Introduction

This report describes a digital, three-dimensional (3-D) faulted stratigraphic model we have constructed to represent the geologic framework of the Edwards aquifer system in the area of San Antonio, northern Bexar County, Texas (Fig. 1). The model is based on mapped geologic relationships that reflect the complex structures of the Balcones fault zone, detailed lithologic descriptions and interpretations of about 40 principal wells (and qualified data from numerous other wells), and a conceptual model of the gross geometry of the Edwards Group units derived from prior interpretations of depositional environments and paleogeography.

Model area

The aquifer system is developed in carbonate rocks that have been strongly modified by karst dissolution, diagenetic alteration, and locally intense fracturing within the en echelon strands of the Balcones fault zone (Maclay and Small, 1986; Barker et al., 1994; Hovorka et al., 1996). Fault structure has been shown to affect the direction of ground-water flow in the system, and to strongly influence the distribution of hydraulic head in the system (Maclay and Small, 1983). Recharge to the system is strongly influenced by fault structure, particularly by stream capture where drainage crosses the Balcones fault zone, and the major spring discharge systems are all fault-controlled (Barker et al., 1994).

Our principal goal in developing this digital 3-D model was to accurately represent the faulted lithologic units that make up the hydrogeologic framework in this area. We selected the recharge zone area of northern Bexar County (Fig. 1) as the focus of this effort because of detailed subsurface information available for numerous wells drilled by USGS for the Edwards Aquifer Authority in 1998. We used sophisticated commercial geologic modeling software (Earth Vision; Dynamic Graphics Corp., Alameda CA) to construct and evaluate this model; we have found Earth Vision to be one of few systems capable of modeling geologic systems with complex fault structures. This model allows us to view and evaluate the overall shape and form of the aquifer in this region, and to assess the amount of dislocation of the aquifer across the many strands of the Balcones fault zone.

The digital model depicts the complicated intersections of numerous major and minor faults in the subsurface, as well as their individual and collective impacts on the continuity of the aquifer-forming units of the Edwards Group and the Georgetown Formation (Table 1). The model allows for detailed examination of the extent of fault dislocation from place to place, and thus the extent to which the effective cross-sectional area of the aquifer is reduced by faulting. This geologic framework model is useful for visualizing the geologic structures within the Balcones fault zone and the interactions of en-echelon fault strands and flexed connecting relay ramps. The model also aids in visualizing the lateral connections between hydrostratigraphic units of relatively high and low permeability (Table 1) across the fault strands.

Average annual spring discharge (Comal, San Marcos, Hueco, and other springs) from the Edwards aquifer was about 365,000 acre-ft from 1934-1998, with sizeable fluctuations related to annual variations in rainfall. Withdrawals through pumping have increased steadily from about 250,000 acre-ft during the 1960s to over 400,000 acre-ft in the 1990s in response to population growth, especially in the San Antonio metropolitan area (Slattery and Brown, 1999). Average annual recharge to the system (determined through stream gaging) has also varied considerably with annual rainfall fluctuations, but has been about 635,000 acre-ft over the last several decades.

The area depicted in this 3-D model (Fig. 2) encompasses 30 monitoring wells drilled by USGS for the Edwards Aquifer Authority (EAA) in 1998 to evaluate influx to the aquifer system and to record fluctuations in water levels. These wells are part of an extensive system of monitoring wells maintained by EAA throughout the Edwards aquifer to aid in managing this critical resource for the region.

The Edwards aquifer consists of the Lower Cretaceous Edwards Group and Georgetown Formation that are exposed along the Balcones fault escarpment, which marks the edge of the Edwards Plateau in south-central Texas (Fig. 2; Maclay and Small, 1986; Barker et al., 1994; Hovorka et al., 1996). These units are chiefly carbonate, marl, and evaporite beds deposited in shallow marine waters and in the tidal-intertidal zone (Table 1; Maclay and Small, 1986; Hovorka, 1996). The underlying Trinity Group strata (Glen Rose Limestone and older units) are similar, but are hydrologically less transmissive and form a regional confining unit beneath the Edwards aquifer.

