Short term averages - what do they tell us?

[Chris Knight]

Most of us are familiar with 30 year means to show trends in global warming. Rolling means are one, fairly simple way to remove the noise from a temperature time series to show when there was a noticeable change in the trend.

Following a thread by Basil Copeland on Watts Up With That?, I was intrigued by the graphs - which I couldn't exactly reproduce, but found some interesting patterns, which may attract some interesting comments.

Taking the HadCRUT3 time series dataset at http://hadobs.metoffice.com/hadcrut3/diagn...cs/global/nh+sh, I followed what I could of the methodology in the thread and came up with this:

By taking monthly annual differences and performing a rolling mean over the year for each month, some of the microdetail in the record is brought out.

The rest of the 20th century record looks like this:

And the 19th century:

Whatever is going on, the 21st century so far is unusual.

[Devonian]

Most of us are familiar with 30 year means to show trends in global warming. Rolling means are one, fairly simple way to remove the noise from a temperature time series to show when there was a noticeable change in the trend.

Following a thread by Basil Copeland on Watts Up With That?, I was intrigued by the graphs - which I couldn't exactly reproduce, but found some interesting patterns, which may attract some interesting comments.

...

Whatever is going on, the 21st century so far is unusual.

I think this is a rather confusing way of showing what we already know - the weather has been warm and less extreme these last few years. Why? No cold winters, (which would be the extreme months) and warm summers.

Edited March 14, 2008 by Devonian

[Chris Knight]

I think this is a rather confusing way of showing what we already know - the weather has been warm and less extreme these last few years. Why? No cold winters, (which would be the extreme months) and warm summers.

I don't think it shows warmth - it shows that year on year over the last 150y, there is a neutral trend of warming or cooling over the globe, whatever time period anomalies are based against. There are warmer extremes and cooler extremes, and regularly they cancel each other out entirely.

To test this I applied the same algorithm to something we do know has a real upward trend - the 50-odd year record of CO₂ in the atmosphere at Mauna Loa, and plotted them against the same time period with the HadCRUT data. This is the result:

[VillagePlank]

Since 2002 the pattern has changed!

Look at how CO2 follows temperature and during the last six years the level of correlation, at a first visual inspection, seems to disappear completely. Yikes!

[Devonian]

I don't think it shows warmth - it shows that year on year over the last 150y, there is a neutral trend of warming or cooling over the globe, whatever time period anomalies are based against. There are warmer extremes and cooler extremes, and regularly they cancel each other out entirely.

Hang on, I think I've misunderstood what you did Did you take a rolling average and then plot the anomalies to that rolling average? If so one would surely expect a simple variation around the mean?

To test this I applied the same algorithm to something we do know has a real upward trend - the 50-odd year record of CO₂ in the atmosphere at Mauna Loa, and plotted them against the same time period with the HadCRUT data. This is the result:

Are you saying this is significant?

[Hiya]

Very good presentation of data. This firmly lays to rest the myth of temperature following carbon dioxide and provides us with a very interesting start to the 21st century.

I don't know if you have looked at the correlation of global temp with sunspots, but there are fairly obvious sine curves in temp with sunspots until ~1988, then it all goes off the rails. Since 2001 the sine curve has returned to normal with sunspots. Coincidence?

[Devonian]

Very good presentation of data. This firmly lays to rest the myth of temperature following carbon dioxide and provides us with a very interesting start to the 21st century.

The $64, 000 question is, will you be able to say that in 2010, 2020, 2030...

I don't know if you have looked at the correlation of global temp with sunspots, but there are fairly obvious sine curves in temp with sunspots until ~1988, then it all goes off the rails. Since 2001 the sine curve has returned to normal with sunspots. Coincidence?

Nope, lets see it can we?

A famous scientist of the past found a correlation between the level of lake Victoria and sunspots. Indeed, for several cycles it was clear. Then, in the 20's I think it was, the two lines diverged.

Fact is if a clear solar/temp correlation were there we'd have found it ages ago, any we do find don't stick or long.

[Chris Knight]

Since 2002 the pattern has changed!

Look at how CO2 follows temperature and during the last six years the level of correlation, at a first visual inspection, seems to disappear completely. Yikes!

Mauna Loa is only a single NH station, measuring CO₂. The temperature is global.

Using some subsets of HadCRUT -Extratropical above 30deg north and tropical, and 30N to -30S give better fits to recent end of the curve, worse fits at the beginning for Mauna Loa. Southern Hemisphere temperature series fit less well along the whole series. I'll try the Global CO₂ curve at some time.

I would say that the Arctic has a much greater role in the CO₂ balance than previously thought, and sea ice may be one of the most important sinks.

Is it significant? Who can say. It will be all change with the next clutch of VEI 5 eruptions or the next big El Niño. As to solar grand minima, it's all guesswork at the moment.

[Chris Knight]

The $64, 000 question is, will you be able to say that in 2010, 2020, 2030...

Nope, lets see it can we?

A famous scientist of the past found a correlation between the level of lake Victoria and sunspots. Indeed, for several cycles it was clear. Then, in the 20's I think it was, the two lines diverged.

Fact is if a clear solar/temp correlation were there we'd have found it ages ago, any we do find don't stick or long.

I'll stick my neck out as far as sunspots are concerned. Regarding the MM and Dalton minima, the climate was under the influence of much more Volcanism and El Niño events than the 20th Century was. Wet cool summers, mild winters. (Hmm... rings a bell?). Then long fiercely hot dry dusty plague ridden summers, and bitingly cold hard winters. Famines, floods, storms. Rising food prices and taxes to pay for other peoples misfortune and conflict. Wars and persecution for beliefs. The world during those times was in turmoil, social, economic, religious and political - relatively more than today, due among other things to climate. Spots weren't seen not because they were missing, but that they were unobserved, records lost or destroyed. There was so much expansion of knowledge, and so few enlightened people. Sunspots were so unexciting, and purposeless when the whole of the cosmos was expanding into view around the scientists of the time. Like last summer, sunny days were for recreation, not for spending hours drawing the daily changes of spots on a circle of paper, with an assistant manually tracking the sun across the sky with a primitive telescope and mount. Been there done that, lets do something else this year.

The sunspot reconstructions have not corrected (adequately) for the missing portions of the record. We are embarking on a great experiment over the next quarter century or so as far as solar observation goes.