REGIONAL-SCALE ENSEMBLE REGIONAL-SCALE ENSEMBLE FORECASTS OF THE FORECASTS OF THE 7 FEBRUARY 2007 7 FEBRUARY 2007
LAKE EFFECT SNOW EVENT LAKE EFFECT SNOW EVENT
Justin Arnott and Michael EvansJustin Arnott and Michael EvansNOAA/NWS Binghamton, NYNOAA/NWS Binghamton, NY
Richard GrummRichard GrummNOAA/NWS State College, PANOAA/NWS State College, PA
George YoungGeorge YoungPenn State University, University Park, PAPenn State University, University Park, PA
NROW IX, 7-8 November 2007NROW IX, 7-8 November 2007
MotivationMotivationPast LES ForecastingPast LES Forecasting
LES a Pattern Recognition LES a Pattern Recognition ProblemProblem
• GFS unable to resolve GFS unable to resolve bandsbands
• Rely on tools such as Rely on tools such as BUFKITBUFKIT
Motivation, continuedMotivation, continued
• 12 km NAM grossly 12 km NAM grossly resolves lake-parallel bandsresolves lake-parallel bands– Each NWS office can run a Each NWS office can run a
local version of this modellocal version of this model
• Individual runs often have Individual runs often have problems with band problems with band location/orientationlocation/orientation
• Can multiple simulations of Can multiple simulations of the NAM (an ensemble) the NAM (an ensemble) provide added value?provide added value?– This question has This question has
prompted the development prompted the development of the Northeast Regional of the Northeast Regional EnsembleEnsemble
Present/Future LES ForecastingPresent/Future LES Forecasting
What is the Northeast Regional What is the Northeast Regional Ensemble?Ensemble?
• 12 km Workstation WRF12 km Workstation WRF– 24-36 hr run length24-36 hr run length
• 2007-2008: 7-8 2007-2008: 7-8 MembersMembers– 2 CTP members2 CTP members– 1 Operational1 Operational
• Goal: Improve operational forecasts of lake effect Goal: Improve operational forecasts of lake effect snowfallsnowfall
Case Day: 07FEB2007Case Day: 07FEB2007
• Part of a ~10-day prolific lake effect Part of a ~10-day prolific lake effect snow event east of Lake Ontariosnow event east of Lake Ontario
• Band moved significantly throughout Band moved significantly throughout the daythe day– Excellent test for the ensembleExcellent test for the ensemble
The Ensemble – 07FEB2007The Ensemble – 07FEB2007
OfficeOffice CorCoree
IC/IC/BCsBCs
MicroMicro CPSCPS #Z Lev#Z Lev
OperationOperationalal
NMNMMM
NAMNAM FerrieFerrierr
BMJBMJ 6060
BGMBGM ARARWW
NAMNAM LinLin KFKF 3131
CLECLE ARARWW
GFSGFS LinLin KFKF 4040
CTP-1CTP-1 NMNMMM
NAMNAM LinLin BMJBMJ 3131
CTP-2CTP-2 ARARWW
NAMNAM LinLin BMJBMJ 3131
BTVBTV NMNMMM
GFSGFS FerrieFerrierr
BMJBMJ 3131
07FEB2007 – Synoptic Setup07FEB2007 – Synoptic Setup
07FEB2007 – Synoptic Setup07FEB2007 – Synoptic Setup
TLAKE: +4C
07FEB2007 – Synoptic Setup07FEB2007 – Synoptic Setup
Radar LoopRadar Loop
Operational NAM PerformanceOperational NAM Performance
Operational NAM PerformanceOperational NAM Performance
• Captures basic band evolutionCaptures basic band evolution– Slow with initial southward band movementSlow with initial southward band movement
• Problems with inland extent of the bandProblems with inland extent of the band– Frequently too far inlandFrequently too far inland
• Can the ensemble add value to this Can the ensemble add value to this simulation?simulation?
