Content uploaded by Don Rosenberry
Author content
All content in this area was uploaded by Don Rosenberry
Content may be subject to copyright.
MOVEMENT OF ROAD SALT TO A SMALL NEW HAMPSHIRE LAKE
D. O. ROSENBERRY1, P. A. BUKAVECKAS2,D.C.BUSO
3,G.E.LIKENS
4,A.M.
SHAPIRO5and T. C. WINTER1
1U.S. Geological Survey, MS 413, Bldg. 53, DFC, Lakewood, CO 80225, U.S.A.
2Water Resources Laboratory, University of Louisville, Louisville, Kentucky 40292, U.S.A.
3Institute of Ecosystem Studies, Hubbard Brook Experimental Forest, Mirror Lake Road, RFD, Box
779, Campton, New Hampshire 03223, U.S.A.
4Institute of Ecosystem Studies, Box AB (Route 44A), Millbrook, New York 12545–0129, U.S.A.
5U.S. Geological Survey, 12201 Sunrise Valley Drive, MS 431, Reston, VA 20192, U.S.A.
E-mail: rosenber@usgs.gov
(Received 10 February 1997; accepted 11 December 1997)
Abstract. Runoff of road salt from an interstate highway in New Hampshire has led to contamination
of a lake and a stream that flows into the lake, in spite of the construction of a diversion berm
to divert road salt runoff out of the lake drainage basin. Chloride concentration in the stream has
increased by over an order of magnitude during the 23 yr since the highway was opened, and chloride
concentration in the lake has tripled. Road salt moves to the lake primarily via the contaminated
stream, which provides 53% of all the chloride to the lake and only 3% of the total streamflow to
the lake. The stream receives discharge of salty water from leakage through the diversion berm.
Uncontaminated ground water dilutes the stream downstream of the berm. However, reversals of
gradient during summer months, likely caused by transpiration from deciduous trees, result in flow
of contaminated stream water into the adjacent ground water along the lowest 40-m reach of the
stream. This contaminated ground water then discharges into the lake along a 70-m-wide segment
of lake shore. Road salt is pervasive in the bedrock between the highway and the lake, but was not
detected at all of the wells in the glacial overburden. Of the 500 m of shoreline that could receive
discharge of saly ground water directly from the highway, only a 50-m-long segment appears to be
contaminated.
Key words: cation exchange, ground water, lake contamination, road salt, stream contamination
1. Introduction
Contamination of land and water resources from runoff of deicers applied to road-
ways is common throughout the northern and especially northeastern United States
(e.g. D’Itri, 1992; National Research Council, 1991). Many case studies of road-
salt contamination of rivers (Peters and Turk, 1981), wetlands (Wilcox, 1986),
lakes (Judd, 1970; Bubeck et al., 1971; Cherkauer and Ostenso, 1976; Driscoll
et al., 1991), and ground water (Bowser, 1992, Church and Friesz, 1993) have
been published. However, for many studies, evidence of the effects of road salt
on water supplies is circumstantial (Jones and Jeffrey, 1992). Often this is due to
a lack of data prior to application of road salt. The long-term record of chemical
data at Mirror Lake, New Hampshire, extending in time both before and after the
Water, Air, and Soil Pollution 109: 179–206, 1999.
© 1999 Kluwer Academic Publishers. Printed in the Netherlands.
180 D. O. ROSENBERRY ET AL.
Figure 1. Location, and streams in and adjacent to Mirror Lake watershed.
construction of an interstate highway through the drainage basin, provided aunique
opportunity to accurately document the effect of road-salt contamination on the
water chemistry of a dilute freshwater lake (Bormann and Likens, 1985). Road-
salt contamination of the lake appeared to be caused by direct input of surface
drainage via the northeast inlet stream, in part due to a leaky diversion berm, and
to contaminated ground water discharge either to the stream or directly to the lake.
The relative proportions of these sources were unknown.
Mirror Lake is a small (15 ha, max. depth 11 m), oligotrophic lake in the White
Mountains of New Hampshire (Figure 1). During 1969–71, interstate highway I93
was constructed through a portion of the Mirror Lake drainage basin. Because of
concerns about contamination of the lake from erosion due to construction activi-
ties and road-salt runoff, a diversion berm was constructed to divert surface runoff
from the highway to a lowland north of the Mirror Lake drainage basin. This
diversion of surface flow reduced the size of the drainage basin of the northeast
MOVEMENT OF ROAD SALT TO A SMALL NEW HAMPSHIRE LAKE 181
Figure 2. Concentrations of sodium and chloride in Mirror Lake, 1967–1994, in µeq L−1. Linear
regression based on data from the period 1974–94.
inlet stream from 20 to 2.5 ha. In spite of these precautions, concentrations of
sodium and chloride in the northeast inlet stream to Mirror Lake have increased by
more than an order of magnitude. In Mirror Lake, concentrations of sodium have
doubled and concentrations of chloride have more than tripled since about 1973,
causing concern about alteration of the ecology of the lake.
From the mid 1960s until the early 1970s, concentrations of sodium in Mirror
Lake ranged between 35 and 70 µeq L−1and chloride concentrations ranged be-
tween 15 and 30 µeq L−1. Increased concentrations of sodium and chloride first
were detected in the lake around 1974–75. By 1980, 9 yr after completion of the
interstate highway through the Mirror Lake watershed, concentrations of sodium
182 D. O. ROSENBERRY ET AL.
Figure 3. Concentrations of sodium and chloride in the northeast inlet stream, 1967–1994, in µeq
L−1. Linear regression based on data from the period 1974–1994.
had increased by about 40% and chloride in the lake had approximately doubled.
Since 1974, a steady increase in concentration of 2.4 µeq L−1yr−1for sodium and
3.3 µeq L−1yr−1for chloride has resulted in present-day (1994) concentrations of
about 95 µeq L−1sodium and 85 µeq L−1chloride (Figure 2). Present-day con-
centrations still are quite low, relative to many other studies of lake contamination
from road-salt runoff. However, if the rate of road-salt loading to the lake continues
to increase, future concentrations could affect the physical and biological processes
within the lake, eventually affecting lake circulation, as has happened at numerous
other lakes that receive road-salt runoff (Bubeck et al., 1971; Hawkins and Judd,
1972; Cherkauer and Ostenso, 1976).
During the past thirty years, concentrations of sodium and chloride have fluc-
MOVEMENT OF ROAD SALT TO A SMALL NEW HAMPSHIRE LAKE 183
tuated between 20 and 125 µeq L−1in the west and northwest tributaries without
any obvious longterm trends. However, atthe northeast inlet stream, concentrations
of sodium have increased from 100 to 1500 µeq L−1and chloride concentrations
have increased from 100 to 2000 µeq L−1from about 1971 to the present (1994)
(Figure 3) (Bormann and Likens, 1985; Likens, 1992). Although other sources
for sodium and chloride in the Mirror Lake watershed may have contributed to the
increased sodium and chloride concentrations in the lake, because of the magnitude
and timing of the change it was suspected that road-salt deicers were the primary
source of contamination.
