The TRIMREX Field Project:
An OSSE Study
Shu-Hua Chen /UC Davis
Other contributors:
Jhih-Ying Chen (NCU), Wei-Yu Chang (NCU),
Wen-Yih Sun (NTFRI), Pay-Liam Lin (NCU),
Po-Hsiung Lin (NTU), Tai-Chi Chen (NCU),
Yu-Chieng Liou (NCU)
This work is supported by NTFRI and NSC in Taiwan
Introduction
Motivation
 Where and when should we use additional instruments
to obtain better analysis and forecasts ?
• Location of the TEAM-R?
• Location of ISS wind profiler?
• Frequency of radiosondes?
• Pattern and frequency of dropsondes?
 Use OSSE to investigate better use of additional
instruments during TIMREX.
Introduction
OSSE
Observing System Simulation Experiments
Def: An observing system experiment in which synthetic
meteorological observations are used as a surrogate for real
observations. The assimilation of these observations and
subsequent free forecasts are conducted the same as with
real observations. The synthetic observation are usually
extracted from a long forecast from a high resolution, stateof-the art model, known as a “nature run”, which acts as a
proxy atmosphere specifically for the OSSE.
(from http://sivo.gsfc.nasa.gov/OSSE)
Introduction
OSSE Flow Chart
Nature run: High-resolution (1km) WRF simulations
Simulate existing & additional observations
WRF Var first guess (NCEP reanalysis)
Assimilate simulated observations
Perform WRF runs and evaluate simulated observations
Applied to TIMREX project
Simulated observations
Nature Run – Configuration
D01
Domain
Grid points
Resolution
1
141 x 121 x 31
27 km
2
331 x 271 x 31
9 km
3
721 x 541 x 31
3 km
4
901 x 961 x 31
1 km
D02
D03
D04
Simulated observations
Nature Run – Physics Schemes
WRF integrates from 00Z June 6 to 12Z June 9, 2003
Radiation
Cumulus
PBL
Microphysics
RRTM long wave
GSFC short wave
Grell-Devenyi ensemble
(domains 1 and 2 only)
MYJ TKE
Purdue
Simulated observations
Nature Run – Simulated Radar Echo
Z = 3 km, June 06/00Z – 09/12Z, 2003, interval – 6h
Simulated observations
Nature Run – Simulated Radar Echo
WRF domain 4 simulation with the 1 km resolution from 18Z June
6 to 00Z June 7, 2003 was use to simulate existing and
additional observations. These observations will then be
assimilated in WRF numerical experiments for evaluation.
Z = 3 km, June 06/18Z, 2003
Z = 3 km, June 07/00Z, 2003
Simulated observations
List of Simulated Observations
Existing observations:
Radar radial velocity
Radiosondes
Surface stations
EPS wind profiler
Additional observations:
S-POL radar radial velocity
TEAM-R radial velocity
NCU ISS wind profiler
Radiosondes
Dropsondes
Simulated observations
Simulated Radar Data Strategy
Scanning angles : 0.5, 1.5, 2.4, 3.3, 4.3, 6.0, 9.9, 14.6, 19.5 degree
Existing radar & SPOL
TEAM-R
Maximum distance
~ 250 km
Maximum distance
~ 40 km
Gate resolution
– 1 km
Gate resolution
– 1 km
Beam resolution
– 1 degree
Beam resolution
– 1 degree
Three potential sites :
22.75 lat, 120.47 long
22.88 lat, 120.64 long
22.88 lat, 120.21 long
Simulated observations
Simulated Doppler Radar Locations
Potential position for TEAM-R
Position for S-POL
Existing radar
Simulated observations
Simulated Wind Profiler Winds
Vr1
z
Vr3
w
β
α
 Vr1  w

 Vr 2  u  sin α  w cos α
 V   v  sin β  w cos β
 r3
Vr2
v
y
x
u
 w  Vr1

 u  ( Vr 2  Vr1 cos α ) / sin α
 v   ( V  V cos β ) / sin β
r3
r1

Simulated observations
Simulated Wind Profiler Locations
ISS wind profiler
EPA Wind profiler
Simulated observations
Simulated Radiosondes
Radiosonde track
wb
w z
tf
S   u 2  v 2  ( w  w b )2 dt
t0
v
 Data taken every 20 Seconds
 Wb = 5 m/s
y
u
x
Simulated observations
Simulated Radiosonde Locations
Regular sites:
Ban-Chiao
Hua-Lian
Ma-Kung
Green-Island
Dong-Sa
Additional sites:
Tai-Chung
Tai-Nan
Peng-Dong
Heng-Chun
ISS
Ship
Simulated observations
Simulated Surface Station Locations
33 CWB surface stations
Simulated observations
Simulated Dropsondes
w z
Dropsonde track
v
tf
S   u 2  v 2  ( w  w d )2 dt
y
u
t0
x
Wd
-wd
Hock and Frnaklin (BAMS, 1999)
Simulated observations
Simulated Flight Patters
Simulated observations
Simulated Flight Patters
Experiment design
Numerical Experiments - Configuration
D01
D02
Domain
Grid points
Resolution
1
133 x 115 x 31
27 km
2
241 x 181 x 31
9 km
3
271 x 250 x 31
3 km
D03
Same physics were chosen as the nature run
Experiment design
Numerical Experiments
WRF Var cycling
WRF integration
6h
-6
-5
-4
-3
1.5 days
-2
-1
0
36
Time (h)
Obs Obs Obs Obs Obs Obs Obs
Case
WRF Var data cycling
WRF simulation
2003
18Z/06 – 00Z/07
00Z/07 – 12Z/08
Experiment design
Temperature at the 1st Half Model Level
18Z/06/06/2003
Nature run (18h simulation)
First Guess (NCEP reanalysis)
Experiment design
Moisture at the 1st Half Model Level
18Z/06/06/2003
Nature run (18h simulation)
First Guess (NCEP reanalysis)
Experiment design
Wind Speed at the 1st Half Model Level
18Z/06/06/2003
Nature run (18h simulation)
First Guess (NCEP reanalysis)
Experiment design
Numerical Experiment Design
EXPERIMENT
Assimilated OBs
BEST_NORAD
3 hourly radiosondes (existing + additional)
Hourly surface stations
Hourly Wind profiler
NONE
No observation
Experiment design
Temperature at the 1st Half Model Level
00Z/06/07/2003 (After WRF Var cycling)
Nature run
BEST_NORAD
NONE
Experiment design
Moisture at the 1st Half Model Level
00Z/06/07/2003 (After WRF Var cycling)
Nature run
BEST_NORAD
NONE
Experiment design
Wind Speed at the 1st Half Model Level
00Z/06/07/2003 (After WRF Var cycling)
Nature run
BEST_NORAD
NONE
Experiment design
On Going
 Still work on the assimilation of simulated radar winds
(close to finish)
 Still work on the assimilation of dropsondes
(close to finish)
 Keep evaluating the OSSE system and hopefully will
finish before the TIMREX starts.