Transcript Statistical relationships between integral parameters of

Retrieval of ice cloud properties
from Doppler cloud radar
measurements
Delanoë J., Protat A., Bouniol D., Testud J.
Centre d’étude des Environnements Terrestre et Planétaires
CloudNET meeting : Delft
18/19th October 2004
1/17
Julien Delanoë
Outline
2/17

Overview: Microphysics calculations
from particle concentration spectra

Normalized Particle size distribution
(PSD)

RadarOnly Algorithm
The ice cloud properties :
M n   N ( D eq) D eq d D eq
n
Deq: Equivalent melted diameter
The cloud parameters also depend on
the moments of PSD
Dm 
IWC 
3/17
4
N
(
D
)
D
 eq eq dDeq
3
N
(
D
)
D
 eq eq dDeq
 w
6
3
N
(
D
)
D
 eq eq dDeq


2
103  N ( D) D 2 dD
re  IWC / 
The three measurements of a
Doppler cloud radar
For a vertically-pointing cloud radar :
• Reflectivity factor:
Ze 
4
2
Kw 
5
1018  N ( D) bsc dD
• Doppler velocity
 v bscN ( D)dD
Vd 
w
  bscN ( D)dD
• Doppler spectral width
 vt 
4/17
 v ² bscN ( D )dD
 Vt ²
  bscN ( D )dD
All
measurements
depend on N(D)
The normalized particle size
distribution
High variability in ice clouds
Scaling the PSD so that it does not depend on IWC, Dm
N(D) = N0* F (Deq/Dm)
Delanoë et al. (JGR, 2004) : shape F can be
approximated
by a single modified gamma formulation for all ice
clouds (less than 5% error)
The unknowns that remain to get cloud properties :
N0* and Dm
5/17
Normalized PSD
Mean normalized spectra For all experiments
Stability of the PSD shape
Dataset:
CLARE 98, CARL 99,
EUCREX, ARM SGP, FASTEX,
CEPEX, CRYSTALFACE
All experiments
6/17
Main result : Stability of the normalized PSD shape
Radar Only Algorithm
RadOn
Principle of the radar
retrieval method
7/17
Principle of the radar retrieval
method (1)
First step : Retrieval of VT from (VD , Z) V D w V T
Hypothesis : for a long enough time span w  V t
Vd m.s-1
For example:
14th April 2003 case
Vd= -0.88 Z 0.155 m.s-1
With altitude correction :
We obtain Vt-Z
Z dbZ
Vt= -0.88*exp(-0.4*zh/H) Z 0.155
in m.s-1
A VT-Z relationship is derived for each cloud
8/17
Principle of the radar retrieval
method (2)
Second step : estimate of the particle habit (D), v(D) from VT-Z relationship

Vt-Z relationship obtained from radar is compared to microphysical
Vt-Z relationships with different density and velocity laws.

Microphysical Vt-Z relationships :
(D)=aDb and v(D)=f(m(D),A(D),ad,bd) (Mitchell 1996)
Where m(D)=(/6) a D3+b , A=/4 D2 and ad, bd are the continuous
drag coefficients (Khvorostianov and Curry 2002).
From coefficients of Vt-Z radar relationship we estimate the best
density diameter relationship.
9/17
Principle of the radar retrieval
method (3)
Third step : Dm Retrieval from Vt
Knowing (D), v(D) and assuming a modified gamma PSD shape we
estimate Vt(Dm) :
v bscF ( D / Dm )dD

Vt ( Dm ) 
  bscF ( D / Dm )dD
Inversion of Vt(Dm) using a regression fit of Dm and Vt
Thus, we obtain a Dm value from Vt
10/17
Principle of the radar retrieval
method (4)
Fourth step : Retrieval of N0* from Dm and Ze
A direct relationship between N0* Dm and Ze
2
5
18
1

10


K
w
*
N0 
Z e   F ( D / Dm ) bsc dD 
4



N0* retrieval
11/17
Clouds parameters

IWC=(w/44)N0*Dm4

 for a modified gamma G(,b):
  N0*a
 re=
 
12/17

2
b  3
  5
G

G(4)   b  
44     4  
G b 

 
(3/2) IWC / 

zmax
zmin
 (s)ds
3
6
b 3
6





1
6

b  3  b 3 1
 Dm
G

b





Evaluation of RadOn using the
microphysical database


Dataset: CLARE 98, CARL
99, EUCREX, ARM SGP,
FASTEX, CEPEX,
CRYSTALFACE
We impose a density law :
=0.00556(D in cm)-1.1 and a
radar frequency at 95GHz.
Entries of the algorithm :Vt and
Ze from in situ data
13/17
Bias in
%
std in
%
IWC
-0.2
22.4

-6.5
25.1
re
9.6
17. 9
Application of RadOn on RASTA
data : 14th April 2003
Altitude km
Altitude km
Vd m.s-1
time


14/17
time
Vt-Z relationship for this day in the interval 3 to 9km:
Vt= -0.65Z 0.155 m.s-1
Mean retrieved density relationship : (D)=0.01536D-0.835
g.cm-3 with D in cm
RadOn retrieval 14th April 2003
Altitude
Altitude
Mean retrieved density relationship : (D)=0.01536D-0.835
15/17
time
time
Density relationship for different
altitude intervals
A single relationship is not realistic in deep ice cloud
(aggregation)
=> Truncate Vt-Z relationship in altitude slices
14/04/04
16/17
a
b
3-5km
0.00606
-1.214
5-7km
0.012837 -0.86
7-9km
0.006788 -0.89
Very close to Locatelli and Hobbs
=0.00556(D in cm)-1.1
Aggregates of side planes columns
and bullets =0.00535(D in cm)-0.9
Future work



17/17
Adapt the RadOn input to use the
categorisation products
Run Radon on all the CloudNET sites
Investigate the /vt retrieval of density
relationship (advantage: density estimated at
each radar gate)