Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Multi-center, multi-model ensemble and its application
Yuejian Zhu Environmental Modeling Center
NCEP/NWS/NOAA
Acknowledgements: Ensemble team and Dr. B. Lapenta
Present for US THORPEX workshop September 20 2012
THORPEX related projects • International projects
– THORPEX-HEPEX – THORPEX-IPY – THORPEX-TPARC – THORPEX-YOTC – THORPEX-TIGGE/GIFS – THORPEX-NAEFS
• National projects – THORPEX-HYDRO – THORPEX (NCEP and ESRL – NOAA THORPEX)
• Ensemble data assimilation • Model error – physical uncertainty • Systematic error – bias correction – NAEFS products • Target observation – WSR
• THORPEX related workshop, regional committee meeting • Best reference:
http://www.emc.ncep.noaa.gov/gmb/ens/THORPEX.html
3
Research and Operational Applications
In Multi-Center Ensemble Forecasting
http://wwwt.emc.ncep.noaa.gov/gmb/ens/index.html
Yuejian Zhu
Acknowledgements: Zoltan Toth (ESRL), Glenn Rutledge (NCDC), Andre Methot (MSC),
Dan Collins, Bo Cui, Richard Wobus (NCEP)
North American Ensemble Forecast System
International project to produce operational multi-center ensemble products
• Combines global ensemble forecasts from Canada & USA
– 40 members per cycle, 2 cycles per day from MSC & NWS
• 6-hourly output frequency
• 1*1 degree resolution (~100km)
• NDGD resolution (~5km, CONUS only)
• Forecasts out to 16 days
• Generates products for
– Weather forecasters
• E.g., NCEP Service Centers (US NWS)
– Specialized users
• E.g., hydrologic applications in all three countries
– End users
• E.g., forecasts for public distribution in Canada (MSC) and Mexico (NMSM)
• Operational outlet for THORPEX research using TIGGE archive
– Prototype ensemble component of THORPEX Global Interactive Forecast System (GIFS)
5
NCEP CMC NAEFS
Model GFS GEM NCEP+CMC
Initial uncertainty ETR EnKF ETR + EnKF
Model
uncertainty/Stochastic
Yes (Stochastic Pert) Yes (multi-physics) Yes
Tropical storm Relocation None
Daily frequency 00,06,12 and 18UTC 00 and 12UTC 00 and 12UTC
Resolution T254L42 (d0-d8)~55km
T190L42 (d8-16)~70km
(d0-d16) ~ 66km 1*1 degree
Control Yes Yes Yes (2)
Ensemble members 20 for each cycle 20 for each cycle 40 for each cycle
Forecast length 16 days (384 hours) 16 days (384 hours) 16 days
Post-process Bias correction
(same bias for all
members)
Bias correction
for each member
Yes
Last implementation February 14th 2012 August 17th 2011
NAEFS Current Configuration
Updated: February 14th 2012
NH Anomaly Correlation for 500hPa Height Period: January 1st – December 31st 2010
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
An
om
aly
Co
rre
lati
on
Forecast (days)
GFS GEFS NAEFS
NAEFS – 10.2d
GEFS – 9.7d
0.6 skill line
GFS – 8.0d
Benefit for forecast: 1. Ensemble mean
will extend 1.7 days forecast ability
2. NAEFS will add additional 0.5 day forecast skill
3. Post process will add another additional
7
Purpose • Improve reliability while maintaining resolution in NWP forecasts
Reduce systematic errors (improve reliability) while
Not increasing random errors (maintaining resolution)
• Retain all useful information in NWP forecast
Methodology • Use bias-free estimators of systematic error
• Need methods with fast convergence using small sample
• Easy implementation for frequency upgraded forecast system
Approaches – Computational efficiency • Bias Correction : remove lead-time dependent bias on model grid
Working on coarser model grid allows use of more complex methods
Feedback on systematic errors to model development
• Downscaling: downscale bias-corrected forecast to finer grid
Further refinement/complexity added
• No dependence on lead time
NAEFS Statistical Post-Process (SPP)
Bias corrected NCEP/CMC GEFS and NCEP/GFS forecast (up to 180 hrs), same bias
correction algorithm
• Combine bias corrected NCEP/GFS and NCEP/GEFS ensemble forecasts
• Dual resolution ensemble approach for short lead time
• NCEP/GFS has higher weights at short lead time
NAEFS products
• Combine NCEP/GEFS (20m) and CMC/GEFS (20m), FNMOC ens. will be in soon
• Produce Ensemble mean, spread, mode, 10% 50%(median) and 90% probability
forecast at 1*1 degree resolution
• Climate anomaly (percentile) forecasts also generated for ens. mean
Statistical downscaling
• Use RTMA as reference - NDGD resolution (5km/6km), CONUS and Alaska
• Generate mean, mode, 10%, 50%(median) and 90% probability forecasts
Current NAEFS SPP System
8
9
Variables Domains Resolutions Total 10/8
Surface Pressure CONUS/Alaska 5km/6km 1/1
2-m temperature CONUS/Alaska 5km/6km 1/1
10-m U component CONUS/Alaska 5km/6km 1/1
10-m V component CONUS/Alaska 5km/6km 1/1
2-m maximum T CONUS/Alaska 5km/6km 1/1
2-m minimum T CONUS/Alaska 5km/6km 1/1
10-m wind speed CONUS/Alaska 5km/6km 1/1
10-m wind direction CONUS/Alaska 5km/6km 1/1
2-m dew-point T CONUS 5km 1/0
2-m relative humidity CONUS 5km 1/0
Total cloud cover?
