by Philip S. Prince

Well, actually 98 years, but close enough…

Northwest of Hamilton Knob in Wythe County, Virginia, two narrow outcrop bands of Cambrian dolomite are entirely surrounded by Devonian-aged shales and sandstones. These dolomite bodies are tiny klippen that are separated from the Devonian units by the Pulaski Thrust Fault, which has been folded into two anticlines separated by a narrow syncline in which the klippen occur. I mapped in this area during the last year and a half, and I think this might be the most extreme example of folding of the Pulaski Fault system within the Valley and Ridge. In the cross section below, the large black arrow points to one of the klippen. The curving black line is a projected estimate of the folded geometry of the Pulaski Fault prior to erosion to present-day levels.

The GIF below shows the approximate line of section above and points out two of the “micro-klippen.” The klippe crossed by the section line is just large enough to impact land use, allowing it to be seen as a brown, cleared area surrounded by the forested, acidic Devonian shale/sandstone landscape. The smaller klippe is too small to impact vegetation or land use.

These klippen are made even more interesting by the fact that they were mapped and correctly interpreted in 1924! The image below compares a 1924 map by Campbell, Howell, Holden, and Kimball to my 2021 mapping, which was produced entirely through new fieldwork and lidar analysis. M, D, S, O, and C denote Mississippian, Devonian, Silurian, Ordovician, and Cambrian outcrop zones, respectively. The curving black lines in the 1924 map are the Pulaski Thrust, which correctly encloses a narrow sliver of Cambrian rock (peach) between two extensive Devonian (dark green) outcrop fields (south of “Creek” in the Beaverdam Creek label and the 45 dip measurement). Comparing these maps separated by 98 years and considerable changes in land use, geologic exposure, tectonic understanding, and imaging technology is very interesting to me. The same patterns and structural relationships are obviously present in both to a high level of detail, which is impressive given the much larger scale of the 1924 map project. The map is part of a massive, regional overview of coal resources in the Virginia Valley and Ridge, which is why I never took the time to check it out until now. This “Valley Coalfield” report is linked here; the map comes at the very end of the 381-page document.

Identification and mapping of the distinct rock in the klippen in 1924 is not terribly surprising. Due to land use at the time, vegetative cover was much less extensive and outcrop was more visible. Iron ore also occurs along carbonate-Devonian shale contacts in the area, so special attention was paid when those rock types were expected in an area. The interpretation of the klippe as the result of folding of a thrust fault, however, is quite interesting in light of understanding (or lack thereof) of fold-thrust belts at the time. I am impressed that this intensely folded thrust model arose out of the prevailing theories about the nature of thrust faulting at the time, but it provides just about the only way to explain the architecture of the rocks in the area. The cross sections below compare the 1924 interpretation (line of section is the straight black line in the 1924 map) with my 2021 interpretation, with colors adjusted to match. The 1924 section was developed slightly north of mine, but the overall context of the klippe related to adjacent fold structures is the same. Black arrows point out the Cambrian micro-klippe in each section.

In 1924 and today, a key component of understanding these micro-klippen is explaining the tightness of the syncline that preserves them. After my work in the area, I think it is largely attributable to mechanical contrasts in the rock mass. The dark green horizons in the sections above are shale-rich and weak compared to the overlying light green Mississippian sandstones and underlying Siluro-Devonian sandstones (light brown). The base of the Pulaski Thrust Sheet itself (pale orange) is probably mechanically weak as well, consisting of thin-bedded carbonate which is often extensively brecciated. As a result, the forelimb of the overall Draper Mountain anticline was able to tightly buckle and nearly enclose a tiny pod of the Pulaski Sheet carbonates. This behavior would not have occurred with a different mechanical combination of strata. Campbell et al. had a generally similar idea, and explain the concept very nicely in pages 83-84 of the Valley Coalfield text. “Engulf” is indeed a nice word to describe their model.

