Tropical Cyclone Processing for the CFS Reanalysis

1
Tropical Cyclone Processingfor the CFS Reanalysis

• Bob Kistler

2
(No Transcript)
3
NWP TC Review

• tcvital files
• Worldwide observations of TCs
• Tropical Storm Relocation
• TC in guess is located and bodily moved to
  tcvital location
• Bogus vortex winds
• If TC not found in guess, bogus vortex wind
  profiles are created to spin-up storm in GSI

4
tcvitals

• NHC 16L NOEL 20071101 0000 229N 0785W 360
  015 0995 1008 0334 26 093 0185 0185 -999 0185 D
• Formatted storm text message
• identification, location, movement, strength
• historical sources
• Mike Fiorinos bogus wind profiles
• HURDAT archive
• /nwprod archive
• Mike Fiorinos mdeck merge of 1,2,3

5
tcvitals

• type tcvcard ! Define a new type for a TC
  Vitals card
• sequence
• character4 tcv_center !
  Hurricane Center Acronym
• character3 tcv_storm_id ! Storm
  Identifier (03L, etc)
• character9 tcv_storm_name ! Storm
  name
• integer tcv_yyyymmdd ! Date
  of observation
• integer tcv_hhmm !
  Time of observation (UTC)
• integer tcv_lat
  ! Storm Lat (10), always gt0
• character1 tcv_latns !
  'N' or 'S'
• integer tcv_lon
  ! Storm Lon (10), always gt0
• character1 tcv_lonew ! 'E'
  or 'W'
• integer tcv_stdir -99 !
  Storm motion vector (in degr)
• integer tcv_stspd -99 !
  Spd of storm movement (m/s10)
• integer tcv_pcen -999 ! Min
  central pressure (mb)
• integer tcv_penv -999 ! val
  outermost closed isobar(mb)
• integer tcv_penvrad -999 ! rad
  outermost closed isobar(km)
• integer tcv_vmax -9 !
  max sfc wind speed (m/s)
• integer tcv_vmaxrad -99 ! rad
  of max sfc wind spd (km)

1 NHC 2 16L 3 NOEL 4
20071101 5 0000 6 229N 7 0785W 8
360 9 015 10 0995 11 1008 12 0334 13
26 14 093 15 0185 16 0185 17 -999 18
0185 19 D
6
Hurricane Initializations in NCEP Numerical
Models Qingfu Liu Acknowledgement Naomi
Surgi Steve Lord Hua-Lu
Pan EMC/NCEP/NOAA/SAIC
7
Brief History of Hurricane Forecast in NCEP
global model (GFS)

• Prior to 1995, no storm genesis in NCEP global
  model.
• Bogus data are added in 1989.
• tropical storms in the initial conditions
  defined by the bogus data weakened and becomes
  open wave in medium range forecast
• In Oct. 1995, the newly implemented version of
  the NCEP global model started to generate and
  deepen tropical storms. The new version not only
  forecast real storm genesis, but also created
  many false storms. The false storms passed to
  next cycle through first guess, and continue to
  develop and lead to many bad forecast in a row.
• In 2000, cloud momentum mixing was added to the
  simplified Arakawa-Schubert scheme, significantly
  suppressed the false storm formation

8

• Two vortices in the global analysis
• Storms existing in the first guess often are weak
  and large compared to the observed storms, and
  the center located away from the real storm
  center. Adding bogus storm data (and forcing the
  GSI to accept the data) will often lead to two
  vortices in the same region, one from the first
  guess and the other from the bogus data.
• This two vortices interact each other and often
  leads to very bad forecast.
• Prompted development of TC relocation

9
? Hurricane initialization in GFS GSI analysis
cycle at time t
GSI analysis at time t-6
9 hour forecast at time t-3, t, t3
Relocate hurricane vortices
t ? t6
GSI analysis at time t
5 or more day forecast
10
Hurricane Relocation in GFS

• Locate vortex center in t-3, t and t3 forecast
  fields
• ? TC vital message
• lon, lat, max surface wind, min central press,
  etc.
• ? Tracker software
• 700 850 mb max relative vorticity, min wind
  speed and min geopotential height
• sea-level press
• Separate hurricane vortex from environmental
  fields
• Relocate hurricane vortex to the observed storm
  position (divergence, vorticity, surface
  pressure, temperature, and mixing ratio)
• If the storm is too weak in the forecast fields,
  storm bogus data will be added in GSI analysis

11
Advantages Compared to Storm Bogus

• Removes the possibility of double vortices in the
  analysis field
• Guess field is more consistent with observation
  data, and less data are rejected
• Model generated vortex is more consistent with
  model dynamics, and less adjustment needed once
  forecast starts
• Hurricane dynamic structures, such as tilted
  hurricane vortex in shear or baroclinic
  environment, will be maintained in hurricane
  relocation

12
Bogus vortex winds

• Program SYNDATA
• requires full tcvital report
• i.e central p, max wind, radius max winds
• Will fail with incomplete HURDAT or Fiorino
  report
• 2nd execution performs QC near TC vortex
• Flag non-representative reports
• central pressure reports
• dropwindsone winds

13
(No Transcript)
14
TC processing in earlier reanalyses

• NCEP/NCAR I and II none
• Wind bogus tested produce double vortices
• NASA / DAO - none
• ERA-15 - none
• ERA-40, NCEP Regional Reanalysis
• Fiorino wind bogus was availalble
• Was not used
• JRA-25 Fiorino wind bogus was used

15
1. Fiorino Wind Bogus

• 6 hour frequency Jan 1979 Feb 2001
• All basins
• 5 bufr wind profiles
• 1000,925,850,700,500, 400 Hpa
• central location , /- 1 deg N,S,E,W
• Storm name same from storm id TC021 21L
• No tcvital strength information
• Neither radius of max winds nor max wind
• No central pressure
• Storm location has been converted to a tcvital
  report
• End result is a report pair
• tcvitl.gdas.yyyymmddhh - tcvital entry
• tcbogw.gdas.yyyymmddhh - 5 wind profiles

16
2. HURDAT

• 1850s present Best Track
• 4xdaily for reanalysis period 1978 - present
• Atlantic (L) and Eastern Pacific (E) only
• Uses historical names (e.g. Camille, Katrina)
• Has central pressure and max wind
• Lacks radius of max winds
• Storm id (03L) usually equates with Fiorino id
• Supercedes Fiorino tcbog report
• Has W added when tcbogw winds are available

17
3. NCEP tcvital production archive

• From 1997 on
• Complete tcvital reports
• Supercede HURDAT and Fiorino wind tcvital
• 1997 -2002 Indian, Pacific oceans spotty
• 1997-1998 non Y2K
• Multiple differing reports
• GFS (t245) then GDAS(t550)
• Correction of error or later info received
• Upgrade from depression to storm name
• Required manual inspection and editing

18
Bogus vortex winds

• New program TCBOG_SYNDATA
• Alternative to existing SYNDATA
• Runs identical QC of central p., drop. winds
• invoked for incomplete tcvital records
• Hurdat reports marked W
• Fiorino tcvital reports
• If relocation indicates vortex spin-up
• matches tcvitl and tcbogw reports
• Converts the tcbogw profiles for input to GSI

19
(No Transcript)
20
Fiorino mdeck

• Just received on Monday, 11/05/2007
• Awaiting comparison with existing sets
• Fills a gap in production archive
• missing Jan Mar 2002
• Merge of historical best tracks
• HURDAT/TCVITAL type files for all basins
• Contains historical names