Auxiliary material for COS hydrolysis in Antarctic ice cores Introduction This auxiliary material contains: 1) a description of the thermal history calculations for different ice core, 2) a description of the ice flow assumptions that are used for the thermal history calculations, and 3) ice core data pairings used in the minimizations and their impact on the results. There are four supplemental figures (Figures S1 through S4) and three supplemental tables (Tables S1 through S3) that accompany this readme file. 1. Ice thermal history calculations Calculating thermal histories for the ice samples is an under-constrained problem. Many pairs of an accumulation-rate history and an ice-flow strain history will produce the same depth-age relationship [e.g. Waddington et al., 2005]. Similarly, many pairs of a surface-temperature history and a geothermal flux will match the modern borehole temperature profiles reasonably well (Figure S1 and S2). The ice-flow histories can be constrained by using internal layers imaged by radar [e.g. Nereson et al., 1998], geologic and topographic constraints [e.g. Morse et al., 1998], and glaciological modeling [e.g. Price et al., 2007]. We use the ice-flow parameterizations suggested in previous work and described in section 2. To assess the uncertainty in the inferred thermal histories, we vary the isotope calibration factors that we use to calculate the surface-temperature histories. The geothermal flux that minimizes the misfit with the modern borehole temperature profiles for the different isotope calibration was found. At a site like WAIS Divide, the thick ice and high accumulation rate preserve information about past temperature changes [Cuffey et al., 1995; Dahl-Jensen et al., 1998; WAIS Divide Project Members, 2013]. For the thinner and lower accumulation sites like Siple Dome and Taylor Dome, there is little memory of past temperature changes. Therefore, a wide range of combinations of isotope calibration factors and geothermal fluxes match the modern borehole temperature profile sufficiently well. We use upper and lower limits on the isotope calibration factors and then test the sensitivity of the corrections for COS loss to hydrolysis under these conditions. For Taylor Dome, the recommended isotope calibration was 0.5‰ per ˚C with an uncertainty of 0.2‰ per ˚C [Steig et al., 1998]. We also calculated the thermal histories for isotope calibration factors of 0.3 ‰ per ˚C and 0.7 ‰ per ˚C (Table S1). In all three cases, the borehole fits for Taylor Dome agree with the measured borehole temperature profile within 0.6˚C at all measurement depths (Figure S2a). The COS records corrected with the different thermal histories from the three different scenarios display only minor differences (Figure S3a), indicating the thermal histories of the Taylor Dome samples from last 8 ky are well-constrained with the available borehole temperature data. For Siple Dome, the recommended isotope calibration is 0.7‰ per ˚C with an uncertainty of 0.2‰ per ˚C [Taylor et al., 2004]. We also calculated the thermal histories for isotope calibration factors of 0.5 ‰ per ˚C and 0.9 ‰ per ˚C (Table S1). The 0.7‰ per ˚C and 0.9‰ per ˚C cases both match the measured temperatures within 0.6˚C while the 0.5‰ per ˚C has a maximum misfit greater than 1˚C regardless of what geothermal flux is used (Figure S2b). The differences between the COS hydrolysis corrections for the different isotope calibrations are larger at Siple Dome than the Taylor Dome (Figure S3b). However, the differences are still small over the last 5 ky, especially between the two cases that provide a better fit to the borehole temperature measurements. 2. Ice-flow Assumptions Calculating the thermal histories requires different assumptions at each ice core site. The WAIS Divide site is located 25 km from the divide and the ice flow is best approximated as a flank site. We use the same ice-flow model inputs as described in WAIS Divide Project Members [2013]. Siple Dome is a complicated site because there have been changes in the ice thickness and the development of an ice-flow divide at the core site. We assume constant ice thickness during the Holocene and the onset of divide flow occurring between 4 and 3 ky [Nereson et al., 1998]. We use a Dansgaard-Johnsen kink height of 20% for flank flow and 70% for divide flow [Waddington et al., 2005]. We also prescribe 200 m of thinning between 15 and 12 ky based on results of Price et al. [2007]. The accumulation rate is derived from depth-age relationship for these ice-flow assumptions. The rough basal topography beneath Taylor Dome makes it difficult to observe if a Raymond Bump is present [Morse et al., 1998]. The Taylor Dome ice core was drilled 1 km (approximately two ice thicknesses) away from the topographic high [Morse et al., 1998]. We use a Dansgaard-Johnsen kink height of 30% of the ice thickness because the vertical ice-flow will be similar to a flank site, but with a small influence from the divide [Nereson and Waddington, 2002]. We assume no changes in ice thickness. 