I used the conceptual model below (it’s a very slow moving GIF…wait for it) for my own explanation of the micro-klippen. Early broad folding of the Pulaski and Max Meadows Thrust Sheets occurs when the Draper Block moves onto an underlying flat, probably developed in middle Devonian shale. The future micro-klippen are rotated forward on the forelimb of the block. In this geometry, weak units on both limbs of the Draper Block are poised to slip on or close to bedding planes, setting the stage for a localized “pop-down” syncline in the zone of weak, shale-rich interbeds. This model requires relatively little displacement on the out-of-sequence faults due to the pre-existing ramp anticline geometry, as well as the position of the future micro-klippen on the upper Devonian-Mississippian ramp. Interestingly, Campbell et al. draw the klippen in this ramp position as well, suggesting we placed formation boundaries on similar marker beds.

So, has re-mapping this area offered any improvement on a such a good but 98-year old product? Yes. New mapping is spatially much better; the 1924 map cannot be draped onto an existing surface map with decent accuracy. Subsurface interpretations are also better, applying more rigorous geometric controls obtained by outcrop mapping and detailed lidar topography. Details of the Pulaski Thrust Sheet and particularly its immediate footwall are also better in the newer map, due to its smaller spatial scale and (again) the ability to track individual beds using lidar. The present map results also have much more context within the overall study of fold-thrust belt evolution, both in the Appalachians and in other systems around the world. Even so, the 1924 product is surprisingly high-quality and interpretively ahead of its time. I will actually be using it in my current mapping to the north of this area, as it shows in great detail where coal beds are cut of by thrust faults that are extremely difficult to identify in the field or with lidar.

Beyond the scientific details, I like the 1924 map because its details capture specific challenges of working in the area that persist today. My favorite is marking of the “Red shale band in Watauga” with a dashed line, running just left of, then below, the center of the 1924 map image below. The Watauga Formation, today known as the Rome Formation, is intensely folded and stratigraphically baffling in this area. Determining whether you are mapping near its base or top feels impossible, as marker beds are few and far between. Campbell and his associates obviously jumped on the red shale band in hopes of using it as some sort of recognizable frame of reference within the confusing sea of folds. I did exactly the same thing, as the red shale band is very recognizable in lidar-derived imagery as well as in the field (lower image). It can be tracked over significant distances and does indeed highlight some larger folds within the Rome Formation, some of which help define contacts with the overlying Elbrook Formation as we map it today.

by Philip S. Prince

Topographic superlatives are almost always a bit arbitrary, and comparing river steepness is about as arbitrary as it gets. How long of a stretch of river needs to be considered? How big does a stream need to be in order to be considered a “river?” Since a vertical waterfall is the maximum steepness possible, is the biggest river with a freefalling waterfall the winner? How tall does the waterfall have to be? These questions all have merit, and each one can steer this search in a slightly different direction. One reach of Appalachian river, however, stands out no matter which of these parameters is applied: the gorge section of the Tallulah River of northeast Georgia.

At the upper end of Tallulah Gorge, the Tallulah River drops about 400 ft (120 m) in 1 mile (1.6 km) or so (some sources say 500 ft over this distance). Much of this descent occurs across four huge waterfalls, which are easily visible in Google Earth…and which seem precariously close to US Highway 441. Oceana is the only falls directly accessible to hikers and, in April and November of each year, to whitewater paddlers. Hikers can cross a bridge over the top of Hurricane and approach its base, but the upper two falls can only be experienced from a distance due to their extreme ruggedness. Vertical rock walls are present in much of the upper gorge.

This impressive sequence of falls at the head of the gorge occurs where the Tallulah’s drainage area is ~190 square miles, or 500 square kilometers. When presented as a highly vertically exaggerated stream profile, shown below, the short but extremely steep stretch of river looks almost like a dam, and absolutely dwarfs every steep zone in the neighboring Chattooga River, a slightly larger river which the Tallulah meets just below its gorge. Not all of the Chattooga is shown here, but there isn’t anything Tallulah Gorge-like upstream of what is shown.

