Figure 2.1: In situ sea-ice sampling sites and ship tracks for SIPEX-II (orange,n= 8 with 5 usable transects), SIPEX (green, n= 14 with 12 usable transects) and ARISE (red, n= 9 with 12 useable transects). Sites used for analysis in this paper are marked
with inverted triangles. On each voyage site numbering increased from west (right) to east (left). Bathymetry and contours from the GEBCO digital atlas, Antarctic coastline
from the SCAR Antarctic Digital Database v6.
2.2
Data description
In situ data from three research voyages aboard RSV Aurora Australis to the East Antarctic pack-ice zone in spring are used for the following analyses: ARISE 2003 (Massom et al., 2006), SIPEX 2007 (Worby et al., 2011) and SIPEX-II 2012 (Heil et al., 2014) all in the region 110 - 130 E. Figure 2.1 shows the relative location of voyage tracks and field sampling sites. The general strategy for data collection is to measure snow depth, ice thickness and ice freeboard in holes drilled through the sea ice at 1 m intervals along 100 to 200 m ‘transects’. The ideal scenario does not always occur, and in this study profiles vary from 50 to 500 m in length with sample spacing up to 5 m between drill holes. Such floe-scale measurements usually coincide with a multi-disciplinary observation program on the same ice floe, and the overall measurement site is referred to as an ‘ice station’. These sites are inherently biased toward sea ice which is both ship-navigable and thick enough to work on, meaning that the extremes of the sea-ice thickness distribution are generally not represented. However, some randomness is introduced to the sampling strategy by logistical constraints, with sampling lines being determined in an uncontrolled fashion by the direction of surface wind relative to the orientation of the ship once parked at the ice floe.
Measurements of snow depth were collected by forcing graduated probes into undisturbed snow until the end of the probe contact an unbreakable hard layer. In general this layer was the ice/snow interface. Snow depth was also measured when snow was cleared for drilling through ice. If a disparity existed between probed and snow pit measurements, the latter was recorded. Ice thickness and freeboard measurements were collected by
2.2. DATA DESCRIPTION 24
voyage µZi ˜x M o Zi µFi x˜ M o Fi µZs x˜ M o Zs N valid (N total)
ARISE 1.21 0.65 0.5 1.08 0.04 0.03 0.02 0.12 0.21 0.13 0.05 0.2 1132 (2148)
SIPEX 0.93 0.79 0.66 0.51 0.06 0.035 0 0.10 0.15 0.12 0.01 0.13 1558 (2415)
SIPEX-II 2.33 1.75 0.97 1.63 0.12 0.06 0.05 0.19 0.41 0.4 0.5 0.23 439 (1135)
Combined 1.23 0.87 0.5 1.07 0.06 0.03 0.02 0.13 0.21 0.14 0.05 0.19 3129 (5698)
Table 2.1: Summary statistics for in situ sea-ice observations from ARISE, SIPEX and SIPEX-II. Measurements in metres,Zi = ice thickness,Fi = ice freeboard, Zs = snow
depth. The symbol ˜x denotes the population median,M othe mode.
weighted tape measures or rulers with expanding hooks which were designed to lodge at the undersurface of the ice. Ice thickness was measured at the ice surface, freeboard where the measuring device passed through the water surface. In the case of negative ice freeboard, measurements of the di↵erence between the ice surface and water level were taken after flooding through the drilled hole had ceased. Where ice freeboard exceeds 0.15 - 0.2 m, the sea surface is not directly visible in the drill hole. Measurements rely on observers identifying the moment a weighted tape measure or ruler contacts the sea surface as it is lowered. For all measurements, a realistic maximum precision in field conditions was 5 mm. For ice thickness, a further assumption is made that devices designed to measure ice thickness are catching on the underside of the ice and not in some interstitial space between blocks of ice.
From these three voyages there were 5698 sampling sites, of which 3129 were suitable for use in this study - the requirement for usage being that coincident measurements of ice thickness, freeboard and snow depth occurred at the site. All three measurements are required to develop empirical models for snow depth, so any drill sites where fewer than these three measurements were recorded are excluded from this analysis.
Table 2.2 gives a transect-by-transect overview of the in situ data used in this study. Observations are often irregularly spaced, and at some sites external factors (time, safety) constrain sampling to larger intervals with points 2 or 5 m apart. Considering the di↵erences between sampling sites in similar locations (e.g., ARISE site 6, SIPEX site 12, SIPEX-II site 8), it is attractive to treat each of the three voyages as independent datasets which tell a story about inter-season variability. However, given uncertainty about the forcing during ice formation and distribution prior to each voyage, sampling goals, and sampling strategies, a rigorous inter-voyage comparison is not possible. For the purposes of this paper - deriving and assessing an empirical model of snow depth from total freeboard, the dataset is treated as 3129 independent observations, and ask the reader to re-imagine the data presented here as a contiguous set of observations with roughly metre spacing. While not strictly accurate, there are no specific georeferencing data for each observation - and selection of sampling sites is a pseudo-random process. Given these factors, the data used here are treated as a contiguous set.
Of relevance to this study are the freeboard-related values presented in columns 12, 13 and 14 of Table 2.2. The ratio of ice freeboard to total freeboard Fi/F for each observation is derived to show the contribution of ice to the total freeboard signal, giving mean values in