You are on page 1of 4

RADAR CALIBRATION

SOME SIMPLE APPROACHES


BY D A V I D A T L A S

I -«

In considering new and promising methods to calibrate radar, it is worth remembering some of the
old and perhaps forgotten methods that were used over the last half century.

D
uring the Radar Calibration Workshop at the shop. While formalizing these remarks in writing I
81st Annual Meeting of the American Meteo- thought it would be useful to elaborate upon them and
rological Society in Albuquerque, New Mexico, discuss some newer approaches. Thus this paper at-
in January 2001,1 was surprised at the relatively little tempts to synthesize a range of techniques. A com-
attention given to some of the simplest and proven mon thread that runs throughout is the calibration of
methods. This stimulated some extemporaneous re- the overall system by use of standard or well-defined
marks that I presented toward the end of the work- targets external to the radar.
In part, I was troubled by the apparent lack of fa-
miliarity of some of the younger generation with early
A F F I L I A T I O N : ATLAS—NASA Goddard Space Flight Center,
activities in this realm. I was also reacting to the re-
Greenbelt, Maryland
C O R R E S P O N D I N G A U T H O R : David Atlas, Distinguished Visiting
cent findings of the variability in the calibrations of
Scientist, N A S A Goddard Space Flight Center, C o d e 910,
the Weather Surveillance Radars-1988 Doppler
Greenbelt, MD 20771 (WSR-88Ds) around the nation that have been uncov-
E-mail: datlas@radar.gsfc.nasa.gov

In final form 22 May 2002 Above: In the early 1970s, Atlas used BBs to cali-
©2002 American Meteorological Society brate the vertically pointing frequency modulated-
continuous wave ( F M - C W ) radar.

