ABSTRACT
INTRODUCTION
Thermokarst lakes cover nearly 25% of the landscape in icerich,
permafrost lowlands of Arctic Alaska (Hinkel et al.,
2005), Siberia (Grosse et al., 2005), and Canada (Mackay,
1988; Marsh et al., 2009). Thus, they play a key role in hydrology,
permafrost, carbon, and habitat dynamics in the
continuous permafrost zone (Grosse et al., 2013). Thermokarst
lakes form as a result of permafrost degradation and
ground subsidence, and expand through thermal and mechanical
erosion (Jorgenson and Shur, 2007; Grosse et al.,
2013). They may persist for several thousand years, yet are
susceptible to drainage (Mackay, 1992). Their formation,
expansion, and drainage have actively reshaped ice-rich permafrost
lowlands since the onset of the Holocene (Mackay,
1992; Hinkel et al., 2003; Jones et al., 2012). The conversion
of these aquatic ecosystems to terrestrial or wetland ecosystems
is thought to be very rapid, as a result of catastrophic
drainage (Mackay, 1981).
Drained thermokarst lake basins may presently occupy as
much as 50% to 75% of the ice-rich permafrost landscape in
Arctic lowlands (Hinkel et al., 2003, 2005; Grosse et al.,
2005; Jones et al., 2012). Remote sensing analyses of the
continuous permafrost zone indicate that roughly 1 to 2 lakes
may drain annually in several Arctic regions (Mackay, 1988;
Hinkel et al., 2007; Marsh et al., 2009; Jones et al., 2011).
Lake drainage in the continuous permafrost zone is thought
to occur laterally and result from a number of different geomorphic
and hydrologic processes. Factors that likely initiate
drainage include (1) tapping of lakes by rivers, streams, adjacent
lakes, or the sea; (2) headward gully erosion; (3) anthropogenic
disturbance; (4) thaw slump formation; and (5)
bank over-topping resulting from elevated water levels following
heavy precipitation events and/or snow-damming
of outlets (Mackay, 1988; Brewer et al., 1993; Mackay,
1992; Weller and Derksen, 1979; Hinkel et al., 2007; Wolfe
and Turner, 2008; Marsh et al., 2009; Jones et al., 2011;
Grosse et al., 2013).
Mackay (1988) noted that elevated lake water levels and
diversion of water through interconnected ice-wedge systems
likely played a major role in the occurrence of
*
catastrophic drainage events. Mackay (1992) indicated that
these conditions were most likely to occur during the snowmelt
period and thus lakes would likely tend to drain during
early summer. An observation of a catastrophic drainage
event in the late 1980s in northern Alaska by an Inupiaq
Elder indicated that a lake likely drained during this early
season period because abundant lake ice was observed in
the lake prior to it overtopping its bank and catastrophically
draining (Hinkel et al., 2007). Wolfe and Turner (2008) reported
the partial drainage of a thermokarst lake in the Old
Crow Flats, sometime between 06 June and 23 July 2007,
that likely resulted from near record precipitation between
March and May that increased the lake water level and triggered
erosion of a drainage outlet (Turner et al., 2010;
Turner et al., 2014). Brewer et al. (1993) noted the sudden,
natural drainage of a thermokarst lake in northern Alaska in
1989 due to elevated lake water levels and over-topping of
the lake along an ice-wedge trough, but in this case the lake
drained later in the summer, presumably as a result of heavy
late-summer rains. Marsh et al. (2009) also reported on the
catastrophic drainage of a lake in late summer and suggested
that deeper than normal active layers, thaw slump
formation, and moderately high lake levels were the primary
drivers of catastrophic lake drainage in the western Canadian
Arctic. Although these observations are important for
characterizing the mechanisms and controls on catastrophic
drainage events, they all lacked in situ observations in the
lake leading up to and during the drainage event, and so
they provide little environmental context.
Previously, the most detailed observations from a catastrophic
drainage event come from an experiment to study
permafrost aggradation below Lake Illisarvik on Richards
Island in western Arctic Canada by J.R. Mackay (1981,
1992, 1997). Mackay (1981) excavated a 25 m long ditch
down to the frost table in late June 1978, from the sea coast
to within 20 m of the pre-drainage lake margin. Water was
pumped from the lake and down the artiﬁcial ditch, creating
a 1.5 m deep channel. The ditch margin was allowed to
thaw in July and then deepened in August to a depth of 4
m at the coast and tapering up to a depth of 1.5 m within
a few meters of the pre-drainage lake margin. On 12 and
13 August 1978, the remaining distance between the lake
and the artiﬁcial channel was ditched and channelized, and
the lake drainage was initiated. Within 4 hours the lake level
had dropped more than 1 m and within 10 hours the lake
level had dropped by 2 m, after which measurements of lake
level decline were hindered by a soft lake bottom (Mackay,
1981; Mackay, 1997). Peak discharge during the drainage
event was estimated by Marsh and Nuemann (2001) to be
36 m3
/s. This human-induced drainage experiment provided
the ﬁrst evidence that catastrophic drainage events are capable
of producing peak discharges that exceed snowmeltgenerated
peak ﬂows of the spring freshet in Arctic headwater
basins (Marsh and Neumann, 2001; Marsh et al., 2008).
