A Marine Season Metric for Foxe Basin, Nunavut, Canada: Insights into the Evolving Nature of Sea-Ice Breakup and Freeze-Up

Abstract

A new marine climate metric, marine season, is introduced for Foxe Basin, Nunavut, Canada capturing the time of the year that the Basin is influenced by open water. The metric is developed with a day-to-day temperature variability framework using the Hall Beach (Sanirajak) climate record (1957–2023). Day-to-day minimum temperature variability provides a clear signal of the marine season. The new metric is compared to the more traditional breakup and freeze-up dates of sea ice that uses a 5/10th sea-ice spatial coverage threshold. While the two metrics are in general agreement, some important differences occur related to the time required for the breakup (full ice coverage to 5/10th sea-ice coverage). The timing from onset of the marine season to 5/10th ice coverage has shortened in time in a statistically significant fashion, indicating a more rapid breakup in recent years. In contrast, the freeze-up period, 5/10th to full sea-ice coverage has increased. The longer ice-free season, as determined by sea-ice data, arises primarily from open water changes in the breakup (shorter) and freeze-up (longer) period timing. These are novel insights that suggest that the basic sea-ice regime, oscillating from a full sea-ice platform and ice-free conditions has not changed, but rather the observed changes are in the nature of the transitions between these two states, breakup and freeze-up.

1. Introduction

Foxe Basin is a Canadian inland sea and part of the Hudson Bay System, which also includes Hudson Bay and Hudson Strait (Figure 1). The seasonal sea-ice platform of Foxe Basin is integral to the hunting and well-being of local communities [1]. Foxe Basin’s physical environment has been studied: sea-ice cycle [1,2,3,4,5,6], sea-ice thickness [2,7], atmospheric temperature [1,2,4,6,8] as well as three dimensional aspects of the circulation [9,10]. The changes in the sea-ice cycle have led to studies with a navigation lens [5,11,12,13,14].

Key to Foxe Basin’s climatology is a full seasonal sea-ice cycle. Sea ice is mainly produced in the Basin, although some sea ice enters via Hecla and Fury Strait at the northwest reach of the basin. Sea ice forms (freeze-up) in the basin in late October/early November forming a complete sea-ice platform that lasts until the ice breaks up in July [1,2,3,4,5,6]. Sea-ice patterns during breakup follows atmospheric and oceanic currents in the basin, responding to both radiative and dynamic (advective) forcing, characterized by the advective redistrubution of sea ice. This also allows for the regular appearance of polynyas (open water), in particular, at the northwest part of the basin near Hecla and Fury Strait and along the western coast of the basin [9,10,15] during winter. The freeze-up is largely radiatively driven with considerably less movement of the sea ice [6] and occurs more rapidly than the more dynamic breakup.

The published rates of change of the two main metrics of the seasonal sea-ice cycle, the timing of breakup and the timing of freeze-up in Foxe Basin are presented in

Table 1. Summary of rates of change of breakup and freeze-up dates in the published literature.

	Breakup	Freeze-Up	Period
Gagnon & Gough (2006) [2]	−0.5 days/y	No trend	1959–1989
Laidler et al. (2009) [1]	−0.6 days/y	0.6 days/y	1982–2005
Galbraith & Larouche (2010) [3]	−0.9 days/y	0.9 days/y	1990–2009
Hochheim & Barber (2014) [4]	−0.7 days/y	0.9 days/y	1980–2010
Andrews et al. (2018) [5]	−0.7 days/y	0.6 days/y	1980–2014
Kowal et al. (2024) [6]	−0.4 days/y	0.5 days/y	1971–2018

. These typically use the navigation-informed 5/10th ice coverage as the threshold for both breakup and freeze-up. This threshold is a measure of spatial sea-ice coverage with 10/10 during complete sea-ice coverage and 0/10 for the open water season. Although the studies use differing methodologies and timeframes, there is a consistency among the studies indicating both an earlier breakup and later freeze-up of a similar magnitude during the last fifty years.

In this work, a temperature variability framework that has been successfully employed to detect marine climates [16,17] is used. A new metric, “marine season” is introduced within this framework. This season is similar to the “ice-free” season that appears in the literature [4,5,6] (the difference between breakup and freeze-up dates) but tends to be longer and captures the longer period of time when the Basin is influenced by varying degrees of open water (before the 5/10th threshold is reached for breakup and after the 5/10th threshold is reached for freeze-up). The ice-free season determinations are typically done using the 5/10th sea-ice coverage threshold, important for navigation, but they may miss periods of time that do not meet this threshold but in which the marine climate is experienced [6].

