TC Genesis, Track, and Intensity Forecating

1
TC Genesis, Track, and Intensity Forecating

• Todd Kimberlain
• NOAA/NWS/NCEP/HPC

2
THE SHIPS MODEL
Statistical/dynamical model relating tropical
cyclone intensity change to various
climatological, persistence, and environmental
predictors.

• () SST POTENTIAL (VMAX-V) Difference between
  the maximum potential intensity (depends on SST)
  and the current intensity.
• (-) VERTICAL (850-200 MB) WIND SHEAR Current
  and forecast.
• () PERSISTENCE If its been strengthening, it
  will probably continue to strengthen, and vice
  versa.
• (-) UPPER LEVEL (200 MB) TEMPERATURE Warm
  upper-level temperatures inhibit convection

3
TROPICAL CYCLONE INTENSITY FORECAST MODELS

• Statistical Models
• Decay SHIFOR (Statistical Hurricane Intensity
  FORecast with decay).
• Based on historical information - climatology and
  persistence (uses CLIPER track).
• Measure of skill of intensity forecasts
• Statistical/Dynamical Models
• SHIPS (Statistical Hurricane Intensity Prediction
  Scheme)
• Based on climatology, persistence, and
  statistical relationships to current and forecast
  environmental conditions.
• DSHIPS (Decay SHIPS)
• Same as SHIPS except when track forecast points
  are over land when a decrease in intensity
  following an inland decay model is included.
• Dynamical Models
• GFDL, GFS, UKMET, NOGAPS.
• Based on the present and the future by solving
  the governing equations for the atmosphere (and
  ocean).

4
THE SHIPS MODEL (cont.)
Statistical/dynamical model relating tropical
cyclone intensity change to various
climatological, persistence, and environmental
predictors.

• () THETA-E EXCESS Related to buoyancy (CAPE)
  more buoyancy is conducive to strengthening
• () 500-300 MB LAYER AVERAGE RELATIVE HUMIDITY
  Dry air at mid-levels inhibits strengthening
• () 850 MB CIRCULATION TENDENCY Tangential wind
  change of global forecast models representation
  of the tropical cyclone within 6º radius of
  the center (new for 2007).
• (-) ZONAL STORM MOTION Intensification is
  favored when TCs are moving west

5
THE SHIPS MODEL (cont.)
Statistical/dynamical model relating tropical
cyclone intensity change to various
climatological, persistence, and environmental
predictors.

• (-) STEERING LEVEL PRESSURE intensification is
  favored for storms that are moving more with the
  upper level flow. This predictor usually only
  comes into play when storms get sheared off and
  move with the flow at very low levels (in which
  case they are likely to weaken).
• () 200 MB DIVERGENCE Divergence aloft enhances
  outflow and promotes strengthening
• (-) CLIMATOLOGY Number of days from the
  climatological peak of the hurricane season
• () GOES cold IR Pixel Count - Tb standard
  deviation (measure of symmetry of deep convection
  around the center
• () OHC Ocean Heat Content from satellite
  altimetry (UM/NHC algorithm)

6
A STATISTICAL TECHNIQUE TO AID IN THE FORECAST OF
RAPID INTENSIFICATION
The 7 predictors used to estimate the probability
of Rapid Intensification (defined as an increase
in maximum wind speed of at least 25 kt over 24
h)
7
Little progress with intensity
8
(No Transcript)
9
INTENSE WARM CORE CAN BE 16 K WARMER THAN NORMAL
TROPICAL VALUES
10
(No Transcript)
11
VERTICAL WIND SHEAR
45000 ft
DEEP CONVECTION
30000 ft
20000 ft
10000 ft
EXPOSED CENTER
5000 ft
1000 ft
12

• Concentric
• Eyewall
• Cycle
• Black Willoughby (1992)

13
CAT 3
CAT 4
CAT 5
CENTRAL PRESSURE VS. TIME FOR HURRICANE
ALLEN, 1980 LARGE FLUCTUATIONS LARGELY DUE TO
EYEWALL REPLACEMENT CYCLES
14
FACTORS AFFECTING TROPICAL CYCLONE INTENSITY

• Sea surface temperature / upper ocean heat
  content.
• Environmental winds, esp. vertical wind shear.
• Trough interactions.
• Temperature and moisture patterns in the
• storm environment.
• Internal effects (e.g. eyewall replacement
  cycles).
• Interaction with land.

