“Seasonal inversions” in Victorian climate history

Two years ago, it occurred quickly to me that an extremely unusual thing was happening in Melbourne. For 2023, October was cooler than September — and by mean maximum temperature, substantially so. October 2023 average 19.2˚C, and September 2023 averaged 20.1˚C — a figure of course inflated by the greenhouse gas emissions of Australia and the Gulf States, but this should not alter the abnormality being described.

The same thing happened in 2024, when Melbourne’s maximum temperature for August was 0.5˚C hotter than for September.

Studying climate history, I have always noticed that such abnormalities, whilst unusual, are not unprecedented. Although the Bureau of Meteorology says that:

“Temperature data prior to 1910 should be used with extreme caution as many stations, prior to that date, were exposed in non-standard shelters, some of which give readings which are several degrees warmer or cooler than those measured according to post-1910 standards.”

I have known of a number of cases since the earliest temperature records were taken from Melbourne in 1856. If the note of the Bureau is correct, the fact that temperature records before 1910 “should be used with extreme caution” would not invalidate these “seasonal inversions” at least if they be observed at every available station in an area.

For this post, I intend to tabulate cases where any of the following occurred over Victoria as a whole since 1910:

1. April hotter than March
2. May hotter than April
3. June hotter than May
4. September cooler than August
5. October cooler than September
6. November cooler than October

Seasonal inversions can occur in any one or more of maximum, minimum or mean temperature. Logically, the last of these is the most accurate. However, over the years my primary focus has been on memorising and collecting mean maximum temperatures, and I have known of inversions in minimum temperatures that did not extend to maximum or mean temperatures.

Months Since 1910 with Victoria-Wide Seasonal T_mean Inversion:

		Tmax	Tmean	Tmin
1928	September	18.56	12.50	6.44
	October	17.83	12.07	6.29
1940	October	22.13	15.08	8.03
	November	21.49	14.82	8.22
1957	May	16.03	10.27	4.59
	June	16.64	11.50	6.37
1962	May	14.66	9.97	5.36
	June	14.46	10.26	6.05
1969	August	14.59	9.49	4.28
	September	13.41	8.87	4.31
1977	August	16.18	10.51	4.73
	September	15.72	9.99	4.25
2001	September	18.42	12.97	7.51
	October	17.94	12.68	7.41

Months Since 1910 With Victoria-Wide Seasonal T_max Inversion Only

		Tmax	Tmean	Tmin
1982	August	17.18	10.73	4.17
	September	16.77	10.78	4.76
2015	October	25.29	17.64	9.99
	November	24.62	17.72	10.92

Months Since 1910 With Victoria-Wide Seasonal T_min Inversion Only

		Tmax	Tmean	Tmin
1911	April	18.20	12.08	5.95
	May	15.62	11.05	6.56
1912	May	16.57	10.06	3.63
	June	12.86	8.70	4.52
1913	October	20.36	14.12	7.88
	November	21.18	14.30	7.51
1920	May	14.49	9.18	3.95
	June	12.73	8.85	4.97
1923	October	18.64	13.04	7.42
	November	21.26	14.18	7.18
1931	April	18.22	12.90	7.57
	May	15.26	11.39	7.62
1932	September	16.62	11.54	6.46
	October	17.59	11.68	5.76
1936	August	14.03	9.58	5.02
	September	15.35	9.70	4.03
1942	April	19.32	13.81	8.29
	May	15.30	11.78	8.36
1945	May	15.64	10.23	4.90
	June	14.02	9.70	5.37
1945	August	13.89	9.67	5.35
	September	15.68	10.28	4.86
1948	March	22.78	15.34	8.09
	April	19.00	13.56	8.11
1954	March	23.56	15.91	8.43
	April	20.40	14.65	8.90
1955	October	19.01	13.49	7.96
	November	20.40	13.92	7.53
1958	April	21.60	14.54	7.48
	May	16.63	12.55	8.57
1958	August	12.75	9.07	5.27
	September	14.83	9.67	4.49
1961	May	16.18	10.29	4.49
	June	14.11	9.45	4.79
1963	April	19.56	13.27	6.96
	May	15.19	11.58	8.07
1963	October	21.56	15.38	9.19
	November	22.83	15.87	8.98
1991	May	16.79	11.29	5.89
	June	14.62	11.14	7.65
1992	October	19.02	13.83	8.65
	November	19.64	14.04	8.53
1995	April	17.69	12.32	6.93
	May	15.20	11.13	7.15
1999	October	20.53	14.66	8.78
	November	21.37	14.74	8.20
2004	May	15.93	10.61	5.37
	June	13.23	9.43	5.62
2008	May	16.30	11.25	6.29
	June	14.74	10.58	6.42
2013	September	18.93	13.52	8.11
	October	20.00	13.70	7.39
2023	May	15.39	10.57	5.84
	June	13.69	9.93	6.16

