Global Climate and Weather Modeling

1
Global Climate and Weather Modeling

• John H. Ward
• NCEP Production Suite Review
• December 9, 2009

2
Outline

• GCWMB Responsibilities
• Global Model the Production Suite
• Most Recent Changes
• Upcoming Changes
• Dropouts

3
GCWMB Responsibilities

• GCWMB presentation will be broken into four
  separate parts
• Overview of recent changes and test results from
  upcoming implementations for
• Global Data Assimilation
• Global Forecast System
• Ensemble Forecast System
• Long Term planning
• Data Assimilation Steve Lord
• Climate Forecast System Suru Saha
• Ensemble Forecast System Yuejian Zhu

4
NOAAs NWS Model Production Suite
Oceans HYCOM WaveWatch III
Climate CFS
Hurricane GFDL HWRF
Coupled
MOM3
1.7B Obs/Day
Satellites 99.9
Dispersion ARL/HYSPLIT
Regional NAM WRF NMM
Global Forecast System
Global Data Assimilation
Severe Weather
Regional Data Assimilation
WRF NMM/ARW Workstation WRF
Short-Range Ensemble Forecast
North American Ensemble Forecast System
Air Quality
WRF ARW, NMM ETA, RSM
GFS, Canadian Global Model
NAM/CMAQ
Rapid Update for Aviation
NOAH Land Surface Model
5
Most Recent Changes

• Data Assimilation
• Global GSI Upgrades 2Q09
• Ensemble Forecast System
• NAEFS Upgrade 1Q08
• Climate
• CFS/GODAS 1Q08
• Global Forecast Model
• GFS Upgrade 3Q07

6
2Q09 GSI Upgrade

• Addition of METOP/IASI data
• Variational Quality Control
• Change in land/snow/ice skin temperature variance
• New Background error covariances
• Reduce number of AIRS water vapor channels
• New version and coefficients for CRTM
• Modification of height assignment for height
  based wind observations
• Retune observational errors
• Modification of surface land use file
• Situation Dependent Background Variances

7
1Q08 GEFS Upgrade

• Bias corrected GFS forecast
• Use the same algorithm as ensemble bias
  correction up to 180 hours
• Combine bias corrected GFS and ensemble forecast
• Dual resolution ensemble approach for short lead
  time
• GFS has higher weights at short lead time
• NAEFS new products
• Combine NCEP/GEFS (20m) and CMC/GEFS (20m)
• Produce Ensemble mean, spread, mode, 10
  50(median) and 90 probability forecast at 11
  degree resolution
• Climate anomaly (percentile) forecasts also
  generated for ens. mean
• Statistical downscaling
• Use RTMA as reference - NDGD resolution (5km),
  CONUS only
• Generate mean, mode, 10, 50(median) and 90
  probability forecasts

8
1Q08 CFS Upgrade

• Deep water GODAS
• Extend the GODAS assimilation to 2175 meters
• Correct a temperature bias in the global
  intermediate waters
• CFS upgrade
• Reduce the 8-day lag in the initial conditionsto
  a 1-day lag for both ocean and atmosphere.
• Introduce 2 new members (T62L64) out to 9 months.
  These 2 new members would initiate from perturbed
  initial conditions similar to the current 2
  members.
• Both these upgrades aim to improve upon the week
  3-6 / monthly forecast leads

9
3Q07 GFS Upgrade

• Observation changes
• Full resolution AIRS
• COSMIC GPSRO
• Analysis changes
• GSI (Gridpoint Statistical Interpolation)
• Physics changes
• Modularized radiation package
• Dynamics changes
• Hybrid sigma-pressure vertical coordinate
• Post processing and product changes
• Output hourly GDAS files
• Change to internal model history file
• More fields output in model flux file

