Charles Jones and Leila M. V. Carvalho

1
Changes in the Activity of the Madden-Julian
Oscillation during 1958-2004
Charles Jones and Leila M. V. Carvalho University
of California Santa Barbara

• Extensively studied over the years but no
  comprehensive theory
• Behavior on time scales longer than interannual
  is unknown

Case to Case
Seasonal Variations
Interannual Variations
Long-term Behavior
Time scales
???
????
?
??
2

• This Study
• Are there linear trends in the activity of the
  MJO? (intensity number of events)
• Does the MJO exhibit regimes of high and low
  activity?
• Are there significant seasonal differences in the
  activity of the MJO on time scales longer than
  interannual?

Note MJO refers to summer and winter events
3

• Identification of MJO events
• NCEP/NCAR U200, U850 (1958-2004) 20-100 day
  anomalies
• OLR (1979-2004) 20-100 day anomalies
• EOF analysis
• Summer domain south Asian monsoon (Lawrence and
  Webster 2001)
• Winter domain Equatorial region (Kessler 2001)

Summer Winter
1979-2004 EOF (OLR) Covariance matrix PC1, PC2 (20) RPC1, RPC2 (43)
1958-2004 combined EOF (U200, U850) Correlation matrix PC1, PC2 (19) PC1, PC2 (40)

• Event amplitude PC1 exceeds 1 sigma
• within 20 days PC2 exceeds 1 sigma
• Events registered at pentads in which PC1 exceed
  1 sigma
• 1958-2004 158 events (75 in summer 83 in
  winter) (U200,U8500)
• 1979-2004 90 events (49 summer 41
  winter) (OLR)

4
(No Transcript)
5
(No Transcript)
6
(No Transcript)
7

• 158 events (1958-2004) identified with CEOF
  (U200, U850) (correlation matrix)
• Each bar represents an event registered at peak
  in PC1
• Amplitude is the variance (15S-15N all
  longitudes) of eastward wavenumbers 1-6 20-100
  day anomalies (Hendon et al. 1999)

8

• Low-Frequency diagram
• Consider XT, T1, 3431 pentads, XT1 event, XT
  0 no event

• Define moving window SK and compute percentage of
  MJO events in SK PK,T MK,T / N where MK,T is
  number of events and N total number of events
• Similar for summer PSK,T MSK,T / NS where
  MSK,T is number of summer events and NS total
  number of summer events
• Similar for winter PWK,T MWK,T / NW where
  MWK,T is number of winter events and NW total
  number of winter events
• SK odd number and varied from smallest (1 pentad)
  to largest possible (3431 pentads)

9
Low-frequency diagram
Cone of Influence
Cone of Influence
10
Winter Summer
11
(No Transcript)
12
(No Transcript)
13
Overall
Mean MJO Low-frequency Variability
Mean PK,T Mean PSK,T Mean PWK,T SK
145 to 657 (1.98 to 9 years)
Summer
Winter
14

• Monte Carlo simulations
• Summer
• XT, XT1 summer event, XT 0 no event
• Randomize seasons 999 times
• Each batch
• compute LF diagram
• mean RSK,T (SK 145 to 657 (1.98 to 9 years)
• correlation PSK,T and RSK,T
• Frequency distribution of correlation gt Cr
• Same for winter

15

• Summary and Conclusions
• MJO exhibits substantial changes in regimes
  longer than interannual
• Two regimes of high activity (1974-1978,
  1988-1992), and low activity (1981-1986).
• Markedly different changes in summer and winter
  activity
• Summer increased from early 1960s to mid
  1970s and decreased steadily until the 1990s
• Winter more regular changes with peaks in 1967,
  1976 and 1989 and lows in 1971, 1983 and 1997.
  
• Changes in summer and winter MJO activity are
  statistically different from random occurrences
• There are positive linear trends in (1958-2004)
• U200 amplitudes of summer and winter MJO events
• Number of summer MJO events
• Linear trends are not higher than random
  occurrences (5 level)
• Trends in U200 amplitudes of the MJO are
  consistent between EOF metric (this study) and
  equatorial zonal mean intraseasonal index (Slingo
  et. 1999)

16
(No Transcript)
17
(No Transcript)
18
Slingo et al. (1999) Intraseasonal Index (QJRMS)

• Interannual variations in the MJO NCEP/NCAR
  1958-1997
• Zonal mean of U 200hPa (10S - 10N)
• 20-100 day anomalies square and apply 100 day
  smoothing
• 40-yr integration of HadAM2a model driven with
  observed SST reproduced general trends in MJO
  activity
• Index contains up to 40 unrelated MJO
  variability (Hendon et al. 1999)

19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)