2014 Annual Weather report

2014 was the first year that I have had a fully functional weather station and thus the first year for which I can write a report on the year as a whole!  As a consequence of this I am starting out with pretty much of a shotgun approach looking at things of potential interest and seeing what emerges, rather than having clear hypotheses to test.

1. If methodologist colleagues from StatCan read this, please forgive me.  
2. If the result is too incoherent, will the rest of you also forgive me!

Of course this only covers a single year so analysis is restricted, but one has to start somewhen.

Rainfall

Because I have had rain gauges of various types ever since we moved here I have actually got a n 8 year series for this item.  A basic chart, comparing the current year, previous year and average,  is part of my monthly reports and is reproduced here for completeness.

Apart from the record low fall in January and the record high fall in December the rest of the year was about normal.

I have also started to look at rainfall rates and have shown the maximum rate per month in the chart below.

Clearly there were storms around in February, May, November and December!  I need to give a little more thought to how to present"average rates": doing a simple average comes out with values close to zero because there are usually many recordings of zero fall per half hour per month.  I have averaged those readings >0 to give this result, which is not entirely good.

I think what it shows - allowing for the different scales - that some months (eg November and December) have sustained periods of rain with some storms while in others (eg September) much of the rain comes in storms.

The next chart shows the amount of rain x year and the number of days in which more than 5mm were recorded.

For reasons that escape me, when I use 2 vertical scales on a column chart they always appear 'stacked'.  I hope it is clear what is shown.

I have used days >5mm as an indicator, since a meteorologist friend has in the past advised that 5mm is the basic amount needed to get some moisture into the soil.

It has been interesting in the past to observe the flows in Whiskers Creek following various levels of rain:

• sometimes a sudden storm of 10mm will cause the Creek to go from zero to a flood; while
• on other occasions we can get 50mm in a day and the Creek doesn't flood and drops to no flow in a very few days.

Obviously this will vary according to the moisture already in the soil.

Temperatures

The first chart shows the maximum and minimum temperatures recorded each month and the average of daily maxima and minima for each month.

In terms of comfort, the higher maximum temperatures are far more tolerable is they are accompanied by modest minima so I have looked temperature ranges for each month.  In this chart the red line is highest daily range for each month while the blue line is the average of the daily ranges.

On first glancing at the values for May and December I wondered if the chart was a proof of Trewin's Law which states:

"Any interesting statistic is probably a processing error."

However what appears to be happening is that in May the daily maxima were still in the upper teens while the minima had dropped to close to zero.  This gives a wider range than April (with higher minima) and June (with lower maxima).  .In December the maxima are similar to November but the minima are higher and thus the range is lower.

Humidity

For reasons which I cannot now remember I have always used the 5pm (1700Hrs) value when working with humidity.  In contrast the Bureau of Meteorology uses 3pm (1500hrs) - I suspect because it is close to the end of a standard workday but still allowing enough time to do reports etc.

So I have decided to chart the average of daily humidity reading each month for the two times.

The similarity of pattern at this level is very clear.  The two series have a correlation coefficient of 0.98 which indicates significant similarity.  A day-to-day Chart is just about impossible to interpret at this scale but the two series have a correlation coefficient of 91% which with 413 observations (including a few from late 2013) is pretty strong.    For the future I shall move to 1500hrs.