Antarctica CryoSat-2 Digital Elevation Model (DEM)
Product User Guide
Science Lead:
Andrew Shepherd
University of Leeds, UK
[email protected]
Technical Officer:
Tommaso Parinello
ESA ESRIN, Frascati, Italy
[email protected]
Consortium:
University of Leeds, Centre for Polar Observation and Modelling
University College London, Centre for Polar Observation and Modelling
Contact:
Thomas Slater
[email protected]
1
Contents
Change Log .............................................................................................................................................. 3
Acronyms ................................................................................................................................................ 4
1. Abstract ............................................................................................................................................... 5
2. Data and methods............................................................................................................................... 5
2.1. Cryosat-2 elevation measurements ............................................................................................. 5
2.2. DEM generation ....................................................................................................................... 5
2.3. Airborne elevation measurements .......................................................................................... 8
2.4 DEM evaluation ......................................................................................................................... 9
3. Results ............................................................................................................................................... 10
3.1. Comparison of DEM to airborne elevation measurements: observed grid cells ................... 10
3.2. Comparison of DEM to airborne elevation measurements: interpolated grid cells .............. 12
4. Product Format ................................................................................................................................. 13
5. Acknowledgements........................................................................................................................... 14
6. References ........................................................................................................................................ 15
2
Change Log
Version Author
number
1.1
Thomas
Slater
1.2
Thomas
Slater
Affected Section
Reason/Description
Status
All
Document Creation
Title page
Acronyms
Acknowledgements
Addition of listed
sections
Released on
20/03/2017
Released to
AS on
25/04/2017
Files:
Antarctica_DEM_CryoSat2_v1.2.nc
(‘nvals’ and ‘chisq’ parameters)
Antarctica_DEM_CryoSat2_nvals_v1.2.tiff
Antarctica_DEM_CryoSat2_chisq_v1.2.tiff
3
Fixed incomplete
data coverage on
terminus of Ross Ice
Shelf in ‘nvals’ and
‘chisq’ parameters
Released to
ftp on
26/04/2017
Acronyms
Acronym
Explanation
ATM
Airborne Topographic Mapper
DEM
Digital Elevation Model
EAIS
East Antarctic Ice Sheet
LRM
Low Resolution Mode
OCOG
Offset Centre Of Gravity
RLA
Riegl Laser Altimeter
SARIn
Synthetic Aperture Radar Interferometric mode
SIRAL
SAR Interferometer Radar Altimeter
WAIS
West Antarctic Ice Sheet
4
1. Abstract
This document describes a new Digital Elevation Model (DEM) of the Antarctic ice-sheet and iceshelves produced using data acquired by the CryoSat-2 radar altimeter. Posted on a 2 x 2 km grid, the
DEM provides an elevation measurement for 97% of the floating ice-shelves and 91% of the grounded
ice-sheet, up to a latitude of 88oS. The remaining grid cells are interpolated to provide a continuous
grid covering 100 % of the ice-shelves and ice-sheet North of 88o S. The model utilises both Low
Resolution Mode and Synthetic Aperture Radar Interferometric Mode acquisitions from the baselinec distribution of the level 2 product. In total, approximately 2.5 x 108 elevation measurements were
used, acquired between July 2010 and July 2016.
The accuracy of the presented model is assessed through comparison to approximately 1.9 x 107
airborne laser altimeter measurements from NASA IceBridge campaigns (more information available
at https://nsidc.org/data/icebridge) over the time period December 2008 to December 2014, in
various locations across Antarctica.
2. Data and methods
2.1. Cryosat-2 elevation measurements
The initial input data were recorded by the SIRAL (SAR Interferometer Radar Altimeter) instrument,
mounted on the CryoSat-2 satellite, over the 6 year time period between July 2010 and July 2016.
Over the interior of the continental ice-sheet, data was selected from Level 2 Low Resolution Mode
(LRM) data files. Over the ice-sheet margins and ice-shelves the chosen data came from Synthetic
Aperture Radar Interferometric (SARIn) mode acquisitions (information regarding the geographical
extent of the CryoSat-2 mode mask is available online at https://earth.esa.int/web/guest//geographical-mode-mask-7107). The Level 2 product has a series of geophysical corrections applied