15


Shear
Instability
TROPICAL ATLANTIC

Genesis parameter

Moisture
Courtesy Mark DeMaria
16
LARGE-SCALE CONDITIONS ASSOCIATED WITH TC
FORMATION

• A PRE-EXISTING DISTURBANCE CONTAINING
  ABUNDANT DEEP CONVECTION
• WARM SST
• A SUFFICIENTLY UNSTABLE ATMOSPHERE DEEP
  LAYER OF MOIST AIR
• SMALL VERTICAL SHEAR OF THE HORIZONTAL WIND
• APPEARANCE OF CURVED BANDING FEATURES IN
  THE DEEP CONVECTION

17
LARGE-SCALE CONDITIONS ASSOCIATED WITH TC
FORMATION

• FALLING SURFACE PRESSURE 24-HOUR PRESSURE
  CHANGES OF USUALLY 3 MB OR MORE
• UPPER-TROPOSPHERIC ANTICYCLONIC OUTFLOW OVER
  THE AREA

18
INNER CORE MAY ORIGINATE AS A MID-LEVEL
(NEAR 700 MB) MESO-VORTEX THAT HAS FORMED IN
ASSOCIATION WITH A MESOSCALE CONVECTIVE
SYSTEM (MCS)
PRE-GORDON DISTURBANCE, 9/13/00 1145 UTC (24
HOURS PRIOR TO GENESIS)
19
Multiple mid-level mesoscale vortices during
genesis stage. (Reasor et al. 2005 J. Atmos. Sci.)
8/19/96
8/19/96
(Hurricane Dolly)
8/19/96
8/20/96
20
Madden-Julian Oscillation

• Discovered in the early 1970s by Roland Madden
  and Paul Julian.
• An eastward propagating wave that circles the
  globe in about 40-50 days involving tropical
  convection.
• Detected in the Outgoing Longwave Radiation (OLR)
  fields across the tropics.
• Later papers showed that it is an important
  modulator of TC activity, especially in the
  Pacific Ocean.

21
-Idealized Diagram of the 40-50 day Tropical
Intraseasonal Oscillation -Became known as the
Madden-Julian Oscillation in the late
1980s -Generally forms over the Indian Ocean,
strengthens over the Pacific Ocean and weakens
due to interaction with South America and cooler
eastern Pacific SSTs
(Madden and Julian 1972)
22
(No Transcript)
23
(No Transcript)
24
Daily Rainfall (mm)
25
MJO Effects in the Atlantic Basin

• The MJO can lose much of its strength before
  entering the Atlantic basin.
• In addition, the MJO is weakest during the late
  summer, near the peak of Atlantic activity.
• Western part of the basin most strongly affected
  (Maloney and Hartmann 2000).
• Mo (2000) showed the Atlantic basin is most
  active when tropical convection is suppressed in
  the Central Pacific Ocean and enhanced in the
  Indian Ocean.

26
Active MJO EOF and corresponding TS and H tracks

• Active MJO in the western Caribbean Sea and Gulf
  of Mexico produces more storms due to
• Increase in low-level convergence (ITCZ moves
  farther north)
• Low-level vorticity is also increased due to
  westerly low-level flow meeting easterly trades
• Upper divergence is stronger than average during
  the westerly phase, with a drop in shear as well

Inactive MJO EOF and corresponding TS and H tracks
Adapted from Maloney and Hartmann (2000)
27
Gray shades are positive OLR values
Dateline
When convection is suppressed in the western
Pacific Ocean, there are more tropical storm
formations in the Atlantic basin (Mo 2000).
28
African Easterly Waves

• Origins barotropic instability with AEJ
• Often appears as inverted V Riehl (1948)
• Roughly 60 of Atlantic TCs come from waves
• Max vorticity in low-levels (700-850) and
  decreases with height
• April/May-November period 3-4 days
• Propagation 10-15 kts rule of thumb 7 degrees
  longitude/day
• Wavelength 1500-2500km
• Well-defined pressure couplet

29
How to Track Easterly Waves

• Zeroeth order persistence and extrapolation
• Build a case
• Wind shifts pressure falls/rises sharp PW
  gradients in satellite imagery sharp theta-e
  gradients in analyses
• Follow distinct features in satellite imagery
• Use Hovmoeller diagrams, esp use of meridional
  component of the wind
• http//www.nhc.noaa.gov/index_station.shtml

30
Genesis Parameters

• Low-level vorticity 850mb circulation -
  determined from a line integral of the wind
  component tangent to the boundary of each 5 by 5
  degree area.
• Thermodynamic effect instability parameter
  CAPE ranges from 4 to 3
• VERTICAL INSTABILITY The vertical average
  temperature difference between the equivalent
  potential temperature of a parcel lifted from the
  surface to 200 hPa, and the saturation equivalent
  potential temperature of the environment, for
  each 5 by 5 degree area.
• GOES COLD PIXEL COUNT The percent of GOES-east
  channel 3 pixels colder than 40 degree C in each
  5 by 5 degree area. All full disk images within 3
  hours after and 6 hours before each synoptic time
  are include, so that this parameter represents
  the amount of sustained deep convection.
• http//www.ssd.noaa.gov/PS/TROP/genesis.html

