Introduction
Volcano-like, conical mounds of ice up to 5 m high form each winter along portions of the southern shore of Lake Erie and Lake Ontario. Particularly large and regular examples of ice volcanoesFootnote * form on the west side of Point Gratiot at Dunkirk, New York, when the lake is open, and a shelf of ice is attached to the shore.
When observed in active formation, eruptions of slush, ice and spray spouted two or more times the height of the cones. In some cases this was 10 m or more in the air. Flowing, sliding and rolling down the flanks of the cone, the ice was added rapidly. Accumulation of some of the larger cones is a matter of a few hours to a few days at most. At Dunkirk they occur in linear groups parallel to depth contours. In December 1970, there were three rows ranging from less than a meter in height near shore to 5 m in height 50 m offshore (Fig. 1). During mid-January 1972 a single row of irregularly spaced, poorly shaped ice volcanoes was built overnight about 50 m offshore.
The authors have observed the cones at several localities within a few miles of Dunkirk, on the west side of Presque Isle at Erie, Pennsylvania, and at a number of points along the Lake Ontario shore from Rochester to Oswego, New York. The cones have been reported by oral communications as far west as Ashtabula, Ohio.
Although the writers may have failed to note some earlier works, certainly there is no abundant literature on the subject. We should be grateful for any communications regarding their distribution, formation or previous studies.
Form and Structure
Viewed from shore, the ice volcanoes appear to be simple cones, even in spacing and form (Figs. 1 and 2). As shown in the map cross-sections (Fig. 2) and photographs (Figs. 3, 4 and 5), the lake side of these features is much more complex. Each of the conical mounds has a central channel opening toward the lake. The channels are separated by intervening head-lands which jut lakeward beyond the line of mounds. The headlands are topped by levees -which slope gently away from the lake margin. At Dunkirk, headlands stand 2–3 m and the cones 5 m or more above the adjacent lake ice.
The internal structure exposed in the channels and on the headlands is shown in the cross-sections (Fig. 2) and in Figure 4. Layering is shown by orientated tabular ice blocks and interbedded layers of clear, white and dirty granular ice (Fig. 4). Most of the tabular blocks lay parallel to the conical surface or parallel to the gently sloping shoreward face of the headlands and levees. Silt, sand, fish, wood fragments and shale bedrock fragments (Fig. 6) are also buried within the ice layers.
Origin of the Ice Volcanoes
In January 1972, the authors were able to observe the ice volcanoes forming at Dunkirk (Fig. 7). At the time they started to form, the air temperature was below freezing and there was floating slush and blocks of ice in the water. The wind was blowing onshore, from the west and north-west, at about 14 m/s. The cones in the first row were small (less than a meter) and irregularly shaped, and formed where the waves broke on the beach. As the wind subsided the water no longer splashed up over the cones. Instead, blocks of ice drifted into shallow water and became frozen as a shelf to the lake side of the first row of cones. A zone of shore-fast ice blocks, averaging 50 m wide, formed before the wind velocities again reached 14 m/s. At this time, a second line of cones began forming on the lake edge of the shore-fast ice. This second line of cones roughly paralleled the depth contours which are at a slight angle to the beach. The cones were spaced fairly regularly about 15 m apart and built to heights of 5 m overnight. Strong, onshore, wind-generated waves and above freezing air temperatures modified these cones by enlarging the channels and in some cases breaking completely through the shore side of the cones.
A third storm with onshore winds exceeding 28 m/s raised the level of eastern Lake Erie about 1.25 m or more and completely broke up the cones and the zone of shore-fast ice and piled everything on to the beach. A few irregularly shaped cones formed on the beach but in general the ice was simply piled up to heights of about 3 m. By the time the wind quieted, the water temperature was low enough that a zone of shore-fast ice formed out for a distance of about 100 m. A new line of cones was started but did not build very high before the ice completely closed in and prohibited further cone formation for the rest of the winter.
We conclude from our observations that the following factors are responsible for the formation of the ice volcanoes: (1) The air temperature must be below freezing while the water is mainly unfrozen, with only slush and a few floating ice blocks; (2) The wind direction must be onshore and the velocity greater than 11 m/s; (3) Some shore-fast ice must be present for the waves to break against.
As the wind-generated waves impinge on the lake-side edge of the shore-fast ice, slight irregularities in the ice edge tend to focus the wave energy into a smaller area. The result is that slush and ice blocks are tossed up around the indentation more than along the rest of the ice edge. While the concentrated wave energy tends to erode the indentation, shoreward, making it a deeper and more V-shaped channel, water and floating ice are thrown even higher around the eroded channel. As the water sprays up into the cold air, it either freezes before it falls or as it drains down the outer or shore-face slope of the resulting cone. Ice blocks and other debris thrown out of the channel are immediately frozen in place on the cone. The height of the cone is a function of the height to which the water can be thrown out of the channel.
The fairly regular spacing of the cones along the edge or the shore-fast ice may be analogous to the regular spacing of beach cusps. This suggests that the spacing might be controlled by the direction and size of the approaching waves. The exact relationship is unknown.
The headlands, which extend lakeward between the cones, are partially erosional and partially constructional features. The channels are cut deeply into the shore-fast ice and cones are built at the ends of the channels shoreward of the original position of the ice front. In addition, waves do break over the entire edge of the ice and, thus, build an ice and debris surface that slopes away from the open lake and channel edge in much the same fashion that levees slope away from a river channel. This tends to build the headlands higher and perhaps farther lakeward.
Since the row of cones and the 50 m or so of shore-fast ice were completely destroyed this year during a violent storm, it appears that the wind, and the wind-generated waves, can reach a point where the channels are cut faster than the cones and headlands are built. Erosion is also made easier by a higher water level which raises some of the ice from the bottom and also, perhaps, by higher temperatures, which cause some melting. The combination of these factors ultimately results in the destruction of the cones and all of the ice is broken up and either piled on shore or carried along shore. If this happens just before the lake begins to freeze more solidly, it is possible that a new line of cones will not be built. This, of course, would vary from year to year as does the size of the cones and the numbers of rows of cones. If several rows of cones remain in place, they tend to protect the shoreline from erosion during winter storms.
