Part ONE:  Discussions on the afternoon of 2 December, i.e about 6 to 12 hours prior to the event

John McBride

Date: Tue, 2 Dec 2003 05:33:04 +0000 (GMT)
From: John McBride <jmb@BoM.GOV.AU>
To: synoptic_discussion@BoM.GOV.AU
Subject: [synoptic_discussion] convective outbreaks breaking out all over.... the country has
    convective chicken pox
 As no doubt you have all noticed, there is impressive convection right now all over the eastern half of the wide, brown land of ours.

I was tempted to send out a note asking why that is... but.. I know the response... you blokes and blokesses will tell me its as obvious as all hell and will be scrambling over one another to talk about convergence lines and triggers and lifted indices and moisture inflows and so on......
 

So.... I'll ask the question another way.... why is all of that stuff (convergence lines, triggers etc) there today... and how is it different to yesterday?  The satellite image at 0900 yesterday morning would have looked similar to this morning, except the front/cloud-mass was further west.  What has changed?

Cheers

John mcB

Bob Moore

Date: Tue, 02 Dec 2003 17:00:06 +1100
From: bobm <r.moore@BoM.GOV.AU>
To: John McBride <jmb@BoM.GOV.AU>
Cc: synoptic_discussion@BoM.GOV.AU
Subject: Re: [synoptic_discussion] convective outbreaks breaking out all over.... the country
    has convective chicken pox

The surface trough was further north in NSW yesterday. Convection got  going from about Albury to Cobar and out to Wilcannia. As the high  slipped east into the Tasman overnight and this morning the trough  and  the high humidity air moved southwest into southern Riverina and Vic  today. Although now the line of storms seems to be similar to yesterday
- Wagga to Brkn Hill.

bob m

Elly Spark
From: ellys <e.spark@BoM.GOV.AU>
Cc: John McBride <jmb@BoM.GOV.AU>, synoptic_discussion@BoM.GOV.AU
Subject: Re: [synoptic_discussion] convective outbreaks breaking out all over.... the country     has convective chicken pox

John,

Apart from the location of the surface trough, the surface moisture is  on its way up.

Yesterdays 4pm moisuter anal over NSW had Dewpoints >15C confined to the  coast, much of inland NSW at 5-10C and NE and SW NSW ar 5-10C.

Same time today and we have most of NSW at >15C and the renmainder with  the exception of Moree at 10-15C. - Nice feedback loop - On this basis  we should see more activity tomorrow. (unless the extra cloud kills the  convection). So have we go a perpetual rain/thunderstorm machine going  here? Probably -- util the front coming on Friday clears the moisture out.

If you look at the TS diagnostic available for  NSW  (check out  "LAPSSTRM.BCH" at
http://serva.nsw.bom.gov.au/web/menu/nswmenu.htm#Thunderstorms click on
and
MESOVIEWER for NSW at
http://gale.ho.bom.gov.au/cgi-bin/bm/wefor/cme/severewx/thunder.pl?type=intro&specific=&day=TDY&region=NSW
you'll be dazzled by the CAPE/LI and potential TS over NSW for next few  days.

NOTE: Todays dewpoints are considerably less than forecast by the  models.. Laps, for example, had large areas of NSW forecast with surface  dewpoints of 20-22C for the same time today. Note that Laps is going for  dewpoints at 18/20 over much of the inland Wednesday afternoon and  20/22C on Thursday afternoon. [ US also 20/22 tomorrow and 20/23
Thursday,  UK 15/18 tomorrow,19/21 Thursday]

Last year dewpoints like this were non existent - and so was the rain...  another feedback loop.

Re models however, the overforecasting of surface (and 850hPa) dewpoints    is a perpetual problem in trying to forecast (a) fire wx and (b) thunderstorms. Any modellers looking for a challenge could usefully  spend their time on what must be a very difficult problem. (US model  probably worse, UK perhaps marginally better than LAPS - this comparison  is a very subjective opinion)

cheers
Elly Spark
Senior Forecaster (NSW)

Rob Webb

Hi Elly et al

One thing that jumps out of the current moisture products is the behaviour of the low-level moisture when there is precipitation. There is a massive increase in surface moisture over a wide area in the models which isn't reflected in observations. This being said, we were only comparing today what we were using 5-10 years ago and that stops me saying too much. Over the inland I used to look closely at the model dewps at a timestep before the convection starts to get a feel for what was there and also the deeper afternoon moisture (say 850hPa).

The moisture today here in Vic from my first observation actually seemed underdone by about 3-4C (not that I was looking). May have made which wouldmake a big difference to those trying to forecast a marginal day for Melbourne.

