1

1 Instability time-scales of Antarctic outlet glaciers from similitude analysis

2 3 A. Levermann1,2,3* & J. Feldmann1,3 4 5 1Potsdam Institute for Climate Impact Research, Potsdam, Germany. 6 2LDEO, Columbia University, New York, USA. 7 3Institute of Physics, Potsdam University, Potsdam, Germany. 8 9 *Correspondence to: Anders Levermann ([email protected]). 10 11 Abstract 12 Instabilities in have the potential to significantly raise sea level globally. Recent 13 observations and ice-dynamic modeling suggest that such an instability might have been 14 triggered in West Antarctica. Although other regions have similar topographic predispositions 15 it is so far unclear how fast these instabilities would unfold if they were initiated. The relative 16 initial response time of different ice streams after destabilization can be estimated from their 17 current force balance that reflects their relation of surface-mass balance, basal friction and 18 topographic geometry. Combining observations with a similarity analysis of the ice-dynamic 19 equation we show here that the already-triggered instabilities are likely to be the fastest in 20 Antarctica. Consistent with earlier studies the response time of is found to 21 be about twice as fast as that of while the main outlet of East Antarctica’s 22 Wilkes Basin would be about twice as slow if destabilized. 23 2

24 Main text 25 Sea-level rise poses a future challenge for coastal regions worldwide (1). The contribution 26 from mass loss of the West Antarctic Ice Sheet has been increasing over the last two decades 27 (2). Large parts of the Antarctic Ice Sheet rest on a retrograde marine bed, i.e. on a bed 28 topography that lies below the current sea level and is down-sloping inland (3, 4). This 29 topographic situation makes these parts of the ice sheet prone to a so-called marine ice-sheet 30 instability (MISI) (5, 6). Currently these instabilities represent the largest uncertainty in 31 projections of future sea-level rise (7, 8): If the grounding line (that separates grounded from 32 floating ice) enters a region of retrograde bed slope, a positive ice-loss feedback can be 33 initiated (9–12). If it is not hindered by the buttressing effect of ice shelves and topographic 34 features (13–15) or strong basal friction (8, 16), it unfolds and yields a self-sustained ice 35 discharge from the region. 36 The ongoing retreat of two major outlets of West Antarctica, Pine Island Glacier (PIG) and 37 Thwaites Glacier (TG), initiated by warm-water entrainment into their ice-shelf cavities and 38 resulting basal melting (17), is pointing toward a developing marine ice-sheet instability (9, 39 10, 18, 19) with the potential of raising global sea level by more than 3 m (12, 20). Similar 40 warm-water exposition might apply to other Antarctic ice shelves in the future (8, 21, 41 22) bearing the potential to trigger unstable grounding-line retreat. It has been demonstrated 42 that East Antarctica's Wilkes Subglacial Basin (WSB), would, once destabilized, contribute 43 by 3 – 4 m to global sea level (11). The marine part of the adjacent Aurora Subglacial Basin 44 stores ice of around 3.5 m sea-level equivalent (22). Other studies investigated the response 45 of the Filchner-Ronne tributaries to enhanced melting (15, 23, 24) and suggest that instability 46 might not unfold there, possibly due to the stabilizing buttressing effect of the large ice shelf 47 and narrow bed troughs (13, 14). In total, the marine catchment basins of Antarctica, that are 48 connected to the ocean store ice masses of about 20 m of sea-level equivalent (Figure S1). 49 Besides the question of whether local instabilities are already triggered or will be triggered in 50 the future the question remains on which time scales a potentially unstable retreat would 51 evolve if they had been initiated. It has been argued that ice-sheet models are not suitable for 52 an analysis of these time scales because they lack necessary detail in certain physical 53 processes such as basal friction (7). 54 Here we try to contribute information from a different approach. The currently observed 55 shape of the outlet glacier represents the balance of the different forces that act on the ice. 3

56 That is the balance of the forces caused by the bed and ice topography, the surface mass 57 balance and the frictional forces from the ice interaction with itself and its surrounding. This 58 static situation carries the information for the initial ice-sheet response after destabilization. 59 We combine observations with properties of the fundamental ice-dynamic equations. Eleven 60 Antarctic outlets with similar topography are analyzed to estimate their relative response 61 times to instability. 62 The time-scaling laws used here are derived from the governing equations of shallow ice- 63 stream flow (25) (see Supporting Online Material). In their dimensionless form these 64 equations have three independent parameters, analogous to the Reynolds number in the 65 Navier-Stokes equation (26). If all three of theses parameters are the same for topographically 66 similar situations the solution of the ice-dynamical equation will be similar. The parameters 67 are combinations of characteristics of the ice dynamics, physical ice properties, boundary 68 conditions and geometry (concept of similitude/similarity (27), for a detailed derivation 69 compare ref. (25)). The resulting scaling laws are consistent with analytic solutions of the ice 70 equations (6). 71 Here the individual geometry and physical conditions of the examined ice streams, that are 72 obtained from observational data (4, 28) and model inversion (29), are used to calculate 73 response times after destabilization relative to the reference, PIG. The applied method 74 assumes conditions of unstable retreat in which ice-shelf buttressing is a minor constraint 75 taking into account bed friction as a possible stabilizing factor. The investigation is 76 constrained to the short initial time period after destabilization during which the grounding 77 line passes the retrograde bed slope near the coast. 78 The procedure is schematized in Fig. 1: For each of the considered tributaries a transect is 79 defined that represents a center line of the and covers its potentially unstable bed 80 section close to the grounding line (Figs. 1a and b). Fitting a linear slope to the retrograde bed 81 section (blue line), the bed elevation b at the starting point of the slope and the slope 82 magnitude s serve as characteristic geometric scales for the sloping bed. Basal friction C is 83 averaged along this section to represent local conditions whereas the surface mass balance a 84 is obtained from averaging over the catchment that feeds the ice stream (Figures S2-S12). For 85 each tributary these physical parameters are taken relative to the reference values of PIG (Fig. 86 1c). 4

87 The resulting scaling ratios (Fig. 1d, values given in Table 1) are then used to calculate 88 response-time ratios τ=T' /T via two independent scaling conditions. The first equation 89 results from the shallow-shelf approximation of the stress-balance which accounts for the fast 90 plug-like flow of ice streams (see Supporting Online Material) and involves the ratios for

91 vertical scale β=b'/b , slope σ =s'/ s and friction γ=C' /C which yields

τ = σ− 2 γ . (1) 92 As a consequence, for example, more slippery bed conditions and a steeper bed slope yield a 93 faster response. An independent second relation is derived from mass conservation and 94 incorporates the ratios for the vertical scale and the surface mass balance, δ=a' /a , i.e.,

~τ =βδ− 1 . (2) 95 Here an initially stable situation with stronger accumulation but initially thinner ice at the 96 grounding line, for example, results in a faster response of the instability after destabilization. 97 The two time-scaling values τ and ~τ are used as independent estimates of the response 98 time and to a certain extent give a measure for the uncertainty range for each analyzed ice 99 stream that accounts for observational uncertainty and approximations in our approach. Ice 100 streams with large uncertainty in their calculated time scaling are reported, but considered 101 unsuitable for the approach used here (see Supporting Online Material). For each of the six 102 remaining tributaries (Table 1, blue) the relative error of the two calculated response-time 103 ratios with respect to their mean value is smaller than 28 %. 104 The individual retrograde bed slopes vary in magnitude by a factor of 2 and bed elevation 105 differs by up to ~1000 m (Fig. 2a). The lengths of the slope sections indicate how far the 106 considered retrograde slopes reach inland before an (intermediate) section with vanishing or 107 up-sloping bed follows (Figs. 1a,b and S2-12). After applying a geometric scaling according 108 to the obtained ratios of vertical and horizontal scales (Table 1) the bed profiles collapse 109 towards the reference showing similar down-sloping, while still exhibiting their characteristic 110 pattern (Figs. 2b and S7). This assures that the chosen geometric measures adequately reflect 111 the characteristics of the retrograde bed slopes. 112 Our results reveal that TG has the smallest response time to destabilization of all analyzed 113 outlets (Fig. 3), being between 1.25 to almost 2 times as fast as PIG, which is the second 114 fastest in the analyzed ensemble. For the chosen outlet of WSB the two calculated ratios 115 indicate a response that would be twice as slow as PIG. East Lambert Rift (ELR) and Institute 5

