Mantle Fluids in the San Andreas Fault System, California

See allHide authors and affiliations

Science  14 Nov 1997:
Vol. 278, Issue 5341, pp. 1278-1281
DOI: 10.1126/science.278.5341.1278


Fluids associated with the San Andreas and companion faults in central and south-central California have high3He/4He ratios. The lack of correlation between helium isotopes and fluid chemistry or local geology requires that fluids enter the fault system from the mantle. Mantle fluids passing through the ductile lower crust must enter the brittle fault zone at or near lithostatic pressures; estimates of fluid flux based on helium isotopes suggest that they may thus contribute directly to fault-weakening high-fluid pressures at seismogenic depths.

The San Andreas fault (SAF) is known to be a weak fault, and explanations for its weakness include either low-friction fault-zone materials or superhydrostatic fluid pressures within the fault zone (1-3). Abnormally high fluid pressures have been measured in pores at shallow crustal depths within the SAF system (4). For example, fluid pressure in the Varian–Phillips well, 1.4 km from the main trace of the SAF, is ∼12 MPa over hydrostatic values at a depth of 1.5 km (5).

Models of fault weakening by elevated fluid pressures call on different fluid origins. Crustal fluids, connate or meteoric, may be drawn into the fault zone in response to fault rupture and become trapped by mineral reactions; the high fluid pressures required to weaken the fault are reestablished by compaction of the sealed fault-zone materials (6-8). In this model, the base of the seismogenic zone, defined by the brittle-ductile transition, is treated as an impermeable boundary. In an alternative model, fault-weakening fluid pressures are generated by a high flux of deep crustal or mantle fluids that are continually supplied to the seismogenic zone from the ductile lower crust at superhydrostatic pressures (9).

To investigate fluid source and influence on SAF dynamics, we conducted a chemical and helium isotopic study of selected springs, seeps, and wells associated with the SAF and companion faults. Elevated discharge temperatures or salinity, unusual solute chemistry, high gas or water flow rates, or other indications of deep circulation or extensive water-rock interaction were used to select sampling sites. We found that the fluids can be divided into four categories based on geochemistry (Fig. 1) and into meteoric or evolved connate based on water isotopes (Fig.2).

Figure 1

Ternary plot of major anion concentrations in deep fault-zone fluids from the San Andreas system. Covariations delineate four chemically distinct fluids: chloride-rich (yellow), sulfate-rich (red), bicarbonate-rich (blue), and bicarbonate- and chloride-rich (green). The range in total dissolved solids is 260 mg/liter for some of the bicarbonate fluids to 46,800 mg/liter for the chloride-rich brines. Symbol shapes refer to sample groups defined by location: San Andreas Group (SAG, squares), western Transverse Ranges (SYG, triangles), and the southern Coast Ranges (CRG, diamonds).

Figure 2

Isotopic composition of fault-zone waters (symbols as in Fig. 1). Dashed line labeled MWL (meteoric water line) depicts expected variations in the isotopic composition of meteoric waters. Box defines the range in isotopic compositions of oil field waters from the San Joaquin Valley, California (31).

Fluids enriched in crustal helium are characterized by a3He/4He ratio of ∼0.02 Ra, whereas mantle helium has a value of ∼8 Ra (Ra is the3He/4He ratio in air, 1.4 × 10−6) (10). The helium isotopic compositions in our samples varied from 0.12 to 4.0 Ra (11) and indicate that fluids associated with the SAF system contain mantle He contributions of ∼1 to ∼50% (Fig. 3). Mantle helium occurs throughout the entire investigated segment of the SAF (Fig.4) but its occurrence is not closely restricted to the main trace of the SAF. The highest ratio (4 Ra, ∼50% mantle) was found in Mercey Hot Springs, which emerge from the southern extension of the Tesla–Ortigalita fault that defines the eastern limit of the central Coast Range. Elevated3He/4He ratios were also found west of the SAF, although there is a modest trend of decreasing3He/4He ratios with distance away from the SAF along the Santa Ynez fault in the western Transverse Range (Fig. 3, triangles). These springs are bicarbonate- or sulfate-rich, and the westward decrease in 3He/4He parallels the trend of decreasing CO2/CH4. However, when all samples are considered, there is not a strong correlation between helium isotopes and fluid chemistry, which implies that the influx fluids affect 3He/4He ratios but not other aspects of the fluid chemistry that are controlled by the adjacent crustal provenance lithologies.

