Feeling the heat

We are currently experiencing a disgusting spell of hot weather.  I am reminded of a couple of lines from the poem "Integrated Adjective" by John O'Grady:

".... drinkin' beer and arguin' an' talkin' of the heat, 

An' boggin' in the bitumen in the middle of the street; "

This post started off "talking'of the heat,"and in particular the heat of December but then wandered off into general issues of maximum and minimum temperatures.  I have discovered that there appear to be a few anomalies in some of the temperatures recorded for Carwoola between 2000 and 2006(ish) but they appear to vary somewhat according to the variable being examined. 

My conclusions thus far are that

• there seems to have been a change of an unknown nature to site conditions in the early noughties;
• in most cases, more recent data appears to correlate well with data for Canberra airport;
• for the principal use to which the Carwoola series is applied - general interest information for residents of the area - the series is probably fit for purpose; but
• if it is applied to matters requiring more rigour only the data since 2010 should be used.

In what follows I have marked some text in blue to indicate a general grizzle about parts of the BoM system.  They can be ignored  at no risk to the overall narrative.

The main attribute of this heatwave so far has been its duration, and forecast continuation, but it led me to contemplate maximum temperatures in Carwoola for past Decembers.  That in turn led me to ponder the equivalent annual maxima.  On raising the findings of these musings with some folk who know about meteorology severe doubts (which I now share) were cast on the peak annual maximum of 45.1^oC recorded on 24 January 2001.  So I explored the BoM climate data online series to find some local benchmarks, finding two sites which together fill my needs;

• Station 070014 Canberra Airport Comparison NSW (now closed but contains readings from March 1939 to November 2010); and 
• Station 070351 Canberra Airport NSW (still open containing data from September 2008 until November 2017).

In case you wonder why 'NSW' is mentioned in the station name it is apparently because the ACT is part of the BoM administrative region and it is too difficult to change the database system to reflect the Constitution (and/or the ACT being self-governing).  Words - or at least polite ones - fail me!!!

Whatever: here is the graph of the series for Carwoola and Canberra. 

It is important to note that the BoM series is of "Highest Temperature" as the series including the word "Maximum" is Monthly Means.  (As another aside each observation in the Highest series not only has the temperature but also the date on which it was recorded.  While interesting, it becomes a major pain to select only the temperature data if, as in this case, it is that in which one is interested.  Again, my search for polite words comes up empty!)

Eyeballing the chart shows that the two series match quite well apart from the years 2000 to 2006.  This led me to investigate the history of the Carwoola series. 

• It was at one site until 2010; when 
• the device was moved to a second location; and then 
• since 2014 a third location (and different device) has been used.  

I have been able to contact the recorder for the period 2010-13 and he suspects that there may have been a temporary change in the earlier site (eg a tree being pruned, leading to more direct sunlight hitting the sensor) over the period of difference.  However it basically gets into the category "Interesting but unknown"!  

For what it is worth (not much) the overall Correlation Coefficient is 0.64.  The level of agreement since 2013 is very good.

I decided to repeat the exercise for annual maxima .

Much the same picture is presented although while Carwoola is higher from 1999 to 2007 the pattern of variation in the BoM series between 2000 and 2006 is matched rather better.  As a consequence the overall Correlation Coefficient is higher at 0.73.

Having looked at the highest temperatures I wondered what might be revealed by looking at the Average Maxima for December and Annual periods might show.  (For simplicity I have only used a single line in these graphs: the comments above about site shifts still apply.)

The first chart is for December and closely resembles the Highest Temperature Chart.

The next chart is for Annual Mean Maximum.

 Much of the pattern is similar to other series.  The significantly lower readings from 2014 onwards are interesting.  As a case study I compared the daily maxima for 2015 between the two series.  In that year nearly every day from February to August showed a Carwoola maximum more than 1^oC below the Canberra value.  Only in November and December was there more than a handful of days in which Carwoola was warmer.   Given that Carwoola is about 200m higher than Canberra this would seem quite believable.

For the sake of completeness I have also looked at the Average Monthly Minimum Temperatures, with an expectation that Carwoola would be cooler than Canberra.  This would reflect both the increased elevation of Carwoola and, in Summer, the greater exposure to the cooling Easterlies in the evening.

My expectation is met!  I find the "closing of the gap" between 1997 and 2000 interesting, as is the slide in Canberra minima following the shift to the new site in 2010 (just because its BoM data doesn't make it free from error) and the subsequent resumption of situation normal.