The Edwards Group was partly exposed and eroded in late Albian time (late Early Cretaceous) due to sea-level drop and flexural uplift of the San Marcos Platform (Maclay and Small, 1986; Barker et al., 1994). As much as 100 ft (30 m) of section was locally removed from the Edwards, and dissolution, karst collapse, and diagenetic alteration was developed over a widespread area in the southeast part of the model area in this study. Limestone of the Georgetown Formation, which lies disconformably on the Edwards Group, records renewed marine transgression in early Cenomanian time. Post-Georgetown erosion occurred during subsequent renewed uplift of the platform. Significant sea-level rise a few mil-lion years later led to deposition of the Del Rio Clay across the entire platform area, and the Del Rio forms the base of the regional upper confining unit over the Edwards aquifer (Barker et al., 1994).

The carbonate and evaporitic strata of the Edwards Group and Georgetown Formation show widespread evidence of complex alteration, recrystallization, dissolution, and cementation. These processes, singly and in combination, profoundly affected the porosity and permeability structure of the original depositional units and collectively produced the lithologic framework of the present-day Edwards aquifer (Maclay and Small, 1986; Hovorka et al., 1996). The details of these processes are beyond the scope of this report, but they are well summarized and discussed in reports by Hovorka et al. (1996, 1998) and by Maclay (1995).

Two significant events are recognized as the principal contributors to formation of the karst aquifer system in the Edwards. The first event was uplift of the San Marcos Platform in latest Albian (Late Early Cretaceous) time, which led to local erosion, dissolution by meteoric water, and karst formation. The second event dates to the Miocene and younger uplift of the Edwards Plateau (Fig. 1) along the Balcones fault zone (Fig. 3; Barker et al., 1994; Collins, 2000). This Cenozoic uplift produced more than 1000 ft (300 m) of differential displacement and led to widespread stripping of the post-Early Cretaceous strata from the Edwards and Trinity Groups. In the process, the uplifted Edwards Group beds were exposed to meteoric-water circulation, which leached out significant volumes of evaporite minerals and dolomite (Maclay and Small, 1986).

Hydrostratigraphic Units

Subsurface studies of the Edwards Group and aquifer by Rose (1972), and amplified by Maclay and others in the late 1970s (see Maclay and Small, 1986), identified subunits of the Kainer and Person Formations that seemed to have hydrostratigraphic distinction and lateral continuity throughout Bexar County and surrounding areas (Fig. 2). From base to top of the Kainer Formation, these units comprise the Basal Nodular member, the Dolomitic member, the Kirschberg Evaporite member, and the Grain-stone member. Ascending units of the Person Formation comprise the Regional Dense member, the combined Leached and Collapsed members, and the combined Cyclic and Marine members.

The overlying, disconformable Georgetown Formation is included in the definition of the Edwards aquifer because it is hydrologically connected to the Edwards Group, although the Georgetown is not particularly transmissive by itself. These hydrostratigraphic units of the Kainer, Person, and Georgetown Formations are the units modeled in three-dimensions in this report. These same units were mapped across the land surface of Bexar County by Stein and Ozuna (1996).

The hydrostratigraphic sub-units of the Kainer and Person Formations are locally distinct and identifiable, but not consistently so. Contacts between the Cyclic and Marine members (combined) and the Leached and Collapsed members (combined) are difficult to identify reliably, both in surface and subsurface conditions (T. Small, oral commun., 2003). Similar uncertainties apply to the boundary between the Grainstone and the Kirschberg Evaporite members, as well as the Basal Nodular and Dolomitic members of the Kainer Formation. Much of the uncertainty in identifying boundaries is due to extensive and irregular post-depositional modification of the units.