However, because the ultimate goal of this experiment focused
on the growth of permafrost in the unfrozen lake sediments
following drainage (Mackay, 1997) and not
necessarily the drainage itself, the degree to which this
human-induced lake drainage reﬂected the duration and
magnitude of a natural drainage event remained uncertain.
Despite the prevalence of thermokarst lakes and drained
thermokarst lake basins in the Arctic, detailed observations
of the environmental conditions preceding a natural
catastrophic drainage as well as in situ observations on
the duration and magnitude of a natural drainage event
have previously not been recorded. In this study, we describe
the preceding conditions, timing, and processes associated
with a natural thermokarst lake drainage event in
northern Alaska during the early summer of 2014. We
place these observations in the context of lake drainage
events and climatic conditions occurring in the study
region since 1955. Our observations from a natural,
thermokarst lake drainage event may provide useful information
for predicting future catastrophic lake drainage
events and their broader impacts.
STUDY AREA
The Arctic Coastal Plain of northern Alaska contains the
second largest lake-district in the state (Arp and Jones,
2009). Lakes in the northern portion of this region, the
younger outer coastal plain (Hinkel et al., 2005), are classiﬁed
as “true” thermokarst lakes (Jorgenson and Shur,
2007). The permafrost is ice-rich, with an estimated volumetric
ground-ice content that exceeds 80% for both primary
landscape surfaces and drained lake basins
(Kanevskiy et al., 2013). Excess ice content in the ground
is typically restricted to the upper few meters of permafrost
(Sellmann et al., 1975; Kanevskiy et al., 2013) and thus the
majority of the lakes on the younger outer coastal plain
range from 1 to 3 m in depth (Hinkel et al., 2012). Previous
estimates from the region indicate that lake surface area
comprises 22.6% of the landscape and drained lake basins
account for an additional 46.6% of the landscape (Hinkel
et al., 2005). The mosaic of lakes and drained lake basins
indicates that active landscape processes associated with
permafrost degradation and aggradation have operated
throughout the Holocene (Hinkel et al., 2003; Jorgenson
and Shur, 2007).
Lake 195 (L195), the lake that catastrophically drained in
2014, is located approximately 170 km southeast of Barrow,
on the younger outer coastal plain of northern Alaska
(Figure 1). This thermokarst lake expanded at a rate of 0.5
m/yr between 1979 and 2002, preferentially at the northern
and southern margins (~1 m/yr) (Arp et al., 2011). The segment
of Arctic coastline adjacent to L195 has also experienced
an increase in erosion rates since the early 2000s
(Jones et al., 2009a, 2009b). Thus, the lake was thought to
have a high likelihood for drainage and we hypothesized
that it would drain by the year 2020 as a result of tapping
by coastal erosion (Arp et al., 2010). Prior to drainage it
had a surface area of ~80 ha, a maximum water depth of
1.4 m, a mean water depth of 1.1 m, and an estimated water
Copyright © 2015 John Wiley & Sons, Ltd.
volume of 871,990 m3
. L195 is also considered critical habitat
for molting Black Brant (Branta bernicla nigricans) as
this migratory Arctic goose has shifted its habitat to coastal
inﬂuenced lakes since the 1970s (Flint et al., 2008).
METHODS
In situ Observations
In anticipation of the drainage of L195, we established a
lake data buoy in July 2009 that measured water level and
water temperature data at a 2-hr interval. The 2-hour interval
was set based on the expectation that the lake would drain
around the year 2020 as a result of tapping by coastal erosion
and thus within the memory limitations of the data logger.
Hobo U20-001-01 data loggers recorded pressure and
temperature of the lake bed. Atmospheric pressure was also
recorded with a Hobo-U20-001-01 data logger. This was
used to convert the lake bed data logger measurements to
a water level measurement (+/- 0.5 cm).
A meteorological station that measures air temperature,
wind speed and direction, precipitation, snow depth, soil
moisture, ground temperature, surface pressure and solar radiation
was established within 10 km of L195 in August
2007 (Urban and Clow, 2014). The observations from the
weather station provided information on the environmental
conditions preceding the catastrophic lake drainage event.
We utilized air temperature data measured with a Campbell
Scientiﬁc CSI model 107 thermistor probe, rainfall data
measured with a Texas Electronics TE 525 rain gauge positioned
60 cm above the ground within an ETI Instrument
Systems Lexan altershield, and snow depth data measured
with a CSI model SR50 ultrasonic distance sensor mounted
2.5 m above the ground surface.
Lake basin geometry was surveyed after the drainage
event, on 14 August 2014, using a Differential GPS system
with a vertical accuracy of 1-2 cm. Additional data points
Figure 1 (a) MODIS satellite image showing a portion of the Arctic Coastal Plain of northern Alaska and key landscape divisions indicated with a dashed
white line: YOCP is younger outer coastal plain, OCP is outer coastal plain, ICP is inner coastal plain, and AF is the Arctic foothills (modiﬁed from Hinkel
et al., 2005). The study area is outlined with the yellow polygon and the white box indicates the location of L195. Landsat OLI image pair from (b) 12 July
2013 and (c) 15 July 2014 showing the drainage of L195. This ﬁgure is available in colour online at wileyonlinelibrary.com/journal/ppp.
121
were added in a GIS environment based on common basin
geometries from other outer coastal plain lakes (Hinkel
et al., 2012) and a bathymetric map was created using a triangulated
irregular network (TIN) point to raster transformation.