As far as I can tell, no other Appalachian river (at least in unglaciated areas) sustains a comparable gradient over 1 mile at a similar drainage area. Plenty of larger rivers have single waterfalls or sets of rapids (an example at the end of this post), but they do not approach the Tallulah’s sustained 1-mile gradient. Likewise, plenty of smaller named rivers descend more rugged topography and probably locally exceed the Tallulah’s gradient, but they are MUCH smaller rivers. The nearby Whitewater (Ww), Thompson (T), Horsepasture (Hp), and Toxaway (Tox) River basins are outlined below for comparison. Their combined drainage area would easily fit within the Tallulah basin.

The drainage area and gradient of Tallulah’s steep zone equate to impressive stream power that is not reflected in neighboring river systems. In many parts of the world, a large and atypically steep river like the Tallulah might be attributed to surface uplift due to active faults or landscape response to glacially-sculpted topography. Both of these factors are off the table in the Tallulah area. So, why is the Tallulah so big and steep? There are two reasons, and both become apparent with the aid of digital topography and geologic maps.

First and foremost, the Tallulah drainage basin has not always flowed to the Atlantic Ocean, a concept first noted by Johnson (1907). The Chattooga and Tallulah Rivers are the former headwaters of the Chattahoochee River system, which now drains the elevated plateau (yellow and brown) at lower left in the image below.

The old Chattachoochee headwaters loomed above the neighboring lower elevation stream networks of the Atlantic-draining Piedmont (green), creating a highly asymmetric drainage divide. Asymmetric divides are unstable over time, and headwaters of the Savannah system ate their way into the plateau edge, capturing the Chattooga and Tallulah and routing them off of the plateau into the lower elevation Savannah system. The GIF below animates the general idea of this process, which can otherwise be tough to visualize.

This stream capture event gave the Tallulah and Chattooga an additional ~660 ft (200 m) to descend, causing the rivers and their tributaries to begin carving deep gorges, which gradually ate their way upstream. The Chattooga (and any of its tributaries), however, don’t host any features like Tallulah’s steep stretch, which highlights Tallulah’s most significant aspect: its bedrock.

Tallulah Gorge is developed into quartzite, a metamorphic rock composed primarily of quartz (it’s not just a clever name). The quartzite, which is exposed within the Tallulah Falls Dome structure, is significantly harder/tougher/less erodible than surrounding rock types, and the Tallulah crosses plenty of it on its way to the Chattooga. In the GIF below drapes a great geologic map (Thigpen and Hatcher, 2009) over the Tallulah basin. Areas of quartzite exposure are yellow, and are widespread within the geologic dome. This map does a great job of showing the regionally atypical geology crossed by the Tallulah.

The quartzite bedrock is so resistant to stream erosion that the Tallulah has been unable to keep up with the Chattooga’s progressive downcutting in response to the stream capture. While the lower Chattooga has evolved towards a smoother transition into the rest of the Savannah system, the Tallulah has been comparatively unable to cut into its own distinct bedrock to match the Chattooga. As a result, much of the steepening related to the capture event is still “stuck” at the upper end of Tallulah Gorge, where the river is working hard to cut through the quartzite to allow the Tallulah to better match the Chattooga’s channel elevation.

This hard work takes the form of the large waterfalls, particularly the upper three falls, where the river is at its narrowest and its flow most concentrated. These falls represent the river concentrating its erosional energy as much as possible to try to catch up, erosionally speaking, to the Chattooga and the rest of the Savannah system.

The Tallulah’s work is made even more challenging by the downstream dip, or tilt, of the quartzite layers, which is particularly noticeable at Oceana Falls. I wore a GoPro while paddling in the gorge last November to film the short video linked here. Oceana is the first rapid shown (I run it twice), and downstream-dipping quartzite horizons are clearly visible to my left. Because of this orientation, the river is less able to access weaker horizons under the quartzite, which would cause undercutting and collapse of the hard layers. The block diagram below shows one interpretation (Hatcher, 1973) of structure beneath Tallulah Gorge. The geology shown here is very large-scale when compared to the size and depth of the gorge itself, but it does give an overall sense of structure. and the position of the gorge with respect to the only slightly exposed quartzite layers. Tallulah Gorge is presently located where the yellow layer starts to tilt down towards the right side of the diagram.

The quartzite (yellow, again) is not particularly thick. It is, however, in just the right place, and is dipping just the right way, to obstruct the Tallulah’s incision and force the river to become so atypically steep, despite its size and the relatively modest elevations of the surrounding landscape.

At the end of the day, the quartzite bedrock is really what sets the Tallulah apart and controls its unique steepness. The Tallulah’s valleys upstream of the gorge are notably higher than neighboring Chattahoochee valleys, suggesting a gorge and steep stretch of river has always been present where the Tallulah steps down from the quartzite into the surrounding landscape; the capture event has simply exaggerated this step. Without the quartzite, the Tallulah landscape would look much more like the neighboring Chattooga system, which drains schist and gneiss bedrock.

Quartzite always does interesting things in the Appalachian landscape, and the Tallulah is not the only river that interacts with this particularly tough rock type. Another noteworthy quartzite-crossing river is the Linville River, which is another stream capture victim. While the Linville descends farther than the Tallulah, it never matches the Tallulah’s one-mile maximum gradient. The Linville also drains a smaller area into the steepest portion of its gorge. Even so, the presence of quartzite within Linville Gorge absolutely impacts the length of the river’s steep stretch, presumably by slowing its downcutting and adjustment in a similar fashion to the Tallulah scenario.

To repeat this post’s opening disclaimer, the Tallulah is noteworthy amongst rivers its size, but comparing it to significantly larger streams flowing over different bedrock is still an apples-and-oranges game to some degree. A river feature that has always impressed me is Cumberland Falls on the Cumberland River in Kentucky. This is a single, vertical waterfall is about 70 feet (22 m) high and occurs at around 2,000 square miles (5,200 square kilometers) drainage area. The image below shows the location of Cumberland Falls (red star) and the upstream drainage area, which dwarfs Tallulah (yellow star near the bottom).

Cumberland Falls is developed on sandstone overlying weaker shale, and is likely migrating upstream very quickly, geologically speaking. It is also the only feature of its type on the river, so the 1-mile scale steepness is nowhere near that of the much smaller Tallulah system. A direct comparison between Cumberland and Tallulah is complicated by completely different geologies and river system settings, but both are intriguing features that reflect geologic details of the host area.

The Tallulah River and its unique character are yet to be a significant presence in scientific literature, but I expect this to change in coming years. The gorge is dramatic in person, and its connection to geologic details makes it an equally impressive example of factors that control landscape appearance on Earth’s surface. I recently ran across a great Chattooga Conservancy website that gives some Tallulah history, including the great 1841 painting shown below. The painting appears to show the upper two falls and Hurricane, which is partly obscured at lower right. The faintly tan color used for the quartzite cliffs is a good representation of their actual appearance.

According to the Chattooga Conservancy site, an author named David Hillhouse wrote in 1819 that the falls of Tallulah Gorge were surpassed only by Niagara amongst rivers he had seen. I would say that there is geologic and topographic evidence to support his opinion.

by Philip S. Prince

Lidar-derived imagery is a game-changer in the study of landslides and other slope movements, and it frequently produces dramatic visuals like the Rutherford County, North Carolina, debris slide shown below. The slide has developed in talus dominated by sillimanite schist, which is likely sliding atop underlying gneissic bedrock.

Over the past year of focused landslide mapping here in southern Appalachia, I have noted that slides that are this crisply defined in lidar hillshade are often found to be active upon field inspection, with soil-exposing scarps, ground ruptures and fissures, and trees back-rotating or being pushed over. The (very) humid temperate climate of the region is not well suited to preserving the fine details of many landslides, particularly when they do not involve intact bedrock masses. Numerous kinematic features are beautifully defined on the slide, with detail on the slide’s toe and laterals being particularly impressive and comparable with confirmed active slides in the area. The slide is equally dramatic using slope-shade imagery, which can offset bias effects related to lighting angle in hillshade imagery.

