Figure 1 : Location of the Edwards Aquifer in relation to the Balcones fault zone and other physiographic features. Note how the aquifer cuts across the surface drainage basins.Figure 2 : Schematic cross-sections of stages in the development of the Edwards Aquifer (modified from Rose, 1972, and Abbott, 1975). Figure 3 : Topography and rainfall of the central segment of the Edwards Aquifer (modified from Woodruff and Abbott, 1979). Figure 4 : Recharge-discharge relations and potentiometric levels, central segment of the Edwards Aquifer (modified from Woodruff and Abbott, 1979). Table 1 : Long-term recharge and discharge from central segment of the Edwards atesian aquifer. Figure 5 : Drainage-basin evolution within the central and Barton Springs segments of the Edwards Aquifer:
Three river systems dissect the southern margin of the Edwards Plateau in south-central Texas: the Nueces, the San Antonio, and the Guadalupe (Fig. 1). The Edwards Plateau is a karstic upland, and its dissected margin consists of plateau outliers, narrow incised stream courses, and intervening areas of steeply sloping terrain known locally as the Central Texas Hill Country.
The upper reaches of these three drainage basins constitute an important hydrogeologic entity; they include the catchment watersheds, the recharge areas, and the points of discharge for the central segment of the Edwards artesian aquifer. The Edwards Aquifer is a major cavernous limestone system that extends for over 400 km along the Balcones fault zone from Val Verde County on the Mexican border to Bell County in north-central Texas (Fig. 1). The central part of the aquifer is the most prolific water-yielding segment and thus is the main focus of this report; it constitutes the main water supply for a region that includes the city of San Antonio and a population of more than one million people. Some attention also will be given to the 390 km2. Barton Springs segment that lies immediately north of the central aquifer segment.
Major exchanges on a regional scale occur between surface stream flow and groundwater levels in the central segment of the aquifer (Sayre and Bennett, 1942; Pettit and George, 1956; Arnow, 1963; Klemt and others, 1975; Woodruff and Abbott, 1979). In brief, most recharge occurs within the semiarid western part of this aquifer segment, while most discharge occurs in the subhumid eastern portion. Interactions between the surface and subsurface water regimes are likely to have occurred during earlier developmental stages of both the aquifer and the surface drainage network. It is our purpose to show that drainage-basin evolution and aquifer development have operated mutually. That is, within larger structural geologic and climatic controls, physiographic development near the Balcones fault zone predetermined both geographic configuration and magnitudes of recharge and discharge in the Edwards Aquifer. Moreover, aquifer development has influenced the evolution of surface drainage configurations by the diversion of surface flow via recharge in one area while augmenting stream flow via spring discharge elsewhere. We propose that these relations are due in large measure to stream piracy that chiefly occurred within the San Antonio and Guadalupe watersheds. Similar piracy also affected landform and hydrologic development farther north within the Barton Springs segment of the Edwards Aquifer (Woodruff, 1984a).
Stream piracy greatly increased the catchment area of the through-flowing Hill Country rivers where they cross resistant limestone strata within the fault zone (Woodruff, 1977; Woodruff and Abbott, 1979). The higher average rainfalls that occur in the pirated basins also enhance the ability of these streams to cut deep canyons. Deeply incised canyons were necessary to provide spring sites for groundwater that otherwise would have been trapped and then equilibrated chemically with host rocks and therefore would have ceased the dissolution of the surrounding limestones (Abbott, 1975). If piracy had not occurred, the dynamic hydrologic situation would not have developed as rapidly and the formation of cavernous porosity would have been retarded. As it happened, a region-wide circulation system developed during two diverse time periods. The first was near the middle of the Cretaceous Period when the Edwards Limstone was deposited, exposed subaerially, and buried. The second was during the Miocene Epoch when Balcones faulting occurred, and the erosion of the fault-rejuvenated stream exhumed the Edwards Limestone. Eventually, extensive cavern systems developed as the main conduits for transmission of groundwater.
The structural and stratigraphic frameworks of the study area are the basic controlling factors for both surface and subsurface drainage development. Most aquifer development occurred within rocks of the Edwards Group that were deposited on the San Marcos platform and in the Edwards-equivalent limestones of the Devils River trend (Fig. 1).
The stratigraphic and lithic characteristics of the (Albian Stage) Edwards Limestone originated with the differing depositional environments, and resultant facies, of late Early Cretaceous time. To summarize Rose (1972), the San Marcos platform acted as an area of lesser subsidence during the time of Edwards deposition. That platform was the site of accumulation of about 150 m of shallow marine and tidal-flat sediments. At the same time, along the Devils River trend roughly a 300 m thickness of grainstone and rudist boundstone was formed. Subsequent uplift along the northwest-trending axis of the San Marcos platform caused more than 30 m of the uppermost Edwards Group to be removed by erosion during late Early Cretaceous time (Fig. 2). Subaerial erosion of carbonate rock was acompanied by pore-space enlargement and cavern development resulting from circulation of shallow meteoric waters. The part of the Edwards Group that makes up the present aquifer in Bexar, Comal and Hays Counties (where discharge dominates today) was on the San Marcos platform and received significant enhancement of porosity during Cretaceous time. The parts of the aquifer in Medina, Uvalde and Kimey Counties (where recharge dominates at present) were southwest of the axis of uplift and apparently received little, if any, solution enlargement of porosity during Cretaceous time. During the remainder of the Early Cretaceous and throughout Late Cretaceous time, the entire region was covered by shallow marine-shelf waters. Deposition of argillaceous and micritic sediments resulted in the Edwards Group being covered on the San Marcos platform by a 260-m thickness of low-permeability rocks. This burial sealed off the Edwards Group and precluded the circulation of groundwater necessary to further increase porosity.
