Content uploaded by Jothiram Vivekanandan
Author content
All content in this area was uploaded by Jothiram Vivekanandan on Apr 29, 2014
Content may be subject to copyright.
P8B.5 A NEW HIGH-ALTITUDE AIRBORNE MILLIMETER-WAVE RADAR FOR
ATMOSPHERIC RESEARCH
Gordon Farquharson∗
, Eric Loew, Jothiram Vivekanandan, and Wen-Chau Lee
National Center for Atmospheric Research, Boulder, Colorado
1. INTRODUCTION
Clouds play an important role in the radiation budget for the
earth-atmosphere system. To improve climate models, obser-
vations of cloud formation and associated dynamics are re-
quired. The recent launch of CloudSat (Stephens et al., 2002)
will provide a wealth of information about clouds on a global
scale. However, space-borne instruments are unable to pro-
vide simultaneous in situ measurements about the observed
clouds, and thus, measurements from a combination of in situ
sensors and airborne cloud radars are still necessary to study
cloud processes.
Airborne millimeter wave radars have been used for at-
mospheric remote sensing since the early 1990s (Pazmany
et al., 1994; Horie et al., 2000; Wolde and Pazmany, 2005),
but most of the aircraft platforms on which these are borne lack
a comprehensive a suite of in situ measurement capabilities.
The National Center for Atmospheric Research has recently
started operating a modified Gulfstream V aircraft, and has in-
stalled an array of sensors for atmospheric research (Laursen
et al., 2006). The aircraft is called the High-Performance
Instrumented Airborne Platform for Environmental Research
(HIAPER) and is funded by the National Science Foundation
(NSF). As part of the instrument suite for HIAPER, NSF is
funding the development of the HIAPER Cloud Radar (HCR).
The HCR will serve the atmospheric science research com-
munity by adding millimeter-wave remote sensing capabilities
to the HIAPER aircraft. The HCR measurements in conjunc-
tion with other HIAPER instrumentation will provide the most
complete picture of cloud physics available to atmospheric sci-
entists. The HCR is scheduled for completion in 2009.
2. SYSTEM DESCRIPTION
The HIAPER Cloud Radar will be implemented in a phased
approach. The first phase (Phase A) will consist of a single-
polarimetric W-band (94 GHz) Doppler radar. Pulse compres-
sion and polarimetric capability will be added in Phase B, and
Phase C will add a second wavelength (Ka-band) radar. In
order to accommodate these extensions to the system, the
Phase A system will be designed to accommodate the Phase
B and C requirements.
The radar will be mounted in a 20 inch diameter (0.5 meter)
wing pod. A conceptual drawing of the pod showing the major
components is shown in Figure 1. A lens antenna is used to
illuminate a rotatable reflector plate which allows the beam to
be steered over almost 270 degrees from zenith through nadir
in the plane normal to the fuselage of the aircraft. The radar
front-end electronics (polarimetric switching network, LNAs,
and calibration network) will be located directly behind the an-
∗Corresponding author address: Gordon Farquharson,
NCAR/EOL, 1850 Table Mesa Dr., Boulder, CO 80305; e-mail:
gordonf@ucar.edu
tenna. The transmitter and associated high power electronics,
the intermediate frequency and receiver electronics, and the
data system will be contained in a pressure vessel. The rest
of the radar hardware (data storage, a real-time display, and a
user interface) will be located in a 19 inch aircraft rack.
A block digram of the transceiver is shown in Figure 2.
A digital waveform generator is used to control the spectral
shape of the transmitted signal by applying an arbitrary am-
plitude taper or phase (frequency) coding. The output of the
waveform generator is centered at 156.25 MHz, and is up-
converted to 94.03125 GHz and amplified by an extended in-
teraction klystron amplifier (EIKA). The directional coupler af-
ter the EIKA couples a small portion of the transmitted signal
into the receiver through a well-calibrated attenuation network
to measure changes in the absolute calibration. For this cali-
bration scheme, knowledge of only the loss through the cal-
ibration path and the receive path before the LNA must be
well characterized. This calibration scheme also allows for
an adaptive pulse compression scheme to be implemented as
a copy of the transmitted pulse is recorded. Measurements
using a noise source or signal generators in the laboratory,
or measurements of a corner reflector in the field will be re-
quired to determine the absolute calibration of the system.