The day-to-day thermal variability framework (DTD) developed by [18], based on [19], has been found to provide a useful suite of metrics for detecting thermal variations that reflect the local environments where data is collected. These include urban, rural, peri-urban, marine, island, airport and mountain environments [20]. The absolute difference between a day’s surface mean temperature with the previous day’s surface mean temperature (DTD) was introduced as the basic measure of thermal variability [18]. Day-to-day thermal metrics have been applied to the examination of climate variation [21,22,23,24,25] including the detection of fog and cloud cover [1]. It has been considered in other disciplines, such as, health [26,27,28,29,30], economics [31,32], agriculture [33] and transportation safety [34].

The day-to-day variability of the minimum temperature of the day (DTD_Tmin) was found to be a clear indicator of coastalization or marine climates [16,17]. At the surface, insolation is partitioned into sensible heat, subsurface heat, and latent heat. Marine (coastal) environments, under the same radiative conditions as in-land settings, diverts more energy into latent heat, evaporating of surface water, and providing the potential for fog and clouds. This mechanism reduces the day-to-day variability, as shown in [16]. This measure has been used to detect the marine influence on temperature in a number of locations [16,17,22]. For two Canadian studies [16,17], due to sea and lake ice in winter which tends to suppress the marine effect, an adjusted, ADTD_Tmin, was used that excluded the winter months of December, January, and February [16,17] in the calculation. For the ice-free season, ref. [16] found that coastal stations fell below a threshold of 2.35 °C for ADTD_Tmin. Ref. [17] examined 31 locations in the Canadian Prairies. Included among the 31 were two climate stations from Churchill, Manitoba (Churchill A, Churchill Marine). For these two stations, a marine climate was only detected when the excluded months were extended to match the months when sea ice dominates Hudson Bay. This is illustrated for Foxe Basin in the annual cycle of DTD_Tmin for Sanirajak (Hall Beach) (Figure 2) using an average of the period 1971–2000. The 2.35 °C DTD_Tmin threshold for marine climates is included in the figure to show the bifurcated behaviour with, and without, sea ice. For the marine season, beginning in late spring the DTD_Tmin metric is substantially lower than for the ice-covered months. For this work the marine season is defined as the part of the year that DTD_Tmin is below the 2.35 °C threshold.

To explore this new metric, the following will be done. First the creation of a time series of the “marine season” using data from Hall Beach A (more recently known as Sanirajak A) from 1957 to 2023. These time series will be compared to the breakup and freeze-up time series generated by using sea-ice chart data from 1971 to 2018 as reported in [6]. Finally, the newly created time series will be analyzed for temporal trends and compared to the trends found in data from [6].

2. Materials and Methods

2.1. Surface Air Temperature Data

Four climate stations (Hall Beach (Sanirajak), Igloolik, Cape Dorset (Kinngait), and Coral Harbour) (see Figure 1) with substantial climate records in the Foxe Basin region were identified (

Table 2. List of climate stations in the Foxe Basin region including latitude, longitude, elevation and available data.

Station	Latitude	Longitude	Elevation	Period
Hall Beach A	68.78 N	81.24 W	9.1 m	1957–2014
Sanirajak	68.78 N	81.24 W	8.2 m	2004–2023
Igloolik	69.38 N	81.80 W	21.3 m	1977–2003
Igloolik A	69.37 N	81.82 W	52.7 m	1984–2015
Cape Dorset	64.23 N	76.53 W	48.2 m	1963–2012
Coral Harbour A	64.19 N	83.36 W	62.2 m	1933–2015

). In aggregate there is data during the period of 1944 to 2023, but are heteorogeneously distributed across the four stations. The data was extracted from the climate data archive at Environment and Climate Change Canada (https://climate.weather.gc.ca/historical_data/search_historic_data_e.html) (accessed on 8–15 September 2024).

