U.S. Dept. of Commerce / NOAA / OAR / PMEL / Publications
The studies described above document that about 10% of the
In the heady early days of hydrothermal discovery, many workers believed that the chemistry of hydrothermal solutions was simply a function of exit temperature, and that this function could be established empirically by careful sampling at a few sites [e.g., Edmond et al., 1982]. Further discoveries demonstrated a much more complex universe of chemical variations [e.g., Von Damm, 1990], emphasizing the need for continued exploration and the compilation of as large a hydrothermal data base as possible. (We leave unwritten the many biological, physical, and geological reasons for vent field exploration.)
The search for new vent fields has historically followed an inefficient path: A tectonic segment, or segment portion, is chosen on geological considerations. A remote or manned search for geologic or biologic visual clues of active venting is then mounted to pinpoint discharge locations. Finally, vent fluids are sampled from a submarine and plume waters from a surface ship. Consideration of the several studies reviewed in this paper demonstrates that a more efficient approach would postpone detailed seafloor exploration until after systematic plume mapping had revealed the relative magnitude and approximate location of discharge sources. Moreover, chemical and physical observations of plume waters can provide first-order information on the composition of the source fluids, which are likely indicative of the present magmatic and hydrothermal conditions of the vent field.
The efficacy of plume reconnaissance can be documented from several recent
examples. On the EPR near 10�N, an expensive and exhaustive
While the above example is an after-the-fact plume success story, instances of plume mapping directing the discovery of seafloor discharge sources are common. On the CoAxial segment of the JDFR, plume surveying immediately after the remote detection of seismic activity in 1993 directed seafloor imagery mapping by a remotely operated vehicle towards the eventual discovery and sampling of a new lava eruption [Embley et al., 1995]. On the MAR, systematic plume surveying between 27� and 30�N in 1993 discovered optical and chemical evidence for three possible hydrothermal sites, including a strong signal at 29�10.1'N, 43�10.3'W [Murton et al., 1994]. The Broken Spur vent field was confirmed at this site three months later during an Alvin dive series [Murton et al., 1994].
Fig. 21. Comparison of the global extent of plume studies and seafloor vent fluid sampling. Shaded areas on spreading centers mark the locations of multi-segment plume mapping investigations on the Reykjanes Ridge, Mid-Atlantic Ridge, northeast Pacific ridges, Gulf of California, northern and southern East Pacific Rise, and the north Fiji Basin. Numbers refer to Table 1 and mark sites of vent fluid collection and analysis along spreading centers; fluid discharge sites noted only from visual observations are not included.
The number of fluid discharge sites sampled by submersible or remote vehicle is steadily growing but remains only a tiny fraction of all global sources (Table 1, Figure 21). Plume mapping and sampling has allowed us to expand our knowledge of the distribution and geologic setting of hydrothermal venting beyond the limits imposed by camera and submersible operations. Figure 21 graphically compares the relative coverage of vent and plume data sets. The expansion of hydrothermal information afforded by plume studies is essential for regional-scale issues such as the effect of hydrothermal cooling on heat transport within the crust [Phipps Morgan and Chen, 1993], the movement of melt within the crust [Purdy et al., 1992], magmatic differentiation [Sinton et al., 1991], and the effect of hydrothermal heating on abyssal waters [Hautala and Riser, 1993].
TABLE 1. Sites of Hydrothermal Vent Fluid Collection
As an example of the kind of linkage between geological and hydrothermal data
made possible by plume studies, we examine a simple but crucial hypothesis:
A functional relationship exists between axial hydrothermal heat flux and ridge-crest
spreading rate, a convenient analogue of the time-averaged magmatic budget.
Incorporation of hydrothermal cooling into models of the genesis of oceanic
crust [e.g.,
Phipps Morgan and Chen, 1993] requires generalizations of this kind.
To use plume distributions to test this hypothesis, we assume that hydrothermal
heat output is proportional to plume incidence, the percentage of ridge-axis
length overlain by a significant hydrothermal plume. Simply put, more hydrothermal
cooling produces more plumes. (We operationally define a "significant"
plume as one with plume tracer values greater than three standard deviations
above the background (non-plume) mean.) Our assumption of the equivalence of
heat flux and plume incidence is reasonable over the narrow crest and shallow
axial depressions characteristic of ridges spreading faster than
Fig. 22. (a) Full-rate seafloor spreading vs. axial magma chamber depth
from seismic observations [Purdy
et al., 1992] (solid circles; LAU, Valu Fa Ridge in the eastern Lau
Basin; JDFR, Cobb segment on the Juan de Fuca Ridge; NEPR, northern East Pacific
Rise, 9�-13�N; SEPR, southern EPR, 14�-20�S) and from the Phipps
Morgan and Chen [1993] crustal thermal model (solid line). (b) Full-rate
seafloor spreading vs. plume incidence (solid squares) and hydrothermal cooling
in the neovolcanic zone according to a crustal thermal model (Y. J. Chen and
J. Phipps Morgan, The effect of magma emplacement geometry, spreading rate,
and crustal thickness on hydrothermal heat flux at mid-ocean ridges, submitted
to Geophys. Res. Lett., 1994) (dashed line). Plume incidence data available
for the Reykjanes Ridge (RR) [German
et al., 1995], Mid-Atlantic Ridge (MAR) (11�-40�N) [German
et al., 1994], JDFR [Baker
and Hammond, 1992], NEPR (8�40-13�10'N) [Bougault
et al., 1990; Baker
et al., 1994], and SEPR (13�50'-18�40'S) [Baker
and Urabe, 1994]. "Plume incidence" is the percentage of axis
length overlain by a significant (plume tracer values greater than background
mean plus three standard deviations) plume. The heavy solid line is a least-squares
fit forced through the origin:
Continuous surveys from the JDFR (Figure 7) [Baker and Hammond, 1992], the northern EPR (Figures 10, 11) [Bougault et al., 1990; Baker et al., 1994], and the southern EPR (Figure 12) [Baker and Urabe, 1994] have yielded direct measures of plume incidence. On slow-spreading ridges, in the current absence of conclusive continuous surveys, we approximate plume incidence from estimates of vent field frequency derived from discontinuous plume surveys along the MAR and Reykjanes Ridge [German et al., 1994; German et al., 1995]. If we assign a reasonable plume size of 10 km to each of the 15 suspected sites of venting on the MAR between 11� and 40�N [German et al., 1995], we calculate a 6% plume incidence. Similarly, the one vent field found on the Reykjanes Ridge [German et al., 1994] yields a 2% plume incidence. When data from all five regions are plotted together, a strong linear relationship between plume incidence and spreading rate emerges (Figure 22).
It is vital to realize that Figure 22 is valid
only over long time or space scales. Vigorous hydrothermal activity can occur
on segments of any spreading rate, so an instantaneous observation at a particular
location may not be representative of its long-term hydrothermal history. We
obtain a representative hydrothermal history of a particular ridge-crest location
only by observing it through geologic time, which is impractical, or by simultaneously
observing many individual segments spreading at the same rate. Figure
22 thus implies that, over long time scales, (1) one-third of the global
ridge axis spreading at
How does the plume vs. spreading rate relationship compare to other observed
and modeled parameters? Axial magma chamber (AMC) depth from seismic observations
[Purdy
et al., 1992] shows a rapid decrease from 60 to
Still at issue is whether our meager coverage of the global ridge crest is
representative and the trends described above are general. Over
Return to previous section or go to next section