10
Upcoming changes

• GSI/GFS Fall Bundle
• Implementation Scheduled for 12/15/09
• GEFS Resolution Increase
• NCO parallel 30-day test expected to begin
  12/15/09
• GFS Shallow Convection Q2
• Initial parallel testing at T382 shows some
  degradation in key parameters, additional tests
  required
• GFS Downscaled Numerical Guidance Q2
• GFS Resolution Increase Q3
• Considering T574L64 Eulerian possibly T878L64
  Semi-lagrangian
• CFS V2.0
• Reanalysis Complete
• Reforecasting to begin on Mist Dew
• GEFS Initial Perturbations Q4
• NAEFS Q4

10
11
GSI Changes

• Adding new observation data sources.
• Tropical storm pseudo sea-level pressure obs
• NOAA19 hirs/4,AMSU-A, MHS brightness temp obs
• NOAA18 sbuv/2. Monitor N19 GOME, and OMI ozone
  (no assimilation)
• RARS (currently only EARS) 1B data
• EUMETSAT-9 atm motion vectors
• Implementing improved techniques in GSI analysis.
• Use uniform thinning mesh for brightness temp
  data
• Improvements to assimilation of GPS RO data (QC,
  retune ob errors, improved forward operator )
• Add dry mass pressure constraint
• Merge GMAO EMC codes
• Update background error covariance
• Proper use of different spectral truncation
  between background and analysis

12
GFS Changes

• Restructure the Global Model code
• Code unification between GFS GEFS
• Upgrade to ESMF 3.1.02p
• Modify low cloud definition
• Introduce more accurate algorithm for several
  diagnostic variables
• Output additional parameters for TIGGE ICAO
• Output additional hourly files to 12 hours for
  RTMA Guam
• Output additional 3 hourly files to 60 hours for
  GOGART
• Will provide the ability to run a Null Test as
  part of EMC testing
• Restructuring of parallel code in 2008 made it
  impossible to run a baseline test of parallel
  code that will produce bitwise compatible result
  with production. This makes it difficult to
  verify experiments are being configured properly.
  

13
Post Consolidation

• Currently two versions of the Global Model post
  processor are running in Production
• NCEP Post GFS, GEFS
• Global Post GDAS
• NCEP Post will replace Global Post in GDAS
• Reduce code maintenance
• NCEP Post will be a component of NEMS, which will
  allow for a consolidation of post processing
  across NCEP modeling systems

14
Expected Benefits

• Data Assimilation
• Improve tropical storm track intensity
  forecasts
• Slight incremental improvement in overall
  forecast accuracy
• Model
• Facilitate testing and implementation of future
  upgrades
• Reduce code maintenance
• Post
• Eliminate the need to maintain multiple versions
  of the Global Post
• Improvements expected to be small and primarily
  due to the assimilation changes

15
GFS Changes only
Full Package
Improvements are small and primarily due to the
assimilation changes
Difference
16
500 MB Anomaly CorrelationJune 08 September 09
Northern Hemisphere
Southern Hemisphere
17
Precipitation Scores June 08 September 09
Day-1
Day-2
Day-3
18
Fit-to-Obs Anl GuessJune 08 September 09
Significant Improvement
19
500 MB Anomaly Correlation2008 Hurricane Season
Northern Hemisphere
Southern Hemisphere
Significant Improvement
20
500 MB Anomaly CorrelationWinter 2008/09
Northern Hemisphere
Southern Hemisphere
Significant Improvement
21
Tropical Vector Wind RSME2008 Hurricane Season
22
GFS Atlantic Hurricane Track Error2008 Hurricane
Season
GSI/GFS Bundle Red Operational GFS - Green
23
GFS Atlantic Hurricane Intensity Error2008
Hurricane Season
GSI/GFS Bundle Red Operational GFS - Green
24
GFDL Atlantic Hurricane Track Error2008
Hurricane Season
Small number of cases 50
25
GFDL Atlantic Hurricane Track Error2008
Hurricane Season
26
HWRF Atlantic Hurricane Track Error2008
Hurricane Season
All test as good or better than Production except
_at_ 120 hrs
Production
27
HWRF Atlantic Hurricane Intensity Error2008
Hurricane Season
28
GFS EPAC Hurricane Track Error2008 Hurricane
Season
GSI/GFS Bundle Red Operational GFS - Green
29
GFS EPAC Hurricane Intensity Error2008 Hurricane
Season
GSI/GFS Bundle Red Operational GFS - Green
30
HWRF EPAC Hurricane Track Error2008 Hurricane
Season
HWRF 2008 with new GFS shows marked improvement
after 72 hours
HWRF 2009 with new GFS shows degradation until
120 hrs
31
HWRF Atlantic Hurricane Track Error2009
Hurricane Season
32
2Q10 GEFS Upgrade