116 Ice Stream (IIS) that feed Amery Ice Shelf and Filchner-Ronne Ice Shelf, respectively, are 117 found to be 3 to 6 times slower than PIG. By far the slowest response shows Support Force 118 Glacier (SFG) which is (more than) 10 times slower than PIG. 119 The relatively large response times of ELR, SFG and IIS are consistent with results from 120 dynamical modeling that find the glaciers of Filchner-Ronne and Amery ice shelves to be far 121 less sensitive to instability (15, 23, 24, 30) than WSB (11) and PIG and TG in the Amundsen 122 Sea sector (9, 10, 12). In these cases it is possible that no instability unfolds if ice-shelf 123 buttressing remains relevant. 124 125 According to observations, PIG and TG show the largest ice discharge rates in the Amundsen 126 Sea Sector of West Antarctica since the last four decades, including phases of rapid ice speed- 127 up, dynamic thinning and grounding-line retreat (18, 19). PIG's most recent acceleration was 128 found to coincide with the ungrounding of a region of only lightly grounded ice and the 129 grounding line that is now at the upstream end of this area stabilized during the last few years 130 (19). Meanwhile PIG's deceleration in ice discharge is more than compensated by TG's 131 speed-up which can be attributed to reduced ice-shelf buttressing that previously stabilized 132 the upstream grounded ice (19). It thus remains an open question whether the current TG 133 destabilization will result in retreat rates larger than PIG's recent retreat rates and hence 134 confirm our finding of a faster time scale of TG compared to PIG. Recent regional model 135 simulations of these regions (9, 10) indicate that TG is up to a factor two faster than PIG after 136 the basal ice-shelf melting has ceased and the glaciers have relaxed into a constant ice 137 discharge. 138 139 Besides the time scaling there is a further physical parameter, the ice softness A, that can be 140 obtained via a separate scaling law (see Supporting Online Material). The calculated ice- 141 softness ratios differ by two orders of magnitude (Table 1) and for most of the ice streams 142 discussed here a lower softness value corresponds to a larger time scale. In other words, 143 stiffer ice tends to delay response to instability. 144 Our results are based on scaling laws that are derived from an approximation of the full- 145 Stokes stress balance. The approximation accounts for the dynamics of shallow ice streams 146 that are characterized by dominating longitudinal stresses and rapid sliding and typically have 147 an onset of several 10 to 100 km upstream of the grounding line. The here conducted analysis 6

148 focuses precisely on such regions. The examined sections are all situated in the proximity of 149 the grounding line and indeed exhibit slippery bed conditions and fast ice flow as indicated 150 by inferred basal friction and velocity data, respectively (Figs. S2-12 f and g). 151 We find that the presented method is not applicable to all of the tributaries analyzed here. 152 Each of the discarded ice streams exhibits extremely slippery bed conditions, with basal- 153 friction values of up to three orders of magnitude smaller than the reference (Table 1). On the 154 other hand these outlets also have the smallest bed slopes of the ensemble. Both slope and 155 friction are hence subject to large relative uncertainties that mutually amplify in the 156 calculation of ~τ and hence amplify the mismatch with τ . 157 Observational data of bed and ice geometry still show substantial gaps in coverage, though 158 the regions analyzed here lie at the lower end of the uncertainty range (ref.(4), their Figs. 11 159 and 12). We use the bed elevation at the starting point of the retrograde slope rather than at 160 the current grounding line location to estimate the vertical scales of the tributaries. Note that 161 the error in bed elevation (~100 m) is smaller than the uncertainty in grounding-line position 162 (~1 km). Measurements of the Antarctic surface mass balance are in general sparsely 163 distributed over the continent. In contrast, results from numerical modeling are known to 164 have biases with respect to observations. We thus use a surface-mass-balance dataset obtained 165 from a regional climate model that was calibrated by the available observational data to 166 reduce uncertainty (28). 167 The method used in this study allows to compare the relative time scales of Antarctic outlets 168 based on their specific geometry and physical conditions. We emphasize that this approach is 169 not capable of inferring absolute response times of individual tributaries and is strongly 170 limited by not accounting for ice-shelf buttressing, i.e. it only estimates the time scale after no 171 significant ice shelf is present anymore. Continued observation of the retreat of PIG and TG 172 that extend the already available data (18) might allow to deduce long-term trends and thus 173 give estimates for the absolute time scales of instability. These could be used to calibrate the 174 method presented here that includes several simplifications. Uncertainties will reduce as 175 observational datasets will further improve. 7

176 Figures

Figure 1: Schematic visualizing the applied similitude analysis to obtain the response-time ratio τ between different Antarctic outlets as a function of their specific physical properties, here exemplified for Pine Island Glacier and Thwaites Glacier. Blue variables are obtained from observations and model data. The equations for τ (Eqs. 1 and 2) result from a dimensional analysis of the stress balance (25). 8

Figure 2: a Bed topography of reverse slope sections of the six examined Antarctic outlets that fulfill the time-scaling criteria. b Bed topography from panel a scaled according to similitude theory using the horizontal and vertical length-scale ratios (see Table 1) and normalized to the dimensions of the PIG slope (that has the same aspect ratio in both panels). For better visibility the x and y axis are cut (see Fig. S4 for full extent of the ELR bed section). 9

