John McBride

Noel Davidson wrote a few days ago:

"Such high latitude events have preceded every monsoon onset event that we
have looked at (~ 8 now)".

My immediate response is "Which is chicken and which is egg?".
The development of the monsoon is forced by the seasonal cycle, whereas
the sudden onset is a trigger that sets it off once the (seasonally
forced) large scale is  ready.  An onset is going to occur every year
around that time of year....
So... why does your high latitude jet-stream mechanism occur every year
around that time of year?

Second comment is that a number of people (shall remain nameless but one
has the acronym hhh and another is a bit of a youthish wonder) would think
the  onset is kicked off (once the large scale is ready) when the next MJO
wanders through.... how does this fit with your model?

cheers

John McB

Harry Hendon

Hi, what the wonder man recently showed was that onset at Darwin
(as precisely defined by wasyl) never occurred in the active phases of the mjo
(here, 4 phases of the mjo are defined., with each phase spanning about 12 days (1/4 cycle).
other than that, onset date is all over the chart with respect to the active phases of the mjo.

Noel Davidson

G'Day Mate, You raise some interesting questions :
--------------------------------------------------------------
"Such high latitude events have preceded every monsoon onset event that we
have looked at (~ 8 now)".

JMB 1 : My immediate response is "Which is chicken and which is egg?".
The development of the monsoon is forced by the seasonal cycle, whereas
the sudden onset is a trigger that sets it off once the (seasonally
forced) large scale is  ready.  An onset is going to occur every year
around that time of year....
So... why does your high latitude jet-stream mechanism occur every year
around that time of year?

1. There is great interannual variability in the time of onset and thus there must be shorter-term controls than just the seasonal cycle. I believe the main controls are still the MJO and extra-tropical interaction (see below).

2. We have looked at 40+ days prior to Wasyl's onset dates. Within that period in each year there have been 1 - 3 major cyclogenesis events near 55S, 100E. ie, these events appear to develop in a REGULAR BUT NOT FREQUENT manner during Dec and Jan, and so have a similar variability to that of onset (only 8 cases tho').
In some years there has been only one distinct event that PRECEDES onset - no chicken and egg here?? When there are two events these can correspond with a "false" onset (ie, with a brief, active monsoon) prior to the real action
OR(underline) if there is no STJ, an event may have NO IMPACT ON TROPICS.

3. One more thing - although we often think of onset as a sudden event - and for westerly winds it often is - another feature we have noted (also found by Hendon et al for the AMEX onset) is the build-up of large scale ascent and
tropospheric moistening (pre-conditioning) from ~ 5 days prior to the development of westerly winds. ie, onset really occurs over these sorts of time scales (5+ days). But what sets up these favourable conditions of ascent and moistening???
- MJO and extra-tropical interactions.

JMB 2 : Second comment is that a number of people (shall remain nameless but one
has the acronym hhh and another is a bit of a youthish wonder) would think
the  onset is kicked off (once the large scale is ready) when the next MJO
wanders through.... how does this fit with your model?

I believe that onset does coincide with the passage (not necessarily the arrival) of an MJO, but a shorter term and additional (possibly stronger) modulation is the extra-tropical interaction. So not necessarily any conflict here.
ALSO and this is really letting my imagination run wild : if the source area for the MJO is the equatorial Indian Ocean - as some claim - then maybe these major high latitude cyclogenesis events over the Indian Ocean actually
triigger active MJO events which then take on a life of their own and propogate eastwards. What about that for a scenario?
It would be really good to know the climatology of high-latitude cyclogenesis events, and (as you say) to know the impact on lower latitudes (although this is what we are trying to do via idealised simulations).
Last thing : although we have used these simulations to focus on onset, there are much broader implications regarding extratropical - tropical interaction, which are rather interesting (what about the infamous NW cloud-band??).

Gotta go. Hope I haven't said anything too stupid otherwise hhh and bw will beafter me.

Go the Bullies, oops the're gone. Go the Lions.

ND

Matt Wheeler
Okay then, howabout we have a bake-off? Noel's high-latitude wave  trains versus the MJO, to see which has been better associated with the onset of the Australian monsoon for the last 20 years or
so. I've looked at this for the last 23 years in which I've used an equatorial-band EOF as an indicator of the MJO (no time filtering used) and Wasyl Drosdowsky's (J. Climate, 1996) definition of the
onset of the monsoon at Darwin (a very local and point-specific definition of the monsoon based on the zonal wind).

The result I get is shown in the attached figure - admittedly, it is not an easy-to-follow plot, but let me try to explain it.