• Continue using current operational (n-1) GFS
  model
• Upgrade horizontal resolution from T126 to T190
• 4 cycles per day, 201 members per cycle
• Up to 384 hours (16 days)
• Use 8th order horizontal diffusion for all
  resolutions
• Improved forecast skills and ensemble spread
• Upgrade to ESMF (Earth System Modeling Framework)
  for GEFS Version 3.1.0r
• Allows concurrent generation of all ensemble
  members
• Needed for efficiency of stochastic perturbation
  scheme
• Add stochastic perturbation scheme to account for
  random model errors
• Increased ensemble spread and forecast skill
  (reliability)
• Add new variables (26 more) to pgrba files
• Based on user request
• From current 52 (variables) to future 78
  (variables)
• For NAEFS ensemble data exchange

33
NAEFS future configuration Updated December 2008
34
Horizontal resolution change Ensemble control
only (deterministic) From T126 to T190 NH 500hPa
geopotential height
26 4
Gains from short waves
35
Horizontal diffusions Ensemble controls only
OPR(T126)-4th order NHD(T126)-8th order
May 2007
November 2007
36
Resolution and Diffusion for Global Ensemble
E20s T126 4th for all 16d (oper.) E20x T190
8th out to 16d E20e T190 8th (0-180h), then
T126 4th
When reducing resolution from T190 (8th order) to
T126 (4th order), the ensemble forecast
probabilistic skill score tends to t126
immediately, the example shows here for tropical
850hPa temperature. 8th order diffusion for t126
somewhat improves performance (not show here).
Therefore, both the resolution and diffusion play
an important role here.
37
Latest retrospective run (full package)
NH 2-m temperature RMSE Spread
NH 500hPa height RMSE Spread
E20s T126L28 E20g T190L28 (0-180 only)
SH 500hPa height CRPSS
NH 500hPa height CRPSS
38
CRPSS for NH 850hPa temperature
OPER
OPER
EXP
EXP
Extend current 5-day skill to 6-day
Extend current 5-day skill to 6.5-day
39
NH Anomaly Correlation for 500hPa HeightPeriod
August 1st September 30th 2007
GEFSg is better than GFS at 48 hours
GEFSg could extend skillful forecast (60) for 9
days 24 hours better than current GEFS 48 hours
better than current GFS
40
Conclusion

• Based on two sets of retrospective runs (summer
  and winter 2007)
• New package improved the forecast skill (score)
  significantly
• For deterministic (ensemble mean)
• For probabilistic (ensemble distribution)
• The better results is benefited from
• Increase horizontal resolution (include
  diffusion)
• Stochastic perturbation scheme
• Better initial condition (analysis)
• Better forecast model (GFS)

41
GFS Shallow Convection

• New Boundary Layer Scheme
• Include stratocumulus-top driven turbulence
  mixing.
• Enhance stratocumulus top driven diffusion when
  condition for cloud top entrainment instability
  is met.
• Use local diffusion for the nighttime stable PBL.
• Background diffusion in inversion layers below
  2.5km over ocean is reduced by 70 to decrease
  the erosion of stratocumulus along the costal
  area.

