Typhoon track errors of KMA Ensemble Prediction System for TIGGE in 2011 and 2012
Typhoon track errors of KMA Typhoon track errors of KMA Ensemble Prediction System for Ensemble Prediction System for TIGGE in 2011 and 2012TIGGE in 2011 and 2012
SeungSeung--Woo Lee, Juhyung Son, YoungWoo Lee, Juhyung Son, Young--Youn Park, and Hoon ParkYoun Park, and Hoon Park
Numerical Model Development Division, KMANumerical Model Development Division, KMA
IntroductionIntroduction
• The KMA EPS has been run operationally since May 2011 at global domain with N320 resolution (approximately 40 km in mid-latitudes) and 70 vertical levels.
• Total 24 ensemble members including control run and the initial condition perturbations are generated using an Ensemble Transform Kalman Filter.
• The upgraded KMA EPS gives better guidance to forecasters especially in prediction of typhoons.
• In this study, we compare the typhoon tracks from each operational centers participating in the TIGGE project for the year 2011 and 2012.
History of KMA EPSHistory of KMA EPSOperation
period 2001.3.1 - 2003.11.1 - 2006.7.1- 2008.9.1- 2011.3.5- 2012. 6-
Machine 1st (NEC SX5) 2nd (Cray X1E) 3rd (Cray XT6)
Data Asssim 2dOI →3dOI
3dOI →3dVar 3dVar 3dvar(self-
cycle) 4dVar(from global model)
Model GDAPS (JMA GSM) UM(UKMO)
Horizontal Res T106 T213(60km) 40km
Vertical Res 21 30 40 50/70 70
Perturbation method
(Target area)Breeding(Global)
Breeding + Factor Rotation(NH)
ETKF(localized in hor and vert)
ETKFwith SST
perturbation
Model error NO RP, SKEB RP, SKEB2
Run per days 1 (12UTC) 2 (00, 12UTC) 2 (00, 12UTC)
Lead time(days) 10 10 10
Members 16 (16 members + 1 control) (16+1)*2 24
Post Processing NO Bias Correction (Bo Cui) -
Track error in 2011Track error in 2011
Track error in 2011Track error in 2011
1102(SONGDA)1102(SONGDA)
6
Typ. NO. 1102(SONGDA)
Lifetime(UTC) 2011052118~2011052906
Verification [NO. of Init.]
2011052212~2011052812
[13]
1109(MUIFA)1109(MUIFA)
7
Typ. NO. 1109(MUIFA)
Lifetime(UTC) 7.28.06~8.9.00
Verification [NO. of Init.]
7.29.12~8.7.12 [19]
Typhoon list in 2012Typhoon list in 2012
���� ��� ��� ���� ��� �� ��
1 ��� PAKHAR �� �� 2012.03.29 09:00 2012.04.02 09:00
2 �� SANVU �� �� 2012.05.22 09:00 2012.05.28 12:00
3 ��� MAWAR �� �� 2012.06.01 15:00 2012.06.06 15:00
4 �� GUCHOL �� �� 2012.06.12 15:00 2012.06.20 09:00
5 �� TALIM �� �� 2012.06.18 03:00 2012.06.21 06:00
6 ��� DOKSURI �� �� 2012.06.26 21:00 2012.06.30 09:00
7 �� KHANUN �� ���� 2012.07.16 15:00 2012.07.19 12:00
8 ��� VICENTE �� �� 2012.07.21 21:00 2012.07.25 03:00
9 ��� SAOLA �� �� 2012.07.28 09:00 2012.08.03 15:00
10 ��� DAMREY �� ���� 2012.07.28 21:00 2012.08.03 15:00
11 ���� HAIKUI �� �� 2012.08.03 09:00 2012.08.09 21:00
12 ��� KIROGI �� �� 2012.08.08 09:00 2012.08.10 09:00
13 ��� KAI-TAK �� �� 2012.08.13 09:00 2012.08.18 15:00
14 �� TEMBIN �� ���� 2012.08.19 09:00 2012.08.31 00:00
15 ��� BOLAVEN �� ���� 2012.08.20 15:00 2012.08.29 06:00
16 �� SANBA �� �� 2012.09.11 09:00 2012.09.18 09:00
17 ��� JELAWAT �� �� 2012.09.21 03:00 2012.10.01 15:00
18 ���� EWINIAR �� �� 2012.09.24 21:00 2012.09.30 09:00
19 ��� MALIKSI �� �� 2012.10.01 15:00 2012.10.04 15:00
20 �� GAEMI �� �� 2012.10.01 21:00 2012.10.06 21:00
Typhoons in 2012Typhoons in 2012
SAOLA
Typhoon 2012 best tracks from RSMC
DAMREY
TEMBIN BOLAVEN
SAOLA
DAMREY
TEMBIN
BOLAVEN
Verification methodVerification method
• Position Error [km]
The distance between the best track (from RSMC) and the forecast track.
