Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
AATTMMOOSS
Evaluation of raindrop size distribution retrievals based on the Doppler spectra
using three beams
Remote-sensing of the environment (RSE)
Christine Unal
AATTMMOOSS In this talk:
� Differential reflectivity Zdr cannot be used (near-vertical profiling or light rain) Doppler spectra
� first evaluation of rain Drop Size Distribution (DSD) comparing
retrievals from the same radar resolution volume using two different
polarizations
DelftUniversity ofTechnology
2Remote-sensing of the environment (RSE)
polarizations
� second evaluation comparing DSD retrievals in different directions
during stratiform light rain
TARA radarInfluence of the radial wind on the DSD estimates
AATTMMOOSS
FM-CW Doppler-polarimetric S-band TARA radar
DelftUniversity ofTechnology
3Remote-sensing of the environment (RSE)
wind measurement available
AATTMMOOSS
Doppler spectra model
Based on
Moisseev, Chandrasekar, Unal and Russchenberg, 2006: Dual-
polarization spectral analysis for retrieval of effective raindrop shapes
Input
- Drop Size Distribution (DSD) :
� median volume diameter D0
Output
DelftUniversity ofTechnology
4Remote-sensing of the environment (RSE)
� intercept parameter Nw
� shape parameter µ
- Radial wind (v0)
- Spectral broadening (σ0)
AATTMMOOSS
Retrieval technique
Based on
Moisseev, Chandrasekar, Unal and Russchenberg, 2006: Dual-
polarization spectral analysis for retrieval of effective raindrop shapes
Non-linear optimization of D0 , µ , σ0
+ estimation of Nw , v0
DelftUniversity ofTechnology
5Remote-sensing of the environment (RSE)
+
=
AATTMMOOSS
Sensitivity analysis
Parameter D0 Nw µ σ0 v0
Region 0.2 – 3 mm 0 – 8000 mm-1 m-3 -2 – 10 0.1 – 0.9 m s-1 0 – 1.2 m s-1
RMSD 0.12 mm 1350 mm-1 m-3 0.67 0.04 m s-1 0.18 m s-1
CV(RMSD) 17% 54% 8.4% 28%
Simulation results on averaged Doppler spectra (10-55 dBZ reflectivity, average:30)
DelftUniversity ofTechnology
6Remote-sensing of the environment (RSE)
CV(RMSD) 17% 54% 8.4% 28%
Parameter Z LWC Nt
RMSD 0.30 dB 0.13 g m-3 142 m-3
CV(RMSD) 0.91% 22% 7.4%
AATTMMOOSS
Multi-beam retrieval example (non-averaged Doppler spectra)
Vertical beam75 deg 75 deg 69 deg
DelftUniversity ofTechnology
7Remote-sensing of the environment (RSE)
AATTMMOOSS
Multi-beam retrieval profile example
Vertical
Median volume diameter D0 profile in 3 different looking directions
Retrieval procedure + height-smoothing
DelftUniversity ofTechnology
8Remote-sensing of the environment (RSE)
Vertical beam
AATTMMOOSS
Retrievals comparison (light rain)
HH and VV non averaged Doppler spectra
D0 Nw µ v0
DelftUniversity ofTechnology
9Remote-sensing of the environment (RSE)
σ0 Z
LWC
Nt
AATTMMOOSS
Dynamical retrieval v0
Comparison between retrieval of radial wind (-v0) and radial component of mean horizontal wind measured by TARA
DelftUniversity ofTechnology
10Remote-sensing of the environment (RSE)
AATTMMOOSS
Influence of the error on v0 on the DSD retrieval
Doppler spectra are shifted using TARA mean
DelftUniversity ofTechnology
11Remote-sensing of the environment (RSE)
Doppler spectra are shifted using TARA mean horizontal wind measurement before retrieval
AATTMMOOSS Influence of the error on v0 on the median volume diameter (D0)
Mean horizontal wind correction before retrieval1600 m
1400 m
DelftUniversity ofTechnology
12Remote-sensing of the environment (RSE)
800 m
400 m
AATTMMOOSS
Conclusions
� development of an automatic rain retrieval technique based on the complete Doppler spectrum (from drizzle to heavy rain)
� a tool for study cases
� consistency of the retrievals (D0, µ, v0, σ0, LWC) representing the same radar resolution volume
DelftUniversity ofTechnology
13Remote-sensing of the environment (RSE)
� when Zdr is too small (light precipitation, near-vertical profiling), it looks necessary to input the radial component of the mean horizontal wind to reduce the errors on (D0, Nw, µ)
� adding two other looking directions (beams) to estimate the contribution of the radial wind may solve this problem
AATTMMOOSS
Moisseev, Chandrasekar, Unal and Russchenberg, 2006: Dual-polarization spectral analysis for retrieval of effective raindrop shapes
DelftUniversity ofTechnology
14Remote-sensing of the environment (RSE)
AATTMMOOSS
Do
σ0
µ
+
+
Do min Do max
σo min σo max
µ maxµ min
L
L
L
Fitting
Do
σ0
µ
�Ite
rativ
e se
lect
ion
proc
edur
e
�O
ptim
izat
ion
pro
cedu
refr
om c
ost f
unct
ion
stud
y (L
)
� Final outcome( ) ( )
max
min
2mod0 , , 0( ) ,
vmeasHH dB HH dB
v v
L D sZ v sZ v D=
= − ∑
( )max
0min
2mod
0 , , 0( ) ( , )v
measHH dB HH dB D
v v
L sZ v sZ vσ σ=
= − ∑
DelftUniversity ofTechnology
15Remote-sensing of the environment (RSE)
Nw
V0
+
measured spectrum
Fitting
Model spectrum
�es
timat
ion
proc
edur
e
� Output of the model
Max cross-correlation
Variable projection method (Rust, 2003)
Error calculationNon linear least-square algorithm
Estimation from Do, σ0 and µ
+ ( ) ( )max
0 0min
2mod
, , ,( ) ,
vmeasHH dB HH dB D
v v
L sZ v sZ vσ
µ µ=
= − ∑
AATTMMOOSS
Retrieval outputs: dsd, v0 and σσσσ0
� median volume diameter D0
� µ bounded [-1 5]
� spectral broadening σ0 bounded [0 1] m s-1
�intercept parameter Nw free
� radial wind v0 free
DelftUniversity ofTechnology
16Remote-sensing of the environment (RSE)
crucial to get the actual D0
help
Polarimetry with Zdr
Wind estimates with v0
Reflectivity
AATTMMOOSS
Multi-beam raindrop size comparison (Z-D0)
Mean horizontal wind correction before retrieval1600 m
1400 m
DelftUniversity ofTechnology
17Remote-sensing of the environment (RSE)
800 m
400 m
AATTMMOOSS
Multi-beam raindrop size comparison (Nw)
Mean horizontal wind correction before retrieval1600 m
1400 m
DelftUniversity ofTechnology
18Remote-sensing of the environment (RSE)
800 m
400 m
AATTMMOOSS
Multi-beam raindrop size comparison (µµµµ)
Mean horizontal wind correction before retrieval
1600 m
1400 m
DelftUniversity ofTechnology
19Remote-sensing of the environment (RSE)
800 m
400 m