# Global Temperatures – the big picture

Suppose you wanted to measure whether the total number of ants on earth has been increasing. The number of ants at any given place depends on location and on season. Let’s assume that today there are 10,000 botanists at fixed locations across the world diligently measuring the number of  ants passing through each square meter. The daily average population at each location can then be estimated as  the sum of the maximum daytime population plus the minimum nighttime population divided by two. Unfortunately though  100y ago there were only 30 such botanists at work and they used pen and paper to record the data. How can we possibly hope to determine whether the global ant population has been increasing since then? The only way is to do that is to assume that changes  in ant population are the same everywhere because it is a global phenomenon – for example it depends on oxygen levels. Our botanists sample this change at random fixed places. Then as far as possible we should remove any spatial biases inherent in this ever changing historical sampling coverage. We can only do this by normalising at each location the population time series relative to its ‘average’ value within say a standardised  30 year (seasonal) average. Then we can subtract this normal value  to derive the ant population differentials (anomalies). Next  we form a ‘spatial’ average of all such disparate ant anomalies (essentially differentials) for each year in the series. What we can then deduce are annual global ant population ‘anomalies’ , but in doing so we have essentially giving up hope of ever knowing what the total number of ants alive on earth were at any given time.

Measuring global temperatures is rather analogous because they too are based on the same  assumptions, namely that a) temperatures change coherently over vast areas and b) these changes are well reflected by stochastic  sampling over the earth’s surface. The global temperature anomaly is a spatial average over all measurements of localised monthly temperature differentials relative to their average over a fixed period.

Figure 1 shows the decadal smoothed results from GHCN V3/HadSST3. The big picture shows there are four phases:

1. 1880-1910 Falling or flat
2. 1910-1945 Increasing
3. 1945-1975 Falling or flat
4. 1975-2015 Increasing

Figure 1: a) Global Temperature Anomalies. Data are 10 year rolling averages  b) Differences between GHCN anomaly trends.

The individual station anomalies for Tav, Tmin and Tmax have each been computed using their respective seasonal averages between from 1961-1990. Note how all the series get zeroed together at the fixed normalisation period. This is an artefact of the choice of baseline period. We can also observe the following apparently odd effects in Figure 1b.

1. Tav ‘warms’ faster than both Tmin and Tmax after 1980.
2. Tmin warms faster than Tmax after 1970. There is other evidence that nights warm faster than days
3. Oceanic temperatures were warmer than global temperatures  before 1910 and then again between 1930 to 1972, but have since lagged behind land temperatures. This appears to be a cyclic phenomenon.

The zeroing effect in differences is again an  artefact of using temperature anomalies.  However, if one looks at these trends dispassionately one must conclude that there is an underlying  natural oceanic cycle of amplitude ~0.3C and wavelength ~90y which drove global temperatures until ~1930.  Since then a slow but increasing CO2 induced warming effect has emerged which has now distorted this natural cycle.  This has resulted  in an underlying  global warming of about 0.8C.

This entry was posted in AGW, Climate Change, climate science. Bookmark the permalink.

### 34 Responses to Global Temperatures – the big picture

1. Joe Public says:

Hi Clive

The ant population won’t change much, irrespective of the hour their number is estimated

“Ant colonies can be long-lived. The queens can live for up to 30 years, and workers live from 1 to 3 years. Males, however, are more transitory, being quite short-lived and surviving for only a few weeks.[64]”

https://en.wikipedia.org/wiki/Ant

However, diurnal temperature swing can be as much as 56K, so the moment of measurement can be significant.

In those olden days of pencil & paper recording in far-off lands with little supervision, who can be sure the scientist measured at the same time each day?

https://en.wikipedia.org/wiki/Diurnal_temperature_variation

• Clive Best says:

You’re probably right. I just couldn’t think of a better analogy to measuring global temperature anomalies. If you go well back before ~1930 then you just have individual voyages that measured sea temperatures and a few scattered weather stations with almost none at all in the largest continent – Africa.

• Ron Graf says:

Just so you don’t alarm any Chinese or Russians I think Asia is still the largest. Africa is #2.

• Ron Graf says:

Africa is #1 in ants.

• Clive Best says:

Especially if you include termites. 😉

• Clive Best says:

True – especially if you include Europe.!

However, Antarctica is about twice the size of Australia and there were no stations there either.

