Developing an Inner-Core SST Algorithm for use in SHIPS. Principal Investigator : Joseph J. Cione NOAA’s Hurricane Research Division Co-Investigators: John Kaplan (HRD) ; Chelle Gentemann (Remote Sensing Systems) ; Mark Demaria (NESDIS) IHC 2005 Joint Hurricane Testbed March 9, 2005. - PowerPoint PPT Presentation
Developing an Inner- Core SST Algorithm for use in SHIPS Principal Investigator Principal Investigator : : Joseph J. Joseph J. Cione Cione NOAA’s Hurricane Research Division NOAA’s Hurricane Research Division Co-Investigators: Co-Investigators: John Kaplan John Kaplan (HRD) (HRD) ; ; Chelle Gentemann Chelle Gentemann (Remote Sensing (Remote Sensing Systems) Systems) ; Mark Demaria ; Mark Demaria (NESDIS) (NESDIS) IHC 2005 IHC 2005 Joint Hurricane Testbed Joint Hurricane Testbed
Transcript
Developing an Inner-Core SST Algorithm for use in SHIPS
Developing an Inner-Core SST Algorithm for use in SHIPS
Principal InvestigatorPrincipal Investigator: : Joseph J. Cione Joseph J. Cione
NOAA’s Hurricane Research DivisionNOAA’s Hurricane Research Division
Co-Investigators:Co-Investigators: John Kaplan John Kaplan (HRD)(HRD); Chelle ; Chelle Gentemann Gentemann (Remote Sensing Systems)(Remote Sensing Systems); Mark ; Mark
Demaria Demaria (NESDIS)(NESDIS)
IHC 2005 IHC 2005
Joint Hurricane TestbedJoint Hurricane Testbed
March 9, 2005March 9, 2005
Developing an Inner-Core SST Algorithm for use in SHIPSDeveloping an Inner-Core SST Algorithm for use in SHIPS
The Goal of this JHT project is to…The Goal of this JHT project is to…
Improve SHIPS intensity forecasts Improve SHIPS intensity forecasts by by incorporating realistic incorporating realistic SST estimates in the SST estimates in the TC high wind inner core.TC high wind inner core.
Developing an Inner-Core SST Algorithm for use in SHIPSDeveloping an Inner-Core SST Algorithm for use in SHIPS
+ As such, SHIPS is unable to account for any storm-induced + As such, SHIPS is unable to account for any storm-induced ocean cooling that occurs within the high wind inner core ocean cooling that occurs within the high wind inner core environment.environment.
+ Furthermore….The ‘potential term’, PI, is defined in + Furthermore….The ‘potential term’, PI, is defined in SHIPS as:SHIPS as:
PI = MPIPI = MPI(fn of SST only)(fn of SST only) - TC intensity - TC intensity and is a and is a highly significant predictorhighly significant predictor in the statistical in the statistical model…model…
+ Therefore….+ Therefore….even modest improvements to SST may result even modest improvements to SST may result in significant improvements in SHIPSin significant improvements in SHIPS intensity forecasts… intensity forecasts…
Developing an Inner-Core SST Algorithm for use in SHIPSDeveloping an Inner-Core SST Algorithm for use in SHIPS
The Problem…The Problem…
Routine observation of the inner core hurricane ocean Routine observation of the inner core hurricane ocean environment is often impractical and in many cases environment is often impractical and in many cases impossible…impossible…
+ However… recent multi-hurricane observations (1975-+ However… recent multi-hurricane observations (1975-2002) from Cione and Uhlhorn (2003), have provided an 2002) from Cione and Uhlhorn (2003), have provided an improved representation of inner core (<60km) SST improved representation of inner core (<60km) SST conditions…conditions…
+ Using storm-specific information in conjunction with + Using storm-specific information in conjunction with ambient and inner core SST observations from the 33 TC ambient and inner core SST observations from the 33 TC events documented in Cione and Uhlhorn (2003)….events documented in Cione and Uhlhorn (2003)….
an algorithm an algorithm to predictto predict hurricane inner core SST was hurricane inner core SST was developed….developed….
Scatter plot of in-situ SST vs. predicted inner-core SST [using the hurricane inner-core SST coolingalgorithm developed from the 23-hurricane (1975-2002) sample from Cione and Uhlhorn (2003)].
The linear best-fit shown explains 87.5% of the variance (i.e. r2=0.875). SST is given in oC.
Developing an Inner-Core SST Algorithm for use in SHIPSDeveloping an Inner-Core SST Algorithm for use in SHIPS
““Dependent Sample” Testing in SHIPS…Dependent Sample” Testing in SHIPS…
+ + The The Inner-Core SST AlgorithmInner-Core SST Algorithm was recently was recently tested on tested on thousandsthousands of individual SHIPS forecast of individual SHIPS forecast events between events between 1989-20021989-2002……
……For the Inner-core SHIPS re-runs, Reynolds For the Inner-core SHIPS re-runs, Reynolds SST was replaced with the algorithm-derived….SST was replaced with the algorithm-derived….
