On Humidity

Back in the day, when we lived in Adelaide and I used to visit Canberra occasionally for work, it was always nervous when a visit was scheduled for Autumn/Winter since there was often the prospect of the Canberra Airport being closed all day due to fog and thus we would be unable to leave.  (It never happened, but there were days when the visibility didn't improve until the afternoon which made for nervousness.)  This is a major reason for humidity being of interest.

The second reason is that humidity is the other half (to temperature) of Summer discomfort, in a similar way to wind being the second factor to Winter discomfort.  Runners are well aware of this. While living in NYC I always noted the humidity/temperature combination when I went for a run before work.  A Bad Day was when the temperature was above 80F and the humidity above 80% at 6am.  (When we lived in Dar es Salaam that was a Normal Day.)

Thirdly humidity is a factor in bushfire risk.  When the humidity is high the fires will not spread as quickly as they might when it is low.  (A friend who was in Canberra during the 2003 bushfires reported that humidity on the afternoon of 18 January was around 4%.)

However when doing my note about the Weather of January I found that the plot of maximum humidity was not very interesting or helpful since it was usually in the range 90 -99%.

Thus I decided to explore around the topic of humidity to see what could be done as an indicator.  Since maximum was not good, I wondered  how 'minimum' might go.

Plotting the raw data produced a chart resembling a cross section of the Alps (see blue line) so I took a 3 period moving average to calm things down a bit (see red line).  That shows the very dry period in early January with increasing moisture in late February and seems quite useful.  (I applied my favourite 4 period polynomial trend line which gave a value of 45% to the correlation coefficient so was not greatly better than eyeballing the moving average.)

Looking at the humidity through the day showed a rather clear pattern.  Through the three months for which I have data the humidity drops through the day rising again in the evening and overnight: no big surprises in that statement but here is a Chart.  I have plotted the same data for December to indicate whether the months differ but other than the drop occurring a little earlier in the day the picture is very similar.

That led me to wonder if a plot of humidity in the late afternoon (5pm seeming like a reasonable time) would reveal more interesting information.

In brief, no.  If the minimum and 5pm graphs are plotted together the only difference is a slight reduction in the level of variability of the 5pm chart.  (This is a good thing but since the minimum is an extreme it is to be expected.)

So all of that leads me to conclude that plotting the 5pm humidity will give a fair indication of whether a month has been more or less humid.