NAEFS downscaling parameters and products Plan: Q4FY2012 (NDGD resolutions)
All downscaled products based on 1*1 (lat/lon) degree globally
Products include ensemble mean, spread, 10%, 50%, 90% and mode 9
12hr 2m Temperature
Forecast Mean Absolute
Error w.r.t RTMA for CONUS
Average for September, 2007
GEFS raw forecast
NAEFS forecast
GEFS bias-corr. & down scaling fcst.
11
NCEP/GEFS raw forecast
NAEFS final products
4+ days gain from NAEFS
From Bias correction (NCEP, CMC)
Dual-resolution (NCEP only)
Combination of NCEP and CMC
Down-scaling (NCEP, CMC)
11
12
From Bias correction (NCEP, CMC)
Dual-resolution (NCEP only)
Combination of NCEP and CMC
Down-scaling (NCEP, CMC)
NAEFS final products
NCEP/GEFS raw forecast
8+ days gain
12
Tmax Tmin
Temperature
Latest evaluation for CONUS
temperature forecast by apply :
1. Bias correction at 1*1 degree for
NCEP GFS/GEFS, CMC/GEFS
2. Hybrid bias corrected NCEP GFS
and GEFS
3. Apply statistical downscaling for all
bias corrected forecast
4. Combined all forecasts at 5*5 km
(NDGD) grid with adjustment -
NAEFS
Mean abs error Mean abs error
Mean abs error
3.5 days gain
13
Tmax Tmin
Temperature
Latest evaluation for CONUS
temperature forecast by apply :
1. Bias correction at 1*1 degree for
NCEP GFS/GEFS, CMC/GEFS
2. Hybrid bias corrected NCEP GFS
and GEFS
3. Apply statistical downscaling for all
bias corrected forecast
4. Combined all forecasts at 5*5 km
(NDGD) grid with adjustment -
NAEFS
CRPS CRPS
CRPS
7 days gain
14
Dew point T RH
RMS & Spread RMS & Spread
Dew Point T RH
CRPS
CRPS
15
NCEP CMC FNMOC
Model GFS GEM Global Spectrum
Initial uncertainty ETR EnKF (9) Banded ET
Model uncertainty
Stochastic
Yes (STTP) Yes (multi-physics) None
Tropical storm Relocation None None
Daily frequency 00,06,12 and 18UTC 00 and 12UTC 00 and 12UTC
Resolution T254L42 (d0-d8)~55km
T190L42 (d8-16)~70km
L40 ~ 66km T159L42 ~ 80km
Control Yes Yes No
Ensemble
members
20 for each cycle 20 for each cycle 20 for each cycle
Forecast length 16 days (384 hours) 16 days (384 hours) 16 days (384 hours)
Post-process Bias correction for
ensemble mean
Bias correction for
each member
Bias correction for
member mean
Last
implementation
February 14th 2012 August 17th 2011 September 14 2011
NAEFS/NUOPC Configuration
Updated: February 14 2012
17
PQPF forecasts from various global ensembles and combination
18
Raw NCEP
NAEFS + FNMOC Stat. corr.
NAEFS
Combined NCEP – CMC (NAEFS) show further increase in skill (6.2d)
Addition of FNMOC to NAEFS leads to modest improvement (6.7d)
Raw NCEP ensemble has modest skill (3.4d)
Statistically corrected NCEP ensemble has improved skill (4.8d)
0.5 CRPS skill
Value-added by including FNMOC ensemble into NAEFS T2m: Against analysis (NCEP’s evaluation)
Winter evaluations – first try
19
Verification against observation (NH) Raw ensemble forecast, Period: 11/17-12/21/2010
Courtesy of Normand Gagnon (CMC/MSC)
2-meter temperature
10-meter wind speed
T2m analysis difference accumulation (out to 20101026 ~ 10 days)
Challenge: large surface temperature analysis variation from center to center
Good - after adjustment Bad – after adjustment
Still not good – after adjustment
Perfect – don’t need adjustment
NCEP NCEP
NCEP
NCEP
FNMOC
FNMOC
FNMOC FNMOC
Example for combine current NAEFS and downscaled SREF
Future Plan for Improving NAEFS
Probabilistic Products
• Improving global probabilistic products – Adding FNMOC global ensemble
• NAEFS-LAM – Combine NCEP/SREF and CMC/REFS
• Extend additional variables and other regions – Pending on RTMA’s development
– Currently applied to CONUS and Alaska only
– Future: Hawaii, Guan and Puerto Rico
• Adding post processed regional ensemble for downscaling probabilistic products – SREF for Northern American products
• Improving exist bias correction method – Adjust decaying coefficients – varied by lead time, domain and
season
• Adjust 2nd moment – Possible to get help from BMA