1.1. PURPOSE AND SCOPE
The pathways by which highway runoff moves from the catchment to the lake
needed to be identified to understand why mitigation of road salt runoff still re-
sulted in contamination of the lake, and to allow for better estimates of future
concentrations of sodium and chloride in the northeast inlet stream and in the lake.
The study also investigated the relative magnitudes of various sources of sodium
and chloride to the lake to determine whether increased concentrations of chloride
in the lake could be attributed entirely to contributions from the northeast inlet
stream, or if other sources of chloride needed to be identified to account for the
mass of chloride in the lake. The focus of this investigation was limited to the
northeast inlet stream and eastern shoreline of the lake because those areas of the
watershed are most susceptible to contamination by movement of road salt from
the nearby highway, due to their close proximity to the highway, and because this
was where increased sodium and chloride concentrations have been documented.
Data specific to this study were collected from late 1992 through August 1994, in
conjunction with collection of ongoing long-term monitoring data.
2. Physical and climatic setting
The Mirror Lake watershed is located in the northeast section of the Hubbard
Brook watershed, which has been the focus of long-term research on the ecology
and biogeochemistry of small, forested catchments (Likens and Bormann, 1995;
Bormann and Likens, 1979). Hydrologic and limnologic research has occurred
within the lake watershed since the early 1960s (Likens, 1985). The Mirror Lake
watershed has an area of 85 ha (excluding the lake) and a maximum relief of 268 m.
Crystalline bedrock is exposed along parts of the northeast and east shorelines of
the lake; bedrock outcrops are present throughout the watershed. Glacial deposits
cover most of the watershed, and underlie most of the lake. The glacial deposits
generally are a silty, sandy till containing numerous cobbles and boulders. Terrain
generally is steep within the watershed, especially on the north and west sides. Till
on the east side of Mirror Lake, the area of interest for this study, generally is thin,
184 D. O. ROSENBERRY ET AL.
ranging from 0 to 11m thick. Three streams flow into Mirror Lake on the west,
northwest, and northeast sides of the lake. A dammed outlet on the south side of
the lake discharges water to Hubbard brook, which empties into the Pemigewasset
River 3 km southeast of the lake (Figure 1).
The northeast inlet stream has an average gradient of 0.04 along the 190 m
distance between the diversion berm and the lake, but the gradient steepens along
the 50 m segment closest to the lake. A much steeper gradient (about 0.1) exists
across the berm between the ponded water east of the berm, and the marshy area
directly west of the berm; commonly there is a head difference of 1 m from one
side of the berm to the other.
The climate of the Mirror Lake area is humid continental (Likens and Bormann,
1995). Annual precipitation averaged 1400 mm and annual evapotranspiration av-
eraged 525 mm during 1963–1993. Approximately 30% of annual precipitation
falls as snow, which generally melts during March and April. Snowmelt contributes
to heavy springtime streamflow; about 50% of annual streamflow occurs during
March through May. The growing season for the mixed deciduous-coniferous for-
est in the watershed is from mid May to mid October. Streamflow generally is low
during this period (Likens and Bormann, 1995).
3. Methods
For this study, two transects, each consisting of two wells and a stream-stage gage,
were installed to determine the interaction between ground water and surface water
at the northeast stream (Figure 4). Wells were hand augered, 5.1 cm in diameter,
contained a 0.5 m long wire-wound PVC screen, and were completed about a meter
below the water table. Wells 28 and 30 were completed east of the stream, and wells
29 and 31 were completed west of the stream; all wells were located within 5 m
of the stream. A flume (S1 in Figure 4) located at transect 1 provided continuous
record of stage and discharge in the stream. This also was the site where long-term
discharge and water-chemistry data were collected. At transect 2, a manually-read
stage gage (S2) provided stream-elevation data. Stream-stage gages and sampling
sites also were established west (downgradient) and east (upgradient) of the di-
version berm, designated S3 and S4 respectively (Figure 4). Transect 1 is located
approximately 150 m downstream of the berm (20 m upstream of the lake) and
transect 2 is located approximately 70 m downstream of the berm (100 m upstream
of the lake). Water-level data were recorded weekly; water-table gradients between
the northeast stream and adjacent ground water were monitored at these transects
from October 1992 to July 1994. Fluxes were estimated using the Darcy equation:
Q=KIA,whereQ= discharge (L3T−1), K= hydraulic conductivity (LT−1), I=
hydraulic gradient (dimensionless) and A= cross-sectional area through which
ground water flows to or from the stream (L2).
Wells 28 and 30, were installed in poorly sorted silty sandy till containing many
MOVEMENT OF ROAD SALT TO A SMALL NEW HAMPSHIRE LAKE 185
Figure 4. Location ofwells and staff gages in study area.
cobbles and boulders; the till at well 28 was particularly silty. Well 31 was installed
in till that was less silty than at wells 28 or 30. Well 29 was installed in sand that
contained very little silt. Because of low water levels in wells, in-situ hydraulic
conductivity was determined by slug test only at well 31, which had a value of
5×10−4cm s−1, typical of silty sandy till. Based on observations of geology
during well installation, hydraulic conductivity was estimated to be the same for
well 30 as for well 31, half an order of magnitude greater at well 29, and an order
of magnitude less at well 28.
Three wells were installed adjacent to I93 to obtain information on the vertical
gradients within the ground water system between the glacial deposits and the
underlying fractured bedrock. Only the water-table well, labeled IS1, is shown in
Figure 4. Two of the wells were installed in the glacial deposits at depths of 5.6 and
11.6 m below land surface; each well has a 0.6-m-long well screen. The third well
was cased through the unconsolidated glacial deposits and extends as an open hole
in the bedrock to 151.8 m below land surface. Additional wells (COWT, 32, CS-
20R) were installed to determine ground water gradients in the area between I93
and the eastern side of the lake (Figure 4). Numerous wells completed in bedrock
186 D. O. ROSENBERRY ET AL.
are located near well COWT, but are not shown in the figure. In addition, numerous
other stream gages, wells, rain gages, and other meteorological instruments exist in
the Mirror Lake watershed; only those pertinent to this study are shown in Figure
4. All wells, stream-stage gages, and the lake gage were surveyed to a common
elevation. Shapiro et al. (1995) provide an overview of ground water flow and
chemical transport through fractured rock in the watershed.
A hydraulic potentiomanometer (Winter et al., 1988) was used during July and
August 1994 to investigate whether salt-contaminated ground water was discharg-
ing into Mirror Lake along the eastern shoreline. This device was used to determine
direction of flow between ground waterand the lake, and to collect water-chemistry
samples from directly beneath the lake bed at distances from 0 to 4 meters from
shore. Each hydraulic-head measurement indicated a vertical potential between the
point at which the 8-cm-long screen was exposed to the aquifer and the lake. Head
measurements relative to lake level were madeat three to four distances from shore
at each sampling site; with the probe inserted 0.5 m into the lakebed, the smallest
vertical hydraulic-head gradient that could be detected was 0.002. The device also
was used along the northeast inlet stream to collect hydraulic-head and chemistry
data from beneath the streambed.
