1. the University
of Oklahoma
Spectral Polarimetry for Identifying and Separating Mixed Biological Scatterers
Svetlana Bachmann1,2, Dusan Zrnic2,
1 Cooperative Institute for Mesoscale Meteorological Studies, The University of Oklahoma; 2 NOAA/OAR National Severe Storm Laboratory, Norman, OK
Case – KOUN (Norman, OK) - 09/07/ pm CDT (04Z) – Clear Air with Mixed Biological Scatterers
Summary A B
Signal Processing
Example – radial at azimuth 180
Spectral densities of polarimetric variables (Spectral Polarimetry)
We examine Doppler spectra and show that separation of birds from insects is possible if the antenna is pointing in (or near) the direction of wind. Further, spectral densities
of dual polarization variables contain discernible signatures that allow discrimination between birds and insects and lead to the measurement of winds aloft.
Spectral fields
DC removed from I and Q
Ground Clutter removed
Notch and Interpolate spectral coefficients corresponding to near-zero velocity
Pulse Pair velocity estimator
Ranges km, Azimuths 
Polarimetric variables are computed for each Doppler velocity (spectral density) using H- and V-channel spectral coefficients. Spectral densities are plotted for all range locations in a radial. H
PH, dB
Velocity, m s–1
Velocity, m s–1 PH
Example - radial at azimuth 180 V
Zdr
|hv|
dp
Range, km
Zdr
Velocity, m s–1
Altitudes
Flight velocity
Flight direction
Polarimetric values (Zrnic and Ryzhkov, 1998)
below 3 km,
mainly under 900 m
8 – 22 m s–1
Favorable winds
–1 dB < Zdr < 3 dB
dp > 100°
several
hundred meters
1 m s–1 or less
(small insects, spiders, etc.)
Passive wind tracers
Zdr < 10 dB
dp < 40°
Facts
Migrating Birds
Wind-carried Insects
Velocity, m s–1
Velocity, m s–1
Zdr, dB
Range, km
PPI at 0.5 elevation
Continuous & Sporadic Bands
above the noise power are evident in
spectral fields for each radial of the PPI
|hv|
H channel Reflectivity
Velocity, m s–1
Velocity, m s–1
|hv|
100 80 60 40 20
Velocity, m s–1
Insects
Birds
Range, km 20 15 10 5 -5
<-10
Z, dB
dp
Velocity, m s–1
Velocity, m s–1
dp, 
Azimuth, deg
Average spectral densities from 30 to 70 km range for each radial of PPI
Polarimetric variables
computed from selected spectral coefficients
Migrating and wandering birds are not passive wind tracers and they contaminate wind velocity estimates. Wind-carried insects are useful scatterers that provide echoes needed for wind profiling.
Zdr, (dB)
VAD from
spectral density of Zdr
v, m s–1 30 20 10
-10
-20
-30
100 80 60 40
H channel Velocity
Radar set up
30 km
70 km
H channel Velocity
Elevation Scan (PPI) – 360 radials at 0.5 elevation angle
|hv|
-35
100 80 60 40 20
Radial – two polarization channels
468 range locations
Ra = 117 km
v, m s–1 30 20 10
-10
-20
-30
Velocity, m s–1 H
Range, km
Noise V
Ground Clutter
Contaminated
Good Separation 35 N S W E NE SE SW NW
Range location –
two time series of I & Q samples
Recovered
128 I & Q
samples H channel
va = 35 m s–1
Azimuth
Wind
Velocity
Insects
Birds
Histograms of polarimetric variables azimuth 180, ranges 20 – 70 km
128 I & Q
samples V channel
Zdr, dB
vmax ~ 30 m s–1
Conclusions
vmax ~14 m s–1
|hv|
Polarimetric variables computed from contiguous sections of Doppler spectra allow almost perfect discrimination of birds and insects.
Wind data contaminated by migrating birds can be recovered:
by computing the mean Doppler velocity from selected
spectral coefficients:
window positioning - from block analysis of spectra or from polarimetric variables;
window span - depends on number of spectral coefficients, needed accuracy of the estimator, and azimuth angle.
by reconstructing VAD of wind from spectral densities of
polarimetric variables.
Range
Horizontal
Vertical
For a range location
Spectrum H
and Spectrum V
dp, deg
Adaptive Window of
selected spectral coefficients
in Spectrum H A
Range
location
Polarimetric Variables
Zdr, |hv|, dp
computed from contiguous sections
of spectral coefficients
discriminate Insects and Birds
Dual
Polarization
Doppler
Radar B