Friday, April 03, 2009

Analysis of Australian Temperature - Part 1

Australian temperatures are on the increase, there is little doubt about that. Maximum and minimum temperatures have risen by about 0.7 degrees in the last 100 years. Our analysis will look not only at maximum and minimum temperatures, but also at time based temperatures, which we previously argued are a better more consistent representation for temperature analysis.

Using 21 weather stations around Australia that have accurate and consitant time based temperature data from 1950 onwards, we also found that the maximum and minimum temperatures have significantly increased in the last 50 years. This is shown below. Click on them for a larger graph.

You should note a couple of things. Firstly, the maximum temperature has increased since 1950 at a rate of 1.5 degrees which is larger than normal. This is for a couple of different reasons. One is that temperature has warmed up quicker in the last 50 years than the 50 before that, and secondly, because we have had to use some urban weather station in our analysis which is not normally used because of the heat island effect. Thus the increase is greater than normal.

But that shouldn't worry us or hinder any analysis that we undertake. Secondly you will notice that the minimum temperature has increased at a rate to what was expected or at a lower rate than the maximum temperature. It is still a significant increase, although there seems to be no major changes in the temperature since around 1972. Before this we saw mostly negative anomalies, and after positive.

Either way, both maximum and minimum temperature are shown to have increased significantly since 1950 as expected.

So what about time based temperature? Well as shown below we have 8 time based temperature since 1950 for Midnight, 3am, 6am, 9am, Noon, 3pm, 6pm and 9pm. It must be noted that for some reason the amount of data for 9pm is less than the other time variables.

There are many things that we can talk about with regards to the above graphs. Readers will first note that night time temperatures (Midnight, 3am, 6am) show little increase in temperature. In fact, they average just 0.43, 0.26 and 0.28 degree increase in temperature per 100 years. This is a lot lower than the expected, which clearly indicates that night time temperatures in Australia are only increasing a very small amount.

But how can this be when the minimum temperature is shown to be increasing a lot more? Well the answer is simple and surprising to many, in that, the minimum temperature more often than not occurs during the day. Basically, as soon as the sun sets, the temperature decreases over night. When the sun rises it starts to heat up the atmosphere, and only after 30 mins to an hour after sunrise to we fall to a minimum and the temperature starts to increase again for the day.

Hence, the minimum temperature has little to do with night time temperature and is a lot more influenced by the sun and cloud cover.

Readers will also note the sudden increase in temperature at 9am, noon and 3pm. Which then starts to fade a way a little from 6pm onwards. This is highlighted in the graph below:

So it is clear that temperatures in the middle of the day have increased at a greater rate than temperature outside this time. What would cause this? It is also clear that the maximum (which generally occurs around 3pm depending on the season), and also the minimum (which generally occurs around 6am to 9am depending on the season) are highly influenced by the increase in temperature in the middle of the day.

This graph alone strongly suggests that analysis of maximum and minimum temperatures solely is not an accurate way to measure temperature. A better method would be to take the average of each of the times to come up with an average temperature increase since 1950.

Currently, climate change analysis simply averages the maximum and minimum temperatures to come to a conclusion that the world has increased by 0.7-1.0 degrees over the past 100 years. With our data, this equates to 1.17 degree increase per 100 years. However if we average all of the time based temperatures, which no doubt, would be a more accurate way of measuring temperature over time, we find an average increase of only 0.66 degree per 100 years. This is shown below:

Hence we can conclude that 44% (1 - 0.66/1.17) of all increase in temperature in Australia since 1950 can be accounted for, simply by a better mathematical method of measuring temperature!


We've just reduced the amount of temperature increase that Australia has seen, simply by looking closer to the data and analysing it in more depth. But we are going to get a lot more in depth that this. And that will have to wait till part 2.


Anonymous said...

You should check out "Climate Change and Global Warming" by A. Marterman for a better method of analysis of temp records. GO:

Also check "Still waiting for Geenhouse by the late John Daly. GO:

Click on "Station Temperature Data" and check the plot for Alice Springs. You should use Marterman's method for analysis of the temp records from the Old Post Office to determine if there are any seasonal effects forund by Marterman in his anlysis of the CET.

Harold Pierce Jr

Anonymous said...

Good analysis, and congratulations, picked up the story... I'll post a link over at too. Mike Smith

Anonymous said...

Opps! error in first link. Should be:

If link doesn't work google the title.

Use his method on Tmax and Tmin.

Harold Pierce Jr.

Magnus said...

One can recognize that the higher temperature increase is mostly from the max temperature record.

May this be due to changes in the equipment (the wind shield/sun screen) as in examples in this Climat Audit comment? In night time that should not matter, but in day time. A simple hypothesis...

Eyrie said...

You are certain this data is raw data and not BoM processed by something like the GISS algorithm?

Jonathan Lowe said...

Not sure about that Magnus. I don't have any data to prove that, and Eyrie, yes the data is raw BoM data and taken from this website:

I will look up those links thanks multiple anonymous's

Magnus said...

Jonathan Lowe. I'm not sure about that either. But consistently increasing bias for max temperatures makes me suspect changes in the equipment. Changes in the cover do take place, but these mey be handled with corrections. I have no idea. Since the max temperature seems to be the problem I suspected changes in the absorption of the sun light.

Of course this may be wrong. I'm not sure about anything, and you don't have to be that either... ;)

Magnus said...

Btw, it should be easy to test if the equipment hypothesis is a good one or not. I can't see any reason to not do that. :)

Magnus said...

Another argument for sun light absorption may be that the highest bias is at noon and 3 hours before and after noon. Not when the temperature is highest between noon and 6 pm.

Magnus said...

Oh! Not bias. Trends.

Anonymous said...

Wasn't there a significant change in the way that the raw BOM data was collected at some point in your sequence? Didn't they go from manual observations to automated around the 1980's?

Jonathan Lowe said...

Obviously the BOM did go form observed to automatic temperature recording at one stage. There might be a bias there, but I would expect it to be very minimal.

Magnus said...

Sorry I'm persistent... It may not help you (or it may?). My guess is still that changes in the equipment gives a much stronger warming trend especially at noon and 3 hours before and after noon due to energy absorption from the sun. But something else should explain the marginally higher warming trend at other times of the day and at night. (The automatic temperature recording??)

Jonathan Lowe said...

you might be right Magnus, it may increase the temperature during the day because of the change of equipment, but until proper research is carried out in this area we will never know. It might even decrease the temperature because of new equipment. A proper scientific test needs to be carried out to determine what the case is.

sandymcc said...
This comment has been removed by the author.
sandymcc said...

I was surprised by your upward trends.
See some of my charts at
click on the PDF link
It seemed to me that the cities here are warming but not in the country.
Which stations did you use?

Jonathan Lowe said...

good analysis, but that is only for 3 sample country stations and 1 city station? What about the rest of them?

Anonymous said...


Go to Wolphram Alpha

Search on "Australia, Temperature"

In "history" section, use drop down menu to select "ALL"

... there's no change whatever in Australia's average temp since 1950, which is as far back as the data they have goes.

Anonymous said...

Sorry, link didn't work. Here it is, again, I hope


etc., as above.

Jonathan Lowe said...

where's the history section?

Richard Wakefield said...

Nice work. I've done a but of AU data:


Jonathan Lowe said...

Some good analysis these Richard. In regards to melbourne, the BOM suggest that all records before 1910 are void because of handling errors and techology, so we have to go by that even though the temperatures in the late 1800s were hotter than today. I believe they are correct in this assumption however. Also, with Melbourne there's a massive, MASSIVE urban island effect. See here:

and especially here: