Weather report July 2018

As I estimated the last week for the Gazette report there are a few small changes, notably for the average temperature (dropped a small amount) and rainfall (went up a small amount).

Note that for future months, for reasons explained below, the temperature material for past years will be restricted to the period since 2010. Rainfall will continue to cover the period since May 1984.

My view before analysing the data was that it was a very dry and pretty cold month.  After analysis those views are confirmed:

1. Both the negligible rainfall and relative humidity data show a dry month.
2. For temperatures the average maximum was a little above average but minima and frost-days were both indicative of a cold month. Bringing those together it appeared that average temperatures were .... average!

Rainfall

My weather station recorded 8.8mm for the month.  That is the lowest recorded for July in Carwoola since our Carwoola records started in 1984.  There are however several lower totals for July offered by BoM for the Canberra Airport Comparison site (site 70014) which closed in 2010. So, bad as things are, they could be worse, on that measure at least!

The basic situation is shown in this graph:

I have in past months commented on the total rainfall since January 2017.  By last month that was 18 months which matches one of the standard BoM series, so here is a scatterplot of rolling 18 month totals since 1985.

The last two values almost overlay each other, and each was the lowest value recorded.  Here is a BoM map for 18 month rainfall deciles.  

It appears, from the metadata shown on the above linked page, that the 'average' to which the current values are compared relates to observations since 1900.  Noting that Julys in the the 70s were drier than the current arid period it is believable that Carwoola is on the cusp of the 1st and second deciles (rather than lowest on record as suggested by the data since 1985).

Using the Carwoola data I extended the range to 19 month totals and the scatterplot looks very similar, except July 2018 drops well below June 2018 (as the month of July 2018 was much drier than July 2017).

It is still bloody dry!  As at the end off July we are at 47% of average rain, year-to-date and looking at a total of 337.1mm for the year. 

Temperatures

The basic situation is shown in this chart of daily maxima and minima.

By and large it is what was expected   Most maxima were in the range 10^oC to 15^oC and most nights having a frost.

Before getting to subcategories I will note that as I have become more familiar with the data set, and with other data available I have found that some of the older records do not appear compatible with (a) records taken in Carwoola since 2010 nor (b) observation from Canberra Airport (BoM site 70014 - now closed).  In contrast the more recent Carwoola records do show a good correlation with observations from the currently active site 70351. This is illustrated in the discussion of minimum temperatures below.   I have thus decided to only use Carwoola records since 2010 for my analysis of temperatures

Maximum temperatures 

The average maximum for the month was 11.1^oC a touch lower than year and almost exactly on the average.

 Looking at average maximum temperatures for July since 2010 shows some variation around ~11^oC but no significant trend.

The highest maximum temperature in the month was 16.3^oC recorded on the 5^th.  This is the second highest (of 279)  temperature recorded in July since 2010.  (The highest was 17.3^oC recorded on 19 July 2016.)  It is notable that the 15.3^oC recorded on the 28^th of last month was the 4th highest since 2010.

In contrast the lowest maximum in the month (7.1^oC on the 7^th) was the tenth lowest July maximum.  As might be expected from the low average maximum, most of the lowest daily maxima occurred in 2015.

Minimum Temperatures

In recent months I have commented how the minimum temperatures have all, like the children of Lake Woebegone, been above average.  In the past month I have re-evaluated that view and concluded that minimum temperatures in the early years of the series I was using were not consistent with later readings.  The comparison is illustrated in this chart:

The major concern is the area ringed in red where the Carwoola data is much below the Airport data for a few years but then 'jumps'.  By contrast the later data, ringed in green, while still somewhat below the Airport data does show a similar pattern.  The middle set of Carwoola data does show a pretty good correlation with the BoM information (but I have doubts there about the maximum data).  Thus, I am only using the data since 2010.

Here is the basic chart.

The subplot with minimum temperatures is the number of frost-days.  In July we recorded 23 days with temperature below 2^oC (a ground frost) of which 21 went below 0^oC (an air frost.) The ground frosts were a smidgin above average since 2010 while the air frosts were well above average.

The lowest temperature recorded was -6.2^oC which is - to my surprise the lowest July temperature recorded since 2010.  It is the 3^rd lowest temperature for any month since 2010 with the lowest being -6.8^oC on 5 August 2014.

Average temperatures

There are two ways of obtaining the average temperature for a period.  

1. The first, and more reliable is to take the mean of all available temperature readings for the period.  In my case for each day that is the sum of the 24 hourly readings divided by 24.
2. Where that level of detail is not available the average can be estimated as the highest value plus the lowest value, and the sum divided by 2.

Both methods are illustrated in the following chart.

The two lines follow one another quite well, and tend to turn in the same direction most of the time, so it is not surprising that they correlate well: the correlation coefficient is 0.97.  Its not perfect, so where important it is far better to use the actual data if it is available.

 One application of average temperatures is to calculate the 'anomaly' which is the difference between the current value of a variable and its long term mean.  For average temperature I have calculated two values of the monthly anomaly as shown below.  For years since 2013 I have the actual data available whereas for 2010-12 I only have the estimated mean data.

While the actual values of the anomaly differ somewhat - and for a couple of months have a different sign - the overall pattern is very similar: April was a bugger of a month!

In looking at the daily average temperatures I noticed a rather low value (0.5^oC, on 22 July, as the outcome of a day ranging from -5.3^oC to +8.8^oC.  This is the 5th lowest exact mean temperature I have recorded.  This led me to wonder about low mean temperatures for each month.  I arbitrarily defined a low mean temperature as 4^oC or less.  Means this low have been recorded, from 2014-18 in 6 months (albeit only once each in September and October)  The number of occurrences by year is shown below.

July 2018 is right on the average following June 2018 which was slightly below average.

Humidity

The levels of humidity recorded in July were quite low.  Which is of course quite consistent with appallingly low rainfall.  A bit below last year (al;so dry) and a lot below the average.

The dryness is especially obvious when the values for 0900hrs are checked.   I'd normally expect them to be above 80% at this time of year rather than below 70%!

Wind

The month was quite windy, as I'm sure everyone noticed.  My station is relatively sheltered from the wind so the speeds here are well understated, but it does give an idea of relative windiness.