A M E R I C A N M E T E O R O L O G I C A L SOCIETY SEPTEMBER
Unauthenticated 2002 01/25/21
| Downloaded B A I T S ' I07:49
1 3 1 3 AM UTC
ered by comparison with the radar measurements of tenna. These frequencies are readily distinguished
precipitation by the radar on board the Tropical Rain- from ground clutter and precipitation echoes. It is an
fall Measuring Mission (TRMM); (Bolen and excellent calibration device because it is always avail-
Chandrasekar 2000). The remarkable stability of the able regardless of the weather.
TRMM precipitation radar has made it a traveling
standard against which ground-based weather radars 88s. We first used BBs fired vertically from a BB pis-
can be calibrated. tol as standard targets to calibrate the vertically point-
There were a few papers presented at the work- ing frequency modulated-continuous wave (FM-CW)
shop that resorted to the more traditional methods radar at the Naval Electronics Laboratory Center at
such as calibration with a standard target. David Point Loma, California (Stratmann et al. 1971). After
Brunkow of Colorado State University spoke about having failed to support a calibration sphere from a
the use of a metal sphere. Ron Rinehart of the Uni- balloon in a stable position on the axis of the radar
versity of North Dakota used an oscillating dihedral beam we searched for another approach. In a joking
corner reflector. Also Isztar Zawadzki recounted his manner I suggested the use of a BB gun. Although
work with rain gauges and a Joss-Waldvogel (J-W) there was no prior literature on the subject it was
disdrometer. Surely, few of the participants were cheap, straightforward, and worth a try. We were very
aware that the early workers in Canada (Stewart pleased by how well it worked. If enough BBs are used
Marshall, Bob Langille, and Walter Palmer) and in (one at a time), the statistics of echo strength mimic
my group at the Air Force Cambridge Research Labo- the radiation pattern of the beam. The maximum echo
ratories (Vernon Plank, A1 Chmela, and I) used fil- corresponds to the antenna gain on the beam axis.
ter papers powdered with Gentian violet dye (which When using a conventional radar, one should tilt the
left purple stains on our clothes and teeth) to mea- beam close to the horizon outside the region of
sure the sizes of tens of thousands of drops by hand ground clutter. With Doppler radar, the Doppler shift
in the late 1940s and early 1950s (Hitschfeld 1986). can be used to distinguish the moving BBs from
Oh what a blessing it was to display the drop size dis- clutter.
tribution in a comfortable laboratory , while the J-
W disdrometer was observing the size of each drop Metalized Ping-Pong balls. T h i s is a n e x t e n s i o n of t h e
automatically outdoors. BB method. One can fly a light aircraft across and
Historically, it was the Weather Radar Group at the above a fixed radar beam and drop the balls sequen-
Massachusetts Institute of Technology (MIT), under tially at about 10-15 m intervals so that only one tar-
the leadership of Alan Bemis and the seminal work get is in the beam at any time. The metalized balls
by Polly Austin and Ed Williams (1951), that found are good targets of known radar cross section. The
the large underestimates of the radar echoes from successive echoes present a quantitative measure of
gauge measurements of rain in comparison to the the antenna pattern. Tracking of the aircraft and
then-available theory. It was this difference that mo- timing of each drop positions each target relative to
tivated Richard Probert-Jones (1962) in England to the maximum echo on the beam axis. The Ping-
formulate the proper radar equation for meteorologi- Pong balls are cheap and nonhazardous. One may
cal scatterers (Hitschfeld 1986). For almost a decade also use metalized wiffle balls (with holes in them).
we all struggled to understand the source of this dis- The idea is to prevent either type of ball from falling
crepancy. And here we are today still struggling with fast enough to create a hazard. Note that either of
the optimum methods of radar calibration. these types of balls may be within the Mie region
depending on the radar wavelength so that their
C A L I B R A T I O N METHODS. Frequency shift re- cross sections should be computed carefully. It is also
flector (FSR). The FSR was invented by John Chisholm possible to release such targets sequentially from a
(1963). It has been used mainly as a ground-based bucket carried on a constant-level balloon moving
target for precise locations on airports and geographi- with the winds perpendicular to the fixed radar
cal siting. It employs a parabolic reflector with a horn beam. A similar method was used to measure the
at the focus that is shorted by a diode at a frequency/ cross section of a free-falling artificial hailstone re-
(e.g., 30 or 60 MHz). The frequency/is generated by leased from a balloon and measured by a tracking
a battery-driven modulator. The echo from the tar- radar (Willis et al. 1964).
get is returned at F ±/, where F is the transmitted fre-
quency. The echoes at ±f are exactly 6 dB below that Airborne modulated target. This approach combines the
corresponding to the known cross section of the an- concepts in the frequency shift reflector (FSR) and

1 3 1 4 | BAI1S- SEPTEMBER 2002 Unauthenticated | Downloaded 01/25/21 07:49 AM UTC