At about the end of the Cretaceous, slow upwarping of the northwestern margin of the subsiding Gulf of Mexico basin lifted the region of the present-day Edwards Aquifer above sea level. Continued deformation gave a generally southeastward dip to the sedimentary rock units of central and south Texas. At this time, deep groundwater might have augmented earlier-developed porosity, except this groundwater system would have been largely static, having no means for egress (Abbott, 1975). This stasis would have resulted in chemical equilibration between host rock and the waters contained therein, thus preventing extensive cavern development at that time.
The dominant geologic feature in the region is the Balcones fault zone, a system of en echelon, mainly down-to-the-coast, normal faults that extend about 545 km from Del Rio on the Mexican border to near Waco in north-central Texas. Faulting probably occurred primarily during the late Early Miocene (Young, 1972), as evidenced by the abundance of reworked Cretaceous fossils and limestone fragments in the fluvial sandstones (calclithite) of the Oakville Formation (Wilson 1956; Ely, 1957). The strike of individual faults within the study region is predominantly northeast-southwest, but the overall structural alignment subtly changes to a more east-west trend in the southwestern part of the region. Faulting within the San Antonio, Guadalupe, and Colorado River basins has juxtaposed the approximately 150 m-thick Edwards Limestone against the older Cretaceous Glen Rose Formation that consists largely of alternating beds of limestone, dolomite, and marl. On the downthrown (eastern) side throughout the region, the Edwards Limstone abuts against less resistant chalk, clay, and marl units of younger Cretaceous age.
Displacement along the main fault-line scarp is as little as 60 m in the westernmost part of the region, whereas a maxium displacement of about 185 m occurs in the Guadalupe River basin (Klemt and others, 1975). Similarly, total stratigraphic displacement decreases from east to west. Total displacement in Comal County is as much as 520 m over a width of 39 km (George, 1952). This fault-bound exposure of limestone has resulted in compartmentalization of the aquifer into a narrow belt that includes most of the recharge and discharge areas within the eastern basins. Farther west, however, in Uvalde County, total displacement is only 215 m (Welder and Reeves, 1962). Because of this lesser fault displacement, the aquifer is not confined to a narrow outcrop belt (Barnes, 1974a, b, c). Instead, the Edwards Limestone crops out continuously across much of the Nueces basin, and thus can receive recharge waters through a larger area than in the San Antonio or Guadalupe watersheds.
The most evident geologic controls on the physiography of the region include the topographic changes across the main line of displacement of the Balcones fault zone. There, an escarpment separates the low-relief terrain of the Gulf Coastal Plain from the ruggedly dissected Hill Country to the north and west (Fig. 3). The orientation of the escarpment changes from northeast/southwest in the Guadalupe, San Antonio and Colorado River watersheds to a more east-west trend in the Nueces River watershed. In response to these changes in relief and in orientation, the topographic position and geometry of drainage nets also change from west to east. The highest topographic elevations occur in the headward reaches of the westernmost part of the Nueces watershed. Likewise, the component rivers of the Nueces system cross the Balcones Escarpment at generally higher elevations than do streams within the San Antonio and Guadalupe basins. The escarpment is less pronounced in the Nueces basin where fault displacement is less and where streams trend throughout their entire upper courses in a generally southward direction. The most extensive alluvial plains exist south of the main fault line in the Nueces basin rather than in the areas near the escarpment to the northeast (Barnes, 1974b). Moreover, streams of the Nueces system have generally broader alluvial valleys throughout their reaches, despite the fact that they transect large outcrop areas of resistant limstone strata. Yet, studies by Rose (1972) show the overall properties of the limestones within these western basins to be simlar to rocks occurring in those parts of the San Antonio and Guadalupe basins where steep-walled canyons have been eroded, and where little or no alluviation has occurred.
Stream regimes also change in the vicinity of the Balcones fault zone. Upstream from the fault zone there are moderately wide alluvial valleys separated by broad interfluves consisting of plateau remnants. Within the fault zone, streams are incised as narrow canyons into the resistant limestone strata. Little or no alluvial deposits occur within these incised reaches. Immediately downstream from the Balcones escarpment, broad alluvial plains occur that have both modern depositional surfaces associated with active streams and high relict deposits blanketing uplands far from present fluvial activity.
Medina River, Cibolo Creek, and Guadalupe and Blanco Rivers display a distinctive geometric response to geologic controls near the Balcones fault zone. These streams flow in their upper reaches in a roughly eastward direction, the projection of which is at an acute angle to the strike of Balcones faulting. Where these streams cross the resistant limestone in the fault zone, there is an abrupt change in course to a trend roughly perpendicular to the fault-line scarp. Associated with these abrupt elbow turns is incision into steep-walled canyons within the resistant limestone. Asymmetrical drainage basins also occur, and there are relict erosional and depositional features on drainage divides near the elbow turns. Woodruff (1974, 1977) has postulated that stream piracy occurred in these reaches of the San Antonio and Guadalupe basins as a result of streams with steeper gradients eroding normal to the Balcones fault zone. Similar features and processes have been noted in the Barton Creek watershed (Woodruff, 1984) and farther north near the Jollyville plateau (Woodruff, 1985). No piracy of such large magnitude is evidenced in the Nueces basin; there are no elbow turns, assymmetrical basins, or relict fluvial features on divides between major streams. Thus, it is presumed that component streams of the Nueces system have flowed generally southward throughout their developmental history.
Besides bedrock conditions, another controlling factor affecting water regimes and landform development is climate. In the western part of the region the climate is semiarid, with mean annual rainfall as low as 48 cm in some areas (Fig. 3). This, coupled with high evaporation rates, means that streamflow and erosional potential of western streams are necessarily lower than that of the subhumid eastern basins. The Guadalupe River basin, for example, lies in the center of an oblong 80-cm isohyet, whereas the 50-cm isohyet follows the West Nueces River (Fig. 3).
