Current Status and Plans for Convection Allowing Models in ... · • Stan Benjamin ESRL/GSD ... Can GFS and GEFS take over NAM/RAP and SREF capabilities in the next couple years?

N A T I O N A L O C E A N I C A N D A T M O S P H E R I C A D M I N I S T R A T I O N

2018 NPSR 2018 NPSR

Current Status and Plans for Convection Allowing Models in the NCEP Production

Suite Jacob R. Carley

NOAA/NCEP/Environmental Modeling Center

With acknowledgements to EMC colleagues and the UFS CAM SIP Working Group

1


2018 NPSR

UFS CAM SIP WG Membership • Curtis Alexander ESRL/GSD ** • Jack Kain EMC ** • Lou Wicker NSSL** • Lucas Harris GFDL**

• Dave Stensrud PSU • Ming Xue OU/CAPS • Xuguang Wang OU/SoM • Jamie Wolff NCAR/DTC • Glen Romine NCAR/MMM • Bill Putman NASA/GMAO • Gary Lackmann NC State • Vittorio Gensini NIU

Co-Chair **

• Eric Rogers EMC • Jacob Carley EMC • Geoff DiMego EMC • Adam Clark NSSL • Israel Jirak SPC • SJ Lin GFDL • Stan Benjamin ESRL/GSD • Sundararaman Gopalakrishnan AOML • Andy Hazelton AOML

2


2018 NPSR

● Cover CONUS + OCONUS in a large variety of capacities

● NAM + Nests ● HRRR CONUS and Alaska ● HiRes Windows

3

Current Status - Quite a bit

○ ARW and NMMB members for each domain ● Much of this has been consolidated in the HREF

○ High Resolution Ensemble Forecast system ○ Motivated by the successes of the SSEO from SPC

NAM+Nests

RAP/HRRR

HiResWs


2018 NPSR 4

The HREF v2 0 +12 +24 +36 -12

HiresW – ARW (RAP init)

HiresW – NMMB (RAP init)

NAM NEST - NMMB (NAM init) **CONUS ONLY**

HIRESW - “mem2” ARW (NAM init)

*Covers CONUS, Alaska, Hawaii, and Puerto Rico Thanks to Matthew Pyle and Israel Jirak for the material on this slide

2018 HWT results: HREF rated higher than other formally constructed CAM ensembles for severe weather guidance in subjective evaluations

8 members for CONUS 4 distinct members 4 time lagged


2018 NPSR

● NAM was frozen at version 4 (March 2017) ● HiRes Window systems are currently frozen ● RAP/HRRR has one more upgrade scheduled ~FY20 ● The HREF was a natural step forward in aggregating

much of the CAM output into an ensemble ○ But the underlying components of HREF are frozen or about

to be - so what’s next?

5

What’s Next?


2018 NPSR

FV3 → Adaption for GFS First then CAM

Courtesy: GFDL

6

• NOAA GFDL FV3 selected for dynamic core component of NGGPS – *Going into operations with GFS v15*

• Early 2019

• Opportunity to unify efforts and enhance collaboration for CAM applications – EMC/GSD/NSSL/AOML/GFDL/DTC/etc


2018 NPSR 7

FV3-CAM Timeline → Rapid Refresh Forecast System

FY18/19

FY21

~FY22

Freeze all non-FV3 CAM systems

*Timeline may be revised as development matures/progresses*

Q4FY20

Q2FY20

Development underway standalone/regional, nesting,

DA, physics, etc.

HREFv3 Replace poorer performing members with FV3-CAM

RRFSv1 Implement RRFSv1 pending favorable

evaluation

CAM Development Continues Demonstration FV3-CAM ensemble DA + forecast system - evaluate against HREF. Continue physics testing/advancement.

Rapid Refresh Forecast System → To replace HREF, HRRR, NAM + nests, HiResWs


● FV3 is a global model with a nest capability ● Need a standalone grid option for CAM applications ● Why?

○ Extra resources are not needed for a global parent ○ Rapid updates in DA are much more feasible

● Obvious potential drawback

8

Highlights → Standalone Alone Regional

○ Boundary data from an external forecast cannot be as accurate as those provided by a parent to a nest every timestep during the integration

● Recent tests show standalone is 2x faster!


2018 NPSR 2018 NPSR

3 km NAM 3 km FV3-SAR

32h Forecasts of Hurricane Michael

**Example → meant to demonstrate progress on key components**

9

Highlights →Current DA Testing


2018 NPSR

● Can GFS and GEFS take over NAM/RAP and SREF capabilities in the next couple years? ○ GFS and GEFS comparable

resolution ○ Some outstanding issues

currently inhibit this ■ e.g. BL thermo issues ■ Spread? (GEFS? Single core

CAM ensemble? 10

Challenge: NAM/RAP/SREF?