31
SHIPS Output

• ATLANTIC SHIPS INTENSITY
  FORECAST
• GOES/OHC INPUT
  INCLUDED
• HELENE AL082006
  09/15/06 12 UTC
• TIME (HR) 0 6 12 18 24
  36 48 60 72 84 96 108 120
• V (KT) NO LAND 50 55 61 66 71
  78 82 86 86 87 87 87 86
• V (KT) LAND 50 55 61 66 71
  78 82 86 86 87 87 87 86
• V (KT) LGE mod 50 55 60 64 68
  75 80 82 83 82 82 83 84
• SHEAR (KTS) 7 8 5 4 8
  6 10 11 18 8 12 12 12
• SHEAR DIR 28 63 84 17 3
  48 328 296 303 258 262 213 236
• SST (C) 28.1 28.2 28.1 27.9 27.7
  27.4 27.4 27.7 28.1 28.3 28.4 28.5 28.6
• POT. INT. (KT) 139 140 139 135 133
  128 128 132 137 140 141 142 143
• ADJ. POT. INT. 137 136 133 129 125
  118 117 120 124 125 124 124 124
• 200 MB T (C) -53.0 -52.2 -52.6 -53.0 -52.8
  -52.2 -52.2 -52.3 -52.5 -52.1 -51.9 -51.3 -50.9
• TH_E DEV (C) 9 8 9 9 9
  10 10 10 9 9 9 9 9
• 500-300 MB RH 55 56 52 52 52
  51 49 53 53 52 52 50 46
• 850 MB VORT 110 103 94 77 77
  73 59 58 42 44 55 86 93

32

• INDIVIDUAL CONTRIBUTIONS TO INTENSITY CHANGE 6 12
  18 24 36 48 60 72 84 96 108 120
  --------------------------------------------------
  -------- SAMPLE MEAN CHANGE 1. 2. 3. 4. 6. 7. 8.
  9. 10. 11. 11. 12. SST POTENTIAL 1. 2. 3. 4. 6.
  6. 7. 7. 8. 8. 8. 8. VERTICAL SHEAR 1. 1. 3. 4.
  5. 7. 8. 8. 8. 8. 9. 9. PERSISTENCE 2. 3. 4. 5.
  5. 6. 5. 5. 4. 3. 1. 0. 200/250 MB TEMP. 0. 0.
  -1. -1. -2. -2. -3. -3. -4. -4. -5. -6. THETA_E
  EXCESS 0. 0. -1. -1. -1. -2. -2. -2. -3. -4. -5.
  -5. 500-300 MB RH 0. 0. 0. 0. 0. 0. 0. -1. -1.
  -1. -1. -1. 850 MB ENV. VORT. 1. 1. 2. 2. 3. 4.
  5. 5. 5. 5. 6. 7. 200 MB DIVERGENCE 0. 1. 1. 1.
  2. 3. 4. 4. 5. 7. 7. 7. ZONAL STORM MOTION 0. 0.
  1. 1. 1. 2. 2. 2. 2. 2. 3. 3. STEERING LEVEL PRES
  0. 0. 0. 0. 0. 0. 0. 1. 1. 0. 0. 0. DAYS FROM
  CLIM. PEAK 0. 0. 0. 0. 1. 2. 2. 3. 3. 4. 4. 4.
  --------------------------------------------------
  -------- SUB-TOTAL CHANGE 5. 10. 16. 20. 27. 32.
  36. 37. 39. 39. 39. 38. INTENSITY ADJUSTMENTS
  FROM SATELLITE INPUT 6 12 18 24 36 48 60 72 84 96
  108 120 ------------------------------------------
  ---------------- MEAN ADJUSTMENT 0. 0. 0. 0. 0.
  0. 0. 0. 0. 0. 0. -1. GOES IR STD DEV 0. 0. 0. 0.
  0. 0. 0. 0. 0. 0. 0. 0. GOES IR PIXEL COUNT 0. 1.
  1. 1. 1. 1. 0. 0. 0. -1. -1. 0. OCEAN HEAT
  CONTENT 0. 0. 0. 0. 0. -1. -1. -1. -1. -1. -1.
  -1. ----------------------------------------------
  ------------ TOTAL ADJUSTMENT 0. 1. 1. 1. 0. 0.
  0. -1. -2. -2. -2. -2. ---------------------------
  ------------------------------- TOTAL CHANGE (KT)
  5. 11. 16. 21. 28. 32. 36. 36. 37. 37. 37. 36.
  HELENE 9/15/06 12 UTC 2006 ATLANTIC RAPID
  INTENSITY INDEX ( 25 KT OR MORE MAX WIND
  INCREASE IN NEXT 24 HR) 12 HR PERSISTENCE (KT)
  10.0 Range-45.0 to 30.0 Scaled/Wgted Val .7/
  1.1 850-200 MB SHEAR (KT) 6.2 Range 42.5 to
  2.5 Scaled/Wgted Val .9/ .9 D200 (107s-1)
  38.8 Range-20.0 to 149.0 Scaled/Wgted Val .3/
  .4 POT MPI-VMAX (KT) 81.9 Range 8.1 to 130.7
  Scaled/Wgted Val .6/ 1.1 850-700 MB REL HUM ()
  70.6 Range 57.0 to 88.0 Scaled/Wgted Val .4/ .1
  area w/pixels lt-30 C 80.0 Range 17.0 to 100.0
  Scaled/Wgted Val .8/ .4 STD DEV OF IR BR TEMP
  21.8 Range 37.5 to 5.3 Scaled/Wgted Val .5/ .4
  Scaled RI index 4.3 Prob of RI 27 is 2.3 times
  the sample mean(12) Discrim RI index 4.4 Prob
  of RI 36 is 3.1 times the sample mean(12)

