Hurricane Model Transitions to Operations at NCEP/EMC

1
Hurricane Model Transitions to Operations at
NCEP/EMC

• 2009 IHC Conference, St. Petersburg, FL
• Robert Tuleya,
• V. Tallapragada,Y. Kwon, Q. Liu, Zhan Zhang,
• Yihua Wu ,J. OConnor and N. Surgi

JHT sponsored
2
Project Goals and Emphasis

• Participate in the yearly operational
  implementation of HWRF
• Upgrade HWRF system
• Trouble shoot problems
• Goal increase both track and intensity skill
• Continued collaboration with URI, Florida State,
  GFDL, and others

3
HWRF Hurricane Forecast System
NHC storm message Position domain
Get OBS, Model Input initial boundary
conditions
Ocean Initialization Initialize wake, loop
currents eddies
Wrf si (used for topographical parameters) Wrf
realreplace with interpolations from native
model data
storm analysis and data ingest 6hr 1st guess
vortex relocation 3DVAR gsi for both nests
HWRF model Coupled with POM
Next cycle
Synoptic fields for many variables Create file
for track, intensity, etc
4
HWRF GFDL
Grid configuration 2-nests (coincident) 3-nests(not coincident)
Nesting Force-feedback Interaction thru intra-nest fluxes
Ocean coupling POM (atlantic only) POM
Convective parameterization SAS mom.mix. SAS mom.mix.
Explicit condensation Ferrier Ferrier
Boundary layer GFS non-local GFS non-local
Surface layer GFDL ..(Moon et. al.) GFDL ..(Moon et. al.)
Land surface model GFDL slab/NOAH GFDL slab
Dissipative heating Based on D-L Zhang Based on M-Y tke 2.5
Gravity wave drag YES NO
Radiation GFDL (cloud differences) GFDL
5
HWRF 2008 Skill
6
Hourly intensity
Hourly model data
Model atcf data
Model variability may be important ?
7
Season statistics not affected
8
Sfc Temperatures Problems in HWRF remaining b.c.
noise??
3
2
1
4 (fixed)
9
Fay 082018Impact of Tsfc fix (improved track,
more intense)
Tsfc fix
Tsfc fix
Hwrf
Hwrf
HWRF prod
HWRF prod
10
Gustav 083100Impact of Tsfc fix (improved
track, same intensity)
Tsfc fix
HWRF prod
Hwrf
Tsfc fix
HWRF prod
Hwrf
11
Other potential improvements

• Surface flux formulations
• Land surface modeling
• Gravity wave drag
• High resolution

12
Reduced Surface Drag
13
Noah LSM studies - Background
GFDL Slab LSM 1) One level, only predicts
surface temperature, wetness is fixed, no runoff
Noah LSM 1) The operational LSM in NCEP's
operational mesoscale forecast model (Ek et al.,
2003) 2) Multiple soil layers (usually 4 layers
0-10,10-40, 40-100 and 100-200 cm depth) with a
one-layer vegetation canopy 3) Spatially
varying root depth and seasonal cycle of
vegetation cover 4) Frozen soil physics for cold
regions, and improved soil and snowpack thermal
conductivity. 5) The Noah LSM predicts soil
moisture, soil temperature, land surface skin
temperature, land surface evaporation and
sensible heat flux, and total runoff. 6) The
HWRF runoff prediction using the Noah LSM can
then be used as forcing input to EMC's Streamflow
Routing Scheme (Lohmann et al., 2004).
Additionally, 7) The HWRF-Noah forecasts of
soil moisture and runoff are good spatial
indicators of soil moisture saturation (water
logging) and flooding.
14
HWRF Predicted Tracks of Katrina
Other sfc bl physics
Obs
With/without NOAH LSM
15
12 Hour Accum. Rainfall (mm)
HWRF
NAM
OBS
48h
48h
48h
72h
72h
72h
Observed rainfall is the rain gauge
measurement Observed rainfall spreads in larger
area than NAM and HWRF rainfall did
16
Forecasted Stream Flow (m3 s-1)
17
Influence of orography on the atmosphere
Create obstacles and additional turbulence
Gravity wave drag
Change the large scale flows
Generation of vertically propagating gravity waves
Change the track of hurricanes
18
Motivations

