NCEP Climate Forecast Systems T62 vs' T126: Annual Cycle, ENSO and its Decadal Changes

1
NCEP Climate Forecast Systems T62 vs. T126
Annual Cycle, ENSO and its Decadal Changes

• Yan Xue
• Climate Prediction Center
• Acknowledgements
• Suru Saha, Wanqui Wang, Kyong-Hwan Seo, Boyin
  Huang

2
Climate Forecast System (CFS) T62

• 1. Atmospheric component

• Global Forecast System 2003
• T62 in horizontal 64 layers in vertical

2. Oceanic component

• GFDL MOM3
• 1/31 in tropics 11 in extratropics 40
  layers
• Quasi-global domain (74S to 64N)

3. Coupled model

• Once-a-day coupling
• No flux correction
• Sea ice extent taken as observed climatology

3
Wang et al. 2005

• Climatology well simulated
• NINO3.4 amplitude larger
• than observation
• Phase-locking to end of year
• Early onset and late decay
• ENSO too regular

Kirtman 2005

• Models low frequency
• modes differ from observation
• Initialization shock

4
T62 vs. T126

• Climate Forecast System T62 went operational in
  2004.
• Keeping the oceanic component unchanged, the
  horizontal resolution of the atmospheric
  component was increased from T62 to T126.
• Two sets of simulations with T126, each 100 year
  long, were studied here.
• Four sets of simulations with T62, each 32 year
  long, were studied here.

5
S
W
S
W
W
S
S
W
6
Penland and Saha, 2005
T62

T126
OBS
7
Motivation

• Characteristics of observed ENSO change with
  time pre- and post-1976
• Mechanisms for decadal modulations of ENSO
• Background state changes (Fedorov and Philander
  2000)
• Interaction between annual cycle and ENSO
• Atmospheric noise forcings
• Interactions between the tropics and subtropics
• Nonlinear dynamics
• What are the mechanisms controlling ENSO
  characteristics in T62 and T126?
• What are the biases of the mean and annual cycle
  in T62 and T126?
• How do the biases in models influence their ENSO
  characteristics?
• Why do the ENSO characteristics in T126 change
  from decade to decade?

8
Data

• ERSST in 1950-2005
• FSU wind stress in 1978-2005
• CMAP precipitation in 1979-2003
• Depth of 20oC isotherm from NCEPs Global Ocean
  Data Assimilation System in 1979-2005
• Monthly and daily fields from T62
• Monthly and daily fields from T126

Methodology

• Spectrum analysis
• Linear regression
• Composite analysis
• Interannual Coupling Strength and Thermocline
  Coupling Strength
• Intraseasonal variance in 20-90 days

9
(No Transcript)
10
(No Transcript)
11

• Amplitude of SST anom. in T62 and T126 (1.2oC)
  is similar to obs except the maximum center is
  detached from the coast.
• Spatial structure of SST anom. is similar to
  obs except its meridional width is a little too
  narrow.

12

• Amplitude of prec. anom. in T62 and T126 (2.5
  mm/day) is similar to obs except there are two
  maximum centers located to north and south of the
  equator.
• Spatial structure of prec. anom. is similar to
  obs except convection south of equator extended
  too far eastward, subsidence over Indonesian
  continent is too weak and subsidence to north of
  ITCZ too strong.

13

• Amplitude of westerly anom. in T62 (1.2
  dyn/cm2) and T126 (0.9 dyn/cm2) is smaller than
  that of obs (1.5 dyn/cm2), but easterly anom. to
  north of westerly center is too strong.
• Spatial structure of zonal wind stress anom. is
  similar to that of obs except its meridional
  width is too narrow.

14

• Amplitude of positive D20 anom. in T62 (20
  meter) and T126 (20 meter) is a little larger
  than that of obs (15 meter).
• Positive center in the eastern Pacific is too
  equatorially confined, and the negative center in
  the north-western Pacific is located too close to
  the equator, and extended too far eastward.

15
Lag

• SST anom. in T62 propagates eastward, while SST
  anom. in T126 is largely stationary, close to
  obs..
• Duration of SST anom. in T62 and T126 are too
  long compared to obs..
• Onset of ENSO is too early in T62.

16
Lag

• Prec. anom. in T62 propagates from the western
  Pacific to eastern Pacific, close to obs., while
  prec. anom. in T126 has no propagation.
• Prec. anom. in the far western Pacific in T62
  and T126 is too weak.
• Prec. anom. in T62 is too large in the Indian
  Ocean.

17
Lag

• Zonal wind stress anom. in T62 propagates from
  the western Pacific to central Pacific, similar
  to obs., while T126 anom. is largely stationary.
• Easterly anom. in the far western Pacific in
  T126 is too weak.
• Anom. in the Indian Ocean in T62 is too strong.

