NCEP Regional Ensemble Status and Plans

1DET Mesoscale Ensemble Workshop 18-19 August 2010

NCEP Regional Ensemble Status and Plans

Geoff DiMego and Jun Du

NOAA/NWS/NCEPEnvironmental Modeling Center

NCEP


Presentation Outline

• Status

• Plans

• Challenges


Regional Ensemble Prediction Systems (EPSs) at NCEP

EPS Model (s) Domain (s) Member Resolution Forecast length

Update frequency

Intended purposes

SREF (2011)

NEMS(nmm), WRF(nmm, arw)

North America (NA)

21 32-35km 3.5 days 6hrly Short-range weather

HREF* (dyn dwn) (2011)

WRF(nmm, arw)

East CONUS

West CONUS

Alaska

44 4-5km 2 days 12hrly

24hrly

24hrly

high impact weather

VSREF* (t-lag) (2011)

RR(arw), NAM(nmm)

CONUS 11 12-13km 0.5 day hourly aviation

NARRE (2012-13)

NEMS(nmm, arw)

NA 6 10km 1 day hourly aviation

HRRRE (2014-15)

NEMS(nmm, arw)

CONUS, AK 6 3km 1 day hourly Convective storms

NAEFS_LAM (2013)

NCEP SREF + CMC REPS

NA 41 ~30km 2 days 12hrly Short-range weather

HWRF ens (2015?)

HWRF Atlantic & E. Pacific

21 27-9-3km 5 days 6hrly hurricane

* See poster


IC aspect:• Multi-analysis (NDAS, RR, GDAS)• Perturb ctl anl (bred vector, global ETR, regional ETR)• Perturbed LBCs (from global EPS)• Land surface initial states such as soil moisture and

temperature (tested)

Model aspect:• Multi-model (NEMS-NMMB,WRF-ARW, WRF-NMM)• Multi-physics (various convection and cloud schemes)• Stochastic parameterization (tested)

Residual Part:• Post processing including:

bias correction downscaling (HREF)

Current SREF Methodology


5

Convergence of NAM & RUC into hourly NARRE & HRRRE

• Based on a new common modeling infrastructure called NEMS (NOAA Environmental Modeling System)

• 84 hour guidance from NAM & SREF will continue as extensions to longer range of the 00z, 06z, 12z and 18z cycle runs


6

CURRENT2009-2010

NAM• WRF-NMM (Egrid)• GSI analysis• 4/Day = 6 hr update• Forecasts to 84 hours• 12 km horizontal• 60 layers with 2 mb top• 12 hr pre-forecast

assimilation period with 3hr updates (catch-up)

RUC• Non-WRF RUC model• RUC 3DVAR analysis• 24/Day = hourly update• Forecasts to 18 hours• 13 km horizontal• 50 layers with 50 mb top• Continuous forward cycle with

no pre-forecast assimilation period (no catch-up)

RUC-13 CONUS domain


7

2011NAM

• NEMS based NMMB• Bgrid replaces Egrid• Parent remains at 12 km• Multiple Nests Run to 60hr

– 4 km CONUS nest– 6 km Alaska nest– 3 km HI & PR nests

• Reinstate Fire Weather/IMET Support/DHS run to 36hr– Locate single 1.20-1.33 km1.20-1.33 km run– In either CONUS or Alaska

Rapid Refresh• WRF-based ARW• NCEP’s GSI analysis• Expanded 13 km Domain

to include Alaska• Experimental 3 km HRRR

RUC-13 CONUS domain

WRF-Rapid Refresh domain – 2010

Original CONUS domain

Experimental 3 km HRRR


8

2012-2013North American Rapid Refresh

ENSEMBLE (NARRE)• NMMB (from NCEP) & ARW (from ESRL) dynamic

cores• Common use of NEMS infrastructure and GSI analysis• Common NAM parent domain at 10-12 km• Initially ~6 member ensemble made up of equal

numbers of NMMB- & ARW-based configurations• Hourly updated with forecasts to 24 hours• NMMB & ARW control data assimilation cycles with 3

hour pre-forecast period (catch-up) with hourly updating• NAM & SREF 84 hr forecasts are extensions of the

00z, 06z, 12z, & 18z runs – for continuity sake.– SREF will be at same 10-12 km resolution as NARRE by then– SREF will have 21 members plus 6 from NARRE for total of 27