2018 NPSR

Summary ● RAPv5/HRRRv4 dev and final implementation

○ Now to Q2 FY20 ● Transition work of existing Meso/CAM capabilities to FV3-CAM

○ Now to Q4 FY20 ○ HREFv3

● Rapid Refresh Forecast System [RRFS] ○ Q4 FY20 to ~FY22 ○ Rapid update ensemble DA and forecast system ○ Replaces NAM/NAM nests/RAP/HRRR

● EMC/GSD/NSSL/GFDL/DTC/etc. Collab. ● Thanks!

11

Contact: [email protected]


2018 NPSR 12

BREAKOUT SESSION 3 TOPICS

1. What do you see as impediments to using the same CAM framework for all hi-res applications and timescales?

1. How are users currently using the HREF and experimental CAM ensemble guidance? What are your impressions?


2018 NPSR 13

Supplementary


2018 NPSR 2018 NPSR

Highlights →Current DA Testing

● The GSI system has been developed to interface directly with the FV3 native grid for a tile or nest input

● Configuration similar to 3 km NAM CONUS nest, using hourly RAP obs ● Very much in experimental/sandbox testing mode

Coldstart from FV3-GFS Parallel

Hourly hybrid 3DEnVar assimilation ● Using global EnKF members ● Assimilates hourly RAP obs ● *NO* cloud analysis and reflectivity derived latent heating ● *NO CAM ensemble at this time*

60 hr 3 km forecast

6 hour forecast at 3 km

14


2018 NPSR

Development of NGGPS/UFS Planning Documents

26 Sept 2018 NGGPS Update 15

Forecast Field Application Vertical Attribute Temporal Attribute Validation Source Priority/Importance Maturity/Readiness Deterministic Methodology Determinisitic Scores Deterministic Stratifications Ensemble Methodology Ensemble Scores Ensemble Stratifications

Downward Shortwave Radiation

Air Quality/Energy Land Surface

Surface Instantaneous/Average

ARM, Surfrad (Oak ridge, ameriflux), USCRN

3 1 Grid-to-Point RMSE, BIAS sub-regions, Need to average in time

Ceiling Aviation Column Instantaneous METARs 1 1 Grid-to-Point

CSI, BIAS, FSS, POD, FAR, AUR,

Performance Diagram

Forecast Length [0-36 hr], Threshold [500,

1000, 3000, 5000, 10000 ft], Domain [W and E

CONUS]

Point probability, HIRA? (see MET

documentation for neighborhood around ob

points)

Briar Score, Briar Skill Score (other ensemble reference), Reliability,

Sharpness, CRPS, CRPSS (other

ensemble reference)

Smoothing [? km], Probabilities [0, 5, 10,

...100]

Echo Top Height Aviation Column Instantaneous MRMS Echo Top 2 1 Grid-to-Grid


Performance Diagram

Forecast Length [0-36 hr], Threshold

[20,25,30,35,40 kft], Scale [03, 40 km], Domain [W and E

CONUS]

Neighborhood probability



ensemble reference)

Neighborhoods [10, 20, 40, 80 km], Smoothing [? km], Probabilities [0,

5, 10, ...100]

Visibility Aviation Surface Instantaneous METARs 1 1 Grid-to-Point


Performance Diagram

Forecast Length [0-36 hr], Threshold [0.5, 1.0,

3.0, 5.0, 10.0 mi], Domain [W and E

CONUS]

Point probability, HIRA? (see MET

documentation for neighborhood around ob

points)



ensemble reference)

Smoothing [? km], Probabilities [0, 5, 10,

...100]

CAPE/CIN Environmental Mixed, Most-

Unstable, Surface-Based

Instantaneous RAOB 2 1 Grid-to-Point RMSE, BIAS

Forecast Length [0-36 hr], Diurnal [00,12 Z],

Domain [W and E CONUS]

Ensemble arithmetic mean, standard

deviation of ensemble members vs mean

(ensemble spread), ob error, ensemble spread + ob error (total spread),

standard deviation of ensemble mean vs obs

(ensemble error), bias of ensemble mean vs obs (ensemble bias), RMSE

and BIAS of each member (ensemble

design)

Spread-skill ratio, rank histogram?

Vertical Temporal Attributes

Validation Source

Scores and Stratifications (scales, times,

thresholds)

Priority and Readiness Rankings

Draft UFS CAM Verification Metrics (1 August 2018)

● Pre-implementation decision making is initial emphasis

● Holistic set of CAM forecast fields/applications:

○ Environmental, Severe, Precip/Winter, Aviation, Energy, Air Quality

● Total of thirty CAM forecast fields identified

● Eleven high-priority and high-readiness CAM fields identified