Figure 3

Helium isotopic composition plotted as a function of approximate distance from the main strike of the SAF (symbols as in Fig. 1: MHS, Mercey Hot Spring; MM, Middle Mountain Oil Well; VP, Varian–Philips Well; SC, Stone Canyon Well). Associated flow rates and mantle contributions are calculated as described in text.

Figure 4

California map showing faults associated with the San Andreas system and other regional fault systems (SAF, San Andreas fault; SYF, Santa Ynez fault; TOF, Tesla–Ortigalita fault; HF, Hayward fault) and the distribution of Tertiary volcanics. Symbol shapes are the same as in Fig. 1 but are shaded in accordance with the helium isotopic composition of the sampled site numbered as follows: 1, Epsom Salt Spring; 2, Byron Springs Well; 3, Alum Rock; 4, Gilroy Hot Spring; 5, Stone Canyon Observation Well; 6, Mercey Hot Springs; 7, Coalinga Mineral Spring; 8, Middle Mountain Oil Well; 9, Varian–Phillips Well; 10, Jack Ranch Highway-46 Well; 11, Paso Robles Spa; 12, Gaviota Warm Spring; 13, Wheeler Hot Spring; 14, Willits Warm Spring; 15, Sespe Hot Spring; 16, Warm Springs; 17, Arrowhead Hot Well.

Crustal fluids with high 3He/4He ratios have been recognized in regions with active magmatic systems (12) or that are undergoing extension (13). However, aside from some wells with elevated 3He/4He ratios in New Zealand (14), little data exist for seismically active areas, like the SAF system, that are undergoing compression. A critical question is whether the high 3He/4He fluids represent an active mantle-fed flux or input of residual He from earlier magmatism. Magmatism and development of major volcanic centers followed the northward development of the SAF from 30 to 8 million years ago (15) (Fig. 4). However, because magmas are extensively degassed during emplacement and eruption, resulting in low residual He and high U/He ratios, it is unlikely that the volcanic rocks are a significant source of 3He or high3He/4He ratios for circulating fluids. And although magma degassing during emplacement will stain the near-field hydrologic system (16), because of the high U and Th concentrations in local rocks (1 to 5 parts per million) (17), it is also unlikely that stained fluids could retain high 3He/4He ratios (∼1 to 2 Ra) after 8 to 30 million years of radiogenic 4He ingrowth (18). Furthermore, the fluid helium isotopic compositions are not correlated with fluid chemistry or related to the age of the local volcanic areas. Thus, the high 3He/4He ratios represent influx of mantle fluids. The hydrodynamics thus require that (i) the brittle-ductile transition is a permeable boundary and (ii) mantle fluids crossing this boundary and entering the seismogenic zone must be at or near lithostatic pressure.

As mantle fluids flow up through the fault zone, the fluid3He/4He ratios become diluted with radiogenic4He that is produced locally in the crust. This generates a vertical gradient in the helium isotopic composition in the fault zone that depends on the vertical rate of fluid flow and the radiogenic4He production rate in the crust. The solution for the integrated upward fluid flow rate (q) for steady-state one-dimensional flow scaled to crustal thickness (19) isEmbedded Image Embedded Image(1)where H crust is the combined thickness of the brittle and ductile crust, ρs and ρf are the densities of the solid and fluid phases,P(He) is the present day production rate from the U+Th in the fault zone materials, and (R/Ra)mantle,crust,meas are the helium isotopic compositions in the mantle (8 Ra), produced in the crust (0.02 Ra), and measured in the fault zone fluid. [4He]f,mantle is the initial concentration in the mantle fluid and is calculated from the measured 4He concentration and helium isotopic composition of the sampled fluid. Using the composition of the Varian–Phillips well, for which we have the most reliable He concentration data (1.9 × 10−9 mol of 4He per gram of fluid and 2 Ra), as representative of fault-zone fluids and a crustal thickness (H crust) of 30 km, calculated upward fluid flow rates through the fault zone vary from ∼1 to 10 mm/year (Fig. 3). The flow rates could be significantly higher, because neither mixing with radiogenic 4He-rich fluids from the neighboring brittle crust nor hydrodynamic dispersion is included in the calculation. The calculated flow rates also assume continuous flow and therefore are time averages over the flow path defined byH crust; actual flow rates could be significantly higher if flow is episodic.

To assess the impact of flow rate on possible pressurization of the fault zone, total volatile fluxes must be calculated. A flow of ∼3 mm/year corresponds to a 3He flux through the SAF of ∼2 × 10−15 mol cm 2year 1. If the entire network of faults that make up the San Andreas system is considered (20), the total3He flux may be as large as ∼6 mol/year, comparable to that in several arc volcanoes (21). One can assume that the mantle 3He is associated with other more abundant mantle volatiles—certainly CO2, and perhaps water. For instance, the upper mantle, as defined by midocean ridge basalts, has a CO2/3He ratio of ∼1010(22). The San Andreas system thus may have a CO2flux of ∼1011 mol/year.

At seismogenic depths, the density and viscosity of CO2 are comparable to those of water, and a significant CO2 flux into the fault zone could help generate and maintain the high pore pressures necessary for fault weakening (9, 23). The pore pressure induced by the CO2 flux depends on permeability. The permeability of the lower crust is unknown. One extreme is to assume a constant permeability. Then, for a reasonable porosity (0.01) and rock compressibility (10 4MPa 1), the 3He calculated CO2 flux implies a permeability of ∼10−21m2 to maintain high pore pressures through the ductile region and into the seismogenic zone (23). Although this is at the lower end of permeability measured in various rock types, it is probably a lower limit because of the likely addition of nonmantle CO2 to the system from metamorphic decarbonation reactions in the lower crust. CO2/3He ratios in most of our samples exceed, at times by >1000, the mantle value of 1010, and δ13C values range from −3 to −15 per mil, indicative of additional nonmantle CO2 sources. Alternatively, if we assume that permeability decreases exponentially with depth at a rate scaled to a linear increase in stress, then at the base of the seismogenic zone superhydrostatic pressures imply fault zone permeability of ∼10 18 m2(9). In this context, fault weakening by mantle fluid inflow at the rate implied by the 3He flux appears plausible.

We do not know if our sampled features tapped fluids directly from a fault zone or from outside in the adjacent crust. Despite this uncertainty, the flux of mantle He through the system requires a brittle-ductile boundary that is permeable and that supports the fault-weakening model calling for high fluid pressures in the fault zone driven by a mantle source (9). Movement of mantle fluids up through the ductile root zone into the San Andreas system requires that fluid pressures in the lower crust are at lithostatic values and that fluids are directed by the local stress regime in a manner similar to rising magmas (24). If we have sampled fluids directly from fault zones, then the observation that mantle helium is mixed with fluids from the adjacent crust implies that elements of both models (6-9) are important. Because weakness of the fault system is generated by high fluid pressures in the fault zone relative to the surrounding crust, infiltration of crustal fluids into the fault zone requires episodic reversals of the fluid pressure gradient. Within the seismogenic zone of the fault, local increases in fault-zone porosity induced by fault failure may temporarily lower the fluid pressure, permitting infiltration of crustal fluids (25).

Numerous models for the lithospheric structure of the San Andreas system have been proposed. In some (26), the San Andreas and other major strike-slip faults extend to Moho depths. In others (1, 27), the San Andreas and other faults are limited to the brittle upper crust and sole out into a regional decollement, which may extend as far east as the Sierra Nevada. In the latter models, major faults could tap fluids from an extensive decollement, which in turn might tap mantle fluid sources in regions far removed from the surface expression of the faults. These models may provide a reasonable explanation for the enigmatic 3He excesses in Mercey Hot Springs and in San Joaquin gas fields (16, 28). A mantle fluid source region extending as far east as the Sierra Nevada is also consistent with recent models for the lack of subcrustal lithosphere beneath the eastern Sierra Nevada (29), and the hypothesized presence of mantle CO2 in soda springs emerging near the crest of the Sierra Nevada (30).


View Abstract

Navigate This Article