Hovorka (1996) further argues that the hydrostratigraphic-unit boundaries are indistinct because they are not truly stratigraphic contacts and may reflect lateral facies changes as well. Hovorka (1996) has documented numerous high-frequency upward-shoaling cycles based on sedimentary fabric, fossils, etc. (about 10 or 11) within the Kainer and Person Formations. These sedimentological cycles produce repetitive and similar lithologic sequences throughout the section that make lithic correlation difficult, especially where stratigraphic context is limited.

All investigators seem to concur that the Basal Nodular member at the base of the Kainer Formation and the Regional Dense member at the base of the Person Formation are distinct, identifiable, and stratigraphically significant units (compare, for example, Maclay and Small, 1986, Fig. 8 and Hovorka et al., 1996, Fig. 23 interpretations of the Castle Hills well and surroundings). These units mark substantial increases in water depth related to sea-level rise at the time of deposition (Hovorka, 1996).

For the purpose of this modeling study, we elected to depict the 8 hydrostratigraphic units of the Edwards aquifer, as defined by Maclay and Small (1986). This was largely based on the practical consideration that numerous drillholes within the area had been logged and interpreted in this scheme (Small and Maclay, 1982), and that these units are recognized as useful by water management agencies in the area (J. Waugh, San Antonio Water System, oral commun., 2002). Our subsurface depiction of the aquifer structure is also consistent with the geologic mapping on the outcrop in this area (Stein and Ozuna, 1996). These hydrostratigraphic units and their general characteristics are summarized in Table 1.The base of the Edwards aquifer is formed by the top of the Glen Rose Formation, which consists of several hundred feet of thin, alternating beds of dense limestone, dolomitic limestone, marl, and sparse evaporite deposits. The Glen Rose has little vertical permeability, limited lateral permeability along evaporite beds, and sparse fractures (Maclay and Small, 1986). It forms the regional lower confining unit beneath the Edwards aquifer.

The Basal Nodular member of the Kainer Formation consists of 50-60 feet of dense nodular, shaly limestone, mudstone, and grainstone. The unit has limited porosity and permeability, relatively few fractures, and generally behaves as a confining bed in the subsurface. Caves and conduits have been noted within the recharge zone where dissolution has occurred (Maclay and Small, 1986).

The Dolomitic member of the Kainer Formation (Fig. 4) consists of 110-130 feet of dolomitized wackestone deposited in tidal and sub-tidal environments; evidence of burrowing is commonplace (Maclay, 1995). The overlying Kirschberg Evaporite member of the Kainer Formation comprises 50-60 feet of tidal and supratidal limestone, dolomite, and evaporite deposits. The Kirschberg has extensive matrix and fracture porosity and highly permeable zones related to dissolution and collapse (Maclay and Small, 1986). The upper part of the Kainer Formation is described as the Grainstone member and consists of 50-60 feet of grainstone, wackestone, and thin beds of marl. This member represents shallow-water lagoonal deposition under moderate to high-energy conditions; matrix porosity is locally significant and cavernous, honeycombed zones are notable in the middle of the unit (Maclay and Small, 1986).

The basal hydrostratigraphic unit of the Person Formation is designated the Regional Dense member and consists of 20-24 feet of dense, argillaceous deep-water limestone; it forms a persistent confining bed within the Edwards aquifer. The overlying 70-90 feet consist of tidal and supratidal limestone and dolomite packstone that are designated the (combined) Leached and Collapsed members due to widespread honeycomb porosity and collapse breccia (Maclay and Small, 1986; Hovorka et al., 1996). The topmost unit of the Person Formation is designated the (combined) Cyclic and Marine members and consist of 80-100 feet of reefal limestone and dolomitic grainstone and packstone, evaporite beds, and argillaceous limestone (Maclay and Small, 1986). This upper unit of the Person Formation shows variable thickness and considerable porosity due to karst dissolution and brecciation related to Late Early Cretaceous erosion (Maclay, 1995, Table 3).

