National Weather Service The Short-Range Ensemble Forecast: SREF Applying Uncertainty and Probabilistic Forecasts of Winter Storms Matt Steinbugl, NOAA/NWS.

National Weather Service

The Short-Range Ensemble Forecast: The Short-Range Ensemble Forecast: SREFSREF

Applying Uncertainty and Probabilistic Applying Uncertainty and Probabilistic Forecasts of Winter Storms Forecasts of Winter Storms

Matt Steinbugl, NOAA/NWS Des Moines Matt Steinbugl, NOAA/NWS Des Moines (formerly)(formerly)

Rich Grumm, NOAA/NWS State Rich Grumm, NOAA/NWS State CollegeCollege


Short-Range Ensemble Forecast Short-Range Ensemble Forecast ObjectivesObjectives

• ConveyConvey and and applyapply uncertainty uncertainty to the forecast processto the forecast process• RecognizeRecognize and and assignassign probabilities to crucial winter probabilities to crucial winter weather forecast parametersweather forecast parameters

This will allow forecasters:This will allow forecasters:• To To increase confidenceincrease confidence in a in a forecast through a probabilistic forecast through a probabilistic approachapproach• To To make better decisionsmake better decisions while while allowing users better decision allowing users better decision making capabilitiesmaking capabilities


Why Ensembles? – Why Ensembles? – Uncertainty/ChaosUncertainty/Chaos


Why Ensembles?Why Ensembles?

•Uncertainty in initial conditions Uncertainty in initial conditions and model calculations can alone and model calculations can alone lead significant outcome changes lead significant outcome changes (run-to-run)(run-to-run)

•Need to account for non-linear Need to account for non-linear processesprocesses

•Atmosphere is chaotic in natureAtmosphere is chaotic in nature


Why Ensembles?Why Ensembles?

•Needed to deal with inherent Needed to deal with inherent forecast uncertaintyforecast uncertainty

• Improve significant winter Improve significant winter weather forecastsweather forecasts

•Recognize high Recognize high uncertainty/high probability uncertainty/high probability outcomes and relate these to outcomes and relate these to each phase of the forecast each phase of the forecast processprocess

Risk Risk of of heavheavy y rainrain

Prob of Prob of 4” 4” snowsnow



What is the SREF?What is the SREF?

Multi-model based ensemble Multi-model based ensemble prediction system (EPS) with each prediction system (EPS) with each member having different dynamical member having different dynamical cores and physics packages.cores and physics packages.

21 individual members:21 individual members:

5 ETA (BMJ) + 5 ETA (KF) + 5 RSM 5 ETA (BMJ) + 5 ETA (KF) + 5 RSM ++6 WRF NMM/ARW (BMJ/KF) = 6 WRF NMM/ARW (BMJ/KF) = 21 21 membersmembers

-3 hourly output out to 87hrs-3 hourly output out to 87hrs-Produced at NCEP 03Z, 09Z, 15Z -Produced at NCEP 03Z, 09Z, 15Z and 21Z and 21Z


Deterministic (GFS) vs. Deterministic (GFS) vs. Probabilistic (SREF)Probabilistic (SREF)

ModelModel Initial Initial ConditionCondition

s (ICs)s (ICs)

Model Model corescores

RemarksRemarks

GFSGFS 1 IC1 IC 1 model 1 model corecore

run-to-run-to-run run

(jumpine(jumpiness)ss)

SREFSREF Multiple Multiple ICsICs

Multiple Multiple corescores

More More consistenconsisten

cycyComparing deterministic models is a 50/50 Comparing deterministic models is a 50/50

proposition!!!proposition!!!


SREF PerformanceSREF Performance

Combo Combo GFS/NAGFS/NAMM

SREF SREF MeanMeanE

rro

Err

orr


Case Study DataCase Study Data

•Examine 2 significant winter Examine 2 significant winter weather events across the weather events across the Eastern United StatesEastern United States

•Determine the following:Determine the following:

-Amounts/timing of pcpn?-Amounts/timing of pcpn?

-PYTPE?-PYTPE?

-Temps for Snow vs. Ice?-Temps for Snow vs. Ice?

-Pattern Recognition?-Pattern Recognition?

-Atypical/typical event?-Atypical/typical event?


Case Study #1 22-23 Dec 2004Case Study #1 22-23 Dec 2004


SpaghettiSpaghetti//Probability charts - 0° Probability charts - 0° isothermisotherm

Mean Mean and and probabilprobabilityity

SpreSpreadad

22mm

850850mbmb

http://http://eyewall.met.psu.eyewall.met.psu.edu/ensembles/edu/ensembles/java/java/ModelDisplay.htModelDisplay.htmlml


Mixed/Conditional Probability Mixed/Conditional Probability charts PYTPEcharts PYTPE

RaiRainn

Ice Ice PelletsPellets

SnoSnoww

FZRFZRAA

http://http://eyewall.met.psu.eyewall.met.psu.edu/ensembles/edu/ensembles/java/java/ModelDisplay.htModelDisplay.htmlml


Probability/Mean charts – Probability/Mean charts – 0.50/1.00” QPF0.50/1.00” QPF

0.50 0.50 inchinch

1.00 1.00 inchinch


So, what happened ???So, what happened ???

Our guests can look at the handoutsOur guests can look at the handouts

Please don’t share with NWS folks….Please don’t share with NWS folks….

