Efforts in Tropical Storm Modeling, Prediction and Projection

Efforts in Tropical
Image: NASA.
Storm Modeling,
Prediction and
Projection
G.A. Vecchi 1 , M. Bender 1 , J.H.
Chen 1 ,
T. Delworth 1 , Y. Fan 1 , I.M. Held 1 ,
H.S. Kim 2 , T.R. Knutson 1 , S.J. Lin 1 ,
R. Msadek 1 , A. Rosati 1 , G.
Villarini 2,3 ,
M. Zhao 1
1.NOAA/GFDL, Princeton, NJ
2.Willis Research Network
3.Princeton U., Dept. Civil Engineer
Gabriel.A.Vecchi@noaa.gov
Presentation at “Climate 2012”

Data Homogenization for North Atlantic Hurricanes
Adjustments to storm
counts based on
ship/storm track locations
and density
Vecchi and Knutson (2008, J. Clim.)
Landsea et al. (2009, J. Clim.)
Vecchi and Knutson (2011, J. Clim.)

GFDL Hi-Res Global Models
In order to represent processes and phenomena, GFDL
has been developing high-resolution global models.

GFDL High Resolution Coupled Model
Development
Scientific Goals:
•Developing improved models (higher resolution, improved physics,
reduced bias) for studies of variability and predictability on intraseasonal
to decadal time scales
•Explore impact of atmosphere and ocean on climate variability and
change using a high resolution coupled models
•New global coupled models: CM2.4, CM2.5, CM2.6
Ocean Atmos Computer Status
CM2.1 100 Km 250 Km GFDL Running
CM2.3 100 Km 100 Km GFDL Running
CM2.4 10-25 Km 100 Km GFDL Running
CM2.5 10-25 Km 50 Km GAEA/GFDL Running
CM2.6 4-10 Km 50 Km GAEA Running

N. Tropical Atlantic Climate Improves
from better Atlantic ITCZ simulation
Climatological Rainfall Interannual SST Standard Dev.
Doi et al. (2012)
Figures: Takeshi Doi

Coupling appears to improve GFDL high-res model’s
MJO
Figure: Daehyun Kim
(LDEO, Columbia U.)

Response of TCs in high-resolution global coupled model
(GFDL CM2.5, Delworth et al. 2012, J. Climate; Kim et al. 2012 in prep.)
Observed Tracks
Coupled Model Tracks
More storms
Fewer storms
CM2.5 Tropical storm density response to CO2 doubling

The GFDL High-Resolution Atmosphere Model (HiRAM)
• Non-hydrostatic Finite-Volume dynamical core on the cubed-sphere
• Designed for resolution between 1– 100 km, capable of direct cloud
simulation
• A PDF based 6-category cloud micro-physics with finite-volume vertical
sub-grid reconstruction allowing vertically & horizontally sub-grid cloud
formation
• A “non-intrusive” shallow convective parameterization (Bretherton
scheme modified by Zhao et al. 2009)
• Options to couple with ocean and wave models
Slide: S-J Lin

Geographical distribution of TC tracks (1981-2009)
Observation
HiRAM-C180
AMIP simulation
Zhao et al.
(2009)

Response of TC frequency in single 50km global
atmospheric model forced by four climate projections for
21st century
Red/yellow = increase
Blue/green = decrease
Adapted from Zhao et al. (2009, J. Climate)
Regional increase/decrease much larger than global-mean.
Pattern depends on details of ocean temperature change.
Sensitivity of response seen in many studies
e.g., Emanuel et al 2008, Knutson et al 2008, Sugi et al. 2010, etc

C180-HiRAM NA Hurricane Projections including
CMIP5
CMIP3
CMIP5
Adapted from Zhao et al. (2009, J. Clim.) and Held et al. (2012,

Strongest cyclones projected with double downscaling
Adapted from
Bender et al (2010, Science)
Global Climate Models -> Regional Model -> Hurricane
model
Large-scale TS Frequency Intensity

Overall frequency decrease projected for North Atlantic,
but strongest storms may become more frequent
Adapted from Bender et al (2010,
Science)

Courtesy: Yalin Fan – Fan et al. (2011)