42
GFS Shallow Convection

• New mass flux shallow convection scheme
• Detrain cloud water from every updraft layer
• Convection starting level is defined as the level
  of maximum moist static energy within PBL
• Cloud top is limited to 700 hPa.
• Entrainment rate is given to be inversely
  proportional to height and detrainment rate is
  set to be a constant as entrainment rate at the
  cloud base.
• Mass flux at cloud base is given to be a function
  of convective boundary layer velocity scale.

43
GFS Shallow Convection

• Updated deep convection scheme
• Eliminate Random cloud top, and cloud water is
  detrained from every cloud layer of the single
  cloud.
• Finite entrainment and detrainment rates for
  heat, moisture, and momentum are specified.
• Similar to shallow convection scheme, entrainment
  rate is given to be inversely proportional to
  height in sub-cloud layers and detrainment rate
  is set to be a constant as entrainment rate at
  the cloud base.
• Above cloud base, an organized entrainment is
  added, which is a function of environmental
  relative humidity.

44
GFS Shallow Convection

• Some degradation in key verification stats
• SH 500 MB AC
• Precip Threat
• Tropical Vector Wind
• Hurricane Tracks
• Reruns with additional tuning show improvement
• increased momentum background diffusivity
• convective overshooting
• increased cloud water detrainment in upper cloud
  layers

45
(No Transcript)
46
(No Transcript)
47
(No Transcript)
48
500 MB Anomaly Correlation
Northern Hemisphere
Southern Hemisphere
Some degradation In SH
49
500 MB Anomaly Correlation(Rerun with additional
tuning)
Northern Hemisphere
Southern Hemisphere
50
Precipitation Scores
Day-1
Day-2
Day-3
Higher Bias and lower Threat For low precip
amount
51
Precipitation Scores
Original
With Additional Tuning
52
Fit-To-Obs
Higher Temp Bias throughout Troposphere
53
Fit-To-Obs(Rerun with additional tuning)
Some improvement in RMSE Bias
54
Tropical Vector Wind RSME
Higher Upper Level RSME
55
Tropical Vector Wind RSME(Rerun with additional
tuning)
56
GFS Atlantic Hurricane Track Error2008 Hurricane
Season
Shallow Convection Red Fall GFS Bundle -
Green Operational GFS - Blue
57
GFS Atlantic Hurricane Track Error2008 Hurricane
Season
Shal Conv with additional tuning Red Fall GFS
Bundle - Green Operational GFS - Blue
58
GFS EPAC Hurricane Track Error2008 Hurricane
Season
59
GFS Downscaled Numerical Guidance

• The NAM smartinit code has been adapted to work
  for GFS to produce downscaled GFS surface fields
  for up to eight days.
• The methodology involves
• Horizontal interpolation from 35 km GFS native
  Gaussian grid to 2.5 km NDFD grid for Guam and
  Hawaii
• Vertical interpolation from model levels onto
  high resolution terrain using smartinit

60
GFS 12 h surface temperature forecast over Hawaii
before (L) and after (R) downscaling
61
GFS 12 h surface temperature forecast over Guam
before (L) and after (R) downscaling
62
GFS Resolution Increase

• Resolution
• T574L64 for fcst1 (0-180hr) and T382L64 for fcst2
  (180-384 hr)
• ESMF 3.1.0rp2
• Radiation and cloud
• Gravity-Wave Drag Parameterization
• Removal of negative water vapor
• Snow analysis
• Post processing and misc

63
500 MB Anomaly Correlation
Northern Hemisphere
Southern Hemisphere
64
500 MB Anomaly Correlation
65
Precipitation Scores June 08 September 08
Day-1
Day-2
Day-3
T574 Red T574 with Shallow Conv Green 2008
Operational GFS Black
66
GFS Atlantic Hurricane Track Error2008 Hurricane
Season
T574 Red December 09 GFS - Green 2008
Operational GFS - Blue
67
GFS EPAC Hurricane Track Error2008 Hurricane
Season
T574 Red Current Operational GFS - Green 2008
Operational GFS - Blue
68
GFS EPAC Hurricane Track ErrorHurricane Elida
Dropout Date in this period
T574 Red December GFS - Green 2008 Operational
GFS - Blue
69
Tropical Vector Wind RSME
70
Fit-To-Obs
71
Dropouts