The receiver down-converts the received signal to a center fre-
quency of 156.25 MHz. This intermediate frequency signal is
sampled at a rate of 125 MS/s, which allows for a narrow- or
a wide-band digital down-converter to be implemented in the
digital receiver. The in- and quadrature-phase sampled will be
sent by a gigabit Ethernet connection to the cabin for process-
ing, display, and storage. All oscillators in the transceiver are
phase locked to a 125 MHz stable oscillator. The 125 MHz sig-
nal is also used to generate the timing for the digital waveform
generator and the data acquisition system.
A summary of the system parameters for the Phase A radar
are listed in Table 1. The system will operate at 94.03125 GHz,
with a peak radiated power of 900 W. The PRF and pulse
length will be variable, but typical values of 10 kHz and 0.25
µ
s
respectively are used in the simulations presented. For these
parameters, the range resolution for the pulse length is 37.5
meters and the Nyquist velocity is ±8 m s−1. The beam width
and gain of the antenna are 0.56 degrees and 48 dB respec-
tively, and the receiver noise figure is 10 dB.
3. SYSTEM PERFORMANCE
The design of the HCR includes expansion to a fully-
polarimetric dual-wavelength radar. The design analysis there-
fore covers not only the sensitivity and accuracy of the Phase
A system, but also an analysis of the performance of polari-
metric and dual-wavelength measurements.
Lens
antenna
Pressure vessel:
transmitter, receiver,
data system
Rx
end
Rotatable
reflector front
Figure 1: Side view of the 20” HIAPER wing pod showing the layout of the radar electronics. The front of the pod is on the left
hand side of the figure. The reflector plate is positioned such that the beam clears the leading edge of the wing when pointing
vertically.
SSB
92.625 GHz1250 MHz
Stable Oscillator
SSB
waveform
From
generator
System
Clock
(125 MHz)
156.25 MHz
IF @
PLO PLO
125 MHz
156.25 MHz
20 MHz
1406.25 MHz
20 MHz
94.03125 GHz
2 GHz
94.03125 GHz
2 GHz
1406.25 MHz
50 MHz
156.25 MHz
20 MHz
Calibration Network
Figure 2: Block diagram of the radar transceiver. All oscillators are phase locked to the 125 MHz oscillator.
Table 1: System Parameters
Frequency 94 GHz
Peak radiated power 900 W
PRF 10 kHz
Pulse width 0.25
µ
s
Antenna gain 48 dB
Receiver noise figure 10 dB
3.1 Sensitivity
The minimum detectable reflectivity for Rayleigh scattering as
a function of range from the radar is calculated from
Z=1018 210 log(2)
λ
2lr
π
3P
tG2
ac
τθ
bK2
w
P
nR2p2/MI(1)
where
λ
is the wavelength, lris the loss due to the finite band-
width of the receiver, P
tis the transmitted (radiated) power,
Gais the antenna gain, cis the speed of light,
τ
is the pulse
length,
θ
bis the antenna beam width, K2
wis the related to the
refractive index of water, Pnis the system noise power, Ris the
range from the radar, and MIis the number of independent
samples.
The minimum detectable reflectivity at 0 dB SNR is calcu-
lated by using the a mid-latitude summer profile of pressure,
temperature and water vapor density (top three panels of Fig-
ure 3). The atmospheric attenuation due to oxygen and water
vapor is then calculated using the Millimeter-wave Propaga-
tion Model (MPM93, Liebe et al. (1993)) as a function of height
from the profile data (lower left-hand panel in Figure 3). The
attenuation profile is integrated to determine the cumulative at-
tenuation as a function of range for a radar at 12 km altitude
(lower middle panel in Figure 3), and the minimum detectable
reflectivity (lower right-hand panel in Figure 3) is calculated us-
ing Equation 1 for each range assuming a 120 ms dwell time
and a background brightness temperature of 300 K. A 120 ms
dwell corresponds to 28 meters along the flight track at the
long-range cruising speed (236 m s−1) of the aircraft.
