Landfast ice thickness in the Canadian Arctic Archipelago from Observations and Models 1

5 Observed and modelled landfast ice thickness variability and trends spanning more than five 6 decades within the Canadian Arctic Archipelago (CAA) are summarized. The observed sites 7 (Cambridge Bay, Resolute, Eureka and Alert) represent some of the Arctic’s longest records of 8 landfast ice thickness. Observed end-of-winter (maximum) trends of landfast ice thickness (19579 2014) were statistically significant at Cambridge Bay (-4.31±1.4 cm decade), Eureka (-4.65±1.7 10 cm decade) and Alert (-4.44±1.6 cm decade) but not at Resolute. Over the 50+ year record, the 11 ice thinned by ~0.24-0.26 m at Cambridge Bay, Eureka and Alert with essentially negligible 12 change at Resolute. Although statistically significant warming in spring and fall was present at all 13 sites, only low correlations between temperature and maximum ice thickness were present; snow 14 depth was found to be more strongly associated with the negative ice thickness trends. Comparison 15 with multi-model simulations from Coupled Model Intercomparison project phase 5 (CMIP5), 16 Ocean Reanalysis Intercomparison (ORA-IP) and Pan-Arctic Ice-Ocean Modeling and 17 Assimilation System (PIOMAS) show that although a subset of current generation models have a 18 ‘reasonable’ climatological representation of landfast ice thickness and distribution within the 19 CAA, trends are unrealistic and far exceed observations by up to two magnitudes. ORA-IP models 20 were found to have positive correlations between temperature and ice thickness over the CAA, a 21 feature that is inconsistent with both observations and coupled models from CMIP5. 22 23 The Cryosphere Discuss., doi:10.5194/tc-2016-71, 2016 Manuscript under review for journal The Cryosphere Published: 30 March 2016 c © Author(s) 2016. CC-BY 3.0 License.


