Some weather graphs

This is possibly a post for people with "special interests" but the title tells it all.  The post is probably mainly a scribbling pad by me, to try to get my ideas together but if it ends up comprehensible I may publish it to get views from others

When reporting on the weather each month I have tended to use a chart containing a couple of line graphs to indicate the maximum and minimum temperatures.    This is an example from January 2014.

That is all very well but in reality the temperatures don't move like that, with a lot of up and down each day.  So we could put in a single line:

.. but apart from difficulties with axis labels I find it difficult to spot what is going on there.  It looks like a scrumhalf's pulse rate during a close Grand Final!

On mucking about with the EXCEL charting function I noticed one of the styles they offer seemed to have a central "box" with lines emerging from the top and bottom which seemed to be a fair way of presenting the data.

It turns out this is a special graph for presenting movements of shares on a stock market and requires 4 columns representing opening price, high price, low price and closing price for each day.  However computers are largely stupid - or at least unquestioning - so as long as there are four columns per row it won't fuss too much about what the data actually represents.

So my first effort (A) was to have rows for each day with columns for: 

• temperature at 0030;
• maximum temperature during the day;
• maximum temperature during the day; and
• temperature at 2330.

A small surprise arose with the boxes varying in colour.  This represents whether the 0030 temperature was higher (yellow) or lower (blue) than the 2330 temperature, which is somewhat interesting.  The length of the box is the difference between the two.  Also of interest is the position of the box relative to the minimum and maximum lines.  So there is a lot of information in that chart - or is it too much?

I then wondered what would happen if the two timed temperatures were made the same (eg by copying the 0030 values to the 2330 field).  That gave (B):

In effect the boxes compress to a single marker.  The information which disappears is the comparison of the start and finish temperatures, but I am not sure what that really adds to knowledge anyway!  Unusual events in which the days minimum occurs close to 0030 (vide 21 January) can be picked up (as would be the case if the day's maximum occurs then - the red dot would be at the top of the line rather than the bottom).

A further approach is to set both time based temperatures to the daily average temperature.

This graph would be interesting if the distribution of temperatures was subject to kurtosis. This might arise if (for example) most of the day was between 15 and 25^oC but a cold front came through depressing the temperature to 8^oC for a few minutes.  In this case the marker would be close to the top of the line.  Looking at the data for January I found that the only large deviations were around the 20th and 21st where the maximum is a long way above the average (suggesting a cloudy day with a brief spell of sunshine).  However that doesn't strike me from the Chart so it doesn't fulfill that objective.

My personal view is that A and B are more interesting, and suitable than my traditional double line approach.   I'd welcome comments on this - if readers haven't nodded off by the second paragraph!