Water and chemical budgets for Mirror lake were determined to estimate the
relative contribution of sodium and chloride from the northeast inlet. Information
on methods for determining monthly and annual water and/or chemical budgets
for Mirror Lake is presented in Likens (1985), Winter (1984), Winter (1985), and
Rosenberry and Winter (1993). Lake chemistry data were collected near the center
and deepest portion of the lake.
Precipitation, stream, and lake-water samples were collected weekly for se-
lected chemical components. Samples were collected from wells 28–32 and CS-
20R approximately monthly from November 1992 to August 1994. Samples were
collected from well COWT intermittently during the study. In-situ temperature, pH,
specific conductance, and acid-neutralizing capacity (ANC) also were measured
in wells 28–32 during each sampling event. Samples collected with the hydraulic
potentiomanometer system were analyzed for selected major ions, pH and specific
conductance. Sodium, calcium, magnesium and potassium were determined by
atomic adsorption spectrophotometry. Chloride, sulfate, nitrate, phosphorus and
ammonium were determined by ion chromatography or by automated spectro-
colorimetric methods. ANC was determined by titration with hydrochloric acid
using a granplot analysis of automated potentiometric pH data, and pH was deter-
mined by potentiometry with glass electrodes calibrated at pH 7 and 4 (Wetzel and
Likens, 1991; Likens and Bormann, 1995).
MOVEMENT OF ROAD SALT TO A SMALL NEW HAMPSHIRE LAKE 187
Figure 5. Potential pathways for movement of road salt between I93 and Mirror Lake.
4. Results and Discussion
4.1. PATHWAYS FOR CONCENTRATION
Three potential pathways for movement of road salt to Mirror Lake are shown in
Figure 5. (Pathway 1) A narrow zone of salt-contaminated water seeps through
the diversion berm and discharges directly to the present-day headwaters of the
northeast stream. Ground-water gradients are assumed to be toward the stream,
188 D. O. ROSENBERRY ET AL.
which would restrict salt contamination to a narrow region along the stream. Salt
travels via the stream to the lake both in the stream and also in the narrow, shallow,
hyporheic zone directly beneath and adjacent to the stream. (Pathway 2). Water
that has seeped through the berm discharges both to the headwaters of the stream
as well as to ground water on either side of the stream, directly downgradient of
the berm. Salt-contaminated ground water flows toward the lake, but lateral ground
water gradients toward the stream cause the plume to merge with the stream before
flowing as stream water and hyporheic water to the lake. Because ground water
wells located north of the lake have water levels much higher than in the vicinity
of the northeast stream, and because the stream is incised relative to the hillside
directly west of the stream, it is assumed that ground water gradients toward the
stream on the west of the stream prevent a salt plume from spreading west of the
stream. (Pathway 3) Salt contamination is not restricted to the northeast stream
headwaters region. A broad region of salt-contaminated ground water moves from
the interstate highway toward the entire eastern shoreline of Mirror Lake. As with
the second pathway, an assumed gradient toward the stream on its west side pre-
vents movement of a plume to the west of the stream. The third pathway provides
the greatest potential for future large increases in concentrations of sodium and
chloride in the lake.
4.2. HYDROLOGY AND CHEMISTRY OF THE NORTHEAST STREAM
Streamflow in the northeast stream normally reaches a maximum in April, declines
throughout the summer to a minimum in July or August, then rises to a smaller
peak in the fall before declining again during winter (Figure 6). From June through
October, the northeast stream often flows only following rain events. Chloride
concentration typically is diluted during spring runoff, but concentration is little
affected by increased streamflow in the fall. The mass of chloride that enters Mirror
Lake from the northeast stream has a bimodal peak; the greatest mass enters during
April, coincident with the streamflow peak, and a slightly smaller peak also occurs
during October and November (Figure 6). A similar bimodal pattern for flux of
calcium, sulfate and silicate from all streams into Mirror Lake was reported in
Likens et al. (1985).
Chloride concentrations in the northeast stream usually decreased with distance
from I-93 (Figure 7A, Table I), Median chloride concentrations were 4202 µeq
L−1at S4, 3308 µeq L−1at S3, 2552 µeq L−1at S2 and 1472 µeq L−1at S1.
Occasionally chloride concentration increased downstream from the berm between
sites S3 and S2. For example, from March through June 1993, chloride concentra-
tion at site S2 was higher than at site S3 (Figure 7A). Also, occasionally chloride
concentrations were lower in the ponded water on the highway side (east) of the
berm, an example of which was from December 1993 through April 1994, when
chloride concentration on the east side of the berm at site S4 was lower than at both
sites S3 and S2 (Figure 7A).
MOVEMENT OF ROAD SALT TO A SMALL NEW HAMPSHIRE LAKE 189
Figure 6. Monthly flow (m3mo−1), chloride concentration (µeq L−1), and chloride mass (kg mo−1)
in the northeast stream (medians of monthly values from 1988–1993).
Assuming that chloride is conserved, declining concentration in the stream indi-
cates that uncontaminated ground water discharged into the stream and diluted the
stream water. The average dilution of chloride between sites S3 and S1 indicates
that streamflow increased in volume by about 35% over a stream reach of approx-
imately 150 m. Estimated ground water fluxes, using average gradients at the two
transects and a hydraulic conductivity ranging from 5 ×10−4cm s−1to 1 ×10−3
cm s−1, indicated a similar increase in streamflow, between 7 and 33%. Estimated
fluxes ranged from 65 to 330 m3mo−1between sites S3 and S1.
Water-chemistry samples from wells adjacent to the stream indicated that ground
water in the vicinity of the stream is not contaminated with road salt, except in the
vicinity of well 29 (Figure 7B). Chloride concentrations at well 29 were approxi-
mately an order of magnitude greater than concentrations at wells 28, 30 and 31,
which had chloride concentrations similar to uncontaminated ground water west of
Mirror Lake. However, concentrations at well 29 still were much lower than in the
stream, indicating that some of the water flowing to the well did not originate from
the stream.
190 D. O. ROSENBERRY ET AL.
TABLE I
Median values for selected chemical constituents (and ratios relative to sodium)
from 11 samples collected from the northeast inlet stream, January 1991 – August
1994, in µeq L−1
Site Ca Mg K Na H SO4NO3Cl ANC
S1 514 160 33 1394 3 143 4 1472 71
S2 656 206 46 1881 4 143 4 2552 183
S3 1794 568 92 1960 1 143 2 3308 1400
S4 781 225 98 3678 2 143 4 4202 1263
Site Na/Ca Na/Mg Na/K Na/Cl
S1 2.7 8.7 42.2 0.9
S2 2.9 9.1 40.9 0.7
S3 1.1 3.5 21.3 0.6
S4 4.7 16.3 37.5 0.9
Notes:
1. Ammonium and ortho-phosphate were measured only at site S1 and were
assumed to be negligible at all other sites.