Change in lake volume during the drainage event
was calculated using the Grid-Volume tool in Surfer 10™
relative to recorded changes in lake depth measured by the
water level logger. Lake outlet discharge was estimated to
equal change in storage during our 2 hour measurement interval
and as a result our observed peak discharge is likely
to be lower than the true peak discharge. Thus, the peak instantaneous
discharge was estimated based on a linear regression
of peak discharge values observed at 2 hour and
lower recording intervals (i.e. 4, 6, and 8 hours) and approximated
at the zero intercept.
Remotely Sensed Imagery Analysis
Lakes, drained lake basins, and other landscape features in
the study area (Figure 1) were mapped using a cloud-free,
Landsat TM image acquired on 21 August 2010. The image
was clipped to the younger outer coastal plain north of
Teshekpuk Lake. Extant lakes were automatically classiﬁed
using Landsat TM band 5 and an object-oriented classiﬁcation
technique (Frohn et al., 2005). The remainder of the image
space was classiﬁed as land and individual drained lake
basins were manually delineated in a GIS using the Landsat
TM image band combination 5-4-3 (Jones et al., 2012) as
well as a 5 m resolution IfSAR-derived digital terrain model
(Wang et al., 2012). All data are reported to a minimum
mapping unit of 10 ha and the drained lake basin dataset
was updated to include the drainage of L195 in 2014.
Lakes evident in U.S. Geological Survey topographic
map sheets provided the baseline for determining lake
drainage events in the study region since 1955. Cloud-free
Landsat imagery available for the study region, beginning
in 1977 and then every six to eight years since, was assessed
to determine lake drainage events between 1955 and 1977,
1978 and 1985, 1986 and 1992, 1993 and 2000, 2001 and
2007, and 2008 to 2014. Lake surface area was also derived
from Landsat imagery for L195 on an annual basis between
2007 and 2013.
Long-term and Short-Term Climate Data Analysis
In order to place the L195 drainage event in the context of
preceding weather and climate conditions, we analyzed air
temperature, snow depth, and rainfall data collected at a meteorological
station located 10 km from the lake (Urban and
Clow, 2014). Since L195 drained in early July, following a
period of abnormally wet and cold weather that we hypothesized
had initiated this catastrophic drainage event, we
summarized the average air temperature and total rainfall
from 1 June to 15 July 2014. This was compared to the same
early summer period going back to 2008, when rainfall data
were ﬁrst collected from this station. Late winter snow depth
(1 May) from this station was converted to a snow-water
equivalent, assuming a snow density of 0.20 g/cm3
and
50% loss to sublimation, providing an estimate of snowmelt
runoff (Arp et al. 2011). Snow-water runoff was then combined
with total rainfall over this period to represent early
season recharge to L195 between 2008 and 2014.
In addition, we used meteorological data recorded at
Barrow (National Weather Service, Station ID 700260
27502) to place the early summer conditions into a longer
term context and to compare this record to other lake drainage
events north of Teshekpuk Lake since 1955. Data from
this station, located 170 km northwest of L195, were also
summarized by average air temperature from 1 June to 15
July for the period 1955 to 2014. Total precipitation
(snowfall and rainfall) was also summarized over the same
period. The climate data from Barrow were then analyzed
according to the time periods that bracketed lake drainage
events in the study area north of Teshekpuk Lake. Barrow,
which is also on the younger outer coastal plain, is considered
to have very similar climatology as the area north of
Teshekpuk Lake (Arp et al. 2011), though day to day
weather patterns can vary between these locations.
RESULTS
During the early summer of 2014 we observed a natural,
catastrophic thermokarst lake drainage event using a data
logger deployed in a lake that we anticipated would drain
during the next decade. These observations were coupled
with observations from a nearby meteorological station providing
information on the environmental conditions that
preceded the drainage event. The 2014 lake drainage event
was then placed in the context of lake drainage events and
early summer weather conditions in the region since 1955.
2014 Catastrophic Lake Drainage Event
Analysis of Landsat OLI imagery shows that L195 drained
between 12 July 2013 (Figure 1b) and 15 July 2014
(Figure 1c). Annual Landsat-based observations between
2007 and 2013 reveal a ﬂuctuating, yet increasing surface area
for L195 prior to drainage (Figure 2a). Measurements from
the lake data buoy shows that L195 began to drain on 05 July
2014 and almost entirely emptied by 07 July 2014, with 75%
of the water volume loss occurring in the ﬁrst ten hours
(Figure 2b). Unusually intense early summer rainfall events
were recorded on 21 and 22 June before complete ice-out,
as well as several smaller events in the week preceding
drainage (Figure 2b). Water level data show a slight increase
in the depth of L195 following ice-off, corresponding to these
precipitation events. From 01 June to 15 July 2014, 31.5 mm
of rainfall were recorded, which was over four times higher
than the average rainfall recorded for the period between
2008 and 2013 (6.2 mm). In addition, the combination of
rainfall and estimated snowmelt runoff indicates that the early
summer water input to the lake in 2014 was more than twice
that of the 2008-2013 average (Figure 2a).
Copyright © 2015 John Wiley & Sons, Ltd.