Standard Target Radar (STADAR; Atlas 1967). from tethered balloons or kytoons. Some have used
STADAR employs a rotating standard target on the three tethers to stabilize the position of the balloon.
aircraft that modulated the total echo of the aircraft During experiments in England we used a tethered
and the target at a frequency corresponding to the balloon with a standard 12 in. diameter metal sphere
rotation frequency. The original idea was aimed at using and an ice ball (i.e., a simulated hailstone) of unknown
a simple CW radar to detect the range to the target by cross section suspended below the balloon at a suffi-
the intensity of the echo from the rotating target of cient vertical spacing to separate the known and un-
known cross section using the radar equation to com- known targets. Swinging the beam from one to the
pute the range. However, it would be greatly improved other allowed us to measure the cross section of the
by using an FSR on board the aircraft so that the echo simulated hailstone with accuracy of better than -0.5
is returned at a frequency that is different from that of dB. This was more easily done at the time because of
the carrier frequency and thus separated from the air- the use of relatively wide beam height-finder radars
craft echoes. such as the MPS-4 and the TPS-10 (Atlas et al. 1960).
For greater use it is best to do this in the light winds
Balloon-borne or airborne standard target This is an old of early morning or evening.
scheme that must go back to World War II. However,
we first used it in 1953 when we suspended a metal- Use of a radar profiler and disdrometer. T h e use of a
ized sphere below a helicopter and carried it across Doppler radar profiler (at vertical incidence) along-
the beam of our 24-GHz radar in a study of the radar side a disdrometer allows the measurement of the
characteristics of fog (Atlas et al. 1953). That study drop size distribution (DSD) at the surface, computa-
was aimed at determining the relationship of the ra- tion of its associated value of reflectivity, and compari-
dar reflectivity to the liquid water content and drop son to the reflectivity measured by the radar at heights
sizes of fog. Many others have used this method but of 300-400 m just beyond the radar recovery time.
found it difficult to track the target in a narrow beam. This calibrates the radar remarkably well. The method
At the present time the use of the global positioning was first used by Joss et al. (1968). They measured the
system (GPS), either on the balloon or the airplane, reflectivity at a height of only 200 m above their zenith
would facilitate tracking. pointing radar while measuring the rain and DSD with
Calibration with a 24-in. metal sphere suspended gauges and a disdrometer. In 46 periods of uniform
from a balloon was done quite reliably by Atlas and stratiform rain they found excellent agreement between
Mossop (1960) by tracking the balloon with a long, the actual and the disdrometer-deduced values of Z with
easily identified tail by theodolite. Today one might a standard deviation of only 6% or 0.25 dB in the ratio
mount a television camera on the bore sight axis of between the two. It is also remarkable that the radar
the antenna and use the wide angle lens to find the calibration was maintained to this accuracy for a pe-
balloon and then change to telephoto mode to find it riod of 4 months.
accurately and adjust the radar position accordingly. This approach has been extended by Gage et al.
(2000) and others. An analogous technique is that of
Metalized spherical target released from aircraft. D u r i n g Kollias et al. (1999), who used a vertically pointing
experiments at Wallops Island, Virginia, to measure 94-GHz Doppler radar. At this frequency the Mie
the cross sections of individual insects and birds, the backscatter function results in a well-defined mini-
latter targets were released from an aircraft flying into mum in the Doppler spectrum at a specific drop size.
the wind while being tracked by the radar (Glover The difference between the measured Doppler speed
et al. 1966). The targets were released on countdown and the known fall speed for that drop size in still air
and the tracking gate was stopped until the aircraft is then a measure of the air motion; hence, the Dop-
moved out of the gate and the unknown target could pler spectrum in still air may be recovered and the
be gated and tracked. Then the aircraft moved upwind DSD and its reflectivity may be computed. The abso-
while the target moved downwind. This approach lute number of drops depends upon the overall radar
requires the use of a tracking radar that can control calibration and the attenuation by the rain. Thus one
the weather radar. A metalized spherical, constant- still needs to use a disdrometer adjacent to the radar
altitude balloon can be released from the aircraft and to account for the attenuation. Once the zenith point-
expanded upon release by the use of a gas cartridge. ing radars are calibrated in this fashion, they may be
used as transfer standards for other radars.
Tethered balloon or kytoon. Many investigators have Ulbrich and Lee (1999) have used the reflectivity
used metal spheres of known cross section suspended computed from drop size distributions measured with

