The current drought, said to be one of the worst in Australian history, has generated much discussion in the context of climate change and the ‘greenhouse effect’. There is general agreement that global warming has occurred, and with it, changes in rainfall patterns. These patterns have varied from one part of the country to another. There appear to have been changes in the major synoptic circulation such that widespread heavy rain events are not now as common as they were in the 1950s to 1990s.

This paper is an attempt to draw together the various changes that have occurred in rainfall across the continent in the last 100 years or so (Click the links to various charts and graphs; they should be easily viewable through any standard Windows graphics viewer).

I have sought to address the following questions:

1. How do the data in each decade measure up against long and short term averages?

2. Have heavy rain events in NSW become less frequent?

3. Have rainfall patterns changed? An amateur’s attempt at statistical breakpoint analysis

4. How did extreme rainfalls in the Sydney basin compare during wet and dry spells?

5. Conclusions, and future rainfall events..will global warming be the main player?

The following maps (courtesy Bureau of Meteorology) show the trends that have occurred at various stages across the country since 1900. As you can see, the long term, apparent upward trend, can be misleading:

The following chart shows rainfall anomalies in all States with respect to the average rainfall for 1900-2005:

while the following two charts show the anomalies across a selection of the Greater Sydney Basin, with respect to both the average for 1900-2005, and the standard 30-year normal (1961-1990):

A few observations from these:

1. Across the country as a whole, rainfall from 1950-2000 was substantially heavier than it was from 1900-1950;
2. Anomalies with regard to the 1900-2005 mean were substantially less in Victoria and South Australia during the early years of last century compared to elsewhere;
3. From 1950 there were 2 particularly wet decades, beginning 1950 and 1970, in the eastern States;
4. Rainfalls have tended to increase in the tropical north and west from 1980 into this century, while suffering a very sharp drop in eastern States since 2000;
5. The drop since 2000 also happened in the early 1900s.
6. With regard to the long term means in the States, were mostly negative before 1950
7. From 1950-2000 they were almost totally positive
8. Greatest positive anomalies occurred between 1950 and 1959
9. Anomalies were smaller from 1960-1999, with a tendency to greater variability in the 1990s
10. Similar comments apply to the anomalies with respect to the 1960-1990 standard normal for capital cities and Sydney’s suburbs. There is substantial negative variation prior to 1950, and since 2000.
11. In regard to both comparative means, Melbourne and Adelaide appear to show a lot less tendency for change. In fact there appears also to be little change from the first half to the second half of the 20th century (respectively dry and wet in other localities).

The following charts show the relative frequency of occurrence of heavy rainfalls at several locations in and around Sydney. The graphs are of the total number of months in each decade in which the total rain received exceeded 350mm, 250mm, 150mm, and 50mm.

Although this is a very limited sample of data, the following is evident:

1. No clear pattern is evident that would suggest that heavy rain events are generally diminishing. However their occurrence does certainly go up and down. The charts suggest weak cycles of fluctuations which themselves vary with location, and do so over a shorter time span.

2. High-end extreme rainfalls have decreased in Sydney, having peaked in the wetter years from 1950. Southwest of the greater Sydney area (eg Moss Vale, near Warragamba Dam), heavier rainfalls were more occurrent in the first half of the 20th century, whereas since 1950 they have dropped off there at the same time greater Sydney was peaking.

3. In the Blue Mountains, variations were more comparable to Sydney but the greatest incidences of extremes occurred in the 1970s there, as against the 1950s in Sydney. Falls of a lower magnitude (eg >50mm) have not decreased as markedly.

4. Synoptic patterns causing the change to higher rainfalls from 1950 across Sydney were limited in their effects, with local orography playing a significant role.

5. There was a greater concentration of rainfall from coastal cyclonic systems.

6. Clearly the drop in extreme rainfall amounts is a significant factor in overall decreases in rainfalls since 2000. A likely major cause is a change in the influence of the Indian Ocean, and/or the frequency of ‘northwest cloud bands’.

7. A preliminary regression analysis of some of the data (not shown in this paper) suggests that there is significant negative trend in the occurrence of rainfalls over 250mm at Robertson and Moss Vale. The data followed a cubic fit. There is a danger in drawing conclusions from the assumption of linearity, as many press articles and scientific climate research papers are inclined to do.

Before dealing with this, I need to show the statistical background to the conclusions reached.