I recently visited this slide as part of my work with Appalachian Landslide Consultants (ALC), PLLC, and was surprised by its actual field expression. The following photographs show what the slide looks like in the field. The accompanying hillshade images show where each photograph was taken, with the points of the yellow arrows resting on the photographer’s position (this is to avoid covering the details of the slide with the arrow). Each arrow points in the direction the photographer was looking.

Surprisingly, no bare soil was exposed on the scarps of the slide. The headscarp is shown below. Up and right from the center of the photograph below, ALC geologist Aras Mann is seated on the scarp surface. Trees show minimal deformation here.

Many of the trees growing on the headscarp are very young, suggesting its re-vegetation may be quite recent. Even so, it is presently entirely vegetated with an O-horizon present everywhere.

The toe of the slide is impressively defined in lidar hillshade, but it is much more subtle from the ground surface. In the photograph below, Aras contemplates the surprising field appearance of the toe. The dashed yellow lines highlight the faint step on the land surface. The few leaning trees behind Aras are located just beyond the tip of the yellow arrow, where a portion of the toe appears to have slumped slightly due to oversteepening.

The toe bulge is certainly large enough to be noticed, particularly if one already knows that it is there. The bulge is about 6 feet (1.85 m) tall where it is crossed by the logging road. Note the two dead (or dying) trees at the right of the image…could they be evidence of creep within recent years?

Notably, the logging road does not appear to have been deformed by any noticeable slide movement.

Looking back upslope toward the toe from a logging road further downslope, only a faint step in the landscape is visible. The photograph below was taken from the logging road (point of the arrow). Again, the yellow dashed lines highlight the step.

The outermost portion of the toe (lower right of the slide in the images) has slumped due to oversteepening. The resulting headscarp is visible in the landscape in the photograph below (dashed yellow lines again). Some fallen trees were present here, but I did not consider them obviously attributable to slide movement.

The transition from toe to left-lateral scarp was also surprisingly subtle. Again, the toe bulge was large enough to be noticeable, but the crisp lateral visible in lidar hillshade was mostly obstructed by tree growth.

Interestingly, a similar pattern of dramatic lidar appearance and subtle field expression was observed elsewhere in the area. About 1,000 feet (300 m) northeast, another eye-catching (in lidar, at least) debris slide was found to be similarly vegetated. The slide described above is located down and left of center in the image below; its neighbor is pointed out just right of center.

This is another debris slide with crisp scarps and interesting patterns related to extension in the interior of the slide, near the center of the image.

Field investigation revealed another completely forested slide with no clear evidence of recent or ongoing movement. Older trees on the steepest portions of scarps showed some curvature, but, again, this slide looked nothing like what we expected. The slide does maintain several closed sags and depressions, although they too are completely forested. In the photograph below, Aras is standing in one such depression just below the western headscarp of the slide.

The slope which hosts these two slides is home to several other debris slides, all developed in the schist talus. The schist is well foliated, with fragments maintaining surprisingly planar and smooth foliation surfaces which easily slide past each other and along the underlying bedrock. Other slides are pointed out in the GIF below.

Many are equally well defined in lidar, and are similarly vegetated and unremarkable in the field. We have no constraints on the age of the slides, but they may reflect logging history in the area. The majority of these slopes were heavily and continuously logged during the past ~150 years, with logging in this area clearly occurring within the past 50 years. The slides may have developed after clear-cuts, with the rapid return of vegetation common in the region quickly making the area look less disturbed than it really is.

(DEM imagery captured from QL1 lidar data acquired by NC Dept. of Public Safety, raw point clouds assembled and processed by Corey Scheip of the NC Geological Survey Landslide Hazards Program)