Climatic differences are especially reflected in magnitudes of streamflow. For example, Guadalupe River, draining 3,932 km2 where it crosses the Balcones escarpment, has a mean flow of 10.54 m3/s. This is about three times larger than the combined discharge of the Nueces and West Nueces Rivers where they flow together south of the fault Zone; they have a watershed of 5,043 km2 with a mean discharge of 3.12 m3/s. However, these values reflect water losses caused by infiltration into the aquifer by streams of the Nueces River system above and beyond those owing to rainfall deficiencies or increased rates of evaporation. Comparing stream-gage data upstream and downstream from the fault zone, it is seen that, where the East and West Forks of the Nueces River converge below the recharge zone, their basin areas increase 74 % while their mean total discharge decreases 59 % (U.S. Geological Survey, 1974). No such recharge loss is included in the Guadalupe River water budget. These differences between the Guadalupe and Nueces flow regimes demonstrate the self-ramifying conditions that are evident in many limestone aquifers. Where water is maintained predominantly in surface flow, more stream erosion and thus incision can occur. Where water infiltrates underground, not only is there a lessened amount available to perform surface erosion, but, because of soluble bedrock, these recharging waters enlarge their flow paths, thus ensuring further underground infiltration.
The Edwards Aquifer consists of two components, an unconfined (water-table) aquifer in the plateau lands and Hill country upstream from the main faultline scarp of the Balcones fault zone, and a confined (artesian) aquifer within the eastern and southeastern part of the fault zone. Recharge to the water-table aquifer results from precipitation occuring throughout much of the Edwards Plateau; this catchment area extends beyond the drainage basins composing this study region (Fig. 4). Groundwater beneath the Edwards Plateau moves mainly toward the southeast down the regional dip of the aquifer; part of this water discharges through myriad seeps and springs that provide base flow for headwater streams in the Nueces, San Antonio, Guadalupe, and Colorado River basins. Surface streams sustained by this spring-derived base flow eventually cross the highly fractured, cavernous limestones in the Balcones fault zone. There, infiltration into the confined aquifer occurs. In addition, about 6 % of the recharge into the Edwards occurs by underflow from adjacent rock units such as the Glen Rose Formation. This recharge moves directly into the confined aquifer without having been discharged first as surface flow (William B. Klemt, writ. comm., 1977).
Major recharge occurs in two types of terrane--stream bottoms underlain by faulted or cavernous limestone, and low-relief uplands underlain by karstic limstones. The more important of the two recharge areas is where stream courses across permeable limestone. Water-budget studies in the Barton Springs segment of the aquifer have shown that about 85 % of incident rainfall is cycled through evapotranspiration, about 9 % runs off, and the remaining 6 % recharges the aquifer. Of the recharge fraction, about 85 % occurs within stream bottoms (Slade, 1984; Woodruff, 1984b). Recharge zones along bottomlands are especially apparent because stream discharges decrease through these reaches, dry or nearly dry stream beds commonly are incised into bedrock, and there is a concomitant attenuation of alluvial deposits. About 55 % of the estimated annual recharge into the central segment of the confined aquifer is supplied by the component streams of the Nueces basin (Fig. 4). Most natural discharge occurs from springs along the Balcones Escarpment, notably from Comal Springs and San Marcos Springs. Most well discharge occurs in the San Antonio area, and well pumpage is increasing with growing population demands. Total discharge from wells now often exceeds total discharge from springs (Klemt and others, 1975). In the eastern part of the central aquifer segment, the yields fran water wells are greater, discharge from springs is more voluminous, and water levels tend to fluctuate more uniformly, compared to aquifer-discharge characteristics farther west (Table 1). These observations imply that cavernous porosity is best developed near the distal end of the groundwater flow system in the areas farthest removed from the major loci of recharge. The same relations are seen in the smller Barton Springs segment (Slade, 1984). Furthermore, a county-by-county enumeration of vadose caverns conducted by the Texas Speleological Survey documents an increase in the number of caves from west to east along the Balcones fault zone. As of August, 1977, there were 22 surveyed caves in Kinney County, 63 in Uvalde County, 39 in Medina County, 81 in Bexar County, 92 in Ccmal County, and 86 in Hays County. Thus, caves surveyed in the recharge zone number 123 compared to 259 in the discharge end of the fault zone (R. Fieseler, writ. comm., 1977).
Similarly, caverns have been shown to be major conduits for groundwater flow in at least part of the artesian aquifer. Blind catfish have been found in waters discharged from wells as deep as 610 m in the San Antonio area (Hubbs, 1971). A notable cave fauna also exists in the waters of San Marcos Springs (Upa and Davis, 1976; Holsinger and Longley, 1980).
QUESTIONS AND HYPOTHESES
There are anomalies in this regional hydrologic picture. The drier western areas subsidize (by recharge) the water supplies for the areas with higher perennial rainfall and streamflow. Evidently, topography is one main control of the recharge-discharge couplet; recharge occurs primarily at higher elevations such as occur in the Nueces Basin, and discharge occurs mainly at low points. Moreover, the total catchment area of the component streams in the Nueces basin represents 49 % of the areas of the three major basins of the region. A much larger drainage-catchment area, plus a base flow augmented by spring discharge from the unconfined aquifer on the Edwards Plateau, apparently compensates for a decrease in precipitation in these westernmost basins.
But why has erosion been less in the Nueces watershed? Why was there an initial impetus for transfer of water from the semiarid west to the subhumid eastern part of the region? Why does the largest single, integrated basin in the region (the Guadalupe River) contribute insignificant amounts of recharge where it crosses the fault zone? Why did the eastern river system incise more vigorously into lower topographic levels to create initial discharge sites that controlled aquifer development, while the Nueces basin was left "high and dry?"
The hypotheses posed here are that present surface-drainage conditions, aquifer recharge-discharge relations, and direction of potentiometric gradient can be explained by several geologic determinants and by the activity of processes that have occurred as a result of these determinants.
