Verification of NHC Forecasts of Extratropical Transition

1
Verification of NHC Forecasts of Extratropical
Transition
JACK BEVEN
TROPICAL PREDICTION CENTER NATIONAL HURRICANE
CENTER
WHERE AMERICAS CLIMATE AND WEATHER SERVICES
BEGIN
2
What is extratropical transition?
My conceptual model
ET is the process by which a tropical cyclone
loses tropical characteristics in a baroclinic
environment. This can end in a variety of
cyclone structures.
Hart and Evans Cyclone Phase Space
3
Why are analyzing and forecasting extratropical
transition important?

• Tropical cyclones undergo significant structural
  changes during ET, which can include
  (re-)intensification, expansion of the wind
  field, and modified precipitation distribution
  (all seen in Noel of 2007)
• Forecast/warning procedures differ greatly for
  tropical and non-tropical cyclones

4
Hurricane Lili (1996)
Fcst Time Mean Error (hr)
12 hr 39.43
24 hr 99.00
36 hr 27.00
48 hr 42.00
72 hr 24.00
Obviously we have little skill with Lili in
forecasting extratropical transition Lixion
Avila
5
Verification for ET Forecasts

• Based on yes-no 2-D contingency diagram based on
  whether ET occurred/was forecast at any time in
  the forecast period
• Two versions One includes forecasts for all TCs,
  while the other includes only TC where ET was
  forecast and/or did occur both versions done
  for 5-day and 3-day forecasts
• Includes verification of the timing of the ET
  forecast how wrong temporally was the forecast
  of ET when ET occurred

6
Verification Table for 2005
ET Observed
663 forecasts Yes No
Yes 155 (successfully forecast ET) 57 (false alarms)
No 49 (missed transitions) 402 (successfully forecast non-ET)
ET Forecast
7
Verification does not include

• Forecast track, intensity, or wind radii of the
  cyclone during the extratropical phase

8
Verification Skill Measures
Skill Measure Range Good Values
Critical Success Index 0.0 to 1.0 Close to 1.0
Probability of Detection 0.0 to 1.0 Close to 1.0
False Alarm Ratio 0.0 to 1.0 Close to 0.0
Percentage Correct 0.0 to 1.0 Close to 1.0
Event Bias 0.0 to ? Close to 1.0
Gilbert Skill Score -0.33 to 1.0 Close to 1.0
Heidke Skill Score -1.0 to 1.0 Close to 1.0
True Skill Statistic -1.0 to 1.0 Close to 1.0
9
5-Day ET Timing Errors/Biases
10
3-Day ET Timing Errors/Biases
11
2005 Errors/Biases by Storm
12
3-Day Individual Time Errors
Mean ET Forecast Error (hr)
13
3-Day Individual Time Biases
Mean ET Forecast Bias (hr)
14
3-Day Skill Scores
15
3-Day Skill Scores
16
Methodology Issues

• TPC discrete forecast intervals and timing
• Possible best track inconsistencies in
  determining if transition occurred and transition
  time, especially before the development of the
  cyclone phase space
• Forecasts of system becoming ET, then no actual
  forecast positions as an ET cyclone

17
Issues with Storms

• Systems that dissipated upon merging with fronts
  (Cindy 1999, Florence 2000, Cristobal 2002, Kyle
  2002, TD14 2002, Irene 2005)
• Track forecast problems (Katrina 1999, Gamma
  2005, Gordon 2006)
• Intensity forecast problems (Mindy 2003, Bonnie
  2004, Debby 2006)
• Aborted transitions (Alberto 2000)
• Slow or interrupted transitions (Lili 1996,
  Gabrielle 2001, Maria 2005, Helene 2006)
• Unclimatological transitions (Michelle 2001)
• Unclimatological non-transitions (Humberto 2001,
  Juan 2003)
• Transitions or lack thereof over mid-America
  (Lili 200, Ivan 2004, Arlene 2005, Katrina 2005)
• Environmental mis-reads? (Henri 2003, Hermine
  2004, Matthew 2004, Philippe 2005, Epsilon 2005)

18
Conclusions

• NHC forecasts of extratropical transition for the
  most part have skill according to the measures of
  the 2-D contingency table verification
• ET forecasts have shown a general increase in
  quality since 1996. This is likely due to better
  NWP models, as well as the introduction of the
  Hart and Evans Cyclone Phase Space diagrams