Special Investigation: Something's Up Down Under (pt.1: Time and ENSO)

Several years ago while I was working on a tropical cyclone database at the University of Washington, I noticed something strange about storms that made landfall in Australia, particularly in the western and northwestern regions. It seems that storms often live a relatively long time after making landfall. Below is an image of all recorded tropical cyclones from the IBTrACS database for both the South Indian and South Pacific ocean basins. The track of storms in the region under investigation are highlighted in bright red (please ignore the few sporadic straight lines that extend outside of the highlighted area).

So, I decided to conduct a little study to try to figure out what makes storms in this region last so long. In the past ten years alone I have found 18 storms that exhibited this behavior, thus I decided to use these as my sample data set for my investigation.

Debbie (2003)	Raymond (2004)	 Emma (2006)
Glenda (2006)	George (2007)	Jacob (2007)
Helen (2007)	Laurence (2009)	Paul (2010)
Carlos (2011)	18U (2011)	Heidi (2012)
Darwin Tropical Low (2012)	17U (2012)	Tropical Low 90S (2013)
Christine (2013)	06U (2014)	09U (2014)

To try to figure this behavior out, I attempted three different types of analyzes: correlation to time of year, correlation to large scale meteorological and oceanic phenomena, and the average weather patterns (climate) around Australia. By no means is this research to academic standards, simply an independent study.


Time of Year
To see if there were any correlations between these odd tropical cyclones and the time of year, I organized my 18 sample storms by months. It turned out that all of the cases occurred between December and March. Below is the histogram of the distribution.

Right away, it is obvious that these storms tend to occur towards the beginning and end of Australia's tropical cyclone season, amounting to over 61% of the sample. In fact, February, with just 17% of the sample, is often the most active month for the total number of tropical cyclones in the Australian region.
It should be noted that these numbers can be a bit deceiving. For example, 06U and 09U in 2014 are incredible examples of storms that lived over land for a long time, yet both occurred in late January to early February. It seems best to use this analysis to get a general feel for tendencies in the timing of these storms while remembering that other factors clearly have a role, likely at a smaller scale, that can have a stronger influence than the time of year.




Large Scale Phenomenon
For this analysis I used two different two different large-scale weather events: the El Nino/Southern Oscillation (ENSO) and the Madden-Julian Oscillation (MJO). I know from previous experience that both of these have an effect on tropical cyclone activity.

ENSO is a cycle on the time scale of many months to over a year and consists of a shift in sea surface temperatures (SST) across the equatorial pacific. During an El Nino event, the usually cold water off of Peru becomes noticeably warmer, while the waters around Australia and Indonesia become cooler. La Nina events are essentially the opposite with warmer water towards the western side of the pacific. While the changes in SSTs are confined to the equatorial Pacific, the effects ENSO has on weather is worldwide, thus it is a major topic of research right now.

 

The role ENSO has in the strange behavior of tropical cyclones over Australia I believed would have to do with changes in cloudiness brought on by the cycle. The warmer the SSTs are, the warmer the air directly above the ocean becomes. This makes the air more buoyant and it will begin to rise, eventually creating convective clouds (cumulus and cumulonimbus). These are the clouds involved with tropical cyclone formation.

A typical El Nino event where the water off the
South American coast becomes warmer and the
bulk of the deep convection occurs in the central Pacific.

In a typical La Nina event, warm water is pushed westward,
 creating enhanced convective activity over
 equatorial Australia, Papua New Guinea, and Indonesia.

To classify the state of the cycle, NOAA typically uses the SST from a region in the central Pacific known as the Nino 3.4 region. When there are five consecutive 3-month SST means that deviate from the climatological average (anomaly) by at least 0.5°C, an El Nino (+0.5°C) or La Nina (-0.5°C) event is declared.

Locations of the Nino regions, Nino 3.4 includes some of Nino 4 and Nino 3 (NOAA)

 

I took the Nino 3.4 data from the past ten years and plotted it on the same graph as the occurrences of my 18 sample storms. I also plotted horizontal lines at  +0.5°C and -0.5°C for reference. The green line is the SSTs, the black bars show when and how many of the tropical cyclones in the sample occurred. Note that the vertical axis applies for both the tropical cyclone count and the SSTs.

 

 

When tallied up, 6 storms occurred when conditions were neutral, 3 occurred during El Nino conditions, and 9 occurred during La Nina conditions. The strongest remark that can be made with this data is that the tropical cyclones that last a long time over land, such as those in the sample, appear to be quite rare under El Nino conditions. With that being said, how does one explain the three cases during El Nino? It turns out that the very strong El Nino in 2009 and 2010, during which two of the odd cases occurred, actually had a relatively small impact on Australian weather. Thus, only the case in December 2004 is really an outlier. This analysis of ENSO's effect provides a much clearer correlation than the comparison by months, thus I feel this comparison makes a much sounder scientific argument.

 

 

NEXT TIME: I'll wrap-up the large-scale section with an analysis of the MJO.