AMERICAN METEOROLOGICAL SOCIETY SEPTEMBER


Unauthenticated 2002 01/25/21
| Downloaded BAI1S" I07:49
1 3 1 5 AM UTC
a disdrometer at the surface to check the calibration Austin, P. M., and E. L. Williams, 1951: Comparison of
of the WSR-88D at Greer, South Carolina, about radar signal intensity with precipitation rate.
60 km away from their site at Clemson University. Weather Radar Research Tech. Rep. 14, Dept. of
They found that the radar gain was consistently 5 dB Meteorology, Massachusetts Institute of Technology,
too low. This is a straightforward technique, particu- 43 pp.
larly when used in relatively steady rainfall when the Bolen, S. M., and V. Chandrasekar, 2000: Quantitative
bright band is high. It is similar to the schemes used cross validation of space-based and ground-based
by Joss et al. (1968) and that reported by Zawadzki at radar observations. /. Appl. Meteor., 39, 2071-
this workshop. 2079.
Chisholm, J., 1963: Frequency shift reflector. U.S. Patent
Measurement of DSD by aircraft. O n e m a y use obser- No. 3,108,275.
vations of the drop size distribution on board an air- Gage, K. S., C. R. Williams, P. E. Johnston, W. L.
craft for comparison to ground-based radar measure- Ecklund, R. Cifelli, A. Tokay, and D. A. Carter, 2000:
ments. This has been done by Marks et al. (1993) to Doppler radar profilers as calibration tools for scan-
calibrate and obtain the Z-R relation in a hurricane. ning radars. /. Appl. Meteor., 39, 2209-2222.
In the latter case, the radar was on board the aircraft Glover, K. M., K. R. Hardy, T. G. Hardy, W. N. Sullivan,
and measured the reflectivity at a modest distance and A. S. Michael, 1966: Radar observations of insects
ahead. The DSD was then measured a few minutes in free flight. Science, 154, 967-972.
later as the aircraft penetrated the radar-measured Hitschfeld, W., 1986: The invention of radar meteorol-
location. ogy. Bull. Amer. Meteor. Soc., 67, 33-37.
After 56 years of research in radar meteorology, we Joss, J., J. C. Thams, and A. Waldvogel, 1968: The accu-
have still failed to find a reliable and universally ap- racy of daily rainfall measurements by radar. Pre-
plicable method of radar calibration. Various radar prints, 13th Radar Meteorology Conf., Montreal, QC,
configurations require different approaches. I hope Canada, Amer. Meteor. Soc., 448-451.
that this brief essay will serve as a menu of simple Kollias, P., R. Lhermitte, and B. Albrecht, 1999: Verti-
methods to fit the needs of various investigators and cal air motion and rain drop size distributions in
operational users. convective systems using a 94 GHz radar. Geophys.
Res. Lett., 26, 3109-3112.
ACKNOWLEDGMENTS. I appreciate the discussions Marks, F. D., Jr., D. Atlas, and P. T. Willis, 1993: Prob-
with Dr. Merrill Skolnik, former Superintendent of the ability matched reflectivity-rainfall relations for a
Radar Division of the Naval Research Laboratories. He re- hurricane from aircraft observations. /. Appl. Meteor.,
mains skeptical about the accuracy that may be achieved 32, 1134-1141.
by some of the techniques described. This work was done Probert-Jones, J. R., 1962: The radar equation in meteo-
under the aegis of the NASA Tropical Rainfall Measuring rology. Quart. J. Roy. Meteor. Soc., 88, 485-495.
Mission. Stratmann, E., D. Atlas, J. H. Richter, and D. R. Jensen,
1971: Sensitivity calibration of a dual-beam vertically
pointing FM-CW radar. /. Appl. Meteor., 10, 1260-
REFERENCES 1265.
Atlas, D., 1967: STADAR, standard target radar. U. S. Ulbrich, C. W., and L. G. Lee, 1999: Rainfall measure-
Patent No. 3,357,014. ment error by WSR-88D radars due to variations in
, and S. C. Mossop, 1960: Calibration of a weather Z-R law parameters and radar constant. /. Atmos.
radar by using standard target. Bull. Amer. Meteor. Oceanic Technol, 16, 1017-1024.
Soc., 41, 377-382. Willis, J. R., K. A. Browning, and D. Atlas, 1964: Radar
, W. H. Paulsen, R. J. Donaldson, A. C. Chmela, and observations of ice spheres in free fall. /. Atmos. Sci.,
V. G. Plank, 1953: Observation of the sea breeze by 21, 103-108.
1.25 cm radar. Proc. Conf. on Radio Meteorology,
Austin, TX, Amer. Meteor. Soc., Paper XI-6.
, W. G. Harper, F. H. Ludlam, and W. C. Macklin,
1960: Radar scatter by large hail. Quart. J. Roy. Me-
teor. Soc., 86, 468-482.

1 3 1 6 | BAI1S- SEPTEMBER 2002 Unauthenticated | Downloaded 01/25/21 07:49 AM UTC

You might also like