SCIENTIFIC PREPARATIONS FOR CORE-H2O, A DUAL FREQUENCY SAR MISSION FOR
SNOW AND ICE OBSERVATIONS
Helmut Rott1, Don Cline2, Claude Duguay3, Richard Essery4, Christian Haas5, Michael Kern9,
Giovanni Macelloni6, Eirik Malnes7, Jouni Pulliainen8, Helge Rebhan9, Simon Yueh10
1Institute for Meteorology and Geophysics, University of Innsbruck, Austria. Helmut.Rott@uibk.ac.at
2NOAA-NOHRSC, Chanhassen, MN, USA
3University of Waterloo, Canada
4University of Edinburgh, UK
5Alfred-Wegener-Institut, Bremerhaven, Germany, and University of Alberta, Canada
6IFAC-CNR, Firenze, Italy
7NORUT IT, Tromsö, Norway
8Finnish Meteorological Institute, Helsinki, Finland
9ESA-ESTEC, Noordwijk, Netherlands
10JPL-CalTech, Pasadena, CA, USA
The COld REgions Hydrology High-resolution Observatory
(CoRe-H2O) satellite mission has been selected for scientific
and technical studies within the ESA Earth Explorer
Programme. The mission addresses the need for spatially
detailed snow and ice observations in order to improve the
representation of the cryosphere in climate models and to
improve the knowledge and prediction of water cycle
variability and changes. CoRe-H2O will observe the extent,
water equivalent and melting state of the snow cover,
accumulation and diagenetic facies of glaciers, and
properties of sea ice and lake ice. The sensor is a dual
frequency SAR, operating at 17 GHz and 9.6 GHz, VV and
VH polarizations. This configuration enables the
decomposition of the scattering signal for retrieving
physical properties of snow and ice. Scientific preparation
activities include experimental field campaigns,
improvement of radar backscatter models, and the
development of inversion algorithms.
Index Terms— Snow, ice, SAR, backscatter, inversion
Snow and ice are important elements of the climate system,
being very sensitive to changes in temperature and
precipitation, and interacting with other climate variables
through complex feedback mechanisms. Snow and glacier
melt is a basic resource of water for many densely populated
areas of the world, the abundance of which is jeopardized
by climate change. In the 2007 IPCC report it is emphasized
that substantial uncertainty remains in the magnitude of
cryospheric feedbacks within climate models . Accurate
inventories of the snow and ice masses and their temporal
dynamics are needed to improve the parameterizations and
modeling of water and energy exchange.
There is a lack of spatially distributed information on
the accumulation of snow on land surfaces, glaciers and sea
ice. The Integrated Global Observing Strategy (IGOS)
Partnership recommends in its Cryosphere Theme Report
the development and implementation of satellite systems for
spatially distributed measurements of snow water equivalent
(SWE) and other snow properties . In order to close gaps
in spatially detailed cryosphere observations, the satellite
mission COld REgions
Observatory, CoRe-H2O, has been proposed to ESA in
response to the 2005 Call for Earth Explorer Core Missions
and selected for pre-feasibility studies.
2. MISSION REQUIREMENTS
The CoRe-H2O mission will deliver spatially detailed
observations of snow, ice, and water cycle characteristics in
cold environments. The satellite measurements address
applications in hydrological modeling, climate modeling,
numerical weather prediction, glacier mass balance studies,
process studies for sea ice and lake ice, and several other
topics in snow and ice research.
III - 31 978-1-4244-2808-3/08/$25.00 ©2008 IEEE IGARSS 2008
2.1. Measurement Requirements
The observation requirements have been specified for
primary and secondary geophysical parameters .
Primary observation parameters:
? Extent and water equivalent of the snow cover.
Secondary observation parameters:
? Snow depth
? Extent of melting snow areas
? Snow accumulation and facies type of glaciers
? Sea ice type with emphasis on thin ice
? Snow accumulation and melting properties of sea ice
? Lake ice properties
? Surface water extent in northern regions.
The required spatial resolution for the geophysical
products is in the range of 100 m to 500 m, depending on
the application for regional or global studies.
Regarding the temporal coverage two mission phases
? Orbit Phase 1: 3-day repeat cycle to provide
observations at typical time scales of synoptic
meteorological systems over a subset of the snow and
ice areas. The emphasis is on initialization and
validation of mesoscale
hydrological models, and snow process models.
? Orbit Phase 2: 15-day repeat cycle to provide near-
complete coverage of the global snow and ice cover.
Orbit phase 1 is planned for a duration of 2 years, followed
by orbit phase 2 for about 3 years.
In addition, the mission will offer opportunities for
scientific studies in related fields, in particular during the
season with reduced snow and ice extent. One promising
application of great interest for hydrology is the spatially-
detailed retrieval of rainfall .
2.2. Technical Requirements
Table 1. Mission and instrument requirements
SAR frequency 9.6 GHz and 17.2 GHz
Polarizations VV and VH
X-Band: ? -23 dB for VV
? -28 dB for VH
Ku-Band: ? -20 dB for VV
? -25 dB for VH
? 100 km
? 0.5 dB
? 1.0 dB
Incidence angle Swath within 30° to 45°
? 50 m x 50 m (? 5 looks)
? 10 MHz
Theory and experimental studies point out that a dual
frequency, X-band (9.6 GHz) and Ku-band (17.2 GHz)
SAR operating at co- and cross-polarizations should be well
able to achieve the measurement requirements.