Fracturing and displacement of pre-existing strata set into motion the overall drainage evolution of the region. These structural events affected rocks that, in turn, reflected their specific histories of deposition, diagenesis, and weathering. In brief, faulting established the structural grain that: (1) controlled cavern development during post-Miocene time, (2) established the topographic breaks that localized rapid stream incision, and (3) provided gross lateral boundaries for the aquifer host rock in the eastern river basins.
There are several processes that acted on this structurally prepared ground. Mechanical and chemical erosion by surface streams occurred in response to a change in base level. Erosion, however, did not occur equally in all areas. The diversion of large volumes of surface flow in eastern basins by stream piracy locally enhanced capabilities for surface erosional processes. Downwasting in the eastern basins also was abetted by higher rainfall rates. Because of increased downcutting, low topographic levels were reached that intersected the aquifer and allowed a few loci for discharge of groundwater to become established. Near the intersection of the Balcones Escarpment with the major drainage courses, the surface flow is augmented by spring discharge. Meantime, in areas such as the Nueces basin, surface flow and erosion were diminished by infiltration. Ultimately, both recharge and spring discharge were increased by localized dissolution of limestone because of continued exposure of soluble rock to through-flowing waters.
SURFACE-DRAINAGE RESPONSE TO BALCONES FAULTING
The generalized, mainly geometrical, region-wide evidence for stream piracy (Fig. 3) shows how the determinants for aquifer development can be tied to surface erosional processes. These presumed piracy events are ancient, perhaps as old as early Miocene. No clear-cut response is expected in present stream regimens or deposits, nor is any such response observed. Any evidence for ancient piracy events would be expected to occur only on the drainage divides and not in valley bottoms or in present stream regimens. Yet, fluvial deposits on drainage divides underlain by resistant bedrock in a highly dissected terrain are highly susceptible to the effacing actions of erosion.
There are alluvial deposits at various levels across the inner Gulf Coastal Plain where substrate consists of easily erodible clay and marl strata (Barnes 1974a, b, c). But these gravel deposits occur in the western as well as in the eastern river basins, so that these gravel deposits on the Coastal Plain do not substantiate the piracy hypothesis. The geometry and topographic position of these deposits can, however, augment the geometrical stream-net and drainage-net picture already proposed. This is done by extrapolating depositional trends on the Coastal Plain "up-gradient" to relict fluvial features on drainage divides as demonstrated by Woodruff (1977).
The Hill Country is a carbonate-rock terrane in which fluvial deposition is restricted in areal extent. This is an area in which alluvial materials are much more easily eroded than underlying limestone bedrock. For these reasons, no inverted topography exists such as that along the Coastal Plain. Nonetheless, scattered fluvial features do occur on uplands in the Hill Country and, notably, these features occur on drainage divides near some of the abrupt elbow turns in present streams. They exist in the exact areas the piracy hypothesis and regional geometrical features predict they would occur.
Preservation of fluvial features is associated spatially with upland karstic plains. This is probably because the karstic plains have afforded avenues for subsurface infiltration of water rather than the channelling of this incident precipitation into surface drainage courses with concomitant erosion. Because of gentle slopes, resistant bedrock, and predominant subsurface infiltration of water on these karstic plains, the drainage density is low (Woodruff, 1975, p. 25). Thus, karstic plains are an ideal locality for the preservation of relict landforms.
Karstic plains also occur along lowlands adjacent to active streams where stream discharge is attenuated and recharge occurs via the "recharge caverns" of Thrailkill (1968). Recharge caverns occur along some low-lying reaches of Cibolo Creek, and Medina River, and in similar local terranes within the Nueces River basin (Fig. 5c). In each of these areas, the substrate is resistant limestone. The caverns afford direct avenues for groundwater recharge and surface stream activity is correspondingly lessened. The "equilibrium landscape" of Hack (1965) occurs where the landforms are adjusted to the ongoing processes. The probable equilibrium landscape for resistant, solution-prone limestones is the low-relief karstic plain. The topography, soils and other characteristics of these terranes are determined by the processes of concentrated groundwater infiltration coupled with attenuated erosional or depositional activity by surface streams. Thus, a low-relief carbonate-rock terrane coexisting with fluvial features at topographic levels far above active stream courses may be out of adjustment with present landforms and processes. It is deduced that the landform-substrate assemblage on upland karstic plains represents a former level of slow fluvial downcutting with high rates of groundwater recharge, which resulted in the development of an extensive cavern network.
The disequilibrium between topographically high, relict karstic plains and adjacent deeply incised areas is significant in both a geomorphic and a hydrogeologic context. Geomorphically, the disequilibrated association is in itself an indicator of piracy in the eastern basins, as no such high-relict karstic plains exist near the fault zone in the western basins. For example, the Guadalupe River has an enormous erosional impetus because of basin size and subhumid climatic conditions. In terms of downcutting, this river has acted as would be expected from its erosional capabilities; it is deeply incised. However, the presence of relict upland plains in the river's mid-basin is all the more perplexing. That is, if any stream in the region should have a throughly dissected basin, with maximum hillslopes and minimun hilltops, it should be the Guadalupe River. That the most dissected basins occur within the Nueces system, and extensive remnant highlands occur in the Guadalupe watershed, means that there must have been either drastic changes in underlying bedrock or differences in prior drainage history. Bedrock is the same regionwide, but stream piracy affords a means for explaining this observed condition (Woodruff and Abbott, 1979).
The hydrogeologic significance of these disequilibrated landscapes relates to inferred changes in locations and magnitudes of the recharge-discharge couplets. Before piracy occurred, high rates of recharge probably occurred in the eastern as well as the western basins, as evidenced by karstic plains on the eastern divides. Although the locus of discharge of these waters can only be inferred, an eastern river such as the Guadalupe would have had a somewhat lessened discharge and lower erosion rate owing to water losses into the cavernous aquifer. Only after integration by the Guadalupe of its present headwaters and subsequent breaching of the karstic recharge plain would marked increases in downcutting have occurred. Then, the Guadalupe River was draining the cavernous aquifer system that it had previously recharged.
