Transcript The NCEP North American Mesoscale Modeling System: Recent

The NCEP North American
Mesoscale Modeling System:
Recent Changes and Future Plans
Eric Rogers Geoff DiMego Thomas Black Michael Ek
Brad Ferrier George Gayno Zaviša Janjić Ying Lin
Matthew Pyle Vince Wong Wan-Shu Wu
Mesoscale Modeling Branch, EMC/NCEP
Jacob Carley
Perdue University
AMS 23st WAF / 19th NWP Conference, Omaha, NE
1 June 2009
Where the Nation’s climate and weather services begin
Overview of Presentation
 Summarize
March 2008 and December
2008 NAM implementations
 Discuss some of the specific changes in
more detail
 Planned NAM changes in 2010
 More details can be found in extended
abstract and at underlined hyperlinks
PRE-2008 NAM HISTORY
 June
2006 : WRF-NMM and GSI analysis
replace Eta model / 3DVAR analysis in the
NAM
 December 2006


Extensive tuning of convection / microphysics
New divergence damping scheme with extra
damping of external mode gravity waves; 5x
higher damping of both internal and external
modes during NDAS to reduce noise in
analysis increments
MARCH 2008 NAM CHANGES


Expanded (~18%) computational domain
Model changes




Gravity wave drag / mountain blocking
Unified Noah Land-surface physics
Modified advection of passive substances (q, cloud water, TKE)
Two minor changes in radiation
• Improved computation of surface longwave radiation
• Fix bug for climatological O3

Analysis changes




Retuned background error covariances
New observations : AIRS & GOES SFOV radiances, EUMETSTAT
& MODIS satellite winds, surface mesonet winds from sites on
NCEP use list
Adjust NDAS soil moisture outside of CONUS using 00Z
12-36 h NAM QPF
Smoother model terrain
Gravity Wave Drag & Mountain Blocking
(GWD)

“Mountain blocking” (Lott & Miller, 1997; Alpert 2004)



Wind flow around subgrid orography
Low-level flow is blocked below a dividing streamline (air
flows around, not over barrier)
Gravity wave drag (Alpert et al., 1988, 1996; Kim &
Arakawa,1995)
 Mountain wave stress, pressure drag
 Vertical distribution of the wave stress,
changes winds aloft (momentum deposition)

Parallel testing : GWD can improve NAM synoptic
forecasts, greatest impact seen on upper troposphere
winds/height and lower troposphere temperatures
Impact of GWD: 12Z 20 Dec 2006 Runs: 72-h
SLP (500 hPa and QPF in extended abstract)
GWD
Control
Control
1016 hPa
1024 hPa
1032 hPa
In GWD run:
• High pressure breaks down over Canada & New England
• Low pressure forms over western NC and SC
Example of the impact of GWD on lower tropospheric vertical
profiles : valid 12z 10/28/07
Black= Omaha (OAX)
raob
Red= Ops NAM 48-h fcst
Blue = Parallel NAM (with
GWD) 48-h fcst
Model soundings are from
grid point closest to OAX
Much better
depiction of
low-level
inversion in
GWD run
Hurricane Dean: 24-h SLP forecast valid 12z 18 Aug 07
OPS NAM
Pll NAM
TPC Best Track Position (Franklin, 2008)
Hurricane Dean: 48-h SLP forecast valid 12z 19 Aug 07
OPS NAM
Pll NAM
TPC Best Track Position (Franklin, 2008)
Hurricane Dean: 72-h SLP forecast valid 12z 20 Aug 07
OPS NAM
Pll NAM
TPC Best Track Position (Franklin, 2008)
Hurricane Dean: 84-h SLP forecast valid 00z 21 Aug 07
OPS NAM
TPC Best Track Position
(Franklin, 2008)
Pll NAM
TPC predicted position (issued 12z
8/17) valid 06z 8/21
DECEMBER 2008 CHANGES


NDAS “Partial Cycling”
Model Changes




Analysis Changes



Vertically mix/diffuse each hydrometeor species
Radiation change : double absorption coefficients for
water and ice in clouds
Land-sfc model changes to address problematic 2-m
dew point temps over snow cover
New observations : MetOp-a radiances and
TAMDAR/AMDAR aircraft data
New version of radiative transfer model w/default
climatology
Use hi-res (23 km) AFWA snow depth analysis
NDAS Configuration