33
Unconventional Shear

• Westerly shear between 200-500 mb
• Easterly shear between 500-850mb

34
Westerly Shear with Easterlies

• Westerly 200-850 dominates the troposphere

35
Unconventional Shear

• Easterly shear between 200-500 mb
• Westerly shear between 500-850mb

36
Shear

• Negative influence impedes intensification
• Ventilation advects heat and moisture away from
  the tropical cyclone
• DeMaria (1996) shows that shear has a
  thermodynamic effect through tilting and
  stabilitization
• 8-10-kt shear could be enough to interrupt
  development but not appreciably weaken the TC
• 20-30-kt shear causes well-defined convective
  asymmetries (wave 1) and significant weakening

37
More on Shear

• When the shear changes sign, it must go through a
  minimum
• In a sheared flow, there is a natural restoring
  force that reduces the tilt of the TC in the
  vertical
• Resistance of a TC to vertical shear is a
  function of latitude, size, and intensity
• Changes in motion are usually indicative of not
  only a change in the vertical shear but likely
  the intensity
• Convection removed more than 1-degree from the TC
  center reflects an erosion of the warm-core
• Once warm-core has dissipated, regeneration is
  difficult and must follow process of stage 1 and
  2
• For weak to moderate shears, there tends to be a
  delayed response in the storms intensity

38
Maximum Potential Intensity

• Represents a theoretical upper bound on TC
  intensity based on available thermodynamic
  profiles and SSTs
• A small number of TCs realize their maximum
  potential intensity
• What factors prove to be limiting factors on
  intensity?

39
Deep Layer Mean

• (u850u500)/2350mb (u500u200)/2300mb/650mb
  DLM
• DLM Mean Steering Flow
• Well-correlated with TC motion
• Like a block of wood in a river of air
• http//cimss.ssec.wisc.edu/tropic/real-time/atlant
  ic/movies/wg8dlm5/wg8dlm5java.html

40
200 mb Velocity Potential fields one way to
track the MJO
Blue divergence Red convergence Center of the
blue area tracks the most upper divergence, which
is usually well-linked to thunderstorms
41

• Most genesis points are near or behind the
  upper- level divergence center.

42
Another way to track the MJO is looking at the
raw wind fields Diagram shows 850 mb zonal wind
anomalies for the past 6 months between 5S and
5N. Can be used to infer areas of low-level
large-scale convergence and divergence, as well
as trade wind surges or westerly wind bursts
which influences ENSO.
43
(No Transcript)
44
ATLANTIC PRESSURE-WIND RELATIONSHIPS
1) GLFMEX Vmax(kt)10.627(1013-p)0.5640 n
664 r0.991 2) lt25N Vmax(kt)12.016(1013-p)
0.5337 n 1033 r0.994 3) 25-35N
Vmax(kt)14.172(1013-p)0.4778 n 922
r0.996 4) 35-45N Vmax(kt)16.086(1013-p)0.43
33 n 492 r0.974 5)For Kraft
Vmax(kt)14.000(1013-p)0.5000 n 13 r ??
P(MB) GLFMEX lt25N 25-35N 35-45N
KRAFT P(MB) P(IN)
960 100 100 94 90
102 960 28.35

• Hurricanes with a small Radius of Maximum Winds
  (RMW) will typically have stronger winds than a
  system with the same central pressure but larger
  RMW.