The Georgetown Formation forms the top of the Edwards aquifer and comprises 60 feet or less of dense, marly limestone deposited under marine conditions (Maclay and Small, 1986). The Georgetown has low porosity and permeability and generally behaves as a confining bed in the section.

Geologic Controls on Ground-Water Flow

The Miocene uplift of the Edwards Plateau was accomplished by displacements across en echelon strands of the Balcones fault zone (Fig. 3). These normal faults generally trend east-northeast and chiefly show down-to-the-south offsets. Some shorter strands show down-to-the-north offset and form the south margins of small graben and horst blocks in the complex fault zone.

Fracturing and dissolution along all fault strands contributed to development of high-permeability ground-water flow zones (Fig. 4; Maclay and Small, 1986; Maclay, 1995). Flow is strongly controlled by the trend of the Balcones fault zone for two reasons. The first relates to the fracture-induced permeability enhancement along fault zones. The second reason is that fault strands, depending on amount of offset, place hydrologically dissimilar parts of the aquifer side-by-side and therefore act as barriers or conduits for cross-fault flow (Maclay and Small, 1986). Over time, then, faults have acted to divert southeast-directed down-dip flow toward the east-northeast where major springs discharge from the Edwards aquifer.

The total thickness of rocks of the Edwards aquifer is about 450 to 500 feet in northern Bexar County, on the average (Maclay and Small, 1986). If one excludes the Basal Nodular member from the bottom and the low-permeability Georgetown Formation from the top, the average effective thickness of the aquifer would be on the order of 400-430 feet. The aquifer is contained above and below by thick sections of relatively impermeable rock (Del Rio Clay through Navarro Group above, more than 800 feet; Glen Rose Limestone below, about 900 feet). Displacements across strands of the Balcones fault zone range from a few feet to about 1000 feet. Thus, any fault offset of the Edwards aquifer places part of the permeable zone adjacent to less permeable rock and diminishes the effective thickness of the aquifer. For every 50 feet of fault displacement the aquifer thickness is reduced by about 10 percent (see Maclay, 1995, fig. 14, for schematic depiction of fault-offset effects).

Three-Dimensional Geologic Framework Model

Our data for elevation control on the hydrostratigraphic units (Fig. 5) come from several sources. We started from the tabulation of tops of hydrostratigraphic units from the 30 primary monitoring wells logged during construction in 1998; these contacts were interpreted from geophysical logs and cuttings using consistent criteria by the USGS drillers and geologists. We included similar hydrostratigraphic-unit data from several of the cored test-holes that were drilled in the 1970s by USGS during early scientific investigations of the Edwards aquifer system (Small and Maclay, 1982). Additional data for about 50 water wells were extracted from the Texas Water Development Board database; these data were generally limited to tops of major formational breaks in the Lower Cretaceous section (Georgetown, Person, and Kainer Formations), but provided vital elevation control on the position of the aquifer in many fault blocks.

The drillhole data noted above tend to be concentrated within the southern part of the aquifer recharge zone and in the confined zone to the south (Fig. 5). We supplemented the model with elevation-control data for some of the lower Kainer Formation units based on mapped geologic relations (Fig. 5; Collins, 1993, 1994, 1995). We also estimated the position of the top of the Georgetown Formation in the deep subsurface in southern parts of the model where few deep holes have been drilled (areas south of the bad-water line); positions were estimated from elevations of higher stratigraphic units and adjusted for approximate thicknesses of intervening formations.

Initial 3-D geologic models based on the fault patterns and the primary drillhole data (30 monitoring wells and 3 test-hole cores) indicated the need for additional well control in several fault blocks. We also noted patterns in some of the calculated thickness data that suggested some of the initial contact picks might need to be reviewed. Such review provided objective basis to support revision of some contacts, for example, where an anomalously thick interval of one unit in a hole was paired with an anomalously thin interval of an adjacent unit. We sought and incorporated the additional well data from TWDB sources, and incorporated the map-control data in many of the northern fault blocks.