The wetter than normal early summer conditions in 2014
likely resulted in bank overtopping, which promoted thermal
and mechanical erosion along an ice-wedge network and the
drainage of L195 (Figure 3). Erosion and melting of ice during
the drainage event formed a thermo-erosional gully that
was 9 m wide, 2 m deep, and 70 m long. It is likely that this
gully was completely eroded during the drainage event because
we visited the site within ten days of drainage and
again 40 days after drainage and documented no noticeable
changes to the gully. Based on these measurements and the
estimated water volume of 871,990 m3
prior to drainage,
we estimate that the drainage resulted in a measured peak
ﬂow of 25.0 m3
/s and an estimated, peak instantaneous discharge
of 29.4 m3
/s, or 17.5% higher than was recorded
(Figure 4). Although the drainage event lasted for 36 hours,
75% of the lake discharge occurred in the ﬁrst ten hours.
Landscape Features and Drainage Events Since 1955
Thermokarst lakes (22.5%) and drained thermokarst lake basins
(61.8%) currently account for ~84% of the landscape
area on the younger outer coastal plain north of Teshekpuk
Lake (Figure 5). Between 1955 and 2014, nine lakes
completely or partially drained (>25% loss in lake surface
area) or 0.17 lakes drained per year. Of these nine lakes, four
drained between 1955 and 1977, two between 1977 and
1985, one between 1985 and 1992, one between 1992 and
2000, none between 2000 and 2007, and one between
2007 and 2014. This corresponds to a drainage rate of
0.18, 0.25, 0.14, 0.13, 0, and 0.14 lakes per year,
respectively.
Lakes drained into three different settings. Two of the
lakes drained directly into the sea, ﬁve drained into adjacent
lakes, and two drained into river systems. It appears that one
of the lakes was tapped by the lateral expansion of
Teshekpuk Lake (Weller and Derksen, 1979), one directly
by coastal erosion, whereas the other seven appear to have
drained through a thermo-erosional gully that presumably
developed as a result of water spilling out of the lake. The
average length of seven thermo-erosional drainage gullies
was 160 m but ranged from 30 m to 500 m.
Long-Term Climatic Conditions
The weather station at Barrow provides the best long-term
regional dataset available during the observation period
(1955 to present). Analysis of early summer air temperature
and precipitation data (01 June to 15 July), within the time
periods constrained by the U.S. Geological Survey topographic
map and the Landsat imagery used in the study,
shows varying patterns of precipitation relative to air temperature
(Figure 6). Early summer air temperatures in
2014 were cool (1.9°C) compared with the 1955 to 2013 average
(2.4°C). Total precipitation at Barrow during 2014
was 39.9 mm, the sixth highest amount recorded over the
last 59 years, and twice the average between 1955 and
2013. During 2013, the year before L195 drained, total precipitation
was 34.3 mm, which was also wetter than normal,
but was considerably warmer (4.7°C) relative to the longterm
mean (Figure 6).
DISCUSSION
Field Observations of Drainage Events
Although catastrophic lake drainage events are cited as a
common Arctic landscape process, in situ observations of
the timing and discharge associated with a natural drainage
have previously not existed. Mackay (1981, 1997) induced
and directly observed the catastrophic drainage of Lake
Illsarvik in 1978 by excavating an artiﬁcial drainage channel
in ice-rich permafrost. This experiment provided the ﬁrst direct
data on the duration and magnitude of a catastrophic
thermokarst lake drainage event in the Arctic. Subsequently,
Marsh and Neumann (2001) constrained the duration of a
natural lake drainage from a downstream gauging station.
Their data indicate that the event likely lasted 16 hours; however,
due to instrument failure the magnitude of the event
was not captured. Our observations of a natural drainage
event indicate that for an 80 ha lake with a water volume
Figure 2 Hydroclimatic conditions prior to the catastrophic drainage of
L195. (a) Lake area derived from mid-summer Landsat data between 2007
and 2013 (open circles and dashed line) plotted along with estimated early
summer (1 June to 15 July) snowmelt runoff and rainfall (blue squares and
dashed line) and air temperature (red triangles and dashed line) from 2008
to 2014. (b) Changes in lake surface elevation leading up to and during the
drainage event (black line), lake ice-out (black dashed line) lake water temperature
following ice-off (red line), and rainfall events (blue bars) between 21 June
and 07 July 2014. This ﬁgure is available in colour online at wileyonlinelibrary.
com/journal/ppp.
123
Copyright © 2015 John Wiley & Sons, Ltd.
of 871,990 m3
, that drainage can indeed be catastrophic,
with the majority of it occurring in the ﬁrst ten hours.
The observation network indicates that the cooler and wetter
than normal early summer conditions in 2014 were likely
responsible for the drainage of L195. The drainage was
likely a result of bank overtopping and thermo-erosion along
a course of interconnected ice wedges from the lake margin
to the sea. Mackay (1988) indicated that thermo-erosion
along an ice-wedge network due to lake bank overtopping
was one of the most likely pathways controlling catastrophic
lake drainage events. Since the drainage of L195 occurred in
the early summer, limited seepage was likely occurring
through the active layer prior to drainage, which could have
allowed the lake to ﬁll and spill over and erode the drainage
pathway. However, we cannot rule out the possibility of subsurface
thermal erosion and resulting ice-wedge tunneling as
playing a role in the drainage of L195 (Marsh et al., 2009).
