Recent decline of Grinnell and Terra Nivea ice caps, Baffin Island, Canada
R.G.
1Department
1
Way
of Geography, University of Ottawa, Ottawa, Ontario, K1N 6N5
Results
Discussion & Conclusion
Introduction
In the Canadian Arctic, ice losses and melt rates have begun to exceed those recorded over the past
several millennia (Fisher et al, 2012; Zdanowicz et al, 2012). Regional changes in ice volume are
increasingly important for sea level rise, particularly over the past decade when Canadian Arctic
contributions have nearly doubled (Gardner et al, 2012). Although there are efforts to monitor ice
cap changes in the eastern Arctic, the eastern low-Arctic has been comparatively under-studied.
The region contains shrinking small mountain glaciers (Torngat Mountains) (Way, 2013) and two
small ice caps at the southern tip of Baffin Island (Mercer, 1956). These ice caps (Grinnell Glacier
and Terra Nivea) are the southernmost in the eastern Canadian Arctic and despite being located
only ~200 km from Iqaluit, Nunavut, have not been comprehensively studied (Sharp et al, In Press)
(Figure 1). Positioned at the southern fringe of Arctic glacier viability, they are useful for
monitoring the impacts of climate change. In this study, we evaluate multidecadal area changes of
both Grinnell Glacier and Terra Nivea ice caps using satellite remote sensing. Results are compared
with meteorological data to assess the sensitivity of Grinnell Glacier and Terra Nivea to changes in
summer air temperature and to interpret glacier susceptibility to future climate change.
Figure 2: Multi-temporal ice cap outlines for Terra Nivea (left) and Grinnell Glacier (right) corresponding to 1975 (red) and
2013 (blue).
Area estimates derived from satellite imagery for Terra Nivea ranged from 199.1 km2 (1975) to
153.4 km2 (2013) and for Grinnell Glacier from 134.3 km2 (1975) to 109.1 km2 (2013) (Figure 2). For
both ice caps, area changes occurred along all margins with the greatest change for Terra Nivea
observed along its northwest margin and for Grinnell Glacier in the southwest. On multidecadal
timescales both ice caps show retreat at each time interval since the beginning of the satellite
record with an overall area decline of ~21% (Figure 3). The decadal estimates of Terra Nivea ice cap
area ranged from 199.14 ± 4.69 km2 (1973/1975) to 154.8 ± 1.3 km2 (2010/2013) indicating a decline
in ice area of ~44 km2 over the past four decades. For Grinnell Glacier, the corresponding decline
in ice area was ~24 km2 from 134.3 ± 1.8 km2 (1973/1975) to 110.0 ± 0.9 km2 (2010/2013). Ice cap
outlines derived from historical aerial photography by Canada’s Glacier Atlas (Ommanney, 1980)
and Sharp et al (In Press) extend the record back further to 1958. This data shows negligible
changes (<5 km2) in total ice cap area from 1958 to 1973/1975, when the satellite record begins,
with the 1958 estimate within the 1973/1975 uncertainty limits (Figure 3).
Figure 4: 3D oblique of Terra Nivea’s northeastern ice margins for 1984 (top panel) and 2013 (bottom panel).
Obliques were derived by draping 7-4-2(1984)/7-5-3(2013) false colour composite images over the National
Topographic database DEM. Areas of interest noting significant ice decline are shown by black rectangles.
Mercer (1956) previously suggested Grinnell Glacier and Terra Nivea were particularly
sensitive to changes in summer air temperature as opposed to winter precipitation due to
their geographic location and thin ice cover. Comparing the regional air temperature
record (Figure 5) with the multidecadal ice area estimates from this study largely supports
this interpretation. Direct observations of widespread marginal recession and the
emergence of Nunataks on the interior of both ice caps support Mercer’s interpretation and
suggest that large areas of these ice caps are currently at disequilibrium with current air
temperatures (Figure 4). In contrast to the rapid warming observed in Figure 5, the region’s
cold-season precipitation shows no trend over the period of this study and suggests that
changes in winter snowfall are not the main driver of recent decline in ice area (not shown).
Figure 1: Satellite image of Grinnell Glacier (A) and Terra Nivea (B) ice caps. Satellite image is
a late-summer (September 11th, 2010) true colour composite from Landsat 5 Thematic Mapper.
Methods
Ice cap outlines were derived satellite remote sensing and existing historical databases (see
Omanney [1980]). General glacier characteristics were collected in a GIS framework using a DEM
prepared by the Canadian National Topographic Database. A total of 10 satellite images were
acquired over the period 1973 to 2013 (Table 1). The first three principal components of each image
were classified using maximum likelihood supervised classification and manually collected
training areas (n=250) separating the image into five major classes (bare ground, ice, lake, ocean,
vegetation). False and true color composites were used in conjunction with the first three principal
components to inform ice cap delineation. Misclassified pixels were manually corrected using onscreen digitization and were typically associated with lake ice and excessive snow cover (e.g. Paul
et al, 2013). Decadal estimates are provided rather than interannual to account for problematic lateseason snow cover on some images. Therefore, pairs of images from adjacent years were used with
decadal estimates being the mean of the two individual outlines and the upper and lower
uncertainty bounds showing the upper and lower ice area estimates respectively.
Figure 3: Time series of multi-decadal area changes from satellite remote sensing for Terra Nivea (red dots) and Grinnell
Glacier (grey dots). Ice cap areas derived from historical aerial photography (Ommanney, 1980; Sharp et al, In Press) are
shown as black squares.
Table 1: General characteristics of satellite images used in this study.
In contrast with the lack of change from 1958 to 1973/1975, since the beginning of the satellite era
both ice caps have lost ice at a rapid pace. Between 1973/1975 and 2010/2013 the rate of cumulative
change for both ice caps was was -1.69 km2/yr with Terra Nivea losing ice at a faster rate (-1.1
km2/yr) than the smaller Grinnell Glacier (-0.57 km2/yr) (Figure 3). The largest are change for both
ice caps occurred between the early and late 2000s with Grinnell Glacier and Terra Nivea losing
9.5 km2 and 15 km2 respectively. Over the full study period both Grinnell Glacier and Terra Nivea
changed very little in mean elevation (+10 m to +40 m) though minimum glacier elevations
changed by +100 m and +80 m respectively. For Grinnell Glacier, these changes reflect the
transition of two outlet glacier termini from being tidewater to land-terminating while for Terra
Nivea this change follows the recession of outlets upvalley into higher elevations (Figure 4).
Figure 5: (Black line) Warm season air temperature (June to September; 1890-2012) for southern Baffin
Island (62-63°N, 66-67°W) derived by re-adding the surface climatology to the Berkeley Earth Surface
Temperature dataset (Rohde et al. 2013). (Red line) Five-year centred running mean of air temperatures.
Assuming the rate of ice losses observed over the past four decades continued for both ice
caps throughout the 21st century, total ice area in the region would reduce by 57% by 2100.
For Grinnell Glacier this would result in a 46% decline in ice area while Terra Nivea would
lose 62% of its ice cover. Enhanced regional warming in the summertime would likely
increase these rates of decline but projections of change in regional precipitation could
potentially offset some ice losses (Lenaerts et al, 2013). This study shows that both Grinnell
Glacier and Terra Nivea are undergoing rapid areal decline and suggests that projected
Arctic warming over the next century will lead to considerable reductions in ice area and
volume, in agreement with the results of Lenaerts et al (2013) for the entire Canadian Arctic
Archipelago.