RAP / HRRR weather guidance for Air Traffic Flow Management:

Progress and Challenges

Friends and Partners of Aviation Weather NBAA Convention, October 23-24, 2014

Stephen S. Weygandt Curtis Alexander, Stan Benjamin,

Geoff Manikin*, Tanya Smirnova, Ming Hu NOAA Earth System Research Laboratory

*NOAA National Centers for Environmental Prediction

Rapid Refresh and HRRR NOAA hourly updated models

13km Rapid Refresh (RAP)

3km High Resolution Rapid Refresh (HRRR)

Version 2 -- NCEP implement 25 Feb 2014

Version 3 – GSD Planned NCEP -- 2015

RAP

HRRR Initial – NCEP implement 30 Sept 2014

Version 2 – GSD Planned NCEP --2015

GSD NCEP HRRR: MORE RELIABLE (now 99+%) AVAILABLE SOONER (1h fcst by +50 min, 15h fcst by +1:30

Higher resolution models

Rawins Satellite Aircraft Radar Surface Profiler

Accurate prediction of hazardous weather What are the key requirements?

Rapid updating

Use of all Observations (Data assimilation)

13-km Reality

ORD

LAX

ORD

3-hr forecast

LAX

6-hr forecast Wind

Errors

3-km

-- Fresher forecasts -- Time-lagged ensembles

High-Impact Prediction Needs: Higher Resolution Models

20 km RUC 2002

13 km RUC 2005 (RAP 2012)

3 km HRRR 2014

40 km RUC 1998

AWRP PMR 11 June 2013 4 Model Development & Enhancement NBAA 22 Oct 2014 4 Friends and Partners Aviation Weather

13-km 6hr forecast HRRR 6hr forecast

Observations Used Data assimilation: using all the observations including radar data

Model Pre-Forecast Time (min) MODEL START

-60 -15 -45 -30 0

Model Integration

Observed Radar

Valid at -45 Valid at -30 Valid at -15 Valid at 0

+30

30-min fcst 3-km radar DA

30-min fcst NO 3-km radar DA

Radar Obs 05:30z 18 May 2013

0 1 2 3 4 5 6

PersistenceExtrapolationNo radar DARAP radar DAHRRR radar DA

HRRR adds skill to nearly 1 hour

(2013-2014)

Cross over in skill

Forecast Skill Comparison

Forecast Length (Hours)

Fore

cast

Ski

ll

L

ess

Skill

M

ore

Skill

Improved 0-3 hr Convective Fcsts

Model prediction skill is moving toward nowcast skill Radar data assimilation is key part of this

NO radar DA

13-km RAP radar DA

3-km HRRR radar DA

Nowcast

Combined

00 UTC

Real-time

Model ensembles: Key to assessing likelihood

Member 3

Observations

Member 2

Member 1

Supercell probability past hour

19 UTC

00 UTC

18z 14 July 2014 09z + 9h fcst 18z

NSSL Radar

mosaic

Thunderstorm Probability (%)

Reflectivity

HRRR forecast radar

HRRR forecast probability

HRRR deterministic and probabilistic forecasts

21z 14 July 2014 09z + 12h fcst 21z

NSSL Radar

mosaic

Thunderstorm Probability (%)

Reflectivity

HRRR forecast radar

HRRR forecast probability

HRRR deterministic and probabilistic forecasts

Evolution of hourly updated NOAA modeling

Feb 2014 –Rapid Refresh v2 – oper at NCEP Improved surface forecasts, convective environment fields • Hybrid ensemble-variational GSI assimilation • Model – improved cloud / PBL / LSM / soil, updated version of WRF ARW

Sept 2014 – HRRR (3km) – oper at NCEP Hourly updated storm-scale forecasts, greater accuracy (timing, location, structure) for thunderstorms and many other features • 3-km / 15 min radar reflectivity assimilation

North American Rapid Refresh Ensemble (NARRE)

• 2 cores (ARW, NMMB), 6 members

• Hourly updating with GSI-hybrid EnKF

• Hourly forecasts to 24-h, 84-h 4x / day

Rapid Refresh NARRE

2015 – RAPv3/HRRRv2

~2017 – NARRE with hybrid 4d-ens/var DA

~2019 – HRRR Ensemble (HRRRE) – eventually with ~3km ensemble DA

HRRRHRRR

Future oper @ NCEP

AK HRRR