Our ﬁndings support the notion of Mackay (1992) that
early summer weather conditions likely have a major inﬂuence
on catastrophic lake drainage events. Our sensor
Figure 3 (a) Oblique aerial photographs of L195 in 2009 (top) prior to drainage and in 2014 (bottom) following drainage. (b) Ground photographs of a thermoerosional
gully (black boxes in Figure 3a) in 2010 (left) prior to drainage and again in 2014 (right) following the catastrophic drainage of L195. This ﬁgure is
available in colour online at wileyonlinelibrary.com/journal/ppp
Figure 4 Observed outlet discharge generated by the natural drainage of L195
(black circles and line) and the outlet discharge generated by artiﬁcial drainage
of Lake Illisarvik (grey squares and line) (Mackay, 1997; Marsh and Neumann,
2001). Asterisk indicates the estimated, peak instantaneous discharge associated
with the L195 drainage.
observations documenting the drainage of L195, along with
local knowledge observations from an Inupiaq Elder
(Hinkel et al., 2007), and aerial reconnaissance and ﬁeld observations
at Lake Zelma in the Old Crow Flats (Wolfe and
Turner, 2008), further support this notion of early summer
conditions being critical to the catastrophic drainage of
thermokarst lakes. However, the observations of Brewer
et al. (1993) and Marsh et al. (2009) highlight the importance
of late summer conditions as driving some catastrophic
lake drainage events. Better constraining the
seasonality associated with catastrophic lake drainage
events will be important for predicting the response of these
systems to future environmental change in the Arctic.
Remotely Sensing Lake Drainage Events
Reconstructing lake drainage events between 1955 and 2014,
based on analysis of U.S. Geological Survey topographic
maps and Landsat imagery, identiﬁed nine lake drainage
events in the 1,750 km2
study area (Figure 5). While retrospective
analyses of remotely sensed data lack the detailed
observations that we were able to describe prior to and during
the drainage of L195, there appear to be patterns associated
with early season weather conditions and lake drainage events
(Figure 6). Lakes may have a higher tendency to drain when
relatively cool early summer air temperatures and enhanced
early summer precipitation prevail, similar to the conditions
observed during 2014 and the associated drainage of L195.
Hinkel et al. (2007) identiﬁed three lake drainage events
for this area between 1977 and 2000, providing a drainage
rate of 0.13 lakes per year. By extending the period of record
back to 1955 as well as forward to 2014 we have identiﬁed
an additional six lake drainage events, with ﬁve of
these occurring prior to 1977. This seemingly indicates a
Figure 5 Landscape features derived from a Landsat TM image acquired on 20 August 2010 and updated with a Landsat OLI image acquired on 15 July 2014.
Lakes that drained since 1955 are shown in red, older drained basins in grey, extant lakes in blue, and other terrain features in green. This ﬁgure is available in
colour online at wileyonlinelibrary.com/journal/ppp
Figure 6 Mean early summer (01 June to 15 July) air temperature and precipitation
derived from the Barrow, AK, climate station between 1955 and
2014 summarized by the time periods used to bracket lake drainage events in
the study area. The early summer of 2014 experienced 92% greater precipitation
relative to the record we analyzed going back to 1955. This ﬁgure is
available in colour online at wileyonlinelibrary.com/journal/ppp
125
Copyright © 2015 John Wiley & Sons, Ltd.
reduction in the drainage rate of lakes in the study area over
the past ~60 years. Marsh et al. (2009) documented
thermokarst lake drainage events in the western Canadian
Arctic between 1950 and 1973, 1973 and 1985, and 1985
and 2000 and reported a decrease in the drainage rate of
lakes from 1.13 to 0.83 to 0.33 lakes per year, respectively;
they hypothesized that this decrease was linked to changes
in climate during this period. Other studies documenting
lake drainage events over time have reported steady (Jones
et al., 2011) to increasing (Grosse et al., 2009; Lantz and
Turner, 2015) patterns, highlighting the need for more concentrated
and detailed ﬁeld observations and data collection
to determine the processes controlling catastrophic lake
drainage events around the Arctic.
Our landscape-scale assessment of drained lake basins
north of Teshekpuk Lake revealed ~15% greater coverage
than previously mapped for the study area (Hinkel et al.,
2005). This difference is not related to an increase in the
number of lake drainage events since the early 2000s. Instead,
it can be explained by the manual digitization of
multiple overlapping drained basins that the automated
techniques described in Frohn et al. (2005) struggled with,
a cloud-free Landsat image (cloud cover was an issue in
the imagery used by Frohn et al. (2005)), and the use of
a high-resolution digital terrain model for distinguishing
the perimeters of these sometimes subtle landscape
features.
Peak Discharge Associated with Drainage Events
The observed peak discharge of 25.0 m3
/s and the estimated
peak instantaneous discharge of 29.4 m3
/s associated with
the catastrophic drainage of L195 are similar to the 36 m3
/s
peak discharge measured at the much smaller (340,000 m3
)
Lake Illisarvik (Marsh and Nuemann, 2001; Mackay,
1997). Marsh and Nuemann (2001) also provided estimates
of peak discharge for a lake with an estimated volume of
112,500 m3
that catastrophically drained in 1989. Based on
a record from a downstream hydrograph, this drainage occurred
over a 16 hour period, with an estimated peak discharge
of 29 m3
/s likely occurring 6 hours after the onset
of drainage (Marsh and Nuemann, 2001). Thus, these three
observations from lakes that vary in surface area and volume
indicate peak discharge ﬂoods that range from 25 to 36 m3
/s
for catastrophic drainage events. However, lakes and drained
lake basins can be several orders of magnitude larger than
these three examples (Figure 5). Thus, the catastrophic
ﬂoods generated during future lake drainage events from
larger or deeper lakes or during past lake drainage events
require further analysis (Marsh et al., 2008). Nonetheless,
the fact that 50-75% of the landscape area in several Arctic
lowland regions (Hinkel et al., 2003; Grosse et al., 2005;
Hinkel et al., 2005; Jones et al., 2012) represents drained
lake basins means that these events have likely had a dramatic
inﬂuence on downstream ecosystems and erosion of
the landscape, and pose a potential threat to infrastructure
(Marsh et al., 2009).
