H. Rott –CoReH2O IGARSS 2011
CoReH2O – A Dual Frequency Radar Satellite for Cold Regions Hydrology
H. Rott1, D. Cline2, C. Duguay3, R. Essery4, P. Etchevers5, I. Hajnsek6, M. Kern7, G. Macelloni8, E. Malnes9, J. Pulliainen10, S. Yueh11
1 University of Innsbruck & ENVEO IT, Austria
2 NOAA, NWS, Hydrology Laboratory, USA
3 University of Waterloo, Canada
4 University of Edinburgh, UK
5 Meteo-France, Saint Martin d’Héres, France
6 DLR-HR, Germany & ETH Zürich, Switzerland
7 ESA-ESTEC, Noordwijk, NL
8 IFAC-CNR, Firenze, Italy
9 NORUT IT, Tromsǿ, Norway
10 Finish Meteorological Institute, Helsinki, Finland
11 JPL-Caltech, Pasadena, USA
H. Rott –CoReH2O IGARSS 2011
Outline of the Presentation
• Summary of mission objectives
• Observation requirements
• Retrieval concept for snow mass
• Inversion of RT model
• Examples for performance analysis
- with simulated data
- with experimental data
• Conclusions
H. Rott –CoReH2O IGARSS 2011
Objectives: Improved Snow and Ice Observations
For climate research
• Snow and ice – two essential climate elements not well represented in climate models
• In particular, snow mass is poorly known
Hydrology and surface/atmosphere exchange processes
• High-resolution data are needed to account for spatial variability of snow
Glacier mass balance – climate interactions
• An essential climate variable measured only for few glaciers
• Global data are needed to quantify response to climate forcing
Snowmelt and glacier runoff - a crucial water resource
• Snow cover and glacier retreat caused by climate change may affect the
water supply to hundreds of millions of people.
• New models using spatially detailed snow observations are needed to
improve water management and support adaptation to changes.
H. Rott –CoReH2O IGARSS 2011
Observation Requirements
Primary parameters Spatial scale
Regional/Global
Sampling
(days)
Accuracy
(rms)
Snow water equivalent 200 m / 500 m 3-15 3 cm for SWE 30 cm,
10% for SWE > 30 cm
Snow extent 100 m / 500 m 3-15 5% of area
Glacier snow
accumulation 200 m / 500 m 15 10% of winter maximum
Secondary parameters
Diagenetic
facies types,
glacial lakes
Glaciers
Ice area; freeze
up and melt
onset
Lake and river ice
Melting snow
area, snow
depth
Snow
Snow on ice (SWE,
melt onset and area);
type and thickness of
thin ice
Sea ice
H. Rott –CoReH2O IGARSS 2011
Parameter Ku-band SAR X-band SAR
Frequency 17.2 GHz 9.6 GHz
Polarization VV, VH
Swath width, Inc angle ≥ 100 km; 30° to 45° range
Spatial resolution ≤ 50 m x 50 m (≥ 4 ENL)
NESZ ≤ -25dB VH ≤ -27dB VH
Radiom. Stability / Bias ≤ 0.5 dB / ≤ 1.0 dB
Antenna concept Single reflector with multiple beam feed array
Peak RF power 1.2 kW; 1.8 kW (2 concepts) 1.8 kW; 3.5 kW
Nr. of ScanSAR beams 6 6
CoReH2O – Instrument Design Parameters
H. Rott –CoReH2O IGARSS 2011
SWE Retrieval Algorithm - Iteration
A semi-empirical radiative transfer model
is used for forward computations to
enable efficient iteration for 2 free
parameters: SWE, re
H. Rott –CoReH2O IGARSS 2011
Semi-empirical RT-Formulation for Snow over Soil
Semi-empirical RT Model (sRT) – Single Layer
Basic Equation:
Ground
Snow
Air
d ts s,
q
q'
trP
P
s
s
v
sas
Scattering
g
tt
g
pqtpqt
v
pqi
as
pqi
t
pq t qqsqqsqsqs 2
t
et
g
pq
t
etpqtpqi
as
pqi
t
pq
SWEkSWEk
qqs
qqqqsqs
cos
'2exp
cos
'2exp1cos75.0 2
tetset SWEkdkL qqq sec'expsecexp
One-Way Loss Factor:
s
esae
kkkk
'´'' Extinction coefficient for unit mass
Formulation for forward computation:
T(q).. Power transmission coefficient; … Scattering albedo
H. Rott –CoReH2O IGARSS 2011
Frequency dependence of scattering is parameterized based on experimental data and numerical simulations for closely packed snow grains:
Wavelength exponent A = 3.2 is used as default value for seasonal snow, based on experimental data and numerical simulations (e.g. Tse et al., 2007). Further work needed to establish relations to snow morphology/snow type.
sRT – Parameterization of Snow Volume Backscatter
Initial value of Scattering coefficient:
The sRT scattering coefficient, ks , at f1 (17.2 GHz VV) is related to “effective grain size” re
which is used as input parameter for specifying the scattering efficiency in this channel.
