ORIGIN OF THEATER-HEADED TRIBUTARIES TO ESCALALANTE AND GLEN CANYONS, UTAH. R. P. Irwin III1, C. M. Fortezzo2, S. E. Tooth3, A. D. Howard4, J. R. Zimbelman1, C. J. Barnhart5, A. J. Benthem4, C. C. Brown, and R. A. Parsons5. 1Center for Earth and Planetary Studies, National Air and Space Museum, Smith-sonian Institution, MRC 315, 6th St. at Independence Ave. SW, Washington DC 20013-7012, irwinr@si.edu, zim-belmanj@si.edu. 2U. S. Geological Survey, Flagstaff AZ 86001, cfortezzo@usgs.gov. 3Aberystwyth University, Ceredigion, SY23 3DB, U.K., set@aber.ac.uk. 4Department of Environmental Sciences, University of Virginia, Charlottesville VA 22904, alanh@virginia.edu, ajbjc@virginia.edu. 5Department of Earth and Planetary Sciences, University of California, Santa Cruz CA 95064, barnhart@pmc.ucsc.edu, rparsons@pmc.ucsc.edu.

Introduction: Theater-headed tributary canyons

to the Escalante and Colorado Rivers in the Glen Can-yon National Recreation Area, Utah, have been cited as important analogs to fluvial valleys on Mars [1–4] (Fig. 1). The morphological similarity between these canyons and Martian valley networks, combined with difficulty modeling an atmosphere that would support precipitation runoff [5,6] on early Mars and spectral evidence of unweathered basalt in the cratered and dissected highlands [7,8], suggested to some investiga-tors that subsurface water cycling may have been ade-quate to form and supply the Martian valleys [9,10]. These terrestrial analog canyons have overhanging headwalls up to 200 m high, near-vertical sidewalls, and roughly constant widths from the headwall to the junction with the trunk river (Fig. 1). Alcoves are common in the headwall and sidewalls, and they con-tain active seeps or evidence of salt weathering (Fig. 2). These tributaries are incised into the lower Jurassic Navajo Formation, a cross-bedded aeolian sandstone with thin lenses of impermeable finer-grained sedi-ments. The canyon floors are interbedded fluvial sandstones and shales of the underlying Kayenta For-mation [3]. The study area occurs on the southwest-dipping limb of Waterpocket Fold (strike 131° dip 1.4° was surveyed on the Navajo/Kayenta contact in East Bowns Canyon). Construction of the Glen Canyon Dam in 1967 flooded the main canyon and the lower Escalante River with Lake Powell, which was 33 m below full level during our first visit (March, 2008).

Laity and Malin [1,2] conducted a field study of this area in the early 1980s, concluding that these tributary canyons primarily result from seepage weath-ering (several processes of chemical and physical breakdown) at the base of the headwall and seepage erosion (transport of the resulting debris by the small spring-fed stream), collectively termed groundwater sapping. They found that runoff from the contributing watershed was of minor importance due to high per-meability of fractured Navajo sandstone.

In March and May of 2008, we visited these can-yons to re-evaluate the roles of seepage erosion, over-land flows, and stream flows in weathering and sedi-ment transport.

Fig. 1. Study area including Fence (F), Explorer (E), 1N/1S, West Bowns (WB), East Bowns (EB), and Long (L) Canyons.

Fig. 2. Headwall of Fence Canyon. Note the incised contributing stream, alcove, dark seep, plunge pool that is swept clear of debris, and vegetated floor. Re-lief is ~150 m.

SECOND WORKSHOP ON MARS VALLEY NETWORKS 43

Methods: We examined the valley floor, the base of the headwall, and the contributing plateau surface at Explorer Canyon, two branches of a smaller unnamed tributary immediately to its north (here called 1S and 1N) (all are tributaries to the Escalante River, which joins the Colorado in Glen Canyon/Lake Powell), and two branches of Bowns Canyon (which drains directly to Lake Powell). We also surveyed the Fence Canyon headwall from the plateau surface and small branches of Iceberg and Long Canyons from the canyon floor and plateau, respectively (Fig. 1). Collected data in-clude spring pH, hardness, and discharge (to evaluate the spring’s capacity to weather and erode the head-wall rock); compressive strength by Schmidt hammer [11] of Navajo and underlying Kayenta beds (to de-termine whether vertical strength differences might facilitate undercutting of the headwall); Selby [12] bulk strength of Navajo sandstone (primarily to evalu-ate the effect of joints on headwall strength); discharge estimates for flash floods and size of transported rocks (to compare the erosional capacity of floods with that of spring-fed streams); and vertical profiles of valley headwalls and alcoves using a laser rangefinder (to quantify morphology and evaluate factors that may influence alcove development).