2. Windchaser says:

However, if one looks at these trends dispassionately one must conclude that there is an underlying natural oceanic cycle of amplitude ~0.3C and wavelength ~90y which drove global temperatures until ~1930.

Based on n=1? Eeek. That is not sound science. It’s kinda assuming the conclusion; dispassionate science would say that you should look at alternate hypotheses, first, before drawing broad and firm conclusions like that.

You can’t go from “the ocean temperatures cycled once” to “therefore, there is an underlying natural oceanic cylce of X amplitude and Y wavelength”.

• Clive Best says:

We cannot rule out some longer 200 year cycle but it seems very unlikely since there is no past evidence of this happening.

• Windchaser says:

Who says it’s a cycle at all? Instead of the results of stochastic or semi-stochastic changes in forcings.

3. billbedford says:

How can Tav be larger than both Tmin and Tmax in your first graph?

• Clive Best says:

Tav is the daily value (Tmax+Tmin)/2 then averaged over one month, whereas Tmax and Tmin are the monthly averages of their respective daily values. Clearly the absolute value of Tmax must be greater than the absolute value of Tav. Now comes the trickery of using ‘Anomalies’.

1. The 12 monthly averages for each series is calculated from 1961 – 1990.
2. We subtract these ‘normal’ values from the entire temperature series to get anomalies

So if the Tav normals happen to be relatively lower than those for Tmax within this 30y band then the Tav anomalies appear larger than the Tmax anomalies. Likewise for Tmin.

Physically though in deg.C. Tmax < Tav < Tmin

• Bryce Payne says:

Did you not mean Tmin<TavTav>Tmin?

• Clive Best says:

Aargh! ….Sometimes I type too fast and forget to check

Physically though in deg.C. Tmax > Tav > Tmin

• Danny says:

Hi Clive,

Looking at your graphs I had similar thoughts as billbedford: how can it be that Tav get “outside” of Tmin and Tmax. All operations done on the data are linear, so shouldn’t Tav always stay somewhere between Tmin and Tmax?

But the graph shows 10-year averages, so maybe not so obvious to be 100% sure…

It happened to be that I downloaded last week the temperature series for Macquarie Island; I got the monthly minimum and maximum temp, so I had to calculate the average temp myself as well as do the conversion to anomalies. After reading your blogpost I made some extra calculations: Tmax and Tmin anomalies and also Tmax – Tmin anomalies. This can be seen on the following graph (all anomalies):

?dl=0

What can we see?
– Tav is always at the midpoint between Tmin and Tmax
– (Tmax-Tmin)/2 gives the difference between both Tmax/Tav and Tav/Tmin
– Tmin anomalies can be greater than the associated Tmax anomalies

So, looking at my graph, I am almost sure there must be some error in your calculations… unless I am missing something…

• Danny says:

Oops… the link didn’t come through. Hopefully this will work:

http://www.dropbox.com/s/ihwwmc56yaywph2/Anomalies.png?dl=0

• Clive Best says:

Hi Danny,

You are looking at one site, so clearly Tav must always equal (Tmax+Tmin)/2 and will lie exactly between them. If you make the anomalies based on a 30y set of normals then the same relationship should hold. Now suppose that Tmax remains constant but only Tmin is increasing. Then the anomalies for Tmax will remain zero while those for Tmax and Tav will increase, although Tmax should increase faster. So both will rise above Tmax.

Likewise for Tmin remaining constant and only Tmax is increasing. Now average these anomalies over 7000 stations. I think it is possible to have the anomalies for global Tmin to rise above Tav.

• Danny says:

Well I don’t think so.
We both agree that for one site Tav is always at the midpoint between Tmax and Tmin and equal to (Tmax + Tmin)/2. Now let’s introduce Diff, the difference between Tmax/Tav (or Tav/Tmin): (Tmax – Tmin)/2

Tav + Diff = (Tmax + Tmin)/2 + (Tmax – Tmin)/2 = Tmax
Tav – Diff = (Tmax + Tmin)/2 – (Tmax – Tmin)/2 = Tmin

So we now can replace Tmax by (Tav + Diff) and Tmin by (Tav – Diff)
Just notice that Diff is always positive if Tmax and Tmin are measured temperatures, but can be positive or negative if Tmax and Tmin are expressed as anomalies of the measured temperatures.

Now let’s average such Tmax/Tav/Tmin-anomaly-triplets over 7000 (world) sites.