In order to provide a link to common formulations, the initial value of ks is computed with the Rayleigh approach for frequency f1 =17.2 GHz as f(re).
In the iteration ks is a free parameter to match forward computations and measurements.
2
12
2
211
12 2
1.....,,;,....,
2
1
q
j
jj
j
iriiiqi
n
i i
xxZcccxxxJs
Cost function
For iteration
Forward model a-priori SWE, re
H. Rott –CoReH2O IGARSS 2011
Input Parameters for sRT Forward Model
Symbol Name Source / Role in retrieval and forward model
Snow pack (single layer)
SWE Snow water equivalent Free variable
re Effective grain radius Free variable , related to ks at f1 = 17 GHz
Ts Mean snow pack temperature Configuration Parameter: from auxiliary data / for
computing ka (”)
s Mean snow pack density Configuration Parameter: auxiliary data or default value/
for computing T(pq) and q(t)
rmsas Std. deviation of surface height at
air/snow interface
Configuration Parameter: Pre-scribed / for computing
sas (small contribution to total backscatter)
sg (f, pq) Backscatter coefficient at ground
surface
From pre-snowfall backscatter measurements in 4
channels
RT model parameters (empirical)
As Coefficient for frequency
dependence of ks
Relation based on experimental data for linking ks(f2) to
ks(f1). Presently used default value As=3.2
Ap Cross- to co-polarized ratio for ks
(depolarization factor)
Relation based on experimental data for deriving ks (pq)
from ks(pp); presently linked to grain size
H. Rott –CoReH2O IGARSS 2011
Performance Analysis for SWE Retrieval - Simulations
-10
-9
-8
-7
-6
-5
-4
-3
-2
FP01 FP02 FP03 FP04 FP05 FP06 FP07 FP08 FP09
SIG
MA
_0
[d
B]
Basic Test ID
SIMULATED RADAR BACKSCATTER - X_vv
xvv_snow_mean
xvv_ref_mean
-10
-9
-8
-7
-6
-5
-4
-3
-2
FP01 FP02 FP03 FP04 FP05 FP06 FP07 FP08 FP09
SIG
MA
_0 [d
B]
Basic Test ID
SIMULATED RADAR BACKSCATTER - Ku_vv kuvv_snow_mean
kuvv_ref_mean
Input for simulation
FP-ID SWE [m] re [mm]
FP01 0.1 0.3
FP02 0.1 0.5
FP03 0.1 0.7
FP04 0.3 0.3
FP05 0.3 0.5
FP06 0.3 0.7
FP07 0.5 0.3
FP08 0.5 0.5
FP09 0.5 0.7
Example for test case using
Synthetic Scene Generator
H. Rott –CoReH2O IGARSS 2011
Performance Analysis – Effect of Snow Density
Retrieval statistics for different snow cover states using Synthetic Scene Generator
H. Rott –CoReH2O IGARSS 2011
Performance Analysis with NOSREX Data
SnowScat s°
17 GHz, 10 GHz
SWE time series
GWI
Field campaign Sodankylä 2010-11
H. Rott –CoReH2O IGARSS 2011
Retrieval Tests – Effect of Background s°
Snow Density Snow Temperature
RV – Grain radius (mean, stdev)
Cost-function (0 without RV-SWE)
Reference Backscatter
200 kg/m³ -5 0.5, 0.4 mm 0 December
200 kg/m³ -5 0.5, 0.4 mm 0 October
Retrieval input data
H. Rott –CoReH2O IGARSS 2011
Conclusion
• The CoRe-H2O mission addresses a particular gap in present cryosphere
monitoring: spatially detailed observations of snow mass (SWE).
• A dual frequency, dual polarized Ku- and X-band SAR sensor is proposed as
tool for SWE measurements.
• The baseline retrieval method for SWE is based on iterative inversion of a
semi-empirical RT model, applying a statistical concept.
• Experimental data are essential for calibrating and testing the forward model
and inversion algorithm.
• Important contributions to the experimental data base are supplied by the
NOSREX Campaign (17& 10 GHz in situ), CAN-SCI (17 & 10 GHz in situ),
CLPX PolScat (14 GHz), TerraSAR-X (9.6 GHz).
• Activities for scientific mission preparation are dealing with assimilation of
CoRe-H2O products in snow process models, including the extraction of
auxiliary data for input to the SWE retrieval, and further field campaigns (with
the new 17 & 10 GHz airborne SnowSAR of ESA and in situ sensors).