Coral Harbour A, in spite of its more fulsome coverage, was eliminated as being more responsive to Hudson Bay sea-ice conditions. Cape Dorset (Kinngait) had too much missing data [6] to use for the marine season analysis. Finally, the polynynas located near Igloolik introduced a marine influence that persisted beyond the ice-free season for the Basin, an interesting point that may be explored in future work. Thus, the spatial paucity of useable temperature data leaves us with a single climate station, Hall Beach A. Fortunately this station has the most robust (length and lack of missing data) and is the spatially most relevant of this group of stations. The community officially changed its name from Hall Beach to Sanirajak in 2020. The station Sanirajak is also located at the Hall Beach airport, about 50 m from the Hall Beach A station. Due to this proximity, the Hall Beach A record is extended using Sanirajak data for a period from 1957 to 2023, treating this as a single record.

2.2. Sea-Ice Data

The sea-ice data was accessed from a data set generated in previous work [6]. In that work, the Basin was analyzed on a 24-point grid and breakup and freeze-up time series starting from 1971 were generated. They used sea-ice charts produced for navigation purposes with a threshold of 5/10 sea-ice coverage as the threshold for both breakup and freeze-up. For this work, the 24 grid data is averaged to produce a single breakup and freeze-up date per year for comparison with the marine season time series. Since the sea- ice charts are available weekly, one week (7 days) is the error associated with the data. Subsets of the 24 point grid were explored but the highest correlation was found when using the full 24 point grid.

2.3. Marine Season Creation Using DTD_Tmin

2.3.1. DTD_Tmin Calculation

The DTD metric used arises from a day-to-day temperature variability framework. The formulation for DTD temperature variability metrics, DTD, are taken from [18]:

DTD = Σ_i (T_i − T_i−1)/(N − 1)

(1)

where i is the daily counter over the time period of interest, and a total of N − 1 pairs of values are variation of minimum temperature of the day which, as noted above, has been shown to be adept at identifying marine climates [16,17]. For this work N = 9 (days), to suppress high frequency variability.

2.3.2. Marine Season

For each of the years of the Hall Beach A/Sanirajak climate record (1957–2023), the DTD_Tmin daily data was filtered using a nine-day running mean to remove high frequency noise and plotted with a 2.35 °C line as was done in Figure 2. From these plots the onset and end of the marine season were determined and tabulated. The marine season length was the difference between these two. The nine-day window can also be used as an estimate of the uncertainty associated with this methodology, not dissimilar to the uncertainty associated with the sea-ice data (±1 week).

2.3.3. Comparison with Sea-Ice Data

The three marine season metrics are compared with the sea-ice chart derived breakup, freeze-up and ice-free season that used the 5/10th spatial sea-ice coverage as the threshold taken from [6]. A correlation analysis is done using linear regression. The time series are tested for autocorrelation and normality.

2.3.4. Trend Analysis

The temporal trends are calculated using linear regression and the Mann-Kendall test. The data is tested for normality (Shapiro-Wilks test) and autocorrelation. The linear regression trends are compared to those from a coincident time period taken from metrics derived from sea-ice charts [6].

3. Results

3.1. Marine Season

Table 3. Marine season (Onset DTD, End DTD, MSL DTD) represents the onset of the marine season, the end of the marine season, and the length of the marine season, respectively. SI indicates sea-ice metrics for Foxe Basin taken from [6] for the years 1971–2018. Delta is the difference between the two data sets using coincident periods.

