Prediction of Tropical Cyclones

1
Prediction of Tropical Cyclones
2
Outline

• Tropical Cyclogenesis
• Convection in mature tropical cyclones
• Severe convection in landfalling tropical cyclones

3
Hurricane Community

• Very long history
• Ideas continually rediscovered
• Research and operational communities used to work
  closely together
• Many things are hotly debated
• Collision of theory and empiricism

4
Important Papers

• Eliassen 1952
• Charney and Eliassen 1964
• Ooyama 1964, 1969
• Gray 1968, 1979
• Shapiro and Willoughby 1982
• Schubert and Hack 1982
• Emanuel 1986
• Rotunno and Emanuel 1987
• Many recent articles

5
Characteristics

• Most form 5º-20º N or S gt Rotation needed
• Pre-existing disturbance required
• Easterly Waves
• Monsoon Troughs
• MJO
• Cold-troughs on DT with surface fronts
• Many more disturbances than tropical cyclones
• Moist environment (esp 2-6 km MSL)
• High SST (26º C or higher, maybe)
• Weak vertical shear (or maybe not)
• Organized deep convection

6
Genesis Parameters
habsolute vorticity HRH at 600 mb
Vpotpotential intensity Vshear850-200 mb
shear Vpot F(SST,Outflow temperature, ambient
RH, CK/CD)
From Camargo, Emanuel and Sobel, unpublished
See also Gray 1979, DeMaria 2001
7
How Well Does It Work?
8
Development Theories
Conditional Instability of the Second Kind
(CISK) Charney and Eliassen 1964 Wind Induced
Surface Heat Exchange (WISHE) Emanuel 1988
Rotunno and Emanuel 1987

• WISHE
• Finite-Amplitude Instability
• Does not require CAPE
• Relies on surface enthalpy flux
• Surface fluxes
• Release of heat in clouds
• Intensified warm core
• Stronger surface winds and fluxes

CISK Linear Instability Requires CAPE Relies on
frictional convergence
Craig and Gray, 1996 JAS
9
Maximum Potential Intensity
Emanuel 1988
10
Eliassen Balanced Vortex
absolute vorticity x inertial frequency
Note that with v0, Cgt0 (i.e. an f-plane is
inertially stable)
11
Efficiency of Heating

• Efficiency is
• Localized to where dm/drgtgt0
• Strong function of intensity
• Always very low well outside the core

Aweak depression Ehurricane
12
A Larger C

• Axis of easterly wave
• Flank of westerly phase of MJO (a large-scale
  region)
• Monsoon trough
• MCV
• Front
• ITCZ

13
Atlantic Genesis Issues

• Large-scale influences (waves, troughs)
• Mesoscale (MCSs, MCVs, fronts)
• Convective scale (convective bursts, VHTs)
• Numerical prediction capability
• Critical observations?

14
Developing versus Non-developing Systems
Zehr and McBride 1981
Non-developing
Developing
15
Atlantic Genesis (1975-2005)
Black 1975-1993 Red 1994-2005
16
Not Easy to Predict
17
Tropical Cyclogenesis (N. Pacific)
Eastern Pacific
Western Pacific
Dependence on MJO Phase
From Maloney and Hartmann 1999
18
Easterly Waves
Courtesy Chris Thorncroft (I think)
19
Easterly Wave Genesis
600 mb
Variance at Dakar
850 mb
Thorncroft and Hodges, 2001 J. Clim.
20
Wave-Vortex Duality
21
Wave Contraction

• PV gradient ends (and then a miracle occurs)
• du/dxlt0

Webster et al. 2005, Lorenz Symposium
22
MCVs and Convective Bursts or Hot Towers
Montgomery et al. 2006 JAS
23
What do Those Little Things Do?

• Wind shear provided by MCV
• Tilting gives rise to vortex pairs
• Stretching intensifies cyclonic member
• Diabatic heating projects onto azimuthal mean
• Vorticity becomes axisymmetrized (eddy decays in
  favor of the mean flow)

24
85 GHz Polarization Corrected Temperature (PCT)
and GOES Vis/IR
0039 UTC 20 Sept.
1310 UTC 20 Sept.
0026 UTC 21 Sept.
200 km
0354 UTC 21 Sept.
1257 UTC 21 Sept.
0013 UTC 22 Sept.
Images obtained from NRL http//www.nrlmry.navy.m
il/tc_pages/tc_home.html
25
Humberto (2001) via MM5
Azimuthal mean momentum equation
26
Tropical Transition

