Sunday, 21 January 2018

Ranging around the matter of temperatures

In my weather reports I usually comment on maximum and minimum temperatures and less frequently mean temperatures.  I can't recall saying much about the range of temperatures.  As a result of some comments on a weather discussion group I decided to look into this aspect of our climate a little more.  The first lot of commentary refers to the period from 2014 using my Davis Weather Station, so the collection mechanism is constant.

I classified the daily ranges to arbitrary groups.
That looks pretty close to a normal distribution to me.

The total range of temperatures I have recorded is from -6.8oC to 40.7oC or 47.5oC.   So that basically sets a limit on daily ranges!

The greatest daily range I have recorded since starting up my weather station here in 2014 has been 29.2oC.  That occurred on 17 January 2014 and was a range from a minimum of  9.7oC to a maximum of 38.9oC.

The lowest range was 1.5oC, on 1 September 2016 from 4.2oC to 5.7oC.

When looking at these extreme ranges it seemed that low ranges mainly occurred at lower temperatures.  By way of example of the 106 lowest ranges (those equal to or less than 6.2oC) only 3 were associated with a maximum temperature of greater than 20oC.  In contrast looking at all maximums almost exactly half were greater than 20oC.

At the other end of the scale, of the 100 days with a range greater than or equal to 22.2oC, only 1 had a minimum below 0oC. This contrasts with 18% of all days having a minimum below zero.  I think this is fairly logical: if the weather conditions lead to a frost it implies weather coming from the South but to get a large range in temperature requires at least a warm maximum, more likely to be associated with weather coming from the north.

It is also logical (IMHO) that a high proportion (~63%) of the days with top 100 ranges have a minimum below 10oC.  This comes about through arithmetic.  Relatively few days have a maximum above 30oC (and in most of those the range is less than 22oC) so to achieve a range >22oC effectively demands a minimum below 10oC.

I also looked at the monthly average ranges.  Note that the vertical scale has been truncated to clarify - perhaps overstate - the pattern.
This pattern shouldn't be surprising in view of the relationship between size of range and temperatures (big range with higher temperatures, small range with lower temperatures).

I decided to have a look at the level of variation in the data for each month using the Standard Deviation (SD) function within ACCESS.  The surprising (to me at least) outcome of this  was how similar the level of variation was.  This becomes apparent when expressing the SD as a % of the mean.  8 of the 12 months had an SD within the range 29.85% - 33.77% of the mean average range.  The only outstanding value was June where the SD was 39.75% of the mean.  As I have trouble establishing whether June is the last month of Autumn or the first month of Winter on both meteorological or biological grounds this is probably to be expected (I have concluded it's the last month of Autumn). 

Longer range analysis

As well as my personal data I also have access to maximum and minimum temperatures collected since 1993 from two other sites in Carwoola.  Although I have some doubts about the maximum temperatures from the early -mid noughties I thought it might be interesting to look at the ranges from the earlier years (1993-2013) in comparison to my data.
The similarity of pattern is immediately obvious, as is the greater range shown in the earlier records.  I strongly suspect that this is mainly a function of the different collection methods (site/equipment) rather than indicating a change in the weather.

No comments: