Hydrothermal circulation of fluids within the oceanic crust is not restricted to regions in the immediate vicinity of spreading centers, but continues off-axis for possibly tens of millions of years. Although relatively little is known about this process, it is now recognized that fluid circulation and the associated alteration reactions and mineral precipitation play a profound role in the evolution of the physical properties of the oceanic crust. For example, based on a review of drilling and downhole logging results, lab measurements of the physical properties of oceanic rocks, and data from ophiolites, Carlson and Herrick [1990] concluded that the seismic velocity structure of the oceanic crust is strongly influenced by changes in porosity and the degree of alteration. Most studies have demonstrated a general decrease in porosity with age [ Wilkens et al., 1991; Jacobson, 1992] as would be expected from mineral precipitation within veins, vesicles and cracks. However, a recent compilation of available physical properties data from all DSDP/ODP sites showed no simple trend, although problems with the sparse distribution of drill sites in terms of crustal age, as well as possible bias in recovery rates in different lithologies, may well have affected the results [ Johnson and Semyan, 1994].
Permeability is the most critical variable controlling patterns of fluid flow in hydrothermal systems both in space and time. Early observations of the wavelengths of elongate, off-axis heat flow anomalies combined with numerical models of fluid flow through homogeneous porous media and assumptions of high crustal permeability down to depths of >1 km, implied that hydrothermal circulation systems reached depths of several kilometers (references to all these studies are not included here for brevity, but a summary can be found in Fisher et al. [1990]). However, measurements of the permeability of the ocean crust made using drill-string packers indicate that the permeability structure of the oceanic crust is strongly layered and only the top few hundred meters are highly permeable [e.g. Anderson et al., 1985; Becker, 1989, 1991; Larson et al., 1993]. Little is known about the lateral heterogeneity of the crustal permeability structure or the spatial and temporal persistence of layers with distinct hydrogeological properties. Precipitation of secondary mineral phases and thermoelastic stresses cause narrowing and closure of cracks; analytical models that attempt to quantify these processes are just being developed [e.g. Germanovich and Lowell, 1992; Lowell et al., 1993; Lowell and Germanovich, 1994].
Despite the difficulties associated with using sea floor measurements to infer the existence and character of subsurface hydrothermal circulation cells, considerable progress has been made in the last few years by combining detailed geophysical/geochemical data from field surveys and drilling with increasingly sophisticated numerical modeling techniques. At the Juan de Fuca Ridge, results from drilling an axial hydrothermal system can be combined with off-axis geophysical and geochemical studies to further our understanding of how hydrothermal systems change as they are transported away from the ridge crest. The circulation of the axial hydrothermal system at Middle Valley---a sediment-covered spreading rift on the northern Juan de Fuca Ridge---is currently being modeled using downhole physical properties data collected during ODP Leg 139. Permeability measurements made proximal to an active vent site indicate the presence of several discrete zones of exceptionally high hydraulic conductivity which, if typical of oceanic basement at spreading centers, must dominate circulation patterns in axial hydrothermal systems [ Becker et al., 1994].
About 50 km to the south on the eastern flank of the Juan de Fuca Ridge,
a comprehensive program of detailed heat flow studies, seismic surveys,
and sediment coring of sea floor ranging in age from 0.6-6.0 Ma has
been conducted as part of the FlankFlux program to investigate
off-axis hydrothermal circulation [ Davis et al., 1992; Wheat
and Mottl, 1994]. Pleistocene turbidites bury the ridge flank beyond
a crustal age of about 600,000 years (
18 km from the ridge axis) and form
a sharp boundary between sediment-free and sediment-covered oceanic
crust. Further to the east, the buried basement varies from being
very smooth in some areas to extremely rough in areas where
occasional volcanic edifices penetrate through the sediment.