Piracy-induced incision plus high surface discharge from a large subhumid river basin combined to produce the highest erosion rates of the entire region. This ultimately accomplished two things: 1) erosion provided topographically low points for spring discharge that became engrained as base levels toward which most of the artesian aquifer flowed; 2) in the Guadalupe River basin, incision occurred at such a high rate that most of the upper aquifer levels were completely breached, and discharge from the aquifer (instead of recharge into it) became the major process. Part of the aquifer system draining to San Marcos Springs apparently does extend beneath Guadalupe River, and the Edwards Limestone crops out along a short reach of this deeply-incised river. Yet no long-term recharge is shown to have occurred for the Guadalupe River (Table 1). This anomaly may be explained by the relatively small volume of caverns within the part of the aquifer that underlies the Guadalupe River. Thus, the pore space beneath the Guadalupe River may be essentially full of water under normal climatic conditions, and only during extreme drought conditions might this cavern system be able to accept recharge from the Guadalupe River. This thesis is substantiated to some extent by the lesser fluctuations of discharge from San Marcos Springs during times of drought compared to the normally larger Comal Springs (Brune, 1975). Not only is Comal Springs probably more adversely affected by increased discharge by well-pumpage in the San Antonio area, but San Marcos Springs may be recharged to some extent during very dry years by the Guadalupe River, a condition that does not occur during wetter times simply because the cavernous pore space within the small segment of the aquifer that underlies the Guadalupe River is normally filled to capacity.
Where the Guadalupe River crosses the Balcones fault zone, its drainage area increases by 15 %, while its mean annual discharge increases by 34 % (U.S. Geological Survey records, courtesy of Texas Natural Resources Information System). This increased rate of flow across the fault zone is a result of higher rainfall in the middle part of the basin, input of water from Hueco Springs, and presumably little or no recharge into the Edwards Aquifer. During the peak drought year of 1956, however, discharge increased across the fault zone by only 12 %, a decrease in expected flow of about 84 hm3. This is a 36 % decline compared to long-term yearly averages. Assuming that the drought's effects on the water budget had equal impact throughout the river basin, there are only two means for effecting this relative decrease in discharge: diminished flow from Hueco Springs, or infiltration into the Edwards aquifer. Hueco Springs did indeed experience decreased flow; no spring discharge was recorded during 1956 (Texas Board of Water Engineers, 1959). But in order to attribute all of the decreased Guadalupe River discharge through these reaches to diminished spring discharge would require a decline of 2690 1/s in the long-term rate of flow fran Hueco Springs. There is no exact figure for mean annual discharge from Hueco Springs because of its erratic flow. However, Brune (1975, p. 38) cited 3710 1/s as the maximum discharge for Hueco Springs. He further listed 33 spring-discharge rates measured over a period of 48 years, the highest of which was 2322 1/s while the lowest was zero. Moreover, all 136 discharge measurements for Hueco Springs cited by the Texas Board of Water Engineers (1959) present values less than the average rate required to account for the computed water loss from the Guadalupe River. It is likely that some fraction of the diminished flow in the Guadalupe River during drought periods is a result of recharge into the Edwards Aquifer.
INTEGRATION OF LANDFORM
EVOLUTION AND DEVELOPMENT OF THE EDWARDS AQUIFER
Determinants that shaped the present recharge-discharge geometry of the Edwards aquifer began during Early Cretaceous time with differential uplift across the San Marcos platform and associated early development of cavernous porosity in the Edwards Limestone in the area that later was to become part of the San Antonio and Guadalupe River basins. Subsequent burial by younger Cretaceous rocks of low permeability precluded further significant porosity development during Cretaceous time. During Miocene time, Balcones faulting created a network of fractures that crisscrossed the Edwards Limestone along a strike distance of 545 km. These fractures not only provided gross structural boundaries for much of the 400-km-long aquifer, but also they superimposed additional permeability conduits upon both primary (interparticle and other) and secondary (Cretaceous dissolution) porosity systems. Fault displacement was greatest in the very areas along the San Marcos platform that had experienced Cretaceous augmentation of primary porosity. Lesser fault displacements occurred in the areas southwest of the San Marcos platform. This affected aquifer development in three ways.
1. The western part of the fault system was dropped "down-to-the-coast" the least and thus became the structurally highest part of the Edwards Limestone. This was later reflected in a higher topographic level for Edwards outcrops in southwestern areas, especially in the Nueces watershed.
2. The greater fault displacements in the east caused the Edwards Limestone to be bounded structurally by rocks of lesser permeability and solubility. The compartmentalization into discrete lithic packages channeled groundwater flow and thus concentrated porosity enhancement within major fault blocks (Abbott, 1975).
3. The geometry of faulting set in motion the aforementioned surface drainage responses (Fig. 5). Stream piracy resulted from the geometrical relations between pre-faulting surface drainage nets in the western and eastern basins and the strike of the fault zone. Greater fault displacement in the eastern areas resulted in more rapid incision by streams flowing normal to the incipient escarpment. These rapidly eroding streams were the first to exhume the already porous Edwards Limestone. This exhumation might have occurred initially before headward stream reaches captured major regional streams. But at any rate, piracy and the resulting vigorous downcutting provided exposures within the Edwards Limestone at low topographic levels.