I might say from afar that someone in VRO must be suffering from Severe Weather Trauma after today's TS in Melb. They just seemed to keep coming, when one headed east, a new one developed to the west. It would have been a
really tough situation to be warning for given the way they storms were moving, splitting etc. I'm just glad it wasn't me.

Part TWO:  Discussions in the following week; i.e post-analysis of the event

Tony Bannister
 Hi Peter,
you and Ken saved the Bureau's bacon last week, well done.  I understand the afternoon ts stuff, Ken said you guys saw a southerly push.  What was in your mind to check on overnight ts rather than phew the afternoon stuff has died down, set up for the following afternoon.

cheers
--------------------------------------------------------------
Tony Bannister
Weather Forecasting Group
Bureau of Meteorology Research Centre

Peter Newham

 Thanks Tony....it was definitely a tricky one.

I guess I have been on enough night shifts now to know that thunderstorms don't always die when the sun goes down....We have been caught out a couple of times in the past with strorms redeveloping overnight in particularly unstable situations.

Given that the moisture and instability was still present over much of the state overnight (I am sure you have seen the 11Z sonde) and a near-stationary surface trough was running across Victoria from the NW-Melbourne-SE, I was quite concerned that there may still have been a few showers or thunderstorms overnight. When doing the 12-12Z Melbourne airport
TAF at about 10Z I had two main concerns.....Firstly: low cloud forming due to the moist low levels with a very shallow inversion (and actually just before the thunderstorms developed some cloud at 300ft started to form!) and

....Secondly: The risk of thunderstorms at Melbourne airport.
We had a CODE GREY for thunderstorms overnight, but no thunderstorm INTER on the TAF at this stage. I discussed with Ken my concerns of the risk of a storm at Melbourne Airport. I saw it as probably only at 10-20% chance of a thunderstorm forming anywhere near Melbourne Airport, but being a bit conservative when it comes to aviation forecasting I thought a prob 30 INTER might be a good idea. After the discussion with Ken it was decided not to amend the TAF, but I was still concerned so continued to monitor the radar every 10 minutes for the next couple of hours. (A good reason why the 3D
radar must stay near the aviation desk!).

The only thing that had been lost from the previous afternoons storms was surface heating......The moisture and potential instability was there and a strengthening low-level trigger (the trough through Melbourne).  Overnight the winds either side of the trough strengthened (something the models didnt properly diagnose and was very difficult to predict with such a lazy
trough), particularly near Melbourne (probably one reason why such huge storms occured over Melbourne and the rest of the state remained fine!)  One of the key ingredients to signficant storm development may have been the development of the northerly low level jet over Melbourne overnight. At 11Z the 3000ft wind at Melbourne was 010/15KT, but by 17Z it had increased to 360/25KT, also the 5000ft increased from 010/12KT to 330/29KT. At the same time the trough that had moved to the southwest to be just southwest of Melbourne in the evening, moved to the northeast over Melbourne, with
southwesterly winds of 5-10 knots increasing for a period to 10-15 knots to the south of the trough. This increase in convergence appears to have been enough to get the storms going.

Stuart was up for much of the night watching the storms and made a couple of interesting observations about storm structure due to being able see with the frequent lightning. He said it appears that the storms cells had reached some kind of steady state, with the strong low level northerly inflow rising up over a quasi-stationary gust front on the northern side of the storm. On the radar it did appear that the storms continued to "back-build" to the north of the initial cells....and hence I had to have lightning alerts out for Melbourne Airport for 2 hours!

Anyways just a few thoughts....we are still scratching our heads a little bit.  Although we weren't advertising much chance of storms overnight, at least Ken and I were onto the situation quickly..... I noticed the cell developing near Craigiburn and said to Ken......"you may need to reissue that severe thunderstorm warning"..it was a little bit tounge in check when the cell first started to develop, but within about 20 minutes there was not much doubt we had a severe storm on our hands (50dbz to 10km and VILS increasing to around 80 (they eventually reached 160!)......the rest is
history

Pete.

Harald Richter
> Peter Newham wrote:

> >Given that the moisture and instability was still present over much of the
> >state overnight (I am sure you have seen the 11Z sonde) and a
> >near-stationary surface trough was running across Victoria from the
> >NW-Melbourne-SE, I was quite concerned that there may still have been a few
> >showers or thunderstorms overnight. When doing the 12-12Z Melbourne airport
> >TAF at about 10Z I had two main concerns.....Firstly: low cloud forming due
> >to the moist low levels with a very shallow inversion (and actually just
> >before the thunderstorms developed some cloud at 300ft started to form!) and
> >....Secondly: The risk of thunderstorms at Melbourne airport.