These structures are believed to provide permeable pathways for discharge
of fluids to the sea floor [ Davis et al., 1989]---a
conclusion consistent with other studies of both relatively young (<1 Ma
in age) crust near the ridge axis and older crust on the ridge flanks
that suggest the geometry of fluid circulation is forced by
thermal perturbations caused by basement topography [ Fisher et
al., 1990; Johnson et al., 1993]. In areas of extensive
basement outcrops and discontinuous sediment, measured heat flow is
only about 15% of that predicted for a cooling lithospheric plate
[ Davis et al., 1992]. Vigorous flow of seawater cools the basement
to temperatures of 10-20 C, and the high water/rock ratios and
short residence time of the fluid in the crust results in little change
in its chemistry. As the sediment thickness increases to tens of
meters, direct access of bottom seawater is limited and the vertical flux
of fluids through the sea floor decreases to a few mm/year. This results
in a hydrothermal system with a lower water/rock ratio, longer
residence times, higher temperatures and greater chemical exchange.
In addition, heat and fluid can be advectively transported in the
basement over large distances (up to 20 km) in these areas. Finally,
the sediment cover becomes thick enough (about 160 m in the case of
the turbidites) to effectively prevent exchange of seawater between
the basement and the ocean, although active circulation within the
crust maintains uniform temperatures at the basement-sediment
interface [ Davis et al., 1992; Wheat and Mottl, 1994] .
Subtle heat flow variations in an area where the sediment thickness
and basement topography are relatively uniform suggest a horizontal
cell dimension of about 700 m. If an aspect ratio of one is assumed
[ Lister, 1990; Rosenberg et al., 1993], then the depth
of penetration of the circulation system corresponds to a strong
upper crustal seismic reflector in the region [ Davis et al., 1992].
The most extensive dataset relating to off-axis hydrothermal circulation
is from the flanks of the Costa Rica Rift where a variety of
geophysical, geochemical and geothermal data have been collected, and
a number of holes drilled, including DSDP/ODP Hole 504B located on 
6 Ma
old crust (comparable to the oldest area studied on the Juan de Fuca
Ridge). Heat flow varies with topography and has a wavelength of 4-7 km,
and the agreement between the observed mean and the predicted value
suggest that conduction dominates heat transfer and convection is weak
[ Langseth et al., 1988]. Further modeling has suggested that heat
flow is more strongly correlated with basement relief and
differential sediment thickness, both of which are reflected in
the bathymetry [ Fisher et al., 1990, 1994]. Vertical advective
flow through sediments of a few mm/year continues to sediment thicknesses
of 310 m, and indicates that the physical properties of the
sediments influence the duration of the vertical fluid flux [
Mottl, 1989]. Porosity and permeability measurements in Hole 504B
indicate that the upper 200 m of crust, which contains many open
fractures, has permeabilities of 10
- 10
m
,
which decrease with depth as the fractures become filled with
secondary minerals [ Pezard, 1990]. In the low porosity sheeted
dikes, permeabilities decrease by several orders of magnitude
to 10
-10
m
[ Becker et al., 1989].
The permeability structure and refined modeling of heat and fluid flow
in this area suggest that circulation within the basement is restricted
to an upper, permeable layer of volcanics extending down to depths of
only 200-300 m, and that much of the circulation is concentrated
along narrow high permeability zones [ Becker et al., 1989;
Fisher et al., 1990, 1994]. Moreover, numerical simulations of
convection in porous media using the same data suggest that the
convection occurs in numerous cells with aspect ratios of less than
one within the upper permeable layer [ Rosenberg et al., 1993].
These studies demonstrate that hydrothermal circulation can persist for long periods of time as the crust is transported away from the ridge axis. In addition, they suggest that active circulation in the crust can continue once advective exchange of fluids between the crust and the ocean is prohibited by a thick sediment cover. Understanding the influence of the permeability structure of the upper crust on the geometries of the circulation cells will require a better knowledge from field measurements of the three-dimensional permeability field and its temporal evolution, and application of increasingly sophisticated models that numerically simulate changes in the physical properties of the crust in off-axis hydrothermal systems.