2. Obvious rusty flocculate in samples from sites S3 and S4 indicate missing
cation may be Fe+3.
Water-level data collected at the two transects perpendicular to the northeast
stream indicated that water-table gradients were not always toward the stream as
had been expected, and as the water-chemistry data indicated. During fall through
spring, gradients typically were toward the stream at both transects (Figure 8).
However, during early to mid summer, gradients reversed as the water table dropped
below the stream level at both transects. Gradients were away from the stream dur-
ing most of the summer, until cessation of transpiration likely allowed fall recharge
to raise the water table above the stream stage. Maximum flow velocities between
the stream and ground water at wells 28-31, assuming a porosity of 0.2 (based
on the poorly sorted till adjacent to the stream) and using the maximum gradients
recorded between the wells and the stream, ranged from about 0.02 m day−1at well
28 to about 0.5 m day−1at well 29. Gradients at these wells frequently were much
smaller than the maximum recorded gradients; in addition, gradients reversed sea-
sonally. Therefore, net movement of salt-contaminated ground water in the vicinity
of the stream was significant only in the vicinity of well 29.
4.3. CONTAMINATION OF GROUND WATER FROM THE NORTHEAST INLET
Discrete measurements of water levels at transect 1, along with large levels of
chloride at well 29, indicated that the lower reach of the stream was losing water,
a condition not expected prior to this study. The steeply sloping hillside to the
MOVEMENT OF ROAD SALT TO A SMALL NEW HAMPSHIRE LAKE 191
Figure 7. Concentration of chloride (µeq L−1): (A) at sites S1-S4, (B) at wells 28–31.
west of the stream, together with ground water levels north of the lake that were
30 m higher than the lake, had led to the assumption that gradients toward the
stream from the west would be steep. Wells 28 and 29 were instrumented with
continuous water-level-monitoring equipment from mid July to early October 1994
to determine if diurnal ground water fluctuations would show evidence of transpi-
ration, which could remove ground water and create a gradient from the stream
to ground water. During much of the period of record, water-level fluctuations in
the wells showed little evidence of diurnal fluctuations. Water levels usually were
rising or falling rapidly in response to individual rainfall events, which would mask
much smaller fluctuations due to transpiration. However, an extended dry period
during the first two weeks of August provided evidence that evapotranspiration
192 D. O. ROSENBERRY ET AL.
Figure 8. Stream and ground-water-head altitudes at two transects across the northeast stream,
January 1993 through August 1994, in m.
was removing ground water (Figure 9). At well 29, where the geologic deposits
are relatively permeable, response to evapotranspiration was subtle but immediate.
At well 28, where the till conducts water much less readily than the sand at well
29, water-level fluctuations were larger, but they were delayed. At well 29, on all
but one day during August 2–12, the water-level declined in early afternoon during
times of maximum evapotranspirational withdrawals. On August 9, 11, 12, and 13,
the water level recovered slightly during the nighttime and early morning hours.
Light showers on August 2, 5, and 9 may have reduced slightly the water-table
decline by reducing transpiration. These data indicate that evapotranspiration was
responsible for drawing ground water down to below stream level at transect 1. This
process probably occurred at transect 2 as well, but hourly data were not collected
from that site.
Exchange of stream water with water in the hyporheic zone adjacent to the
stream (White, 1993) may also have contributed to contamination of ground water
MOVEMENT OF ROAD SALT TO A SMALL NEW HAMPSHIRE LAKE 193
Figure 9. Hourly fluctuation of stream stage and ground water head at site S1, well 28, and well 29,
August 1–14, 1994, in m.
in the vicinity of well 29. The pool and riffle nature of the northeast inlet stream
may have created short stream-reach segments where the gradient was from pooled
stream water to ground water, even when gradients at transects 1 and 2 were toward
the stream. In such an event, once stream water with large concentrations of salt
flowed into the ground water system, that salty water could travel through the
ground water system downvalley toward the lake essentially parallel to the stream,
contaminating ground water downstream even in locations where the downstream
ground water was at a higher head than the stream water at that downstream lo-
cation. Evidence that this process was occurring is presented in Figures 7B and
8. During April 1993, the concentration of chloride increased sharply at well 29
(Figure 7B) while at the same time the head at well 29 was higher than the stream
stage at S1 (Figure 8). Since stream water directly adjacent to well 29 could not
flow to well 29, it is possible that stream water entered the ground water system
somewhere upstream, and flowed downvalley to well 29, causing the increase in
chloride concentration observed during April 1993.
194 D. O. ROSENBERRY ET AL.
4.4. EVIDENCE OF SALT CONTAMINATION IN THE NORTHEAST INLET
STREAMBED
Vertical hydraulic-head gradients were measured with the hydraulic potentiomano-
meter during late July 1994 beneath the streambed, and chemistry samples were
collected at the same time. At two sampling locations within 10 m of transect 2,
vertical gradients were small (0.01, 0.02) and indicated seepage from the stream to
ground water. A measurement made in a small pool in the stream halfway between
S2 and S3 indicated a large gradient (0.13) from the stream to ground water. In
the vicinity of the berm gradients also were large; west of (below) the berm the
gradient was 0.24 from ground water to the stream, and the gradient was 0.07 from
ponded water to ground water east of the berm.
Large chloride concentrations were measured in ground water to depths of 0.4
and 0.6 m directly beneath the stream, which was as far as the probe was inserted.
Chloride concentration beneath the streambed was largest directly east of the berm
(5590 µeq L−1) and directly west of the berm (5080 µeq L−1). Sub-streambed
chloride concentrations ranged between 2300 and 2500 µeq L−1along a 50-m
long reach centered on transect 2. Even at transect 1, where the hydraulic gradient
was very small, chloride concentration beneath the streambed still was large (1780
µeq L−1) relative to uncontaminated ground water. It is likely that in the fall, when
reduced evapotranspiration and increased recharge cause a reversal in the hydraulic
gradient between the stream and contiguous ground water, some of the salt beneath
the streambed is flushed into the stream, and contributes to the increase in chloride
concentration in the stream during October and November.