It is obvious that T_min seasonal inversions (32 Victoria-wide since 1910) are significantly more common than T_mean (seven) or T_max (eight). There have been on average four T_min inversions every fifteen years, but T_max and T_mean inversions occur only on average every fifteen to sixteen years. For this reason, when using Melbourne data for likely inversions predating 1910, I will consider only T_mean and T_max.

All Months with Seasonal T_mean and/or T_max Inversion in Melbourne:

Year	Months	Tmax	Tmean	Δmax	Δmean	Notes
1864	September	18.48	13.69	-0.69	-0.24
	October	17.79	13.45
1866	October	20.48	14.94	-0.16	-0.48	Minimum temperature in November 1866 cooler than for May!
	November	20.32	14.46
1873	October	21.31	15.41	-1.48	-0.84	Extremely large seasonal inversion. Data inadequate to demonstrate accuracy, but unlikely that so large an inversion — fifth-largest on record in Melbourne — could be due to different shelters.
	November	19.83	14.57
1883	May	16.20	11.55	0.62	-0.64	Inversion replicated in Hobart data — it is the first full year of temperature recording there.
	June	15.58	12.19
1886	September	18.53	13.07	-1.07	-0.49	Unusual La Niña seasonal inversion due to major Tasman Sea block that saw Brisbane and Cairns have wettest September on record. Melbourne had hottest September until 1919, and inversion strongly replicated in Hobart data.
	October	17.46	12.58
1888	March	20.67	15.79	-0.47	0.97
	April	21.14	14.82
1899	September	17.76	13.13	0.25	-0.07
	October	18.01	13.06
1904	March	20.12	15.81	-1.67	-0.34	Coolest March on record — after cool pluvial summer that saw Wilfrid Rhodes, utterly hopeless on his other Australian tours, manage 31 Test wickets — followed by extremely dry April. Hobart had just two raindays all April. Seasonal inversion replicated by Hobart and Adelaide data.
	April	21.79	16.15
1905	August	14.59	10.35	-1.06	-0.78	Extraordinary cool spring spell. Both September and October easily coolest on record, and the last fortnight of September did not reach a maximum of 14˚C! Inversion replicated in Hobart data.
	September	13.53	9.57
1919	March	20.67	15.79	-0.47	0.97	Second-wettest March in Melbourne followed by dry April with springlike flowers
	April	21.14	14.82
1928	September	19.50	14.50	-0.94	-0.87	Major seasonal inversion replicated in all-Victoria data. Hottest September in Melbourne until 2013, with first-ever 30˚C day in that month, and then extremely wet October in exposed Upper Northeast and West Coast.
	October	18.56	13.63
1936	August	15.33	11.17	0.50	-0.20	Extremely wet winter followed by extremely cool anticyclonic September. Mean minimum for September was 0.01˚C cooler than July [September 6.13˚C; July 6.14˚C] and second-coolest on record [September 1949 was 6.136667˚C against September 1936’s 6.13333˚C and September 1905’s 5.57˚C].
	September	15.83	10.97
1940	October	23.32	16.36	-2.55	-0.77	Hottest October mean maximum in Melbourne until 2015, followed by cool, rainy November in southern Victoria and Wimmera. Largest seasonal inversion in Melbourne on record by Tmax, although weaker than September-October 1928 over Victoria as a whole.
	November	20.77	15.59
1948	September	18.75	13.13	-0.29	0.15
	October	18.46	13.28
1957	May	16.77	12.39	-0.94	-1.17	Largest seasonal inversion on record by mean temperature in both Melbourne and all-Victoria data. First half of June saw all records for heat in that month broken
	June	17.71	13.56
