WITTKOP, BENNETT, CHORMANN AND WUNSCH

GEOLOGY OF THE MAY 2006 SUNCOOK RIVER AVULSION

by

Chad Wittkop, Department of Geology, University of Wisconsin – Eau Claire, Eau Claire, WI 54701

Derek Bennett, Rick Chormann and David Wunsch, New Hampshire Geological Survey, NH Department of

Environmental Services, 29 Hazen Drive P.O. Box 95, Concord, NH 03302-0095

INTRODUCTION

The May 15-16, 2006 Suncook River avulsion in Epsom, NH, was the highest profile geologic event occurring

in the state since the collapse of the Old Man of the Mountain in 2003 (e.g. Zezima, 2006; for Old Man collapse see

Fowler, 2005). Rivers commonly change course in the form of meander cutoffs and small-scale avulsions within

floodplains of braided systems, but the Suncook avulsion was unusual because the new channel cut through an area

outside the documented 100- and 500-year floodplains (National Flood Insurance Program, 1978).

This unique event serves as a natural laboratory for study of the processes driving river avulsion, the creation of

a new river valley, and the impact of human activity in shaping these events. This field guide examines the causes

of the avulsion by combining field observation with analysis of aerial photography, GPS data, and geologic maps in

a GIS environment, and documents the major features visible in the field at the avulsion site.

STUDY AREA

The Suncook River originates in the town of Gilmanton at the outlet of Crystal Lake, which collects the inflow

of several smaller brooks and streams draining the southern flanks of the Belknap Mountains in Gilmanton and

Alton. Over its 35 mile length, the Suncook River drains an area of 256-square miles in southeastern New

Hampshire including portions of 16 towns in four counties. The Suncook River flows south-southwest and joins the

Merrimack River at Suncook Village, approximately six miles south of Concord, NH.

The May 2006 avulsion site is located in the town of Epsom, approximately 9.5 miles upstream of the

confluence of the Suncook River with the Merrimack (Figure 1). Just west of the avulsion site, the Suncook

formerly split into two channels—a primary (west) channel and a smaller, secondary (east) channel—forming an

island (Bear Island) 0.8 miles long and 0.3 miles wide. The lower member of the Silurian Rangeley Formation, a

stratified metapelite, occurs at depths of less than ten feet from the surface around the northwestern portion of Bear

Island (Lyons et al., 1997; Goldsmith, 1998). Between the study site and its confluence with the Merrimack, the

Suncook valley follows the trace of the strike-slip Pinnacle fault, which forms the sharp eastern boundary of the

valley in the study area (Lyons et al., 1997).

During late-glacial times, an arm of glacial Lake Hooksett—a large glacial lake whose level was controlled by a

spillway in the Merrimack Valley just south of the Suncook confluence—extended northeast into the Suncook

valley, depositing a series of coalescing coarse-grained (sand to cobble gravel) ice-contact deltas as well as finer

grained (fine sand to silt) lake-bottom sediments containing rhythmic bedding (Goldsmith, 1998). Where

undisturbed, coarser-grained ice-contact delta deposits form a series of hummocks with as much as 60 feet of relief

above the low-lying glacial lake beds, stream terraces, alluvium, small ponds, and wetlands of the valley bottom

(Figure 2).

At Huckins Mill at the northern end of Bear Island, two dams at heights of 13 and 5 feet height blocked the

main and secondary channels of the Suncook respectively, creating a 31-acre impoundment (New Hampshire

Department of Environmental Services Dam Inventory). The dams were constructed in the late 19th

century and

reconstructed in the 1930s (Orff, 2006).

MAY 2006 FLOOD

May 2006 was the second wettest month on record in New Hampshire. A sustained event from May 10-15

contributed up to 17 inches of precipitation in southern New Hampshire and northeastern Massachusetts (National

Climatic Data Center, 2006). Though the Suncook River is not presently monitored as part of the USGS stream

gauging network, highest-ever flows were recorded on 12 rivers in central and southern New Hampshire as a result

1

WITTKOP, BENNETT, CHORMANN AND WUNSCH

of this precipitation event, including tributaries of the Merrimack River of similar size as the Suncook (USGS News

Release, 2006). Peak flow measurements averaged 24 times normal, at or exceeding the 100-year return interval.

Figure 1. Location of new and abandoned Suncook River channels. Flow is from north to south. 10-foot

contours reproduced from 1967 USGS 7.5-minute Gossville quadrangle.

METHODS

The avulsion site and surrounding areas were inspected several times following the May 15-16 event to

photograph and describe the features observed. Differential GPS data were collected on May 20, 24 and June 9,

2006 using a Trimble backpack system and supplemented with field notes, photos, and Brunton compass bearings of

the new channel’s path. Following field collection, GPS data were differentially corrected against a GPS base

station maintained by the NH Department of Transportation. These data were compiled into a GIS database using

ArcMap 9.1® software.