SFG IIS ELR

PIG

TG

WSB

Figure 3: Map of inverse response time of Antarctic tributaries (two circles for each, corresponding to the two independent estimates, τ , ~τ , see Table 1) relative to Pine Island Glacier (PIG, grey circle) as obtained from similitude analysis. Rectangles denote examined regions as displayed in the Figures of the Supporting Online Material. Clock-wise: Thwaites Glacier (TG), Wilkes Subglacial Basin (WSB), East Lambert Rift (ELR), (SFG) and (IIS). 10

α σ γ τ ~ Tributary β δ ζ τ c1 c2

PIG 1 1 1 1 1 1 1 1 1 1

TG 0.86 1.03 1.2 0.77 1.28 1.72 0.53 0.8 0.42 0.2

MAIS 3.54 0.99 0.28 0.003 0.34 31.74 0.03 2.87 -1.93 0.98

BIS 20.37 1.04 0.05 0.003 0.34 0.79 1.13 3.05 15.26 0.45

MIS 6.06 1.25 0.21 0.78 0.41 0.03 18.17 3.04 0.12 0.71

WSB 0.67 1.11 1.67 5.76 0.41 0.35 2.07 2.73 1.61 0.14

TOG 2.37 2.84 1.2 0.22 0.53 0.29 0.15 5.31 -5.07 0.95

ELR 0.41 0.54 1.31 10.15 0.16 1.08 5.93 3.42 0.49 0.27

SFG 2.67 1.89 0.71 6.65 0.2 0.01 13.35 9.55 0.69 0.17

FIS 8.75 2.56 0.29 0.04 0.2 0.14 0.43 12.93 -20.85 0.94

IIS 2.5 1.77 0.71 2.18 0.33 0.04 5.34 4.36 1.29 0.1

Table 1: Scaling ratios for all examined Antarctic outlets, clockwise, starting with the reference Pine Island Glacier (PIG). The last two columns give the values from application of the criteria according to which tributaries are accepted (blue) or discarded (grey). The other tributaries are abbreviated as follows: Thwaites Glacier (TG), MacAyeal Ice Stream (MAIS), (BIS), (MIS), Wilkes Subglacial Basin (WSB), (TOG), East Lambert Rift (ELR), Support Force Glacier (SFG), (FIS) and Institute Ice Stream (IIS)

179 11

180 References and Notes 181 182 1. C. B. Field et al., IPCC, 2014: Climate Change 2014: Impacts, Adaptation, and 183 Vulnerability. Part A: Global and Sectoral Aspects. Contribution of Working Group II 184 to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change 185 (Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 186 2014). 187 2. A. Shepherd et al., A Reconciled Estimate of Ice-Sheet Mass Balance. Science (80-. ). 188 338, 1183–1189 (2012). 189 3. N. Ross et al., Steep reverse bed slope at the grounding line of the Weddell Sea sector 190 in West Antarctica. Nat. Geosci. 5, 393–396 (2012). 191 4. P. Fretwell et al., Bedmap2: improved ice bed, surface and thickness datasets for 192 Antarctica. Cryosph. 7, 375–393 (2013). 193 5. J. Weertman, Stability of the junction of an ice sheet and an ice shelf. J. Glaciol. 13, 3– 194 11 (1974). 195 6. C. Schoof, Ice sheet grounding line dynamics: Steady states, stability, and hysteresis. 196 J. Geophys. Res. 112, F03S28 (2007). 197 7. T. F. Stocker et al., “IPCC, 2013: climate change 2013: the physical science basis. 198 Contribution of working group I to the fifth assessment report of the intergovernmental 199 panel on climate change” (Cambridge University Press, 2013). 200 8. C. Ritz et al., Potential sea-level rise from Antarctic ice-sheet instability constrained by 201 observations. Nature, 1–14 (2015). 202 9. L. Favier et al., Retreat of Pine Island Glacier controlled by marine ice-sheet 203 instability. Nat. Clim. Chang. 5, 117–121 (2014). 204 10. I. Joughin, B. E. Smith, B. Medley, Marine Ice Sheet Collapse Potentially Under Way 205 for the Thwaites Glacier Basin, West Antarctica. Science (80-. ). 344, 735–738 (2014). 206 11. M. Mengel, a Levermann, Ice plug prevents irreversible discharge from East 207 Antarctica. Nat. Clim. Chang. 4, 451–455 (2014). 208 12. J. Feldmann, A. Levermann, Collapse of the West Antarctic Ice Sheet after local 209 destabilization of the Amundsen Basin. Proc. Natl. Acad. Sci., 201512482 (2015). 210 13. D. Goldberg, D. M. Holland, C. Schoof, Grounding line movement and ice shelf 211 buttressing in marine ice sheets. J. Geophys. Res. 114, F04026 (2009). 212 14. G. H. Gudmundsson, J. Krug, G. Durand, L. Favier, O. Gagliardini, The stability of 213 grounding lines on retrograde slopes. Cryosph. 6, 1497–1505 (2012). 12

