Transcript Document

Tropical Cyclone Intensities: Recent observational studies and simulated response to CO2-induced warming

Acknowledgements: Dr. Chris Landsea (NOAA/NHC) Dr. Kerry Emanuel (MIT)

Thomas R. Knutson

NOAA/Geophysical Fluid Dynamics Lab Princeton, New Jersey

A Look at Tropical Atlantic SSTs…

Main Development Region

All Forcings (n=8) Why have Tropical Atlantic (MDR) SSTs warmed?

GFDL CM2 coupled model historical simulations (1860 2000); Aug-Oct season Natural Forcings Only (n=4) Anthropogenic Forcings Only (n=4) 5-yr running means

Hurricane –region SSTs in the 21 st Century

What is the response of hurricane intensities to increasing tropical SSTs?

NW Pacific Basin: Intensity vs. SST Sea surface temperature (deg C) Source: Baik and Paek,

J. Meteor. Soc. Japan

(1998). Used with permission.

The most intense storms occur at high SSTs

Potential Intensity theories simulate an increase in the intensity of hurricanes for higher sea surface temperatures

Source: Kerry Emanuel, MIT.

Future Hurricane Intensities: Simulations using a high-resolution hurricane model

 9 km grid spacing near storm- partially resolves eye of hurricane  Similar versions of this model are used operationally for hurricane prediction at NOAA’s National Centers for Environmental Prediction

GFDL hurricane model: simulated max. surface wind speeds in the NW Pacific  Observed (1971-92)  Control (51 cases) m/sec Note: Earlier (18km inner-nest) version of the model was used for these simulations

GFDL Simulations: Hurricanes are more intense for warmer climate conditions …(~4% per deg C) Note: Min. central pressures are averages over day 5 of integrations.

Source: Knutson and Tuleya,

J. Climate

, 2004.

GFDL Simulations: Hurricanes have significantly more near-storm rainfall for warmer climate conditions …(~12% per deg C) Average rainfall in a 32,700 km 2 region of highest 6-hour accumulation (equivalent to 100km radius region).

Tropical Cyclone-generated sea surface cooling Cool wake Tropical cyclone SST (deg C) Model: GFDL Coupled Hurricane-Ocean model

 What do the historical tropical cyclone data show in terms of long-term trends?

Emanuel (2005)

Original PDI from Emanuel (2005) Revised PDI from Landsea (2005 - updated)

Source: Chris Landsea, NOAA/NHC

Storm-Maximum Power Dissipation Index – Atlantic Basin Source: K. Emanuel, MIT, 2006

Atlantic Basin: SSTs vs number of tropical cyclones Source: Kerry Emanuel, MIT

Emanuel’s Multi-basin Tropical Cyclone Power Dissipation Index (PDI) has increased substantially over past 50 years, along with tropical SSTs Source: Kerry Emanuel, MIT, http://wind.mit.edu/~emanuel/anthro2.htm

. SST anomaly (deg C) with arbitrary vertical offset. PDI scaled by constant.

Webster et al.: The percentage of hurricanes which reach Category 4-5 has increased in all basins, comparing two recent 15 year periods… 45 40 35 30 25 1975-1989 1990-2004 20 15 10 5 0 North Atlantic East Pacifc West Pacific Southwest Pacific North Indian South Indian Question: Are the historical data adequate for this conclusion?

Source: Adapted from Webster et al.,

Science,

Sept. 2005.

Knaff and Sampson’s reanalysis of 1966-87 NW Pacific max intensities produces a reduced Cat 4 5 trend, relative to “best track” Figure 4. The annual counts and as occurring in the western North Pacific Basin for the years 1966-2004 are shown. Notice the change in the trends as a result of the reanalysis. Again, blue and red bars and lines are associated with the best trac e MSLP data, respectively. Source: Knaff and Sampson, AMS Hurricanes Conference Proceedings, 2006

Sriver and Huber’s PDI from reanalysis, although weaker, is well-correlated after 1978 with Emanuel’s PDI from “best track” data (Atlantic + NW Pacific) Figure 1 5 4 3 ERA Global PD ERA Global PDI ERA at+wp PDI Emanuel at+wp PDI 2 -1 -2 1 0 -3 1960 1970 1980 1990 Source: Sriver and Huber,

Geophysical Research Letters

, in press.

2000

Comparison of models with observations…

 GFDL Model wind speed intensity, V (and hurricane theory) vs SST: ~

4-5% per o C

 Emanuel (2005) for Atl, NW Pac, NE Pac: V 3 increases 50% for 0.5

o

C, so V:

~30% per o C

 Emanuel (2006) for Atlantic only: Century-scale data: V increases

~10% per o C;

Data since 1980 only:

~20% per o C

 Factor of 2 to 6 discrepancy in sensitivity…  Implications for future projections??

Resolving the Discrepancy?

Possibilities:

1. Past trend of intensity over-estimated? (i.e.:

the data is wrong

) 2. Hurricane model/theory not sensitive enough to SST change? (i.e.,

the models are wrong

) 3. Other factors besides SST which can affect potential intensity are playing a role? (i.e.,

our simple analysis is wrong

)

GFDL Zetac Nonhydrostatic Regional Model: 18km Tropical N. Atlantic Simulation Simulated hurricanes

Note: Uses large-scale interior nudging

Summary of Main Points:

Global Warming and Hurricanes

    Tropical SSTs (including tropical North Atlantic):   Substantial warming (~0.6

o C) occurred in 20 th global mean temperature century, roughly tracking Substantially greater 21 st century warming (~2 o C) is anticipated due to anthropogenic forcing (greenhouse gas emissions, etc.) Intensity simulations with a high-resolution hurricane prediction model:   Historical hurricane observations give conflicting information on past trends:   Several Atlantic hurricane activity measures are dominated by multi-decadal “cycles” or noise (e.g., landfalling PDI)—not trends. Some basin-wide indices show unprecedented levels in recent years, correlated with rising SSTs. Data quality issues remain unresolved at this time.

Hurricane intensity sensitivity implied by some studies greatly exceeds that of current model simulation and theory, a discrepancy that remains unresolved at this time.

Ongoing work at GFDL: high-resolution seasonal Atlantic simulations   Maximum intensities increase (roughly 4% -- per deg Celsius SST increase) Near-hurricane precipitation increases (roughly 12% per deg Celsius) Future frequency changes? highly uncertain Future regionally specific effects? highly uncertain