Volume 34, Issue 24
Hydrology and Land Surface Studies
Free Access

Soil and topographic controls on runoff generation from stepped landforms in the Edwards Plateau of Central Texas

Bradford P. Wilcox

Bradford P. Wilcox

Ecosystem Science and Management, Texas A&M University, College Station, Texas, USA

Search for more papers by this author
Larry P. Wilding

Larry P. Wilding

Soil and Crop Sciences Department, Texas A&M University, College Station, Texas, USA

Search for more papers by this author
C. M. Woodruff Jr.

C. M. Woodruff Jr.

Woodruff Geologic Consulting, Austin, Texas, USA

Search for more papers by this author
First published: 12 December 2007
Citations: 55


[1] The rugged Hill Country of Central Texas is part of the extensive Edwards Plateau region. Significant portions of the Texas Hill Country overlie the Glen Rose Formation, which is characterized by a stair-step topography formed by the weathering of interbedded carbonate materials having different weathering susceptibilities. This process has sculpted the strata into a series of “risers” and “treads” that mimic stairways. In this paper, we document the soil hydrology within the riser–tread catena. Our results are counterintuitive in that we find the highest infiltration and deepest soils on the steep riser slopes. In addition, we find that the riser subsoils are saturated or very wet for extended periods. On the basis of these findings, we suggest that (1) groundwater recharge on these hillslopes is minimal and occurs only in highly fractured zones; (2) the water-holding capacity of the subsoils is sufficient for supporting the woody vegetation; and (3) runoff generation occurs as a combination of surface and subsurface flow, with the risers serving as sinks or recharge zones and the treads as source areas.

1. Introduction

[2] It is often assumed that in rangelands of subhumid to dry climates, runoff generation is exclusively Horton overland flow—that is, runoff occurs only when rainfall exceeds the infiltration capacity of unsaturated soil. In many of these rangelands, however, runoff generation can be more complex. For example, in the United States frozen soil runoff [Seyfried and Wilcox, 1995], saturation overland flow [Lopes and Ffolliott, 1993], and shallow subsurface flow [Wilcox et al., 1997; Wilcox et al., 2005] have all been documented. In addition, on rangelands where springs occur, there is a groundwater flow component to runoff [Huang et al., 2006].

[3] Runoff generation has been found to be especially complex on rangelands underlain by limestone bedrock [Wilcox et al., 2006]. Limestone and other carbonate rocks in the Edwards Plateau proper often display solution features and/or fractures that facilitate the subsurface flow of water. Also, soils on these landscapes often have relatively high permeability. Karst landscapes, areas in which dissolution of bedrock is one of the dominant geomorphic processes, occupy 10–20 percent of the earth [Palmer, 1991]. Although extensive areas of carbonate and karst terrains exist within subhumid and semiarid landscapes, we know relatively little about how runoff is generated from these landscapes.

[4] One of the more important karst rangelands in the United States, and perhaps in the world, is the Edwards Plateau region in Central Texas. In spite of the relatively dry climate of this region, the margins of the plateau uplands are drained by springs that feed perennial streams [Woodruff and Abbott, 1979]. Even so, water is becoming a limited resource as population pressures increase—which makes the need to better understand runoff generation and its relationship to vegetation, soil, and geologic properties even more urgent.

[5] The Edwards Plateau region is bordered on the south and east by the Balcones Escarpment and encompasses two major physiographic features—the Texas Hill Country and the Edwards Plateau proper (Figure 1). The Plateau tablelands are capped with a thick sequence of Cretaceous limestones, consisting mostly of the Edwards Group [Barnes, 1992]. Following the main displacement events of the Balcones Fault Zone, the eastern and southern margins of this limestone cap progressively eroded, forming the Hill Country [Anaya, 2004; Woodruff and Abbott, 1979]. The principal bedrock unit underlying most of the Hill Country is the Glen Rose Formation, a stacked, repetitious sequence of limestone and dolomitic beds [Barker and Ardis, 1996; Stricklin et al., 1972] with varying degrees of weathering potential [Woodruff and Wilding, 2007]. It thus exhibits a distinctive “stair-step” topography. The more resistant layers (underlain by carbonate rocks having generally homogeneous fabrics) form gently sloping “treads.” The intervening “risers” are underlain by limestone that (when fresh) is well indurated, with low porosity, but they exhibit heterogeneous fabrics consisting of fossils and myriad clay partings. These interbeds weather rapidly when exposed to water (Figure 1). The soil sequence thus produced is a counterintuitive one: the deepest soils are on the risers and the more shallow soils are on the treads [Wilding, 1993, 1997; Woodruff and Wilding, 2007]. Riser soils comprise an upper depositional zone consisting of colluvium and redeposited pedisediments and a lower (inner) residual soil/regolith.

Details are in the caption following the image
(top) The major physiographic regions in central Texas including the Hill Country, Edwards Plateau, and the Llano uplift. (middle) The stair-step topography that is typical for the Glen Rose limestone in the Texas Hill Country. (bottom) Common features of soils along the riser-tread continuum.

[6] Significant soil areas of this landscape have been mapped in the Brackett series, which has been described as thin, strongly calcareous, and exhibiting only slight pedogenic development [Werchan et al., 1974]. Wilding, however, has taken exception to this characterization [Wilding, 1993, 1997], arguing that these soils are often deeper, better developed, and richer in organic matter than previously thought; and that the soil patterns change in predictable ways depending on the topographic setting. He noted [Wilding, 1997] that the map unit design needed to be changed because the landscape was composed not as a single soil (monotaxa consociation), but as a complex assemblage of different soils that vary greatly in depth and are found as disjunct bodies linked together in a systematic pattern—from deep (on the upper risers) to moderately deep (on the mid risers) to shallow (on the lower risers) to very shallow (on the treads) (Figure 1).

[7] Soils on the risers are generally classified as Udic Calciustolls, or occasionally as Petrocalcic Calciustolls, and range in thickness from 1 to 3 m. They are developed in a thin veneer of limestone/dolomite-rubble colluvium, which serves as the parent material for organic-matter-rich, gravelly A horizons. These are uncomformably superposed over partially weathered in-situ materials in which pedogenically enriched carbonate horizons (calcic or petrocalcic) are formed. In the common parlance this carbonate-rich layer is called “caliche.”

[8] On the gently sloping treads, soils are thinner (<0.5 m) and classified as Lithic Haplustepts, Lithic Calciustepts, Lithic Calciustolls, and Lithic Petrocalcic Calciustolls. Generally lighter in color, they contain less organic carbon than the riser soils. But light soil colors belie the fact that even on treads surface horizons commonly have more than 2.5% organic carbon. The sola are formed in thin, loamy-gravelly, carbonatic pedisediments derived as erosional products from upslope transport. These sediments are admixed with weathered material or superposed directly over the hard bedrock, in which many of the joint planes are partially plugged with pedogenic calcite cement.

[9] Live oak–juniper woodlands have been present in the Hill Country for millennia, but have probably become more extensive and denser in the last century [Amos and Gehlbach, 1988; Huxman et al., 2005; Wilcox and Thurow, 2006]. An ecological question of considerable importance—and one that continues to spark debate—is the source of the water for these trees. A prevailing hypothesis has been that the soils in the region are thin and lack the water-holding capacity needed to support woody vegetation, and therefore the trees must be routinely accessing water from deeper groundwater sources [Jackson et al., 1999]. Alternatively, the “caliche” residuum is another source of water that has been little appreciated [Duniway et al., 2007; Hennessy et al., 1983].

[10] Relatively little is known of the dynamics of runoff generation in this landscape. Small-plot infiltration studies on the Edwards Plateau proper have repeatedly demonstrated that these soils have inherently high infiltration capacities, especially where vegetative ground cover is good [Knight et al., 1984; Thurow et al., 1986]. For the Central Texas Hill Country the hillslope hydrology of the tread/riser slopes have yet to be comprehensively investigated. Larger-scale rainfall simulation studies in the Edwards Plateau region [Munster et al., 2006; Taucer, 2006], including some at the small-catchment scale [Huang et al., 2006; Wilcox et al., 2005], indicate that runoff can occur both as overland flow and as subsurface flow. The purpose of this paper is to evaluate and describe the hydrological properties—including infiltration, runoff generation, and soil wetness—associated with the soil continuum of the tread/riser topographic sequence.

2. Methods

[11] From 1991 to 2006, more than 100 trenches were excavated by backhoe at three locations in the Central Texas Hill Country of Travis County, Texas [Woodruff and Wilding, 2007]. The trenches, each 10–15 m long, were excavated down to indurated bedrock perpendicular to the slope, incorporating a riser and tread element with horizontal dimensions up to 30 m. At selected trench locations, soil infiltration capacity, runoff and erosion, and soil matric potential were determined.

2.1. Infiltration

[12] Soil infiltration capacity was estimated using a portable, drip-type infiltrometer similar to the one described by Blackburn et al. [1974]. The raindrop-producing module, which measures 0.5 m2, is constructed from a pair of plexiglass plates spaced about 1 cm apart, sealed with caulking compound, and bolted to aluminum angle. The module has more than 1000 hollow steel needles, evenly spaced, for producing the raindrops.