Tav(world) = (Tav1 + Tav2 + … + Tav7000)/7000

Tmax(world) = (Tav1 + Diff1 + Tav2 + Diff2 + … + Tav7000 + Diff7000)/7000
= (Tav1 + Tav2 + … + Tav7000)/7000 + (Diff1 + Diff2 + … + Diff7000)/7000
= Tav(world) + Diff(world)

Tmin(world) = (Tav1 – Diff1 + Tav2 – Diff2 + … + Tav7000 – Diff7000)/7000
= (Tav1 + Tav2 + … + Tav7000)/7000 – (Diff1 + Diff2 + … + Diff7000)/7000
= Tav(world) – Diff(world)

So all world temperatures will consist of Tmax(world)/Tav(world)/Tmin(world)-triplets with the characteristic that Tav(world) will always be exactly in the middle between the 2 others…

4. Mr Broccoli says:

Why are you giving botanists disgusting, entymological jobs to do?

• Ron Graf says:

I’m sure people ridiculed the silly English monks who started recording a daily readings on their newly invented thermometer.

5. LouMaytrees says:

Clive,
Why do you start your + 0.8*C claim beginning at 1930 when your # 1 is labeled 1880-1910 and which clearly shows global temps start rising then (1910), adding an additional + .2*C to that total?

Also your graphs only go to 2012, some 2013 at most while # 4 lists 1975 thru 2015? Does that also under distort the ” … underlying global warming of about 0.8C” big picture total?

• Clive Best says:

Because I subtract the effect of the oceanic oscillation. 1910 coincides with the minimum of the signal.

The plots shown are 10 year running averages so the stop 5 years before the last available year – 2018. However the effects of these years are still apparent in the average. Essentially it removes the el Nino signals.

• LouMaytrees says:

Thank you.

6. Hugo says:

Next question would be what are reliable CO2 levels and what is the normal and or best proper or optimum co2 level. And at which height and location. and is it human based or is it of other origin like water land, sedimentary rocks or volcano or ice melting or …. Did we save our selves from an ice age boundary if it would have dropped under .03

• Clive Best says:

Mauna Loa measurements and those elsewhere are pretty accurate so CO2 has definitely increased by ~30% . There is no ideal value for CO2 although it is probably not too healthy to breath air much higher than 1500ppm for long periods.

We may well have saved ourselves from the next ice age.

• Hugo says:

Dear mr. Clive. Let me say, I really love your work and thank you for your response.
Do allow me the following remarks.
And please correct me when I am wrong.

Mauna Loa measurements and those elsewhere are pretty accurate so CO2 has definitely increased by ~30% .

You are correct. and I do not dispute their accuracy of measurement for the last 30 years or so. However a 30 procent rise you can not project that globally. Its a local figure only.

Mauna Loa measurement is at app 4000 meters and in the middle of an ocean near or on top of one of the most active volcanos on earth.

I have seen it fluctuates per day per hour and per season.
Statistics and careful modeling can not be the answer. to a presumed global issue. I state insufficient data.

Statistics and daily measuring. You probably know how difficult it is to keep data clean. With climate its extremely difficult to get clean un biased data. And when dat seems to be under control an earth quake shifts mount Everest for a meter up or down. Which can shift any baseline Sea-level, no satellite can do the job properly, you have to go there, local measurements are needed

Unless the grid is global at a meter or ok a one hundred or 1 thousand meter scale from minus 9000 metes up to a 100 km plus we can not claim global.

And as local variables can be higher or lower its real value does not stand grid, as its merely a floating number a number on its self. and that is how we should interpretate that. Maybe we should try not to globalize but regionalize.

• JonA says:

I think you meant glaciation. We’re still in an Ice Age 😉

• Hugo says:

Yes ok. you are right ice age is the popular term. Glaciation. Well actually no. We do not know that. if we are in it.
As it seems the 2 major data collection points are the USA and Russia. and then some vague Africa countries.
Funny to see that after Trump stepped out of Kyoto. Suddenly USA co2 emissions dropped.
I suppose Hawaii will try to keep it up, yes they are usa but no they are nor directly trump funded. So I wonder when they will change too. Pretty soon I think. Hawaii seems to be a strong hole. for measurement. But its the most flawed of all.

