As promised, a brief report on the Sumatra floods event.

According to the reports in the Indonesian Press (Jakarta Post Online) , the rainfall, flooding and landslide occurred on Sunday night 2 November.

First a couple of press reports, both from the online Jakarta Post website.  On my web-page (http://www.bom.gov.au/bmrc/clfor/cfstaff/john_mcbride.htm   follow link to monsoon discussion, and then 2003 November 5)  I have placed links  to the text of a number of Jakarta Post Reports

November 4  Floods hit Langkat, 92 killed

November 5a   Survivors found as death toll from flash flood rises

November 6a   U.S. provides US$50,000 assistance for Langkat flash flood victims

November 6b   Environmental minister urges ban on resettlement in flood-hit area

November 6c   President Megawati conveys condolences to Langkat flash flood victims

    As described in the 4 November Report,  the floods were triggered by days of heavy rain and the flooded Bahorok River hit the resort town of Bukit Lawang between 10pm Sunday night and 1 am Monday morning.  In UTC time, this means the event occurred around 1200 2 November 2003.

It seems  of the order of 100 people were killed; and as you'll see in the Press Reports, the Indonesia Press and local Government is blaming upstream deforestation for the disaster.  Despite that, there is no doubt it was a major rainfall event.  Hopefully any Indonesian correspondents to this list may be able to supply us with some actual rainfall data?  Also it would be interesting to know what evidence there is for the role played by deforestation.

Going over the synoptics of the situation, the major factor was a mesoscale low sitting over northern Sumatra during the 24 hours leading to the event. This is shown in the 850 hPa analysis for 0000 2 November,  on my  webpage:  The mesovortex is located at approximately 98E, 4N, i.e over the tip of Sumatra.

As I pointed out in my earlier message, the vortex seems to be part of a vortex pair, with a southern hemisphere cyclonic vortex to the southwest.  The satellite signature is interesting.  I have put on my page links to a series of Eumetsat images for the preceding couple of days:
31st 0100
31st 1330
1st 0100
1st 1330
2nd 0100
2nd 1330

The images show there has been intense diurnal deep convection surrounding Sumatra for the preceding two days with the convection being located over land in the 1330 pictures and over the adjoining seas , but lined up parallel to the coastline on the 0100 pictures.  The scale of the convective organization is quite large, covering all of Sumatra  over about a ten degree latitude by longitude box.  Look for instance at the image for  0100 on the 1st where there is an intense ring of maritime deep convection surrounding the northern half of Sumatra along its coastline.

Looking at the series of charts, the peak of the life-cycle of the meso vortex was around 0000 to 1200 UTC 2 November; and looking at the rainfall prognoses from the operational TLAPS model the maximum rainfall was at this same time.  Thus I am assuming it was a "short-term response" flash flood. That is to say the heavy rain occurred during those preceding 12 hours.  Once again, perhaps our Indonesian colleagues could  correct or confirm this?

In terms of why this event occurred, I believe it was a direct result of the existence of  the mesoscale vortex, which in turn seems to be the product of the larger scale monsoon environment.  I attribute the heavy rainfall to the existence of the vortex simply because the heavy rainfall was "captured' by the operational Numerical Weather Prediction forecasts.  For example, on  my web-page I have placed a link to the relevant 24hr and 48 hr rainfall forecast for the TLAPS model.  Looking at these precip progs, the 48 hour prog is more impressive than the 24 hr prog; but still it is clear the model was producing intense vertical motion off Sumatra in association with mesovortex.  As part of our model-testing we also run a no-convection version of the model, whereby the rainfall is given by Smagorinsky-type large scale condensation whenever the grid scale saturates.  The 48 hour precip prog for this version is also on my page, showing the heavy precip is not brought about by subtleties of the convection parameterisation, but rather is a result of the mesoscale vortex and its associated convergence field.

 So the question is, what caused the mesoscale vortex?   The short answer is that I do not know.  Mong in her email of 4 November suggested it may be related to the existence of Typhoon Melor which was over Taiwan at around the same time.  This suggestion makes a certain amount of sense as the ITCZ at the time was in a "spun-up" mode whereby it had broken up into a number of discrete vortices, one of which was Melor, one was the vortex over Sumatra, and there was also a  vortex further west at the southern tip of India.  (This "spun-up" structure can be seen on the operational TLAPS 900 hPa vorticity analysis on my  page, where northern hemisphere cyclonic relative vorticity is yellow through red.)