• Along Track vs. Cross Track bias
Along-track bias : The bias in the direction of TC movement
Cross-track bias : The bias in the rectangular direction of TC movement
Track error in 2012Track error in 2012
Center No. of samples
CMA 122CMC 20
ECMWF 122UKMO 121NCEP 39JMA 122KMA 123
Track error in 2012Track error in 2012
• Along vs. cross track errorFaster
(AT bias)
Slower(AT bias)
Rightward(CT bias)
Leftward(CT bias)
Center No. of samples
CMA 122CMC 20
ECMWF 122UKMO 121NCEP 39JMA 122KMA 123
Along vs. cross track biasAlong vs. cross track bias
Track error in 2012 Track error in 2012 (same samples)(same samples)
• Same samples for each center Center No. of samples
CMA 39CMC 20
ECMWF 39UKMO 39NCEP 39JMA 39KMA 39
Track error in 2012 Track error in 2012 (same samples)(same samples)
• Along vs. cross track errorFaster
(AT bias)
Slower(AT bias)
Rightward(CT bias)
Leftward(CT bias)
Center No. of samples
CMA 39CMC 20
ECMWF 39UKMO 39NCEP 39JMA 39KMA 39
Along vs. cross track bias Along vs. cross track bias (same samples)(same samples)
Track error for 4 typhoonsTrack error for 4 typhoons
• 4 typhoons which has a track near KoreaNumber Name
1209 SAOLA1210 DAMREY1214 TEMBIN1215 BOLAVEN
Track error for 4 typhoonsTrack error for 4 typhoons
• Along vs. cross track errorFaster
(AT bias)
Slower(AT bias)
Rightward(CT bias)
Leftward(CT bias)
Number Name
1209 SAOLA1210 DAMREY1214 TEMBIN1215 BOLAVEN
Along vs. cross track bias Along vs. cross track bias (4 typhoons)(4 typhoons)
Track error for BolavenTrack error for Bolaven
• Track error increase largely after 60 hours for KMA eps
Track error for BolavenTrack error for Bolaven
• Along vs. cross track errorFaster
(AT bias)
Slower(AT bias)
Rightward(CT bias)
Leftward(CT bias)
Along vs. cross track bias(Bolaven)Along vs. cross track bias(Bolaven)
Track error for BolavenTrack error for Bolaven
• Deterministic vs. Ensemble Prediction
EPS Track for BOLAVEN for each CenterEPS Track for BOLAVEN for each Center
BLUE : KMARED : JMAGREEN: ECMWF
Along vs. cross track bias(Bolaven)Along vs. cross track bias(Bolaven)
SummarySummary
• Even though total sample numbers are different for each operational center, ECMWF and NCEP EPS show best performance in typhoon forecast.
• Hybrid 3DVAR/ENKF in NCEP GFS may play an important role in typhoon forecast performance.
• KMA EPS show reduction in track error for 4-typhoons that affect directly compared with whole typhoons in 2012.
• Most of all operation centers predict westward tilted and faster movement compare with analyzed track for typhoon Bolaven.
Thank you for your attention!Thank you for your attention!