2

WITTKOP, BENNETT, CHORMANN AND WUNSCH

Figure 2. Surficial geologic map of the area around the avulsion site (redrawn from Goldsmith, 1998) showing

locations of glacial lake sediments, stream terraces and alluvium, glacial till, and areas of shallow bedrock

(less than 10 feet depth).

GPS and field data were compared with 1-meter pixel resolution digital orthophotography aquired by the NH

Department of Transportation from imagery collected in April 2005, the 1967 USGS 7.5-minute topographic map of

the Gossville quadrangle, a 10-m digital elevation model (DEM) derived using tagged vector contour data digitized

from the USGS topographic map, and a surficial geologic map (Goldsmith, 1998). Estimates of new channel

dimensions and the volume of sediment removed were computed from measured distance and area determined using

GIS. Planes of equal elevation were computed from the DEM with the Raster Calculator function in the Spatial

Analyst extension of ArcMap.

It is important to note that, with the exception of the high-resolution orthophotography and the GPS data, this

analysis is based on best-available 1:24,000-scale data, and such data are generally not intended for site-specific

studies at the scale of the events discussed here.

3

WITTKOP, BENNETT, CHORMANN AND WUNSCH

Points of interest around the avulsion site—including the banks of the new channel, areas of significant erosion,

and high-water marks—were located using differential GPS. High-water marks were identified from physical

evidence such as erosional features, limits of silt deposition, and locations of organic debris (twigs and leaves)

festooned in branches and underbrush. Maps of the new channel were drawn by comparison of these data with

detailed field notes, photographs, and compass bearings. The map of the new channel was further refined after

comparisons with oblique aerial photos obtained by the Army Corps of Engineers in July 2006 and January 2007,

and satellite imagery obtained by NASA in summer 2006.

RESULTS

The Suncook River now flows through a gravel pit to the northeast of Bear Island before rejoining a portion of a

preexisting secondary channel that formed the eastern boundary of the island (Figure 1; approximate location of

gravel pit contained within area of interest of Figure 4). 1.97 miles of former channel were abandoned, including

1.52 miles of the primary channel that formed the western boundary of Bear Island. Aside from small pools and

seeps and contribution from a small tributary, the abandoned portions of the Suncook are not expected to maintain

significant year-round flow, and by mid-summer of 2006 much of the abandoned reach was completely dry.

Bedrock is exposed in the abandoned channel around the northern end of Bear Island. The presence of rocky

substrate is unique to this reach of the Suncook and has served to restrict downcutting. Surficial geologic maps

show that shallow bedrock does not occur upstream of this area in the former channel (Goldsmith, 1998; Figure 2).

The new channel is 1.03 miles long, of which 0.44 miles is newly eroded. As a consequence of the overall

shortening of the Suncook’s course, the average gradient of the river increased 44%, from 16 feet per mile to 23 feet

per mile. In addition, the flow of the Suncook is now concentrated into a single channel, whereas it had previously

split into two channels around Bear Island. As a result, the average velocity of the river will likely increase both

upstream and downstream of the avulsion site, enhancing the river’s ability to erode laterally and horizontally. By

summer 2006, the new channel of the Suncook had downcut as much as 5 feet below the thalweg of the abandoned

channel at the point of avulsion.

Erosion and subsequent downcutting of the new channel created nearly continuous exposures of glacial Lake

Hooksett bottom sediments, and in places exposed Holocene wetland sediments containing macrofossil remains and

nodules of the blue mineral Vivianite (Fe3[PO4]2·8H2O). Eroded wetland sediments featured a sculpted texture

reminiscent of larger-scale erosional features seen in bedrock-lined river channels.

A distinct high-water mark was observed along the western face and remnants of the southern face of the gravel

pit that is dissected by the new channel, indicating that water pooled there to a depth of as much as 5 feet prior to the

avulsion.

GIS ANALYSIS

A GIS analysis of high-water marks and DEM data was undertaken to evaluate hypotheses for the initiation of

the new channel—namely whether the new channel initiated from the north when the Suncook River broke through

its banks, or whether the new channel resulted from headward erosion after flood waters found an outlet at the

southern end of the gravel pit. Local flood-levels were reconstructed by intersecting planes of equal elevation with

the DEM surface and comparing the limits of inundated areas with the distribution of GPS-located high-water marks

as determined in the field.

A 335-foot-elevation flood surface (Figure 3A) would overtop the banks of the Suncook River at the northern

portion of the new channel, but a flood limited to this elevation would not have found an outlet farther south, nor

would it have reached all GPS-located high-water-mark elevations. National Flood Insurance maps (1978) use 335

feet as the base flood elevation for the reach of the Suncook River above Huckins Mill, and 336 feet where the new

channel intersects the old. A local flood maximum at an elevation of 340 feet (Figure 3B) equals or exceeds the

levels of GPS-located high-water marks. However, according to 1967 topographic data, this water level would not

have been high enough to overtop the gap in the glacial ridge presently breached by the new channel. A 345-foot

flood surface (Figure 3C) would have found an outlet through two gaps in the glacial ridge where the new channel

now exists, but neither field nor anecdotal evidence support flood waters having reached this elevation.