Year	Breakup SI	Onset DTD	Delta	Freezing SI	End DTD	Delta	SI IFS	MSL DTD	Delta
1958		160			275			115
1959		169			274			105
1960		161			276			115
1961		153			272			119
1962		143			265			122
1963		158			274			116
1964		140			290			150
1965		149			284			135
1966		132			284			152
1967		152			285			133
1968		149			282			133
1969		158			279			121
1970		142			273			131
1971	212.5	147	−65.5	285.0	272	−13.0	72.5	125	52.5
1972	220.1	172	−48.1	264.5	270	5.5	44.4	98	53.6
1973	207.4	149	−58.4	287.3	279	−8.3	79.8	130	50.2
1974	204.4	160	−44.4	294.3	274	−20.3	89.9	114	24.1
1975	194.6	146	−48.6	298.3	284	−14.3	103.6	138	34.4
1976	197.2	166	−31.2	276.6	289	12.4	79.4	123	43.6
1977	199.4	150	−49.4	296.9	282	−14.9	97.5	132	34.5
1978	192.0	164	−28.0	264.5	284	19.5	72.5	120	47.5
1979	199.6	158	−41.6	284.6	272	−12.6	85.0	114	29.0
1980	183.6	138	−45.6	300.0	281	−19.0	116.4	143	26.6
1981	177.3	154	−23.3	309.8	290	−19.8	132.5	136	3.5
1982	195.3	142	−53.3	291.0	275	−16.0	95.7	133	37.3
1983	204.9	148	−56.9	270.3	278	7.7	65.4	130	64.6
1984	186.3	151	−35.3	301.7	295	−6.7	115.4	144	28.6
1985	184.6	134	−50.6	304.4	285	−19.4	119.8	151	31.3
1986	185.7	151	−34.7	286.4	291	4.6	100.7	140	39.3
1987	192.2	155	−37.2	279.4	291	11.6	87.2	136	48.8
1988	182.8	138	−44.8	285.2	280	−5.2	102.3	142	39.7
1989	185.4	148	−37.4	279.9	280	0.1	94.5	132	37.5
1990	203.2	155	−48.2	269.1	269	−0.1	65.9	114	48.1
1991	192.2	157	−35.2	275.9	288	12.1	83.7	131	47.3
1992	191.7	162	−29.7	278.1	272	−6.1	86.4	110	23.6
1993	200.6			280.5			79.9
1994	174.6	141	−33.6	305.0	276	−29.0	130.5	135	4.5
1995	180.4	151	−29.4	296.0	276	−20.0	115.6	125	9.4
1996	186.0	158	−28.0	303.4	293	−10.4	117.4	135	17.6
1997	186.2	155	−31.2	300.5	283	−17.5	114.3	128	13.7
1998	179.6	149	−30.6	317.1	296	−21.1	137.5	147	9.5
1999	189.2	145	−44.2	286.5	281	−5.5	97.3	136	38.7
2000	198.2	159	−39.2	313.2	293	−20.2	115.0	134	19.0
2001	188.2	145	−43.2	300.4	285	−15.4	112.3	140	27.8
2002	188.2	154	−34.2	300.4	277	−23.4	112.2	123	10.8
2003	175.0	141	−34.0	315.6	274	−41.6	140.6	133	−7.6
2004	196.2	161	−35.2	287.5	277	−10.5	91.3	116	24.8
2005	186.3	152	−34.3	314.3	301	−13.3	128.0	149	21.0
2006	172.8	136	−36.8	327.6	280	−47.6	154.8	144	−10.8
2007	182.5	159	−23.5	303.0	269	−34.0	120.5	110	−10.5
2008	193.8	136	−57.8	304.3	273	−31.3	110.5	137	26.5
2009	183.0	154	−29.0	307.3	286	−21.3	124.3	132	7.7
2010	172.4	137	−35.4	337.5	306	−31.5	165.1	169	3.9
2011	181.6	149	−32.6	313.2	291	−22.2	131.5	142	10.5
2012	181.2	151	−30.2	310.4	280	−30.4	129.2	129	−0.2
2013	192.9	147	−45.9	300.3	293	−7.3	107.3	146	38.7
2014	189.0	147	−42.0	307.1	276	−31.1	118.1	129	10.9
2015	183.1	149	−34.1	294.2	285	−9.2	111.1	136	24.9
2016	182.6	148	−34.6	314.5	278	−36.5	131.8	130	−1.8
2017	187.0	152	−35.0	311.3	274	−37.3	124.3	122	−2.3
2018	177.6	138	−39.6	284.4	294	9.6	106.8	156	49.2
2019		142			289			147
2020		155			291			136
2021		154			292			138
2022
2023		153			283			130

lists the marine season calculated using Hall Beach A/Sanirajak temperature data using the DTD_Tmin threshold of 2.35 °C. For comparison purposes (next section) the sea-ice metrics derived from sea-ice charts (1971–2018) are also included [6].

3.2. Comparison to Sea-Ice Chart Derived Metrics

The onset of the marine season (orange line) is presented in Figure 3, along with the breakup dates from [6] (blue line). The grey line represents the difference between the two. This difference represents the time from onset to reaching the 5/10th sea-ice coverage threshold. On average the onset is 39 days earlier, although that difference diminishes with time, a point pursued in the next section. This indicates that the climatic effects of the sea-ice breakup are experienced well before the ice reaches the 5/10th threshold. The difference between the two is an indication of how quickly the breakup takes place. A Pearson r correlation calculation for the two times series yielded a statistically significant r value of 0.50 (p = 0.0004). This is an indication that the two are related but also signals that the nature of the breakup from onset to 5/10th sea-ice coverage varies from year to year and that breakup over time is taking less time.