This structure has been a feature of the monsoon ITCZ of late; and I suspect it may be a function of  the retreat of the northern summer monsoon which occurs at this time of year,.  Thus, as the ITCZ retreats, it has a band of north-easterlies on its northern side stretching from Western Pacific across to India.  The configuration is highly cyclonic.  On my webpage I have placed links to images of the 850 hPa flow for individual days one week apart for the four weeks preceding the current event:  On the 26 October    there is a vortex north of Borneo and another in the Bay of Bengal at about 10N.  A week earlier on 19 October   there has been a resurgence/regeneration of the monsoon westerly current; so there are strong vortices the entire length of the ITCZ/monsoon trough, in particular east of the Philippines, south of IndoChina and off the east coast of India.  A week earlier still, 12 October   there has been a relaxation of the structure; but on the preceding week 5 October  its quite apparent with a vortex over the Philippines, one southwest of IndoChina and one over the west coast of India.   My feeling is that as we get later in the season, the east-west extending filament of cyclonic vorticity is moving further equatorward, hence the current vortex over Sumatra.   One could speculate that the individual vortices along the trough are caused by a vortex sheet roll-up as happens all northern summer long in the East Pacific ITCZ, and as documented by Ferreira and Schubert,   Journal of the Atmospheric Sciences, Boston, MA. Vol. 54, no. 2, pp. 261-285. 1997, and as documented in real-time in some of my synoptic discussions (see for example: http://www.bom.gov.au/bmrc/clfor/cfstaff/jmb/02Oct24a.html   and  http://www.bom.gov.au/bmrc/clfor/cfstaff/jmb/02Oct28a.html).  The idea this process is also occurring in the northern autumn transition over the monsoon longitudes  is mere speculation at this stage, however.

As satellite retrievals and the science of data assimilation have advanced over the years, we seem to have fairly reliable large scale tropical analyses in real-time.  I am getting the impression that intense equatorial vortices are less rare than we once may have thought.  Some of you may remember we actually had a tropical storm sitting over Singapore in December 2001 as was documented in our synoptic discussions at the time ( http://www.bom.gov.au/bmrc/clfor/cfstaff/jmb/singapore.htm  ).

So: summary of the current event:

a) Almost certainly it was a direct result of the mesovortex lying over northern Sumatra.  The fact that the operational models gave intense rainfall, even when running without a cumulus parameterisation indicates the rain is a property of the balanced convergence fields associated with this vortex.

b) I have hypothesized the vortex is a result of an instability of the ITCZ at this time of year as it retreats towards the equator and has the strong northeasterly monsoon current on its polar side.

For the record, the Madden Julian Oscillation (MJO) was not active at the time and so did not play a role.  See the real-time diagnostic of Matt Wheeler http://www.bom.gov.au/bmrc/clfor/cfstaff/matw/maproom/OLR_modes/h.6.ALL.EQ.html, the value of the diagnostic at the time of the event being linked on my webpage (here)

The event was a few days ago now, and the subsequent sequence is interesting.  The vortex and its "twin" to the southwest straightened up and acquired the signature of a classical n=1 Rossby wave, with its equatorial westerly jet and twin vortices.  These waves have a westward phase velocity, and accordingly the system has propagated westwards. I have links on my page to last night's GASP 850 hpa analysis  for the eastern hemisphere, where you can see a couple of cyclonic vorticity centres  (yellow) in the northern hemisphere on either side of India and their cyclonic twins (blue) in the southern hemisphere.  The linked Eumetsat image for last night has the characteristic twin vortex structure we have learned to recognize in recent years.  (see for example the real-time discussion at http://www.bom.gov.au/bmrc/clfor/cfstaff/jmb/02May07b.html
 

Despite the length of the above analysis I haven't said anything at all about the upper-level flow.  I do not understand this yet, partly due to the fact I haven't spent much time looking at it.  I have attached a link to a 200 hPa absolute vorticity analysis for the time of the event.   As would be expected the Sumatra vortex is not obvious at this level, consistent with it being a warm-cored tropical-type weather system.  The subtropical ridge (which in a classical model should overlie the ITCZ) is a fair way north, along about 20N in fact; so it is a long way from the low-level ITCZ as characterized by the cyclonic vorticity filament on the charts discussed earlier.  There is a nice symmetric easterly current along the equator across the Indian Ocean, which I speculate may be conducive to equatorially-trapped wave type dynamics. However, as I said before I haven't looked much at the upper charts for this event; so perhaps the less said the better.

So... that's it.  Hopefully someone out there will be stimulated to add their two bob's worth: agree, disagree, add some interpretation, etc.

cheers

John McBride