While other people go to the pub on a Friday afternoon, Guo Ming chose instead to wander around to my office and ask me about the El Nino.  Guo Ming wanted to know whether I was aware of any research looking at the reasons fro the very different responses of the 82-83 versus the 97-98 El Nino's on Oz rainfall.

 As Guo Ming pointed out, the 002-003 is an interseting third case.We have

82-83: strong El Nino   major drought
97-98: strong (H-U-G-E) el Nino... very weak response in Oz rainfall
02-03 weak to moderate El Nino... huge drought... a doozy.

So... why the different responses...??? once the question has been raised, there are all sorts of ways my mind is starting to race and aspects to look at... no doubt, however, the David Jones, Harry Hendons, and Jeff Callaghans out there have already asked the question and have some advanced thoughts on the matter.

Ahh well... Gou Ming has gone...and I'm  off to the pub

John McB

Bill Wright

Hi John,

        I have in fact researched that very question backed by a LWRRDC research grant. It looked at synoptic and oceanic features that differentiated between "normal" El Nino events (in terms of rainfall response in the June-December period over a broad swathe of inland eastern Australia), and abnormal ones (such as 1969 and 1997) in which the "normal" El Nino suppression of rainfall was weak or non-existent. The results (which I have still to submit for publication) suggest that:

        1) the nominal strength of El Ninos is a second-order effect;
        2) abnormally wet El Ninos tend to be associated with higher ocean
temperatures around northern Australia;
        3) there have been more wet El Nino years in the last 30 years than
in the rest of the record, going back to 1900 (which is in line with Neville
Nicholls' finding about trends in El Nino rainfall);
        4) A "wet" El Nino is likely when the stratospheric QBO is in its
positive phase early in the El Nino year, whereas a "dry" El Nino is much
more likely when the QBO early in the year is negative. This is in line with
the work of Janette Lindsay and colleagues on the stratification effects of
the QBO on South African rainfall, and seems to be related to interference
with the Walker circulation.
        5) The results supported Wasyl's work suggesting that SSTs over the
Indian Ocean also modified the El Nino influence - a warm Indian Ocean tends
to be associated with abnormal rainfall in El Nino years.
 
The catch is that preliminary inspection indicates that each of (2) through (5) suggested that the year 2002 SHOULD have had a weak influence on Australian rainfall. I'm looking at this more closely to see if I can learn why...
 
I also feel the actual location of the warmest waters is of significance, but didn't look at that closely. For instance, dry El Nino years such as 1982 and 2002 have strong anomalies in the Nino 3.4 and Nino 4 region, whereas "wet" El Ninos such as 1997 are very strong in the east, but only marginal in the latter regions. In that respect at least, 2002 behaved itself. When I have more time....

There has also been some published work about El Nino strength and the time-evolution of individual events, but this was not directly related to impacts.

Cheers

Bill Wright

John McBride
Part of Bill's thesis is that it is the details of the structure El nino  that  matter....

Usually the strength of the El Nino is specified  by only one parameter. E.g. The amplitude can be specified as the  magnitude of the SOI, Nino-3, or the amplitude of Wasyl and  Lynda's  first EOF of SST... all of these are strongly related to  each other,however, so we generally use just one of them.

However, the structure of the El Nino SST pattern is complex enough that we probably require more than one parameter to measure it.

For example, the El Nino  pattern has an eastern  pacific warm region, a Western Pacific  boomerang-shaped cold  anomaly and the Indian Ocean warm anomaly.... so  possibly three  parameters are required to specify these.... Thus, for example, the current NCC operational forecast model does not forecast with simply EOF -1.  Rather it uses four parameters:
EOF-1, EOF-1 for three months prior, EOF-2 and EOF-2 from three months prior.

Thus... the thesis can be that with the same value of the SOI (or EOF-1) you can get quite different rainfall responses, depending on the other measures of the structure (EOF-1 prior, EOF-2, EOF-2 prior). I shall title this  (above) thesis the COMPLEX-SST STRUCTURE hypothesis.

Bill's other hypothesis is that there are other, independent large scale controls going on that can counter or strengthen the rainfall's response to ENSO... for example the stratospheric QBO.  This hypothesis I'll title the "INDEPENDENT BOUNDARY FORCING" hypothesis.
 

The counter hypothesis, representing the opposite extreme, is that the only large scale boundary condition controlling Australian rainfall is ENSO; but that it only accounts for 50% of the variance.  The reason it only accounts for 50% of the variance is that the atmospheric dynamics responding to the underlying sea surface temperature pattern are chaotic.
For this reason, we forecast and simulate with ensembles.

This CHAOS (or ensemble) hypothesis would have that for the same sea-surface temperature   distribution (e.g the 97-98 temperature pattern, one could run 100 ensembles in a GCM of the winter rain, and there would be a severe drought in a  large sub-family of the ensembles, but there would be normal rainfall in some members of the ensemble).  Of course, we would need to know how to specify the sets of initial conditions for the ensemble runs for  to be able to equate the POPULATION density distribution of the rainfall in the ensembles to the PROBABILITY density function of rainfall
in the atmosphere, which doesn't seem trivial.... nevertheless, I expect such experiments have been done, say by the AMIP and CMIP scientists... so possibly, the simple fact that the atmospheric response is chaotic is the reason for the different  ENSO responses. (AS I said, this is the CHAOS hypothesis).
 

We can learn more..... someone should pull down mean and anomaly charts for each of the three seasons from NCEP reanalyses off the CDC NOAA interactive site.  Also, one should have a look at the Bob Leighton storm-track type diagnostics for each of the three years.  These approaches will give some insight as to the synoptics of how the
atmosphere responded differently.

Cheers

John McB