As soon as the Edwards Limstone was breached by the pirate stream, pent-up groundwater was released fran the proto-aquifer, thus beginning the engrainment of the flowpaths of groundwater mving toward these few discharge points (see Figs. 2c, 2d). Notwithstanding the presence of primary porosity, Cretaceous solution-enlarged porosity, and Miocene fracture porosity, an effective groundwater flow system could not have developed until the overlying blanket of fine-grained sedimentary rocks was breached, thus exposing the soluble limestone to both recharge and discharge. Continual region-wide groundwater circulation developed as component streams of the Nueces system exhumed the structurally and topographically higher Edwards Limstone in the western basins. Downcutting was less rapid there because of lesser fault displacement and because there was no piracy-initiated increase in discharge. Ultimately, because of lower rates of downcutting by western streams coupled with lesser fault-displacement, these western basins were thoroughly dissected. This resulted in broad expanses of limstone being exposed along stream courses at relatively high topographic elevations.
Interconnection among the drainage catchment areas by fault-generated fracture systems allowed long-distance interbasin groundwater transfer from the topographically higher western areas to the lower, more permeable discharge points to the northeast. This set in motion the continuously circulating, self-ramifying groundwater flow system that converged toward the loci of the few springs. In this way, the initial discharge sites became the "drains" for the central segment of the aquifer, drawing on waters throughout the several drainage basins that en compass more than 17,695 km2. In the same way, Barton Springs became the drain for the underground watershed comprising the surface catchment areas of Barton, Williamson, Slaughter, Bear, Little Bear, and Onion Creeks.
The ancient engrainment of the aquifer helps explain noteworthy features of the present Edwards Aquifer system such as the paucity of springs and the origin of the "bad-water line." Although the Edwards Aquifer in the fault zone is about 400 km long, there are only about a dozen large springs discharging from the system (Sayre and Bennett, 1942). Six of the major springs occur in the 280-km-long central segment: Leona Springs in Ulvalde county; none in Medina County; San Antonio and San Pedro Springs (four km apart and rising along the same fault) in Bexar County; Comal and Hueco Springs in Comal County; and San Marcos Springs in Hays County (Fig. 4). These springs issue forth at progressively lower elevations to the northeast, and all but Leona Springs occur near major pirate streams. They probably discharge from enlarged lower-level conduits of the initial flow system honed to the earliest discharge sites. That is, the loci of discharge probably migrated to progressively lower topographic levels near fault traces as dictated by changing base levels. The controlling base levels were in turn established by vigorously downcutting pirate streams. The general lowering of interfluvial areas and exposure of the Edwards Limstone over large areas have not caused more springs because the plumbing system was engrained long ago toward the few original discharge sites.
The southeastern boundary of the aquifer is a "bad-water line" that separates the potable water of the cavernous, high-yield aquifer from the high-salinity water on the downdip side. Although the bad-water line is roughly parallel to the trend of the Balcones fault zone, its detailed course generallly disregards individual faults and facies boundaries (Abbott, 1975; Woodruff and others, 1982). It can be understood as the solution-engrained original flow boundary of groundwater that moved toward the earliest discharge sites. This hydraulically controlled boundary probably marks the down-dip potentiomtric boundary as originally affected by the subtle draws of low-elevation springs.
Initially, recharge probably occurred through karstic plains at high topographic levels within the San Antonio and Guadalupe watersheds (Fig. 5c), and much of the cavern development was confined to individual fault blocks in a trend subparallel to the cavern system that lies beneath the western edge of the Gulf Coastal Plain. When streams trending normal to the fault scarp effected piracy, the greater volumes of flow rapidly cut through the high-level karstic plains and stranded sone of the subjacent cavern systems that formerly recharged their local areas. Upland karstic plains are the relicts of this former equilibrated landform/process couplet. The post-piracy couplet enhanced stream incision, while the developmnt of caverns occurred at lower topographic levels, as is presently seen along Cibolo Creek and Medina River (Fig. 2d and fig. 5c). These low-lying karst plains provide a periodic influx of recharge undersaturated with respect to calcite and dolomite, which is so important for the continued solutional growth of the Edwards Aquifer.
The recharge-discharge geometry described here does not fit the general concept that holds that dissolution is concentrated near recharge sites where groundwater is least saturated with respect to calcite and dolomite and thus is most aggressively able to form caverns. If this common view of cavern formation held for the Edwards artesian aquifer, then the cavern systems in Kinney, Uvalde, and Medina Counties should be more highly developed, and yields from springs and wells should be greater there. Likewise, cavern development in Bexar, Comal, and Hays Counties should be much less than it is, because groundwater that has traveled a long distance should be saturated or supersaturated with respect to calcite and dolomite and hence unable to accomplish any further dissolution. If the groundwater had been mostly saturated during the developmental history of the aquifer, then caverns in the eastern areas would be poorly developed, and yields from wells and springs would be low. Since the caverns are best developed near the distal end of the groundwater system, then clearly the groundwater passing through Bexar, Comal, and Hays Counties has primarily been undersaturated with respect to calcite and dolomite.
Today, groundwater within the Edwards Aquifer is at least seasonally undersaturated, as shown by negative saturation indices calculated for calcite and dolomite from well and spring water samples near Comal, San Marcos, and Hueco Springs by Pearson and Rettman (1976) and by Abbott (1977a). The mechanism that maintains at least seasonal undersaturation in the ground-water appears to be the mixing-of-waters effect explained by Thrailkill (1968). Introduction of recharge, even if it is saturated or supersaturated, may cause undersaturation in the main groundwater body if the recharge is cooled upon entering or if it is mixed with water in equilibrium with a lower partial pressure of carbon dioxide. The necessary process is the addition of carbon dioxide into the main groundwater body. The resulting increase in carbonic acid pemits further dissolution of limestone and dolomite. Maintenance of chemical undersaturation over the long distances within the confined Edwards Aquifer must be a result of the introduction of high-volume vadose flows recharged through sinks in stream bottoms in Bexar and Comal Counties. Although the recharge supplied by Cibolo, Dry Comal, and other creeks crossing the middle part of the aquifer comprises less than one-third of the water in the aquifer, its delivery occurs at strategic points that promote undersaturation; moreover, because rainfall in the region commonly occurs as downpours from convective thunderheads, much of the recharge occurs rapidly and is undersaturated with respect to calcite and dolomite.
