From Watts Up With That?

Geoff Sherrington.

There were 10 significant weather station/city sites explored for their heatwave properties in the first article of this series, 5 days ago.

Australian Heatwave Stories Cop Severe Criticism – Watts Up With That?

That first article shows 160 graphs of hottest heatwave temperatures over the years when records have been kept, for Adelaide, Alice Springs,  Brisbane, Cape Leeuwin, Darwin, Hobart, Longreach, Melbourne, Sydney and Perth. This second article deals with some patterns of interest.

I wrote the articles because many Establishment sources make claims like this one from the Climate Council of Australia, 2014:

“Climate change is already increasing the intensity and frequency of heatwaves in Australia. Heatwaves are becoming hotter, lasting longer and occurring more often.”

HEATWAVES: HOTTER, LONGER, MORE OFTEN

An Internet search using “heatwaves longer hotter more often” returns these 6 hits and more.

Climate change study: Australia is in the crucible of slower, longer heatwaves | SBS News

Heatwaves: hotter, longer, more often – Macquarie University

Heatwaves: hotter, longer, more often – Environmental Health Australia (Western Australia) Inc

Australian heatwaves more frequent, hotter and longer: Climate Council report – ABC News

nationalacademies.org/based-on-science/global-warming-makes-heat-waves-hotter-longer-and-more-common

Heatwaves to be hotter, longer and more frequent, climate change report says – ABC News

HOTTER?

Using practically all years of data recorded, I examined each of 4 heatwave durations of 1, 3, 5 and 10 days. The raw data are from the Climate Data Online CDO source by the Bureau of Meteorology.

The first half of the heatwave numbers is compared to the last half. The first half numbers are subtracted from the second half numbers. If there is warming, the difference is positive. If there is cooling, the difference is negative.

This example summarises the method by a graph.

The first half average equals the second half average, so in this simple case no 5-day heatwave warming is shown for Adelaide.

These results summarise all 40 cases.

ADELAIDE							DIFF LAST MINUS FIRST
FIRST HALF 1-DAY		41.42	SECOND HALF 1-DAY		41.68	0.26
FIRST HALF 3-DAY		39.14	SECOND HALF 3-DAY		39.19	0.06
FIRST HALF 5-DAY		37.43	SECOND HALF 5-DAY		37.37	-0.07
FIRST HALF 10-DAY		34.57	SECOND HALF 10-DAY		34.54	-0.03
ALICE SPRINGS							DIFF LAST MINUS FIRST
FIRST HALF 1-DAY		43.08	SECOND HALF 1-DAY		42.94	-0.15
FIRST HALF 3-DAY		41.16	SECOND HALF 3-DAY		41.37	0.21
FIRST HALF 5-DAY		40.22	SECOND HALF 5-DAY		40.54	0.32
FIRST HALF 10-DAY		39.90	SECOND HALF 10-DAY		40.27	0.37
BRISBANE							DIFF LAST MINUS FIRST
FIRST HALF 1-DAY		37.45	SECOND HALF 1-DAY		36.04	-1.41
FIRST HALF 3-DAY		34.08	SECOND HALF 3-DAY		32.77	-1.32
FIRST HALF 5-DAY		33.78	SECOND HALF 5-DAY		32.52	-1.26
FIRST HALF 10-DAY		32.42	SECOND HALF 10-DAY		31.43	-1.00
C_LEEUWIN							DIFF LAST MINUS FIRST
FIRST HALF 1-DAY		33.88	SECOND HALF 1-DAY		34.91	1.03
FIRST HALF 3-DAY		28.92	SECOND HALF 3-DAY		30.07	1.15
FIRST HALF 5-DAY		27.06	SECOND HALF 5-DAY		27.92	0.86
FIRST HALF 10-DAY		25.32	SECOND HALF 10-DAY		25.99	0.67
DARWIN							DIFF LAST MINUS FIRST
FIRST HALF 1-DAY		36.95	SECOND HALF 1-DAY		36.67	-0.28
FIRST HALF 3-DAY		35.97	SECOND HALF 3-DAY		35.64	-0.33
FIRST HALF 5-DAY		36.94	SECOND HALF 5-DAY		36.67	-0.27
FIRST HALF 10-DAY		35.12	SECOND HALF 10-DAY		34.66	-0.46
HOBART							DIFF LAST MINUS FIRST
FIRST HALF 1-DAY		34.86	SECOND HALF 1-DAY		36.49	1.62
FIRST HALF 3-DAY		29.68	SECOND HALF 3-DAY		30.62	0.94
FIRST HALF 5-DAY		27.15	SECOND HALF 5-DAY		28.04	0.89
FIRST HALF 10-DAY		24.86	SECOND HALF 10-DAY		25.86	1.00
LONGREACH							DIFF LAST MINUS FIRST
FIRST HALF 1-DAY		44.12	SECOND HALF 1-DAY		43.94	-0.18
FIRST HALF 3-DAY		43.18	SECOND HALF 3-DAY		42.85	-0.33
FIRST HALF 5-DAY		42.44	SECOND HALF 5-DAY		42.17	-0.27
FIRST HALF 10-DAY		41.33	SECOND HALF 10-DAY		41.13	-0.20
MELBOURNE							DIFF LAST MINUS FIRST
FIRST HALF 1-DAY		40.55	SECOND HALF 1-DAY		40.46	-0.10
FIRST HALF 3-DAY		36.58	SECOND HALF 3-DAY		36.93	0.35
FIRST HALF 5-DAY		34.00	SECOND HALF 5-DAY		34.18	0.17
FIRST HALF 10-DAY		30.69	SECOND HALF 10-DAY		30.99	0.30
PERTH							DIFF LAST MINUS FIRST
FIRST HALF 1-DAY		40.58	SECOND HALF 1-DAY		42.40	1.82
FIRST HALF 3-DAY		37.90	SECOND HALF 3-DAY		40.17	2.27
FIRST HALF 5-DAY		36.38	SECOND HALF 5-DAY		38.65	2.28
FIRST HALF 10-DAY		34.01	SECOND HALF 10-DAY		36.41	2.40
SYDNEY							DIFF LAST MINUS FIRST
FIRST HALF 1-DAY		38.02	SECOND HALF 1-DAY		38.62	0.60
FIRST HALF 3-DAY		32.68	SECOND HALF 3-DAY		33.32	0.64
FIRST HALF 5-DAY		30.72	SECOND HALF 5-DAY		31.35	0.63
FIRST HALF 10-DAY		28.78	SECOND HALF 10-DAY		29.53	0.75
					GRAND AV	0.35
					POSITIVES	24.00
					NEGATIVES	16.00