4.5. DISCHARGE OF SALT TO THE NORTHEAST INLET HEADWATERS AREA
Water chemistry data (Table I) collected at the four sites along the northeast inlet
(locations shown in Figure 4) indicate that ground water may have discharged to
the northeast stream via three different flow paths. (1) Water may have flowed
through the berm, from the pond east of the berm to the headwaters area west
of the berm. This flow path is indicated by the difference in water chemistry be-
tween sites S4 and S3. The median water chemistry of samples collected at site
S3 indicated 130% more calcium, 152% more magnesium, and 47% less sodium
than samples collected at site S4. The change in the proportion of base cations
indicates that significant cation exchange occurred in the berm material, perhaps
due to the existence of many freshly exposed weathering surfaces created during
the emplacement of berm fill. (2) Between sites S3 and S2, sodium concentra-
tions in the stream usually changed only slightly while calcium and magnesium
concentrations usually decreased, changing considerably the proportions of base
cations at site S2 to ratios similar to the water ponded east of the berm at site
S4 (Table I). This indicates a second possible ground water flow path; water from
the pond east of the berm flowed beneath the berm through undisturbed glacial
and streambed deposits and discharged farther downstream, between sites S3 and
MOVEMENT OF ROAD SALT TO A SMALL NEW HAMPSHIRE LAKE 195
S2. Water that flowed via this flow path contained sodium and chloride and little
else, did not undergo significant cation exchange, and when mixed with the stream
water, resulted in the large decrease in calcium and magnesium concentration in
the stream at site S2 relative to site S3. This flow path also is a possibility because
it is likely that some water could flow beneath the berm due to the potentially
higher hydraulic conductivity of the streambed material beneath the berm relative
to the hydraulic conductivity of the berm material. Another explanation for the
change in stream water chemistry between sites S3 and S2 is the ‘valence dilution
effect’ associated with cation exchange reactions (Bohn et al., 1985). At elevated
concentrations of salt, the ratio of monovalent to divalent cations is increased on the
soil/sediment surface. With dilution of the salt between sites S3 and S2, divalent
cations are preferentially retained over monovalent cations on the soil-exchange
complex, reducing the concentrations of calcium and magnesium in the stream
water relative to sodium. (3) Since chloride concentration decreased from S3 to
S2, and from S2 to S1, a third flow path could have been uncontaminated ground
water that discharged from areas of higher head contiguous to the stream. This flow
path likely occurred along most of the stream reach between the berm and the lake.
Prior to this study, there were concerns that cation exchange in the northeast-
inlet subcatchment was leaching calcium from the soils in the vicinity of the stream.
This study indicates that the bulk of cation exchange takes place within the berm
and along a short reach of the stream upstream of site S2. Downstream of site
S2, the data indicate little cation exchange takes place; all cations decrease nearly
uniformly in response to dilution of the stream from discharge of uncontaminated
ground water. During the first five years following completion of I93, ratios of
sodium to chloride diminished from 2–3 (similar to ratios from the pristine water-
shed at nearby Hubbard Brook (Likens and Bormann, 1995)) to less than 1. Since
1980, ratios have steadily increased from an annual mean of 0.4 in 1980 to 0.7 in
1993 (significant at p < 0.01), indicating that cation exchange may be diminishing
with time. It is possible that the cation exchange in the berm is decreasing with
time due to a limited supply of base cations (calcium, magnesium, potassium) in
the berm material that is being depleted. Glacial deposits in the local area, from
which the berm was constructed, typically contain only 0.2 to 17% calcium by
weight; the average is between 1 and 2% (Billings and Wilson, 1965).
4.6. HYDROLOGY AND CHEMISTRY OF GROUND WATER BETWEEN MIRROR
LAKE AND I93
Water-level measurements at wells located between Mirror Lake and I93 indicate
that ground water flows from the vicinity of I93 to the lake along most of the
eastern shoreline (Figure 10). Springs seeping out of the roadcut east of I93, at an
elevation much higher than the wells west of the highway, indicate that the gradient
across the highway also is toward the lake. Water-table gradients are quite steep,
typically 0.04 between the interstate wells and the northeast stream outlet, and 0.1
196 D. O. ROSENBERRY ET AL.
Figure 10. Water-table altitude (m) and chloride concentration (µeq L−1) east of Mirror Lake,
averaged over study period.
between wells CS-20R and 32. Between the spring east of I93 and well CS-20R the
gradient is especially steep and approaches 0.3. Assuming a hydraulic conductivity
of 1 ×10−4cm s−1(Rosenberry and Winter, 1993) and a porosity of 0.2, ground
water velocity in the area is between 6 and 16 m per year. If salt-contaminated
ground water were flowing toward the lake it would take about 7 to 18 yr to reach
the easternmost shoreline of Mirror Lake, assuming that flow was restricted to the
glacial deposits. If flow were in the fractured bedrock, arrival times could be much
shorter or much longer, depending on the number, orientation, and connectivity of
fractures in the underlying bedrock.
A lack of wells east of well COWT precludes a determination of hydraulic gra-
dients along the southernmost part of the eastern shoreline of the lake. However,
surface topography, and the presence of Hubbard Brook and the Pemigewasset
River to the south and east, suggest that flow is either to the south or southeast,
which would prevent salt-contaminated ground water from reaching the lake along
the southern part of the eastern shoreline.
MOVEMENT OF ROAD SALT TO A SMALL NEW HAMPSHIRE LAKE 197
Median concentrations of chloride for water from selected wells and wetlands
are shown in Figure 10. Values of chloride larger than the typical 50–70 µeq L−1
were found at most wells. Above-normal values were not detected at well COWT,
which is a shallow water-table well completed in glacial deposits. However, water
samples collected from deeper boreholes in the fractured bedrock beneath well
COWT had chloride values between 110 and 1300 µeq L−1. The chloride concen-
tration indicated in Figure 10 for water collected from well IS1 is from a sample
collected from an isolated interval of bedrock at a depth of 17.4 m below land
surface. Samples were collected from three additional isolated intervals in this
bedrock borehole, at depths of 112, 123, and 143 m below land surface. Chloride
values for these intervals were 1610, 1520, and 70 µeq L−1respectively, indicating
that road-salt contamination of ground water has penetrated at least 123 m into the
bedrock beneath the highway.
It appears that road salt has contaminated the ground water system at several
locations between I93 and the lake, based on above normal chloride concentrations
at wells CS-20R, 32 and wells completed in the bedrock beneath well COWT.
However, since chloride concentration at well CS-20R is only slightly larger than
uncontaminated ground water in the area (typically 50–70 µeq L−1), salt may not
have contaminated all ground water in the glacial deposits between I93 and the
lake. It also is possible that above normal concentrations of chloride in the vicinity
of well 32 was due not to road salt, but to leachate from a small septic-system
drainfield. However, other data do not support the septic-system source for ground
water near well 32. Septic leachate usually exhibits larger values for ammonium,
phosphorus, and nitrate, and concentrations of all three constituents were small or
undetectable at well 32. Hydraulic heads in the bedrock near I93 are much higher
than heads near and in Mirror Lake, and it is possible that water contaminated with
road salt could move through a network of interconnected fractures in the bedrock,
and discharge to the glacial deposits in the vicinity of well 32, where glacial de-
posits are only a few meters thick. If this were happening, larger concentrations of
sodium and chloride would be present near well 32, but not pervasive in the glacial
deposits between well 32 and I93.