The end of the marine season (orange line) is shown in Figure 4. In contrast to the onset of the marine season results (Figure 3), the sea-ice chart derived 5/10th sea-ice coverage threshold is largely coincident with the end of the marine season and the difference between the two is within the uncertainties of the two methodologies. This suggests that the freeze-up, post 5/10th coverage is rapid, that is, a much faster process than breakup, a result noted in [6] but over time this process has become slower. The difference between the two increases with time, a point explored in the next section. A Pearson r correlation calculation for the two times series yielded a statistically significant r value of 0.36 (p = 0.01). This is an indication that the two are related but also signals that the nature of the freeze-up from 5/10th sea-ice coverage to the end of the marine season varies from year to year.

The marine season length is plotted in Figure 5, along with the ice-free season from [6]. During the early part of the record, 1971–1993, the two are substantially different with the marine season being longer than the ice-free season. However, the two are similar post 1994. This does not mean though that the two are coincident. The marine season is displaced from the 5/10th threshold relatively equally (starting earlier and ending earlier). A Pearson r correlation calculation for the two times series yielded a statistically significant r value of 0.62 (p < 0.0001). This is an indication that the two are closely related but the nature of the difference has changed with time as clearly evident in Figure 5 and explored in more detail in the next section.

3.3. Trend Analysis

Temporal trend analyses using both linear regression and the Mann-Kendall test are performed on the marine season metrics. The time series were tested for autocorrelation and normality (Shapiro-Wilks test). The three time series were not autocorrelated and the distributions were normal. Thus, no corrections were applied to the data. In addition, temporal trend analysis was performed on the sea-ice chart derived data (taken from [6]) and on the difference times series for the three metrics.

From

Table 4. Statistical analysis of marine season metrics. Onset marks the beginning of the marine season and End marks the conclusion of it. Marine season length (MSL) is the difference between End and Onset. “Breakup”, “Freeze-up”, and “Ice-free season (IFS)” are corresponding data derived from sea-charts [6]. “Difference” is used to designate the difference between marine season metric and those derived from sea-ice charts. The measures are compared initially using the coincident period of 1971–2018 and then Onset, End, and MSL are reported for the full period of 1957–2023. M-K refers to Mann-Kendall test and LR refers to linear regression.

	M-K p-Value	LR p-Value	Rate of Change	Range
Onset	0.12	0.06	−0.17 days/y	1971–2018
Breakup	0.0001 ***	<0.0001 ***	−0.43 days/y	1971–2018
Difference Onset	0.02 *	0.006 **	−0.27 days/y	1971–2018
End	0.24	0.20	0.12 days/y	1971–2018
Freeze-up	0.0001 ***	0.0001 ***	0.64 days/y	1971–2018
Difference End	0.0005 ***	0.0007 ***	0.52 days/y	1971–2018
MSL	0.11	0.03 *	0.28 days/y	1971–2018
IFS	0.00004 ***	<0.0001 ***	1.07 days/y	1971–2018
Difference MSL	0.00004 ***	<0.0001 ***	0.78 days/y	1971–2018
Onset	0.17	0.13	−0.09 days/y	1957–2023
End	0.01 **	0.005 **	0.15 day/y	1957–2023
MSL	0.01 **	0.004 **	0.24 days/y	1957–2023

* indicates statistical significance at p < 0.05, ** p < 0.01, and *** p < 0.001.

an interesting insight on the sea-ice evolution emerges. Turning to the coincident analysis (1971–2018), the sea-ice observations from [6] the breakup, freeze-up, and ice-free season had statistically significant trends for earlier breakup, later freeze-up, and increased ice-free season length. However using DTD metric the onset and end of the marine season were not changing in a statistically significant fashion and the marine season length weakly so using linear regression but not so for the Mann-Kendall test. These diverging results suggest that the increase in the ice-free season (1.07 days/year) is largely the result of the changing nature of the breakup and freeze-up. With onset (DTD) not occurring earlier in a statistically significant fashion, and the start of sea-ice season as measured by the 5/10th coverage metric occurring earlier (−0.43 days/y), this indicates that the breakup period (time from 10/10th to 5/10th) is occuring more rapidly with time during the 1971–2018 period. Similarly with the end of the marine season not occuring later in a statistically signficant fashion but the 5/10th coverage metric occuring later (0.64 days/y), this is an indicator that the freeze-up season (5/10th to 10/10th) is occuring more slowly as observed in [1] when timing of 5/10th and 9/10th coverage were compared near Igloolik in northwestern Foxe Basin. The combined impact of these two account for the more pronounced longer ice-free season (1.07 days/y) compared to more modest increase in marine season length (DTD) (0.28 days/year). Turning to the longer time series, 1957 to 2023, for the DTD analysis, the onset while earlier with time, it is not statistically significant as it was for the shorter time series (1971–2018). However, for the end of the marine season and for the length of the marine season, both are later and longer, respectively, an indication with the longer time series that it is not just the nature of breakup (faster) and freeze-up (slower) that accounts for the changes in sea-ice conditions.