This case is part of a training This case is part of a training scenario- scenario-

yet to be completed !yet to be completed !


Case Study #2Case Study #2 23-25 April 2005 23-25 April 2005

Detroit, Detroit, MichiganMichigan


Mixed/Conditional Probability Mixed/Conditional Probability charts PYTPEcharts PYTPE

RaiRainn

Ice Ice PelletsPellets

SnoSnoww

FZRFZRAA


Probability/Mean 0.40” QPF over Probability/Mean 0.40” QPF over 24 hr24 hr

StartiStarti

ng ng 21Z21Z

Startin

Startin

g 9 g 9

hours hours

laterlater


Detroit, MI Plume DiagramDetroit, MI Plume Diagram

http://http://eyewall.met.psu.edu/plumes/eyewall.met.psu.edu/plumes/PlumeDisplay.htmlPlumeDisplay.html

You can get these for DSM, ALO, DBQ You can get these for DSM, ALO, DBQ and BRLand BRL

National Weather ServiceNOHRSNOHRSCC


SummarySummary Ensemble Prediction Systems are an Ensemble Prediction Systems are an important means of:important means of:•Conveying and applying uncertainty through Conveying and applying uncertainty through a probabilistic approacha probabilistic approach

•Visualizing and quantifying uncertainty Visualizing and quantifying uncertainty within the forecast processwithin the forecast process

Using ensembles will allow forecasters to Using ensembles will allow forecasters to relate probabilities to each phase of the relate probabilities to each phase of the warning decision processwarning decision process

In turn, this will allow forecasters to make In turn, this will allow forecasters to make better decisions and users to have better better decisions and users to have better decision making capabilitiesdecision making capabilities


http://eyewall.met.psu.edu/http://eyewall.met.psu.edu/ensembles/java/ModelDisplay.htmlensembles/java/ModelDisplay.html

Spaghetti charts, model variance Spaghetti charts, model variance and normalized anomalyand normalized anomaly

http://eyewall.met.psu.edu/http://eyewall.met.psu.edu/plumes/PlumeDisplay.htmlplumes/PlumeDisplay.html

Plume charts for DSM, ALO, DBQ, Plume charts for DSM, ALO, DBQ, BRLBRL

http://wwwt.emc.ncep.noaa.gov/mmb/SREF/http://wwwt.emc.ncep.noaa.gov/mmb/SREF/FCST/COM_US/web_js/html/FCST/COM_US/web_js/html/

mean_surface_prs.htmlmean_surface_prs.htmlNCEP Environmental Modeling NCEP Environmental Modeling Center SREF pageCenter SREF page


Questions ???Questions ???


SpecialSpecial ThanksThanks

Rich Grumm, SOO CTPRich Grumm, SOO CTP

Karl Jungbluth, SOO DMXKarl Jungbluth, SOO DMX

Peter Manousos, SOO NCEPPeter Manousos, SOO NCEP

Jun Du, NCEP/EMCJun Du, NCEP/EMC

Steve Wiess, SPCSteve Wiess, SPC

Jeremy Grams, SPCJeremy Grams, SPC

David Bright, SPCDavid Bright, SPC


ReferencesReferences

http://http://www.hpc.ncep.noaa.gov/ensembletrainingwww.hpc.ncep.noaa.gov/ensembletraining//

http://wwwt.emc.ncep.noaa.gov/mmb/SREF/WMO06_full.pdfhttp://wwwt.emc.ncep.noaa.gov/mmb/SREF/WMO06_full.pdf

http://wwwt.emc.ncep.noaa.gov/mmb/SREF-Docs/http://wwwt.emc.ncep.noaa.gov/mmb/SREF-Docs/

AWOC Winter IC 6.3: Using Ensembles in Winter Weather AWOC Winter IC 6.3: Using Ensembles in Winter Weather Forecasting Forecasting

http://mcc.sws.uiuc.eduhttp://mcc.sws.uiuc.edu

http://nws.met.psu.edu/severe/index.jsphttp://nws.met.psu.edu/severe/index.jsp

http://nws.met.psu.edu/severe/2006/11May2006.pdfhttp://nws.met.psu.edu/severe/2006/11May2006.pdf

SREF Exploitation at NCEP’s Hydrometeorological Prediction Center SREF Exploitation at NCEP’s Hydrometeorological Prediction Center (HPC) (HPC) http://nws.met.psu.edu/severe/2005/23April2005.pdfhttp://nws.met.psu.edu/severe/2005/23April2005.pdf

Dealing with uncertainties in forecasts – M Steven Tracton Dealing with uncertainties in forecasts – M Steven Tracton NWS/NCEP/EMCNWS/NCEP/EMC

http://weather.unisys.com/archive/index.htmlhttp://weather.unisys.com/archive/index.html

http://eyewall.met.psu.edu/plumes/Plume.pdfhttp://eyewall.met.psu.edu/plumes/Plume.pdf

http://eyewall.met.psu.edu/plumes/PlumeDisplay.htmlhttp://eyewall.met.psu.edu/plumes/PlumeDisplay.html

http://eyewall.met.psu.edu/ensembles/java/ModelDisplay.htmlhttp://eyewall.met.psu.edu/ensembles/java/ModelDisplay.html

National Weather ServiceMRCMRCCC

National Weather Service The Short-Range Ensemble Forecast: SREF Applying Uncertainty and Probabilistic Forecasts of Winter Storms Matt Steinbugl, NOAA/NWS.

Documents