SUMMARY AND CONCLUSIONS
The primary determinants that caused streams in the eastern part of the Balcones Escarpment to incise deeply to a topographically low level were the greater fault displacements across the area of the Cretaceous San Marcos platform compared to areas to the southwest. Faulting markedly affected surface and subsurface drainage evolution in several ways:
1. Progressive downdrop of fault blocks toward the Gulf of Mexico created a disequilibrium along established surface streams graded to their previous base levels.
2. Resistant limestone strata juxtaposed against more erosive rock resulted in additional changes in stream regimes, and the locus of major fault displacement was perpetuated as a fault-line scarp.
3. Fractures associated with faulting offered preferred orientation for surface stream erosion into resistant bedrock. Additionally, fractures acted as initial conduits for recharge and regional transfer of groundwater. These underground-flow conduits and the pre-existing porosity within the Edwards Limestone were selectively enlarged by dissolution.
4. Several lines of evidence, including the frequency of caves and the occurrence of cave fauna, indicate that cavernous porosity is well developed and interconnected. The arcuate path of the Balcones faults runs 240 km in a least-distance path from the western drainage divide in Kinney County to San Marcos Springs; some groundwater travels at least this far.
Streams in the eastern part of the region originally trended at acute angles to the strike of major fault displacement. Thus, when escarpments began to form, the eastward-trending streams were vulnerable to piracy by rapidly eroding, smaller streams with courses normal to the escarpment. After piracy, more rapid incision commenced because of the much larger discharge suddenly available from newly acquired headwaters. No such situation existed in the more arid Nueces basin, where total fault displacement was less and streams already flowed normal to the escarpment. Thus, rather than incising deep valleys, the component streams of the Nueces system established generally wider valleys and more extensively dissected uplands at higher topographic elevations, despite the fact that these streams crossed the same rock types in which incision occurred elsewhere. As alluvial plains were created in the western basins, and as infiltration occurred, low-lying karstic plains were formed as described by Thrailkill (1968) for recharge areas.
The impetus for the transfer of groundwater from the semiarid western region to the subhumid northeastern region resulted from several factors:
1. The porosity and permeability owing to faults and fractures associated with the Balcones fault system provided avenues for meteoric water to enter the Edwards Limstone, and the fault-block trends guided or channeled the movement of water.
2. Although the regional dip is to the southeast, a major component of dip within the Balcones fault zone is toward the northeast.
3. The overlying seal of low-permeability sedimentary rocks was breached by rapidly degrading, scarp-normal pirate streams in the eastern watersheds. When these pirate streams intercepted the Edwards Limestone, they created discharge sites for the pent-up groundwater.
4. The creation of a few discharge sites in the east and the broader dissection in the west set in motion a continously circulating groundwater-flow system that converged toward the few springs. These springs are most voluminous in areas containing cavernous porosity developed during the Cretaceous on the San Marcos platform.
The transmitted effect of a few discharge sites at the low-elevation end of the fault zone drew confined groundwater from more than 200 km away. The originally subtle draw of low-elevation springs was engrained into an integrated cavern system by the dissolution accomplished by groundwater kept undersaturated by periodic influxes of fresh water. The extensive development of caverns near the discharge sites emphasizes the influence of the progressively greater flow of groundwater funneled toward the springs. When the aquifer system is viewed in scale with its great extent, few springs, and small number of recharge streams, it appears to be a long, thin tabular container with few entry and exit points.
The regional potentiometric gradient extends from northwest to southeast down the regional dip, yet most groundwater flows through cavern systems parallel to the Balcones faults that are at right angles to the apparent regional potentiomtric gradient. In effect, the Edwards Limstone is like a tabular container inclined to the southeast, but fluid flow is stopped at the downdip permeability barrier (container wall) expressed as the "bad-water line." The early discharge sites created by the pirate streams are analogous to pulling plugs from the northeastern edge of the southeast-inclined container. Water runs to these exit points rather than down the regional slope.
This paper is a modification of an article originally published in Earth Surface Processes (Woodruff and Abbott, 1979). We thank Richard Dillon of the Bureau of Economic Geology, The University of Texas at Austin, for making available original drafting for modification in this presentation.
Abbott, P. L., 1974, Calcitization of Edwards Group dolomites in the Balcones fault zone aquifer, south-central, Texas: Geology, v. 2, no. 7, p. 359-362.
Abbott, P. L., 1975, On the hydrology of the Edwards Limestone, south-central Texas: Journal of Hydrology, v. 24, no. 3/4, p. 251-269.
Abbott, P. L., 1977a, On the state of saturation of groundwater with respect to dissolved carbonates, Edwards artesian aquifer, south-central Texas: Texas Journal of Science, v. 29, nos. 3-4, p. 159-167.
Abott, P. L., 1977b, Effect of Balcones faults on groundwater movement, south-central Texas: Texas Journal of Science: v. 29, nos. 1-2, p. 5-14.
Arnow, T., 1963, Groundwater geology of Bexar county, Texas: U.S. Geological Survey Water Supply Paper 1588, 36 p.
Brune, G., 1975, Major and historical springs of Texas: Texas Water Development Board Report 189, 94 p.
Barnes, V. E., project coordinator, 1974a, Austin Sheet. The University of Texas at Austin, Bureau of Economic Geology Geologic Atlas of Texas, scale 1:250,000.
Barnes, V. E., 1974b, San Antonio Sheet: The university of Texas at Austin, Bureau of Economic Geology, Geologic Atlas of Texas, scale 1:250,000.
Barnes, V. E., 1974c, Sequin Sheet: The University Of Texas at Austin, Bureau of Economic Geology, Geologic Atlas of Texas, scale 1:250,000.