45
Factors Affecting TC Motion

• Large-scale
• Vortex Moves with Steering Flow ? main
  contributor to TC motion
• Cyclone-scale
• Vortex induces beta-gyres and other asymmetries
  that affect motion
• Convective distribution
• Vertical Structure
• Other
• Binary interaction (Fujiwhara effect)
• Landmass interaction
• Internal dynamics (trochoidal motion)

46
The Large-scale Steering Flow is the Main
Contributor to TC Motion
47
(No Transcript)
48
The Beta Effect

• The circulation of a TC, combined with the
  North-South variation of the Coriolis parameter,
  induces asymmetries known as Beta Gyres.

INDUCED STEERING 1-2 m/s NW
HIGHER VALUES OF EARTHS VORTICITY
H
ßvgt0

• Beta Gyres produce a net steering current across
  the TC, generally toward the NW at a few knots.
  This motion is knows as the Beta Drift.

ßvlt0
L
N
LOWER VALUES OF EARTHS VORTICITY
49
Binary Interaction (Fujiwhara Effect)
Relative rotation diagram of 12-h positions
relative to the midpoints between Bopha (in solid
typhoon symbol) and Saomai (in solid dot) based
on a direct binary interaction interpretation (Wu
et. al, 2003)

• Fujiwhara effectOccurs when two tropical
  cyclones become close enough (lt 1450 km) to
  rotate cyclonically about each other as a result
  of their circulations' mutual advection.
• Named after Dr. Sakuhei Fujiwhara who initially
  described it in a 1921 paper about the motion of
  vortices in water
• Most often occurs in the northwestern and eastern
  North Pacific basinless often in the Atlantic.
• Presents a unique forecast challenge since the
  complex interplay results in different scenarios
  which determine the final result of the
  interaction
• Some of the factors affecting the outcome of
  binary interaction include comparable strength
  of the two cyclones, comparable size of the two
  cyclones, distance apart, background flow

50
Trochoidal Motion (Wobble)

• Related to inner-core structure, convective
  asymmetries, and dynamic instability
• Unable to forecast
• Simply observe
• Beware of the wobble
• Wait for a sustained (several hours) change in
  motion

51
Trochoidal Motion (Wobble)

• Related to inner-core structure, convective
  asymmetries, and dynamic instability
• Unable to forecast
• Simply observe
• Beware of the wobble
• Wait for a sustained (several hours) change in
  motion

52
Hierarchy of TC Track Guidance Models

• Statistical
• Forecasts based on established relationships
  between storm-specific information (i.e.,
  location and time of year) and the behavior of
  previous storms
• CLIPER
• Statistical-Dynamical
• Statistical models that use information from
  dynamical model output
• NHC91 still maintains skill in the eastern
  Pacific
• Simplified Dynamical
• LBAR simple two-dimensional dynamical track
  prediction model that solves the shallow-water
  equations initialized with vertically averaged
  (850-200 hPa) winds and heights from the GFS
  global model
• BAMD, BAMM, BAMS -gt Forecasts based on simplified
  dynamic representation of interaction with vortex
  and prevailing flow (trajectory)
• Dynamical Models
• Solve the physical equations of motion that
  govern the atmosphere
• GFDL, GFDN, GFS, NOGAPS, UKMET, ECMWF, NAM,
  (HWRF)

53
CLIPER (CLImatology and PERsistence) Model

• Statistical track model developed in 1972,
  extended to 120 h in 1998
• Required Input
• Current/12 h old speed/direction of motion
• Current latitude/longitude
• Julian Day, Storm maximum wind
• Average 24, 48, 72, 96 and 120 h errors 100,
  216, 318, 419, and 510 nautical miles
  respectively
• Used as a benchmark for other models and
  subjective forecasts forecasts with errors
  greater than CLIPER are considered to have no
  skill.

54
Beta and Advection Model (BAM)

• Method Steering (trajectories) given by
  layer-averaged winds from a global model
  (horizontally smoothed to T25 resolution), plus a
  correction term to simulate the so-called Beta
  Effect
• Three different layer averages
• Shallow (850-700 MB) - BAMS
• Medium (850-400 MB) - BAMM
• Deep (850-200 MB) - BAMD

55
WHICH BAM TO USE?
200 mb
Typical cruising altitude of commercial airplane
400 mb
700 mb
Surface
TROPICAL DEPRESSION
5,000 ft/850 mb
56
Primary Dynamical Models used at NHC

• U.S. NWS Global Forecast System (GFS) lt
  relocates the first-guess TC vortex
• United Kingdom Met. Office (UKMET) lt bogus (syn.
  data)
• U.S. Navy Operational Global Atmospheric
  Prediction System (NOGAPS) lt bogus (syn. data)
• U.S. NWS Geophysical Fluid Dynamics Laboratory
  (GFDL) model ltbogus (spinup vortex)
• GFDN- Navy version of GFDL ltbogus (spinup
  vortex)
• European Center for Medium-range Weather
  Forecasting (ECMWF) model (no bogus)