Introduction 24
Landfast sea ice is immobile ice that is grounded or anchored to the coast [Barry et al., 25 1979]. In the Arctic, this ice typically extends to the 20-30 m isobath. It melts each summer and 26 reforms in the fall but there are regions along the northern coast of the Canadian Arctic 27 Archipelago (CAA) where multi-year landfast ice (also termed an "ice plug") is present. The two 28 most prominent regions of multi-year landfast sea ice in the CAA are located in Nansen Sound 29 and Sverdrup Channel [Serson, 1972;Serson, 1974] (Figure 1). It has been documented that ice 30 remained intact from 1963-1998 in Nansen Sound and from 1978-1998in Sverdrup Channel 31 [Jeffers et al., 2001Melling, 2002;Alt et al., 2006]. The extreme warm year of 1998 disintegrated 32 the ice in both regions and their survival during the summer melt season in recent years has 33 occurred less frequently [Alt et al., 2006]. Over the entire Arctic, landfast ice extent is declining at 34 7% decade -1 since the mid-1970s [Yu et al., 2013] 35 Records of landfast ice thickness provide annual measures of ice growth that can also 36 almost entirely be attributed to atmospheric forcing with negligible deep ocean influence on local 37 ice formation. While the key forcings on landfast ice and offshore ice are different, the seasonal 38 behavior of landfast ice can nevertheless provide useful information for understanding the 39 interannual variability of ice thickness in both regimes. Presently, there is no pan-Arctic network 40 for monitoring changes in landfast ice but available measurements suggest thinning in recent years. 41 Thickness measurements near Hopen, Svalbard revealed thinning of landfast ice in the Barents 42 Sea region by 11 cm decade -1 between 1966and 2007[Gerland et al., 2008. From a composite 43 time series of landfast ice thickness from 15 stations along the Siberian coast, Polyakov et al. 44 [2010] estimate an average rate of thinning of 3.3 cm decade -1 between the mid-1960s and early 45 At four sites in the CAA, Brown and Cote [1992] (hereinafter, BC92) provided the first 48 examination of the interannual variability of end-of-winter (maximum) landfast ice thickness and 49 associated snow depth over the period . Their results highlighted the insulating role of 50 snow cover in explaining 30-60% of the variance in maximum ice thickness. Similar results were 51 also reported by Flato and Brown [1996] and Gough et al. [2004]. In the record examined by 52 BC92, no evidence for systematic thinning of landfast ice in the CAA was found. Landfast ice 53 thickness records at several of these CAA sites are now over 50 years in length, which represents 54 an addition of more than two decades of measurements since BC92 during a period that saw 55 dramatic reductions in the extent and thickness of Arctic sea ice [e.g. Kwok and Rothrock, 2009;56 Stroeve et al., 2012]. 57 The sparse network of long term observations of snow and ice thickness in the Arctic 58 (clearly exhibited by only four ongoing measurements sites operated by Environment Canada in 59 the CAA) has made the use of models imperative to provide a broader regional scale perspective 60 of sea ice trends in a warming climate. Given the coarse spatial resolution of global climate models, 61 previous studies focusing on the CAA have relied on either a one-dimensional thermodynamic 62 dynamic model [Flato and Brown, 1996;Dumas et al., 2006] or a regional three-dimensional ice-63 ocean coupled model [e.g. Sou and Flato, 2009] and Haas and Howell [2015]. The standard deviations are nearly uniform (at ~0.2 m) across all 128 sites, giving a relatively low coefficient of variation (COV; a measure of relative dispersion 129 defined as the ratio of the standard deviation to the mean) of ~0.1. The thickest ice is found in 130 Eureka with a 1957-2014 mean of 2.27 m that is likely due to climatologically lower air 131 temperatures in the fall and winter (Table 3). 132 Snow depth also appears to grow linearly through the season, peaking in May but unlike 133 ice thickness the monthly variability is high (COV ~0.4) ( Figure 3) (Table 3). The rapid buildup of the snow cover due to storms in the fall and early winter that 136 is evident over the Arctic Ocean multi-year ice cover [Warren et al., 1999;Webster et al., 2014], 137 is not seen in these snow depth records within the CAA. The linear behavior in snow depth is likely 138 maintained by continuous wind-driven redistribution and densification throughout the ice growth 139 season [BC92;Woo and Heron, 1989]. 140 141

Trends 142
The time series of maximum ice thickness at Cambridge Bay, Resolute, Eureka and Alert 143 are illustrated in Figure 4 and summarized in Table 1. Statistically significant (95% or greater 144 confidence level) negative maximum ice thickness trends are present at Cambridge Bay (-4.31±1.4 145 cm decade -1 ), Eureka (-4.65±1.7 cm decade -1 ) and Alert (-4.44±1.6 cm decade -1 ) ( Table 1). A slight 146 negative trend is present at Resolute but not statistically significant at the 95% confidence level 147 (Table 1). Over the 50+ year record, the ice thinned by ~0.24-0.26 m at Cambridge Bay, Eureka 148 and Alert with essentially negligible change at Resolute. These trends in the CAA are similar to 149 trends on the Siberian coast (-3.3 cm decade -1 ) [Polyakov et al., 2010] but lower in magnitude 150 compared to the Barents Sea (-11 cm decade -1 ) [Gerland et al., 2008]. 151 For the shorter record (late 1950s-1989, ~30 years) investigated by BC92 there was a 152 negative trend at Alert (-7.1 cm decade -1 ), no evidence of a trend at Eureka, and a positive trend at 153 Resolute (10 cm decade -1 ) but only the positive trend at Resolute was statistically significant at the 154 95% or greater confidence level. Our results from the present 50+ year record suggest that the 155 negative trend at Alert is robust and the trend at Eureka is now negative and significant. The trend 156 at Resolute is now slightly negative however it is not statistically significant. Typically, ice thickness reaches its maximum in late May with trends toward earlier dates 158 of maximum ice thickness present at all sites (significant at Resolute, Eureka and Alert; Table 3). 159 The significant trends are between -2.0±0.1 days decade -1 at Eureka to -6.2±1.5 days decade -1 at 160 Resolute. At Resolute, the date of maximum ice thickness is now on average more than a month 161 earlier than the early 1960's suggesting a shortened growth season although this is not reflected in 162 the trend in ice thickness. Together, the trends of ice thickness and their recorded dates suggest a 163 systematic thinning of landfast ice at Cambridge Bay, Eureka and Alert. 164 165

