National Weather Service The Short-Range Ensemble Forecast: SREF Applying Uncertainty and Probabilistic Forecasts of Winter Storms Matt Steinbugl, NOAA/NWS Des Moines (formerly) Rich Grumm, NOAA/NWS State College
National Weather Service Short-Range Ensemble Forecast Objectives Convey and apply uncertainty to the forecast process Convey and apply uncertainty to the forecast process Recognize and assign probabilities to crucial winter weather forecast parameters Recognize and assign probabilities to crucial winter weather forecast parameters This will allow forecasters: To increase confidence in a forecast through a probabilistic approach To increase confidence in a forecast through a probabilistic approach To make better decisions while allowing users better decision making capabilities To make better decisions while allowing users better decision making capabilities
National Weather Service Why Ensembles? – Uncertainty/Chaos
National Weather Service Why Ensembles? Uncertainty in initial conditions and model calculations can alone lead significant outcome changes (run-to-run)Uncertainty in initial conditions and model calculations can alone lead significant outcome changes (run-to-run) Need to account for non-linear processesNeed to account for non-linear processes Atmosphere is chaotic in natureAtmosphere is chaotic in nature
National Weather Service Why Ensembles? Needed to deal with inherent forecast uncertaintyNeeded to deal with inherent forecast uncertainty Improve significant winter weather forecastsImprove significant winter weather forecasts Recognize high uncertainty/high probability outcomes and relate these to each phase of the forecast processRecognize high uncertainty/high probability outcomes and relate these to each phase of the forecast process Risk of heavy rain Prob of 4” snow
National Weather Service
What is the SREF? Multi-model based ensemble prediction system (EPS) with each member having different dynamical cores and physics packages. 21 individual members: 5 ETA (BMJ) + 5 ETA (KF) + 5 RSM + 6 WRF NMM/ARW (BMJ/KF) = 21 members -3 hourly output out to 87hrs -Produced at NCEP 03Z, 09Z, 15Z and 21Z
National Weather Service Deterministic (GFS) vs. Probabilistic (SREF) Model Initial Conditions (ICs) Model cores Remarks GFS 1 IC 1 model core run-to-run (jumpiness) SREF Multiple ICs Multiple cores More consistency Comparing deterministic models is a 50/50 proposition!!!
National Weather Service SREF Performance Combo GFS/NAM SREF Mean Error
National Weather Service Case Study Data Examine 2 significant winter weather events across the Eastern United StatesExamine 2 significant winter weather events across the Eastern United States Determine the following:Determine the following: -Amounts/timing of pcpn? -PYTPE? -Temps for Snow vs. Ice? -Pattern Recognition? -Atypical/typical event?
National Weather Service Case Study # Dec 2004
National Weather Service Spaghetti / Probability charts - 0° isotherm Spaghetti / Probability charts - 0° isotherm Mean and probability Spread 2m850mb edu/ensembles/java/ ModelDisplay.html
National Weather Service Mixed/Conditional Probability charts PYTPE Mixed/Conditional Probability charts PYTPERain Ice Pellets Snow FZRA edu/ensembles/java/ ModelDisplay.html
National Weather Service Probability/Mean charts – 0.50/1.00” QPF 0.50 inch 1.00 inch
National Weather Service So, what happened ??? Our guests can look at the handouts Please don’t share with NWS folks…. This case is part of a training scenario- yet to be completed !
National Weather Service Case Study # April 2005 Detroit, Michigan
National Weather Service Mixed/Conditional Probability charts PYTPE Rain Ice Pellets Snow FZRA
National Weather Service Probability/Mean 0.40” QPF over 24 hr Starting 21Z Starting 9 hours later
National Weather Service Detroit, MI Plume Diagram eyewall.met.psu.edu/plumes/PlumeDisplay.html You can get these for DSM, ALO, DBQ and BRL
National Weather ServiceNOHRSC
Summary Ensemble Prediction Systems are an important means of: Ensemble Prediction Systems are an important means of: Conveying and applying uncertainty through a probabilistic approachConveying and applying uncertainty through a probabilistic approach Visualizing and quantifying uncertainty within the forecast processVisualizing and quantifying uncertainty within the forecast process Using ensembles will allow forecasters to relate probabilities to each phase of the warning decision process In turn, this will allow forecasters to make better decisions and users to have better decision making capabilities
National Weather Service java/ModelDisplay.html Spaghetti charts, model variance and normalized anomaly PlumeDisplay.html Plume charts for DSM, ALO, DBQ, BRL COM_US/web_js/html/mean_surface_prs.html COM_US/web_js/html/mean_surface_prs.html NCEP Environmental Modeling Center SREF page
National Weather Service Questions ???
National Weather Service Special Thanks Rich Grumm, SOO CTP Karl Jungbluth, SOO DMX Peter Manousos, SOO NCEP Jun Du, NCEP/EMC Steve Wiess, SPC Jeremy Grams, SPC David Bright, SPC
National Weather Service References AWOC Winter IC 6.3: Using Ensembles in Winter Weather Forecasting SREF Exploitation at NCEP’s Hydrometeorological Prediction Center (HPC) Dealing with uncertainties in forecasts – M Steven Tracton NWS/NCEP/EMC
National Weather Service MRCC