-mass conservation to simulate the cooling of an intrusion emplaced in the upper crust were performed using the permeability maps. Extensive focusing of fluid flow into high permeability zones is predicted by these models. Locally, small, long lived high fluid flux convection cells develop in some simulations without disrupting the overall convection pattern. These local convection cells are of the order of several hundred meters only and are centered on high permeability areas. The fluid involved is mainly of local origin, and no major stable isotope alteration is predicted for these high flux areas. On the other hand, significant prograde and retrograde reactions might be driven by these cells due to temperature and pressure gradients along the flow path. This is an example in which isotopic and reaction progress patterns appear to record different fluid flow histories.

The correlation range for the statistical description of permeability in these stochastic models is on the order of 1500m. The detailed field and isotope data described in a companion presentation (Roselle *et al*., 1996) at Ubehebe Peak, Death Valley (USA) documents that correlation lengths of alteration phenomena are on the order of centimeters in that contact aureole. Such a small correlation length requires a discretisation scale of the grid for the finite element models in the millimeter range. This results in a prohibitive number of elements (over 1012). Current simulations do not take into account any coupling between fluid driven reactions and porosity/permeability evolution of the rocks. Such coupling of permeability to reaction progress of metamorphic reactions would introduce temporal variable permeability fields, further complicating the prediction of hydrothermal fluid flow evolution in contact aureoles.

Ferry, J. M., *Amer. Mineral.* **80**, 1226-1228 (1995).

Gerdes, M.L., Baumgartner, L.P., Person, M. & Rumble, D., *Amer. Mineral. ***80**, 1004-1019 (1995).

Gerdes, M.L., Baumgartner, L.P. & Person, M., *Geology* **23**, 945-948 (1995).

Hanson, B.R., *Amer. Mineral.* **80**, 1222-1225 (1995).

Roselle, G.T., Baumgartner, L.P. & Valley, J.W., (this volume) (1996).

Zimmerman, D.A. & Wilson, J.L., *TUBA*, 104 pp. (GRAM Inc., New Mexico 1990).