The current model of the northern Bexar County area, depicted in this report, accurately honors all of the varied data we used as input. We imposed certain degrees of mathematical smoothing into the modeling of many horizon surfaces in order to maintain reasonably consistent thicknesses of hydrostratigraphic units from place to place. We also determined from regional isopach patterns that the Regional Dense Member could be modeled as a unit of constant thickness (22 ft; 7 m). Similarly, the Georgetown Formation was rendered as a unit of constant thickness (20 ft; 7 m), even though we are aware that it is bounded by unconformities above and below and that it has been removed by erosion in some peripheral areas of the model. Imposition of these conditions, which seem reasonable based on mapped geologic relationships, improved the appearance of subsequent models by damping out some high-frequency noise arising from modeling mathematics.

Model Results

The calculated 3-D model is only an approximation of reality because it is based on scattered data for unit tops, incomplete (and potentially erroneous) conceptual models of the fault structure, and an interpolated topographic surface. Nonetheless, we believe the fit of this model to the mapped geologic relationships is good, and that this correspondence suggests the model reasonably reflects the actual geologic structure of the area. Figure 2 shows a nearly vertical view of the top of the model volume with white contact-traces showing the mapped top and bottom of the Edwards aquifer system, which is the recharge zone for the aquifer (Stein and Ozuna, 1996). The mapped and modeled extents of the aquifer show high degrees of correspondence, in general, and the minor excursions are not unexpected given the moderate topography and low dip of the formation boundaries.

In greater detail, the model allows exploration of the inner geometries of fault blocks, fault-zone juxtapositions, and hydrostratigraphic units in the subsurface. Figure 6 shows a typical view toward the northeast along the Balcones fault zone, in which several minor fault blocks have been removed. The structural model shows the truncation of hydrostratigraphic units within the Edwards aquifer across the fault planes, as well as the continuity of units within fault blocks. All faults diminish the effective thickness of the aquifer because all displacements cause parts of the Edwards system to be placed against less transmissive units of the Glen Rose Limestone (below) and the Upper confining unit (above). Connectivity of the aquifer is maintained, however, where the upper part of the Person Formation (Cyclic + Marine Member and Leached + Collapsed Member) are faulted down against the lower part of the Kainer Formation (Kirschberg Evaporite Member and Dolomitic Member). These conclusions are not new (compare Maclay, 1995, p. 32-35), but this 3-D geologic model allows for a more quantitative assessment of the extents and effects of faulting across northern Bexar County.

The geologic model also provides a device for examining the distribution of water table measurements and exploring potential geologic causes of anomalies and trends. We used long-term average water-table data for several dozen wells in the model area to create a mathematical approximation of the regional water table (potentiometric surface), and then treated that surface like a geologic horizon in the modeling software. Figure 7 shows the result in a perspective view from the south-southwest. The water table is depicted as a fault-displaced surface that varies smoothly within individual fault blocks but is dislocated across fault planes. This result is similar to what would arise if one hand-contoured the water-table data and treated all faults as (partial) barriers, which we believe is the general rule in the Edwards system. This depiction draws attention to a fault block with anomalously high unconfined water levels in the north-central part of the recharge zone. Further examination of the model indicates this block is a small horst in which the Edwards is partially bounded by Upper confining unit beds on both the north and south sides.

Summary

The geologic framework of the Edwards aquifer system in northern Bexar County can be faith-fully modeled in full three-dimensional form, based on geologic information for numerous strands of the Balcones fault zone and well-log data for hydrostratigraphic units. This model allows analysis of the extent of hydrogeologic communication across fault planes, and particularly highlights locations in the aquifer system where the more transmissive units of the upper and lower Edwards Group are in contact. Water-table trends and anomalies can be displayed and analyzed in terms of the geologic framework of adjacent fault blocks.

ACKNOWLEDGMENTS