3. Ice core data pairings used in the minimizations and their impact on the results In the main body of the paper, equation (6) and figure 5 are based on minimization results using six different ice core pairings: Siple Dome (SDM) – WAIS Divide (WAISD), SDM – Byrd, SMD – Taylor Dome (TDM), WAIS-D – South Pole (SPO), WAIS-D – Byrd, and WAIS-D – TDM (Min-1 in Table S2). The cost function (C) used in the minimization is the sum of the differences in mean COS levels between the pairs, with one exception that it also includes a slope comparison for the SDM – WAIS-D pair: C c1 c2 c3 c4 c5 c6 c7 (1) where: c1 abs(SDM WAIS D) c2 abs(SDM slope WAIS Dslope) c3 abs(SDM Byrd ) c4 abs(SDM TDM ) c5 abs(WAIS D SPO) c6 abs(WAIS D Byrd ) c7 abs(WAIS D TDM ) Several additional searches were conducted using varying combinations of pairings between different ice cores (i.e. different variations of C) to explore the impact of the chosen minimization criteria on the search results (Table S2). Minimizations 2, 3, 4, and 6 focus specifically on the impact of having the Byrd and the Taylor Dome measurements as part of the search criteria. These data sets have comparatively larger uncertainties regarding the measurements, the chronology, and the model-derived thermal histories. The results demonstrate that the sensitivity of our results to whether or not the Byrd and the Taylor Dome measurements are used in the minimizations is fairly low (Figure S4a, Table S3). In minimization 5, we removed the slope equality constraint on the SDM – WAIS-D pairing, and in minimization 6, we eliminated the WAIS-D – SPO pairing replacing it instead with the SDM – TDM and the WAIS-D – TDM pairings (Table S2). Neither had a considerable impact on the results (Figure S4b, Table S3). These tests demonstrate that the slope in equation (6) (in the main body of the paper) is not sensitive to the changes in the parameterization of equation (1) (shown here) as long as the SDM – WAIS-D pair is included in C. There is seemingly more than one term in (1) that imposes this requirement. For example, the requirement that both SDM and WAIS-D agree with Byrd also requires that SDM and WAIS-D agree with each other. However, the time periods of overlap between pairings are not exactly the same and each pairing places some independent constraint on the minimization. References Cuffey, K. M., G. D. Clow, R. B. Alley, M. Stuiver, E. D. Waddington, and R. W. Saltus (1995), Large Arctic Temperature Change at the Wisconsin-Holocene Glacial Transition, Science, 270(5235), 455-458. Dahl-Jensen, D., K. Mosegaard, N. Gundestrup, G. D. Clow, S. J. Johnsen, A. W. Hansen, and N. Balling (1998), Science, 282(268), 268-271. MacGregor, J. A., D. P. Winebrenner, H. Conway, K. Matsuoka, P. A. Mayewski, and G. D. Clow (2007), Modeling englacial radar attenuation at Siple Dome, West Antarctica, using ice chemistry and temperature data, J. Geophys. Res., 112 (F03008), doi:10.1029/2006JF000717. Morse, D. L., E. D. Waddington, and E. J. Steig (1998), Ice age storm trajectories inferred from radar stratigraphy at Taylor Dome, Antarctica, Geophys. Res. Lett., 25(17), 3383-3386. Nereson, N. A., C. F. Raymond, E. D. Waddington, R. Jacobel (1998), Migration of the Siple Dome ice divide, West Antarctica, J. of Glaciol., 44(148), 643-652. Nereson, N. A. and E. D. Waddington (2002), Isochrones and isotherms beneath migrating ice divides, J. of Glaciol., 48(160), 95-108. Price, S. F., H. Conway and E. D. Waddington (2007), Evidence for late Pleistocene thinning of Siple Dome, West Antarctica. J. of Geophys. Res., 112(F03021). Taylor, K. C. et al. (2004), Abrupt climate change around 22 ka on the Siple Coast of Antarctica, Quarter. Sci. Rev., 23, 7-15. Waddington, E. D., H. Conway, E. J. Steig, R. B. Alley, E. J. Brook, K. C. Taylor, and J. W. C. White (2005), Decoding the dipstick: Thickness of Siple Dome, West Antarctica, at the Last Glacial Maximum. Geology, 33(4), 281-284.