214 15. M. Mengel, J. Feldmann, A. Levermann, Linear sea-level response to abrupt ocean 215 warming of major West Antarctic ice basin. Nat. Clim. Chang. 5 (2015), 216 doi:10.1038/nclimate2808. 217 16. I. Joughin et al., Basal conditions for Pine Island and Thwaites Glaciers, West 218 Antarctica, determined using satellite and airborne data. J. Glaciol. 55, 245–257 219 (2009). 220 17. A. Jenkins et al., Observations beneath Pine Island Glacier in West Antarctica and 221 implications for its retreat. Nat. Geosci. 3, 468–472 (2010). 222 18. E. Rignot, J. Mouginot, M. Morlighem, H. Seroussi, B. Scheuchl, Widespread, rapid 223 grounding line retreat of Pine Island, Thwaites, Smith, and Kohler glaciers, West 224 Antarctica, from 1992 to 2011. Geophys. Res. Lett. 41, 3502–3509 (2014). 225 19. J. Mouginot, E. Rignot, B. Scheuchl, Sustained increase in ice discharge from the 226 Amundsen Sea Embayment, West Antarctica, from 1973 to 2013. Geophys. Res. …. 227 41, 1–9 (2014). 228 20. J. L. Bamber, R. E. M. Riva, B. L. Vermeersen, A. M. LeBrocq, Reassessment of the 229 potential sea-level rise from a collapse of the West Antarctic Ice Sheet. Science. 324, 230 901–903 (2009). 231 21. H. H. Hellmer, F. Kauker, R. Timmermann, J. Determann, J. Rae, Twenty-first-century 232 warming of a large Antarctic ice-shelf cavity by a redirected coastal current. Nature. 233 485, 225–228 (2012). 234 22. J. S. Greenbaum et al., Ocean access to a cavity beneath Totten Glacier in East 235 Antarctica. Nat. Geosci. 8, 1–5 (2015). 236 23. A. P. Wright et al., Sensitivity of the Weddell Sea sector ice streams to sub-shelf 237 melting and surface accumulation. Cryosph. 8, 2119–2134 (2014). 238 24. M. Thoma, J. Determann, K. Grosfeld, S. Goeller, H. H. Hellmer, Future sea-level rise 239 due to projected ocean warming beneath the Filchner Ronne Ice Shelf: A coupled 240 model study. Earth Planet. Sci. Lett. 431, 217–224 (2015). 241 25. J. Feldmann, A. Levermann, Cryosph. Discuss., 10 (2016), doi:10.5194/tc-2015-226, 242 http://www.the-cryosphere-discuss.net/tc-2015-226. 243 26. O. Reynolds, An Experimental Investigation of the Circumstances Which Determine 244 Whether the Motion of Water Shall Be Direct or Sinuous, and of the Law of Resistance 245 in Parallel Channels. Philos. Trans. R. Soc. London. 174, 935–982 (1883). 246 27. Rayleigh, The Principle of Similitude. Nature. 96, 396–397 (1915). 13

247 28. W. J. van de Berg, M. R. van den Broeke, C. H. Reijmer, E. van Meijgaard, 248 Reassessment of the Antarctic surface mass balance using calibrated output of a 249 regional atmospheric climate model. J. Geophys. Res. 111, D11104 (2006). 250 29. M. Morlighem, H. Seroussi, E. Larour, E. Rignot, Inversion of basal friction in 251 Antarctica using exact and incomplete adjoints of a higher-order model. J. Geophys. 252 Res. Earth Surf. 118, 1746–1753 (2013). 253 30. Y. Gong, S. L. Cornford, a. J. Payne, Modelling the response of the – 254 Amery Ice Shelf system, East Antarctica, to uncertain climate forcing over the 21st and 255 22nd centuries. Cryosph. 8, 1057–1068 (2014). 256 257 14

258 Acknowledgements 259 The study was partially financially supported by the German Federal Environmental 260 Foundation (DBU) the Deutsche Forschungsgemeinschaft (DFG) under Priority Programme 261 1158. We are grateful to M. Morlighem for providing the dataset of Antarctic basal friction.

262 Author contributions

263 A.L. designed the study and J.F. carried out the calculation and prepared the figures. A.L. and 264 J.F. wrote the text.

265 Competing financial interests

266 The authors declare no competing financial interests.