Wednesday, April 08, 2009

Analysis of Australian Temperature - Part 2

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

Part 2:

What is interesting in the analysis done in Part 1, might not have been that 44% of all warming in Australia is accounted for, simply by a better mathematical method of analysing the data, but rather the variations in the temperature throughout the day as shown below:



Why is it that we are seeing a significant warming trend between the hours of 9am and 3pm, and a smaller increase in other times? Surely, if CO2 were the principle mean of global warming, then we should see a constant increase amongst all times. This clearly is not the case in the analysis of Australian temperatures.

We can look even closer at Australian time based temperatures by comparing their anomalies over time as compared to the maximum and minimum temperatures. If global warming has been by and large been a result of an increase in CO2 emissions, then we should have a steady increase when compared to maximum and minimum temperatures over time. This is what we find below:

















What we find in the above is something quite extraordinary. For 9am, noon, 3pm, 6pm and 9pm there is little trend. This indicates that temperatures at these times have increased at around the same rate as the minimum temperature. Readers will note however a large increase in the last 10 years of differences in temperature between 9am and the minimum.

However for midnight, 3am and 6am, we see a very strong decreasing relationship. This indicates that the minimum temperature has increased at a significant rate when compared to these temperatures. Some readers will say that this is understandable considering the analysis done in part 1, however it is highlighted here for significantly.

It seems, that the minimum temperature is more strongly related to the temperature in the middle of the day than at night. Too may times, people quote minimum temperature trends as having something to do with overnight temperatures, when we clearly show here, and the fact that the minimum rarely occurs at night, that this is not the case.

Put frankly, the minimum is a poor measure of overnight temperatures, and its trend has more to do with day temperatures than night. And what of maximum temperatures? An analysis of time vs maximum temperatures is shown below.

















Whilst the above graphs show that maximum temperatures have increased at a greater rate over time when compared to midnight, 3am and 6am temperatures, this was expected. What was not expected is the strong negative relationship between maximum temperatures and time based temperatures during the day. In particular 3pm. Maximum temperatures are shown here to be increasing at a lot larger rate over time when compared to 3pm (and other day temperatures).

This is strange, as we would expect the maximum temperature to be reached between noon and 6pm depending on various aspects, especially the season. What this indicates therefore, is that during the day, Australia is heating up at a greater rate per hour, reaching a peak temperature and then falling in temperature at a greater rate per hour.

One could argue that Co2 could not be the cause of this. The carbon blanket should increase temperatures steadily throughout the night and day at a relatively constant rate. However, what we are seeing here is minimal increases at night, and then steady increase and then decreases during the day, the rate of change of this increase and decrease during the day increasing over time.

What could cause this? Well in Part 3, we will compare time based temperatures as well as look at temperature trends based on season (summer, winter).

8 comments:

Rafe said...

Hi Jonathan,

it seems someone in Mexico has been thinking along similar lines.

From the abstract (emphasis added):

"The diurnal range in surface temperatures (DTR = maximum - minimum temperature) has been widely used as one indicator of potential climate change. On hemispheric space scales DTR trends over about the last half-century tend to be decreasing. This paper analyzes regional scale trends in DTR for Mexico (1940-2001). Our principal finding is that in recent decades (post-1970) DTR trends over Mexico are positive as maximum temperatures are warming at a significantly higher rate than minimum temperatures. Regional land use and land cover changes (LCCs) are identified as potential forcing mechanisms responsible for at least part of the observed DTR behavior."

Jonathan Lowe said...

Great Post Ralph, the greater increase in maximum temperature as opposed to the standard BOM increase in maximum temperature that I found, is most likely due to the urban heat island effect, which is included in my data.

A few points in my dataset when looking at time based temperatures are urban, so this is most likely the reason.

Count Iblis said...

Do you find a significant difference if you compare the data from stations in the Outback to stations in cities?

Jonathan Lowe said...

welcome back count, that analysis will come shortly after this.

College Term Papers said...

This world is quite the big place and to encounter a story such as this one just puts me out of my ordinary. I gotta hand it to whoever wrote this, you've really kept me updated! Now, let's just hope that I can come across another blog just as interesting

Jay Alt said...

Keep plugging away Zog and you may yet invent the wheel -

Historical records of cloud cover and climate for AU, 1991 PA Jones
http://www-cluster.bom.gov.au/amm/docs/1991/jones.pdf

Technical translation services online said...

Would like to see a comparison with for example German data.

Anonymous said...

One thing notable in nearly all these graphs is the clear change of direction in 1992.

This reminds me of Phil Jones' BBC interview comment that there has been "no statistically significant rise since 1995".

Your Australian data suggests this may have happened a little earlier.

In fact if they spent less time "adjusting" the raw data it may be obvious that this is not limited to Aus.

Due to the level of noise in all this it has taken 10 or 15 years for this to become obvious but it seems the last warming cycle peaked early to mid 90's.

Looking at dT/dt in hadSST3 shows a turning point nearer 2004
http://judithcurry.com/2012/03/15/on-the-adjustments-to-the-hadsst3-data-set-2/