My best guess is that there are many factors to consider since what is appropriate for one purpose (e.g. forecasting) might not be appropriate for another purpose (e.g. past trend analysis or attempts at correlation with other data).
Sufficiently slow to make the third of a century ago economically disastrous source reduction (use less fossil fuel) proposals look insane, for both then, and now. The warming trend is so slow that unless it's time to close the patent offices, who can say what could be invented to solve the problems?
Besides, it's not as if source reduction will reduce the carbon. The developing world would simply laugh and enter the abandoned niche using coal while the impoverishment of the compliant would merely serve as an example of how environmental concern is the same as political and economic suicide.
Since we don't know if there are consequences to our adding more carbon dioxide to the atmosphere, caution suggests we examine ways of removing the carbon should we wish to (sequestration). And Iron fertilization would be incredibly cheap, quick and easy.
http://en.wikipedia.org/wiki/Iron_fertil...
(Short version: iron is not very soluble in ocean water and is commonly a deficient nutrient there. Adding iron causes microscopic plants to multiply. The plants absorb carbon and a significant number sink, sequestering the carbon on the ocean floor. Nor is this just theory. Every part has been observed as natural processes or as experimental results. The overall method has been observed both ways.
Amazingly a number of people are all exited about a recent experimental result that allegedly left oxygen depleted zones. It was as if they'd never heard of eutrophication:
http://en.wikipedia.org/wiki/Eutrophicat...
Just add the iron more diffusely and you'll get results in line with the over a decades worth of more positive observed experimental AND natural results. Since this is an imitation of a natural process, either the "dead zones" are an avoidable abnormality or not significant to the environment. Take your pick.)
If you try to correlate CO2 data with temperature, it doesn't work very well. Some argue linear is the best fit but when temperatures suddenly rose in the 1980s to 2000 there wasn't an increase at that time that would explain the rise and when temps failed to rise with ever increasing CO2 since 2000, obviously it didn't fit then either. About the most we can say is that CO2 may have an effect on temperature but it can't be determined with any certainly based solely on looking at the temperature versus CO2 trend lines.
That is why IMO it doesn't really matter what type of trend line you use. As Ottawa implied, there are probably many factors, some acting independently, and therefore using the past to make predictions isn't very scientific IMO. As a geologist, I am all about looking at the past to see what you can tell about the present and future but in this case IMO, the past shows variation above the likely influence of CO2 making it hard to predict the future.
Moving average.
It really doesn't matter. Once one method has been as established, history has proven the greenies will only accept that method as long as it supports their cause. Once we all switch to that established method and it suddenly goes against the greenie agenda, then it is considered no good and a new method is devised.
I think what is more important is the starting and ending dates you select.
Global warming deniers have found that by selecting specific dates, they can make it seem like it's not warming.
A moving, 5-10 year, average, over the last few decades, should give you the most honest picture.
10 year moving average
Linear? Polynomial? Moving average?