Ongoing technical studies show that a ScanSAR with
parabolic antenna and multi-beam feed arrays is a cost-
efficient solution for fulfilling these requirements. Options
with a single parabolic reflector for both frequencies, as
well as with a separate reflector for each frequency on the
same platform have been studied and are both well feasible.
3. BACKSCATTER MODELLING
Theoretical and experimental studies have been carried out
to investigate the relations between Ku- and X-band
backscatter and physical properties of snow and ice. These
studies are the basis for the development of inversion
algorithms to retrieve snow and ice parameters.
Forward calculations on the backscatter signal of a
snowpack on a background medium at polarization pq take
into account the following contributions:
where ?as accounts for scattering at the air/snow
interface, ?v for the direct contribution from the snow
volume, ?gv for ground surface/volume interaction, and ?ag
for backscatter at the ground attenuated by the snowpack.
For characterizing the scatterers, two modelling
approaches have been studied: (i) the discrete dipole
approximation (DDA) enabling to describe arbitrary shapes
of the scatterers and to specify preferred orientations of the
particles; and (ii) ellipsoidal shapes defined by the ratio of
the two axes, with random orientation in the volume.
The models allow for definition of the roughness of the
snow surface layer and of the background target (soil). For
the snow medium the following variables can be specified:
snow depth, density, temperature, grain shape, grain size
(equivalent diameter referring to the volume of a sphere).
The main parameter to be measured by CoRe-H2O is the
mass of snow on ground, the snow water equivalent, SWE
(snow depth x density). For developing SWE retrieval
algorithms, it is important to study the backscatter
sensitivity to physical properties of the snow pack and the
background target. Because the SAR penetration into wet
snow is smaller than one wavelength, measuring SWE by
means of SAR is only possible as long as a snowpack is dry.
Typical numbers for the one-way penetration depth into dry
snow are 3 to 4 m at Ku-band and about 10 m at X-band.
Radar scattering theory and the model calculations
show that the main parameters determining backscatter of
soil covered by dry snow are SWE, the grain size, and the
roughness and dielectric properties of the soil. Figure 1
shows an example on the sensitivity of Ku- and X-band
backscatter as a function of SWE. The calculations were
III - 32
carried out with a second order dense medium radiative
transfer (DMRT) model, developed by J. Du and J. Shi ,
. For this example a snow density of 250 kg/m3 is
assumed, ellipsoidal grains with axial ratio of 0.25 and grain
equivalent diameter of 1 mm (typical for aged seasonal
snow), and frozen soil as background. The simulations point
out that for SWE < 300 mm Ku-band is more sensitive to
SWE than X-band, whereas the X-band signal shows rather
constant sensitivity over a wider range of snow depths.
Figure 1. Backscatter coefficients as a function of snow
water equivalent (SWE) at Ku-band and X-band, 40°
Further modeling studies have been dealing with effects of
the attenuation in the atmosphere and in vegetation, using
radiative transfer models for radar propagation in these two
4. FIELD EXPERIMENTS
Several field experiments were conducted to acquire
experimental data for the development and validation of
backscatter models and snow retrieval algorithms. In winter
2006/07 Ku- and X-band backscatter measurements were
carried out at two test sites in the Austrian Alps with the
ground-based SAR (GB-SAR) system of the University of
Cranfield, UK, during the SARALPS2007 campaign .
In winter 2007/08, polarimetric backscatter data at L- ,
S-, C-, X-, and Ku-band frequencies were acquired over
several test sites in the Austrian Alps with a helicopter-
borne scatterometer during the HeliSnow-2008 campaign.
Of great relevance for the development of backscatter
theory and inversion methods are also the Cold Land
Processes Experiments (CLPX) performed by NOAA and
NASA/JPL in winter 2006/07 in Colorado and in 2007/08 in
Alaska. Ku-band data were acquired with the Polscat, a
conically scanning polarimetric airborne scatterometer
operating at 13.5 GHz. Co-located co- and cross-polarized
X-band SAR data were acquired by TerraSAR-X.
Figure 2 shows an example of co- and cross-polarized
backscatter at 17 GHz, measured on 5 February 2007 during
the SARALPS2007 campaign at the Alpine test site Kühtai
(1750 m a.s.l.). The background medium is mountain
pasture with a bumpy surface. Therefore the angular
dependence of the measured ?° is not smooth. The surface
was covered by 35 to 40 cm of dry snow, with an average
SWE of 100 mm. The backscatter simulations were carried
out with a dual layer DMRT. The bottom layer of the snow
pack was made up by large, metamorphic grains, whereas
the upper layer contained smaller rounded grains. The
comparison shows good agreement between measured and
simulated ?° at co-polarizations, whereas the cross-
polarized ?° is underestimated by a few dB. Comparisons
with the experimental CLPX and HeliSnow data sets show
similar mismatch for cross-polarized ?°, probably caused by
underestimation of multiple scattering effects by the model.