to correct the selected measurements for the following: off-nadir ranging due to slope, dry
atmospheric propagation, wet atmospheric propagation, ionosphere propagation, solid earth tide and
ocean loading tide. For both operating modes data from the baseline-c distribution of the Level 2
product was used. Over the LRM mode mask area elevation estimates were retrieved from CryoSat-2
waveforms using the OCOG retracking algorithm. For SARIn regions the Level 2 SARIn retracker was
used. In total, approximately 2.5 x 108 elevation measurements were used to generate the DEM.
2.2. DEM generation
CryoSat-2 elevation measurements were separated into 3361315 regularly spaced 2 km squared
geographical regions. The plane fit technique (McMillan et al, 2014), was used to separate the various
contributions to the measured elevation fluctuations within each region. This method is designed to
5
best suit CryoSat-2’s 369 day orbit cycle, which samples all data acquired along a dense network of
ground tracks with few coincident repeats. The plane-fit technique (Equation 1) models elevation (𝑍)
as a quadratic function of local surface terrain (𝑥,𝑦), a time invariant term ℎ accounting for anisotropy
in radar penetration depth depending on satellite direction (Armitage et al, 2013), and a linear rate of
elevation change 𝑡. ℎ is set to 0 or 1 for an ascending or descending pass respectively.
𝑍(𝑥, 𝑦, 𝑡, ℎ) = 𝑧̅ + 𝑎0 𝑥 + 𝑎1 𝑦 + 𝑎2 𝑥 2 + 𝑎3 𝑦 2 + 𝑎4 𝑥𝑦 + 𝑎5 (ℎ) + 𝑎6 𝑡
(1)
Coefficients were retrieved in each grid cell using an iterative least squares fit to the observations to
minimise the impact of outliers. Unrealistic estimates for coefficients resulting from poorly constrained
model fits were discarded. The criteria chosen for the removal of poorly constrained model fits are
given in Table 1. This approach provides, on average, in excess of 80 elevation measurements to
constrain each solution. The plane fit method allows data from 6 years of continuous CryoSat-2
observations to be used to create the DEM, without being unduly affected by fluctuations in surface
elevation during the acquisition period. In addition, it also allows for the retrieval of ice sheet elevation
and rate of elevation change from the same data in a self-consistent manner.
Parameter
Selection criteria
dh/dt (m/yr)
-10 ≤ dh/dt ≤ 10
Uncertainty in dh/dt (m/yr)
≤ 0.4
Root mean square (RMS) of elevation residuals from the model ≤ 10
fit (m)
Time period covered by data used in dh/dt fit (yrs)
≥2
Number of data points in grid cell
≥ 15
Surface slope (degrees)
≤5
Table 1: Selection criteria used to remove elevation estimates resulting from poorly constrained
model fits
6
Figure 1: Area coverage per drainage basin provided by plane fit solution of CryoSat-2 elevation
measurements. Here the Zwally definition of drainage basins is used.
The DEM was generated from the mean elevation term, 𝑧̅, in equation 1 within each 2 x 2 km grid cell.
In addition, any elevation estimate deviating by more than 200 m from the median value of all level 2
elevations in its corresponding grid cell was removed. Approximately 8% of the elevation estimates
were eliminated from poorly constrained fits, and less than 0.5% from deviations from the binned
median. After all filters were applied, the plane fit approach provided an elevation estimate in 91% of
grid cells within the ice-sheet, and 97% in the ice-shelves. The data coverage of the grounded ice-sheet
provided by the plane fit method, separated into individual drainage basins, is shown in Figure 1.
Elevation values for grid cells not containing any data were estimated using a local mean filling
algorithm. This algorithm replaces an empty grid cell with the mean value of all data points within a
25 km search radius. On the 2 x 2 km grid the DEM is projected on, a 25 km search radius corresponds
to 13 grid cells. The interpolated DEM provides an increased data coverage of 99% and 100% for the
ice-sheet and ice-shelves respectively. The pole hole is not interpolated due to interpolation distances
exceeding the 25 km search radius, and a desire to keep the DEM a product of CryoSat-2 data only. The
generated DEM and number of CryoSat-2 elevation measurements within each grid cell is illustrated
in Figure 2.
7
Figure 2: DEM of Antarctica from CryoSat-2 radar altimetry (left) and number of CryoSat-2
measurements in each grid cell (right)
2.3. Airborne elevation measurements
Elevation measurements acquired by airborne laser altimeters, as part of NASA’s IceBridge survey,
were used to assess the accuracy of the generated DEM. NASAs IceBridge mission, running since 2009,
is the largest airborne polar survey ever undertaken (Koenig et al 2010). The primary goal of IceBridge
is to maintain a continuous time series of laser altimetry over the Earth’s polar regions, bridging the
gap between ICESat, which stopped collecting data in 2009, and ICESat-2, planned for launch in 2017.
Data from two IceBridge airborne LIDAR instruments were used to evaluate the DEM:

Airborne Topographic Mapper (ATM), acquired between March 2009 and December 2014
over the following regions of the continental ice-sheet: the Antarctic Peninsula,
Bellingshausen, Amunsden and Getz sectors of West Antarctica, and the Transantarctic
Mountains, Oates Land and the plateau region of East Antarctica. The following ice-shelves
were also surveyed: the Larsen C, Pine Island, Thwaites, Wilkins, Abbot, Getz, Ross and RonneFilchner.

Riegl Laser Altimeter (RLA), acquired between December 2008 and January 2013 over the
Antarctic Peninsula, West Antarctica, the Transantarctic Mountains, Queen Maud Land,
Totten Glacier and the Ross ice-shelf.
The locations of the ATM and RLA datasets are given in Figure 3. In total, approximately 1.9 x 107
surface elevation measurements acquired from airborne laser altimetry were compared to the DEM.
These datasets were chosen in order to provide independent elevation measurements over a
contemporaneous time period in a wide range of locations across Antarctica. For all airborne
8
measurements a filter was applied to remove any erroneous step changes in elevation resulting from
the laser altimeter ranging from cloud cover (Young et al, 2008; Kwok et al, 2012).
Figure 3: Locations of ATM (left) and RLA (right) airborne data used to evaluate the DEM
2.4 DEM evaluation
In the following section the comparison between the DEM and the IceBridge airborne dataset is
presented. In all cases the elevation value of the DEM at the exact location of the airborne laser
altimeter measurement was calculated through bilinear interpolation. The interpolated DEM values
were then subsequently corrected for any change in surface elevation which may have occurred
between the acquisition periods of the DEM and airborne datasets due to any ice dynamics or surface
mass balance processes. This correction was calculated by interpolating the gridded elevation change
trends (dh/dt) obtained from Equation 1 to the location of the airborne measurement, through the
same bilinear interpolation method as used for the DEM elevation estimate. Before being used to
correct for changes in surface elevation, the dh/dt trends were corrected for temporal fluctuations in
backscattered power, which can introduce spurious signals in time series of elevation change (Davis
and Ferguson, 2004; Khvorostovsky, 2012).
All elevation differences are presented as the interpolated DEM elevation subtracted from the
measured airborne elevation. Because multiple LIDAR elevation measurements can be found in a
given DEM grid cell, elevation differences were averaged into the same 2 x 2 km grid that the DEM is
projected on. Evaluation results were separated according to whether the IceBridge elevation
9
measurement resides in a grid cell containing CryoSat-2 measurements, hereby referred to as an
observed grid cell, or an interpolated elevation value. This approach allows the accuracy of CryoSat-2
observations and the chosen interpolation method to be assessed independently. In total
approximately 80% of IceBridge elevation measurements resided in an observed grid cell.
3. Results
The generated DEM is provided in Figure 2. After the removal of elevation values generated from
poorly constrained model fits, the DEM provides an elevation measurement for 91% of the ice-sheet
and 97% of the ice-shelves. The remaining grid cells north of 88o S are interpolated to provide a
continuous grid for 100% of the area outside of the pole hole. In total 3% and 8% of grid cells are
interpolated in the ice shelves and ice sheet, respectively.
3.1. Comparison of DEM to airborne elevation measurements: observed grid cells
The ATM is an airborne scanning laser altimeter capable of measuring surface elevation with an
accuracy of 10 cm or better (Krabill et al 2000). Flown at a typical altitude of 500 m above ground level,
ATM illuminates a swath width of approximately 140 m, with a footprint size of 1-3 m and along track
