The punchbowls of southwest Manitoba

Thomas E. Barchyn
University of Calgary
Calgary, Alberta, Canada.

On the banks of the Assiniboine River in southwestern Manitoba, amphitheatre-like features 100s of meters wide are actively advancing into the deep sand delta deposits of Glacial Lake Agassiz. These spectacular landforms are locally known as ‘punchbowls’. The activity of these features belies their relatively geomorphically sleepy surroundings; the newest punchbowl has expanded to a diameter of ~125 m within the last decade. Punchbowls are clearly hazardous here – large portions of the prairie are being eaten at a rate far beyond adjacent river banks. These are spectacular and mysterious features that hint at powerful subsurface forces and capture visitors’ imaginations. Here, I take a brief look at the geomorphology of punchbowls.

Punchbowl image

Figure 1: Panorama of the newest punchbowl. Photo: Thomas Barchyn.

What is a punchbowl?
Punchbowls are found south of Carberry, Manitoba (Figure 2). The features vary in size and shape, but all have an amphitheatre ‘head’ eroding into the bank away from the river. The well-known Devil’s Punchbowl is perhaps the most famous example (red line on Figure 1); however, the banks of the Devil’s Punchbowl are vegetating and a small lake has formed in the bottom of the erosional head, possibly indicating some phase of dormancy.

Figure 2: Punchbowls are clearly evident along the banks of the Assiniboine River. The well visited Devil’s punchbowl is marked with a red line. The new punchbowl is marked with position in 2011 (black) and 2015 (orange). Other features are less straightforward to map from imagery, but nonetheless hint that punchbowls are present elsewhere (blue lines).

Figure 3: Advance in the new punchbowl from 2011 to 2015. Note the differences in stream pattern. This punchbowl was surveyed Summer 2015 with a UAV.

Figure 3: Advance in the new punchbowl from 2011 to 2015. Note the differences in stream pattern. This punchbowl was surveyed Summer 2015 with a UAV.


Figure 4: An oblique view of the new punchbowl in 2015, looking NW from the SE of the feature.

Figure 4: An oblique view of the new punchbowl in 2015, looking NW from the SE of the feature.

The punchbowls cut into the ~30 m deep sandy deltaic deposits of the Assiniboine Delta of Glacial Lake Agassiz. The surface of the delta has been reworked into dunes of the Brandon Sand Hills, in this location also known as the Bald Head Hills or Spirit Sands. However, the aeolian rework is only a thin veneer in this portion of the dune field. The exposed deltaic sediments are dominantly sand, with coarser gravel and pebble strata. Exposed adjacent to the river bank is a hard clay, presumably glaciolacustrine and deposited distally to the delta. A conformable contact with deltaic sands occurs ~5 m above river level. There are large boulders and cobbles around, but the source is not clear (possible dropstones?). The sands host a large and important aquifer. In the dunes of Spruce Woods Provincial Park, the water table is within meters of the surface (~ 2 km away), suggesting an aquifer ~ 30 m thick, perched above the river level on a sub-delta clay.

At the bottom of the punchbowls are streams which are draining the aquifer and evacuating sediment from the punchbowl into the river, leading to further erosion. The slopes of the punchbowl are close to the angle of repose for sand, but the upper slopes are steep and periodically collapse with oversteepening.

How are punchbowls formed?
The basics of punchbowls mechanics must involve both hydrogeology and geomorphology. While the literature on ‘punchbowls’ is sparse, what is referred to here as a punchbowl is most likely a version of a feature referred to as a ‘seepage erosion valley’. Seepage erosion valleys occur where the head of a drainage network forms an amphitheatre bowl and is sourced from high volume ground water seeps (Abrams et al., 2009). Similar amphitheatre head valleys occur in Colorado, Hawaii, Chile, Idaho, Florida, and Mars (see Abrams et al., 2009; Marra et al., 2014). This noted, the punchbowls of Manitoba are undeniably a spectacular example.

The key criterion to develop punchbowls is an aquifer which can produce water at a rate sufficient to erode and evacuate sediment, creating a feedback that allows the punchbowl to continue advancing. The punchbowls here are generally short, but the same feedback of erosion and sediment evacuation in seepage erosion valleys can form entire valley networks (e.g., Abrams et al., 2009). As the seepage erosion valleys cut away from the river channel they move into virgin aquifer with large potentiometric gradients, often increasing the rate of water seepage, erosion, and feature advance.

It is not entirely clear what causes a punchbowl to relax erosion rates and vegetate (e.g., the Devil’s Punchbowl). Given that similar amphitheatre features elsewhere (e.g., Apalachicola River on the Florida panhandle, Abrams et al., 2009) are vegetated, yet still actively migrate, it is perhaps premature to label vegetated punchbowls as relict. The newest punchbowl (see Figures 2, 3) is actively eroding trees, prairie, and anything else in its path. However it is not clear whether youth and aggressiveness correlate, it is possible that more elderly punchbowls (e.g., the Devil’s Punchbowl) become more sedentary with time for some reason. However, it is equally possible that the newest punchbowl is simply a different geometry that is hosting higher potentiometric gradients and greater water flow.

What is clear though is that punchbowls are begging for further study. Punchbowls demonstrate that awesome geomorphology is not confined to the Rockies – rapid and exciting landscape change is found even in the flatlands of Manitoba.


Figure 3: CGRG-GCRG 2015 president Prof. Chris Hugenholtz gets a closer look at the delta sands. This photo does not imply this is a safe place to go. Photo: Thomas Barchyn.

Figure 3: Delta sands exposed along the ~30 m tall scarp of the newer punchbowl. Photo: Thomas Barchyn.


Note: Please contact Spruce Woods Provincial Park if you would like to visit the region. The photos in this post do not suggest that it is safe to get near these features!

Further reading:
Abrams, D.M. et al., 2009. Growth laws for channel networks incised by groundwater flow. Nature Geoscience 2, 193-196.

Marra, W.A., et al., 2014. Valley formation by groundwater seepage, pressurized groundwater outbursts and crater-lake overflow in flume experiments with implications for Mars. Icarus 232, 97-117.

Post Your Thoughts