Experiences with 0-36 h Explicit Convective Forecasting
with the WRF-ARW Model
Morris Weisman
(Wei Wang, Chris Davis)
NCAR/MMM
WSN05
September 8, 2005
Composite NEXRAD Radar 00 UTC 6/10/03 - 12 UTC 6/11/03
WRF Realtime Convective Forecasting
May 1 – July 31 4 km 00 UTC -- 36h
2003, 2004, 2005
WRF Real-time Forecasts: 2004, 2005
4-km from 0000 UTC - 36 h
Version 1.3 (2.0.3.1)
Eta initial and boundary conditions (40 km)
Physics:
Lin et al. (5 cat.) microphysics (WSM6)
YSU PBL (first-order closure)
Noah LSM (HRLDAS)
2000 km X 2000 km domain / 2800 km X 2600 km domain
5.0h (6.5h) on 128 (192) IBM Power-4 processors
Real-time WRF 4 km Forecast
Composite NEXRAD RadarReflectivity forecast
Initialized 00 UTC 9 June 03
Real-time WRF 4 km Forecast
Initialized 00 UTC 10 June 03
Reflectivity forecast Composite NEXRAD Radar
Consider the 24-30 h forecast challenge (i.e., the next diurnal cycle).
How does the 4 km WRF-ARW ‘’guidance’’ compare to the 12 km operational ETA?
Radar 03 GMT
May 30, 2004
03 GMT 27 hr 06 GMT prec. 30 hr4 km WRF-ARW
12 km ETA
Overall Forecast
Good:
OK:
Bad:
WRF RADAR
How well does 4-km WRF represent the climatological behavior of convection? (e.g., diurnal cycle, episodes)
Longitudinal 1 hr Precip. May 10-31, 2004
Stage IV WRF
Diurnal Average Frequency: May 10-July 31 2004
Stage IV 4 km WRF
12 km ETA4 km WRF
Stage IV
Diurnal Average 3 hr Precip.
Initialized 04 June 2005 00 UTC
Reflectivity forecast Composite NEXRAD Radar
24 h forecast Valid 00 UTC 06/05/05
4-km ARW
2-km ARW
4-km NMM
Nexrad
ARW2 BREF
NMM4ARW4
0500 UTC 29 April 2005: 1 km model reflectivity, NEXRAD BREF
Progress
4 km WRF-ARW simulations exhibit:
A surprising ability to forecast mesoscale convective systems (MCS) out to 36 h ( strongly
controlled by 12-km ETA forcing)
A demonstrated skill at depicting MCS mode (bow echoes, mesoscale convective vortices, supercell lines)
A more accurate depiction of diurnal cycle and precipitation episodes
An ability to spin-up convective systems within 3-4 h from a cold start.
• QPF problematic (too much convective precip)• Stratiform regions appear too small (microphysics?)• Convective systems often fail to decay (BL
evolution?)• Initialization (data assimilation)• Verification methods
Challenge:
WRF-ARW 2005 real-time forecasts can be found at:
http://box.mmm.ucar.edu/projects/wrf_spring/
http://rain.mmm.ucar.edu/mm5/
Also archived at:
http://www.joss.ucar.edu/wrf-2004/catalog/http://www.joss.ucar.edu/wrf-2005/catalog/
24 hr precipitation verification
Valid 6/10/03 12 UTC
OBS
10 km WRF4 km WRF 12 km ETA
4 km WRF 12 h Precipitation Forecast
NCEP Stage 4 Data
Reisner Scheme SB2004 Scheme
WSM-6 SchemeLin, et al. Scheme
Valid 10 June 2003 12 Z
Total Precipitation (mm)
(Axel Seifert, 2004)
4 km WRF 12 h Reflectivity Forecast
NEXRAD Composite
Reisner Scheme SB2004 Scheme
WSM-6 SchemeLin, et al. Scheme
Valid 10 June 2003 12 Z
Column Max Reflectivity (dBZ )
(Axel Seifert, 2004)
Reisner
Lin WSM-6
SB2004
12 h Surface Theta
Sounding comparison: 24h forecast valid 00Z 24 May at DDC
ARW4
NMM4
Good forecast…
PBL too shallow, cold, & moist…clouds just broke up!
Model Raob
Model Raob
Sounding comparison: 24h forecast valid 00Z 18 May at DDC
ARW4
NMM4
Too dry in PBL, too moist above; Where is the PBL top?
Good forecast…
Model Raob
Model Raob
Sounding comparison: 24h forecast valid 00Z 28 April at OUN
ARW4
NMM4
Good in PBL, but CIN layer is washed out
PBL too shallow and moist, but CIN layer looks good
Model
Raob
Model
Raob
24 h Reflectivity Forecast valid 4-30-05 00Z
Radar
4 km WRF-ARW
2 km WRF-ARW
4 km WRF-NMM
Diurnal Average (2003)
4 km 10 km
1 10 100 km
Cumulus ParameterizationResolved Convection
LES PBL Parameterization
Two Stream Radiation3-D Radiation
Model Physics in High Resolution NWP
Physics“No Man’s Land”
CAT 3: severe, well organized (> 1 in)
CAT 2: some organization (.25-1.0 in)
CAT 1: weak, disorganized (< .25 in)
Convective Mode/Intensity WRF ETA