• NARRE requires an increase in current HPCC funding


9

2012-2013High Resolution Rapid Refresh ENSEMBLE

(HRRRE)• Each member of NARRE contains

– 3 km CONUS, 4 km Alaskan, 2.5km Hawaii & Puerto Rico nests– The two control runs initialized with radar data & other hi res obs

• This capability puts NWS/NCEP[+OAR/ESRL] in a position to – Provide NextGen Enroute AND Terminal guidance– Provide PROBABILITY guidance with full Probability Density

Function specified– Provide a vehicle to improve assimilation capabilities using hybrid

(EnsKF+4DVar) technique with current & future radar & satellite– Address Warn-on-Forecast as resolutions evolve towards ~1 km

• NAM nests are extensions of the 00z, 06z, 12z & 18Z runs.

• HRRRE requires an increase in current HPCC funding over and above that required for the NARRE


10

2012-2015A Catch-Up Cycle for NARRE & HRRRE could constitute the Analysis of Record

• Catch-up concept: reach back in time to include late arriving observations

• Assimilate ALL in situ and remote data sources • Use state-of-the-art 4-dimensional data assimilation technique

– Likely NCEP’s hybrid Ensemble Kalman-Filter and 3D-/4D-Variational– Able to take quick advantage of its evolution

• Use state-of-the-art nonhydrostatic numerical models– Advanced Research WRF (ARW) core from NCAR & ESRL/GSD– Non-hydrostatic Multiscale Model on B-Grid (NMMB) from NCEP– Interoperable physics from WRF community & NCEP operations– Able to take quick advantage of their evolution

• Extend to include NextGen required parameters• This AoR requires an increase in current HPCC funding


SREF Future Plans2010 2011 2012 2013 2014 2015

SREF SREF SREF SREF SREF SREF*(21mem, 32km, 4-model) (21mem,22km, 3modelsWRF&NEMS) (12km, NEMS-only + CMC) (NARRE extension)

NARRE NARRE NARRE (6mem, 12km, hourly update, 24hr, 2-model)

HRRRE HRRRE HRRRE (CONUS, AK, 3km, nests in each NARRE member, hourly update to 24hr)

VSREF(RUC/NAM) VSREF(=NARRE) after (time-lagged ensemble based probabilistic products for aviation NextGen)

HREF(SREF/HiResWinRuns) HREF(=HRRRE) after (dynamically downscaled from SREF using HiResWindow runs for high impact weather)*NEMS = NOAA Environmental Modeling System (a unified modeling framework)

*SREF (3222km12km, 6-hrly update to 84h for general weather forecasts) *NARRE = North America Rapid Refresh Ens (10-12km, hrly update to 24h for aviation)*HRRRE = Hi-Res Rapid Refresh Ens. (3km, nested in NARRE, hrly update to 24h for high-impact events for targeted CONUS, AK domains)*VSREF = Very Short-Range Ens Forecast for aviation NextGen prob products*HREF = High-Resolution Ensemble Forecast


SREF Future Plans (Cont.)

• IC perturbation – Regional ETR (with DET), comparing to downscaled GEFS

ETR– any other better methodologies (with DET)

• Model perturbation– Stochastic parameterization – GEFS stochastic physics– Any other better ones (with DET)

• Post processing– CDF-adjusting for precipitation bias correction – NAFES post-processing package (Bayesian processor)– Any other better ones (with DET)

• Products – Any thing good from community


Challenges

• Day-1 ensembling techniques

• How can stochastic physics reproduce the effectiveness of multi-model approach?

• Lack of enough computer resource

• Lack of enough human resource


Probabilistic Forecast Performance (Ranked Prob Skill Score): little or no skill comparing to 12km NAM at day 1


Effectiveness of multi-model approach: can it

be reproduced by stochastic physics?


Computer Resource Reality

• Stochastic -> Single executable limits flexibility

• Projected is Problemmatic– Acquisition for 2012 is postponed/delayed

– Funding for the bridge and transition in doubt

– Could be left with no net increase next 4 years

• Nothing close to what is needed for a truly MESOSCALE ensemble


Key to the success of R2O and O2R

• Community needs to understand– operational needs– restrictions of an operational environment

• Establish SREF baseline at DET

• R2O tests new methods in context of this baseline for potential implementation

• Agreement on standard verification measures


Questions?

NCEP Regional Ensemble Status and Plans

Documents

Transcript of NCEP Regional Ensemble Status and Plans