Thursday, April 16, 2009

Analysis of Australian Temperature - Part 3

For all new readers, please read Analysis of Australian Temperature - Part 1 and Part 2 first.

Previously we found that temperatures overnight in Australia are staying relatively steady over the years despite the minimum temperatures increasing. Temperature trends over time suddenly rose when the sun made an appearance, increasing at a greater rate until the middle of the day and then fading away. A case for the sun can be easily argued here.

This can be backed up even more when looking at time based differences over the years. Shown below are the differences in temperature trend over the years for two neighbouring times.















The above graphs show us a few things. Firstly that there is no difference in temperature trends over time at night from 9pm to 6am. Temperatures are steady over time. Then suddenly at 9am we see a sudden increase in temperature. Temperatures remain steady throughout the day (despite a seemingly seasonal trend?), but at 6pm the temperature drastically decreases when compared to 3pm. All this could have been deduced from the previous graphs.

But what happens if we look at the differences in trends from summer and winter?
I will look into this further and in a lot more detail in the next post.

6 comments:

Richard Wakefield said...

I have seen this too in Canada: http://cdnsurfacetemps.wordpress.com/2010/11/24/why-mean-temp-is-meaningless/

Anonymous said...

"The above graphs show us a few things. Firstly that there is no difference in temperature trends over time at night from 9pm to 6am. Temperatures are steady over time. Then suddenly at 9pm we see a sudden increase in temperature."
------------------------------------;

I believe there's a typo in the last sentence in above paragraph.

Shouldn't it read "Then suddenly at 9am, we see a sudden increase in temperature?

Jonathan Lowe said...

you are def right about the typo there anon, i shall correct, thanks for that

Jonathan Lowe said...

good link there Richard, Whilst I agree that Tmean is a terrible way to measure average temperature, that articale doesn't really prove much so imo. If the process is kept consistant, then so should the results. Of course, its just adding in a lot more noise.

Anonymous said...

This is very significant. It needs much wider coverage.

I can't believe this has been up for 3 years and I've not seen it before.

One thing seems to be missing above : the 9pm to midnight graph. Why so?

The pattern of warming and cooling times seems to put this firmly into what one would expect from a reduction in cloud cover / albedo.

Amazing that the last 30 years of climate science has not even got past basics like this yet.

Anonymous said...

Phil Bradley's post at Bishops Hill shows 9pm data is the only time slice that does not have a significant correlation to cloud cover. I offered this explanation:

Since there is a negative correlation in day time and positive correlation at night, this will break any correlation and would probably account for the near zero correlation over the full period.

I suspect correlating winter and summer separately will show 9pm data falls into line with the rest of the data.