• Definition
• Sudden dramatic reduction in model performance
  determined by the Day-5 500 mb Anomaly
  Correlation
• Day-5 score lower than 0.7

72
GFS Low Skill Score Investigations

• Dropout team meets regularly (weekly with NCEP
  director and quarterly with the NCEP director).
• Participation NCO, SDM, NESDIS, COPC ,NRL,
  FNMOC, Joint Center and GSI team
• Evaluate the effect of conventional and
  non-conventional observations in low skill score
  events
• Define a climatology (verification and web page
  stats) of dropout events (Composites)
• Detect/correct problems and improve quality
  control (QC)
• Delivery, status of networks (Counts, RTDMS)
• Bias in reports (ACARS)
• Make controlled experiments using ECMWF analysis
  to determine time and space location of dropouts,
  controlling observational input by ECM runs and
  manipulating prepbufr file.
• Test observations types for their contribution
  and perhaps gain insight into their interaction

73
SH Dropout 2009081200 (IC date)
74
Conventional Obs Exps
Sat Radiance Exps
GFS, GDAS CNTRL, ECMWF and ECM are respectively
the 5-day AC skill of GFS Operations, 6-hour
cycled fnl analysis, ECMWF operations and GFS
model run from ECMWF analysis. GDAS CNTRL
experiments on the PREPBUFR file only remove the
specified data type and assimilate the rest of
the observations.
GDAS CNTRL NO6TYPES Experiment improves 5-day AC
score (0.72) by removing SATWND, SFCSHP, WDSATR,
AIRCAR, ADPSFC, and SPSSMI. By adding AMSUA data
to this list our score improves to 0.79 (up 28
pts from the GDAS CNTRL. New GSI improves the
score to 0.83.
75
10 day Dropout Composite Satellite Radiance
Impact on NH and SH 500 mb AC
In the NH, satellite radiance data has a positive
impact on on 3-day and 5-day forecasts however,
in the SH the conventional observations in the
PREPBUFR file show a negative impact on 3 and
5-day forecasts which is causing the satellite
radiance experiments to have negative impact as
well for the 10 cases used in this composite.
Next slide begins exploration of PREPBUFR
conventional observation contributions.
76
Development of Standard Procedures for GFS
DropoutsBest Practices

• Plot horizontal and vertical maps of differences
  between the GFS vs. ECMWF models (FCST-ANL and
  FCST-FCST)
• Plot maps of GFS ANL-GES differences using
  graphical package that plots regional maps for
  high-resolution analysis
• Run GSI/GFS using GDAS data dump to see if
  extended time window improves forecast
• Run GSI/GFS using ECMWF pseudo-observations as
  initial conditions
• Run GSI/GFS using ECMWF pseudo-observations as
  initial conditions over specific regions where
  GFS-ECMWF differences are largest
• Run GSI/GFS using GDAS data dump and withhold
  data types within PREPBUFR file and satellite
  radiances
• Run the Eady Baroclinicity Index package to
  determine areas of large instability which helps
  understand model predictability
• Run GSI/GFS with the ECM run as the background
  guess
• In depth analysis of all observations in the
  vicinity of an area where the forecasts between
  the GFS and ECMWF diverge. Also, go back to
  previous cycles before the dropout to find out if
  any of the earlier observations contributed to
  problems within the area
• Contact SDMs to find out if any observations
  were missing or problematic at the forecast cycle
  time
• Save all the GFS/GDAS critical data type
  (conventional and non conventional) counts and
  its running monthly mean counts time series for
  any missing/shortage of data from the NCO's Real
  Time Data Monitoring System (RTDMS) web site

77
Questions