The naïve interpretation is that overall, if there is evidence of heatwaves getting hotter, it is small. 40% of cases show cooling, 60 % show warming. The overall warming is about 0.35  ͦC, which is of similar size to the uncertainty of the numbers.

In scientific experiments, it is often the case that a simple first look will either encourage or discourage further investigation. Here, there is little enthusiasm to look further. The pattern is not indicative of future health problems, for example.

Those familiar with time series numbers such as these will question the data quality. As I noted in my first article,

NOTE 1: Many of these stations underwent a site shift, many in the middle of the study period, typically from a Post Office to an Airport. This has affected the distribution over time of these hottest heatwaves. There is no reliable method to adjust this by a single step change because the magnitude of such a step is not now knowable.

A conclusion is that those who seek to show heatwaves getting hotter need to ask how to combat these simple contrary results without the use of subjective adjustments that typically lead to disagreement about methods, motives and manipulations.

Why not accept that there is next to no simple evidence of Australian heatwaves getting hotter and leave it at that?

LONGER?

I took the Top 40 heatwave numbers for all 10 cities, CDO not ACORN-SAT, to examine 1-Day, 3-Day, 5-Day and 10-Day heatwave patterns.

In theory, if the longer 10-Day heatwave types were increasing compared to 1-Day or 3-Day, then counts of various heatwave numbers should show this. I also examined the number of days in the first half of the station history versus the last half of the total years. Adelaide, the first station examined in this way, met the change point in 1956.

	HEATWAVE		1ST PART	2ND PART
	DAY-LENGTH		# of DAYS	# of DAYS
ADELAIDE	1		19	21
	3		20	20
	5		20	20
	10		20	20

This example shows that first, the number of Top 40 heatwaves in the first part of the observations is about the same as in the second half. Second, the table shows similar counts for heatwaves of all durations, so the longer 10-Day types were not increasing compared to 1-Day etc, so the inference is that these Adelaide heatwaves were not becoming longer. (The Top 40 selection is used because if all years were examined, the split would always be equal and could not show anything of interest.)

Not all stations trend the same as Adelaide.

PERTH

HEATWAVE			1ST PART		2ND PART
DAY-LENGTH			# of DAYS		# of DAYS
1		8		32
3		5		35
5		6		34
10		5		35

Perth indicates that while the “hotter” heatwaves were more abundant in later years, the share of longer heatwaves remains much the same as shorter ones.

Here is the complete table of all ten stations. I have added the average temperature of each subset, which is a check on paperwork but does not add to understanding because the differences are so small.