Minimum detectable reflectivity for a vertical profile are tab-
ulated for specific ranges in the second row of Table 2. The
third row shows the minimum detectable reflectivity for a hor-
izontal profile at an altitude of 10 km, assuming a 62 K back-
ground temperature. A significant difference in the sensitivity
of the radar only occurs within the boundary layer where the
concentration of water vapor increases the attenuation of the
electromagnetic wave.
Figure 3: Simulated minimum detectable reflectivity for a vertical profile. The radar is at an altitude of 12 km.
Table 2: Minimum detectable reflectivity for a 0 dB SNR and
integration over a 120 ms dwell time
Range (km) 1 2 5 10 12
MDZ (vertical
profile from 12 km) −37 −31 −23 −17 −12
MDZ (horizontal
profile at 10 km) −37 −31 −23 −17 −16
3.2 Reflectivity Accuracy
The standard deviation in reflectivity ∆ZdB for a Gaussian
Doppler spectrum is approximately given by (Doviak and Zrni´
c,
1993; Hogan et al., 2005)
∆ZdB =10log10(e)
√MN
λ
4
π
(1/2)
σ
w
τ
s
+1
SNR2+2
SNR (1/2)
(2)
where Mis the number of pulse repetition intervals in the dwell
time, Nis number of ranges gates averaged, SNR is the linear
signal to noise ratio,
τ
sis the pulse repetition time,
σ
wis the
spectral width of the scatterers. For a dwell time of 120 ms, a
PRF of 10 kHz, and averaging over one range gate, the stan-
dard deviation in reflectivity varies from 0.44 to 0.19 dB for
signal to noise ratios of 0 to 30 dB and spectral widths of 0.25
to 2 m s−1(Table 3).
3.3 Polarimetric Measurements
A reflector plate will be used to steer the antenna beam.
The reflector plate changes the polarization of the transmit-
Table 3: Standard deviation in reflectivity measurement (dB)
Spectral Width (m s−1)
SNR (dB) 0.25 0.5 1 1.5 2
0 0.44 0.40 0.34 0.30 0.29
10 0.39 0.34 0.26 0.22 0.19
20 0.39 0.33 0.26 0.21 0.19
30 0.39 0.33 0.26 0.21 0.19
ted waveform as it rotates, which therefore changes the basis
in which polarimetric variables are measured. Consider an in-
cident electric field polarized in the ˆxdirection and propagating
in the +ˆzdirection (~
Ei=ˆxEi
0e−jk0z). If the reflector plate is ro-
tated counterclockwise around the z-axis by an angle
ψ
, then
the electric field reflected by the plate is
~
Er=Ei
0n−ˆxsin2
ψ
−ˆycos
ψ
sin
ψ
−ˆzcos
ψ
o
×e−jk0(xcos
ψ
−ysin
ψ
).(3)
The relationship between these vectors is illustrated in Fig-
ure 4 where
ξ
=cos−1cos
ψ
sin2
ψ
+cos2
ψ
. Because of
this change in polarization, measurements made while scan-
ning will necessitate a rotation of the polarimetric variables to
a common polarimetric basis. However, the transformation to
a common basis induces an error in the measured polarimetric
variables due to the limited cross-polarimetric isolation of the
radar. This error is studied through the use of a simple model
which is presented next.
~
Ei
x
~
Er
ψ
z
y
ξ
ˆu
Figure 4: Incident and reflected electric field vectors in the xyz
coordinate system. The incident electric field vector is ver-
tical with respect to the aircraft frame of reference, and the
reflected electric field vector rotates in azimuth and elevation
for a rotation of the reflector plate. The reflector plate has a
counter-clockwise rotation
ψ
from the point of view of the an-
tenna.