separation of 2 m (Levinsen et al 2013).
Data acquired by the RLA were collected as part of the NASA ICECAP program from December 2009
to 2013, mounted to a survey aircraft flown at a typical height of 800 m. Elevation measurements,
referenced to the WGS84 ellipsoid, are provided at a spatial resolution of 25 m along track and 1m
across track with an error of approximately 12 cm (Blankenship et al 2012).
Figure 4: Difference between ATM laser altimeter elevation measurements and observed DEM elevation estimates in
Antarctica (left), the Antarctic Peninsula (middle) and the Amunsden Sea Sector (right). Elevation differences are
overlaid on a shaded relief plot of the DEM.
10
Figure 5: Difference between RLA laser altimeter elevation measurements and observed DEM elevation estimates in
Antarctica (left), the Totten catchment area (middle) and Queen Maud Land (right). Elevation differences are overlaid on
a shaded relief plot of the DEM.
The spatial distribution of elevation differences between IceBridge elevation measurements and DEM
estimates are provided in Figures 4 and 5. For each plot in Figures 4 and 5 the elevation differences
are binned at the DEM resolution of 2 x 2 km, but plotted at a larger scale for ease of viewing. The
spatial distribution of the two IceBridge datasets differ. The majority of ATM locations are
concentrated in low elevation areas where topography is more challenging for radar altimetry, with
58% of the compared grid cells having an elevation of less than or equal to 1000 m. Unlike the ATM
dataset, a larger proportion of RLA elevation measurements were acquired in high elevation regions,
with 66% of observations occurring in regions where the elevation exceeds 2000 m.
IceBridge – CS2 DEM (Observed points only)
Region
Number of
grid cells
Region
coverage
(%)
Mean
difference
(m)
Median
difference
(m)
Standard
deviation
(m)
RMS
difference
(m)
Ice-shelves
14508
3.8
1.54
0.43
11.99
12.09
Peninsula
1466
2.6
-8.63
-6.66
20.10
21.87
WAIS
21261
4.9
-4.19
-2.04
14.04
14.65
EAIS
62834
2.5
0.11
-0.31
23.84
23.84
Table 2: Statistics for comparison between IceBridge elevation measurements and observed DEM
grid cells for individual Antarctic regions
11
Statistics for comparisons between IceBridge elevation measurements and observed DEM grid cells,
calculated for the ice-shelves, Antarctic Peninsula, West Antarctic and East Antarctic ice-sheets (WAIS,
EAIS) are given in Table 2.
3.2. Comparison of DEM to airborne elevation measurements: interpolated grid cells
The accuracy of the chosen local mean filling interpolation method is assessed by comparing IceBridge
elevation measurements residing in a DEM grid cell containing no data with the interpolated value.
Statistics for comparisons between IceBridge elevation measurements and interpolated DEM grid
cells, calculated for the ice-shelves, Antarctic Peninsula, West Antarctic and East Antarctic ice-sheets
(WAIS, EAIS) are given in Table 3. As expected, the interpolated values deviate more from IceBridge
elevation measurements in areas of high slope and complex terrain. This is particularly true for the
ice-sheet margins and the Antarctica Peninsula, where there is little spatial correlation over the 25 km
search radius chosen for the interpolation. Over higher elevation regions with relatively smooth
topography, it is more reasonable to assume spatial correlation over distances of 25 km.
ICEBRIDGE ALL – CS2 DEM (interpolated points only)
Region
Number of
grid cells
Region
Mean
coverage difference
(%)
(m)
Median
difference
(m)
Standard
deviation
(m)
RMS
difference
(m)
Peninsula
4658
8.2
-193.79
-143.05
291.25
349.81
WAIS
3904
0.9
-46.73
-28.01
106.44
116.24
EAIS
11065
0.4
-89.38
-20.10
217.17
234.83
Table 3: Statistics for comparison between IceBridge elevation measurements and interpolated
DEM grid cells for individual Antarctic regions
12
4. Product Format

