CIWSIR: A Proposed ESA Mission to
Measure Cloud Ice Water Path
S. A. Buehler (1), C. Emde (2), P. Eriksson (3), F. Evans (4), C. Jimenez (1), V. O. John (5), T. R. Sreerekha (6)
(1) Lulea Technical University, Sweden; (2) German Aerospace Center (DLR), Germany; (3) Chalmers University,
Gothenburg, Sweden; (4) University of Colorado, Boulder, USA; (5) University of Miami, USA; (6) Met Office, UK
0.3
0.2
0.1
0
−90
−30
−60
0
30
60
90
30
60
90
Latitude
CCMA
GFDL CM2.0
GFDL CM2.1
GISS ER
GISS EH
INMCM3
IPSL
MIROC
MPI ECHAM5
NCAR CCSM3
UKMO HADCM3
δ Ice Water Path (%) @ 2xCO2
40
20
0
−20
−90
−30
−60
0
Latitude
3
CIWSIR calculation for gamma distribution, IMC 0.001 g/m
176.31 GHz
240.7 GHz
315.65 GHz
0.04
440.8 GHz
659.8 GHz
868 GHz
0.03
10 µm
94 GHz Radar
7
6
∆ BT [K]
5
4
0.02
3
2
0.01
1
0
0
200
400
600
800
1000
1200
0
1400
D [µm]
−3
mass of particles [g/(m 3 µm)]
6
x 10
Ivanova 2001
Mitchell 1999
Donovan 2003
5
4
3
Backscattering efficieny [−]
0.05
8
The sensitivity of sub-mm radiances at a
given frequency to cloud ice depends
strongly on the particle size. The top figure
to the left illustrates this by showing for
different measurement frequencies the signal
of a fixed amount of ice, that is put in a
narrow size distribution at the frequency
denoted by the x-axis. At 868 GHz the
instrument is most sensitive to particles of
approximately 200 micron diameter, at 240
GHz to particles of approximately 700 micron
diameter. IR measurements at 10 micron
wavelength see only small particles below
100 micron diameter. Radar measurements
are most sensitive to the largest particles of
1000 micron or more, because of the
characteristic wavelength dependence of the
backscatter coefficient.
2
1
0
200
400
600
800
D [µm]
1000
1200
1400
The middle plot shows some size distribution
parameterizations from the literature. Submm channels can cover a large part of the
distributions, but the small mode below 100
micron will only be seen by IR channels.
Brightness Temperature [ K ]
300
clear
ice cloud
280
The bottom plot shows the sub-mm channel
setup that was proposed in the CIWSIR
Earth Explorer proposal. It consists of 6
radiometers, with labels as indicated in the
plot. Radiometers R1, R3, and R4 have 3
channels each, the others only 1 channel
each, resulting in a total of 12 channels.
260
240
220
200
180
160
0
R1 R2 R3 R4
200
R5
400
600
Frequency [ GHz ]
R6
800
1000
The highest frequency of
the Advanced Microwave
Sounding Unit B (AMSU-B)
is 183+7 GHz (Channel 20).
This is too low to detect
weak ice clouds, but strong
ones can be easily seen.
The left-most plot below
shows AMSU-B Channel 20
data taken on 2002-01-25
over the UK, a day with an
intense winter storm. The
second plot shows the IWP
field from the Met Office
mesoscale model, which
shows up to more than 3
kg/m2 of ice. As the third
plot shows, we can model
the AMSU data very well
with a scattering radiative
transfer model. (The model
used was the Atmospheric
Radiative Transfer
Simulator, ARTS.) The
cloud signal is in this case
up to approximately 30 K.
At 664 GHz (right plot) the
signal is much stronger,
even for small IWP.
An ESA study optimizing the CIWSIR concept and
channel selection is currently ongoing. The study
includes also an industry component to derive
possible hardware configurations. The figures on
the right show retrieval simulation results from this
study. These are neural network retrievals, based
on an elaborate dataset of random atmospheric
states, including random variations in cloud
microphysics. Shown is the performance for the
retrieval of IWP as a function of true IWP (top)
and true D (bottom). The errors are median
errors, rather than RMS errors, because RMS
errors are dominated by rare outlier cases. The
performance is shown for different radiometer
subsets. The bottom plot demonstrates, that both
very high and very low frequency radiometers are
necessary, in order to get good accuracy for very
small and very large particles, respectively.
The assumed noise level in this simulation varied
from channel to channel, from 0.5 K for some low
frequency channels to 1.7 K for the highest
frequency channel. For the full configuration, the
IWP accuracy is mostly below 20%, the D
(median particle size) accuracy is mostly below
10%, and the zmed (median IWP cloud altitude)
accuracy is around 300 m. The detection
threshold (100% relative error) for IWP is between
3 and 5 g/m2, depending on instrument
configuration. For higher noise levels, the
performance degrades gracefully, for example, if
the noise level is doubled, the detection threshold
changes to somewhere between 4.5 and 7 g/m2.
65
11
R123456
R 34
R 234
R1 345
R 2345
R12345
60
55
50
45
40
35
30
25
20
10
9
8
7
6
5
4
3
15
10
0
Median of abs(error) in retrieved IWP−value (g/m2)
0.4
The Cloud Ice Water Submillimeter
Imaging Radiometer (CIWSIR) is an
instrument concept that has been
proposed to the European Space
Agency (ESA) in the last call for
Earth Explorer missions. It exploits
the fact that ice clouds are semitransparent to sub-millimeter
radiation (top plot in figure on the
right). The ice particles in the cloud
scatter away a part of the upwelling
thermal radiation from lower layers of
the atmosphere, thus reducing the
brightness temperature (BT)
measured by a receiver on a satellite.
For not too thick ice clouds the BT
depression is linear in IWP (bottom
plot). The sensitivity depends
strongly on particle size and
frequency, since radiation with a
wavelength close to the particle size
is scattered most efficiently.
Median of abs(relative error) in retrieved IWP−value(%)
2
Ice Water Path (kg/m )
0.5
The uncertainties associated with the
feedback of clouds, particularly ice
clouds, in the climate system, are
responsible for a significant part of the
uncertainty in climate predictions. Even
the present climatology of ice in the
atmosphere is very uncertain. The top
plot in the figure to the left shows that
zonal mean ice water path (IWP)
climatologies vary by one order of
magnitude between the different models
in the IPCC AR 4 archive. For a doubleCO2 scenario, the models do not even
agree on the sign of the zonal mean IWP
change (bottom plot). Another important
ice cloud parameter, which is even more
uncertain, is ice particle size. The size
determines the fall speed of the ice
particles and thus has a large impact on
cloud ice fields in circulation models.
100
200
300
400
500
600
True IWP−value (g/m2)
700
800
900
1000
2
0
2
4
6
8
True IWP−value (g/m2)
10
65
Median of abs(relative error) in retrieved IWP−value(%)
CCMA
GFDL CM2.0
GFDL CM2.1
GISS ER
GISS EH
INMCM3
IPSL
MIROC
MPI ECHAM5
NCAR CCSM3
UKMO HADCM3
R123456
R 34
R 234
R1 345
R 2345
R12345
60
55
50
45
40
35
30
25
20
15
10
100
150
200
250
300
350
400
True IWP−dme (um)
450
500
550
600
Acknowledgements
The top left figure was made by B. Soden, University of Miami.
The schematic drawing on the top right was made by Oliver
Lemke, Lulea Technical University. The poster was put together
on behalf of the CIWSIR Science Community. We gratefully
acknowledge the support of the community for the preparation of
the CIWSIR mission proposal, from which much of the material
here is drawn. We also gratefully acknowledge the help from the
ARTS radiative transfer model community.