Tuesday, 12 September 2017

Winter Weather summary

I recently received a comment on my August Weather report (sorry, can't remember who) to which I replied that I would compile a summary of Winter in total.  I have finally got around to that.

For reasons I have described in an earlier post I define Winter as being July and August.  (In this area June seems to have more in common, both meteorologically and through observed natural history, with March and April than July and August.)


 As would be expected it was a relatively dry year, being well below average rainfall.  However it wasn't extreme.


My data series is missing reports of temperatures for most Winter days in 2001 and 2004 so those years are omitted from the series.  I - or at least Excel - calculated trend lines for the two series but neither of them shown a significant trend.  (The minimum series was closer to significance than the maximums.)

The maximum temperatures were very close to average.  The average minimum in 2017 was also close to the average average minimum (convoluted term shown deliberately to highlight depth of manipulation involved).  I think it interesting that the past 8 years have all been above the longer term average.

The most interesting attribute of Winter temperatures is whether there is a frost or not.  As explained elsewhere a temperature at screen height of 2oC indicates a temperature of 0oC at ground level.  I thus use 3 measures of frost:

  • A light frost (which my Dad used to call a ground frost) with a minimum between +2oC and 0oC;
  • A hard frost (which my Dad used to call an air frost) with a minimum equal to or less than 0oC; and
  • Total frosts ( the sum of the above).

In analysing the data this became complicated by the days on which it was apparent that values were missing (and had been replaced by 0).  I decided that days on which both maximum and minimum temperatures were  0oC were "missing values" while a minimum of that level was dinkum.  As a consequence the number of days per Winter became variable and I have decided to illustrate the 5 of recording days with a frost.
With due allowance for the impact of my messing around as described above, and noting the shape of some non-significant trend lines, I think that what this shows is the impact of the dry years from about 2001 to 2010 in giving a high proportion of hard frosts and the wetter period from 2010 to 2016 giving a higher rate of light frosts.  I shall see what Excel can do in the way of a regression of percentage of frosts against rainfall and update this post with the result.

No comments: