Date: Tue, 11 May 2004 16:15:24 +0930
From: Sam Cleland

Hi John and others

Rossby waves seem to have been a feature in recent weeks. Early April saw good structure in the western Pacific, with cyclonic circulations symmetrical about the equator and bursts of westerly wind seen in the  near-equatorial regions there. This was the set-up that saw the start of  Typhoon Sudal, In the current iteration of Matt's filtered animation, you
can follow the current Indian Ocean equatorial Rossby wave from the Pacific  in early April. The stretch through the Indonesian Archipelago is pretty weak, but I guess any weak low-level vortex is going to struggle through
that type of orography?

In the current situation, it seems as though a MJO event has been superimposed on all of this. Have a look also at Matt's RMM Index at:
http://www.bom.gov.au/bmrc/clfor/cfstaff/matw/maproom/RMM/phase.Last40days.gif.  (Copy current at time of this message here)
Considering the current active broad-scale near-equatorial convection over  Aus longitudes in recent days, and the lead-up of substantial convection in the Indian Ocean over the past two weeks, this seems like a good representation.

There seems to be logic in transition periods being favourable for symmetric set-ups about the equator, (eg equatorial Rossby waves), and I can add the observation that a preliminary frequency-analysis of email discussion list topics showed a "Rossby" peak in May and November! From memory, November 2003 was a very good month for equatorial Rossby waves in the Indian Ocean. This is the set-up for equatorial westerly wind bursts, so why not monsoon onsets?

Cheers

Sam

Matt Wheeler
Date: Tue, 11 May 2004 17:50:08 +1000 (AEST)
From: Matthew Wheeler
Subject: Re: [Monsoon] n = 1 Equatorial Rossby wave in Indian Ocean:  Possible role in
    first appearance of monsoon westerlies

> c) The OLR diagnostics pick up a Kelvin wave travelling rapidly through
> the region between about 3 May and 9 May, which again I find difficult to
> see in the unfiltered wind signature.... though  it probably IS there...
> look at the little westerly burst at 150E on the last analysis in the
> sequence (12 UTC 10 May).  Time  permitting, I'll run a hovmoller of zonal
> winds at the equator and see if I can pick up the Kelvin wave signal).

John, I have a hovmoller of u850 (5S to 5N average) at
http://www.bom.gov.au/bmrc/clfor/cfstaff/matw/maproom/NCEP+GASP/hov.last6m.an.EQ.u850.gif

Copy current at time of email: here

The Convectively-coupled Kelvin wave is quite easy to see in this plot. It started way back in the Pacific, and has gone all through the Atlantic to appear in the Indian Ocean.

(compare to the OLR-only plot, with filtering, at
http://www.bom.gov.au/bmrc/clfor/cfstaff/matw/maproom/OLR_modes/h.6.ALL.EQ.html )

(Copy current at time of mail here)
-Matt.

 Matt Wheeler

Date: Tue, 11 May 2004 18:15:52 +1000 (AEST)
From: Matthew Wheeler
Subject: Re2: [Monsoon] n = 1 Equatorial Rossby wave in Indian Ocean:   Possible role in
    first appearance of monsoon westerlies

> There seems to be logic in transition periods being favourable for
> symmetric set-ups about the equator, (eg equatorial Rossby waves), and I
> can add the observation that a preliminary frequency-analysis of email
> discussion list topics showed a "Rossby" peak in May and November! From
> memory, November 2003 was a very good month for equatorial Rossby waves in
> the Indian Ocean. This is the set-up for equatorial westerly wind bursts,

Yes, the transition seasons often show similar examples of symmetric
vortices, sometimes combined with n=1 ER waves, MJOs, or both. I have
put a small collection of samples from recent years in a powerpoint
presentation on the web at (will be available in < 20 mins)

http://www.bom.gov.au/bmrc/clfor/cfstaff/matw/junk/

-Matt.