12-h spin-up; 4 successive analyses & 3-h model
forecasts; last 3-h forecast is first guess for NAM
analysis
 Prior to July 1998, 12-h NDAS was started from a 6h forecast from the NCEP Global Data Assimilation
System (GDAS)
 Since July 1998, NDAS was fully cycled (3-h model
forecast from previous cycle used to initiate NDAS)
 Rationale for full cycling

Eliminates adjustments/inconsistencies if a model forecast
with different characteristics (resolution, physics) is used
as a first guess…..but
NDAS Configuration

GFS tends to outperform NAM
 WRF-NMM forecasts initialized off GFS initial
conditions sometimes produce better synoptic-scale
forecasts


Issues still exist with respect to “noisy” analysis
increments in the NDAS
Not practical to run ops NAM off global analysis
(GFS runs after NAM, NDAS is necessary to
provide fully cycled soil states)
 Revisit use of 6-h GDAS forecast to provide
atmospheric first guess states to initiate the NDAS
while using fully cycled NDAS soil states (NDAS
Partial Cycling, or “NPCY”)
Impact of NDAS Partial Cycling
during parallel testing

Tended to improve NAM forecast (sometimes
more “GFS-like”, especially at day 2-3)
 Occasional significant improvement in QPF over
ops NAM
 Occasionally, parallel NAM would do worse than
ops NAM when GFS had a “dropout” forecast
(see example in extended abstract)
 NPCY method implemented in Dec 2008 bundle,
is considered temporary while more advanced
techniques to control noise in the NDAS mature,
such as digital filter initialization (Colón, paper
18A.2)
Pll NAM w/NCPY

72-h forecast – GDAS anl
500 hPa height difference
valid 00z 8 Oct 08
 NPCY NAM has lower
errors associated
w/trough over central
US and Gulf of Alaska
Ops NAM
Ops GFS
24-h QPF valid 12z 8 Oct 2008
72-h Pll NAM forecast w/NPCY
72-h Ops NAM forecast
 Parallel
NAM improves 24-h
QPF across Iowa, Northern
Illinois, Wisconsin and U.P. of
Michigan
 Parallel NAM shifts heaviest
precip further east into central
KY/TN than ops NAM
CPC 1/8 deg observed pcp analysis
Impact of March 2008 changes and partial cycling :
Vertical profile of cumulative RMS errors from
December 2007 – March 2008
- Solid = Ops NAM
- Dash = Pll NAM
with 3/08 changes
- Dash-dot = Pll
NAM with 3/08
changes + NDAS
Partial Cycling
Black = 24-h Fcst
Red = 48-h Fcst
Blue = 72-h Fcst
Temp
Height
RH
Vector
Wind
Impact of radiation change

12 February 2008 Mid-Atlantic Freezing Rain Case; Poor
forecast by NAM for DC area
2-m Temperature (°F): Black line is 32oF isotherm
Significant
Cold Air
Damming
RUC Analysis of 2-m T at 00z 13
Feb 2008
CTL
Control WRF-NMM 12-hr forecast of 2m T valid 00z 13 Feb 2008
Impact of radiation change

Increase cloud optical
depths by doubling
absorption coefficients of
water and ice in clouds
 Intended effect is to reduce
the amount of incoming solar
radiation reaching the
ground and warming the
surface, thus keeping the
surface slightly cooler
beneath cloud
Result: more
pronounced damming
signature
Experimental WRF-NMM 12-hr
forecast of 2-m T valid 00z 13 Feb
2008
NAM 2010 Plans

December 2008 bundle is the last major set of
changes to WRF-based NAM
 NAM changes planned in fall 2010






Change to ESMF-based NOAA Environmental
Modeling System (NEMS) framework (Black, paper
2A.6)
Replace E-grid NMM with B-grid NCEP Nonhydrostatic
Multiscale Model (NMMB, Janjić, paper 5A.1)
Use physics packages developed for WRF
Parent NAM domain will remain at 12km resolution and
84-h forecast range
Add ~3-6 km runs nested in NAM over CONUS and
Alaska, out to 36-48 h (try to match or better NDFD grid
resolution)
Possibly add ~3 km nests over Hawaii/Puerto Rico