Figure 2. 17 GHz co- and cross-polarized backscatter
coefficient of snow-covered ground measured by GB-SAR
on 5 February 2007 at the test site Kühtai (thick lines).
Backscatter simulated by DMRT: thin lines.
The airborne and in situ Ku- and X-band data
acquisitions were supported by extensive field observations
on snowpack properties. The analysis of these data sets is
going on, providing an important basis in preparation for the
5. INVERSION OF BACKSCATTER
Various options for retrieving physical snow properties
from the dual-frequency, dual-polarized backscatter data
have been studied. The focus is on the retrieval of SWE
because this is a main geophysical product to be delivered
by the mission. In order to obtain SWE, it is necessary to
separate the signal of the snow volume from that of the
background target and to account for effects of grain size on
The direct inversion of a radiative transfer forward
model is an interesting option to derive SWE from the 4
III - 33
channel measurements . The surface and volume Download full-text
scattering contributions can be separated using the
depolarization factor at both frequencies. The scattering-
efficient grain size and scattering albedo of the snow
volume can be estimated from the optical thickness.
However, though the direct inversion of the theoretical
model is well defined in terms of physics, it suffers from
rather high sensitivity to noise, both in the measurements as
well as in the model parameterizations.
For this reason statistically based retrieval methods,
which are less sensitive to noise, are widely applied for
inversion of remote sensing data. We propose as a baseline
version for SWE retrieval a statistical approach, applied to
inversion of a physical forward model. The cost function to
be minimized follows the approach of Lehtinen :
where ?i represents the forward model, i refers to the
measurement channel (n = 4 for CoRe-H2O), x1 …, xq are
the free model parameters (the state variables: SWE, ?,
….), c1i, …, cri are the configuration parameters (frequency,
polarization, etc.), Zi are the measured backscatter
coefficients, and ?i is the measurement noise. The second
sum takes into account a priori statistical information of
Gaussian distributed parameters with a standard deviation ?j
and the mean value x’j.
For the forward calculation we use a first order
radiative transfer model. The number of free parameters can
be reduced by taking
interdependences between backscatter in the various
channels, such as frequency dependence of scattering and of
absorption. The background signal can be taken from
backscatter time series before the snowfall period. The main
parameters for optimization are SWE and the volume
scattering albedo, ?. Statistical a priori information on SWE
can be obtained from climatology or from meteorological
analysis. Figure 3 shows the cost function for the case of
snow backscatter measured on 5 February 2007 by GB-SAR
into account physical
Figure 3. Example for the cost function to retrieve SWE
and ? from Ku- and X-band VV- and VH-polarized ?°.
The dual frequency Ku- and X-band SAR sensor, proposed
for the CoRe-H2O mission, will enable to observe snow and
ice physical properties that cannot be detected at lower radar
frequencies. Experimental field campaigns, improvements
of radar backscatter models, and the development of
inversion algorithms for retrieval of snow parameters have
been conducted in preparation for the mission. The studies
to date confirm that the proposed mission concept should
provide an excellent basis for accurate repeat observations
of snow characteristics with great spatial detail. Such
information is of great importance for better representation
of snow and ice in land surface processes and hydrology
models, and for improving the parameterization of
surface/atmosphere interactions in numerical weather and
The investigations were supported by the European Space
Agency, ESTEC Contract No. 20756/07/NL/CB. We wish
to thank Jiancheng Shi for making available his DMRT for
backscatter modelling of snow.
 D.A. Randall et al., Climate Models and Their Evaluation. In:
Climate Change 2007: The Physical Science Basis. Contribution of
Working Group I to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change. Cambridge Univ.
Press, Cambridge, UK and New York, NY, USA, 2007.
 J. Key et al., IGOS Cryosphere Theme Report. WMO/TD-No.
1405. UNESCO, WMO, ICSU, 2007.
 H. Rebhan, CoRe-H2O Mission Requirements Document.
Noordwijk, NL, 2008.
Rev. 1.5. ESA-ESTEC,
 F.S. Marzano, S. Mori and J.A. Weinmann, “High resolution
rainfall retrieval over land from satellite synthetic aperture radar
measurements at X, Ku and Ka band”. Proc. 5th European Conf.
on Radar in Meteorology and Hydrology, Helsinki, 2008.
 J. Du, J. Shi, S. Tjuatja, and K.S. Chen, “A combined method
to model microwave scattering from a forest medium”, IEEE
Trans. Geosci. Rem. Sens., 44 (4), pp. 815-824, 2006.
 J. Shi, “Water equivalence retrieval using X and Ku band dual-
polarization radar”, Proc. IGARSS, 2006.
 K. Morrison, H. Rott, T. Nagler et al., “The SARALPS-2007
measurement campaign on X- and Ku-band backscatter of snow”,
Proc. IGARSS, 2007.
 M.S. Lehtinen, „On statistical inversion theory.” In Theory and
Applications of Inverse Problems (H. Haario, Ed.), pp. 46-57.
Pitman Res. Notes in Mathematics Series 167, 1988.
III - 34