Statistical Models
• Build parsimonious statistical models
• Train on high-res dynamical models & homogenized
obs.
• Predictors motivated by high-res dynamical models

uild statistical models for exploration, prediction
and projections
Consistent with high-res dynamical models, understanding on controls
to hurricanes & “cheap”.
Rate e abSST ATLcSST TRO
Statistical NA TS Projections from 17 CMIP5
CGCMs
Villarini and Vecchi (2012)
Knutson et al. (2008) Swanson (2008), Vecchi et al. (2008), Zhao et al. (2009, 2010), Villarini et al. (2010, 2011.a.,.c), Villarini and

GFDL-CM3 indicates aerosols key for NA TS
projections
All Forcing
No future aerosol or O 3
No future aerosol
Villarini and Vecchi (2012)
See also Vecchi and Soden (2012)

Key uncertainty sources to projections of decadal TS
activity
Tropical Atlantic SSTA
NA TS Frequency
Villarini et al. (2011), Villarini and Vecchi (2012, submitted)
Sources of uncertainty (after Hawkins and Sutton, 2009)
•Variability: independent of radiative forcing changes
•Response: “how will climate respond to changing GHGs
& Aerosols”
•Forcing: “how will GHGs & Aerosols change in the
future”

Tropical Cyclone Prediction
• Dynamical seasonal forecasts
• Hybrid long-lead (multi-season to year) seasonal
forecasts
• Hybrid multi-year forecasts

25km HiRAM Seasonal hurricane predictions –
initialized July 1
1990-2010 (Jul-Nov)
Resolution: 25 km, 32 levels
• 5-members initialized on July 1 with
NCEP analysis
• SST anomaly is held constant
during the 5-month predictions
• Climatology O3 & greenhouse
gases are used
0.78
0.88
0.94
1. Chen and Lin 2011, GRL
2. Chen et al., submitted

Merge multiple tools and understanding to build experimental long
lead hurricane forecast system: skill from as early as October of
year before
May & onward
forecasts fed
to NOAA
Seasonal
Outlook Team
Run Hi-Res AGCM
in many different
climates.
Count storms.
Initialized January
Build statistical
model of the
response of
hurricanes in
HiRAM
r=0.66
Apply
Stat
model to
Predicted
SST
Use CM2.1 (and
CFS) to forecast
future values of
Atlantic and
Tropical SST
Vecchi et al. (2011)
Make
Prediction of
Full PDF of
Hurricane
Activity

Experimental decadal hurricane predictions
Hybrid system: statistical hurricanes, dynamical decadal climate forecasts
• Retrospective predictions
encouraging.
• However, small sample size limits
confidence
• Skill arises more from recognizing
1994-1995 shift than actually
predicting it.
• This is for basinwide North Atlantic
Hurricane frequency only.
Vecchi et al. (2012 in prep.), see also Smith et al. (2010,

Idealized Forcing Experiments
If local SST the dominant control, as opposed to relative SST:
•Similar Atlantic Response to Atlantic and Uniform Forcing
•Little Pacific Response to Atlantic compared to Uniform

North Atlantic Response to Idealized SST
Atlantic Forcing
Uniform Forcing
Near-equatorial
Forcing
Similar TS frequency
response to:
0.25° local warming
4° global cooling
Vecchi et al (2012, in prep.)

Idealized CO2 and 2K Experiments under US-CLIVAR
Hurricane WG
Response Averaged for 3 Models Hemispheric Response
Figures: Ming Zhao - Zhao et al. (2012, in prep.)

Summary
• Recorded century-scale increase in Atlantic hurricane
frequency consistent with observing system changes.
Issues with intensity records remain.
It is premature to conclude we have seen hurricane change due
to CO 2
• Statistical and dynamical models allow estimates of future
activity:
• Next couple of decades: internal variability dominant player
(some may be predictable, some not)
• NA Hurr. Response to GHG: likely fewer, probably stronger.
• Aerosol forcing and response a key uncertainty & response.
• Encouraging results from long-lead (multi-season and
multi-year) experimental forecasts using hybrid system:

Snapshots from CM2.6 movie