Ensemble PerformanceEnsemble Performance
Ensemble PerformanceEnsemble Performance
• All members able to simulate a bandAll members able to simulate a band
• Like NAM, ensemble successfully Like NAM, ensemble successfully captures basic band evolutioncaptures basic band evolution
• Probability plots indicate operational Probability plots indicate operational NAM an outlier with inland extentNAM an outlier with inland extent– Ensemble provides added valueEnsemble provides added value
Individual Member PerformanceIndividual Member Performance
• Quantitatively assess each ensemble Quantitatively assess each ensemble membermember – Method: MODE pattern matching software (Davis et Method: MODE pattern matching software (Davis et
al. 2006)al. 2006)
• Identify precipitation “objects” in Identify precipitation “objects” in forecast/observationsforecast/observations
•Match objects based on different attributes Match objects based on different attributes – Distance apart, similarity in area/orientation, overlapDistance apart, similarity in area/orientation, overlap
• Precipitation Obs: NCEP Stage IV AnalysisPrecipitation Obs: NCEP Stage IV Analysis
Individual Member PerformanceIndividual Member Performance
• Example: Example:
Individual Member PerformanceIndividual Member Performance
• The Statistics: Primary Band IdentificationThe Statistics: Primary Band Identification– POD/FAR/CSIPOD/FAR/CSI
MODELMODEL PODPOD FARFAR CSICSI
NAM-NAM-NMM*NMM*
0.900.90 0.000.00 0.900.90
BGM-ARWBGM-ARW 0.850.85 0.080.08 0.790.79
BTV-NMMBTV-NMM 0.310.31 0.230.23 0.250.25
CLE-ARWCLE-ARW 0.480.48 0.080.08 0.430.43
CTP-ARWCTP-ARW 11 0.040.04 0.960.96
CTP-NMMCTP-NMM 0.810.81 0.270.27 0.640.64* 3 hourly time steps
Individual Member PerformanceIndividual Member Performance• The Statistics: Basic Position/IntensityThe Statistics: Basic Position/Intensity
Average Bias (fcst-obs)Average Bias (fcst-obs)MODELMODEL Area Area
(gdpts)(gdpts)Angle Angle
(degs)(degs)Area-Avgd Area-Avgd Intensity Intensity
(mm/grdpt)(mm/grdpt)
NAM-NMM*NAM-NMM* 96.396.3 -2.7-2.7 0.100.10BGM-ARWBGM-ARW 23.823.8 -0.4-0.4 0.050.05BTV-NMMBTV-NMM 2.82.8 -5.3-5.3 -0.04-0.04CLE-ARWCLE-ARW -33.2-33.2 -1.1-1.1 0.100.10CTP-ARWCTP-ARW 68.268.2 1.11.1 0.040.04CTP-NMMCTP-NMM 9.89.8 -0.5-0.5 -0.01-0.01* 3 hourly time steps
ConclusionsConclusions
• Case study suggests ensemble Case study suggests ensemble approach to LES may be valuableapproach to LES may be valuable– Hone in on high-probability impact areasHone in on high-probability impact areas– Highlight outlier (low-probability) Highlight outlier (low-probability)
outcomesoutcomes
• Initial Quantitative Analysis shows Initial Quantitative Analysis shows diversity in “best member” for diversity in “best member” for different variablesdifferent variables– Ensemble mean likely to have increased Ensemble mean likely to have increased
skill over individual membersskill over individual members
Contact Info/AcknowledgementsContact Info/Acknowledgements
• Have Questions? Have Questions? – [email protected]@noaa.gov
Acknowledgements:Acknowledgements:• Ensemble Participants Ensemble Participants
– For agreeing on a common domain/sharing dataFor agreeing on a common domain/sharing data
• Ron Murphy, ITO BGM Ron Murphy, ITO BGM – For gathering 7 February 2007 case dataFor gathering 7 February 2007 case data
• MODE Software designersMODE Software designers– http://www.dtcenter.org/met/users/http://www.dtcenter.org/met/users/