August 2003

1
Hurricane Humberto
Drop Size Ambiguities in the Retrieval of Rain
Ratesfrom the GPM Core Measurements
14GHz reflectivities
35GHz reflectivities
J. P. Meagher and Z. S. Haddad meagher_at_ampersand.j
pl.nasa.gov Jet Propulsion Laboratory California
Institute of Technology
August 2003
Dual-frequency (14 and 35 GHz) radar reflectivity
profiles collected during the CAMEX-4 experiment
over Tropical Storm Gabrielle and Hurricane
Humberto (using JPLs GPM-simulating PR-2
airborne rain radar) were processed using a
Bayesian approach to estimate rain-rate and
mean-drop size profiles for each vertical beam.
Estimated rain rate
Estimated mean drop size
E
E
Six DSD models were considered (E) the
exponential (G1 and G2) two gamma models, each
with a different imposed m-N0 relation (G0) the
gamma DSD with the rain-normalized mass-weighted
drop diameter variance fixed (C) the COARE
database of averaged sampled DSDs (CC) the
COARE database with a hard-wired cloud-water
proportion
G1
G1
E
The Z14-Z35 manifold of radar reflectivities
corresponding to each of the first five models
are shown here (in the CC model, the cloud was a)
monodisperse, b) with liquid water content equal
to 20 of the rain)
The Bayesian procedure tries to estimate 4
variables at each height bin the rain rate, the
mass-weighted mean drop size, the accumulated
14GHz attenuation and the accumulated 35GHz
attenuation. The rain-rate and mean-drop-size
are estimated directly from the two measured
radar reflectivities (at 14 and 35 GHz), and the
two attenuations are updated for the next height
bin. The radar-rain relations were compiled in
the form of a table constructed from Mie
calculations using each of the six DSD models.
The main conclusion seems to be that several
quite different DSD models do indeed produce
plausible precipitation estimates. The
precipitation amounts do differ from model to
model, though the general shape of the vertical
variation of the retrieved drop size seems
similar among the different models
considered. Most important, it does appear that
the a-priori decision about which drop size
distributions should be considered plausible does
have a determining effect on the eventual
retrievals. These a-priori assumptions will
therefore have to be backed up by detailed DSD
measurements at radar-sized resolutions, such as
can be obtained from dual-frequency wind
profilers.
G2
G2
G1
G2
G0
G0
C
G0
C
C
Tropical Storm Gabrielle
14GHz reflectivities
35GHz reflectivities
Reconstruction - Measurement error in dB (14 GHz)
Reconstruction - Measurement error in dB (35 GHz)
E
E
G0
G0
Estimated rain rate
Estimated mean drop size
C
C
E
E
14 GHz Path-Integrated Attenuation
35 GHz Path-Integrated Attenuation
G0
G0
Triple-frequency radar measurements of light
precipitation implications for Particle Size
Distributions
C
C
In January and February 2003 a joint field
experiment was conducted by the US AMSR-E and
Japanese AMSR teams over the Wakasa Bay area of
Japan. As part of the experiment the NASA P3
aircraft flew the JPL APR-2 dual frequency
precipitation radar operating at 14 and 35 GHz,
the University of Massachusetts/JPL ACR 95 GHz
cloud radar, and a number of passive instruments.
The reflectivies at the three different
frequencies (14,35,95 GHz) are shown in Figure
below left. This is a powerful data set for
research for both the CloudSat mission and
TRMM/GPM. For CloudSat research we have the 95GHZ
cloud radar which is very similar to the CloudSat
radar, plus the 14 and 35 GHz channels. With
these three frequencies we are able to
investigate the type of DSD we are likely to see
with CloudSat in areas of light precipitation.
For GPM we have the 14 and 35 GHz radar
frequencies of the proposed core satellite, plus
the 95GHz channel. With the inclusion of the
95GHz radar we are able to observe the light
precipitation that will be below the sensitivity
of the GPM core satellite radars, and estimate
the effect of smaller particles on the radar
attenuation at 14 and 35 GHz to which the 95GHz
radar has great sensitivity. Figure below right
shows scatter plots of the triple frequency
measurements at and below the Bright-Band. In the
BB a number of particles are in the Mie
scattering regime for at least the 35GHz channel,
below the BB the particles have most likely
melted reducing their size, so that they are
mostly in the Rayleigh regime for the 14 and 35
GHz radars.
CC
CC
Reconstruction - Measurement error in dB (14 GHz)
Reconstruction - Measurement error in dB (35 GHz)
E
E
G0
G0
C
C
CC
CC
14 GHz Path-Integrated Attenuation
35 GHz Path-Integrated Attenuation