In terms of rainfall alone, Manilla, NSW, had droughts between 1900 and 1950 that were more severe, and lasted very much longer than those of recent years.
It is hard to say how bad one drought is compared to another because some droughts last longer than others. A drought that lasts two months, and has only 10 mm of rain when it would normally have 100 mm, qualifies as “extreme”. In such a very short drought, rainfall as low as 10% of normal just qualifies as extreme. For a drought lasting twelve months, when there is normally 652 mm of rainfall at Manilla, there has never been a case of a twelve-month rainfall as low as 10% of that (65 mm). (The lowest ever was 288 mm, in 1964-65.) Clearly, using 10% of normal rainfall will not do to define longer-term droughts.
I find the severity of each drought, whether it is long or short, by its percentile rank.
The Bureau of Meteorology defines “Rainfall Deficiency” as:
Lowest on record – lowest since at least 1900 when the data analysed begin.
Severe deficiency – rainfalls in the lowest 5% of historical totals.
Serious deficiency – rainfalls in the lowest 10% of historical totals, but not in the lowest 5%.
On the graph, I use this code:
Extreme rainfall shortage: rainfall in the 1st percentile only.
Severe rainfall shortage: rainfall in the 2nd to 4th percentiles.
Serious rainfall shortage: rainfall in the 5th to 9th percentiles.
All of Manilla’s extreme rainfall droughts that lasted for six years or more happened in the first half of the 20th century. Extreme droughts lasting for thirty years ended during 1940, 1941 and 1947.
Since 1950, the longest extreme drought lasted only five years, ending in 1961. The next longest lasted three years, ending in 1968. The last forty-four years have brought only six extreme droughts, all of less than two years duration: 1971, 1974, 1982, 1984 (2 months!), 1994 and 2002. The twelve years since 2002 may be the longest period without an extreme drought in the whole record since 1883.
Extreme droughts had also been few and short in the earliest years, from 1883 to 1902.
Similar, but much improved graphs
[This graph gives the misleading impression that the longer the duration of rainfall shortage, the later it occurs. That is an artefact.
A shortage of a given duration observed in a particular month must have commenced earlier: earlier by the number of months of its duration, less one.
Graphs that remove this defect, and plot correctly the dates of onset, persistence, and breaking of rainfall shortages are “Rainfall Shortage History: Manilla” and “Rainfall Shortage Jan 2000 – Mar 2019”.
At Lake George, in the southern highlands of NSW, extreme droughts of long duration were similarly restricted to the first half of the 20th century, as shown by rainfall records and lake levels.
The “Millennial Drought of southeastern Australia” was not a drought of long duration at either Manilla or Lake George.
[Note added May 2016
The post “Manilla’s Droughts, 1884 to 1916” has an enlarged graph showing more detail. Of the catastrophic droughts of 1902 (“Federation Drought”) and 1912-16, the first had a sudden termination but the second had a sudden onset.]
This blog has many more posts about drought (or rainfall deficiency) at Manilla, NSW. Please use the “Search” function at the top right.
Monthly rainfall totals (mm) are from Manilla Post Office (Station Number 055031) courtesy of the postmaster, Phil Pinch.
I made a table (in Excel) with the 1570 months of record in a single column on the left. In the second column I put the rainfall for that month. In the third column was the sum of the rainfall for that month and the previous month. In the fourth column was the sum of the rainfalls for that month and the two previous months. Following columns had sums for more months. To limit the size of the table, not all possible sums were listed, but only sums in a logarithmic sequence. The following monthly sums were included:
1, 2, 3, 4, 5, 6, 9, 12, 15, 18, 24, 30, 36, 48, 60, 72, 96, 120, 144, 180, 240, 360.
Some of these are whole numbers of years (not all of them calendar years) up to 30 years (360 months).
I made a second table, replacing the rainfall totals by percentile ranks. I selected a normal period for the percentiles that is less than the whole record. The earliest years could not be used because there were no values for sums of large numbers of earlier months. The latest years were not included, to avoid the need for recalculation. The chosen standard was the years from 1901 to 2000.
From the percentile ranks, only those less than the 10th percentile were used in this analysis of rainfall droughts, sorted into the three classes “Extreme”, Severe” and Serious”.