[13] Runoff plots were constructed by inserting flexible metal frames into the soil, each attached to a water collector at the base of the plot. The tear-shaped plots thus formed were around 0.3 m2. The area of each runoff plot was determined by placing a grid over the plot. Simulated rainfall was applied at rates of 10–35 cm/hr until steady-state runoff was reached (usually 30–60 minutes). The difference between the application rate and the runoff rate was considered to be the infiltration capacity.

[14] Soil infiltration rates were determined at 40 locations adjacent to 11 of the trenches. The number of infiltration tests per trench ranged from 2 to 7. For each trench, infiltration was determined on at least one riser and one tread, for a total of 21 risers (9 upper risers, 8 mid risers, and 4 lower risers) and 19 treads.

2.2. Runoff and Erosion

[15] At one trench location, seven of the runoff plots established for the infiltration studies were subsequently used for monitoring of naturally occurring runoff and erosion. Between 1992 and 1995, rainfall (measured with a manual rainfall gauge) and runoff were monitored after each rainfall event, for 804 days. If runoff did occur, a 100-ml sample was collected for determination of the sediment concentration.

2.3. Soil Matric Potential

[16] Eight sites—six risers and two treads—were instrumented at various depths with ceramic cup tensiometers to measure soil matric potentials from 0 to −900 cm of water tension. The tensiometers were inserted horizontally 20–25 cm into trench side walls, sealed with bentonite, and brought to the surface with vertical PVC connector tubes of the same diameter. Black plastic sheeting was then placed between the side walls of the trench and the backfilled materials to prevent extraneous hydrological effects from the trench excavation. Field tensiometer values were corrected for the length of the installed tensiometer (total of depth into soil and freeboard length) to obtain adjusted soil moisture matric potentials. The deepest tensiometer was placed near the interface of the soil and the underlying indurated bedrock. Data were collected weekly (until soil matric potentials exceeded >−900 cm water tension) and after all rainfall events for about three years (1992–1995).

3. Results

[17] On average, the infiltration rates of the soils on the upper risers were about double those of the tread soils, in spite of the much steeper slopes of the risers (Figure 2). Infiltration rates declined progressively from upper-riser soils to lower-riser soils (those of the lower risers being comparable with those of the treads). At the same time, infiltration rates within each of the topographic positions varied considerably, most likely because of individual differences in soil thickness, vegetative cover, and surface conditions.

Details are in the caption following the image
Box and whisker plots of infiltration rate, soil depth, and percent slope at various topographic positions along the riser–tread continuum. With sufficient data a box and whisker diagram presents the median and mean (red line); and the 5% (lower dot), 10% (lower whisker), 25% (lower edge of box), 75% (upper edge of box), 90% (upper whisker), and 95% (upper dot) distribution of the data.

[18] Soil depths (including soil/caliche), as highlighted in Figure 2, declined progressively down the face of the riser. The average soil depth on the treads was only about one quarter or less that on the upper riser.

[19] The data obtained from the runoff monitoring are consistent with patterns observed from the infiltration experiments (Table 1). Runoff as a percentage of precipitation increased with distance downslope, from upper riser to lower riser (runoff from the lower risers being comparable with that from the treads). Marked differences occurred between the two tread plots: the one consisting mostly of bare ground and rock produced higher amounts of runoff while the one having good vegetative cover produced much less. Erosion was quite low from the upper and mid risers but was very high from the lower risers.

Table 1. Soil and Hydrological Characteristics for a Sequence of 7 Runoff Plots Positioned Along a Topographic Gradient of Risers and Treadsa
Characteristic Topographic Position
Upper Riser Mid Riser Mid Riser Lower Riser Lower Riser Tread Tread
Runoff (%) 2 11 28 32 24 33 12
Erosion (g/m2) 2 8 58 364 306 98 56
Infiltration capacity (cm/hr) 14.8 11.0 7.7 2.9 4.7 0.0 7.7
Slope (%) 26 38 36 16 26 4 6
Soil Depth (cm) 66 97 91 41 66 10 15
  • a Runoff and erosion were monitored for an 804-day period between 1992 and 1995. Rainfall during this period totaled 1625 mm.

[20] Figure 3 presents corrected soil matric potential data and associated precipitation data as a time series (weekly averages) for three of the riser sites. These data are consistent in that all three indicate saturated or very moist conditions at the base of the riser soil for two extended periods (6 months or more). The relatively high positive pressures at sites 4 and 5 suggest that either water is confined and under artesian pressure on a given step, or that there is connectivity with water at a higher elevation. The latter is less probable because positive pressures range only from 0.5–2m in the hillslope sequence of risers that have elevation differences of 28m (Figure 3).

