d multiyear 
ınd for 
03 Envisat 
=0.4m. 
|. Advanced 
js used to 
SAR Wide 
ites that the 
ase in area 
a ice up the 
season into 
)03 by 254 
jerienced a 
extent for 
) Sound. 
November 
to remove 
r to surface 
field sites 
wo ICESat 
ements are 
| tracks by 
about one month. Taking the mean of the 10 nearest ICESat 
shots to the field sites, the methodology performs well. Track 15 
(mean value of 0.22 +/-0.1 m) corresponds well to the ground 
measurement (0.23 m). Track 9 (mean value of 0.47 +/-0.07 m) 
shows larger deviations to the ground measurement (0.22 m). 
Probability Density Functions (PDFs) from ICESat and the HL 
flights show good overall agreement. Two peaks in the 
histogram are seen as FY and MY sea ice. ICESat exaggerates 
the FB of FY ice when compared to the HL. This is likely 
attributed to h being too low in the southern area of the Sound. 
  
TFT TEN ENT 
   
PDF (1/mi 
  
  
  
  
AN 
m 
nl 
“0.0.05 1.0 15 20 0 
Figure 9. Statistics for ICESat (red) and HL (black) freeboards. 
4. DISCUSSION 
This initial analysis shows the applicability of ICESat derived 
freeboard information to small scale assessment of areas of 
particular interest. 
The degree of the method's accuracy is dependent upon 
successful identification of water as the lowest 5 % elevation 
retrievals and minimised interference from remaining errors 
from unaccounted variables. This methodology in small 
geographical areas becomes a play off between data availability 
to create statistics and the magnitude of error. A positive trend 
between ellipsoid and geoid is seen toward the northern end of 
the profiles in h (Figure 3). The satellite repeat pass data 
indicates this is likely attributed to residual errors in the 
modelled geoid. Such deviations will need to be considered in 
any freeboard determination. 
It is known that permanent MY ice was in existence at all times 
in the southern part of the Sound. During certain years, 
especially 2005, MY covered all the western areas of the Sound 
meaning no sea surface was visible along the groundtrack of the 
satellite. This has resulted in significant underestimation of sea 
ice freeboard along these groundtracks. Such a situation 
exemplifies the need for special care when applying the 
described methodology. Using a running mean to remove any 
residual geoid error, tidal influence and DOT (as used by 
Zwally et al. 2008 and Yi et al. 2011) was not used in this 
study. In this smaller spatial scale assessment its influence 
causes artefacts in measurements at differing ice type 
boundaries. 
Scattering of the laser beam by clouds may introduce an 
elevation error which results in the sea surface appearing lower 
(Zwally et al. 2008). Gain corrections have not been applied at 
the time of writing. Furthermore, reflections from clouds do not 
permit the use of laser reflectivity alone for ice-water 
discrimination. We have shown in one example (Figure 4) that 
additional methods like imagery need to be used in support of 
laser reflectivity. 
Data availability decreased through the study period with 2009 
having only 21 % of data available compared to 2003. 
Interannual comparison of freeboard is therefore slightly 
hindered. 
The final stage of conversion to thickness is highly influenced 
by snow depth. This step will be taken when a better knowledge 
of snow distribution across the Sound is gained. This is a 
further work in progress. The ground measurements reveal the 
complexities associated with snow cover. With track 9 the miss 
match is likely due to the size of the ICESat footprint (70 m) in 
comparison to the mean snow depth taken (0.03 m) over an area 
only 20 m in diameter. Snow is observed to be deeper in the 
east than in the west. The western ICESat groundtracks (like 
track 15) are more likely recording the actual sea ice freeboard. 
In the east (e.g. track 9), snow depths across the ICESat 
footprint may have been larger. Furthermore, intervening 
snowfall between measurements may be the source of the 
freeboard difference at cross-over points. This is only 
speculation and therefore errors in the methodology may also 
account for the disagreement in measurements. 
The extent of MY ice is assumed to be closely related to the 
passage of the iceberg B-15A across McMurdo Sound (Arrigo 
et al. 2002) from mid 2002 to mid 2005. The drastic reduction 
in MY sea ice freeboard recorded by ICESat during 2005 is 
likely a result of this. The new MY ice with significantly 
smaller freeboard than the older MY ice to the south introduced 
a bias towards overall lower mean freeboard values. 
The southern areas of the region have also been identified as 
harbouring sub-ice platelets, a component of the sea ice system 
occurring near ice shelves (Leonard et al. 2006). Subsequent 
investigation has revealed that the this sub ice layer is 
influential on sea ice growth rates (Purdie et al. 2006). It is 
possible that this interaction between the solid sea ice and the 
sub-ice platelet layer alters the freeboard. Further investigation 
is required to quantify this effect. This is especially important in 
Antarctica, where the outflow of cold ice-shelf water favours 
the formation of the sub-ice platelet layer. 
5. CONCLUSIONS 
This preliminary investigation provides the first time series of 
satellite derived sea-ice freeboard in the western Ross Sea. 
Using a combination of ICESat elevation data and auxiliary 
satellite imagery this investigation reveals evidence of steadily 
increasing freeboard heights of MY sea ice in McMurdo Sound 
for the period 2003-2009. No significant trends in freeboard 
were detected for FY sea ice. For localized areas the satellite 
derived freeboard height shows reasonable agreement with 
ground measurements. Freeboard and sea ice extent appear to 
be sensitive to local drivers of oceanic change including iceberg 
passages. Further work is needed to quantify the role of other 
drivers such as the sub-ice platelet layer in the region. From 
here, further methods including the analysis of local tide 
information to determine ICESat derived freeboard will provide 
an alternative data set for comparison. Errors associated with 
freeboard and snow depth estimation from satellite remote 
sensing platforms mean that in situ information is crucial in the 
development and validation of retrieval methodologies.