Comparison of observed peak discharge measurements
from L195 and Lake Illisarvik with peak discharge measurements
from various ordered North Slope of Alaska river basins
(Kane et al., 2000; Whitman et al., 2011; Arp et al.,
2012) shows that peak discharge from catastrophic lake
drainage events can exceed snow-melt generated peakﬂows
for Arctic river basins that are more than two orders of magnitude
larger (Figure 7). The smallest catchment that we
could ﬁnd with a long-term gauged station (minimum of ﬁve
years of data) for comparison to L195 (~1 km2
) is Imnavait
Creek (~2 km2
). This site is located in the Arctic Foothills
of northern Alaska and has an average annual peakﬂow of
1 m3
/s (Kane et al., 2000). Normal peak discharge for coastal
plain catchments that are more than one order of magnitude
larger than L195, averaged 3.5 m3
/s (Whitman et al. 2011).
Watersheds more than two orders of magnitude larger than
L195 experience average snowmelt peak discharges between
30 and 100 m3
/s (Kane et al., 2000, Arp et al.,
2012) (Figure 7), though it is recognized that extreme rainfall
events can generate even higher peak ﬂows in Arctic rivers
(Kane et al. 2003). Thus, the event discharge generated
by the drainage of L195 was of a similar magnitude as can
occur from river basins more than two orders of magnitude
larger. Effective ﬂows in many low-order Arctic streams
and rivers are moderated by channel-armoring bedfast ice
during normal snowmelt peak ﬂows, thus reducing erosion
and sediment transport (McNamara and Kane, 2011). This
comparison indicates that the one-time ﬂood events generated
by catastrophic lake drainages may be highly effective
at restructuring downstream river channels and transporting
large amounts of sediment. Such drainage events are presently
infrequent in the study area, but the abundance of
drained lake basins in the region points to this type of ﬂood
as potentially important for Arctic river systems and landscape
modiﬁcation over the Holocene.
Figure 7 The observed peak discharge at L195 compared to the artiﬁcial drainage
of Lake Illisarvik (Mackay, 1997; Marsh and Neumann, 2001) and normal,
climatically-driven annual peak ﬂows (mean +/- 1 standard deviation, n=5) for a
range of watersheds in Arctic Alaska (Kane et al. 2000, Whitman et al. 2011; Arp
et al. 2012).
Copyright © 2015 John Wiley & Sons, Ltd.
CONCLUSION
By deploying in situ sensors in a thermokarst lake that we
anticipated as having a high likelihood to drain, we have provided
detailed information on the environmental conditions,
timing, and discharge associated with a natural drainage
event. The 80 ha thermokarst lake drained on 05 July 2014
over a course of 36 hours, with 75% of the 871,990 m3
volume
of water loss occurring 10 hours after the onset of drainage.
Peak discharge was observed to be 25 m3
/s and the
estimated, peak instantaneous discharge was 29.4 m3
/s.
The lake likely drained as a result of cooler than normal air
temperatures and higher than normal rainfall and snowmelt
runoff during early summer. These conditions elevated the
lake water level and promoted permafrost degradation
through bank overtopping and thermo-erosion along an
ice-wedge network. The discharge generated by this lake
drainage event was of a similar magnitude to that of northern
Alaska river basins whose areas are more than two orders of
magnitude larger. The fact that 62% of the 1,750 km2
area
assessed during this study was mapped as drained
thermokarst lake basins indicates that catastrophic drainage
events have likely had a major inﬂuence on the hydrology
and geomorphology of this ice-rich permafrost dominated
lowland. Our ﬁndings indicate that thermokarst lakes may
be primed for drainage in early summer, when cold and
wet weather conditions prevail, and that drainage may result
from bank overtopping and thermo-erosion along an icewedge
network. These ﬁndings support several of the observations
provided by J.R. Mackay during his many ﬁeld seasons
spent studying lakes and drained lake basins in the
Canadian Arctic.
ACKNOWLEDGEMENTS
REFERENCES
Arp CD, Jones BM. 2009. Geography of
Alaska lake districts: Identiﬁcation, description,
and analysis of lake-rich regions of a
diverse and dynamic state. U.S. Geological
Survey Scientiﬁc Investigations Report
2008–5215.
Arp CD, Jones BM, Urban FE, Schmutz JA,
Jorgenson MT. 2010. Two mechanisms of
aquatic and terrestrial habitat change along
an Alaskan Arctic coastline. Polar Biology
33: 1629–1640. DOI:10.1007/s00300-010-
0800-5.
Arp CD, Jones BM, Urban FE, Grosse G. 2011.
Hydrogeomorphic processes of thermokarst
lakes with grounded-ice and ﬂoating ice regimes
on the Arctic coastal plain, Alaska,
USA. Hydrological Processes 25:
2422–2438. DOI:10.1002/hyp.8019.