Ice thickness linkages with snow depth and temperature 166
The variability of landfast thickness at these Arctic sites was previously found to be largely 167 driven by interannual variations in snow depth and air temperature [BC92;Flato and Brown, 168 1996]. With the 50+ year record at the four sites, we can examine the corresponding linkages to 169 snow depth and temperature which are also summarized in Table 3. 170 For snow depth, there are positive trends at Eureka and Alert and negative trends at 171 Cambridge Bay and Resolute ( Figure 5). The only trend that is statistically significant at the 95% 172 confidence is Cambridge Bay at -0.8±0.4 cm decade -1 (Table 3). In contrast, BC92 found a 173 significant positive trend at Alert (4 cm decade -1 ), a trend of low significance in Eureka, and a 174 negative and significant trend at Resolute (-3.3 cm decade -1 ). Looking at the detrended correlations 175 (r) between snow depth and ice thickness reveals the strongest correlation at Resolute (r=-0. the single pointwise snow depth and ice thickness measurements made at each point in time, which 181 fail to capture spatial heterogeneity in the snow depth/ice thickness relationship. 182 With respect to observed temperature, we find significant warming trends in the spring and 183 fall at all sites over the 50+ year record (Table 3; Figure 7). Significant warming is also present at 184 all sites in the summer except Resolute and at all sites during the winter except Eureka (Table 3). 185 Warming is highest during the fall, at ~0.6C decade -1 at all sites (Table 3). The linkage between 186 temperature and maximum ice thickness weaker than compared to snow depth as only at the 187 Cambridge Bay site is warming in the spring and winter associated with decreases in maximum 188 ice thickness with a detrended correlation of ~0.4. This may indicate that temperature plays more 189 of a role at influencing maximum ice thickness at Cambridge Bay as this site also experienced the 190 lowest detrended correlation with snow depth (r=-0.31). 191 Also of interest is that the observed temperature trends over this period differ considerably 192 than the earlier period investigated in BC92, in which they reported cooling at all the sites, with a 193 significant cooling trend at Eureka. It was noted that the general cooling over their record

Climatology 204
In order to compare seasonal cycles and trends in landfast ice thickness and snow depth 205 between models and observations, we limit our comparison to models with a reasonable concentration is on average above 85% for more than one month but less than 11 months over the 216 1955-2014 period. The Eureka site is however particularly challenging for models because it lies 217 deep in a very narrow channel, which is only resolved by the MPI-ESM-MR in the CMIP5. As a 218 result, for most models, the sample point for Eureka is located on the western shore of Ellesmere 219 Island. 220 The seasonal cycle   The seasonal cycle  of median snow depth from CMIP5 is shown in Figure 9. 230 CMIP5 models indicate a linear increase similar to observations reaching a maximum of ~20 cm 231 in April or May. This is lower than the observed maximum at Resolute, Eureka and Alert but is 232 about twice as much as at Cambridge Bay. While the snow depth reaches zero during the summer 233 at Eureka and Alert in models, the sea ice thickness does not (Figure 8), unlike in observations. 234 This likely reflects the fact that thick, mobile ice is located in the vicinity of these sample points 235 in models. The seasonal cycle over packed ice in these models thus gives a reasonable 236 representation of the seasonal cycle over landfast ice in the CAA, especially in the southern region 237 of the CAA. Overall, this comparison shows how recent improvements in sea ice model resolution 238 allows comparisons with observations that required dynamical downscaling techniques in the 239 previous generation of sea ice models [i.e. Dumas et al. 2005;Sou and Flato, 2013]. in accordance with the temperature trends in these models (not shown). One exception is the ORA-267 IP CGLORS that have positive thickness trends (Figure 12a). This is robust and it appears that the 268 model is not completely equilibrated in the CAA and exhibit large month-to-month adjustments. suggest that care should be taken when using these ORA-IP models to study the interannual 291 variability in the Canadian Arctic. 292 In the CMIP5 models, significant winter snow depth trends are more strongly negative in 293 the North than in the South (Figure 14). This is in disagreement with point observations presented 294 in the previous sections that showed slightly positive snow depth trends at Alert and negative 295 trends at Cambridge Bay. Although only based on limited point in situ observations, this suggests 296 that over the last decades winter precipitation at Alert increased faster than warming temperature 297 could increase melting, a compensation that is clearly not captured in CMIP5 models. 298 299