• Subtropical or Extra-tropical precursors
• Cold-core upper tropospheric lows
• Fronts
• Mesoscale Convective Vortices
• Initially cold-core (vorticity maximizes at 700
  hPa or above)
• May involve easterly waves
• 30-40 of Atlantic genesis cases
• Relies on finite vertical wind shear
• Diabatic-baroclinic development (strong
  precursor)
• Organization of convection (weak precursor)
• Mesoscale vorticity organization may have much in
  common with deep tropical genesis

27
Bracken and Bosart (2000) Monthly Weather Review
Shear exists in composite of developing
depressions over Northwest Atlantic!
28
MM5 Simulation of Michael (2000)
Potential Vorticity (PVU) and Wind on 340 K q
Sfc. Sea-level Pressure Start 00 UTC 15 Oct.
2000 End 00 UTC 17 Oct. 2000
29
MM5 Simulation of Hurricane Michael (2000)

• Nested 12.3-km grid
• Cold start from global analysis (NCEP)
• 48-h integration
• Kain-Fritsch CP
• Blackadar PBL
• Schultz Cloud Physics

Davis and Bosart, 2003 Monthly Weather Review
30
Diabatic Shear Reduction
E-W Cross section of Potential Vorticity and
Absolute Momentum (Mvfx)
14 km
Control
z
Fake-dry
z
x
31
Strong Baroclinic Precursors
Images courtesy of NRL http//www.nrlmry.navy.mil
/tc_pages/tc_home.html
32
150 hPa
200 hPa
10º-20ºN
20º-30ºN
1 July 2005
C
D
E
F
G
H
I
K
L
J
M
N
O
P
R
30 Sept 2005
S
33
Dynamic Tropopause Maps

• Developed for Baroclinic Cyclogenesis
  (generalized Eady model)
• When combined with qe in the PBL, a diagnostic of
  mesoscale convection potential
• When adiabatic, q on PV is conserved

dQ/dy
PV jump
After Morgan and Nielsen-Gammon, 1998, Mon .Wea.
Rev.
Surface of constant PV
Dynamic Tropopause
Large PV
z
q2
q1
Small PV
Isentropes
South
North
34
August, 2005 on the DT
Potential Temperature on PV1.5 PVU
Red Contours 900 hPa Relative Vorticity (1, 2,
4, and 8x10-5 s-1)
35
A PV-qe View
July-Sept, 2005
Longitudes 105W to 10 W Results similar for 60W
to 20W
36
Development of Mariaq, Wind on DT, z at 950 hPa
00 Z 30 Aug
00 Z 01 Sep
00 Z 02 Sep
00 Z 31 Aug
Potential Temperature on PV1.5 PVU
Red Contours 900 hPa Relative Vorticity (1, 2,
4, and 8x10-5 s-1)
37

38
Maria
12 Z 29 Aug
12 Z 31 Aug
qe345 K
qe350 K
qe350 K
39
Cross Sections
12 Z 29 Aug
12 Z 31 Aug
40
Franklin
41
Harvey
42
Nate
43
Vince
44
Observational Needs

• Mesoscale thermodynamics
• Baroclinity
• Moistening of lower-middle troposphere
• Existence of conditional instability (locally
  enhanced)
• Vertical structure of mesoscale vorticity
• Warm core versus cold core
• Penetration to PBL
• Distribution and intensity of convection
• Evolution of the PBL
• Confirmation of synoptic-scale structures

45
Conclusions

• Disturbances on the DT sometimes important for
  wave genesis (e.g. Katrina, Maria, Rita, etc.)
• EW is the seed, DT disturbance is the trigger
• What is the optimal structure of DT disturbances?
• (Vertical penetration, intensity, horizontal
  scale, etc.)
• Troughs of extratropical or subtropical origin
  penetrate southward to ITCZ
• Tight coupling of tropopause and PBL in time and
  longitudinal mean from 12º-40ºN
• Critical to better observe mesoscale dynamics
  associated with transition

46
AHW Configuration

• 2-way nested, vortex tracking grid at 4km,
  additional nest at 1.33 km
• No cumulus parameterization at 4 km or less.
• 12 km fixed grid nested into GFS, parameterized
  convection (Kain-Fritsch)
• 3-category cloud physics (no graupel), YSU PBL
• Initial conditions either GFS or GFDL

47
2005 Landfalling Cases Verified
Katrina (Cat 5)
Ophelia (Cat 1)
Maria (Cat 3)
Rita (Cat 5)
Wilma (Cat 5)
48
Convection in Mature Hurricanes (Rita)
49
Concentric Eye Walls
Heat Source
Shapiro and Willoughby, 1982 JAS
Heating at large radius creates subsidence over
inner eye wall
50
Convective Bursts
51
Axisymmetrization
Montgomery and Enagonio, 1998 JAS
52
Results for 2005 Season
Position
Intensity
53
Wilma
54
Katrina