Arp CD, Whitman MS, Jones BM, Kemnitz R,
Grosse G, Urban FE. 2012. Drainage Network
Structure and Hydrologic Behavior
of Three Lake-Rich Watersheds on the Arctic
Coastal Plain, Alaska. Artic, Antarctic,
and Alpine Research 44: 385–398. DOI:
10.1657/1938-4246-44.4.385.
Brewer MC, Carter LD, Glenn R, Murray DF.
1993. Sudden drainage of a thaw lake on the
Alaskan Arctic Coastal Plain. In In Proceedings
of the Sixth International Conference
on Permafrost, South China
University of Technology Press; 48–53.
Flint PL, Mallek EJ, King RJ, Schmutz JA,
Bollinger KS, Derksen DV. 2008. Changes
in the abundance and spatial distribution of
geese molting near Teshekpuk Lake,
Alaska: Interspeciﬁc competition or ecological
change. Polar Biology 31: 549–556.
DOI:10.1007/s00300-007-0386-8.
Frohn RC, Hinkel KM, Eisner WR. 2005. Satellite
remote sensing classiﬁcation of thaw
lakes and drained thaw lake basins on the
North Slope of Alaska. Remote Sensing of
Environment 97: 116–126. DOI:10.1016/j.
rse.2005.04.022.
Grosse G, Schirrmeister L, Kunitsky VV,
Hubberten HW. 2005. The use of CORONA
images in remote sensing of
periglacial geomorphology: an illustration
from the NE Siberian coast. Permafrost
and Periglacial Processes 16: 163–172.
DOI:10.1002/ppp.509.
Grosse G, Walter Anthony KM, Romanovsky
VE, Plug LJ, Jones BM, Edwards ME.
2009. Negative climate feedbacks from surface
permafrost degradation in the continuous
permafrost zone - Thermokarst lakes
on the run. American Geophysical Union,
Fall Meeting 2009, U44A-07.
Grosse G, Jones BM, Arp CD. 2013.
Thermokarst lakes, drainage, and drained
basins. In Treatise on Geomorphology 8,
Shroder J (ed.). Elsevier: San Diego;
325–353. DOI: 10.1016/B978-0-12-37473
9-6.00216-5
Hinkel KM, Eisner WR, Bockheim JG, Nelson
FE, Peterson KM, Dai X. 2003. Spatial Extent,
Age, and Carbon Stocks in Drained
Thaw Lake Basins on the Barrow
Peninsula, Alaska. Arctic, Antarctic, and
Alpine Research 35: 291–300.
Hinkel KM, Frohn RC, Nelson FE, Eisner
WR, Beck RA. 2005. Morphometric and
spatial analysis of thaw lakes and drained
thaw lake basins in the western Arctic
Coastal Plain, Alaska. Permafrost and
Periglacial Processes 16: 327–341. DOI:
10.1002/ppp.532.
Hinkel KM, Jones BM, Eisner WR, Cuomo
CJ, Beck RA, Frohn RC. 2007. Methods
to assess natural and anthropogenic thaw
lake drainage on the western Arctic Coastal
Plain of northern Alaska. Journal of Geophysical
Research: Earth Surface 112:
F02S16. DOI: 10.1029/2006JF000584
Hinkel KM, Sheng Y, Lenters JD, Lyons EA,
Beck RA, Eisner WR, Wang J. 2012.
Thermokarst Lakes on the Arctic Coastal
Plain of Alaska: Geomorphic Controls on
Bathymetry. Permafrost Periglacial. Processes
23: 218–230. DOI:10.1002/ppp.
1744.
Jones BM, Arp CD, Jorgenson MT, Hinkel
KM, Schmutz JA, Flint PL. 2009a. Increase
in the rate and uniformity of coastline erosion
in Arctic Alaska. Geophysical Research
Letters 36: L03503. DOI: 10.1029/
2008GL036205
Jones BM, Arp CD, Beck RA, Grosse G, Webster
JM, Urban FE. 2009b. Erosional history
of Cape Halkett and contemporary monitoring
of bluff retreat, Beaufort Sea coast,
Alaska. Polar Geography 32: 129–142.
DOI:10.1080/10889370903486449.
127
Copyright © 2015 John Wiley & Sons, Ltd.
Jones BM, Grosse G, Arp CD, Jones MC,
Walter Anthony KM, Romanovsky VE.
2011. Modern thermokarst lake dynamics
in the continuous permafrost zone, northern
Seward Peninsula, Alaska. Journal of Geophysical
Research Biogeosciences 116:
G00M03. DOI: 10.1029/2011JG001666
Jones MC, Grosse G, Jones BM, Walter Anthony
KM. 2012. Peat accumulation in
drained thermokarst lake basins in continuous,
ice-rich permafrost, northern Seward
Peninsula, Alaska. Journal of Geophysical
Research Biogeosciences 117: G00M07.
DOI: 10.1029/2011JG001766
Jorgenson MT, Shur Y. 2007. Evolution of
lakes and basins in northern Alaska and discussion
of the thaw lake cycle. Journal of
Geophysical Research Earth Surfaces 112:
F02S17. DOI: 10.1029/2006JF000531.
Kane DL, Hinzman LD, McNamara JP, Zhang
Z, Benson CS. 2000. An overview of a
nested watershed study in Arctic Alaska.
Nordic Hydrology 31: 245–266. DOI:10.
