Retrieval of Cloud Liquid from GroundBased

1

• Retrieval of Cloud Liquid from Ground-Based
• Remote Sensing Measurements During CLIWA-NET
• Ulrich Löhnert, Susanne Crewell, Clemens Simmer
• Meteorologisches Institut, Universität Bonn,
  Germany
• a) CLIWA-NET Aims, Instrumentation, Data
  Availability
• b) MICCY a state-of-the-art MWR
• c) Liquid Water Path Retrieval
• d) Liquid Water Profile Retrieval

2
EU-Project CLIWA-NET
BALTEX Cloud Liquid Water Network Prototype of a
european cloud observation network 1.)
Development of integrated LWP retrievals 2.)
Validation for satellite retrievals 3.)
Evaluation and improvement of cloud
parameterizations within NWP and climate
models 4.) Design of a low-cost MWR CNN I
August/September 2000 CNN II April/Mai 2001 BBC
August/September 2001
ETL Seminar, Dec. 10, 2001, CLIWANET
3
ETL Seminar, Dec. 10, 2001, CLIWANET
4
Measurement Example
23.8 GHz31.4 GHz

ETL Seminar, Dec. 10, 2001, CLIWANET
5
Microwave Intercomparison Campaign
August 1-14, 2001 in Cabauw, The Netherlands
U. Bern, SwitzerlandChalmers U., Sweden CETP
Velizy, FranceUK Met OfficeU. Bonn,
GermanyGerman Weather ServiceGerman Weather
Service Inst. Radioeng., Russia
ETL Seminar, Dec. 10, 2001, CLIWANET

6
MICAM Impressions
7
CLIWA-NET Contact

• Hompage
• http//www.knmi.nl/samenw/cliwa-net/
• Data BBC campaign
• contact bloemink_at_knmi.nl
• Data CNN I/II campaigns
• contact screwell_at_uni-bonn.de
• uloeh_at_uni-bonn.de

ETL Seminar, Dec. 10, 2001, CLIWA-NET
8
MIcrowave radiometer forCloud CarthographY
MICCY
Crewell et al., Radio Science, Vol. 36,
July/Aug. 2001
ETL Seminar, Dec. 10, 2001, MICCY
9
MICCY-Design
Cassegrain main reflector
ambient load
power supply
scanning mirror
cold load
radiometerbox
ETL Seminar, Dec. 10, 2001, MICCY
10
Instrument Description

• Simultaneous measurements in 22 channels
• 3 frequencies with both orthogonal polarizations
• High spacial (1?) and temporal resolution (1 s)
• Azimuth (0-360?) and elevation controll (0-360?)
• Size 3.1 m x 1.5 m x 2.7 m weight 2000 kg
• Infrared radiometer und video mounted on antenna
  axis

Calibration hot/cold 4-point calibration,
tipping curve, periodic gain-correction
ETL Seminar, Dec. 10, 2001, MICCY
11
Tropospheric Parameter

• Water vapor column and profile
• Temperature profile
• LWP and LWC profile (combination)

ETL Seminar, Dec. 10, 2001, MICCY
12
LWP Algorithm Development
Investigated Aspects

• LWP accuracy using different frequency
  combinations, quadratic terms, opt. thickness
• Accuracy assesment depending on cloud model
• Effects of different gas absorption models
• Algorithm validation during cloud free cases
• Different assumptions about radiometer accuracy

ETL Seminar, Dec. 10, 2001, LWP
13
LWP Improvement with Profiler
Start with 22.985 GHz channel combine to
decrease LWP/IWV RMS
without 90 GHz
ETL Seminar, Dec. 10, 2001, LWP
14
Evaluation for Cloud Free Cases
ETL Seminar, Dec. 10, 2001, LWP
15
Results LWP Retrieval

• 2 channel addition (22.985 28.235) (50.8 90
  GHz) - 2 channels explain 88 of LWP
  variance - 4 channels explain 98
  
• More than 4 frequencies will not significantly
  improve LWP retrieval (MICCY frequencies)
• 50.8 GHz channel is very sensitive to statistics
  of data set possible bias errors (rather use
  90 GHz!)
• Ceilometer/IR radiometer as quality control
• Liebe 93 model best for 90 GHz
• NN can help improve cloud free and low LWP cases
  in combination with Ceilometer/IR

