When Is A Heatwave Not A Heatwave?
When the Bureau of Meteorology defines it out of existence.
In his reply to me on behalf of Dr Vertessy, Bob Baldwin wrote:
“The Bureau has adopted a particular operational heatwave definition motivated by human health considerations, defined as a period of at least three days where the combined effect of high temperatures and excess heat is unusual within the local climate. ……….The bulk of heatwaves at each location are low intensity with local communities expected to have adequate adaptation strategies for this level of thermal stress. Less frequent, higher intensity heatwaves are classified as severe and will challenge some adaptation strategies, especially for vulnerable sectors such as the aged or the chronically ill.”
After some digging, I found this paper which describes the Bureau’s methodology used in their Pilot Heatwave Forecast:
The Excess Heat Factor: A Metric for Heatwave Intensity and Its Use in Classifying Heatwave Severity, John R. Nairn and Robert J. B. Fawcett (2015)
The method is quite easy to follow and implement, and I was able to replicate results for the 2014 Melbourne heatwave exactly and use it successfully for other single locations. It is designed for use with AWAP gridded data of course to give forecast maps. Note this is raw data, not homogenised. I downloaded all data from Climate Data Online.
There are several steps. Readers should read the paper for full details. Briefly, using a daily mean temperature calculated by averaging the day’s maximum and the following night’s minimum, three-day means are calculated. These are then compared by subtracting the previous 30 days’ daily means (as people acclimatise to changed temperatures in this period). Differences that exceed the 95th percentile of all three-day means from 1971 to 2000 are multiplied by the three-day mean to give the Excess Heat Factor, which indicates heatwave. This is then compared with the 85th percentile of all positive EHFs from 1958 to 2011 to give a severity index, and if it exceeds 3 times the 85th percentile this becomes an extreme heatwave event.
From the paper:
“The intent of these definitions is to create a heatwave intensity index and classification scheme which is relative to the local climate. Such an approach is clearly necessary given the abundant evidence that people and supporting infrastructure are largely adapted to the local climate, in physiology, culture and engineered supporting infrastructure.”
Here are the results for Melbourne- with all its UHI effect of course.
Fig. 1: Decadal (running 3653 day) count of positive Excess Heat Factor (heatwave) days in Melbourne
Fig.2: Decadal count of Severe Heatwave Days
Fig.3: Decadal Count of Extreme Heatwave Days
Notice how Melbourne heatwaves of all types have been increasing and extreme events are currently at the highest level “ever”.
How does this apply to various other Australian locations? I decided to check with the extremes- the hottest and the coldest Australian locations, Marble Bar in the north west of W.A. and Mawson Base in Australia’s Antarctic Territory.
The old Marble Bar station closed in 2006. I have concatenated the old Marble Bar data with the new, from 2003. This makes very little difference to the calculations but extends the record to the present.
Fig. 5: As for Melbourne, decadal count of heatwave days
Fig. 6: Severe heatwaves
It is clear that local climate does make a big difference to heatwaves by this definition. In fact, Melbourne has more extreme heatwave days than Marble Bar!
How does this method of detecting and measuring heatwaves deal with Marble Bar’s record heatwave of 1923-24?
According to the Australian Government’s website, Disaster Resilience Education for Schools at
“Marble Bar in Western Australia holds the record for the longest number of hot days in a row: the temperature was above 37.8°C for 160 days in 1923-24.”
I count 158 days consecutively from daily data at Climate Data Online. The total for the 1923-24 summer from 13 October to 19 April was 174 days. That is indeed a long period of very hot weather.
Surprisingly, the BOM does not class that as a long or extreme heatwave. Apparently, according to this metric, there were only four short heatwaves, one of them severe, and none extreme. For the entire period, there was only one severe heatwave day – 3 February.
Fig. 8: Marble Bar 1923-24 summer. I have marked in the old “ton”, 100 F, or 37.8C. Squint hard to see the “severe’ heatwave around 3 February, but the heatwave around 22 February is invisible to the naked eye.
Yes, the old timers at Marble Bar were pretty tough and would be used to hot conditions. But not to recognise this old record heatwave when every day in over five months was considerably above body temperature is laughable.
For comparison, Figure 9 shows 182 day counts of days that were over 100 degrees Fahrenheit, or 37.8 degrees Celsius. (The old record finishes in 2006.)
Fig. 9: Running 182 day counts of days over 100 F. 1923-24 is circled.
Note there were two other years when there were more than 170 days over 100F.
Figure 10 is from Figure 16 in the Nairn and Fawcett paper, and is a map of the level of Excess Heat Factor across Australia during the heatwave of January-February 2009.
Fig. 10: Figure 16 from Nairn and Fawcett (2014)- Excess Heat levels across Australia 21 January – 11 February 2009.
The area around Marble Bar has a level of between 0 and 10. My calculations show this is correct- EHF reached 0.08 on 23 January- a mild heatwave. Readers may be interested to know that maximum temperature was above 40 degrees Celsius from 1 January to 24 January, and minima were not below 24.3.
The authors, and their employer, the Bureau, are in effect telling Marble Bar locals their heatwaves don’t rate because they’re used to the heat.
Now I shall turn to the other extreme- Mawson.
Firstly, plots of the range of minima for each day of the year:
Fig. 11: Scatterplot of minima for each day of the year at Mawson Base
Fig. 12: maxima:
Fig. 13: Decadal count (running 3653 day count) of days with positive Excess Heat Factor, i.e., by definition, heatwave days
Fig. 14: Decadal count of days in severe heatwave:
Fig. 15: Decadal count of days in Extreme heatwave:
Apparently, Antarctica gets more extreme heatwave days than Melbourne, or Marble Bar!
Of course, critics will say this metric was never intended for use in Antarctica, and I agree: no one would seriously claim there are heatwaves there. However, if heatwaves are to be defined as “a period of at least three days where the combined effect of high temperatures and excess heat is unusual within the local climate”, and NOT by comparison with any absolute threshold, then this analysis of its use there is valid. “High” temperature by this definition is relative to the local climate, wherever “local” is. If this metric fails in Antarctica, it fails everywhere.
The Bureau of Meteorology’s metric for heatwaves is a joke. It may accurately detect heatwaves in the southern fringe of Australia, and a further use may be to support Dr Vertessy’s outlandish claims. However, it fails to cope with different climates, particularly extremes. A methodology that fails to detect heatwaves at Marble Bar, and creates them in Antarctica, is worse than useless- it is dangerous.