impact of global hawk - ofcm · impact of global hawk on gfs hurricane forecasts . kate howard. ¹,...

Impact of Global Hawk on GF S H urricane

F orecas ts

K ate H oward¹, J as on S ippel ², V ijay T allapragada ¹ T COR F 71s t IH C - M arch 14th, 2017 ¹ (NOA A NWS NCE P /E M C), ² (NOA A A OM L /H R D)

Recent Observing System Experiments (OSE) for S upplem ental Obs ervations

• E x perim ents run interm ittently to evaluate im pact of s upplem ental drops ondes /rawins ondes for T Cs in GDA S /GF S .

• T he following have been evaluated us ing H ybrid 3DE nV ar

– Is aac (2012)

– S andy (2012)

– K aren (2013)

– J oaquin (2015)

2 T COR F 2017, 71s t IH C - M arch 14 - 16

Hurricane Isaac (2012) • Dropwinds ondes reduced 4-5 day average track forecas t errors by about 30%.

• Drops appear to reduce the cycle-to-cycle variabi li ty in track .

• L ittle change in GF S intens ity forecas t errors due to the drops .

• Control s hows s hallower, weak er vortex relative to “no drop”.

T COR F 2017, 71s t IH C - M arch 14 - 16 3

As noticed by operational forecasters, assimilation of surveillance sondes resulted in significant change in TC intensity forecast for Karen (no drop resulted in forecast of stronger storm)

TCORF 2017, 71st IHC - M arch 14 - 16 4

T ropical S torm K aren (2013)

S ondes 1– 14 A s s imilated for 06Z

Prior OSE Summary

• S im ilar to K aren and Is aac cas e s tudies , m os t of the OS E s s how varying degrees of s m all , pos itive im provem ents on track forecas ts .

• One ex ception: E x perim ent with J oaquin (2015) res ulted in degradation from drops ondes

– Caution: T his cas e s eem s to be ex trem ely s ens itive to IC m odifications

• Prior cas e s tudies did not evaluate inclus ion of Global H awk obs ervations .

T COR F 2017, 71s t IH C - M arch 14 - 16

5

NOAA SHOUT Fl ights for 2016 Hurricane Season

6

Investigate the impact of Global Hawk (GH) dropsonde observations on Global Forecast System (GFS) model hurricane track and intensity forecasts. Global Hawk data is currently not assimilated in the GFS at NCEP, but pos itive res ults from this s tudy would jus tify their inclus ion.

T COR F 2017, 71s t IH C - M arch 14 - 16

Methodology

Model used: Q3FY17 GFS (planned upgrade) - T1534L64 (13km) NEMS/GSM - Hybrid 4DEnVar Ensemble Kalman Filter (EnKF) Data Assimilation System

Experiments: - Control run = same as Q3FY17 pre-implementation parallel configuration - Experimental runs = control plus: - Global Hawk observations turned on in GFS by adjusting prepbufr quality marks. - Observations within storm radius reflagged for non-use by assimilation.

Targeted 2016 Atlantic storm cases: Gaston, Hermine, Karl, Matthew

7 TCORF 2017, 71st IHC - March 14 - 16

Results

PRELIMINARY - FIRST RUNS OF THIS KIND - MORE ANALYSIS TO BE DONE!

• Track errors

- Atlantic basin

- Globally

- Hurricanes Hermine & Matthew

8 TCORF 2017, 71st IHC - March 14 - 16

Track Errors in Atlantic

T COR F 2017, 71s t IH C - M arch 14 - 16

ATLANTIC GH targeted storms: Gaston, Hermine, Karl, Matthew

10 TCORF 2017, 71st IHC - March 14 - 16

ATLANTIC all storms in study period

11 TCORF 2017, 71st IHC - March 14 - 16

Global Impacts from Global H awk Drops onde

A s s im ilation in the A tlantic

T COR F 2017, 71s t IH C - M arch 14 - 16

An impact was also noted for Eastern and Western Pacific storms by including the global hawk obs ervations in the Atlantic. Indicates an overall improvement on a global s cale.

EASTERN PACIFIC

13 TCORF 2017, 71st IHC - March 14 - 16

An impact was also noted for Eastern and Western Pacific storms by including the global hawk obs ervations in the Atlantic. Indicates an overall improvement on a global s cale.

WESTERN PACIFIC

14 TCORF 2017, 71st IHC - March 14 - 16

WESTERN PACIFIC

15

Track Error Reduction

H erm ine & M atthew

T COR F 2017, 71s t IH C - M arch 14 - 16

Hurricane Hermine

17 T COR F 2017, 71s t IH C - M arch 14 - 16 T COR F 2017, 71s t IH C - M arch 14 - 16

~15% error reduction at 120hrs

H urricane M atthew

Hurricane M atthew track s

18 T COR F 2017, 71s t IH C - M arch 14 - 16 T COR F 2017, 71s t IH C - M arch 14 - 16

BONUS

Im pact to H WR F

T COR F 2017, 71s t IH C - M arch 14 - 16

HWRF Track & Intensity Improvement w/ GFS+GH IC/BC

- Gaston

20

Hurricane Gas ton - H WR F

21 T COR F 2017, 71s t IH C - M arch 14 - 16 T COR F 2017, 71s t IH C - M arch 14 - 16

Future/Ongoing work

• First study with significant positive impact from dropsondes. Investigate more fully the underlying reasons for improvement.

• Additional cases:

- Storms in 2012 and 2014 - Non-tropical cases (i.e. El Nino Rapid Response 2016 campaign)

• Early results are very promising, therefore currently in discussion to turn on Global Hawk in operational GFS.

22 TCORF 2017, 71st IHC - March 14 - 16

Thank you!

Special thanks to Daryl Kleist, Mingjing Tong, Andrew Collard, and Dennis Keyser at EMC for their help setting up and running these experiments.

Questions?

23 TCORF 2017, 71st IHC - March 14 - 16

Extra Sl ides

24

Hurricane Matthew

• In plot:

FY17 - Q3FY17 w/ GH data included

AVNO - Q3FY16 operational GFS

• Track error reduction begins at 48hrs.

• ~37% reduction in track error at 168 hrs (drop from 400nm to 250nm).

25 TCORF 2017, 71st IHC - March 14 - 16