Some Preliminary Results from Tropical AirSea Interaction study

1
Some Preliminary Results from Tropical Air-Sea
Interaction study

• Cheng-Wei Chang1 and Jia-Yuh Yu2
• 1. Institute of Geography, Chinese Culture
  University, Taipei, Taiwan,
• cwchang_at_cisk.atmos.pccu.edu.tw, 886-2-28610511
  ext384
• 2. Department of Atmospheric Sciences, Chinese
  Culture University

2
Contents

• 1. Introduction
• 2. Data sources
• 3. SST vs. Wind coupled mode
• 3-1?interdecadal mode
• 3-2?interannual mode
• 4. Hybrid couple model
• 5. Concluding remarks

3
Introduction

• Singular Value Decomposition(SVD)can clearly
  identify the coupled characteristics at the
  interface between ocean and atmosphere
  (Brethertor et al 1992 Chang and Yu 2003).
• Empirical atmospheric models sometimes exhibit
  better simulations in response to SSTA then
  dynamic models in strong boundary forcing
  phenomena, such as ENSO ( Anderson 1999
  Brankovic and Palmer 2000).

4
Singular Value Decomposition

• Step 1 write down space-time matrix
  
• a variable l stations, n
  observations
• b variable m stations, n observations

5

• Step 2 creating cross co-variance of A B

where is a transport of B
6
Step 3 decompose cross co-variance matrix

• where

and
W singular values
7
Step 4 projection of time on U V
F contains singular vectors of A G contains
singular vectors of B
8
Data sources

• NCEP/NCAR re-analysis grid data of atmosphere
  (19562001)
• Reconstruction OI SST (19562001)
• Domain60N45S/0E0W
• Horizontal resolution 55

9
Model Components

• Ocean
• Cane-Zebiak (CZ) model with Niller-Kraus
  vertical mixing scheme
• Atmosphere
• Empirical atmospheric model based on SVD
• projections of the first 7 modes

10
Interdecadal mode
11
Interannual mode
12
SST and Strm (phase2-phase1)
13
(No Transcript)
14
Nino5EQ20?N120?E180? Nino3.45?S5?N170?W
120?W Nino12 10?SEQ90?W80?W
15
Empirical atmospheric model
Rrmse0.291 ACCu0.760 ACCv0.756
16
Empirical atmospheric model
Rrmse0.308 ACCu0.684 ACCv0.652
17
Empirical atmospheric model
Rrmse0.392 ACCu0.695 ACCv0.648
18
(No Transcript)
19
Cane-Zebiak (CZ) model
20
Hybrid Coupled Atmosphere-Ocean Model

• Run 10 years
• Time step 4 hours
• Couple 1 day
• Spin up 1 month(1990.Jan.)
• Adjust -

Real data force(ocean model)
New U , V force
New U , V force
running
EAmodel
EAmodel
Next time
Next time
New SST
New SST
21
Hybrid Coupled Atmosphere-Ocean Model
22
Concluding remarks

• ?SVD1- interdecadal mode
• ?19561975(phase1),19822001(phase2)
• ?SVD2- interannual mode
• ?El Nino ? La Nina
• ?the seasonal change of decadal pattern is
  modest, much different from the ENSO evolution.
  It shows that the decadal oscillation involves
  an entire shift of climate state from one to
  another

23

• ?The decadal oscillation can modulate ENSOs by
  intensifying (weakening) and lengthening
  (shortening) ENSO evolution, as well as
  decreasing (increasing) ENSO frequency in warm
  (cold) epoch
• ?EA model can successfully reproduce ENSO
  variability in the tropics.
• ?Hybrid couple model can simulate the amplitude
  of the ideal ENSO case.
• ?Air-Sea interaction affect the amplitude of ENSO
  case directly.

24
Future work

• Improve mixing parameterization in the ocean
  model
• Try to couple with OGCM and to examine ENSO
  predictability

25
(No Transcript)
26
compute variance percentage for each singular mode

• sum 0.
• do i1,nn
• sum sum sval(i)2
• enddo
• do i1,nsvd
• pcent(i) sval(i)2/sum100.
• print,'mode',i,' percentage ',pcent(i)
• enddo

27
compute principle component vec dd

• do k1,nsvd
• do l1,nt
• pc1(l,k) 0.
• pc2(l,k) 0.
• do i1,nn1
• pc1(l,k) pc1(l,k) vec1(i, k)dd1(i,l)
• enddo
• do i1,nn2
• pc2(l,k) pc2(l,k) vec2(i, k)dd2(i,l)
• enddo
• enddo
• enddo

28
Repeat steps for subsequent modes
Project T onto the first SVD mode of SST(T) to
obtain a dimensional value T
Normalize T through division by the variance of T
association with mode 1
Multiply the first SVD mode of stress W by T to
obtain new W

• The orthogonality of the modes implies that the
  total field W is the sum W and multiply the
  Project Coefficient,d(6.610-4)