4. Discussion

In this work a new metric to describe the period of the year in Foxe Basin when some open water is present, the marine season, is successfully introduced. The metric is based on a novel application of the day-to-day temperature variability framework metric, DTD_Tmin, the change in minimum temperature of the day from day to day. Refs. [16,17] have shown that this metric effectively detects marine environments in other parts of Canada.

Using temperature data (DTD_Tmin) from the Hall Beach A/Sanirajak climate station from 1957 to 2023, a time series of the marine season metrics were created including onset, marine season length and end of the marine season. These were compared with seasonal sea-ice metrics derived from sea-ice charts (1971–2018) from [6]. The comparison of temporal trends provides the novel insight that the statistically significant trends in the ice-free season in the Kowal data set (1.07 days/y) arose largely from changes in the nature of the breakup and freeze-up (0.78 days/y) rather than the more modest change in open water season length (0.28 days/y). Over time, breakup has been occurring more rapidly after onset and freeze-up which typically was quite rapid is occurring more slowly in recent years, a finding consistent with observations by community members in Igloolik reported in [1] which compared 5/10th and 9/10th sea-ice coverage timing that affected the ability to safely travel from Igloolik to Melville Peninsula for caribou harvesting. This also suggests a significant regime change in the region has not occurred. A sea-ice platform forms each year and the timing of that platform has not begun to erode in a significant fashion but rather the change has been manifest in the nature of transition from the ice platform to ice-free conditions.

The DTD framework allows for the exploration of sea-ice conditions in Foxe Basin for a longer period than is possible from satellite derives sea-ice measures (which began in 1971). This was done in a preliminary fashion in this work by looking at trends of onset, end and length of the marine seasson from 1957–2018, which generated different trends analyses compared to using a start date of 1971, suggesting some thermal inertia may be operating [35]. It possible to use DTD to hindcast sea-ice conditions to 1957 and impute conditions when sea-ice data is missing as was done in [36] for Hudson Bay.

This type of analysis can be applied to the other basins with seasonal sea ice including the other components of Hudson Bay system (Hudson Bay, Hudson Strait) to tease out the relevant factors in sea-ice abatement in those areas. This will also enable a better understanding of the interplay between atmospheric forcing and the thermal inertial responses in the system, “climate memory” [5,35].

5. Conclusions

A new marine climate metric was introduced using Foxe Basin, Nunavut, Canada data reflective of the time of the year that the Basin was influenced, by varying degrees, by open water. The metric was developed with a day-to-day thermal variability framework using the Hall Beach (Sanirajak) climate record. This new metric was compared to the more familiar breakup and freeze-up dates of sea ice that uses a 5/10th sea-ice coverage threshold commonly employed in navigation. While the two in general agree, some important differences occur that relate primarily to the nature of the breakup and freeze-up. The onset of the marine season as detected by DTD is not changing in a statistically significant fashion while the timing of the 5/10th sea-ice coverage is occurring earlier in a statistically significant fashion. These two results indicate that the breakup is happening during a shorter period of time. The freeze-up period, the timing from 5/10th to 10/10th sea-ice coverage, however, is taking longer over the study period. The faster breakup and slower freeze-up processes, an important aspect not captured in the 5/10th ice coverage methodology temporal trends, are new insights in how regional climate change is manifested in the Foxe Basin seasonal sea-ice record. These results also suggest that the basic nature of the sea-ice cycle, production of a full ice platform and subsequent ice-free conditions has not fundamentally changed but the changes are largely occuring in the transitions between these two states. This analysis is applicable to other regions of seasonal sea ice, such as Hudson Bay [36] and Hudson Strait as well as large mid-latitude lakes, such as the Great Lakes.