	HEATWAVE	1ST PART	2ND PART	Average T	Average T
	DAY-LENGTH	# of WAVES	# of WAVES	First Part	Second Part	HALF WAY YEAR
ADELAIDE	1	19	21	43.37	43.73	1956
	3	20	20	41.61	41.69
	5	20	20	40.4	39.97
	10	20	20	37.24	37.05
ALICE SPRINGS	1	26	14	44.99	44.61	1952
	3	23	17	43.08	42.96
	5	22	18	42.28	42.07
	10	21	19	41.95	41.85
BRISBANE	1	24	16	39.79	39.4	1956
	3	30	10	35.47	35.1
	5	31	9	35.04	34.88
	10	30	10	33.48	33.28
C LEEUWIN	1	17	23	37.3	37.43	1966
	3	13	27	32.03	32.1
	5	14	26	29.35	29.42
	10	15	25	27.05	26.98
DARWIN	1	27	13	37.97	37.77	1955
	3	27	13	36.86	36.57
	5	26	14	37.98	37.72
	10	31	9	35.83	35.55
HOBART	1	15	25	38.17	38.74	1971
	3	20	20	32.78	33.66
	5	14	26	29.72	29.44
	10	17	23	27.87	27.39
LONGREACH	1	24	16	45.74	45.82	1961
	3	26	14	44.77	44.54
	5	24	16	44.1	43.83
	10	21	19	42.81	42.64
MELBOURNE	1	22	19	43.4	43.51	1941
	3	19	21	39.84	39.78
	5	21	19	37.6	37.23
	10	19	21	33.18	33.6
PERTH	1	8	32	43.08	43.57	1961
	3	5	35	41.49	41.39
	5	6	34	39.79	39.94
	10	5	35	37.81	37.39
SYDNEY	1	16	24	41.21	41.53	1942
	3	14	28	35.87	35.65
	5	13	27	33.48	33.3
	10	12	28	30.89	31.01
AVERAGE		19.425	20.65

The conclusion is that by this method of analysis, there is no indication that heatwaves are becoming longer.

NOTE: This analysis used essentially all of the raw time and temperature data available. Different conclusions are possible if some adjustments are used. For example, if only data from the last 40 years is used, there might be an indication of heatwaves becoming longer.  (I have not tested this).

Alternatively, a different conclusion might be reached if adjusted ACORN-SAT temperatures were used. However, these approaches open the gate to cherry picking, selective argument and subjective changes that have to date been a major cause of disagreement between groups of investigators.

MORE FREQUENT?

The data base that I have constructed produces the year in which a putative heatwave happened and the highest average maximum temperature it reached over 1, 3, 5 and 10 consecutive days. It does not sort the number or intensity of more than one heatwave each year, intense enough to be counted among competitors from other years.

Software to detect more than one heatwave a year is complex and it is dependent on the definition of a heatwave. Different countries use different definitions of “heatwave”. Some definitions are precise enough to allow computing, others are vague and can be flexibly adjusted to suit an outcome. Also, questions arise, for example, should one infill a prospective significant 10-day heatwave that has a missing value in the centre? Infill with what temperature? And, when is a 10-Day heatwave better treated as two adjacent 5-Day heatwaves?

It is possible to crawl manually through the 50,000 or so days for each of these 10 stations to detect the stronger heatwaves apart from the primary one that is there because others in the same year are excluded by the “remove duplicates” function on Excel applied to all years examined in a sorting from highest heatwave temperature to lowest. I have examined the tedium required and rejected that option.

Maybe Artificial Intelligence can provide a workable solution. Over to you, readers.

However, I know of no literature that counts the number of heatwaves over time in a manner that concludes that they are becoming more frequent. Links welcomed from readers.

DISCUSSION:

To the extent that these 10 weather station localities can represent Australian climate features, there is little to no evidence in this analysis with its limitations, that heatwaves are becoming hotter or longer with more analysis required about “more frequent”.

There is little comparable scientific literature in conflict with this conclusion. Many cited studies commence after 1950; many invoke convoluted definitions of heatwave; and many seem affected by a wrong piece of logic.

The wrong logic. There is general scientific acceptance that global temperatures have increased by 1 to 1.5  ͦ C in the last 120 years or so. Many publications contain the assertion or assumption that this temperature increase has also underlined the heatwave maximum temperature pattern, expecting it to show a similar increase. It does not, in the data studied here. The conduct of heatwaves differs from the conduct of general temperature observations. In the case of these 10 Australian stations, there is essentially no change over time demonstrated in the maximum average heatwave temperatures reached.

Remember that sea surface temperatures seldom exceed 30  ͦC . There is a mechanism of convective instability that creates an upper limit. While no mechanism appears to have been advanced for an upper limit on heatwave temperatures over land, it should not be discounted from future research.

Further, there is a widespread theme in the literature that the heatwave is an event related to the location at which it happens. The heatwave patterns of coastal Adelaide and Melbourne are strongly influenced by Hadley cells that cause hot winds from central Australia to move to the South and South-East towards the coastal cities, sometimes elevating local average temperatures by up to 15  ͦ C. How hot it gets is dictated by how hot the centre of Australia is at the time, how close to optimum is the direction of travel of hot air from the centre to the coast, how intense is the weather system that created a fast, low heat loss transport from the hot centre to the coast.

SUMMARY.

Simple raw temperature data analysis does not support the widespread global or national claim that heatwaves are becoming hotter, longer and more frequent.

Because simple analysis does not support that claim, those who continue the claim must rely on non-simple analysis. All studies about Australia use the same data set as here, so arrival at different conclusions must depend on subjective assumptions and on them being acceptable.

Acceptable to whom?

END.