4.7. HYDROLOGY AND CHEMISTRY ALONG THE EASTERN SHORELINE OF
MIRROR LAKE
Directions of seepage along the eastern shoreline of Mirror Lake, as measured by
the hydraulic potentiomanometer, are shown in Figure 11. Each arrow in Figure
11 represents one or more measurements made at each sampling transect; a total
of 66 sampling locations are presented by the arrows shown along the eastern and
northeastern shoreline of Mirror Lake. At most sampling transects, measurements
were made at 3 to 4 distances from shore. An attempt was made to insert the probe
consistently to 0.5 m below the lakebed, but at some locations a shallower insertion
was required due to the presence of rocks and boulders. At some of the sampling
198 D. O. ROSENBERRY ET AL.
Figure 11. Primary direction of flow, and reaches of elevated sodium and chloride concentrations,
along the eastern and northeastern shoreline of Mirror Lake, summer 1994.
transects, seepage direction changed with distance from shore. These small-scale
reversals in seepage often are induced by transpiration from near-shore plants.
In these situations, the direction that was indicated by the majority of sampling
locations was the direction indicated by the arrow on the map. Also, some of the
sampling locations were visited two to three times during July and August 1994. In
cases where the seepage direction was consistent throughout all visits, only one
arrow is indicated for that location, but for cases where the direction reversed
from one visit to the next, two arrows, spaced very closely together but pointing in
opposite directions, are shown on the map.
Vertical hydraulic-head gradients beneath the lakebed generally were small along
the Mirror Lake shoreline between well 32 and the no-longer-functional old crib
dam near the outlet of the lake (Figure 11). Many measurements were near the
MOVEMENT OF ROAD SALT TO A SMALL NEW HAMPSHIRE LAKE 199
Figure 12. Concentrations of sodium and chloride along eastern and northeastern shoreline of Mirror
Lake, summer 1994, in µeq L−1.
detection limit of the probe (about 0.002). In the vicinity of the old crib dam, gra-
dients were large (0.01 to 0.03) and they uniformly indicated seepage from the lake
to ground water. However, along most of the rest of the eastern shoreline of Mirror
Lake, gradients indicated seepage from ground water to the lake. Along the section
of shoreline between well 32 and rocky point, seepage direction was variable and
gradients were small. North of rocky point gradients were larger (0.003 to 0.02),
and northwest of the northeast stream inlet gradients were especially large (0.02
to 0.035). Data from all transects in this shoreline reach, except one located at the
stream inlet, indicated seepage into the lake; gradients at the transect near the inlet
that indicated flow in the opposite direction were questionable because they were
so small.
The hydraulic potentiomanometer also was used to determine gradients and
collect water-chemistry samples at a few other shoreline locations along the west
and south shorelines. Seepage directions, as indicated in Figure 11, were consistent
with data collected in the late 1980s by Asbury (1990).
Median sodium and chloride concentrations of samples collected with the po-
tentiomanometer are plotted with distance along the shoreline in Figure 12, be-
200 D. O. ROSENBERRY ET AL.
ginning with 0 at the old crib dam and ending with 730 m along the midpoint of
the northern shoreline of the lake. Two areas of above-normal sodium and chloride
concentration are indicated in the plot, as well as on Figure 11; one a 50-m shore-
line segment in the east cove (distance 220–270), and the other a 70-m shoreline
segment bracketing the northeast stream inlet (distance 500–570). At the east-cove
shoreline segment, sodium concentration is about 1.7 times the concentrations on
either side of the segment, and chloride concentration is about 3.7 times larger
than concentrations on either side of the segment. At the contaminated segment
near the northeast stream inlet, sodium concentrations are more than 10 times the
concentrations adjacent to the segment, and chloride concentrations are more than
40 times concentrations adjacent to the segment.
Average sodium and chloride concentrations between the old crib dam and the
east cove were 1.5 and 1.9 times as large as along the northern shoreline directly
west of the contaminated segment near the northeast inlet. Sodium and chloride
concentrations along the uncontaminated northern shoreline are similar to concen-
trations in other uncontaminated ground water in the western part of the Mirror
Lake watershed. It is possible that the shoreline segment north of the old crib dam
has been contaminated by septic systems from seasonal cabins located along that
shoreline. However, as mentioned earlier, these systems are small and seasonal,
and concentrations of ammonium, phosphorus and nitrate along that segment of
shoreline either are very low or undetectable, indicating that septic leachate is not
the cause of these slightly greater-than-normal sodium and chloride values. These
concentrations also are approximately equal to present-day sodium and chloride
concentrations in Mirror Lake. Frequent reversals in direction of seepage along this
segment of shoreline may have contributed to sodium and chloride concentrations
in the near-shore porous media that are similar to concentrations in the lake.
4.8. MASS-BALANCE APPROACH
The sector of I93 that includes the Mirror Lake watershed is the most heavily salted
sector of the highway in New Hampshire (State of New Hampshire written com-
munication, 1995). During the five-winter period from 1986 to 1991, application
of sodium-chloride salt varied from 23.2 to 33.7 metric tons per lane mile (State
of New Hampshire written communication, 1995). The average annual application
(1986–1991) to the section of highway that includes the Mirror Lake watershed
was 44.6 metric tons of salt, about 27 metric tons of which was chloride. The
annual flux of chloride past site S1 during 1988-1993, based on stream discharge
measurements and weekly chloride concentration measurements, was about 0.7
metric ton. Using chloride as an indicator of salt flux, less than three percent of the
average annual application of road salt to the Mirror Lake watershed has reached
Mirror Lake via the northeast inlet. The rest of the salt either was diverted as surface
flow past the berm and out of the watershed, is transferred to ground water, or
MOVEMENT OF ROAD SALT TO A SMALL NEW HAMPSHIRE LAKE 201
TABLE II
Total annual mass of chloride for components of Mirror Lake chloride budget,
1989-1993, in kg yr−1
1989 1990 1991 1992 1993 Ave.
Outlet 673 406 728 755 803 673
Ground water 838 330 903 890 973 787
Fluxes out Total out 1511 736 1631 1634 1776 1460
West inlet 325 619 383 378 237 388
Northwest inlet 302 428 286 267 237 304
Northeast inlet 677 954 758 750 708 769
Precipitation 434223436243
Ground water 35 36 35 35 35 35
Fluxes in Total in 1382 2079 1485 1473 1279 1539
Budget balance –129 1343 –146 –172 –497 79
stored in the soil. These data indicate that the berm actually was quite effective at
preventing salt from reaching Mirror Lake.
A chloride budget for Mirror Lake was calculated for 1989–1993 to determine
chloride input from the northeast stream relative to other sources of chloride to
the lake. Weekly data were used to calculate stream fluxes and atmospheric fluxes,
and monthly data were used to calculate ground water fluxes. Assumptions were
that the lake was uniformly mixed regarding chloride concentration, and that the
concentration of chloride that seeped into the ground water and/or that flowed out
of the lake via the outlet stream was the same as the concentration in the lake.
Dry deposition also was considered to be insignificant based on comparisons of
chloride concentration in precipitation collected within the Mirror Lake basin and
farther west in the Hubbard Brook valley.