There is a peculiar dry adiabat in the 850-700 hPa layer in the 11Z MML sounding quickly 'fattening up' the CAPE with height.  Could this be a chunk of well-mixed continental boundary layer air that parked itself atop a low-level convergence
line.  That convergence is quite obvious in the northern suburbs of Melbourne in the 23:30 LT surface plot
The surface analysis shows a pure westerly running into a pure easterly N of Melbourne's CBD.
> >Stuart was up for much of the night watching the storms and made a couple of
> >interesting observations about storm structure due to being able see with
> >the frequent lightning. He said it appears that the storms cells had reached
> >some kind of steady state, with the strong low level northerly inflow rising
> >up over a quasi-stationary gust front on the northern side of the storm. On
> >the radar it did appear that the storms continued to "back-build" to the
> >north of the initial cells....and hence I had to have lightning alerts out
> >for Melbourne Airport for 2 hours!

A quasi-steady storm of this intensity brings up the question of storm rotation. Did Broadmeadows take a look at this storm?  From what I heard, the storm  dropped only smallish hail (below golf balls), long-lasting moderately intense  rain with no wind reports.  It was clear that it was a lightning machine.  What made it such a prolific lightning producer on a day where the Lilydale storms earlier barely managed to discharge once?

I have added a link to  the 23:40 LT radar  image.

A second storm initiated N of Bacchus Marsh and (according to a dinner conversation I had last Saturday) dropped 20 minutes worth of icy Christmas cherries on the eastern part of Bacchus Marsh.  The initiation of that storm is another one of those data void dilemmas, even on Melbourne's doorstep.

Cheers, Harald

John McBride
Thanks Pete,

These soon-after-the-event nots from the person on duty are absolutely terrific.  There have been some great ones in the past from Greg Bond, Ward Rooney and others.  I think we should start collecting them together, along with the cold -hard facts of the event (e.g amount of rain, magnitude of the damage etc) and a couple of relevant charts and possibly soundings.  Whether they should just go on the web somewhere, or in the Aust Met Mag., is not obvious.  But, in my mind they are too valuable for them to be allowed to disappear into the ether.

That said, I had a look at the sounding for the event

The positive area/CAPE is extraordinary for a Melbourne sounding, with the air-parcel going all the way up to 200hPa.  Looking at the sounding, however, one needs a fair amount of lift.  Taking the parcel trajectory that has been plotted on the sounding, being night-time the surface temperatures would have stayed around twenty; so the parcel would have to be lifted almost 200hPa by some convergence mechanism. The mechanism you describe with the change to northerly is interesting; but with that sounding its still a bit hard to see what kicked it off.

I was staring at my screen discussing this with Tony Bannister at lunchtime today.  He says he thinks he still has an automatic archive running of the daily maptool loops over Victoria, left over from his Vic severe weather days.  He's gone off to look for them.   200312022359_loop.gif  200312032359_loop.gif

I assume someone kept (or has retrieved) the radar loop and can direct us to its location?

Looking again at the sounding, once you get on a parcel onto that moist adiabat, its an absolute ripper with (as I said) the undilute air-parcel going all the way up to the 200hPa level.  That "explains" the intensity of the thunderstorm in terms of vigour of the circulation, rapidity of the lightning, etc.  To me, though, it doesn't explain the huge quantities of rain.  The old chestnut on this is that the thunderstorm has to regenerate and be slow moving.... I'm not sure that's the case in this instance, though, as I was lying in bed listening to it immediately prior to jumping out of bed to avoid the stream coming down though the ceiling.  It seemed to me the 100 plus mm occurred over as little as twenty minutes; so "slow-moving" did not seem to be the basic factor.  Any other ideas?  Did anyone keep the operational model rainfall progs?  Is there any hint from
the model moisture fields as to mechanisms for such large localised rainfall?

Cheers

John McB

P.s.  Some after-the-event anecdotes:

a) The event was  reported in the press as a once-in-a-hundred-years event.  I've heard second-hand that Ken Dickinson has over recent years been on shift for three one-in-a-hundred-year events and is starting to no-longer see the joy in the experience.  One of these, was last week's thunderstorm, one I  believe was the Sydney-Hobart, I don't know what the
other was... if the reports are true, perhaps Ken can advise us?

B) Having dried the bed out I was back in it listening the Could-a-beens on the breakfast radio show on Thursday morning.  They have a regular spot where they read the newspaper headlines.  They read aloud the front page headlines from the Sun Herald a quote from John Gill of the weather bureau saying this was a one-in-a-hundred-year event.  The Couldabeens went on to say... "lets see... we cross to the Bureau of Meteorology every day and talk to Ward Rooney, Ken Dickinson, Scott Williams......so,  WHO's JOHN GILL????"