Adam Sobel
Date: Tue, 11 May 2004 09:16:50 -0400
From: Adam Sobel John, Matt, et al.,

this is indeed a beautiful textbook case.  Thanks for bringing it to our attention.

I suppose this is why they call the MJO a "Kelvin-Rossby" wave.  Has some
properties of both, presumably involves some coupling of the two, through nonlinearity
and/or convection, and thus doesn't move at the speed of either one.

However another issue that has always bothered me is the role of background flows
on the Wheeler-Kiladis plots, which involve dispersion curves for resting basic states.
I know I have discussed this with Wheeler and/or Kiladis at one time or another but
I can't remember what they said about it... maybe Matt can explain how much this may
be relevant to interpreting what the filtered diagnostics show.  Seems particularly
relevant to slow-moving disturbances, where any background flow influence would
change the propagation speed a lot, percentage-wise.

Another question which interests me is how much tropical cyclone dynamics (meaning,
active coupling of convection to the vortical flow through wind-evaporation feedback
as operates in tropical cyclones) may be helping to strengthen this pattern.  Whether or
not the tropical depressions get to named strength doesn't necessarily tell us the answer
to this.  People argue a lot about how convection and the large scale flow couple in general,
but in TCs is the one place where we don't debate that it works or how it works.  To the
extent that TC dynamics can project on larger-scale patterns (e.g. through Rossby wave
dispersion, TC is a vortex on a beta plane) , it makes a hard problem a lot easier.

Adam

Adam H. Sobel
Department of Applied Physics and Applied Mathematics
Department of Earth and Environmental Sciences
Columbia University
 

PS does this count as an early monsoon onset over India?  We have westerlies
over a good portion of the west coast for a while during John's sequence.

Adam

Matt Wheeler

Date: Wed, 12 May 2004 09:36:49 +1000 (AEST)
From: Matthew Wheeler

Adam asked about the influence of the background flows on the various
waves that appear in the OLR wavenumber-frequency spectra
(that is, the waves as appearing in the figure on the web page at
http://www.bom.gov.au/bmrc/clfor/cfstaff/matw/abstracts/WK99.html ).

He is correct in pointing out that the dispersion curves for which
the filtering is designed are based on a resting basic state, as are
the dispersion curves in the original spectra. For the slower moving
waves (i.e., small phase speed = freq/wavenumber), the effect of a
background flow will be greater. This, I believe, is the reason why
the n=1 ER wave spectral "peak" is more washed-out than the peak
for the Kelvin or low wavenumber MRG waves. For example, look how
sharp the wavenumber-0 MRG wave spectral peak looks!

In practice, what this means is that the filtering diagnostic for
the slower moving waves (i.e., especially the ER wave) is likely
to miss some instances of them in nature. Conversely, the filtered
fields also contain variance from noise that is not related to the
waves. So it is not perfect. However, I don't see how the filtering,
as a simple diagnostic, can be improved to take into account such
background flows in any simple way. Adam, at what level would you
use for a background "steering" flow? Or should one use a tropospheric
average? A further complication is that as soon as one starts thinking
about equatorial waves and background flows, one realises that there
is vertical shear of that background flow, and vertical shear makes
the primitive equations unseparable, and all this use of shallow
water dispersion curves becomes less certain. In the end, however,
the proof is in the pudding, and the pudding to me is that features
that look like the shallow water dispersion curves DO exist in the
spectra, and at times (like now), filtering along these curves
appears very useful for diagnosing the different modes that are
occuring.

-Matt.

(A side note.... Interestingly, the spectral peak of the MJO appears relatively sharp compared to that of the n=1 ER waves. Yet they both move with relatively the same phase speed. The question thus arises, how can the MJO maintain a relatively contant phase-speed as a convectively-coupled signal in the presence of the same background flows? I think the answer is that when the MJO is active, it is so strong that it defines/generates those background flows. There is nothing bigger than it to affect it in such a way! )