57
The U.K. Met. Office Model

• Non-hydrostatic global model
• 4-D VAR analysis scheme with bogus TC
• Arakawa C-grid east-west horizontal grid
  spacing of 0.5 longitude and a north-south grid
  spacing of 0.4 latitude (40 km at
  mid-latitudes)
• Hybrid vertical coordinate system with 50 levels
• In 2002, completely new formulation including
  new dynamical core, fundamental equations, and
  physical parameterizations
• Run twice daily at 0000Z and 1200Z producing
  forecasts for up to 144 hours (6 days)
• Intermediate runs at 0600Z and 1800Z, but only
  produce forecasts to 48 hours

58
The Global Forecast System (GFS)

• Global spectral model
• T382L64 ( 35-km horizontal grid spacing with 64
  vertical levels) through 180 hours.
• T190L64 ( 80-km grid spacing and 64 levels)
  180-384 hours
• Hybrid sigma-pressure vertical coordinate system
  (May 2007)
• Simplified Arakawa-Schubert (SAS) convective
  parameterization scheme
• PBL First-order closure method
• 3D-Var Gridpoint Statistical Interpolation (GSI)
  (May 2007)
• Rather than bogussing, the GFS relocates the
  first-guess TC vortex to the official NHC
  position.
• Often leads to an incomplete representation of
  the true TC structure
• Run four times per day (00, 06, 12, and 18 UTC)
  out to 384 hours

59
NOGAPS Model

• Global spectral model T239L30 (approximately
  55 km and 30 vertical levels)
• Hybrid sigma-pressure vertical coordinate system
  six terrain-following sigma levels below 850 mb
  and remaining 24 pressure levels occurring above
  850 mb.
• Time step is five minutes, but is reduced if
  necessary to prevent numerical instability
  associated with fast moving weather features.
• 3-D VAR analysis scheme
• Run 144 hours at each of the synoptic times.
• Emanuel convective parameterization scheme with
  non-precipitating convective mixing based on the
  Tiedtke method.
• Like other global models, the NOGAPS cannot
  provide skillful intensity forecasts but can
  provide skillful track forecasts.

60
ECMWF Model

• Considered one of the most sophisticated and
  computationally expensive of all the global
  models currently used by the NHC.
• Among the latest of all available dynamical
  model guidance.
• Hydrostatic global model T799L91 (approximately
  25 km and 91 vertical levels)
• Hybrid vertical coordinate system with as many
  levels in the lowest 1.5 km of the model
  atmosphere as in the highest 45 km.
• (4-D Var) analysis scheme
• Provides forecasts out to 240 hours (10 days).
  
• Even though there is no bogussing or relocation
  (i.e. no specific treatment of TCs in the
  initialization), the model produces credible
  forecasts of TC track.

61
THE GEOPHYSICAL FLUID DYNAMICS LABORATORY (GFDL)
HURRICANE MODEL

• Only purely dynamical model capable of producing
  skillful intensity forecasts
• Coupled with a high-resolution version of the
  Princeton Ocean Model (POM) (1/6 horizontal
  resolution with 23 vertical sigma levels)
• Replaces the GFS vortex with an axisymmetric
  vortex spun up in a separate model simulation
• Sigma vertical coordinate system with 42
  vertical levels
• Limited-area domain (not global) with 2 grids
  nested within the parent grid.
• Outer grid spans 75x75 at 1/2 resolution or
  approximately 30 km.
• Middle grid spans 11x11 at 1/6 resolution or
  approximately 15 km.
• Inner grid spans 5x5 at 1/12 resolution or
  approximately 7.5 km

62
(No Transcript)
63
THE HURRICANE WEATHER RESEARCH FORECASTING
(HWRF) PREDICTION SYSTEM

• Next generation non-hydrostatic weather research
  and hurricane prediction system
• Movable, 2- way nested grid (9km 27km/42L
  68X68)
• Coupled with Princeton Ocean Model
• 3-D VAR data assimilation scheme
• But with more advanced data assimilation for
  hurricane core (make use of airborne doppler
  radar obs and land based radar)
• Operational this season (under development since
  2002)
• Will run in parallel with the GFDL

64
HWRF GFDL
65
Ensemble Forecasts(Classic Method)

• A number of forecasts are made with a
  single model using perturbed initial
  conditions that represent the likely initial
  analysis error distribution
• Each different model forecast is known as a
  member model
• The spread of the various member models
  indicates uncertainty
• small spread among the member model may imply
  high confidence
• large spread among the member model may imply
  low confidence

