Obsessing about the anomaly

 I have commented previously about aligning the anomaly with data from my Weather Station (WS).  I have recently realised that one impact of changing to a WS base is the timing of the observations, since the official measures in Australia work off a day starting at 0900 hrs.  I think this reflects the historical records being based on this time as in most cases observers wouldn't want to get up at midnight to make their daily obs.  

The potential impact of this difference has reared its head in 2 ways recently:

• Discovering thar many of the 'calendar 24 hour minima occur at 2300 hours; and
• A few recent days where the minimum 0000 hrs to 0900 hrs has been high, but a significantly lower temperature has been recorded later in the day (typically, lowest at 2300).  

The first of these impacts is illustrated by this chart of hourly minima from 18th and 19th of February.

After the overnight warmth of the 18th the temperature drops more or less steadily until about 0900 hrs on the 19th, with a low value for the 18th at 2300.

Looking at the whole month of April there are 4 days (about 14%) with an all day minimum lower than the overnight minimum.  

This seems quite significant but on looking at the impact on the overall calculated average minimum for the month as a whole it only causes a very small change from 20.78C to 20.61C.  Also the use of  'all day minima' does get the calculated average closer to the actual average temperature given by averaging the set of hourly temperatures for the month (20.06C).

In summary, and recognising that this 'investigation' only looks at a single month, I don't think the different reference periods gives rise to any concerns.