DeCook, K. J., 1963, Geology and ground-water resources of Hays County, Texas: U.S. Geological Survey Water-Supply Paper 1612, 72 p.
Ely, L. M., 1957, Microfauna of the Oakville Formation, Lagrange Area, Fayette County, Texas: University of Texas (Austin), M.A. thesis (unpublished), 118 p.
George, W. O., 1952, Geology and ground-water resources of Comal County, Texas: U.S. Geological survey Water-Supply Paper 1138, 126
Hack, J. T., 1965, Gecmorphology of the Shenandoah Valley, Virginia and West Virginia, and origin of the residual ore deposits: U.S. Geological Survey Professional Paper 484, 84 p.
Holsinger, J. R. and Longley, G., 1980, The subterranean amphipod crustacean fauna of an artesian well in Texas: Smithsonian contributions to Zoology, no. 308, 62 p.
Hubbs, C., 1971, Texas cave fishes, in Lundelius, E. L., and Slaughter, B. H., eds., Natural History of Texas Caves: Dallas, Gulf Natural History, 174 p.
Klemt, W. B., Knowles, T. R., Elder, G. R., and Sieh, T. W., 1975, Ground-water resources and model applications for the Edwards (Balcones fault zone) aquifer: Texas Water Development Board Open File Report, 93 p.
Pearson, F. J., Jr. and Rettnian, P. L., 1976, Geochemical and isotopic analyses of waters associated with the Edwards Limestone aquifer, central Texas: U.S. Geological Survey Open File Report, 35 p.
Pettit, B. M., Jr. and George, W. O., 1956, Groundwater resources of the San Antonio area, Texas: Texas Board of Water Engineers Bulletin 5608, 842 p.
Reeves, R. D. and Ozuna, G. B., 1985, Compilation of hydrologic data for the Edwards aquifer, San Antonio area, Texas, 1982, with 1934-82 summary: Edwards Underground Water District, Bulletin 42, 131 p.
Rose, P. R., 1972, Edwards Group, surface and subsurface, Central Texas: The University of Texas at Austin, Bureau of Economic Geology Report of Investigations 74, 198 p.
Sayre, A. N. and Bennett, R. R., 1942, Recharge, movement, and discharge in the Edwards Limestone reservoir, Texas: Americal Geophysical Union Transactions, v. 23, part 1, p. 19-27.
Slade, R. M., Jr., 1984, Hydrogeology of the Edwards aquifer drained by Barton Springs: Austin Geological Society Guidebook 6, p. 9-35.
Texas Board of Water Engineers, 1959, Summary of peak flood flow measurements and other measurements of stream discharge in Texas at points other than gaging stations: Texas Board of Water Engineers Bulletin 5807C, 255 p.
Thrailkill, J., 1968, Chemical and hydrologic factors in the excavation of limestone caves: Geological Society of America Bulletin, v. 79, no. 1, p. 19-46.
Tupa, D. D. and Davis, W. K., 1976, Population dynamics of the San Marcos salamander Eurycea nana Bishop: Texas Journal of Science, v. 27, no. 1, p. 179-195.
U.S. Geological Survey, 1974, Water resources data for Texas: Part 1: surface water records, 609 p.
Welder, F. A. and Reeves, R. D., 1962, Geology and ground-water resources of Uvalde County, Texas: Texas Water Commission Bulletin 6212, 252 p.
Wilson, J. A., 1956, Miocene formations and vertebrate biostratigraphic units, Texas Coastal Plain: American Association of Petroleum Geologists Bulletin, v. 40, no. 9, p. 2233-2246.
Woodruff, C. M., Jr., 1974, Evidence for stream piracy along the Balcones escarpment, central Texas: Geological Society of America, Abstracts with Programs, v. 6, no. 7, p. 1010.
Woodruff, C. M., Jr.,1975, Land capability in the Lake Travis vicinity, Texas--A practical guide for the use of geological and engineering data: The University of Texas at Austin, Bureau of Economic Geology Report of Investigations 84, 37 p.
Woodruff, C. M., Jr., 1977, Stream piracy near the Balcones fault zone, central Texas: Journal of Geology, v. 85, no. 4, p. 483-490.
Woodruff, C. M., Jr., 1984a, Stream piracy - possible controls on recharge/discharge geometry, Edwards aquifer, Barton Springs segment: Austin Geological Society Guidebook 6, p. 61-66.
Woodruff, C. M., Jr., 1984b, Water budget analysis for the area contributing recharge to the Edwards aquifer, Barton Springs segment: Austin Geological Society Guidebook 6, p. 36-42.
Woodruff, C. M., Jr., 1985, Jollyville Plateau--geologic controls on aquifer development: Austin Geological Society Guidebook 8, p. 4-9.
Woodruff, C. M., Jr. and Abbott, P. L., 1979, Drainage-basin evolution and aquifer development in a karstic limestone terrane, south-central Texas, USA: Earth Surface Processes, v. 4, no. 4, p. 319-334.
Woodruff, C. M., Jr., Henry, C. D., and Gever, C., 1982, Geothermal resource potential at military bases in Bexar, Travis, and Val Verde Counties, Texas: Appendix H in Geothermal Resource Assessment for the state of Texas, Status of progress, 1980: The University of Texas at Austin, Bureau of Economic Geology, Final Report to U.S. Department of Energy, Division of Geothermal Energy, under contract no. DEAS07-79ID12057, 63 p.
Young, K., 1972, Mesozoic history, Llano region, in Barnes, V. E., Bell, W. C., Clabaugh, S. E., Cloud P. E., Jr., McGehee, R. V., Rodda, P. U., and Young, K., eds., Geology of the Llano Region and Austin Area, Field Excursion: The University of Texas at Austin, Bureau of Econmdc Geology Guidebook 13, 77 p.go to: Contents : Next Article