Observations: Analysis of the collected data is underway, and here we report some general observa-tions and preliminary results.

Compressive strength and Selby index. We col-lected Schmidt hammer rebound (R) values on the Navajo Formation at the top and bottom of the canyon 1S, Explorer, East Bowns, and West Bowns headwalls, as well as the bottom of the canyon 1N headwall, sites on the plateau surface, and some Kayenta outcrops on the canyon floors. Navajo formation R-values are generally moderate (R~45, or ~50 N/mm2 compressive strength) on both wet and dry vertical scarps, but they are significantly lower and more variable on weathered plateau surfaces, which commonly have small sheeting joints within a few centimeters of the surface. There appears to be no significant difference in rock strength between the top and bottom of the Navajo section. Surfaces beneath loose, salt-weathered crusts in al-coves are also hard, suggesting that seepage weather-ing affects only the first cm or so of the surface and that most of the resulting small debris is easily broken down and transported by gravity or water [1,2]. Most detached blocks of Navajo sandstone are mechanically weak and break down readily upon movement by grav-ity or in water.

Selby values for rock strength in headwalls are high (~90) due to the combination of the moderate R-values, wide spacing and horizontal/vertical orienta-tions of joints (or absence thereof in many cases), dis-

continuity of joints, and small joint width. This strength supports high vertical scarps [12].

Headwalls. Headwalls typically include an alcove and a less-weathered, sheet-fractured face above it. If present, thin impermeable lenses or more permeable interbeds concentrate seepage at multiple levels in the section (Fig. 3). Alcoves were not observed in theater headwalls with large vertical joints oriented perpen-dicular to the face, but shallow sheeting joints are more common. Different weathering processes pre-dominate on different sections of the headwall relief. Adjacent to the base of the alcove, there is usually a shallow plunge pool (Fig. 2) formed by the waterfall from above (each observed headwall receives an ephemeral overland stream). The lower slope of the alcove contains loose debris and may be vegetated if there is spring discharge. Seeps issue from the most recessed areas of all of the observed headwall alcoves, but not all of the sidewall alcoves, which may have formed mostly during wetter pluvial epochs. There is no alcove at the 1N headwall; the spring emerges from the base and from a large vertical joint that does not extend to the plateau surface. The scarp is less weath-ered above alcoves, suggesting that stress-release frac-turing is required for headwall retreat in that zone.

Above the canyon headwalls, the contributing overland streams have often (but not always) incised narrow, V-shaped slot canyons, which funnel ephem-eral waterfalls over the headwall. The waterfall thus directly affects only a narrow area at the base of the alcove (Fig. 3).

Fig. 3. Narrow waterfall funneled over the East Bowns Canyon headwall following rain on May 22, 2008.

SECOND WORKSHOP ON MARS VALLEY NETWORKS 44

Springs, flash floods, and channels. Measured dis-charges of spring-fed streams varied between 0.001 and 0.002 m3/s. Previous flash floods have formed channels up to >10 m wide and >1 m deep and have transported imbricated gravels at least tens of cm in size (Fig. 4). Peak discharges estimated using the Manning relationship are ~1–10 m3/s near headwalls and three orders of magnitude higher than spring flows. During rainfall observed in East Bowns Can-yon from the evening of May 21 through the morning of the 22nd, runoff occurred almost immediately from slickrock surfaces bounding the canyon, whereas the headwall waterfall was not active until the following morning (Fig. 3), suggesting that storage in potholes in the contributing channel inhibits flow through the stream except in intense and/or prolonged storms.

Fig. 4. The largest imbricated rocks in the lower Bowns Canyon channel are >1 m in length.

Contributing streams on the plateau surface have ir-regular longitudinal profiles with frequent steep (but not vertical) knickpoints, which are dominantly more narrow and V-shaped in cross-section and have more closely spaced potholes than do low-gradient reaches (Fig. 5). Water remains in potholes long after over-land flow ceases, and weathered debris is flushed out by the next flood. This process facilitates downcutting of the contributing stream. There is a strong morpho-logical contrast between knickpoints affected only by overland flow and those with active seeps described above. Infiltration into the potholes and low-standing patches of algally laminated sands on the plateau may represent areally disproportionate sources of water for Navajo aquifers, as other surfaces yield more runoff.

