For half an hour, I was now playing with an amusing meteorological question. If you want to use the approximation that the global weather is repeating itself, what is the previous year that is most similar to Summer 2014 and the previous year or so?
I don't want to reveal my full methodology because you're invited to test your own approach to the question. But I took the RSS data from 1979, and reduced to the periods "minus two or three years" up to "a July", and compared these periods for different years on one side and "the period up to July 2014" on the other side. I first subtracted the most recent global temperature anomaly, and then summed the squared differences (including the regional columns).
The squared differences from the monthly data "well before the final July" were suppressed exponentially, so that the weight decreased 2.718 times when you returned by a year or two.
When this was done with various choices of the parameters, the period up to July 2009 ended most similar to the period up to July 2014. The victory of this year 2009 wasn't spectacular but someone had to win and you may be interested in the winner, anyway.
Not just because of this victory of 2009 but because of the overall charts with an apparent trend, you may see that recent years have more similar global patterns – regional and/or month-on-month differences – to the present than some older years, like in the 1980s. That's true despite the fact that I have effectively subtracted the global mean temperature. If you draw some data of this sort, you will assure yourself that the climate across the globe, and not just the global mean temperature, is naturally drifting somewhere all the time.
The bulk of the climate change has nothing to do with the global mean temperature so if you focus on some hypothetical causes that affect the temperatures more or less uniformly across the globe, like the changes to the greenhouse effect, you will only explain a tiny fraction of the actual ongoing climate change.
If you start to investigate the approximate statement that the weather patterns through July 2014 are similar to those by July 2009, you may do various tests and predictions. For example, you may open the most recent ENSO report and see that five years ago and in the previous months, the ENSO indices (page 23) indeed behaved similarly to the present ones.
Back in 2009, the 2009-2010 El Niño episode began with the June-July-August 3-month period. However, now in 2014, it is less clear because the last known 3-month period, May-June-July, has the ONI index at just +0.1 while it was +0.4 five years earlier (the previous 3-month periods were much more similar). This can make a difference. The appearance of a new El Niño – which could make the year 2015 about as warm as the near-warmest year 2010 – is an open question for me. The ENSO models predict that it is significantly more likely than not for the new El Niño episode to start in the fall. But I am often puzzled by the extrapolations to the future that the models predict. You often see a nice sine curve in the recently measured data that is just approaching the "level zero" and you expect the curve to continue down – but most of the models predict a strange bounce from zero and an increase! The predicted curves just look heavily discontinuous and the main discontinuity is generally "now". I don't know whether these models have been successfully validated but visually, their predictions look highly unconvincing to me.
The most recent RSS global anomaly – in July 2014 – was +0.350 °C. In July 2009, it was +0.325 °C. You could say that the temperature has warmed up by 0.025 °C per five years which corresponds to +0.5 °C per century but of course, you can't take any similar data too seriously because you get significantly different results if you pick some previous months, especially June 2009 and 2014. The June 2009 temperature was 0.3 °C colder than July but in 2014, these two months were pretty much equally warm.
This huge difference of differences is sort of spectacular. Even if you pick two "most similar years" when it comes to the temperature differences, their global mean temperature's one-month jumps may still differ by a whopping 0.3 °C. I would conclude that changes of the global mean temperature by 0.3 °C per month may still be classified as "rather common noise". In 100 years i.e. 1200 months, the monthly temperatures have drifted by something like 0.7 °C. I am still stunned by the people who would be ready to believe that this is a sign of a problem. People who are ready to believe this crap must be unbelievably stupid. As far as I can say, it shows how incredibly accurately the Earth is able to stabilize the temperature despite the obvious noise and drift that the changing weather brings us.
It's totally clear that the Earth can't cancel the month-on-month temperature fluctuations "exactly". Things are changing, drifting, and many of the changes are accumulating. It's also clear that there are stabilizing mechanisms e.g. because the temperatures in the glaciation cycles still mimic the astrophysical (orbital) predictions within a good accuracy. This accuracy is somewhere in between the "totally clear noise" such as those 0.3 °C on one side, and temperature smaller (but not hugely smaller) than the many degree (e.g. 8 degrees Celsius) one gets in between the extrema of the glaciation cycles. The Earth's global mean temperature may be naturally determined by the external parameters with the accuracy of 1-2 °C. It's the kind of the temperature difference over longer periods (e.g. the cetennial scale) that one should treat as noise.