66
GFS ENSEMBLE FOR RITA 9/19/05 12Z
67
Ensemble Forecasts (multi-model
method)

• A group of forecast tracks from DIFFERENT
  PREDICTION MODELS (i.e. GFDL, UKMET, NOGAPS,
  ETC.) at the SAME INITIAL TIME
• A multi-model ensemble is usually superior to an
  ensemble from a single model
• different models typically have different
  biases, or random errors that will cancel or
  offset each other when combined.
• The multi-model ensemble is often called a
  CONSENSUS forecast.
• Primary Consensus forecasts used at NHC
• GUNA
• CONU
• FSSE

68
Ensemble Forecasts (multi-model
method)

• GUNA a simple track consensus calculated by
  averaging the track guidance provided by the
  GFDI, UKMI, NGPI, and GFSI models. All four
  member models must be available to compute GUNA.
• CONU a simple track consensus calculated by
  averaging the track guidance provided by the
  GFDI, UKMI, NGPI, GFNI, and GFSI models. CONU
  only requires two of the five member models.
• FSSE The FSSE is not a simple average of the
  member models. Rather, the FSSE is constantly
  learning by using the performance of past member
  model forecasts along with the previous official
  NHC forecast in an effort to correct biases

69
2001-2003 Atlantic GUNA Ensemble TC Forecast
Error (nm)
619
358
176
Number of Forecasts
467
229
70
Excellent example of GUNA consensus HURRICANE
ISABEL, 1200 UTC 11 SEP 2003
71
Florida State Super Ensemble

• The limitation of such a technique occurs when
  the past performance of the member models does
  not accurately represent their present
  performance
• For example, the FSSE may have to relearn a
  particular models bias at the beginning of a
  season, after changes were made to that member
  model

72
Corrected Consensus
CONU and CCON Forecast Tracks Hurricane Daniel
00Z 20 July 2006

• Derived statistically, based on parameters known
  at the start of the forecast, such as model
  spread, initial intensity, location, etc.
• Can also be derived using historical biases of
  CONU or GUNA
• Typically a small correction

73
Continuity

• Changes to the previous forecast are normally
  made in small increments
• Official forecast typically trends in a given
  direction (left, right, slower, faster)
• Significant changes in the TC track forecast
  should be avoided since
• Models can shift back and forth from one cycle to
  the next
• Credibility can be damaged by making big changes
• Can confuse the public and/or generate over/under
  reaction
• Occasional exceptions (Katrina)

74
Piecing Together a Forecast

• Evaluate the large-scale synoptic environment
• Analyze in-situ and remotely sensed data before
  looking at model output
• Assess the steering pattern
• Compare observations to model initial fields
• Look for areas where the model fields do not
  match the observations garbage in equals garbage
  out
• Compare the conceptual model with the numerical
  model
• How might variations between the model analysis
  and current data affect the forecast
• Based on the large-scale pattern, what seems most
  reasonable?
• Interpret model tracks
• There is rarely a single Model of the Day so
  dont look for it
• Start with a consensus of high-quality dynamical
  models
• Consider past performance of each member (look at
  model trends)
• When possible, try a selected consensus based
  on a thorough analysis of all guidance
• Always Honor Continuity
• Avoid the WINDSHIELD WIPER effect

75
BAD INITIALIZATION FOR TROPICAL STORM GORDON
9/11/06 1200 UTC
76
Initial vortex too weak
Incorrect initial structure leads to a west bias
77
TRACK FORECAST IMPROVEMENTS IN THE NCEP GLOBAL
MODEL (GFS) DUE TO GPS DROPSONDES, 2000-2002
78
Impact of Dropsondes on Model Forecast
79
Think Conceptually

• Ask yourself what is happening in reality?
• What is happening in the model?
• Is the model forecast realistic?
• What are possible error mechanisms of a model
  (error mechanisms always exists)?
• How might these error mechanisms affect the
  forecast?

80
The Power of Latent Heating
1) A model cyclone which is too intense (weak)
leads to enhanced (limited) heating
Understand the convective parameterization scheme
81
CONCLUDING REMARKS (TRACK FORECASTING)

• ? Multi-level dynamical models are the most
  skillful models for TC track prediction, although
  simple trajectory models, such as BAMD, can still
  be useful.
• Consensus track forecasts such as the GUNA and
  CONU generally produce more skillful forecasts
  than any individual model.
• A selective consensus generated by intelligently
  evaluating each model can be effective but should
  be used carefully.
• ? The HWRF model is the next generation high
  resolution hurricane model which will transition
  into operations this year. GFDL and HWRF will be
  run operationally in parallel this year.