On thetwo axes of the plot are the amplitude (principal component) of each of the two EOFs that explain the MJO - I call them the real-time multivariate MJO number 1 (RMM1) and RMM2. Together they form an index of the MJO, and I require 2 numbers for the index because the MJO is a propagating phenomenon. Based on these 2 numbers,
I define 8 phases of the MJO, as labelled going in a anticlockwise direction around the plot. In phase 2 and 3 the convection of the MJO is out over the Indian Ocean. In 4 and 5 it is over the Maritime continent, and 6 and 7 over the W. Pacifc. Now the points on the plot show the value of the (RMM1,RMM2) pair at the time of Darwin monsoon onset (westerly winds) for each of the monsoon years as labelled. For example the "00" shows that the MJO was in Phase 5
at the monsoon onset of the year 2000/2001 (it happended to be 4th December, 2000, by Wasyl's index). As another example, at monsoon onset for the 96/97 season (26th December), the MJO was very strong (as reflected by the magnitude of (RMM1,RMM2)), and was in phase 6.

So then, *if* the MJO did have a significant impact on the onset of the monsoon at Darwin, we should see a clustering of these points - and we do! Most points lie in phases 4,5,6, and 7. I'd say that in those years the MJO *did* have an influence on the onset of the monsoon. That leaves only 4 years out of 23 in which it didn't.
Perhaps Noel's mechanism dominated in those 4 years?

Noel, can you cook up an index of your high-latitude wave train to see how well it performs? For a fair comparison, you should create your index before looking at Wasyl's onset dates (I can provide them to you later). It should also not include data anywhere near Darwin.
After we do the "lag 0" comparison, that is, the comparison of indices on the SAME DAY as the onset, we should then do a lag comparison to see if there is much predictability lerking behind each phenomenon (i.e. MJO vs. wave-train). I know that the (RMM1,RMM2) values are quite predictable with a lead of about 10 days or so, so I get a similar "clustering" of points if I plot the (RMM1,RMM2) points a few days before onset (except shifted a little in a clockwise fashion). Unfortunately the "clustering" is not as close as one would hope, being spread over 4 out of 8 phases, i.e., 50% of the phase-space. Predictions of monsoon onset from such a relationship would thus have to be probabilistic.

And a note for John McB. The Drosdowsky onset dates are spread over times from late November to early February - over a time span greater than the period of the MJO. Thus the relationship I am seeing is not due to these things happening at the same point of the seasonal cycle each year.

-Matt.

 Harry Hendon

Hi, I think we also need to look at the sensitivity
of the onset/mjo relation
to the definition of the onset. Wasyl's
definition is pretty much at a point. A broader
index may have an even stronger (and cleaner)
relationship with Matt's MJO events.

Also, Matt
can (could) make a composite for 500mb heights
based on the MJO index, which should show
if there is any systematic evolution on the synoptic-scale
off the west coast of Australia.
 

Harry

Noel Davidson

Hello Matt and others,

This is getting out of hand? I have just a few final comments.

1. Wasyl's onset dates are/is a very, very valuable data set. But as I implied previously, onset seems to be not so easy to define. If you look at a larger scale index (strength and depth of westerly winds over Indonesia and N. Aust.) you get earlier dates (and its not just because the sampling domain extends further westward) - WMONEX is one example.
There also seems to be this pre-conditioning phase during which time ascent and moistening occurs - has onset started
at this point? (Easily rationalised for the MJO as well) If you take both these factors into account, onset can
be 10 to 15 days prior to Wasyl's date. Eg, Wasyl's WMONEX onset date is 6 Jan 1979, moistening and ascent commenced
near 22 Dec 1978. Is this one of the 4 that didn't work or can the MJO account for this 15 day difference as well?

2. As I stated previously, the MJO is important, but there is more to the tropics than just this phenomena. We don't understand how this extratropical- tropical interaction occurs but that doesn't mean it doesn't happen. Eg, What about those active events that occur pre-onset, but don't quite qualify as onset? I would say that there is at least one of these every year - perhaps the forecasters might care to comment. Are these MJO related too? Has anyone demonstrated that midlatitude circulation changes don't affect the tropics?

3. Matthew, we have only just started to look seriously at this extratropical-tropical interaction. To cook up an index that accouns for strength and locations of PFJ and STJ would be foolish at best. But I won't forget your challenge - even though it could be a while before something pops out. Perhaps a smart, young fellow like yourself may care to disprove the high latitude scenario?