Fig. 5. The entrenched contributing stream above the canyon 1S headwall.

On the valley floor, spring-fed streams are much

smaller than the channels that they occupy and play little role in sediment transport. The stream beds are dominantly exposed bedrock (Fig. 4) and/or covered with vegetation watered by the small stream, not floored with grain sizes that the base flow could trans-port (Fig. 6). In this way, spring flow inhibits rather than enhances sediment transport through the canyon, but it may play an important role in pre-weathering that facilitates erosion during floods.

Fig. 6. Channel in Explorer Canyon floored with vegetation and boulders.

SECOND WORKSHOP ON MARS VALLEY NETWORKS 45

Structural and topographic influences on valley planform. Large tectonic joints are uncommon to ab-sent in the studied valley headwalls, and stress-release fractures down-canyon are usually parallel to the walls and most evident in promontories. Smaller intersect-ing joints occur on some areas of the weathered pla-teau surface, but they are not thought to penetrate deeply, based on examination of scarps. Some of these joints are exposed on and influence incision of the contributing bedrock channels, and although this was uncommon over most of their length, the small joints may have a cumulative effect. Erosion in potholes and rock weathering appears to be more important.

The studied theater-headed tributaries to the Esca-lante River are growing up-dip, and the contributing watershed is a dip slope, which influences both their runoff and groundwater supply. However, Bowns and Long Canyons are growing along the structural strike, tracking up their entrenched contributing streams rather than the dip of the aquifers. The orientation of these canyons is predominantly topographic rather than structural. Some alcoves, particularly in West Bowns Canyon, are growing in the down-dip direction (westward).

Discussion: Our findings differ from the Laity and Malin [2] conclusions in two important respects. 1) Seepage erosion is a negligible process in sediment transport within the canyons, relative to flash floods, which are supplied by abundant runoff from slickrock surfaces. 2) Seepage weathering depends on flash floods to remove mass-wasted debris from the head-scarp, thus the canyons grow headward along the in-cised contributing streams and not necessarily up ma-jor tectonic fractures or the structural dip (except where the contributing watershed is a dip slope). For these reasons, although seepage weathering may be a contributory factor to the retreat of a vertical valley headwall, it does not operate without overland flow to remove the debris and enable further weathering and/or mass wasting. The rate of headward mass wast-ing is minimized when the headwall planform is rounded, maximizing its strength as it is undercut. The development of alcoves depends largely on the mas-sive, unjointed nature of the Navajo sandstone; other-wise the roof would collapse, and the waterfall could directly attack the base of the headscarp. It is un-known whether bedrock in the Martian highlands was competent enough for similar cavities to develop.

Observations suggest a hybrid model for origin of theater-headed valleys involving seepage weathering at the headwall base, transport of resulting debris and incision of the contributing and canyon-interior streams by flash floods, and important structural re-sponses to valley incision. Theater-headed valleys in

Navajo sandstone do not provide empirical support for models of valley network formation on Mars that lack contributions from surface runoff.

References: [1] Laity J. E. (1983) Phys. Geogr., 4, 103–125. [2] Laity J. E. and Malin M. C. (1985) Geol. Soc. Amer. Bull., 96, 203–217. [3] Howard A. D. et al. (eds.) Sapping Features of the Colorado Plateau, NASA Spec. Pub. 491. [4] Higgins C. G. and Coates D. R. (eds.) Groundwater Geomorphology, Geol. Soc. Amer. Spec. Pap. 252. [5] Squyres S. W. and Kasting J. F. (1994) Science, 265, 744–749. [6] Haberle R. M. (1998) JGR, 103, 28,467–28,479. [7] Christensen P. R. et al. (2001) JGR, 106, 23,82323,872. [8] Gaidos E. and Marion G. (2003) JGR, 108(E6), 5055, DOI:10.1029/2002JE002000. [9] Brakenridge G. R. (1990) JGR, 95, 17,289–17,308. [10] Clifford S. M. (1991) GRL, 18, 2055–2058. [11] Goudie A. S. (2006) Prog. Phys. Geogr., 30, 703–718. [12] Selby M. J. (1980) Zeitschrift fur Geomorphologie, 21, 1–13.

SECOND WORKSHOP ON MARS VALLEY NETWORKS 46