Antarctica_DEM_CryoSat2_z_v1.2.tif: Antarctica DEM in geotiff format

Antarctica_DEM_CryoSat2_nvals_v1.2.tif: number of CryoSat-2 measurements in each grid
cell, a value of 0 denotes an interpolated grid cell in geotiff format

Antarctica_DEM_CryoSat2_chisq_v1.2.tif: goodness of fit statistic in geotiff format

netcdf (Antarctica_DEM_CryoSat2_v1.2.nc) containing z, nvals, chisq arrays and x, y
information

Projection: Polar Stereographic

Ellipsoid: WGS-84

Standard latitude: -71°

Central meridian: 0°

Minimum x value: - 2 820 000 m

Minimum y value: -2 420 000 m

Grid spacing: 2000 m

Coordinate convention: Grid coordinates relate to lower left corner of each grid cell

Number of rows: 2420

Number of columns: 2820

DEM units: metres

Data format: big-endian IEEE 4-byte floating point format, no data value is NaN
13
5. Acknowledgements
This work was supported by the European Space Agency (contract number 4000114834 and
4000112227), and by the UK Natural Environment Research Council (contract number cpom30001).
14
6. References
Armitage, T.W.K. et al. 2014. Meteorological Origin of the Static Crossover Pattern Present in LowResolution-Mode CryoSat-2 Data Over Central Antarctica. IEEE Geoscience and Remote Sensing
Blankenship, D., et al. 2012. Icebridge Riegl laser altimeter L2 geolocated surface elevation triplets,
Digital media, NASA DAAC at the National Snow and Ice Data Center.
Davis, C.H. and Ferguson, A.C. 2004. Elevation change of the Antarctic ice sheet, 1995-2000, from ERS2 satellite radar altimetry. IEEE Transactions on Geoscience and Remote Sensing. 42(11),pp.2437–
2445.Letters. 11(7),pp.1295–1299.
Khvorostovsky, K.S. 2012. Merging and Analysis of Elevation Time Series Over Greenland Ice Sheet
From Satellite Radar Altimetry. IEEE Transactions on Geoscience and Remote Sensing. 50(1),pp.23–36.
Koenig, L., S. Martin, M. Studinger, and J. Sonntag (2010), Polar airborne observations fill gap in
satellite data, Eos Trans. AGU, 91(38), 333–334, doi:10.1029/2010EO380002
Krabill, W., et al. 2004. Greenland ice sheet: Increased coastal thinning, Geophys. Res. Lett., 31,
L24402, doi:10.1029/2004GL021533.
Kwok, R., G. F. Cunningham, S. S. Manizade, and W. B. Krabill (2012), Arctic sea ice freeboard from
IceBridge acquisitions in 2009: Estimates and comparisons with ICESat, J. Geophys. Res., 117, C02018,
doi:10.1029/2011JC007654.
Levinsen, J.F., et al. 2013. Improving maps of ice-sheet surface elevation change using combined laser
altimeter and stereoscopic elevation model data. Journal of Glaciology, Volume 59, Number 215 , 524532(9)
McMillan, M. et al. 2014. Increased ice losses from Antarctica detected by CryoSat-2. Geophysical
Research Letters. 41(11),p.2014GL060111.
Young, D. A., S. D. Kempf, D. D. Blankenship, J. W. Holt, and D. L. Morse (2008), New airborne laser
altimetry over the Thwaites Glacier catchment, West Antarctica, Geochem. Geophys. Geosyst., 9,
Q06006, doi:10.1029/2007GC001935.
15