Annual summaries of the results are shown in Table II. The 5-yr average in-
crease in chloride in the lake, based on chloride fluxes to and from the lake, is 79
kg yr−1. As a comparison, the 5-yr increase in mass in the lake, based on the change
in lake concentration multiplied by the lake volume, is 72 kg yr−1. The long-term
increase in chloride in the lake, based on a best fit of lake chloride concentration
data for the period 1974–1993 and using an average lake volume of 8.62 ×108l
(Figure 2), was 104 kg yr−1. (For the period 1980–1994, the long-term increase in
chloride in the lake was 82 kg yr−1). While the 5-yr-average data compare quite
well with loading based on change in measured concentrations of chloride in the
lake, for four of the five years chloride budgets indicated a loss of chloride. These
202 D. O. ROSENBERRY ET AL.
budget results could indicate that a source of chloride exists for the lake that we
have not yet observed or measured.
Budget data indicate that the northeast stream provides half of all chloride inputs
to the lake. While streamflow from the northeast stream accounts for only 3% of all
streamflow to the lake, chloride input from the northeast stream accounts for 53%
of all inputs from the streams and 50% of all inputs to the lake when precipitation
and ground water also are included.
5. Future Salt Loading to the Northeast Stream and Mirror Lake
Since 1974, when increases in sodium and chloride first were detected, the concen-
tration of sodium in the northeast inlet increased at an average annual rate of 70 µeq
L−1yr−1and chloride increased at an average annual rate of 94 µeq L−1yr−1, based
on simple linear regression (Figure 3). If the concentration of sodium and chloride
were to continue to increase at the same rates, average annual concentration of
sodium could be expected to be near 1900 µeq L−1by the yr 2000 and 3600 µeq
L−1by the yr 2025. Average annual concentration of chloride could be expected to
be near 2700 µeq L−1by the yr 2000 and 5000 µeq L−1by the yr 2025.
Eventually, concentrations of sodium and chloride in the northeast inlet will
begin to stabilize as pore spaces in the soil adjacent to and beneath the headwaters
region of the northeast inlet become saturated with salt-laden ground water. There
is some indication that the increase in concentration of sodium and chloride in
the northeast inlet is abating slightly, 23 yr after salt loading began at I93. Linear
regressions of sodium and chloride concentrations in the northeast stream, based
on 1974–1984 data, indicate an annual increase of 64 µeq L−1for sodium and
180 µeq L−1for chloride. Linear regressions based on 1985–1994 data indicate
annual increases of 73 µeq L−1for sodium, but only 87 µeq L−1for chloride.
The larger annual increase of sodium during the latter period indicates that cation
exchange sites are being depleted in the northeast headwaters region, decreasing
the capability of sodium to be exchanged primarily with calcium, but the smaller
annual increase of chloride for the latter period indicates that the concentration of
salt in the soil and ground water near the headwaters area is closer to equilibrium
with the rate of application of road salt to highway I93.
If the trend of increasing sodium and chloride concentration in the northeast
inlet were to continue, the trend of increasing sodium and chloride concentration
in Mirror Lake would continue also. Linear regression of data collected from the
lake from 1974–1993 indicates an average annual rate of increase of 2.5 µeq L−1
yr−1for sodium and 3.4 µeq L−1yr−1for chloride (Figure 2). A continued rate of
increase would result in a concentration of 118 µeq L−1for sodium and 116 µeq
L−1for chloride by the yr 2000, and a concentration of 181 µeq L−1for sodium
and 201 µeq L−1for chloride by the yr 2025. By the yr 2025 the concentration of
MOVEMENT OF ROAD SALT TO A SMALL NEW HAMPSHIRE LAKE 203
sodium would be triple the concentration in the lake prior to construction of I93,
and the concentration of chloride would have increased 8 fold.
The rate of increase in concentration of salt in Mirror Lake may be diminishing
also, as in the northeast stream water, but evidence for a smaller annual increase in
concentration in the lake during the past 10 yr is more subtle than in the northeast
stream water. The slope of the linear-regression lines through the data collected
from Mirror Lake during 1974–84 indicate an annual increase of 2.9 µeq L−1for
sodium and 4.5 µeq L−1for chloride. This compares with annual increases of 2.6
µeq L−1for sodium and 3.0 µeq L−1for chloride, based on the 1985–1994 data.
The smaller annual increase in chloride for the latter period also likely is influenced
by the dilution that occurred following the very wet period during 1990.
6. Summary and Conclusions
(1) Road salt moves to Mirror Lake primarily via the northeast inlet stream. How-
ever, less than three percent of the 44.6 metric tons of salt applied each year to I93
within the Mirror Lake watershed enters the lake via the northeast stream. A zone
of contaminated ground water directly beneath the streambed, at least 0.6 m. thick,
also may transfer road salt to the lake, but at a much slower rate. Ground water
5 m on either side of the stream does not appear to be contaminated with road salt,
except for an area directly west of the stream extending from the lake to at least
40 m upstream. Road salt evidently is not seeping through a broad reach of the
highway diversion berm, but appears to be seeping beneath a portion of the berm,
discharging near the headwaters of the northeast inlet stream.
(2) Chloride concentrations in the northeast stream decrease downstream from
the diversion berm, indicating that, during winter and spring, uncontaminated
ground water discharges into the stream and dilutes the stream water. However,
gradients between wells and the stream along two transects perpendicular to the
stream are away from the stream during summer months, preventing this dilution
during the summer. Consequently, chloride levels commonly reach a maximum
concentration in the stream during the late summer months, when gradients reverse
and salt-laden water is flushed back into the stream.
(3) Water-table depressions adjacent to the northeast inlet stream caused by tran-
spiration, along with more highly conductive porous media near the lower reach of
the stream, have led to contamination of ground water adjacent to the stream on
the side of the stream opposite the interstate highway. This contaminated ground
water enters the lake along an approximately 70-m-wide shoreline segment that is
centered near the stream inlet. Concentrations in the contaminated ground water are
as much as an order of magnitude greater than concentrations in uncontaminated
ground water.
(4) It is likely that cation exchange takes place between the ponded water east of
the berm and the seeps west of the berm, which causes a reduction of sodium and
204 D. O. ROSENBERRY ET AL.
an increase in calcium, relative to the ponded water, in the stream water directly
west of the berm. Little additional cation exchange takes place downstream of the
berm.
(5) Chloride concentrations in nearly all wells between I93 and Mirror Lake
are large compared to chloride concentrations west of the lake. Steep hydraulic
gradients indicate ground water flow toward the lake along most of the eastern
shoreline of the lake. Large chloride concentrations are pervasive in the bedrock
adjacent to I93, to a depth of at least 123 m, but large chloride concentrations were
not pervasive in the glacial material above the bedrock.