2166/nh.2000.015.
Kane DL, McNamara JP, Yang D, Olsson PQ,
Gieck RE. 2003. An extreme rainfall/runoff
event in Arctic Alaska. Journal of Hydrometeorology
4: 1220–1228.
Kanevskiy M, Shur Y, Jorgenson MT, Ping
CL, Michaelson GJ, Fortier D, Stephani E,
Dillon M, Tumskoy V. 2013. Ground ice
in the upper permafrost of the Beaufort
Sea coast of Alaska. Cold Regions Science
and Technology 85: 56–70. DOI:10.1016/j.
coldregions.2012.08.002.
Lantz TC, Turner KW. 2015. Changes in lake
area in response to thermokarst processes
and climate in Old Crow Flats, Yukon.
Journal of Geophysical Research Biogeosciences
. DOI:10.1002/2014JG002744.
Mackay JR. 1981. An experiment in lake
drainage, Richards Island, Northwest Territories:
a progress report. In Current Research,
Part A. Geological Survey of
Canada: Ottawa, Paper 81-1A: 63–68.
Mackay JR. 1988. Catastrophic lake drainage,
Tuktoyaktuk Peninsula area, District of
Mackenzie. In Current Research, Part D.
Geological Survey of Canada: Ottawa, Paper
88-1D: 83–90.
Mackay JR. 1992. Lake stability in an ice-rich
permafrost environment: examples from the
Western Arctic Coast. In Aquatic Ecosystems
in Semi-arid Region: Implication for
Resource Management, Robarts RD,
Boothwell ML (eds). National Hydrology
Research Institute, Environment Canada:
Saskatoon, Saskatchewan; NHRI Symposium
Series 7: 1–25.
Mackay JR. 1997. A full-scale ﬁeld experiment
(1978–1995) on the growth of permafrost
by means of lake drainage, Western
Arctic Coast: a discussion of the method
and some results. Canadian Journal of
Earth Sciences 34: 17–33. DOI:10.1139/
e17-002.
McNamara JP, Kane DL. 2009. The impact of
a shrinking cyrosphere on the form of arctic
alluvial channels. Hydrological Processes
23: 159–168. DOI:10.1002/hyp.7199.
Marsh P, Neumann N. 2001. Processes controlling
the rapid drainage of two ice-rich
permafrost-dammed lakes in NW Canada.
Hydrological Processes 15: 3433–3446.
DOI:10.1002/hyp.1035.
Marsh P, Russell M, Onclin C, Haywood H.
2008. Modelling discharge during the rapid
drainage of thaw lakes in the western Canadian
Arctic. In In Ninth International Conference
on Permafrost, Fairbanks, June
29-July 3, 2008; 1143–1147.
Marsh P, Russell M, Pohl S, Haywood H,
Onclin C. 2009. Changes in thaw lake
drainage in the western Canadian Arctic
from 1950 to 2000. Hydrological Processes
23: 145–158. DOI:10.1002/hyp.7179.
Sellmann PV, Brown J, Lewellen RI, McKim
H, Merry C. 1975. The classiﬁcation and
geomorphic implications of thaw lakes on
the Arctic coastal plain, Alaska. United
States Army, CRREL Research Report
344: 21 pp.
Turner KW, Wolfe BB, Edwards TWD. 2010.
Characterizing the role of hydrological
processes on lake water balances in the
Old Crow Flats, Yukon Territory, Canada,
using water isotope tracers. Journal of Hydrology
386: 103–117. DOI:10.1016/j.
jhydrol.2010.03.012.
Turner KW, Wolfe BB, Edwards TWD, Lantz
TC, Hall RI, Larocque G. 2014. Controls on
water balance of shallow thermokarst lakes
and their relations with catchment characteristics:
a multi-year, landscape-scale assessment
based on water isotope tracers
and remote sensing in Old Crow Flats,
Yukon (Canada). Global Change Biology
20: 1585–1603. DOI:10.1111/gcb.12465.
Whitman MS, Arp CD, Jones BM, Morris W,
Grosse G, Urban FE, Kemnitz R. 2011. Developing
a long-term aquatic monitoring
network in a complex watershed of the Alaskan
Arctic Coastal Plain. In Proceedings of
the Fourth Interagency Conference on Research
in Watersheds: Observing, Studying,
and Managing for Change, Medley CN,
Patterson G, Parker MJ (eds.). U.S. Geological
Survey: Reston, Virginia; 15–20.
Urban FE, Clow GD. 2014. DOI/GTN-P climate
and active-layer data acquired in the
National Petroleum Reserve—Alaska and
the Arctic National Wildlife Refuge,
1998–2013. U.S. Geological Survey Data
Series 892.
Wang J, Sheng Y, Hinkel KM, Lyons EA.
2012. Drained thaw lake basin recovery on
the western Arctic Coastal Plain of Alaska
using high-resolution digital elevation
models and remote sensing imagery. Remote
Sensing of Environment 119: 325–336.
DOI:10.1016/j.rse.2011.10.027.
Weller MW, Derksen DV. 1979. The geomorphology
of Teshekpuk Lake in relation to
coastline conﬁguration of Alaska’s coastal
plain. Arctic 32: 152–160.
Wolfe BB, Turner KW. 2008. Near-record
precipitation causes rapid drainage of Zelma
Lake, Old Crow Flats, Northern Yukon
Territory. Meridian Spring: 7–12.