In the figure above I have marked the number of years after which a regression between the carbon dioxide concentration in the atmosphere and the area of the Arctic ice sheet is statistically significant at a risk level of one percent (P<0.01) in two consecutive years. In X-axis I marked the year from which each analysis series started.
I used Mauna Loa statistics as the carbon dioxide estimate and the area of Arctic ice sheet as the proxy for the global temperature in the month when it reaches its minimum size (September). The Arctic ice sheet is used because I have previously noted that estimates based on other types of measurements are under a continuous change, and therefore I do not trust them. However, the minimum area or the Arctic sheet correlates statistically significantly with those estimates.
Using one percent risk level instead of more common five percent was due to the fact that I made a total of 740 regression analyses for this analysis. Therefore using the common five percent risk level, I could have been expected almost 40 "significant" results by chance.
Actually, approximately seven incorrect "significance levels" could be expected also using one percent criterion. Therefore I decided to make calculations until I find significance in consecutive two years. The number of one year significance levels was very close to random expectations, that is five.
The theoretical framework of this analysis is the well known theory stating that the global warming is caused by increasing atmospheric greenhouse concentration (carbon dioxide being the most important), and that the warming is accelerating in time. From that theory it was possible to derive a testable hypothesis, saying that a statistical dependence between the two will be reached more and more quickly.
This means that if the hypothesis is correct, i.e. the climatic warming is caused by carbon dioxide concentration, the columns in the figure above should shorten from left to right. If not, the hypothesis becomes falsified and should be discarded.
It is clear that the left side of the figure (yeasts 1979-2001) is according to the hypothesis, and the columns are shortening from left to the right. Then something happens, and in year 2002 there is a jump higher. Actually, in statistics from 2002 and thereafter there are no consecutive two years with statistical significance. I have marked these columns with red color, and their shortening toward right is only because of the number of years available for the analysis.
According to this analysis the theory of accelerating global warming due to greenhouse gases (carbon dioxide) should be falsified. It also suggests that the rise of greenhouse concentration in the atmosphere and global warming in 1979-2001 were only correlative and not in a causative relationship.
It remains to be seen how future statistical information affects this kind of analysis. After all, science does not have final truths, but it corrects itself whenever needed. According to this principle, I will continue similar analyses as this in the following years, and report in this blog. Stay tuned!
Previous thoughts on the same topic:
A scoop to be kept out of daylight
Climate change raises dishonesty in the Finnish media
The original though in Finnish:
Napajää ei tue yhteyttä globaalin lämpenemisen ja ilmakehän hiilidioksidipitoisuuden välillä
No comments:
Post a Comment
You are free to comment on the blog posts, but I ask you to stay on topic and adhere to respectful language and good manners.