1960	October	20.40	15.05	-0.43	-0.46	Despite exceptional rains in November — Adelaide had 75 millimetres in one day — inversion not replicated across Victoria.
	November	19.97	14.59
1962	May	15.77	11.99	-0.15	-0.12
	June	15.92	12.11
1963	October	21.81	17.01	0.21	-0.50	Hottest October in Melbourne by Tmean until 2015, and hottest in Hobart to date. This is the only 1960s October-November inversion replicated there, but very strongly.
	November	22.02	16.51
1965	October	21.35	16.23	-0.22	-0.16
	November	21.13	16.07
1969	August	15.58	11.89	-1.31	-0.90	Major, widespread inversion throughout southeastern Australia. In Adelaide, the mean minimum for September was more than 1˚C cooler than for any of the “winter” months! September’s mean maximum in Melbourne was 0.88˚C cooler than July’s, although this was not replicated throughout Victoria.
	September	14.27	10.99
1973	October	20.27	15.74	-0.34	0.28	Melbourne did not reach 28˚C in all of November, although inversion not replicated for the Victoria-wide data.
	November	19.93	16.02
1977	August	16.45	12.48	-1.08	-0.99	Hottest August mean maximum in Melbourne until 1982, and driest since 1944. September was extremely cool in first half, but varied from record dry in South Gippsland and northern Tasmania to extremely wet in East Gippsland.
	September	15.37	11.49
1982	August	17.73	12.82	-0.72	-0.66	Hottest, and third-driest, August across Victoria as a whole. Hottest in Melbourne until 2024.
	September	17.01	12.16
1984	August	15.42	12.06	0.14	-0.13	Unlike 1977 and 1982, this was a very wet August, but minima were extremely high due to cloudiness. September was also very wet but long periods late in the month were dominated by an anticyclone to the west of Victoria.
	September	15.56	11.93
1993	August	17.38	12.79	-0.05	0.73
	September	17.33	13.52
1995	August	16.99	12.25	-0.11	-0.16	Famous hot and dry August across southern mainland Australia — Sydney went 46 days without rain and Melbourne had more afternoons above 20˚C than in August 1982.
	September	16.88	12.09
2005	March	23.91	18.81	-0.18	0.01	Hottest April and driest autumn on record throughout Victoria.
	April	24.09	18.80
2013	September	20.06	15.62	-0.22	-0.15	Hottest September on record in Victoria.
	October	19.84	15.47
2015	October	24.32	18.34	-1.60	-0.49	Record hot and dry October across Victoria and especially western Tasmania. In normally wet Waratah there was only one rainday and it was the second-driest month on record after March 1893.
	November	22.72	17.85
2023	September	20.11	14.88	-0.86	-0.04	Record hot and dry September followed by very wet October across eastern Victoria where bushfires turned into floods in a few hours. October was though even drier and relatively hotter than September in more westerly parts of Australia.
	October	19.25	14.84
2024	August	18.25	13.83	-0.48	0.11
	September	17.77	13.94

The figures from Melbourne indicate highly probable widespread seasonal inversions before 1910 in:

1. October-November 1873
• only because the magnitude of the inversion in Melbourne is so large that it could hardly have been not replicated more widely
2. September-October 1886
3. March-April 1904
4. August-September 1905
• newspapers from this period do provide conclusive evidence that the spring of 1905 was uniquely cool even for a decade that is probably the coolest globally since the last glacial.

In contrast to the spring of 1905, documentary evidence for the previous three cases was not found when I checked Trove. Even the 1905 newspapers do not note that figures indicate September was cooler than August.