The task, as I saw it, was to examine the historical data for a number of locations to see if there was a point at which the pattern changed from one level to another. This was best done with a breakpoint analysis (sometimes known as a changepoint analysis). A breakpoint analysis does the following:

• It is performed on a series of time ordered data in order to detect whether any changes have occurred.
• It determines the number of changes and estimates the time of each change.
• It provides confidence levels for each change and confidence intervals for the time of each change.
• Sensitivity is improved by ranking the data, which also eliminates the risk of extreme data being excluded

• Formulate the null hypothesis: that the means of the data before and after the ‘change’ are the same

• Test the variances of the data before and after the ‘change’ ..the F-test for the ratio of two variances

• Test the means of the data before and after the ‘change’ ..the 2 t-tests for the difference between two means

• Summarise the results and form a conclusion.

The variances needed to be tested as this would determine the type of t-test to be applied. The tests assume the data are normally distributed and this was verified before testing.

The overall conclusion was that the changes detected by the breakpoint analysis were justified however that analysis appeared to be insensitive to changes over smaller timeframes, such as in the last 10 years or so (see note on the Melbourne results)

The following charts show the results for various locations, preceded by a summary of terms and graphics used in them:

The following was observed:

• Changes in variability in the States were much more evident but were very mixed.

• There have been significant increases in the average rainfall in all States. However most of these were between 1950 and 2000. Of these, the eastern States received the greatest changes, in the 1950s and 1970s.

• In this century, the pattern of increases has shifted to the west and north, while southeastern mainland states in particular have suffered marked declines. Tasmania appears to have fared better probably due to the persistence or intensification of the Southern Ocean westerly wind regime, despite a tendency for the subtropical high pressure belt to be stationed further south of normal in winter. This may be connected to a tightening of the polar jetstream to the south, stated to be due to anomalous ozone at high levels resulting in colder stratospheric temperatures…however there are certain to be other factors.

• In Sydney, changes in average rainfalls were fairly uniform but in the northern suburbs they were more significant in the 1940s than elsewhere.

• Brisbane and Darwin had significant increases in their averages. For Darwin this occurred in 1953 and so was likely due to the same patterns that affected eastern States about that time. This would most likely be sea surface temperatures and El Nino/La Nina.

In the last 105 years there have been 4 major dry spells;1901-1912, 1920-1928, 1935-1948, and 2000-2005. If we take a subset of the 1935-1948 period, the period from 1035-1041 was particularly dry.

The following charts show the movement of the 0.5^th and 99.5^th percentiles of annual rainfalls in the Sydney basin. They show respectively the lowest 0.5% and the highest 0.5% of the falls.

The following points were apparent:

• The 99.5th percentile peaked in the multi-decadal 1949-1999 period, most likely in 1950 when many areas received their highest rainfalls on record. There were of course several other years that came close!

• The period since 2000 is comparable with the lowest 0.5% of readings in the early decades of the last century.

Many attempts have been made to predict the likely outcome of the global warming phenomenon. There are different schools of thought on this but the one that gets the greatest public attention, often through left-wing academics and media, is the doomsday scenario, which comes to us in various degrees. It has not been the purpose of this paper to get into the political side of the debate, but it is my view that the attribution of the warming to human activity has been excessively misapplied. This is because there has not been enough research into solar and other cycles. In addition, it is debatable as to whether the data we have are sufficiently robust to warrant some of the conclusions that are being drawn by climate scientists.

With reference to rainfalls themselves, the following conclusions are apparent:

• The charts suggest that there might be a 100-year cycle of rainfall patterns in the eastern States including their capital cities. However this might be drawing a long bow due to the recent wetter trend in western parts especially the tropics.

• At the beginning of last century, rainfalls were quite low, lasting until 1950

• The cycle appears to have recommenced, with the rainfall this year comparable to 100 years ago. Should it repeat itself, the next 50 years should see a gradual increase in rain events over eastern Australia.

• The ‘greenhouse effect’ may cause the pattern to evolve differently to what it did 100 years ago. However in view of our inadequate understanding of the role of natural cycles in climate change, this would be quite speculative.

• Solar cycles, the pattern of El Nino and the Southern Oscillation Index, and statistical probability, suggest we may be in for a major reversal of the latest dry spell especially as history shows that rainfall has increased. The effect of global warming on the variation and dispersal of rainfall patterns seems likely to remain conjectural until we understand more of the influence on them of solar and natural cycles. On balance, wetter years lie not too far ahead.