4

WITTKOP, BENNETT, CHORMANN AND WUNSCH

A local flood surface of approximately 340-feet elevation is supported by GPS-located high-water marks and

anecdotal evidence. Water would have reached within a few feet of a natural gap in the glacial ridge but remained

below the elevation needed to initiate avulsion, given the 1967 topographic model. However, significant landscape

alteration in this area has occurred since that time in the form of excavations from a large gravel pit.

Assuming a local flood surface at an elevation of 340 feet, estimates of depths of gravel mining can be obtained

by overlaying contours from 1967 topographic data onto recent, high-resolution orthophotos of the gravel pit and

subtracting the flood elevation from the 1967 elevation. This analysis suggests that as much as 20 feet of excavation

had occurred in the main portion of the gravel pit where water ponded during the flood.

A postulated flood surface of approximately 340-feet—including the water known to have pooled in the gravel

pit—is shown in Figure 3D. Because topography at the southern portion of the gravel pit was destroyed during the

avulsion, the exact extent of the 340-foot elevation surface cannot be determined.

Overlaying maps of the new channel and areas of avulsion-related scour on 2005 orthophotography (Figure 4)

shows a significant area of scour corresponding with the location of a small access path at the southern end of the

pit. It is likely that the avulsion initiated there when water pooled in the gravel pit reached high enough for the

access path to act as a spillway. Headward erosion rapidly proceeded east-northeast from this point through easily

eroded lake-bottom and wetland sediments. This headward erosion scenario is supported by eyewitness accounts

that describe development of a cascade in the flooded gravel pit and primary access road (Bill Yeaton, personal

communication; Orff, 2006).

Field inspection in the area where initiation of the avulsion is suspected revealed a portion of undisturbed road

grade surrounded by eroded areas. This undisturbed road grade is located in the former gap between glacial ridges

and suggests that gravel-mining operations just west of the gap removed material from its flanks to an elevation

lower than the roadbed itself. This artificially expanded the floodplain of the Suncook and fatally compromised the

glacial ridge that served as a natural levee dictating the unusual east-northeast course of the Suncook above Huckins

Mill since the Pleistocene. The configuration of the gap suggests that it may have formed by erosion during similar

large flood events throughout the Holocene.

Avulsion-associated erosion introduced an estimated 150,000 cubic yards of sediment into the Suncook River.

Much sediment was deposited downstream of the avulsion channel in silt and sand sheets up to 5 feet thick.

Portions of channel downstream of the avulsion that were once several feet deep are now less than a foot deep. The

large volume of sediment introduced downstream will decrease channel depth and increase the frequency of

overbank flooding. In the avulsion area and upstream, the Suncook continues to downcut in response to the increase

in gradient and convergence of flow. This downcutting is expected to increase rates of bank erosion in the avulsion

area and upstream as the river attempts to create a new floodplain. The sandy to silty glacial lake and stream

sediments at the avulsion site and upstream will offer little resistance to erosion and can be expected to continue to

supply large amounts of sediment to the river.

CONCLUSIONS

DEM analysis of high-water marks suggests that May 2006 floodwaters would have reached within a few feet

of a natural gap in a ridge that had previously served as a natural levee blocking the flow of the Suncook River in

this area. Less than 5 feet of excavation near the gap would have allowed the avulsion to occur given the water

levels observed. Such landscape modification in an area naturally sensitive to disturbance allowed the avulsion to

initiate.

The dynamics of the Suncook River avulsion serve as a potential analogue to catastrophic drainage events and

spillway dynamics of ponded water bodies that formed during the deglaciation of New England.

This study demonstrates the utility and power of “off-the-shelf” GIS and GPS techniques coupled with field

work, and shows that these methods can be used in place of more expensive and specialized techniques such as

surveying or LiDAR to support a useful first-order analysis.

5

WITTKOP, BENNETT, CHORMANN AND WUNSCH

Figure 3. GPS locations of high-water mark field evidence (dots) and DEM-calculated flood surfaces for the area

where the avulsion occurred. 10-foot contours redrawn from 1967 topographic data. ‘G’ denotes the gap

between ridges where the new channel now flows. A) 335-foot elevation plane. This level does over-top

the banks of the Suncook but does not reach all GPS-located high water marks. B) 340-foot elevation

plane. This elevation meets or exceeds all high water mark locations but does not breach the gap in the

glacial ridge. C) 345-foot elevation plane. This elevation exceeds all high water mark locations and

breaches the glacial ridge in two places (arrows). However, local flood levels at this elevation are not

supported by field or anecdotal evidence. D) 340-foot elevation plane including boundaries of gravel pit

flooding (ruled pattern) inferred from aerial photo interpretation and GPS data. Question mark denotes

area of gravel pit destroyed by the avulsion.

6

WITTKOP, BENNETT, CHORMANN AND WUNSCH

Figure 4. Location of new channel and associated areas of scour overlain on 2005 aerial photo of gravel pit.

Access path where avulsion likely initiated is highlighted by white arrow.

7

Gravel Pit

Access

Path