Transcript Document

The importance of ice particle
shape and orientation
for spaceborne radar retrievals
Robin Hogan, Chris Westbrook
University of Reading
Lin Tian
NASA Goddard Space Flight Center
Phil Brown
Met Office
Introduction and overview
• To interpret 94-GHz radar reflectivity in ice clouds we need
– Particle mass: Rayleigh scattering up to ~0.5 microns: Z mass2
– Particle shape: non-Rayleigh scattering above ~0.5 microns, Z also
depends on the dimension of the particle in the direction of propagation
of the radiation
• Traditional approach:
– Ice particles scatter as spheres (use Mie theory)
– Diameter equal to the maximum dimension of the true particle
– Refractive index of a homogeneous mixture of ice and air
• New observations to test and improve this assumption:
– Dual-wavelength radar and simultaneous in-situ measurements
– “Differential reflectivity” and simultaneous in-situ measurements
• Consequences:
– Up to 5-dB error in interpretted reflectivity
– Up to a factor of 5 overestimate in the IWC of the thickest clouds
Dual-wavelength ratio comparison
10 GHz, 3 cm
Error 1: constant 5-dB
overestimate of Rayleigh10 GHz, 3 cm scattering reflectivity
94 GHz, 3.2 mm
94 GHz, 3.2 mm
Difference
• NASA ER-2 aircraft
in tropical cirrus
Error 2: large overestimate
in the dual-wavelength
ratio, or the “Mie effect”
Characterizing particle size
• An image measured by aircraft can be approximated by a...
Sphere (but which diameter do we use?)
Spheroid (oblate or prolate?)
Note:
Dmax  Dlong
Dmean=(Dlong+Dshort)/2
Error 1: Rayleigh Z overestimate
• Brown and Francis (1995) proposed
mass[kg]=0.0185 Dmean[m]1.9
– Appropriate for aggregates which
dominate most ice clouds
– Rayleigh reflectivity Z  mass2
– Good agreement between simultaneous
aircraft measurements of Z found by
Hogan et al (2006)
• But most aircraft data world-wide
characterized by maximum particle
dimension Dmax
– This particle has Dmax = 1.24 Dmean
– If Dmax used in Brown and Francis
relationship, mass will be 50% too high
– Z will be too high by 126% or 3.6 dB
– Explains large part of ER-2 discrepancy
Randomly oriented in aircraft probe:
Particle shape
• We propose ice is modelled as oblate
spheroids rather than spheres
– Korolev and Isaac (2003) found typical
aspect ratio a=Dshort/Dlong of 0.6-0.65
– Aggregate modelling by Westbrook et al.
(2004) found a value of 0.65
Horizontally oriented in free fall:
Error 2: Non-Rayleigh overestimate
Transmitted
wave
Sphere
Sphere: returns
from opposite
sides of particle
out of phase:
cancellation
Spheroid:
returns from
opposite sides
not out of
phase: higher Z
Independent verification: Z dr
• A scanning polarized radar measures differential reflectivity,
defined as: Zdr = 10log10(Zh/Zv)
Dshort/Dlong:
Dependent on
both aspect
ratio and
density (or ice
fraction)
Solid-ice
oblate
spheroid
If ice particles
were spherical,
Zdr would be
zero!
Sphere: 30%
ice, 70% air
Solid-ice sphere
Chilbolton 10-cm radar + UK aircraft
• Reflectivity agrees
well, provided Brown
& Francis mass used
with Dmean
• Differential
reflectivity agrees
reasonably well for
oblate spheroids
Z dr statistics
• One month of data from a 35GHz (8-mm wavelength) radar
at 45° elevation
– Around 75% of ice clouds sampled
have Zdr< 1 dB, and even more for
clouds colder than -15°C
– This supports the model of oblate
spheroids
• For clouds warmer than -15°C,
much higher Zdr is possible
– Case studies suggest that this is
due to high-density pristine plates
and dendrites in mixed-phase
conditions (Hogan et al. 2002,
2003; Field et al. 2004)
Consequences for IWC retrievals
• Empirical formulas derived from aircraft will be affected, as
well as any algorithm using radar:
Radar reflectivity ~5 dB
higher with spheroids
Raw aircraft data
Retrieved IWC can be out by
a factor of 5 using spheres
with diameter Dmax
Empirical IWC(Z,T) fit
Spheres with D =Dmax
Hogan et al. (2006) fit
New spheroids
Note: the mass of the particles in these three examples are the same