4. I have attached some plots from the idealised simulations. If the email bounces I'm sure John will put them in his web site.
Plot1 : Cross -section of initial condition (which is a zonal average,  no zonal asymmetries). Note the PFJ (near 40S, not 50S as I stated previously), the STJ, the MT, the upper tropical easterlies - quite a realistic structure.

 

PLOT2, 3: 950 vorticity and divergence after 6 days of simulation
time. At this time a deep low is located near 45S, 150E.

Note that the topography is only for reference and the model has periodic boundary conditions. Vort and Div are X10**6. Note that the values are not large, but the model is dry with no PBL, so that these may increase 2 to 10 fold after moist processes fire up. Initial values of vort and div are ~8 units and zero. Note how the structure has changed - pretty interesting, hey. Nice environment for a large-scale convective outbreak to get started. (The convergence band that extends into
the tropics is not the "cold front", which can be seen to the south. We don't know how and why these developments take place).

I better go and do something for the greater good of the Bureau, instead of this "hobby, fun stuff"?

See ya,

Matt Wheeler

Noel,

....a reply to your messages of Friday and Monday (note that my  e-mail on Friday was sent before I had a chance to read yours from Friday - this may explain some things about my response? I wasn't trying to be too argumentative.).

First, I agree that the MJO does not explain everything about the monsoon onset. Otherwise, all of Wasyl's onsets should occur in a single phase of the MJO in my phase-diagram (RMM1,RMM2) plot. Instead, they are spread over about 50% of the phases. Hence as Harry has noted, it is more significant to say from this that during the suppressed phases of the MJO we are very unlikely to get a monsoon onset. During the enhanced phases (Maritime Continent to Western Pacific convection), however, it can occur at any time. And it may well be that to pin down this time more precisely, it may
take something like extropical-tropical interaction.

Now for some specific replies.

> 1. Wasyl's onset dates are/is a very, very valuable data set. But as I
> implied previously, onset seems to be not so easy to define. If you look at
> a larger scale index (strength and depth of westerly winds
> over Indonesia and N. Aust.) you get earlier dates (and its not
> just because the sampling domain extends further westward) - WMONEX
> is one example.
> There also seems to be this pre-conditioning phase during
> which time ascent and moistening occurs - has onset started
> at this point? (Easily rationalised for the MJO as well)
> If you take both these factors into account, onset can
> be 10 to 15 days prior to Wasyl's date. Eg, Wasyl's
> WMONEX onset date is 6 Jan 1979, moistening and ascent commenced
> near 22 Dec 1978. Is this one of the 4 that didn't work or
> can the MJO account for this 15 day difference as well?

This is not *quite* one of the 4. That season is labelled in my phase diagram as "78", which lies in phase 7. If we changed the date to 22 Dec 1978, however, I imagine that the (RMM1,RMM2) point would lie in either phase 4 or 5 since with time the (RMM1,RMM2) point tends to rotate around the origin of the phase space in the anti-clockwise direction, taking about 4-8 days per phase (I don't know for sure about the location, however, as we don't have the OLR for 1978 to get the
projection for that date). Phases 4 and 5 correspond to the times that I'd expect large-scale moistening and ascent around Darwin, so this is consistent. If we were to change all the "onset" dates to the times of large-scale moistening and ascent around N.Australia, I imagine that the (RMM1,RMM2) point would tend to lie in phases 3, 4, and 5, that it, shift back in time a little. This would be an interesting thing to look at.

Do you have "onset" dates of your own that I can use? That is, ones that reflect the large-scale moistening and ascent? This could be useful for me.

> 2. As I stated previously, the MJO is important, but there is more
> to the tropics than just this phenomena. We don't understand how
> this extratropical- tropical interaction occurs but that doesn't
> mean it doesn't happen. Eg, What about those active events that

I agree.

> occur pre-onset, but don't quite qualify as onset? I would say that
> there is at least one of these every year - perhaps the forecasters
> might care to comment. Are these MJO related too? Has anyone
> demonstrated that midlatitude circulation changes don't affect
> the tropics?

I agree that it probably does.

>
> 3. Matthew, we have only just started to look seriously at this
> extratropical-tropical interaction. To cook up an index that accounts for
> strength and locations of PFJ and STJ would be foolish at best.
> But I won't forget your challenge - even though it could be a while
> before something pops out. Perhaps a smart, young fellow like
> yourself may care to disprove the high latitude scenario?

Hmmm, I'll think about it. I may try Harry's suggestion first, as I would'nt know where to start with the PFJ and STJ.

-Matt.