Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric...

17
Dynamical Tropical Cyclone Forecasts using the COAPS Atmospheric Model w/ CFS SSTs Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University [email protected] Climate Diagnostic Prediction Workshop October 2009

Transcript of Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric...

Page 1: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

Dynamical Tropical Cyclone Forecasts using the COAPS Atmospheric Model

w/ CFS SSTs

Tim LaRowCenter for Ocean Atmospheric Prediction Studies/

Florida State [email protected]

Climate Diagnostic Prediction Workshop October 2009

Page 2: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

Experimental Setup

Atlantic hurricane season (June-November) simulations from 1986 to 2008 (23 years).

Daily updated SSTs from the CFS model.

FSU/COAPS global spectral model – T126L27 resolution

4 member ensembles for each year. Time lagged ECMWF atmospheric initial conditions centered on 1 June of the respective year.

RAS Convective Scheme (Hogan and Rosmond 1991)

6 hourly output for detection and tracking

Modified Detection/tracking algorithm from Vitart et al. (2003), see LaRow et al. (2008)

Page 3: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

Genesis Density (1986-2005)

Observed

COAPS

Page 4: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

Storm Tracks and Numbers

When using climate models to simulate TC activity, two approaches are commonly used:

1. Count the number of hurricane-like vortices directly, using a detection algorithm, and then rescale the counts if necessary.

2. Use the simulated large-scale fields of the model and estimate the number of storms using a statistical procedure such as the Storm Genesis Potential by W. Gray et al. or K. Emmanuel et al.

In this study we use the detection/tracking algorithm developed by F. Vitart, see Vitart et al. (2003)

Page 5: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

20-Years of Storm Tracks (1 Member) with Saffir-Simpson Pressure Scale

Blue=TS, Dark Blue=CAT1, Green=CAT2, Red=CAT3, Dark Red=CAT4

Page 6: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

Sample Season Storm Tracks Saffir-Simpson Pressure Scale

Blue=TS, Dark Blue=CAT1, Green=CAT2, Red=CAT3, Dark Red=CAT4

Page 7: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

(1986-2008)

Atlantic Total Storm Counts Using CFS SSTs

Correlations:Correlations:Ensemble Mean with HURDAT = 0.71Detrended with HURDAT = 0.69Trends are significant at the 95% confident level

Page 8: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

Hurricane Numbers

In our climate model, hurricanes are identified using the 850hPa wind maximum determined by the detection code and multiplying the wind speed by 0.8 to extrapolate to 10 meters. If the 10-m wind speed is greater than 32 m/s it is classified a hurricane and counted.

Page 9: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

Atlantic Hurricane Counts – with CFS SSTs(1986-2008)

Correlations:Correlations:Ensemble Mean Counts with HURDAT = 0.82Linear detrended correlation = 0.63Model Counts with CFS ASO Niño-3 SSTA = -0.80Model with CFS ASO MDR SSTA = 0.59

Page 10: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

Ensemble Member Summary

Dataset Member 1 Member 2 Member 3 Member 4

OBS SSTs 0.76 0.51 0.71 0.62

CFS SSTs 0.71 0.67 0.64 0.50

Dataset Member 1 Member 2 Member 3 Member 4

OBS SSTs 242 234 234 249

CFS SSTs 247 239 262 244

Total Number of Storms (1986-2005)

Interannual TC Correlations (1986-2005)

Total number of observed TCs 245

Page 11: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

Model Resolution Atl Corr. GP Atl Corr. NTCCCM T42 0.46

NSIPP 2.5x2.5 0.50 0.38ECHAM3 T42 0.00 0.53ECHAM4 T42 0.36 0.52ECHAM5 T42 0.53ECHAM5 T63 0.33ECHAM5 T85 0.47ECHAM5 T106 0.27ECHAM5 T159 0.43 0.33

GFDL (Knutson) 18km 0.8650km 0.83

FSU(LaRow) T126 0.78T126 0.73/0.82

GFDL (Zhao)1

FSU (LaRow)2

Camargo et al. (2005), Camargo et al. (2007), Bengtsson et al. (2007)Knutson et al. (2007), LaRow et al. (2008), Zhao et al. (2009)1Accepted J. Climate

2Submitted to GRL

Atlantic Seasonal TC Studies

Page 12: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

Atlantic Accumulated Cyclone Energy – (1986-2009)

Correlation:Correlation:Using Observed SSTs = 0.85 (Previous Study w/ obs. SSTs)Using CFS SSTs = Using CFS SSTs = 0.69, detrended = 0.57 (Significant @ 95%)

19861987

19881989

19901991

19921993

19941995

19961997

19981999

20002001

20022003

20042005

20062007

2008

0

50

100

150

200

250

300

Atlantic ACE

Years

AC

E/1

00

00

kt^

2

Red=HURDATBlue=COAPS Model

Page 13: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

Gates MEM1 MEM2 MEM3 MEM4 OBS

1 6 14 10 12 322 36 35 32 24 433 19 20 18 19 184 5 5 9 10 5

Gates MEM1 MEM2 MEM3 MEM4 OBS

1 12 6 14 10 322 33 33 37 39 433 18 15 17 22 184 5 4 9 8 5

1 2

3

4

Land Falling Numbers(1986-2005)

Underestimates landfalls in westernGulf of Mexico by about a 1/3

CFS SSTs

OBS SSTs

Page 14: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

2009 Real-Time Forecast

SETUP SSTs from the May 29, 2009 CFS Forecast 4 Member Atmospheric Ensemble T126L27 FSU/COAPS Model Forecast Issued June 2, 2009

FORECAST 8 Named Storms (stdev 2.2) 4 Hurricanes (stdev 0.8) Mean ACE 65

Page 15: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

June 2009 Sea Surface Temperature ForecastsNiño-3.4 Region

Models JJA JAS ASO SON

NASA 1 1.4 1.6 1.7NCEP CFS 1.1 1.3 1.5 1.6

JMA 0.8 1 1.1 1.4SCRIPPS 0.1 0.3 0.4 0.6

LDEO 0.3 0.3 0.4 0.5AUS/POAMA 1.1 1.3 1.5 1.6

ECMWF 0.8 1 1.2 1.3COLA ANOM 0.4 0.6 0.7 0.8ECHAM/MOM 0.5 0.6 0.7 0.8

Atlantic TC Forecast Period

OBS: JJA 0.77CJAS 0.84C

Page 16: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

Conclusions

Ensemble hindcast results from a relatively high resolution atmospheric model (T126L27) have been presented for 23 years of the Atlantic Basin hurricane seasons using CFS SSTs.

Rank correlation of the interannual variability of the tropical storm frequency against observation was found to be high (0.73 for tropical cyclones and 0.82 for hurricanes, 0.69 for ACE).

All correlations significant at the 95% confidence level.

Model appears to simulate the ENSO-Atlantic covariation well.

More members in the ensemble are needed to better quantify skill (on order of 20 members).

Page 17: Dynamical Tropical Cyclone Forecasts using the COAPS ...Tim LaRow Center for Ocean Atmospheric Prediction Studies/ Florida State University tlarow@fsu.edu Climate Diagnostic Prediction

Conclusions Continued

Factors contributing to increase in interannual skill not fully understood. Higher horizontal resolution models (eg. ECHAM) not always best in interannual variability.

Intensity still underestimated in terms of surface wind speeds, even at high horizontal resolution (18-20km) and will probably continue to be as the cyclostrophic balance is not resolved by climate models.

Our 1 June 2009 Real-time forecast appears to be validating well with only 2 hurricanes thus far in the Atlantic and an ACE value of 42.1