1. Track forecast skills of HWRF on Eastern Pacific
   storms are not as good as those on Atlantic
   storms
2. Diagnotics of HWRF indicates the anomalous flows
   developed over Mexican Plateau seems to cause the
   less skillful track forecast of HWRF
3. Proper GWD representation might improve the track
   forecast of HWRF
4. GWD improved NAM

Results NEXT PAGE
19
No GWD
GWD
50nm improvement at t120hr
20
High Resolution HWRF Experiment

• Resolution
• - Control 0.18, 0.06 (27km, 9km)
• - High Res 0.09, 0.03
• Domain
• - Control 216x432, 60x100
• - High Res 432x862, 118x198
• Time Step
• - Control 54, 18 (sec)
• - High Res 27, 9 (sec)
• Ocean Coupling
• - Coupled with HYCOM (resolution remains the
  same)
• - Coupling time every 9 minutes
• Case Study
• - Hurricane RITA, starting from 2005092012

21
9 km
4.5 km
More banding
22
4.5km (HRES) HWRF somewhat more accurate
23
HWRF Accomplishments

• HWRF severely test in the active Atlantic Season.
  HWRF ran in a robust, timely fashion . HWRF
  competitive with best operational guidance.
• HWRF installed GWD and fixed sfc temperature
  issues
• HWRF working on LSM, sfc parameterizations, and
  higher resolution.

HWRF Plans

• Upgrade physics, initialization and test
  ensembles
• More extensive quantitative diagnostics
• Implement new ocean wave model

24
(No Transcript)
25
The NMM-WRF Modeling Systemhttp//www.dtcenter.or
g/wrf-nmm/users/

• Regional-Scale, Moving Nest, Atmospheric
  Modeling System.
• Non-Hydrostatic system of equations formulated on
  a rotated latitude-longitude, Arakawa E-grid and
  a vertical, pressure hybrid (sigma_p-P)
  coordinate.
• Advanced HWRF,3D Variational analysis that
  includes vortex reallocation and adjustment to
  actual storm intensity.
• Uses SAS convection scheme, GFS/GFDL surface,
  boundary layer physics, GFDL/GFS radiation and
  Ferrier Microphysical Scheme.
• Ocean coupled modeling system (POM GFDL).

26
Technical Details of Operational HWRF POM
Coupled System run by NCEP Central Operations
Total No. of working Scripts (.sh , .scr, .pl files) 34
NCO Job Scripts (trigger and queue the scripts through schedule maintenance software SMS) 29
NCO SMS scripts (driver scripts that provide arguments for .sh) 27
Parameter files (namelist files) 25
Working space required for running one 126-hr HWRF coupled forecast 50 GB
Output volume (archived) 4.5 GB
Total run time for end-to-end HWRF forecast 110 min. per forecast
Max. Resources required per forecast 80 processors (5 nodes on NCEP production machine) for 60 min.
Maximum Number of Forecasts 4 per cycle
27
Sporadic SLP noise

• Sea level pressure diagnostic
• Model or post processing ??
• Traced to grid movement

Noise
28
Eliminate SLP Noise

• Modify topographic smoothing zone
• Adjust mass fields
• No more Noise !

29
Background (2)

• The GFDL Slab LSM is the default in HWRF
• The initial soil moisture remains fixed in time
  during the HWRF forecast.
• Moreover, the initial conditions of soil moisture
  in the Slab LSM are a fixed field that never
  change throughout the year and thus are unable to
  capture antecedent soil moisture conditions.
• The Slab LSM does not predict the runoff response
  to HWRF precipitation forecasts, thus cannot
  predict streamflow from HWRF forecasts.

LGM
30
Why do we need GWD parameterization?
NWP models use grid-averaged (smoothed) terrain
data
Coarse resolution models ( gt 4km) cannot resolve
the GWD caused by subgrid scale topography
31
HWRF Track Skill

• Competitive with other guidance
• Better than GFDL NGAPS
• GFS UKMET quite good
• Few long lasting storms in 2007
• EPAC HWRF not as good

HWRF
32
HWRF Intensity Skill

• Competitive with other guidance
• Some improvement over GFDL at early times
• Not a good year for dynamic models after
  accounting for landfall
• EPAC intensity degraded-no ocean coupling

HWRF