2014 Warn-on-Forecast / High Impact Weather Workshop 1

Development of an Hourly-Updated NAM Forecast System: Current Efforts and Future Plans

Jacob Carleyab, Eric Rogersb, Shun Liuab, Xiaoyan Zhangbc, Brad Ferrierab, Eric Aligoab,Matthew Pyleb, and Geoff DiMegob

aIMSG, bNOAA/NWS/NCEP/EMC, cESSICjacob.carley@noaa.gov

1

2014 Warn-on-Forecast / High Impact Weather Workshop 2

Hourly-Updated NAM Forecast System• NAM – North American Mesoscale forecast system

o Runs 4x daily at 00, 06, 12, 18Zo Short-range mesoscale NWP system for the

U.S. which provides guidance to day 3.5

• NAMRR: NAM Rapid Refresho Hourly updateso Future North American Rapid Refresh

Ensemble system (NARRE) NAMRR + RAP/HRRR Foundation

• Development of hourly NAM cycling capabilities on NOAA R&D machine Zeuso Part of DOE-funded wind energy projects

WFIP/POWERo Cycling 12 km NAM and 4 (3) km CONUS

nesto Hybrid ensemble-3DVar via Global Data

Assimilaion System’s EnKF memberso Cloud analysis and diabatic digital filter

initialization with radar-derived temperature tendencies

2014 Warn-on-Forecast / High Impact Weather Workshop 3

NAMRR Overview

Current, example, NAM Data Assimilation System (NDAS) configuration for a single, arbitrary cycle:

Example NAMRR configuration for 12, hourly cycles:

Both systems assimilate a wide range of conventional (e.g. surface, profiler, mesonets, Doppler radar radial velocities, etc.) and satellite observations (e.g. radiances)

2014 Warn-on-Forecast / High Impact Weather Workshop 4

NAMRR: Configuration (still testing)Model Points

(nx,ny,nz)dx DFI? dBZ

Init.?Param. Conv?

Micro-physics

Radiation Analysis/Update frequency

Forecasts per day(max length)

Ops NAM

954x835x60 12 km No No BMJ Ferrier GFDL 3 hourly (3DVar)

4 (84 hrs)

Ops NAM CONUSnest

1371x1100x60

4 km No No BMJ ‘light’ Ferrier GFDL None 4 (60 hrs)

NAMRR 954x835x60 12 km Yes Yes BMJ (updated)

Ferrier-Aligo

RRTMG Hourly (Hyb. Ens-3DVar)*

24 (84 hrs)

NAMRR CONUSnest

1371x1100x60(1828x1466x60)

4 km(3 km)

Yes Yes none Ferrier-Aligo

RRTMG Hourly (Hyb. Ens-3DVar)*

24 (60 hrs)

*Using 80 EnKF members from the Global DA Systems

2014 Warn-on-Forecast / High Impact Weather Workshop 5

June 29-30, 2012 Derecho - NAMRR Test

Poor 4 hour forecast from Ops NAM 4 km CONUSnest

Much improved 4 hour forecast from NAMRR 4 km CONUSnest

Additional improvement with NAMRR 3 km CONUSnest

Obs Ops 4km

NAMRR 4km NAMRR 3km

4 Hour Forecast valid 22 UTC

Prelim. Storm Reports from SPC (06/29/2012)

2014 Warn-on-Forecast / High Impact Weather Workshop 6

June 29-30, 2012 Derecho - 27 Hr Forecast NAMRR Test with 3 km CONUSnest

Observations3 km NAMRR CONUSnestFhr=27

4 km Ops NAM CONUSnestFhr=27

Significant improvement at longer lead times with 3 km

NAMRR relative to Ops

2014 Warn-on-Forecast / High Impact Weather Workshop 7

Impact of Cloud Analysis on 4 km CONUSnest Reflectivity Forecasts

Verification over CONUS for 2012 Derecho short-term forecasts (dBZ >= 30)

Stats from 30 hourly forecast cycles.Lines falling outside the rectangles are significant at the p=0.05 level

Blue = 4 km NAMRR CONUSnest with cloud analysis

Red = 4 km NAMRR CONUSnest without cloud analysis

F01 F12

ETS

F01 F12

Bias

CTL

Cloud Analysis

Perfect frequency bias

2014 Warn-on-Forecast / High Impact Weather Workshop 8

NAMRR Testing of Ferrier-Aligo Microphysics3 km CONUSNEST forecast column max radar reflectivity from the 20120628 12Z cycle using the repository microphysics (top) and the Ferrier-Aligo microphysics (bottom). On grid 227 (5 km).

NAMRR is playing a role in current physics development.

NMMB Repository Ferrier Microphysics

Updated Ferrier-Aligo Microphysics tested in NAMRR

Increased number of forecast reflectivity values >= 50 dBZ. More consistent with observations.

Obs

FCST

2014 Warn-on-Forecast / High Impact Weather Workshop 9

May 20th, 2013 OK Severe Weather Event

ETS

NAMRR Nest

Ops NAM Nest

5 km grid

Fhr = 16Valid 22Z

ObsNAMRR 3kmOps 4km

Fhr = 03Valid 21Z

Perfect frequency

bias

Bias

5 km grid

2014 Warn-on-Forecast / High Impact Weather Workshop 10

Ongoing and Future WorkNear Term

• NAMRR retrospective tests for comparison with upcoming NAM bundle

• Optimization (always improving)

Long(er) Term• Extend NAMRR from 2 to all 6 operational NAM domains

• Initial evaluations of NARRE+HRRRE TL (RAP/HRRR +NAMRR/NESTs)

• Test + evaluate in collaborative R2O settings?• WPC-HMT Flash Flood and Winter Weather Experiments, SPC Spring

Experiment?

• Planned for WCOSS Phase II (Petaflop): 2016-2018?

• Hybrid 4DEnVar

Example of the direct assimilation of Doppler velocity and radar reflectivity via the GSI into a 1.33 km version of the NMMB. The figure compares observed column max reflectivity to a case of no assimilation (CTL), 3DVar, and various hybrid ensemble-3Dvar (Hyb) configurations during 1 hour of assimilating obs at 5 minute intervals (from J. Carley’s PhD dissertation).

2014 Warn-on-Forecast / High Impact Weather Workshop 11

May 20th, 2013 OK Severe Weather – 1.33 km HiRes Nest Experiment

Max Hourly Updraft Helicity + Storm Reports

Column-Max dBZ Obs Column-Max dBZ

Where *might* WoF fit?

- A locatable, on-demand, high-resolution set of nested ensemble members all featuring DA cycles

- Adopt techniques and lessons learned from the WoF community!

- Already doing something like this in Ops NAM with FireWX Nest (No DA, single member)

***Just an idea – but good to think about these things as we all progress***

HRRRE WoF NEST LOCATIONS FOR NEXT CYCLES

14Z Cycle 14Z CycleObs

* Thanks to Eric Aligo for providing this run

2014 Warn-on-Forecast / High Impact Weather Workshop 12

Extra Slides

2014 Warn-on-Forecast / High Impact Weather Workshop 13

Convective Initiation: 3km NAMRR CONUSnest: 2013-08-15 @ 2200 UTC

F07 F06F08

F09F10OBS

SPC MESO DISCUSSION #1709: ...ASCENT INVOF THE WRN EXTENT OF A DIFFUSE FRONT THAT ENTERS THE ERN PART OF THE MCD AREA FROM SERN KS WILL ALSO SUPPORT CONVECTION THROUGH THE AFTERNOON HOURS. LIMITED DEEP SHEAR PER VWP DATA WILL KEEP CONVECTION LARGELY DISORGANIZED

2014 Warn-on-Forecast / High Impact Weather Workshop 14

POWER:Position of Offshore Wind Energy Resources• Collaborative effort with DOE, ESRL/GSD, and ESRL/PSD

o Provide information about observation networks needed to support offshore wind energy development

• Data denial experimentso Summer 2004 New England Air Quality Study deployed ~13

coastal wind profilers and one shipborne Doppler lidar in the New England area

o POWER uses these pre-existing coastal profiler observations in a set of data-denial experiments with the NAMRR and RAP/HRRR systems Shipborne lidar is used for verification

o Two periods of study July 10th-17th, 2004 August 6th-12th, 2004

08/06 08/0808/07 08/09 08/10 08/11

Courtesy of Yelena Pichugina

NOAA Ron Brown ship tracks (August, 2004)

NOAA Ron Brownhttp://oceanexplorer.noaa.gov/technology/vessels/ronbrown/ronbrown.html

2014 Warn-on-Forecast / High Impact Weather Workshop 15

August 6th-12th, 2004: 0-2km AGL vs. POWER profilers

RMSE

Forecast Hour

RAPXNAMRRX

NAMRRX+RAPX

NAMRR+RAPX vs. NAMRRX vs. RAPX

Parent RMSE 4 km CONUSNEST RMSE

Forecast Hour Forecast Hour

CTL

EXP

CTL

EXP

All assimilated extra profilers in this case

CTL: No Extra profiler assim.

EXP: Extra profiler assim.

All assimilated extra profilers in this case