The cross-polarimetric isolation Iin decibels for the antenna
and orthomode transducer (OMT) is modeled as a single num-
ber
ε
where
ε
=10−I/20 .(4)
The HCR will use alternating polarimetric transmitted wave-
forms to measure the full polarimetric matrix. If transmitted
signal amplitude is M, the radiated signal amplitudes for trans-
mit H-polarization and transmit V-polarization (ˆ
hˆvbasis) are
~
Mh=1−
ε
ε
and ~
Mv=
ε
1−
ε
(5)
respectively. The amplitude of the received signals at the an-
tenna are
~
V10 =S~
Mhand ~
V01 =S~
Mv(6)
where Sis the scattering matrix describing the atmospheric
state, and ~
V10 and ~
V01 are the received signals at the antenna
for transmit H-polarization and transmit V-polarization respec-
tively. The amplitudes of these signals at the input to the H-
and V-channel receivers (after the antenna and OMT which
have limited cross-polarimetric isolation) are
~
V10
r=V10
co
V10
cr =1−
ε ε
ε
1−
ε
~
V10 and
~
V01
r=V01
cr
V01
co =1−
ε ε
ε
1−
ε
~
V01 ,
(7)
and therefore the measured S-parameter matrix is
Sm=V10
co V01
cr
V10
cr V01
co .(8)
The S-parameter matrix and the measured signals are
transformed to the common orthogonal polarimetric basis
(ˆ
h0ˆv0) through a rotation matrix Rwhere
R=cos
χ
sin
χ
−sin
χ
cos
χ
(9)
where
χ
=ˆ
h·ˆ
h0is the angle between the polarimetric bases
measured from the haxis. The S-parameter matrix describing
the atmospheric state in the ˆ
h0ˆv0basis is
S0=RSR−1,(10)
and the measured S-parameters are
S0
m=RSmR−1.(11)
An example scattering matrix (Equation 12) describing the
atmosphere is assumed to obtain numerical results from this
model.
S=0.9 0
0 0.8(12)
The effect of limited cross-polarimetric isolation when trans-
forming to a common basis is shown by plotting the actual
and measured differential reflectivity (Zdr ) and linear polariza-
tion ratio (LDR) versus the the angle of rotation
χ
between
the basis sets and for a cross-polarimetric isolation of 30 dB
(Figure 5). The difference between the actual and measured
Zdr values increases from 0 up to around 1.2 dB as the rota-
tion angle increases from 0 to 45 degrees and then returns to
0 dB at 90 degrees because the transmitted basis is aligned
with the common basis again at 90 degrees. The actual LDR
at 0 degrees rotation is −∞whereas the measured LDR is
-24.2 dB due to the finite cross-polarimetric isolation of the
OMT and antenna. The difference varies as the rotation angle
increases. At 90 degrees, the measured LDR is the same as
at 0 degrees, as expected. Thus, a correction based on the
cross-polarimetric isolation will need to be applied to the mea-
sured polarimetric variables when transforming to a common
polarimetric basis.
3.4 Cloud Liquid Water Content Measurements
In general, single-wavelength retrievals of liquid water con-
tent (LWC) cannot be estimated accurately from reflectivity
(Vivekanandan et al., 1999). For example, for Rayleigh scat-
tering, LWC can vary from 0.2 to 0.7 g m−3when the reflec-
tivity is −10 dBZ. However, if the drop size distribution of the
clouds is measured with in situ probes on HIAPER, it may be
possible to generalize the retrieval to the rest of the cloud using
the reflectivity measurement.
Dual-wavelength measurements from the Phase C sys-
tem will provide a more accurate retrieval than the single-
wavelength retrieval. The standard deviation in mean LWC in
a layer between heights h1and h2is directly related to the un-
certainty in radar reflectivity (Hogan et al., 2005) and is given
by
∆LWC =∆Z2
35 +∆Z2
94(1/2)
√2(
κ
94 −
κ
35) (h2−h1)(13)
where ∆Z2
35 and ∆Z2
94 are the variance in reflectivity at 35 and
94 GHz respectively, and
κ
94 and
κ
35 are the one-way specific
attenuation coefficients of liquid water in dB km−1(g m−3)−1
for 94 and 35 GHz respectively.