Inner-Core SSTInner-Core SST
Some major highlights from the dependent SHIPS analysis:Some major highlights from the dependent SHIPS analysis:
* ZER0 degradation in mean SHIPS forecast skill was found* ZER0 degradation in mean SHIPS forecast skill was found when the inner core SST algorithm was used, regardless of sample when the inner core SST algorithm was used, regardless of sample
stratification…stratification…
* The algorithm is very stable…* The algorithm is very stable…Analysis used 1000s of individual forecast Analysis used 1000s of individual forecast events from the 1989-2002 SHIPS storm databaseevents from the 1989-2002 SHIPS storm database
* Biggest improvement in skill was * Biggest improvement in skill was found found laterlater in the forecast period in the forecast period (48h-96h)(48h-96h)
Specifics….Specifics…. For For All Hurricanes All Hurricanes (N=4528) (N=4528)
• an average an average 1.5-3.5%1.5-3.5% improvement in skill improvement in skill over the 12-120h period over the 12-120h period For all For all Major HurricanesMajor Hurricanes (N=1011) (N=1011)
• an average an average 2.7-7.0%2.7-7.0% improvement in skillimprovement in skill over the 12-120h period over the 12-120h period For the For the Top 10% “Rapid Intensifiers”Top 10% “Rapid Intensifiers”
• an average an average 4.5-8.0%4.5-8.0% improvement in skillimprovement in skill over the 12-120h period over the 12-120h period For the For the Bottom 10% “Rapid Fillers”Bottom 10% “Rapid Fillers”
• an average an average 2.0-13.5%2.0-13.5% improvement in skillimprovement in skill over the 12-120h period over the 12-120h period
Developing an Inner-Core SST Algorithm for use in SHIPSDeveloping an Inner-Core SST Algorithm for use in SHIPS
……. Independent Sample Testing in SHIPS…. Independent Sample Testing in SHIPS…
+ Investigate the + Investigate the impactimpact of the of the TC Inner-Core TC Inner-Core SST AlgorithmSST Algorithm on SHIPS 2003/04 Atlantic on SHIPS 2003/04 Atlantic hurricane season forecasts…hurricane season forecasts…
……For the Inner-Core SHIPS re-runs, Reynolds For the Inner-Core SHIPS re-runs, Reynolds SST was replaced with the algorithm-derived….SST was replaced with the algorithm-derived….
Inner-Core SSTInner-Core SST
Highlights from the 2003/04 independent sample SHIPS analysis:Highlights from the 2003/04 independent sample SHIPS analysis:
When the SST inner core algorithm was used in SHIPS….When the SST inner core algorithm was used in SHIPS….
+ No mean degradation of forecast skill was found + No mean degradation of forecast skill was found over any over any forecast interval for Hurricane eventsforecast interval for Hurricane events
+ A significant increase in forecast skill was found + A significant increase in forecast skill was found for 2003/04 for 2003/04 Rapid IntensifiersRapid Intensifiers
+ The largest increase in absolute skill was found over the 72-120h + The largest increase in absolute skill was found over the 72-120h forecast interval, forecast interval, regardless of sample stratification regardless of sample stratification (similar to the (similar to the dependent analysis)dependent analysis)
Specifics….Specifics….
For For All Hurricane CasesAll Hurricane Cases (N=147-207) (N=147-207)• an average an average 7%7% improvement in skill improvement in skill over the 72-120h period over the 72-120h period (2kts)(2kts)
For For All Rapid IntensifiersAll Rapid Intensifiers (N=27-33) (N=27-33)• an average an average 33%33% improvement in skillimprovement in skill over the 72-120h period over the 72-120h period (5kts)(5kts)
++ Run a parallel version of SHIPS that uses the Inner-Core SST Run a parallel version of SHIPS that uses the Inner-Core SST Algorithm Algorithm in real-timein real-time during the 2005 N. Atlantic Hurricane during the 2005 N. Atlantic Hurricane SeasonSeason (Planned activity…Coordination with M Demaria)(Planned activity…Coordination with M Demaria)
++ Test the impact of high resolution (time and space) SSTs on Test the impact of high resolution (time and space) SSTs on SHIPS forecasts SHIPS forecasts (Recently completed..2003 N. Atlantic hurricane season)(Recently completed..2003 N. Atlantic hurricane season)
++ Investigate the possibility of incorporating a sub-surface ocean Investigate the possibility of incorporating a sub-surface ocean predictor into the algorithm predictor into the algorithm (Ongoing…near completion)(Ongoing…near completion)
++ Given the algorithm’s success with RI events, investigate the Given the algorithm’s success with RI events, investigate the feasibility of using a version of the algorithm in the Atlantic RI feasibility of using a version of the algorithm in the Atlantic RI Index Index (Preliminary work recently started with J. Kaplan)(Preliminary work recently started with J. Kaplan)
++ Use hurricane inner core air-sea composite analyses from Cione Use hurricane inner core air-sea composite analyses from Cione et al 2000 to help develop a bulk enthalpy flux predictor for possible et al 2000 to help develop a bulk enthalpy flux predictor for possible future use in SHIPS future use in SHIPS (Future…begin late 2005)(Future…begin late 2005)