(6) Sodium and chloride concentrations were large along a 50-m segment of
the easternmost shoreline of Mirror Lake, and along a 70-m segment of shoreline
centered at the northeast inlet. Concentrations of sodium and chloride were only
slightly larger than normal at the easternmost shoreline segment, but near the north-
east stream they were more than 10 times the concentrations of sodium and more
than 40 times the concentrations of chloride found along uncontaminated shoreline
segments. Slightly-above-normal concentrations of sodium and chloride along the
southern portion of the eastern shoreline of Mirror Lake were about equal to con-
centrations in the lake water. Therefore, it is suspected that occasional reversals in
the direction of seepage have contributed to greater levels of sodium and chloride
in ground water adjacent to the southern end of the eastern shoreline.
(7) Half of all the chloride that enters the lake comes from the northeast inlet,
which contributes only 3% of total streamflow into the lake.
(8) If current rates of increase in concentration of salt were to continue in the
northeast inlet and in the lake, the concentration of sodium in the lake would triple
and the concentration of chloride would increase by 8 times by the yr 2025, relative
to pre-highway concentrations.
Acknowledgments
The authors thank D. Tunnell for collecting lake-shore hydraulic and water-quality
data, and J. Crowdes and E. Morency for collection of stream water and wellwater
samples. N. Caraco, C. Driscoll, G. Granato, J. LaBaugh and W. Martin provided
helpful suggestions for improvement of the manuscript. The authors also thank
the USDA Forest Service for allowing access to the Hubbard Brook Experimental
Forest. The Hubbard Brook Experimental forest is operated and maintained by
the Northeastern Forest Experimental Station, USDA Forest Service, Radnor, PA.
Financial support was provided by the National Science Foundation and the A.W.
Mellon Foundation.
MOVEMENT OF ROAD SALT TO A SMALL NEW HAMPSHIRE LAKE 205
References
Asbury, C. E.: 1990, ‘The Role of Ground-water Seepage in Sediment Chemistry and Nutrient
Budgets in Mirror Lake, New Hampshire’, PhD thesis, Cornell University, 275 p.
Billings, M. P. and Wilson, J. R.: 1965, Chemical Analysis of Rocks and Minerals from New
Hampshire, Part XIX: Mineral resources Survey, New Hampshire Department of Resources and
Economic Development, Concord, New Hampshire, 104 p.
Bohn, H. L., McNeal, B. L. and O’Conner, G. A.: 1985, Soil Chemistry, second edition, John Wiley
and Sons, New York.
Bormann, F. H. and Likens, G. E.: 1979, Pattern and Process in a Forested Ecosystem, Springer-
Verlag, New York, 253 p.
Bormann, F. H. and Likens, G. E.: 1985, in G. E. Likens (ed.), An Ecosystem Approach to Aquatic
Ecology: Mirror Lake and Its Environment, Springer Verlag, New York, pp. 436–444.
Bowser, C. J.: 1992, in F. M. D’Itri (ed.), Chemical Deicers and the Environment, Lewis Publishers,
Chelsea, Michigan, pp. 283–301.
Bubeck, R. C., Diment, W. H., Deck. B. L., Baldwin, A. L. and Lipton, S. D.: 1971, Science 192,
1128.
Cherkauer, D. S. and Ostenso, N. A.: 1976, Water Res. Bull. 12, 1259.
Church, P. E. and Friesz, P. J.: 1993, Transportation Research Record 1420, 56.
D’Itri, F. M. (ed.): 1992, Chemical Deicers and the Environment, Lewis Publishers, Chelsea,
Michigan, 585 p.
Driscoll, C. T., Newton, R. M., Gubala, C. P., Baker, J. P. and Christensen, S. W.: 1991, in Charles,
D. F. (ed.), Acidic Deposition and Aquatic Ecosystems, Springer Verlag, New York, 747 pp.
Hawkins, R. H. and Judd, J. H.: 1972, Water Res. Bull. 8, 1246.
Jones, P. H. and Jeffrey, B. A.: 1992, in F. M. D’Itri (ed.), Chemical Deicers and the Environment,
Lewis Publishers, Chelsea, Michigan, pp. 1–107.
Judd, J. H.: 1970, Water Resources 8, 521.
Likens, G. E. (ed.): 1985, An Ecosystem Approach to Aquatic Ecology: Mirror Lake and Its
Environment, Springer Verlag, New York, 516 p.
Likens, G. E.: 1992, The Ecosystem Approach: Its Use and Abuse. Ecology Institute, Olden-
dorf/Luhe, Germany, 166 p.
Likens, G. E. and Bormann, F. H.: 1995, Biogeochemistry of a Forested Ecosystem, second edition,
Springer Verlag, New York, 159 p.
Likens, G. E., Eaton, J. S., Johnson, N. M. and Pierce, R. S.: 1985, in G. E. Likens (ed.), An Ecosystem
Approach to Aquatic Ecology: Mirror Lake and Its Environment, Springer Verlag, New York, pp.
135–155.
National Research Council: 1991, Highway Deicing, Comparing Salt and Calcium Magnesium
Acetate, Special Report 235, National Research Council, Washinton DC, 170 pp.
Peters, N. E. and Turk, J. T.: 1981, Water Res. Bull. 17, 586.
Pollock, S. J.: 1984, ‘Determining the Effectiveness of Highway Drainage Systems in Preventing Salt
Contamination of Ground Water’, U.S. Geological Survey Water-Resources Investigation Report
84-4166, 22 p.
Rosenberry, D. O. and Winter, T. C.: 1993, Proceedings: XXIVth Congress International Association
of Hydrogeologists–Hydrogeology of Hard Rocks, June 28-July 2, 1993, Ås, Norway, pp. 967–
977.
Shapiro, A. M., Hsieh, P. A. and Winter, T. C.: 1995, ‘The Mirror Lake Fractured-Rock Research
Site – A Multidisciplinary Research Effort in Characterizing Ground-Water Flow and Chemical
Transport in Fractured Rock’, U.S. Geological Survey Fact Sheet FS-138-95, 2 p.
State of New Hampshire: 1995, Annual Records 1987-1991, Department of Transportation, Bureau
of Highway Maintenance, District 3, Guilford, NH.
Wetzel, R. G. and Likens, G. E.: 1991, Limnological Analyses, Springer Verlag, New York, 391 p.
206 D. O. ROSENBERRY ET AL.
White, D. S.: 1993, Journal of the North American Benthological Society 12, 61.
Wilcox, D. A.: 1986, Water Res. Bull. 22, 57.
Winter, T. C.: 1984, ‘Geohydrologic Setting of Mirror Lake, West Thornton, New Hampshire’, U. S.
Geological Survey Water Resources Investigations Report 84–4266, 61 p.
Winter, T. C.: 1985, in Likens, G. E. (ed.), An Ecosystem Approach to Aquatic Ecology: Mirror Lake
and Its Environment, Springer Verlag, New York, pp. 40–53.
Winter, T. C., LaBaugh, J. W. and Rosenberry, D. O.: 1988, Limnol. Oceanogr. 33, 1209.