The specific attenuation of liquid water calculated us-
ing MPM93 is 1.298 dB km−1(g m−3)−1at 35 GHz and
4.568 dB km−1(g m−3)−1at 94 GHz for a pressure of 700 mb
and a temperature of −10◦C. Assuming that both the 35 and
Figure 5: The top left-hand plot shows the actual (solid) and
measured (dashed) values of Zdr versus rotation angle for
a cross-polarimetric isolation of 30 dB. The bottom left-hand
plot shows the difference in actual and measured Zdr. The
top right-hand plot shows the actual (solid) and measured
(dashed) values of LDR versus rotation angle for a cross-
polarimetric isolation of 30 dB. The bottom right-hand plot
shows the difference in actual and measured LDR.
Figure 6: Predicted LWC error from dual-wavelength (35 and
94 GHz) measurements as a function of signal to noise ratio
for dwell times of 120, 240, and 480 ms.
94 GHz have the same signal to noise ratio, pulse repetition
frequency, and assuming that the spectral width of the scatter-
ers is the same, the standard deviation in LWC versus signal
to noise ratio can be plotted for various dwell times (Figure 6).
In this simulation, four resolution gates are averaged, and the
standard deviation in velocity of the scatterers is 0.5 m s−1.
The standard deviation in liquid water content for these pa-
rameters varies from 0.26 g m−3to 0.52 g m−3for dwell time
from 480 to 120 ms and signal to noise ratios above 0 dB. LWC
concentration that is typically found in stratocumulus clouds
(<1 g m−3) is on the order of the accuracy achievable with the
system configuration only for dwell times greater than 480 ms.
This time corresponds to 96 meters along the track of the air-
craft at 200 m s−1, which depending on the homogeneity of
the clouds under being studied, may be sufficient to make a
useful retrieval.
ACKNOWLEDGMENT
The National Center for Atmospheric Research is sponsored
by the National Science Foundation. The views expressed are
those of the authors and do not necessarily represent the offi-
cial policy of the U.S. government.
REFERENCES
Doviak, R. J. and D. S. Zrni ´
c, 1993: Doppler Radar and
Weather Observations. Academic Press, second edition.
Hogan, R. J., N. Gaussiat, and A. J. Illingworth, 2005: Stra-
tocumulus liquid water content from dual-wavelength radar.
J. Atmos. Oceanic Technol.,22, 1207–1218.
Horie, H., T. Iguchi, H. Hanado, H. Kuroiwa, H. Okamoto,
and H. Kumagai, 2000: Development of a 95-GHz airborne
cloud profiling radar (SPIDER) - Technical aspects. IEICE
Trans. Comm.,E83B, 2010–2020.
Laursen, K. K., D. P. Jorgensen, G. P. Brasseur, S. L. Ustin,
and J. Hunning, 2006: HIAPER: The next generation
NSF/NCAR research aircraft. Bulletin of the American Me-
teorological Society,87, 896–909.
Liebe, H. J., G. Hufford, and M. Cotton, 1993: Propagation
modeling of moist air and suspended water/ice particles
at frequencies below 1000 ghz. Proc. of 52nd Specialists
meeting of the Electromagnetic Wave Propagation Panel,
AGARD, 542.
Pazmany, A. L., R. E. McIntosh, R. Kelly, and V. G., 1994:
An airborne 95-GHz dual-polarized radar for cloud studies.
IEEE Trans. Geosci. Remote Sens.,32, 731–739.
Stephens, G., D. G. Vane, R. J. Boain, G. Mace, K. Sassen,
Z. E. Wang, A. J. Illingworth, E. J. O’Connor, W. B.
Rossow, S. Durden, S. Miller, R. T. Austin, A. Benedetti, and
C. Mitrescu, 2002: The cloudsat mission and the A-train – A
new dimension of space-based observations of clouds and
precipitation. Bulletin of the American Meteorological Soci-
ety,83, 1771–1790.
Vivekanandan, J., B. E. Martner, M. K. Politovich, and G. F.
Zhang, 1999: Retrieval of atmospheric liquid and ice char-
acteristics using dual-wavelength radar observations. IEEE
Trans. Geosci. Remote Sens.,37, 2325–2334.
Wolde, M. and A. Pazmany, 2005: NRC dual-frequency air-
borne radar for atmospheric research. 32nd Int. Conf. on
Radar Meteor., AMS.