Conclusions 300
Over the 50+ year in situ observational record, negative trends in maximum (end-of-winter) 301 ice thickness are found at all four sites with statistically significant trends present at Cambridge 302 Bay, Eureka and Alert. Negative trends in the day of maximum ice thickness are also present at all 303 sites and statistically significant at Resolute, Eureka and Alert. Together, these trends suggest 304 thinning of landfast ice in the CAA, where little evidence was found in the shorter record analyzed 305 in an earlier study (BC92). Even though warming is seen at all sites, changes in ice thickness is 306 also attributable to variability in snow depth, which plays a dominant role in controlling the 307 interannual mean and variability of ice thickness. Within the CAA, increases in snow depth are 308 contributing to decreased trends in maximum ice thickness at Eureka and Alert but thus far appear 309 to be exerting less of an impact on maximum ice thickness at Resolute and Cambridge Bay. Freeze 310 onset at these sites is increasing at ~3-6 days decade -1 [Howell et al., 2009] et al., 2008;Kwok and Rothrock, 2009]a linear rate of 326 over -60 cm decade -1 that is mostly due to the loss of multi-year ice. However, the contribution of 327 seasonal ice to that rate is not available. As seasonal ice, becomes the dominant ice type, the focus 328 has shifted to understanding the behavior of seasonal ice thickness. Between 1991and 2003, 329 Melling et al. [2005 found only a small trend (-7 cm decade -1 ), though of low statistical 330 significance, in the seasonal pack in the Beaufort Sea. In the short ICESat record of ice thickness 331 (2003)(2004)(2005)(2006)(2007)(2008), Kwok et al. [2009] also found negligible trend in the seasonal ice cover. This led 332 them to speculate that a thinner snow cover during to the later start of the growth season is 333 conducive to higher ice production as a result of reduced accumulation of that large fraction of 334 snow that typically falls in October and November. However, over the seasonal ice cover there is 335 the additional contribution of ice deformation on the mean of the thickness distribution. 336 While the impact of the snow cover on ice thickness is well known, the significant 337 correlations at Resolute, Eureka and Alert suggest that the higher sensitivity to changes in snow 338 depth could easily mask the warming signal on both fast and offshore ice. The dependency between 339 ice thickness trends and warming trends is only weakly present at Cambridge Bay (r=0.4) and 340 further points out the dominance of snow depth because of the large variability of the thickness 341 The Cryosphere Discuss., doi:10.5194/tc-2016-71, 2016 Manuscript under review for journal The Cryosphere Recent changes in the exchange of sea ice between the Arctic Ocean and the Canadian Arctic 418 Archipelago, J. Geophys. Res. Oceans, 118, 3595-3607, doi:10.1002/jgrc.20265. 419 The Cryosphere Discuss., doi:10.5194/tc-2016-71, 2016 1982-2012 1991-2011 1993-2013 1985-2013 1993-2010 Atmospheric forcing   False Discovery Rate (FDR) method with a global pFDR-values less than 0.10 [Wilks, 2006]. 652

List of Figures
The colorbar is linear from -10 cm dec-1 to 10 cm dec-1 and symmetric logarithmic beyond 653 these values.