18
Lag

• D20 anom. in T62 propagates from the western
  Pacific to eastern Pacific, similar to obs.,
  while T126 anom. propagates little.
• T62 has too strong D20 precursor.
• T62 has too strong amplitude in the Indian
  Ocean.

19
(No Transcript)
20
(No Transcript)
21
late onset
4 warm events
onset gt 0.5oC Yr 0 normal onset (4) Mar0
May0 late onset (1) Aug
early onset
2 year peaks
12 warm events
normal onset (12) Dec-1 Apr0 late onset (3)
Jul0 Sep0 early onset (4) Sep-1 Nov-1
early onset
late onset
early onset
22
12 warm events
early onset
early onset
onset gt 0.5oC Yr 0 normal onset (12) Jan0 (3),
Apr 0 (1), May0 (1), Jun0 (3), Jul0 (2),
Aug0 (2) early onset (2) Oct-1
23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29
(No Transcript)
30
(No Transcript)
31
(No Transcript)
32
(No Transcript)
33
(No Transcript)
34
Air-See Feedback Loop
Interannual Coupling Strength (ICS)
Thermocline Coupling Strength (TCS)
35
warm SST
large SST gradient

• Interannual Coupling Strength (ICS) is
  strongest during Feb-Mar and Aug-Sep, and weakest
  during May-Jul and Nov-Dec.
• ICS of T62 is weaker than that of observation.

36
OBS
T62
T126 CMIP2
T126 CMIP1

• Both T62 and T126 are more stable than that of
  observation.
• T62 is more unstable than T126 during Nov-Feb.

37
TCS
weak upwelling

• Thermocline Coupling Strength (TCS) is
  strongest during Nov-Jan and weakest during
  Mar-Apr when upwelling is weakest.
• TCS of T62 is much stronger than that of
  observation, particularly in Feb-Apr and Jul-Sep.

38
TCS
T62
T126 CMIP2
T126 CMIP1

• TCS of T62 and T126 are both stronger than that
  of observation, particularly in Feb-Apr and
  Jul-Sep.
• T62 is more unstable than T126 in May-Sep.

39
OBS
T62
T126 CMIP2
T126 CMIP1

• Both T62 and T126 are more stable than that of
  observation.
• T62 is more unstable than T126 during Nov-Feb.

40
T126 CMIP1 51-100 yr
T62
T126 CMIP1 1-50 yr
T62 CMIP1
T62 CMIP1
T126 CMIP1 51-100
41
T126 CMIP2
42
S1
W1
S1
S2
W2
W1
43
CMIP1 S1 W1
44
CMIP2 S1 W1
45
CMIP2 W2 S1
46
CMIP2 S2 W2
47
Summary

• T62 and T126 have similar mean biases
• Negative SST biases (-1oC) in north-western
  Pacific, positive biases (2oC) in south-eastern
  Pacific
• Double ITCZ and too little prec. in the
  equatorial western Pacific
• T62 and T126 have different biases in annual
  cycle on the equator
• SST in the eastern Pacific peaks in May in T62
  and in June in T126
• T126 has eastward propagating westerly and rain
  band in early spring in the far western Pacific
• Both T62 and T126 have too deep thermocline in
  June due to excessive westerly in the far western
  Pacific in spring
• T62 and T126 have similar phase-locking of warm
  events to end of year but T62 has onsets in Dec-1
  Apr0 and T126 has onsets in Jan0 Aug0.

48
Summary

• Interannual Coupling Strength of T62 and T126 are
  both weaker than that of observation, but they
  are compensated by their stronger Thermocline
  Coupling Strength.
• Interannual Coupling Strength of T126 is weaker
  than that of T62 in Nov-Feb, probably due to its
  confined rain band to the far western Pacific and
  delayed eastern Pacific warming.
• T126 is more stable than T62, and its onset of
  ENSO is more irregular than that of T62,
  suggesting that the system is near neutral and
  atmospheric stochastic forcings could play
  significant roles on its ENSO evolution.
• Decadal modulations of ENSO are related to
  background state changes cool in the western
  Pacific, warm in the eastern Pacific, westerly
  anom. near the dateline, deep thermocline in the
  eastern Pacific and shallow thermocline in the
  western Pacific favor strong ENSO variability.

49
Future Work

• Explore the impacts of atmospheric noise forcings
  on ENSO.
• Explain why T62 has a much earlier onset than
  T126 does.
• Explore the roles of the Indian Ocean.
• Explore the roles of the subtropical Rossby waves
  on ENSO.
• Explore mechanisms for decadal modulations of
  ENSO.