Transcript Retrieving turbulence parameters from cloud radar

CloudNET: retrieving
turbulence parameters
from cloud radar.
Anthony Illingworth, Robin Hogan , Ewan O’Connor, U of Reading, UK
Dominique Bouniol, CETP, France
New method of estimating turbulence
Previous methods used :
• Doppler spectral width (for ground based radar)
 but also contributions from shear and terminal velocity
• Spectral analysis of w (from airborne and ground observations)
 only gives e at a given level & time – noisy for low w.
New Method:
Vertically pointing narrow beamwidth radar.
Look at 1 second values of mean Doppler v for 30 secs.
And calculate the standard deviation over 30 secs: sv
Beamwidth very narrow, horizontal wind U m/s.
So in 1 second
U m of clouds advects past.
And in 30 seconds 30U m of cloud advects past.
e.g. U=10m/s sample scales 10 to 300m.
i.e sample turbulent spectrum between;
k1 = 2/30U
and k2 = 2  /U
NEED TO KNOW THE HORIZONTAL WIND
Turbulence measurements
• Changes in 1-s mean
Doppler velocity
dominated by changes
in vertical wind, not
terminal fall-speed
– We calculate new
parameter: 30-s standard
deviation of 1-s mean
Doppler velocity, sv
– Can use this to estimate
turbulent kinetic energy
dissipation rate
– Important for vertical
mixing, warm rain
initiation in cumulus etc.
Spectral width sv
contaminated by
variations in particle
fall speed
Measurements of “sigma-v-bar”
• 26 Jan 2004
Stable layer: sv~3 mm/s
Frontal shear layer: sv~3 cm/s
Unstable evaporating layer sv~30 cm/s
TKE dissipation rate
e
• Part of TKE spectrum can be interpreted in terms of
the variance of the mean Doppler velocity:
k2
s v2   S (k )dk
k1
– k1 is min horizontal wavenumber sampled in 30 s (use model winds)
– k2 is max horizontal wavenumber due to beamwidth of radar
• In the inertial sub-range (Kolmogorov)
S (k )  ae 2 / 3k 5 / 3
• Hence by integration
 2
e  
 3a 
3/ 2
s v3k1
k1
k2
Calculation of e
 2
e  
 3a 
3/ 2
s v3k1
1. Use model winds to find the value of k1.
- this may fail in the tropics – unpredictable winds.
ideally have a co-located wind profiler.
2. Remove any linear trends in the one second value of v:
this could be due to gravity waves.
3. Check that changes in v not due to gradients in Z –leading to
changes in terminal velocity, by computing sZ/Z(av).
Reject data if this is too high.
Dissipation rate in different clouds
• Z
Cirrus
Rain
• e
Stratocu
1 –year of CloudNet data
• PDF of
dissipation rate
for different
types of cloud
• Note that aircraft
measurements
have lower limit
of detectability
of ~10–6 due to
aircraft
vibrations
0.02 – to trigger
Coalescence in Cu?
Previous range for cirrus Khain and Pinsky, 1997
found from aircraft