Details are in the caption following the image
Soil matric potential at various depths at three riser sites (measurements taken from June 1992 to June 1995). Data are presented only for observations when soil matric potentials <−900 cm of water tension. The hillslope sequence of risers have decreasing elevations from 281 m (riser 4) to 261 m (riser 5), to 253 m (riser 6).

[21] Results from all eight tensiometer sites are summarized in Table 2. At four of the six riser sites, the deepest tensiometer recorded saturated conditions for more than 25% of the time; and at the other two riser sites, soils were near saturation for at least a comparable percentage of the time. According to the 150-year meteorological record of annual rainfall 1992 was one of the wettest years (1170 mm), 1993 one of the driest (673 mm), and 1994 (1045 mm) and 1995 (863 mm) were at or above the long-term mean of 840 mm for Austin, TX.

Table 2. Topographic Position, Soil Depth, and Water State Classes for Each of the Tensiometer Sitesa
Site No. Topographic Position Soil Depth, cm Time Saturated, % Time Moderately Moist and Very Moist, % Time Slightly Moist and Dry, %
1 T1 Tread 38 4 42 54
3 T4 Tread 30 16 10 75
2 T1 Riser 38 2 46 52
2 T2 Riser 76 3 45 52
3 T1 Riser 43 4 43 53
3 T2 Riser 94 28 17 55
4 T1 Riser 36 23 24 53
4 T2 Riser 66 28 19 53
5 T1 Riser 61 2 45 53
5 T2 Riser 112 9 36 55
5 T3 Riser 145 24 23 53
6 T1 Riser 36 1 44 55
6 T2 Riser 69 7 38 55
6 T3 Riser 91 4 43 54
7 T1 Riser 30 10 26 64
7 T3 Riser 152 28 7 65
7 T4 Riser 257 31 6 64
  • a Data from Soil Survey Division Staff [1993]. All data based on corrected tensiometer values reflect saturation (0 or + matric potential values), moderately moist and very moist (−1 to −899 cm water), and slightly moist and dry (>−900 cm water) conditions.

4. Discussion and Conclusions

[22] The predictable and systematic differences in soil hydrology along the riser–tread have clear implications for our understanding of both runoff generation and ecohydrological relationships. We find that the soils on the steepest slopes have the highest infiltration rates; and that in general, there is a progressive decline in infiltration rates from the upper risers downward—the upper risers commonly exhibiting high or very high infiltration rates and the lower risers and treads relatively lower values. Further, we find a similar dramatic and progressive change in soil depth, and therefore water-holding capacity, from riser to tread.

[23] Long-term monitoring of soil matric potential shows that wet or saturated conditions persist for significant periods of time at the base of the risers, which indicates that (1) water is entering the riser; (2) water does move through the Bk horizon (often referred to as caliche); and (3) very little if any vertical flow penetrates the bedrock–soil interface. Some of the tensiometers recorded relatively high hydrostatic head, which indicates localized artesian conditions from water upslope (probably within the same riser sequence.

[24] With respect to runoff generation, these hillslopes are composed of an alternating sequence of sources and sinks that facilitate both surface and subsurface flow. The data from the runoff plots support the notion that most of the surface runoff generated in this landscape comes from the lower risers and the treads. Locally, it appears that the downslope risers are serving as sinks (recharge) and the upslope treads as hydrological sources (discharge). During periods of high rainfall, overland flow is generated from the lower risers and treads, which is then captured and stored and ultimately used by vegetation or slowly released by “seeps” at the base of the riser. During the observation period, around one-third of the rainfall ran off the lower riser and tread plots. We believe it unlikely that the ephemeral multiple water tables within the risers are interconnected (see above comments), but more work is required to confirm this.

[25] Our results have at least three major implications for resource management. First, largely because of the high infiltration rates and significant water-holding capacities, the risers serve as important buffers to storm water flow. Water that would otherwise immediately run off is instead stored on the landscape. This buffering mechanism can be lost if the risers are covered or paved. Second, the “caliche” residuum (Bk and Bkm horizons) on the risers is likely an important reservoir of water for deeper-rooted vegetation as well as the source of water for the hillslope seeps often noted at the base of risers. Finally, vertical recharge of deeper groundwater is probably limited on these hillslopes, and when it does occur is restricted to fractured bedrock or, more likely, to valley bottoms.


[26] Gratitude is expressed to Bob Ayers, Barton Creek Properties and Stratus Properties for supporting some of this research. Appreciation is also expressed to USDA-Natural Resource Conservation Service and the Texas Commission on Environmental Quality for additional support.