82
LARGE-SCALE CONDITIONS ASSOCIATED WITH TC
FORMATION

• A PRE-EXISTING DISTURBANCE CONTAINING
  ABUNDANT DEEP CONVECTION
• WARM SST
• A SUFFICIENTLY UNSTABLE ATMOSPHERE DEEP
  LAYER OF MOIST AIR
• SMALL VERTICAL SHEAR OF THE HORIZONTAL WIND
• APPEARANCE OF CURVED BANDING FEATURES IN
  THE DEEP CONVECTION

83
LARGE-SCALE CONDITIONS ASSOCIATED WITH TC
FORMATION

• FALLING SURFACE PRESSURE 24-HOUR PRESSURE
  CHANGES OF USUALLY 3 MB OR MORE
• UPPER-TROPOSPHERIC ANTICYCLONIC OUTFLOW OVER
  THE AREA

84
INNER CORE MAY ORIGINATE AS A MID-LEVEL
(NEAR 700 MB) MESO-VORTEX THAT HAS FORMED IN
ASSOCIATION WITH A MESOSCALE CONVECTIVE
SYSTEM (MCS)
PRE-GORDON DISTURBANCE, 9/13/00 1145 UTC (24
HOURS PRIOR TO GENESIS)
85
Multiple mid-level mesoscale vortices during
genesis stage. (Reasor et al. 2005 J. Atmos. Sci.)
8/19/96
8/19/96
(Hurricane Dolly)
8/19/96
8/20/96
86
Stage 1 and 2
87
CLIMATOLOGY OF TROPICAL CYCLONE FORMATION

• IN THE LONG-TERM MEAN, TYPICALLY, THERE IS
  A LAG BETWEEN THE OCCURRENCE OF THE MOST
  FAVORABLE THERMODYNAMIC CONDITIONS (IN TERMS
  OF STATIC STABILITY ) AND THE MOST
  FAVORABLE DYN AMICAL CONDITIONS (IN TERMS OF
  VERTICAL SHEAR).
• THE ATMOSPHERE TENDS TO BE MORE UNSTABLE
  LATER IN THE SEASON.
• THE VERTICAL SHEAR TENDS TO BE WEAKER
  EARLIER IN THE SEASON.

88
(No Transcript)
89
(No Transcript)
90
(No Transcript)
91
(No Transcript)
92
(No Transcript)
93
Identifying and tracking easterly waves

• Identify them over west Africa using satellite
  imagery looking for rotations in low/mid clouds
• Verify their passage over western Africa
  rawindsonde stations check for wind shifts
  using timesections
• Follow the rotating low/mid level clouds across
  the Atlantic
• Recognize the characteristics of the waves during
  that period wavelength, period, speed of
  propagation
• With the above, use continuity extrapolation
• Verify and adjust the locations using
  timesections from eastern Caribbean rawindsonde
  stations
• Use continuity/extrapolation again if you dont
  have timesections from east pacific

94
Verify the passage of the waves Time series of
RAOB data from Dakar
(v)
95
Relative humidity San Juan July 12 Aug 04,
2006
96
Meridiornal Wind Anomalies San Juan July
13-July 31, 2006 time-means at each level removed
97
Some characteristics of easterly waves

• Periodic
• Identified as a series of trough, occasionally
  with closed circulation
• Wavelength varies with time, location, and
  environment. (1500-3000 km)
• Propagate westward 5-8 degrees of longitude/day
• Likely to have convection near the trough/cyclone
  vortices (where the upward motion is located)
• Maximum amplitude at low-mid troposphere (700600
  mb) (second max. amp 300 mb)
• Cold core (updraft is colder than environment)
• Likely to have wind surge as the troughs pass by

98
USEFUL TIPS FOR A BETTER ANALYSIS

• Check the data/analysis over a longer period of
  time over a period that is comparable to the
  time scale of a synoptic system
• Time series analysis use temporal coverage for
  the lack of spatial coverage
• Check Cross sections for structural coherence
• Space-time analysis time series and structure
• Use wind analysis when possible both
  streamlines and isotaches

99
(Active)
(Inactive)
Negative vorticity anomaly
Positive vorticity anomaly
Huge changes in the wind and vorticity field are
noted between the active and inactive phases
100
Remotely Sensed Surface Winds
SMRF
101
INSTALLATION OF SFMRS ON AIR FORCE C-130
HURRICANE HUNTER PLANES ONGOING
102
Stepped-Frequency Microwave Radiometer

• Measures nadir brightness temperature at 6 C-band
  frequencies.
• Geophysical model function relates emissivity to
  wind speed. Emissivity depends on surface foam
  coverage and rain rate.
• Calibrated with GPS dropsonde data.
• First data from C-130s in 2007.

103
SFMR issues

• Shoaling breaking waves in areas of shallow
  water can artificially increase the SFMR
  retrieved wind and invalidate the observations.
• Interaction of wind and wave field can introduce
  azimuthally-dependent errors ( 5 kt).
• Rain impacts not always properly accounted for
  (mainly lt 50 kt).
• Calibration only recently completed. Algorithms
  still under development, and forecaster
  understanding of these issues is primitive.