Well... I found it amusing as I lay in bed at 6.00 am still half asleep.

Kevin Torey

Hi All,

With regard to John's comments below about a convergence  mechanism of sufficient intensity to overcome the convective
inhibition, my thoughts immediately turned to a frontal surge-like  mechanism that Michael Reeder and I have found in some of our  idealized frontal modelling studies.  I was reluctant to comment  immediately because the Melbourne Storm was not associated with  a traditional frontal scenario.  However, Noel Davidson decided to  run his experimental severe weather LAPS on the case and invited  me to look at some of the results.  The similarity between the  potential temperature structure and evolution in these results with  the idealized frontal results was striking.

Here's a quick intro to Noel's fantastic severe weather version of  LAPS.

In conjunction with the data assimilation group, we have been  developing a mesoscale assimilation and prediction system, with  specific application to tropical cyclone formation and other severe  weather events. The system incorprates scatterometer, surface  winds from synops, significant level sonde info (soon), and 2  hourly GMS data (in the form of heating profiles for deep  convection). It also includes the latest model developments. The  system has been tested on numerous TC genesis events, and  some other severe weather events. The diagnostics shown here  are from forecasts from this system for the ML flood.

And now back to the story...

I'll try and summarize the idealized frontal surge in a few sentences  and then show a plot illustrating the potential temperature structure  and evolution in Noels results.

In our frontal study we have imposed a diurnal surface  heating/cooling cycle.  During the day deep turbulent mixing leads
to broadening of the trough that leads the cold front.  The cold  front, defined as a potential temperature gradient, regresses as the  classic 1:100 sloping surface steepens to become vertical (due to  turbulent mixing, mixing out the vertical component of the potential  temperature gradient).  The deep turbulent mixing also reduces the  static stability over a deep layer in the trough.  At night radiational  surface cooling decouples the surface layer and there is a rapid  acceleration of the front into the trough.  Convergence leading the  front intensifies and an associated prefrontal updraught is greatly  enhanced.

The following diagram shows the potential temperature structure  (top row) in an east-west cross section passing through Melbourne at 11 AM, 5 PM and 1AM respectively.  The first panel shows a  cold front/warm front-like pattern between the surface and 900  hPa.  6 hours later (5 PM) the cold and warm fronts have become  steepened by the vertical mixing and an implied trough (semi- neutral layer between the fronts) has broadened.  The last image  illustrates the structure after the surge into the trough.  The surge  is most apparent in the cold front.  Note the return to a sloping  gradient and the intensification of the gradient.

Although this evidence for a cold fornt-like surge is by no ways  conclusive it is consistent with our idealized modelling study
(which, by the way, is currently making its way through the internal  review process).  I have looked in more detail at Noel's results  (horizontal sections etc.) and the evidence for this mechanism is  increasing.  This includes a clear low-level front-like surge  propogating from the SSW that arrived in Melbourne close enough  to exactly on time to initiate the storm. I'll continue to investigate  this over the next few days and hopefully come up with a more  conclusive result.

cheers

Kevin

Added later by John McBride: the maptool sfc wind and temperature images corresponding to Kevin's three charts, as extracted from Tony Bannister's maptool loops above:

John McBride

Ah... these things are so much more obvious after the event.

Motivated by kevin Torey's analysis I extracted the relevant maptool images corresponding to Kevin's analysis times: 0000, 0600, 1400 and have put them on my webpage.

The 1400 UTC image, corresponding to the time of the event is particularly interesting.  You can see quite a strong almost-synoptic scale line of convergence across Melbourne, with ten knot southerlies to the soutwest and twenty knot northerlies to the northeast.... easily sufficient to give the 200 hPa uplift required to kick off the convection.

At first I thought the southerly surge discussed by kevin and seen on the maptool image was a sea-breeze... but on further inspection we have a synoptic scale convergence line.
I have put the 1200 msl anal on my page where you can see the convergence is between the northerly flow associated with the ridge in the Tasman and the southerlies associated with the high in the Bight. I have also put up links to the mesoLAPS 10 m wind analysis (from the NMOC chart archive page) for the same time: small large   The LAPS analysys did not really capture the strong northerlies overland; but it has a nice convergence in Bass Strait between the southerlies from the ridge in the Bight and the northerlies associated with the Tasman high.

So.... combining this large surface convergene with the large CAPE sounding, we understand why there was such a vigorous thunderstorm in terms of CAPE, lightning activity etc.  Personally I still do not have a feel for why the rainfall was so large.....Is anyone out there looking at that?

John McB