7. Hugo says:

During the 70ties we became co2 aware and after 2000 we made it a problem. In medicine they do double blind tests. so and the patient does not know and the doctor does not know. because if not this will influence on results.. It looks like we filter more to upwards and anomalies which are lower are filtered out. the temperatures must rise if not its the end of many stupid research with the word CO2 in it or the connection to global warming.

• Clive Best says:

Perhaps the answer then is to make contact with isolated tribal groups in New Guinea and ask them if they have noticed the climate changing over the last 50 years.

• Hugo says:

Dear mr Clive, Yes the way to go. Are you sure what you are about to unleash.

if you ask a Papoea or a Wachiri or Amazon Indian well they do not care about temperatures or C02 levels, as they have no clue. Actually none of 99.9% of the worlds population has too.

They live between trees. They live the real green life, they care about family and tribe, food and rain, they are children of the forests and the last thing they need is western educated people near.

They know so called civilization and research will kill the forests and them too all in the name of progress and science…. Because so fare every contact and so called progress was fatal.

We cry wolf over Hydrocarbon release. While its only 10 percent of which is commercially discovered, and can be profitably used.

We killed over 50 percent or more of rain forests and more than 80 percent of big land and sea grazers.
But claim Hydrocarbon burning C02 release is the culprit.

• Ron Graf says:

““What is disconcerting to me and so many of my colleagues is that these tools that we’ve spent years developing increasingly are unnecessary because we can see climate change, the impacts of climate change, now, playing out in real time, on our television screens, in the 24-hour news cycle.” Michael Mann

https://www.washingtontimes.com/news/2016/jun/27/michael-mann-climate-scientist-data-increasingly-u/

8. Hugo says:

Dear Mr. Graf. Thank you for your repose

Your tools are not unnecessary actually they are very essential but not in the way you think it is.
Tools and models are vital in testing and enhancing understanding. But do not presume the creation of the tool is the goal.

You know the Paris example is not the best example to prove the point you think you claim
If you want i can explain in depth what happened in Paris and what will happen again and is pretty normal for Paris and actually for Germany Belgium and the Netherlands too.
Paris is actually the bad example its similar to New Orleans claiming its flood was global warming caused too. Currently its situation changed to the opposite.

I state, you did not create a tool for science . i state currently you actually created a tool for funding the tool.
And that awareness became true later on. That’s ICT. All computer guy.s and modelling guys think they know all the variables and yes we can do the model. But like most ICT projects. They tend to be way more complex than thought when started. they always go over buget, you tell me why.

But I look forward in thinking over its variables if you care.

9. It doesn't add up... says:

Off topic: you answered Andrew Mountford with this:

You may like to think about the locations and of wind farms, and when they started production, in the light of the metered trends in industrial and commercial consumption regionally as shown in this chart:

The overall decline is about 20TWh across the GB from 2005 to 2015, or about 10% of industrial/commercial demand. Updating to 2017 data, it’s over 26TWh of decline:

GWh
……………. ………2005 ……..2017 ……Diff ……% Change
Great Britain …….200,889 ……174,322 …….-26,568 ……-13.2%

England ………….166,681 ……144,585 …….-22,097 ……-13.3%
North East …………9,349 ……….7,338 ………-2,011 ……-21.5%
North West ……….23,115 ……..19,096 ………-4,018 ……-17.4%
Yorkshire and
The Humber……… 17,032 ………14,441 …….-2,591 …….-15.2%
East Midlands ………15,294 ………12,933 ……..-2,362 …….-15.4%
West Midlands ………16,939 ……….14,682 ……..-2,257 …….-13.3%
East ……………………17,192 ……….15,402 ……..-1,790 …….-10.4%
London …………….27,550 ……….25,310 ……..-2,240 ………-8.1%
South East ……….24,400 ……….21,977 ……..-2,423 ………-9.9%
South West ………..15,812 ……….13,407 ……..-2,405 ………-15.2%
Wales ………………11,912 …………9,891 ……..-2,020 ………-17.0%
Scotland ………………17,114 ……….14,136 ……..-2,977 ………-17.4%

Most behind the meter generation is concerned with triad avoidance, and is thus limited to cold winter evening rush hours, and is diesel based. There are a few exceptions: the wind turbines installed by Nissan and Ford for example, or onsite generation at oil refineries (which has been in decline as the UK shuts theirs). Wind farms are metered where they are embedded generation (we have to know how much subsidy to pay them) – don’t confuse the fact that those meter readings aren’t visible to National Grid in real time: only small systems of under 30kW export potential are unmetered.