ETL Seminar, Dec. 10, 2001, LWP
16
Liquid Water Profiles Through Sensor Synergy

• 95 GHz cloud radar MIRACLE (GKSS)
• - Backscatter of cloud droplets D6
• - Problems domination of larger drops,
• attenuation

MWR MICCY - Emission of cloud droplets D3 -
Problems vertical resolution, gas absorption
Laser Ceilometer - Backscatter due to Aerosol
and cloud droplets D2 - Problems conversion
of backscatter to extinction, strong
absorption within clouds
ETL Seminar, Dec. 10, 2001, LWC
17
Usable Information
standard meteorological obs. (ground)
19 brightness temperatures (TB)
reflectivity profiles (dBZ)
Optimal Estimation Algorithm (OE) for LWC profile
ceilometer cloud base
virtual measurements (backround, a priori)
ETL Seminar, Dec. 10, 2001, LWC
18
Algorithm Development

• Singular column cloud model
• (1.5 dimensional)
• initiated with real radiosonde data
• from Essen, Germany
• prediction of temperature (T), humidity
• (q) and liquid drop size spectra of LWC

develop dBZ- LWC relations
simulation of dBZ profiles, including path
attenuation
create a priori climatology of T, q, LWC
simulation of TB, without scattering
ETL Seminar, Dec. 10, 2001, LWC
19
Theory Optimal Estimation
x atmospheric profile y measurement vector F
forward model K Jacobian xa a priori
profile SE error covariance of measurement Sa
covariance of a priori

• combination of statistically
• independent measurements
• find maximum probability state
• according to Bayesian solution
• of the problem

moderately non-linear problem
iterative approach
ETL Seminar, Dec. 10, 2001, LWC
20
4 Different Algorithms
RADAR
MRAD

• OE with 19 TBs a priori
• y(TB)
• F radiative transfer
• operator
• cloud boundaries known

• Z-LWC power law (PL)
• ZaLWCb
• dependent on height
• above cloud base

LWP1-LWP5 (Löhnert et al., 2001)
COMB

• OE with LWP a priori
• dBZ information
• y(LWP, dBZ)
• F PL SLWCLWP
• different hypothetical
• LWP accuracies

• OE with 19 TBs a priori
• dBZ information
• y(TB, dBZ)
• F radiative transport
• operator PL

ETL Seminar, Dec. 10, 2001, LWC
21
Results I LWC accuracy
ETL Seminar, Dec. 10, 2001, LWC
22
Results II LWC accuracy
LWP1 dLWP0.01 kg m-3 LWP2 dLWP0.03 kg
m-3 LWP3 dLWP0.05 kg m-3 LWP4 dLWP0.07 kg
m-3 LWP5 dLWP0.09 kg m-3
ETL Seminar, Dec. 10, 2001, LWC
23
Results III T accuracy
ETL Seminar, Dec. 10, 2001, LWC
24
Results IV q accuracy
ETL Seminar, Dec. 10, 2001, LWC
25
Results V Retrieval Consistency (COMB)
ETL Seminar, Dec. 10, 2001, LWC
26
Real Data I (CLIWANET BBC CAMPAIGN, Aug.-Sept.
2001)
ETL Seminar, Dec. 10, 2001, LWC
27
Real Data II (CLIWANET BBC CAMPAIGN, Aug.-Sept.
2001)
ETL Seminar, Dec. 10, 2001, LWC
28
Real Data III (CLIWANET BBC CAMPAIGN, Aug.-Sept.
2001)
ETL Seminar, Dec. 10, 2001, LWC
29
Results Perspectives

• COMB algorithm allows physically consistent
  derivation of
• T, q and LWC profiles
• COMB accuracies comparable to LWP1 accuracies
• (ideal LWP retrieval)
• ground information can improve q retrieval up to
  1 km
• sensitivity studies to different cloud
  statistics
• (real data, LES models)
• - long term studies with CLIWA-NET data

ETL Seminar, Dec. 10, 2001, LWC
30
Comparing LWC Algorithms at ETL (Nov. 01 Feb.
02)

• using SHEBA data (May 1998 Aug, 1998),
  consisting
• of dual channel LWP retrievals and dBZ
  profiles
• compare method according to Frisch et al. (1995,
  1998)
• with optimal estimation method (LWP/dBZ
  combination)
• use LES model data (Jiang et. al, 2001) from
• May 18, 1998 as a priori data
• evaluate differences in LWC retrievals and
  search (and
• explain) systematic differences

ETL Seminar, Dec. 10, 2001, LWC