• Post real-time tests
• Flux Parameterization
• Fix u over water
• Experiment with u (CD)
• Experiment with CK
• Couple to mixed-layer ocean model
• EnKF and 3D-VAR initialization
• Add 1.33-km storm following domain

55
ARW Forecast of Katrina from 00 UTC 27 Aug.
56
Katrina Forecastsfrom 00 UTC 27 August
1 Km 4 Km 12 K RT Obs

• 1.3 km captures part of rapid intensification
• All forecasts intensify late (erroneously)
• Obvious initialization problems (worse w/ GFS)
• Answers depend on flux formulation

57
Key Areas Needing Improvement

• Flux Formulation theory says large sensitivity
  to CK/CD
• Ocean Coupling cooling from wind-driven ocean
  mixing a big deal
• Initialization real storms from t0
• Resolution of inner core dynamics (what does it
  take 4 km, 1 km, 100 m ????)

58
Exchange Coefficients What do we do???????????
59
Surface Flux Experiments
1 Km 4 Km 1 K D 4 K D Obs

• D Donelan Formulation
• V2.1.2 Charnock
• D has less drag than V2.1.2
• Less drag means
• Larger eye
• Stronger winds (usually)
• Higher central pressure

60
Ocean Mixing
Comparison of OML and Full Ocean Models for
Idealized Vortex
Katrina 4-km grid
Mixed Layer Model
Full Ocean Model
Price (1981)
Maxima in both idealized calculations is 3.1 K
61
Initialization

• Reduce model spinup (12 h)
• Improve intensity prediction lt 2 days
• Methods
• EnKF
• 3DVAR
• 4DVAR
• Nudging

GFS IVAN
EnKF IVAN
62
10-m Wind Comparison
63
Radial Wind Profile Katrina
64
Wind Profiles Katrina
NOAA P-3 17 UTC 28 00 UTC 29 August
AHW 19-23 UTC 28 August
65
Reflectivity on 1.33-km Grid
66
Vorticity Profile Katrina
67
Rainbands Katrina
AHW 1.33 km 00 UTC 29 August
AHW 4 km 00 UTC 29 August
NOAA P-3 LF Radar 2347 UTC
NOAA P-3 LF Radar 2317 UTC
68
Summary of Simulation Issues

• Flux Formulation
• Apparently large sensitivity
• Controls eye size, pressure-wind relationship
• CK perhaps greatest unknown (and important!!)
• Ocean Coupling
• Columnar mixed-layer model capture most
  short-term effects of full ocean model
• Need to incorporate altimetry data to initialize
  ML depth
• EnKF promising as initialization method
• Sub-2-km grid spacing necessary for inner core
• Issue of spurious Rossby waves (Katrina and Rita)
• Width of eye wall?
• Rainbands show less cellular character
• 12-km AHW surprisingly good so far Is 4-km still
  no-mans land?
• Verification methods are obsolete

69
Irene Extratropical Transition
70
SSMI Comparison
ARW 34 h Forecast
1047 UTC 18 August
1000 UTC 18 August
200 km
L
L
71
Convection in Landfalling Hurricanes
00 UTC 16 Sept Univ. Wyoming
72
Vertical Shear at Low Levels
73
Tornadoes Spawned in Katrina
00 Z 30 Aug Shelby County Airport, Alabama
00 Z 30 Aug Peachtree City, Georgia
74
Composites
McCaul 1991
75
Numerical Simulations
1
Del City Storm CAPE2600 J/kg
2
Centreville, AL (Danny 1985) CAPE1800 J/kg
3
Composite CAPE600 J/kg
76
Interpretation
Large in high CAPE environments
Important in strong vertical shear
Buoyancy Part
Dynamic Part
77
Dynamic Pressure Contribution
1
2
3
McCaul and Weisman 1996
78
What does this all mean?

• Genesis possibility handled by current models
• Genesis details not handled by any model
  consistently
• Baroclinic influences under-represented in
  present-day forecasts
• Inner core convection the key to intensity change
  (may require very high resolution)
• Initialization of hurricanes generally lousy, but
  many promising techniques
• Nobody knows how to deal with air-sea interface
  at high wind speeds
• Effects of the ocean may be straightforward wave
  drag very difficult
• Prediction of sensible weather at landfall in its
  infancy